Research Report 01 - 2016

EXPERIMENTAL RESEARCHES ON THE SEISMIC BEHAVIOUR
OF MASONRY SPANDRELS: AN INTERNATIONAL PERSPECTIVE
edited by
Nicola Augenti
University of Naples Federico II, Italy
Francesco Graziotti
University of Pavia and EUCENTRE, Italy
Guido Magenes
Fulvio Parisi
EXPERIMENTAL RESEARCHES ON THE SEISMIC BEHAVIOUR
OF MASONRY SPANDRELS: AN INTERNATIONAL PERSPECTIVE
edited by
Nicola Augenti
Francesco Graziotti
Guido Magenes
Fulvio Parisi
According to law, EUCENTRE Foundation trademark cannot be reproduced, copied or utilized,

without the written permission of the EUCENTRE Foundation, which is the owner, except in
accordance with established contract conditions pertaining to the production of this document.
July, 2016
Pavia, Italy
Nessuna parte di questa pubblicazione può essere riprodotta o trasmessa in qualsiasi forma o con qualsiasi
mezzo elettronico, meccanico o altro senza l’autorizzazione scritta dei proprietari dei diritti e dell’editore.
No parts of this publication may be copied or transmitted in any shape or form, and by any type of electronic,
mechanical or different means, without the prior written permission of the copyright holder and the publisher.
© Copyright 2016 - EUCENTRE
prodotto e distribuito da:

produced and distributed by:
EUCENTRE Foundation
Via Adolfo Ferrata, 1 - 27100 Pavia, Italy
Phone (+39) 0382.5169811 - Fax (+39) 0382.529131
E-mail: info@eucentrepress.it - Web: www.eucentrepress.it
ISBN: 978-88-6198-124-9
ABSTRACT
Seismic response of unreinforced masonry buildings is strongly influenced by in-plane behaviour
and damage of spandrels. In the last decade, this issue has been one of the most investigated topics
in earthquake engineering research programmes dealing with masonry structures. Several working
groups in different countries have focused on numerical simulation and experimental testing of
spandrels, considering their variability in terms of masonry type, presence of tensile-resistant
elements such as steel ties and reinforced concrete bond beams, magnitude of gravity loads, and
spandrel geometry. Some research groups have also explored the role of innovative strengthening
systems aimed at increasing strength and/or deformation capacity of spandrel panels above
openings.
This report provides a comprehensive discussion of experimental researches carried out by four
research groups in Italy, Switzerland and New Zealand. In all cases, most of experimental tests were
carried out on full-scale masonry specimens, either focusing on spandrel panels or addressing the
pier-spandrel interaction within in-plane laterally loaded walls. More in detail, this report is asimed at
reflecting and presenting the complementary nature of recent researches on spandrels. Special
emphasis is given to observed damage, force–displacement behaviour and nonlinear capacity
measures of spandrels. Valuable data on energy dissipation capacity and ultimate drift of spandrel
panels and perforated masonry walls are also reported and discussed. In-plane overstrength and
displacement ductility capacity of masonry walls with single openings are characterised through
bilinear idealisation of experimental force–displacement diagrams. All testing programmes show that
the geometrical and construction features of spandrels significantly influence the in-plane seismic
capacity of perforated masonry walls and their repairability after cyclic loading. The latter feature is
measured through the ratio of residual drift to the maximum drift of each wall specimen. Rocking
behaviour of piers notably increases demand on spandrel panels, inducing the formation of plastic
hinges in reinforced concrete bond beams at spandrel-pier intersections. Finally, seismic capacity of
perforated walls is also found to depend on the spandrel-pier connection, highlighting the influence
of boundary conditions on nonlinear behaviour of spandrel panels.
The type and amount of experimental data collected in this report can support the improvement of
macro-element capacity models and building codes for seismic performance assessment of masonry
buildings. Several issues require further numerical and experimental investigation and are identified
in each section of the report. The extension of experimental testing to full-scale perforated masonry
walls with multiple storeys and openings is one of those research needs, starting from preliminary
findings on half-scale specimens presented herein.
i
ACKNOWLEDGEMENTS
The research described in this report was partially funded by:

- the Italian Department of Civil Protection (Dipartimento della Protezione Civile) in the
framework of DPC-ReLUIS 2005–2008 Project (Subproject 1 “Evaluation and reduction of
the vulnerability of masonry buildings”) and DPC-ReLUIS 2010–2013 Project (Subproject
AT1-1.1: “Evaluation of the vulnerability of masonry buildings, historic centres and cultural
heritage”);
- the KGV Prevention Foundation in the framework of the research project, "Nonlinear
deformation behavior of unreinforced masonry structures through testing and numerical
simulations" (PI: Alessandro Dazio);
- the New Zealand Foundation for Research, Science and Technology, under the contract
“Retrofit Solutions for New Zealand's Earthquake Risk Multi-storey Buildings”, 2004.
The opinions and conclusions expressed in this report are those of the authors and do not
necessarily reflect those of the funding institutions.
ii
TABLE OF CONTENTS
ABSTRACT ..................................................................................................................................... I
ACKNOWLEDGEMENTS ............................................................................................................... II
TABLE OF CONTENTS ................................................................................................................ III
LIST OF TABLES .........................................................................................................................VII
LIST OF FIGURES ......................................................................................................................VIII
1 INTRODUCTION ....................................................................................................................... 1
1.1 MOTIVATION FOR THE REPORT .......................................................................................... 1
1.2 AIM AND OBJECTIVES ....................................................................................................... 1
1.3 ORGANISATION OF THE REPORT ........................................................................................ 2
2 EXPERIMENTAL CAMPAIGN ON DOUBLE-LEAF STONE MASONRY SPECIMENS AT THE

UNIVERSITY OF PAVIA AND EUCENTRE PAVIA ........................................................................ 5
2.1 SUMMARY ..................................................................................................................... 5
2.2 DESCRIPTION OF THE EXPERIMENTAL CAMPAIGN ............................................................... 6
2.3 DOUBLE-LEAF STONE MASONRY WALLS ........................................................................... 6
2.3.1 STONES.............................................................................................................. 7
2.3.2 MORTAR ............................................................................................................. 7
2.1. COMPRESSION TESTS ON STONE MASONRY SPECIMENS .................................................... 7
2.3.3 VERTICAL COMPRESSION TEST ............................................................................ 8
2.3.4 DIAGONAL COMPRESSION TEST ........................................................................... 8
2.4 DESCRIPTION OF THE SPANDREL SPECIMENS..................................................................... 8
2.5 PROPERTIES OF THE TEST SETUP .................................................................................... 10
2.6 INSTRUMENTATION OF THE SPECIMENS ............................................................................ 12
2.6.1 ACTUATORS FORCE TRANSDUCERS ................................................................... 13
2.6.2 DISPLACEMENT TRANSDUCERS (POTENTIOMETERS) ........................................... 13
2.6.3 ADVANCED VIDEO ACQUISITION ......................................................................... 13
2.7 TEST PROCEDURE .......................................................................................................... 14
2.7.1 LOADING HISTORY OF THE FIRST SPANDREL SPECIMEN (S1) ............................... 14
2.7.2 LOADING HISTORY OF THE SECOND SPANDREL SPECIMEN (S2) ........................... 14
2.7.3 LOADING HISTORY OF THE THIRD SPANDREL SPECIMEN (S3) .............................. 15
2.7.4 LOADING HISTORY OF THE FOURTH SPANDREL SPECIMEN (S4) ........................... 15
2.8 TEST RESULTS FOR THE FIRST SPECIMEN ........................................................................ 16
iii
2.8.1 TEST OBSERVATIONS ........................................................................................ 17
2.8.2 FORCE–DISPLACEMENT BEHAVIOUR OF THE FIRST SPECIMEN ............................. 19
2.9 TEST RESULTS FOR THE SECOND SPECIMEN .................................................................... 21
2.9.2 FORCE–DISPLACEMENT BEHAVIOUR OF THE SECOND SPECIMEN ......................... 24
2.10 TEST RESULTS FOR THE THIRD SPECIMEN ....................................................................... 28
2.10.2 FORCE–DISPLACEMENT BEHAVIOUR OF THE THIRD SPECIMEN............................. 32
2.11 TEST RESULTS FOR THE FOURTH SPECIMEN .................................................................... 32
2.11.2 FORCE–DISPLACEMENT BEHAVIOUR OF THE FOURTH SPECIMEN ......................... 35
2.12 COMPARISON OF THE RESULTS OF THE FOUR SPECIMENS ................................................ 36
2.12.1 FORCE/DISPLACEMENT ..................................................................................... 36
2.1.1. DISSIPATED ENERGY ......................................................................................... 38
2.12.2 IDEALIZATION OF THE RESPONSE ........................................................................ 40
2.12.3 STIFFNESS........................................................................................................ 41
2.12.4 RELEVANT DEFORMATION THRESHOLDS ............................................................. 42
2.12.5 PEAK AND RESIDUAL STRENGTHS ....................................................................... 43
2.13 CONCLUSIONS ................................................................................................................ 44
2.14 FUTURE DEVELOPMENTS ................................................................................................. 44
2.15 REFERENCES ............................................................................................................. 46
3 MONOTONIC AND CYCLIC TESTS ON TUFF STONE MASONRY SPANDRELS AT THE

UNIVERSITY OF NAPLES FEDERICO II ..................................................................................... 47
3.1 SUMMARY ................................................................................................................... 47
3.2 INTRODUCTION ............................................................................................................... 48
3.3 EXPERIMENTAL PROGRAMME .......................................................................................... 48
3.3.1 MATERIAL PROPERTIES ..................................................................................... 48
3.3.2 GEOMETRY AND CONSTRUCTION FEATURES OF SPECIMENS ................................ 49
3.3.3 EXPERIMENTAL SETUP AND INSTRUMENTATION ................................................... 52
3.3.4 TEST PROCEDURES .......................................................................................... 53
3.4 CRACK PATTERNS .......................................................................................................... 55
3.5 EXPERIMENTAL FORCE–DRIFT CURVES ........................................................................... 62
3.6 DUCTILITY CAPACITY AND OVERSTRENGTH...................................................................... 66
3.7 ANALYSIS OF PIER DRIFTS .............................................................................................. 70
3.8 HYSTERETIC DAMPING .................................................................................................... 71
3.9 RESIDUAL DRIFTS........................................................................................................... 72
3.10 CONCLUSIONS ................................................................................................................ 73
iv
3.11 REFERENCES ............................................................................................................. 76
4 QUASI-STATIC CYCLIC AND MONOTONIC TESTS ON MASONRY SPANDRELS AND

COMPOSITE SPANDRELS AT ETH ZÜRICH ............................................................................. 79
4.1 SUMMARY ................................................................................................................... 79
4.2 INTRODUCTION ............................................................................................................... 80
4.3 EXPERIMENTAL PROGRAMME .......................................................................................... 81
4.4 TEST SETUP AND LOADING HISTORY ............................................................................... 83
4.5 HYSTERETIC BEHAVIOUR AND FAILURE MODES ............................................................... 86
4.5.1 MASONRY SPANDRELS ...................................................................................... 86
4.5.2 COMPOSITE SPANDRELS ................................................................................... 90
4.6 CONCLUSIONS ................................................................................................................ 94
4.7 REFERENCES ............................................................................................................. 95
5 SPANDREL TESTING A THE UNIVERSITY OF AUCKLAND ................................................ 97

5.1 SUMMARY ................................................................................................................... 97
5.2 INTRODUCTION ............................................................................................................... 98
5.2.1 SUBSTRUCTURE TESTING .................................................................................. 98
5.2.2 TESTING MATRIX ............................................................................................... 99
5.2.3 CONSTRUCTION DETAILS ................................................................................. 100
5.2.4 MATERIAL PROPERTIES ................................................................................... 100
5.2.5 TEST SETUP ................................................................................................... 101
5.2.6 TESTING PROCEDURE ..................................................................................... 102
5.2.7 TEST OBSERVATIONS ...................................................................................... 102
5.2.8 DEFORMATION MODES .................................................................................... 104
5.2.9 SPANDREL ELONGATION .................................................................................. 108
5.2.10 FORCE–DISPLACEMENT RESPONSE ................................................................. 108
5.2.10.1 Bilinearisation .............................................................................. 110
5.2.10.2 Ductility Capacity ......................................................................... 110
5.2.10.3 Ultimate Drift ............................................................................... 111
5.2.11 CONCLUSIONS FOR SUBSTRUCTURE TESTING ................................................... 112
5.3 TWO STOREY PERFORATED WALL ................................................................................ 113
5.3.1 MATERIAL PROPERTIES ................................................................................... 114
5.3.2 SCALING ......................................................................................................... 115
5.3.3 TEST SETUP ................................................................................................... 115
5.3.4 TESTING PROCEDURE ..................................................................................... 116
5.3.5 TEST OBSERVATIONS ...................................................................................... 117
5.3.6 FORCE–DISPLACEMENT RESPONSE ................................................................. 117
5.3.7 GLOBAL W ALL BEHAVIOUR .............................................................................. 119
v
5.3.8 SPANDREL BEHAVIOUR .................................................................................... 121
5.3.8.1 Central Spandrel Deformations ................................................... 121
5.3.8.2 Arched Spandrel Damage ........................................................... 122
5.3.9 BILINEAR MODELLING ...................................................................................... 123
5.3.10 DUCTILITY CAPACITY ....................................................................................... 124
5.3.11 ULTIMATE DRIFT ............................................................................................. 125
5.4 CONCLUSIONS .............................................................................................................. 125
5.5 REFERENCES ........................................................................................................... 127
6 CONCLUSIONS .................................................................................................................... 129

6.1 SUMMARY OF EXPERIMENTAL RESULTS ......................................................................... 129
6.2 POTENTIAL FOR FUTURE RESEARCH DEVELOPMENTS .................................................... 131
vi
LIST OF TABLES
Table 2.1. Results of the vertical compression tests [Magenes et al., 2010]. ................................... 8
Table 2.2. Results of the diagonal compression tests [Magenes et al., 2010]. ................................. 8
Table 2.3. Summary of the results of experimental wall compression tests [Magenes et al., 2010]. 8
Table 2.4. Load sequence followed for the test on spandrel specimen S1. ................................... 15
Table 2.8. Spandrel geometry and idealized response. ................................................................. 41
Table 2.9. Spandrel relevant deformation thresholds in terms of equivalent piers rotation. ........... 42
Table 3.1. Global experimental response parameters of subassemblages. ................................... 64
Table 3.2. Idealised SDOF system properties and overstrength factor according to Tomaževič.... 68
Table 3.3. Idealised SDOF system properties and overstrength factor according to IBC and EC8.68
Table 3.4. Average estimates of SDOF system properties and overstrength factor according to
Tomaževič. ............................................................................................................................ 68
Table 3.5. Average estimates of SDOF system properties and overstrength factor according to IBC
and EC8................................................................................................................................. 69
Table 4.1. Masonry spandrels: Loading scheme, spandrel type and details of the axial load
application for the four test units. ........................................................................................... 82
Table 4.2. Composite spandrels: Loading scheme, spandrel type and details of the axial load
application for the four test units. ........................................................................................... 83
Table 5.1. Substructure dimensions. ............................................................................................. 99
Table 5.2. Pseudo-static testing matrix. ...................................................................................... 100
Table 5.3. Substructure material properties................................................................................. 101
Table 5.4. Experimental results for pseudo-static testing of substructures. ................................. 104
Table 5.5. Bilinear parameters for substructures when using the Tomaževič method. ................ 111
Table 5.6. Assessment of drift levels achieved in the substructure testing. ................................. 112
Table 5.7. Pier geometry. ............................................................................................................ 114
Table 5.8. Two storey wall material properties. ........................................................................... 115
Table 5.9. Bilinear parameters in accordance with the Tomaževič [2000] procedure. ................. 123
Table 5.10. Bilinear parameters according to the Magenes and Calvi [1997] procedure. ............ 124
vii
LIST OF FIGURES
Figure 2.1. Example of the double-leaf stone masonry [Giuffrè, 1993] and a picture of a masonry
typology used for this experimental campaign.......................................................................... 6
Figure 2.2. Values of flexural tensile (left) and compressive (right) strength for mortar samples tested
at the same time of the spandrel specimens. ........................................................................... 7
Figure 2.3. Geometry of specimens S1 and S2 double-leaf stone masonry specimens (dimensions
in mm)...................................................................................................................................... 9
Figure 2.4. Geometry of specimens S3 (left) and S4 (right) double-leaf stone masonry specimens
(dimensions in mm).................................................................................................................. 9
Figure 2.5. Picture of specimen S1 during and after construction. ................................................... 9
Figure 2.6. Static scheme of the specimen during a test. .............................................................. 10
Figure 2.7. Scheme of the test setup............................................................................................. 11
Figure 2.8. Comparison of the deformation fields corresponding to an uplift of the right pier (left) and
a rotation of the bases of the piers (right). .............................................................................. 11
Figure 2.9. Front view of the test apparatus. ................................................................................. 12
Figure 2.10. Position of the load cells measuring the forces acting on the specimen and on the
structure (left) and hydraulic jack and load cell of the spandrel’s horizontal rods (right). ........ 13
Figure 2.11. Layout of the displacement transducers on the specimen (front/rear)........................ 13
Figure 2.12. Time history of monotonic and cyclic loading with significant cracking points for
specimen S1. ......................................................................................................................... 17
Figure 2.13. Crack pattern of spandrel at C3 (θn = –0.1%) (left) and C8 (θn = 0.4%) (right)............ 18
Figure 2.14. Crack pattern of spandrel and view of the lintel sliding at C10 (θn = 0.6%)................ 18
Figure 2.15. View of the entire specimen at C10 (θn = 0.6%) and C15 (θn = 1.5%). ....................... 18
Figure 2.16. Evolution of the crack pattern observed during the tests on specimen S1. ................ 19
Figure 2.17. Force–displacement cycles of specimen S1, considering the displacement of the base
of the right pier δ and the equivalent base rotation α. ............................................................. 20
Figure 2.18. Force–displacement cycles of specimen S1, considering the relative displacement of
the two interfaces of the spandrel δs and the spandrel drift θel................................................ 21
Figure 2.19. Cyclic loading time history with significant cracking points for specimen S2. ............. 22
Figure 2.20. Crack pattern of spandrel at C6_75 (θn = 0.25%) and C14_38 (θn = 1.25%). ............ 22
Figure 2.21. Crack pattern of spandrel corresponding to positive (left) and negative (right)
displacements C19_38 (θn = ±3%). ........................................................................................ 23
Figure 2.22. Crack pattern of the spandrel at the end of the test. .................................................. 23
Figure 2.23. Evolution of the crack pattern observed during the test on specimen S2. .................. 24
Figure 2.24. Force–displacement cycles of the second specimen (Ff = 75 kN), considering the
displacement of the base of the right pier δ and the equivalent base rotation α...................... 25
Figure 2.25. Force–displacement cycles of specimen S2 (Ff = 38 kN), considering the displacement
of the base of the right pier δ and the equivalent base rotation α. .......................................... 25
viii
Figure 2.26. Force–displacement cycles of specimen S2 (Ff = 19 kN), considering the displacement
of the base of the right pier δ and the equivalent base rotation α. .......................................... 26
Figure 2.27. Force–displacement cycles of specimen S2 (Ff = 75 kN), considering the relative
displacement of the two interfaces of the spandrel δs and the spandrel drift θel. ..................... 27
Figure 2.31. Crack pattern of spandrel at C6 (θn = 0.3%) and C9 (θn = 0.6%). .............................. 29
Figure 2.32. Rear crack pattern of spandrel at C11 (θn = 0.8%). ................................................... 30
Figure 2.33. Crack pattern of spandrel during positive (left) and negative (right) displacements C17
(θn = ±3%). ............................................................................................................................. 30
Figure 2.34. Crack pattern of the S3 spandrel at the end of the test. ............................................. 31
Figure 2.35. Evolution of crack pattern evolution observed during the different tests on the third
specimen. .............................................................................................................................. 31
Figure 2.38. Crack pattern of spandrel at C5 (θn = 0.25%) and C11 (θn = 0.5%). .......................... 33
Figure 2.39. Rear crack pattern of spandrel at C11 (θn = 0.8%). ................................................... 34
Figure 2.40. Crack pattern of spandrel during positive displacement at C17 (θ n = 3%) and during
negative displacement at C18 (θn = –4%)............................................................................... 34
Figure 2.41. Evolution of the crack pattern observed during the test on specimen S4. .................. 35
Figure 2.42. Crack pattern of the S4 spandrel at the end of the test. ............................................. 35
Figure 2.44. Shear versus relative displacement curves obtained from the tests on specimens S1
(dashed line) and S2 (Ff = 38 kN) (solid line). ........................................................................ 37
Figure 2.45. Shear versus base displacement curves obtained from the tests on specimens S4
(dashed line) and S3 (solid line) (Ff = 28 kN). ........................................................................ 37
Figure 2.46. Shear versus base displacement envelopes obtained from the tests on specimens S1
(grey dashed line), S2 (grey solid line), S3 (black solid line) and S4 (black dashed line). ....... 38
Figure 2.47. Dissipated energy versus base displacement curves obtained from the tests on
specimens S1 (grey dashed line), S2 (grey solid line), S3 (black solid line) and S4 (black dashed
line). ....................................................................................................................................... 39
Figure 2.48. Hysteretic damping ξhyst vs. base displacement curves obtained from the test on S1
(red dashed), S2 (red solid), S3 (black solid) and S4 (black dashed). .................................... 40
Figure 2.49. Idealization of the cyclic response. ............................................................................ 41
Figure 2.50. Idealized backbone curves for S1 (red dashed), S2 (red solid), S3 (black solid) and S4
(black dashed). ...................................................................................................................... 41
Figure 2.51. Strut and tie hypothesis for the flexural behaviour on S1. .......................................... 43
ix
Figure 3.1. Geometry of (a) subassemblage S1a and (b) subassemblage S1c (dimensions in mm).
.............................................................................................................................................. 50
Figure 3.2. Geometry of subassemblage S2 (dimensions in mm). ............................................. 50
Figure 3.3. Geometry of (a) subassemblage S3a and (b) subassemblage S3b (dimensions in mm).
.............................................................................................................................................. 51
Figure 3.4. Experimental setup. .................................................................................................... 52
Figure 3.5. Instrumentation of specimen S1 (dimensions in mm). ................................................. 53
Figure 3.8. Input displacement time-histories for cyclic tests: (a) Subassemblages S1b and S1c and
(b) subassemblages S2, S3a and S3b. .................................................................................. 54
Figure 3.9. Crack patterns of (a) subassemblage S1a, (b) subassemblage S1b and (c)
subassemblage S1c............................................................................................................... 56
Figure 3.10. Crack patterns of subassemblage S2. ....................................................................... 57
Figure 3.11. Crack patterns of (a) subassemblage S3a and (b) subassemblage S3b. .................. 58
Figure 3.12. Damage to spandrel panels at the maximum drift levels: (a) Subassemblage S1a (max
= 0.9%), (b) subassemblage S1b (max = 1.1%), (c) subassemblage S1c (max = 2.5%), (d)
subassemblage S2 (max = 3.1%), (e) subassemblage S3a (max = 1.1%) and (f) subassemblage
S3b (max = 0.9%)................................................................................................................... 59
Figure 3.13. Overall damage at the maximum drift levels: (a) Subassemblage S1a (max = 0.9%), (b)
subassemblage S1b (max = 1.1%), (c) subassemblage S1c (max = 2.5%), (d) subassemblage
S2 (max = 3.1%), (e) subassemblage S3a (max = 1.1%) and (f) subassemblage S3b (max =
0.9%). .................................................................................................................................... 60
Figure 3.14. Damage to the piers of subassemblage S2: (a) Rocking-induced crack of the right pier
and (b) crack width in the left pier. ......................................................................................... 61
Figure 3.15. Damage to the piers of subassemblage S3b: (a) Masonry crushing at the base of the
left pier, (b) diagonal shear cracking of the right pier, (c) rocking-induced crack width at the base
of the left pier and (d) crack width in the right pier. ................................................................. 61
Figure 3.16. Experimental force–drift hysteretic curves and envelope curves of (a) subassemblage
S1a, (b) subassemblage S1b and (c) subassemblage S1c. ................................................... 63
Figure 3.17. Experimental force–drift hysteretic curve and envelope curve of subassemblage S2.65
Figure 3.18. Experimental force–drift hysteretic curves and envelope curves of (a) subassemblage
S3a and (b) subassemblage S3b. .......................................................................................... 65
Figure 3.19. Comparison between experimental force–drift envelopes. ........................................ 66
Figure 3.20. Idealised force–drift diagrams of subassemblages. ................................................... 69
Figure 3.21. Lateral resisting force versus variation between pier drifts. ....................................... 70
Figure 3.22. Hysteretic damping ratio versus subassemblage drift. ............................................... 72
Figure 3.23. Residual drift versus subassemblage drift. ................................................................ 73
Figure 4.1. Shear failure of spandrels with shallow masonry arches in an old URM building after the
L’Aquila earthquake on April 6, 2009: Entire building (left) and detail of a spandrel (right). .... 80
Figure 4.2. Flexural failure of spandrels in an old URM building after the L’Aquila earthquake on April
6, 2009: Entire building (left) and detail of a spandrel (right). ................................................. 80
x
Figure 4.3. Modern URM buildings in Switzerland with composite spandrels (left) and RC slabs
(right). .................................................................................................................................... 81
Figure 4.4. Composite spandrels: Brick Type 1 with non-continuous webs (left) and Brick Type 2 with
continuous webs (right). ......................................................................................................... 83
Figure 4.5. Masonry spandrels: Drawing of the test setup. View from the East without side restraint
(dimensions in mm; LF = Hollow core jacks connected to load follower, which keeps the oil
pressure constant). ................................................................................................................ 84
Figure 4.6. Composite spandrels: Drawing of the test setup. View from the East without side restraint
(dimensions in mm)................................................................................................................ 84
Figure 4.7. Masonry spandrels: Photo of the test setup. View from the North-East. ...................... 85
Figure 4.8. Composite spandrel: Photo of the test setup. View from the East. .............................. 85
Figure 4.9. Loading history for quasi-static monotonic and cyclic spandrel tests. .......................... 86
Figure 4.10. Masonry spandrels at failure. .................................................................................... 88
Figure 4.11. Masonry spandrels: Spandrel shear–force deformation hysteresis. ........................... 89
Figure 4.12. Masonry spandrels: Hysteresis curves of the axial loads in the spandrels. ................ 90
Figure 4.13. Composite spandrels: Spandrel shear–force deformation hysteresis. ....................... 92
Figure 4.14. Composite spandrel TU1 at failure (Brick Type 1). .................................................... 93
Figure 4.15. Composite spandrel TU2-5 at failure (Brick Type 2). ................................................. 93
Figure 5.1. Failure patterns from the Canterbury earthquakes: (a) Spandrel failure in Avonmore
House, Christchurch and (b) pier failure in Kenton Chambers, Christchurch. ......................... 98
Figure 5.2. Typical geometry and layout of substructure for pseudo-static test: Side elevation (left)
and end elevation (right). ....................................................................................................... 99
Figure 5.3. Pseudo-static testing set-up: Side elevation (left) and end elevation (right). .............. 101
Figure 5.4. Loading history for testing of the pier/spandrel substructures. ................................... 102
Figure 5.5. (a) Spandrel cracking in all bed joints of PS2 and (b) Rocking at base of Pier B in PS2.
............................................................................................................................................ 103
Figure 5.6. (a) Rocking at the base of pier B in PS3 and (b) Development of the diagonal cracks in
the final push cycle of PS4. .................................................................................................. 103
Figure 5.7. Rotation in spandrel of PS5. ...................................................................................... 103
Figure 5.8. Crack patterns in substructures after cyclic pseudo-static testing (Pier A on the left): (a)
PS1, (b) PS2, (c) PS3, (d) PS4, (e) PS5 and (f) PS6. .......................................................... 105
Figure 5.9. Spandrel failure modes from experimental testing: (a) Rotation in the spandrel of PS2
and (b) diagonal cracking in the spandrel of PS5. ............................................................... 106
Figure 5.10. Spandrel deformation modes: (a) Mode A – Flexural cracking, (b) Mode B – Diagonal
cracking and (c) Mode C – Bending-bed joint sliding............................................................ 106
Figure 5.11. Photographic examples of spandrel deformation modes observed following the
earthquakes in Christchurch: (a) Flexural cracking in the central spandrel of a two storey
perforated URM façade, (b) flexural cracking in a spandrel element in Avonmore House and (c)
diagonal shear cracking in the spandrel above a high arched lintel in Avonmore House. ..... 107
Figure 5.12. Vertical displacement of spandrel in PS3. ............................................................... 108
Figure 5.13. Substructure PS5: (a) Spandrel extension against drift and (b) spandrel extension
against estimated spandrel axial load. ................................................................................. 108
xi
Figure 5.14. Bilinear idealisation and backbone curves of (a) PS1 and (b) PS2. ......................... 109
Figure 5.17. Two storey perforated test wall: (a) Side view and (b) end view (dimensions in mm).
............................................................................................................................................ 114
Figure 5.18. Two storey wall testing set-up: (a) Side view and (b) end view. ............................... 116
Figure 5.19. Loading history for testing of the two storey wall. .................................................... 117
Figure 5.20. Photos of two storey wall testing: (a) Horizontal crack at the top of Pier AT, (b) crack
propagating from arches horizontally across top of Pier BT and (c) cracking through all bed-joints
in central spandrel component (photos taken from the reverse side of the wall, so that Pier A is
the far right pier, and Pier D is the far left pier). .................................................................... 118
Figure 5.21. Crack pattern for the as-built two storey perforated wall after cyclic pseudo-static testing.
............................................................................................................................................ 119
Figure 5.22. Force–displacement response recorded from (a) top actuator and (b) mid-height
actuator................................................................................................................................ 120
Figure 5.23. (a) Mid and top force–displacement response and (b) force–displacement response for
total force against top displacement. .................................................................................... 120
Figure 5.24. Simplified wall deformation: (a) Push cycle and (b) pull cycle. ................................. 121
Figure 5.25. (a) Separation of Pier D and top spandrel and (b) sliding between Pier A and top
spandrel. .............................................................................................................................. 121
Figure 5.26. Mode C spandrel deformation: (a) Deformation of the coupling spandrel between two
rocking piers, (b) rotation of spandrel stepped over brick courses and (c) sliding between brick
courses due to displacement incompatability (see also Figure 5.10). ................................... 122
Figure 5.27. Crack patterns above arched lintels: (a) Cracking above arched lintel and (b) Arched
lintel with bricks dislodging . ................................................................................................. 123
Figure 5.28. Bilinear idealisation of (a) the total response and (b) top floor response. ................. 124
Figure 5.29. Bilinear idealisation of the first floor response. ........................................................ 125
xii
N. Augenti, F. Graziotti, G. Magenes, F. Parisi (Editors) EUCENTRE 1
Research Report
1 INTRODUCTION
1.1 Motivation for the Report
The structural role of masonry spandrels in the seismic performance of unreinforced masonry (URM)
buildings has been for a long time a matter of theoretical speculation, without any clear experimental
reference to validate modelling assumptions from a quantitative viewpoint. The progress of analytical
methods that developed during the 1980s and 1990s for seismic response analysis of masonry
buildings has allowed to point out the importance of such a structural role and the sensitivity of
analysis results to the assumptions on the behaviour of spandrel elements. Those developments
called therefore for a better understanding of the real behaviour of masonry spandrels, and
consequently for specific experimental tests aimed to build up data-driven seismic capacity models.
Besides, until recently, seismic codes at both national and international levels were not providing
specific capacity models and verification rules for URM spandrels, as a sound experimental basis
was not available.
In recent years, several research programmes involving experimental tests on full-scale masonry
spandrel specimens or spandrel-pier subassemblages were carried out in Italy, Switzerland and New
Zealand. The primary objective of those researches was to develop a first set of reference
experimental data in the field. The Editors of the present report, whose research was carried out in
great part under the Italian Reluis-DPC 2005-2008 and 2010-2013 framework programmes,
considered as a useful initiative to organically collect and make available to the scientific community
the main results of the research efforts carried out internationally into a joint publication that could
represent a reference and startpoint for researchers involved in the study and modelling of masonry
structures. Clearly, the collaboration of the colleagues from Switzerland and New Zealand was
essential in this endeavour, and their contribution to this report is therefore highly appreciated.
1.2 Aim and Objectives
The primary aim of this report is to present and summarise the main results of recent experimental
research projects dealing with the influence of spandrels on nonlinear seismic response of in-plane
laterally loaded masonry walls.
All contributors to this research recognised the need for two main features of specimens: (i) full-scale
size in order to consider geometrical conditions alike those of real masonry buildings, and (ii) ad hoc
specimen shapes not limited to that of single (prismatic) masonry panels subjected to axial loads
parallel to mortar bed joints. The second feature motivated a significant number of full-scale
experimental tests on spandrel-pier specimens that pointed out the role of nonlinear interaction
between spandrels and piers. Such an interaction is originated from masonry interlocking at
spandrel-pier intersections, lintels or masonry arches above openings, ties along spandrels, and
reinforced concrete (RC) bond beams running on top or bottom of spandrels.
The objectives of these researches can then be summarised as follows:
i) To develop macro-element capacity models for spandrel panels to be implemented in three-
dimensional capacity models of masonry buildings,
ii) To assess the failure modes and related crack patterns of spandrels under different loading and
construction conditions, the latter delineated by the masonry assemblage scheme (i.e. masonry
bond), presence of lintels or arches above openings, and presence of ties and RC bond beams
on top or bottom of spandrels,
iii) To investigate how the interaction between spandrel and piers affects the in-plane nonlinear
response of perforated masonry walls under increasing lateral drift demand,
2 Experimental Researches on the Seismic Behaviour of Masonry Spandrels
iv) To demonstrate the implications of materials, geometry and construction features of spandrels
on the in-plane seismic capacity of masonry walls with openings, the latter defined through
simplified bilinear or multilinear force–drift diagrams,
v) To study the influence and effectiveness of innovative strengthening systems made of inorganic
matrix-grid composites in increasing the in-plane strength and inelastic deformation capacity of
spandrels.
In order to achieve these objectives, the four experimental programmes presented in this report
followed two complementary lines of research. Indeed, two experimental programmes were focused
on the derivation of force–displacement (or force–drift) diagrams of URM and composite URM-RC
spandrel elements. The other two experimental programmes investigated the effects of spandrel
behaviour and spandrel-pier interaction at the larger scale of masonry walls with openings. That
modus operandi allowed the contributors on one hand to characterise the in-plane capacity of single
spandrel elements for macro-element capacity models of perforated walls, and on the other hand to
assess the contribution of spandrels to nonlinear response of entire perforated walls at different
performance levels corresponding to increasing levels of in-plane damage.
1.3 Organisation of the Report
This report is structured as a series of chapters in which the research carried out at each institution
is described and commented. Every chapter includes the main information related to materials,
geometry and construction features of specimens, instrumentation, testing procedures, and
experimental results. The latter mainly consist of drawings and photographs that show the observed
damage, force–displacement (or force–drift) diagrams, and quantification of seismic capacity
measures. All chapters end with a section that summarises the experimental programme and outputs
by each research group with some considerations of possible future developments in the field.
In detail, Chapter 2 presents detailed information about an experimental campaign on double-leaf
stone masonry spandrels that was carried out at University of Pavia and EUCENTRE, Italy. Double-
leaf stone masonry is one of the most popular masonry types in the European historical built heritage.
Starting from material testing, the geometry of spandrel specimens, test setup and loading history
are described. That discussion is followed by four sections in which test results for as many
specimens are discussed. Those results are compared each other in the last section both in terms
of force/displacement measures and energy dissipation capacity. The implications of steel ties on
nonlinear behaviour of spandrel panels are characterised.
Chapter 3 presents the experimental programme on tuff stone masonry spandrel-pier
subassemblages that was performed at University of Naples Federico II, Italy. Tuff stone masonry
has been used since ancient times to build up monumental buildings and infrastructure systems,
such as bridges and water distribution systems. Nowadays, this type of masonry assemblage is still
used to construct new buildings in Mediterranean countries and other earthquake-prone countries
inside and outside Europe (e.g. Japan). In Chapter 3, the discussion of specimens and experimental
results emphasise the role of the spandrel type in masonry walls with openings, considering how
their in-plane seismic response changes in as-built, pre-damaged and repaired/strengthened
conditions. In detail, the influence of masonry spandrels externally strengthened with inorganic
matrix-grid composite systems is investigated. Furthermore, testing results are differentiated in terms
of (i) the masonry supporting element above the opening, i.e. wooden lintel or masonry arch, and (ii)
the presence or absence of RC bond beam running on top of the spandrel.
Chapter 4 presents the experimental campaign on spandrels made of solid and perforated clay brick
masonry assemblages that was carried out at the Swiss Federal Institute of Technology Zurich,
Switzerland. That experimental programme further investigated the influence of steel ties, RC bond
beams running on bottom of spandrels, and masonry arches or timber lintels above openings. Solid
and perforated clay brick masonries significantly contributed to fill the international research gap in
terms of masonry types worldwide.
Chapter 5 presents a comprehensive discussion of the experimental programme on clay brick
masonry subassemblages that was carried out at University of Auckland, New Zealand. Valuable
Research Report
data on the influence of spandrel geometry and relative strength/deformation capacity on the in-
plane lateral behaviour of coupled piers are reported, emphasising the spandrel-pier interaction and
mutual boundary conditions. Effects of low and high levels of axial loads on piers, as well as spandrel
depth and aspect ratio, are investigated.
This report ends with Chapter 6 where the main experimental findings are summarised and
challenges for future research on spandrels are identified.
Research Report
2 EXPERIMENTAL CAMPAIGN ON DOUBLE-LEAF STONE MASONRY SPECIMENS AT

THE UNIVERSITY OF PAVIA AND EUCENTRE PAVIA
Francesco Graziottia, Guido Magenesa,aAndrea Pennaa
2.1 SUMMARY
Seismic response of unreinforced masonry (URM) structures is influenced by the behaviour of

spandrel elements, as also demonstrated by numerical simulations and post-earthquake
observation. Until recently, the understanding of the seismic behaviour of these structural
components was not supported by adequate experimental tests. Only in the last few years some
experimental programs have been performed with the purpose of studying the in-plane cyclic
behaviour of masonry spandrel beams, made of clay bricks or perforated blocks. The experimental
study presented in this work aims at investigating the seismic behaviour of stone masonry spandrels,
both in the presence or in the absence of a horizontal tensile resistant element. For this reason, an
experimental apparatus was designed in order to test full-scale masonry spandrel specimens. The
test setup allows one to investigate the behaviour of spandrels reproducing realistic boundary
conditions in the element; in particular, attention was paid to the connection between spandrel and
piers.
Cyclic quasi-static tests were performed on four specimens in EUCENTRE Lab, University of Pavia,
Italy. The specimens were made of double-leaf stone masonry with 2 different geometries and
different horizontal force and timber lintel configurations. The deformation field of the masonry was
monitored also by high precision infrared camera able to position a series of passive markers glued
on the specimen.
a Department of Civil Engineering and Architecture, University of Pavia and EUCENTRE, via Ferrata 3, 27100 Pavia, Italy.
2.2 Description of the Experimental Campaign
Within the framework of the RELUIS Research Program and the EUCENTRE Executive Project
2005-2008, a large experimental campaign was planned. Its main focus was a shake table testing
of three full scale prototypes of stone masonry buildings. The prototypes were made of the same
type of masonry of the two specimens used for this work on masonry spandrels. The experimental
campaign includes a set of characterisation tests to define the mechanical properties of masonry:
several wall specimens were built for vertical compression, diagonal compression and in-plane
shear-compression tests [Magenes et al., 2010; Galasco et al., 2009].
2.3 Double-Leaf Stone Masonry Walls
The masonry used in this study consists of a double-leaf typology, i.e. stone walls are formed with
two vertical leaves of stones placed side by side along the wall length, as shown in Figure 2.1. They
can be connected using occasional large elements arranged transversely along the wall length
(through stones). The mechanical behaviour of the whole panel significantly depends on the quality
and number of such connections, as well as on the intrinsic properties of the materials used and the
construction quality.
Figure 2.1. Example of the double-leaf stone masonry [Giuffrè, 1993] and a picture of a masonry typology
used for this experimental campaign.
Double-leaf stone masonry is rather common in Italy. Particularly in areas with a great availability of
suitable stones to work with, many existing buildings were built with this technique. The need for the
identification of effective strengthening methods for this kind of buildings is evident. Indeed, they
have shown to be particularly vulnerable to seismic actions. The degree of connection between the
two layers is usually difficult to be determined, even based on in-situ testing techniques. The
traditional construction rules provide a number of through stones, but in a large number of buildings
these are missing, and connection between the two layers is guaranteed only by the
interaction/interlocking of adjacent leaves. Partial or ineffective bond between the layers can lead to
particularly dangerous failure modes. In particular, buckling of the leaves may occur with loss of wall
integrity. Another element of uncertainty in the assessment of this kind of wall is related to the
regularity of the material used and of the layout. In the presence of roughly dressed stones or of
extremely variable dimension of units, a large amount of mortar is needed: this, in general, adversely
affects the strength properties of masonry. If the mortar is not adequately confined by the units,
localized compression cracks can develop in the vertical and horizontal joints and eventually isolated
pillars, with possible local instability problems, can form. Investigating old buildings, the tendency to
use stones with an outer well worked side of a few centimetres but a very irregular texture has often
been noted. The fundamental characteristic of a double-leaf stone masonry wall is the strong
heterogeneity, which makes it very difficult not only to predict mechanical properties of masonry
based on the characteristics of its components, but also to identify the unavoidable imperfections
related to manufacture. In order to predict the structural behaviour of such masonry typology,
experimental information is essential, derived from tests designed to characterise its mechanical
behaviour.
Research Report
2.3.1 Stones
The choice of materials and construction techniques for this research has been designed to recreate
conditions that are representative of what can be found in existing buildings constructed with the
technique of double-leaf stone walls. The material is called Credaro stone and, from a geological
point of view, it is a sedimentary rock made of calcareous sandstone. In particular, the stones come
from a quarry near Bergamo. This stone was widely used in the past as a building material, as
demonstrated by the presence of several buildings constructed with this stone in the historical
downtown of Bergamo. The Credaro stone is characterised by good mechanical properties, with a
density of 2579 kg/m3, a compressive strength of 165-172 MPa and a flexural strength of 19 MPa. It
also has a significant resistance to freezing cycles.
2.3.2 Mortar
To satisfy the requirement of having a mortar as consistent as possible with that of historical
buildings, it has been necessary to derive an appositely produced type of mortar. This is because
the desired mortar should have a compression strength not exceeding 2 MPa, while current mortars
typically show mechanical properties which are higher than this value.
The mortar used was prepared starting from a product realized by the company Tassullo. This is a
pre-mixed natural hydraulic lime mortar, obtained by modifying a commercial product used for
strengthening interventions on historic buildings.
According to the procedure established by the EN1015-11 (1999), the specimens are prisms of
160×40×40 mm3. The mortar specimens start the curing process inside the moulds which are placed
in sealed polyethylene bags. Mortar specimens matured for 2 days in cellophane bags and 5 days
outside; after those 7 days the specimens were extracted and left in air.
The results obtained are shown in Figure 2.2. The significant scatter in the data is due to various
factors, including the inherent variability of material properties, different curing conditions (humidity,
temperature), different conditions for mortar packaging, etc. This scatter (compressive strength
varying from slightly above 1.2 MPa up to about 2.2 MPa for first two specimens and from 3 MPa to
6.5 MPa for the other two specimens) demonstrates that even if the mortar has been prepared on
site with a pre-mixed material supposed to have nominally constant characteristics, it still presents
rather variable properties within the volume typical of a real building.
1.6 7
1.4 6
1.2 5
1
fc,m [MPa]
ft,m [MPa]
4
0.8
3
0.6
2
0.4
0.2 1
0 0
0 1 2 3 4 0 1 2 3 4
Specimen No. Specimen No.
Figure 2.2. Values of flexural tensile (left) and compressive (right) strength for mortar samples tested at the
same time of the spandrel specimens.
2.1. Compression Tests on Stone Masonry Specimens
The experimental campaign included a set of characterisation tests to define the mechanical
properties of masonry: wall specimens were built for vertical compression and diagonal compression.
The adopted mortar had the same characteristics of the mortar used for this work, as well as the
rock blocks [Magenes et al., 2010].
2.3.3 Vertical Compression Test

Table 2.1 summarises the results of the whole vertical compression test campaign according EN
1052-1 (1998).
Table 2.1. Results of the vertical compression tests [Magenes et al., 2010].
Test fm [MPa] ε (σv,max) σv,el [MPa] ε(σv,el) E [MPa] εvert(σv,el) εorizz(σv,el) vhor εtrasv(σv,el) vtrasv
V1 3.14 0.0035 1.05 3.5 E-4 3000 3.5 E-4 7.0 E-5 0.20 4.7 E-5 0.13
V2 3.09 0.004 1.03 4.0 E-4 2400 4.0 E-4 8.9 E-5 0.22 5.8 E-5 0.14
V3 3.76 0.008 1.25 5.0 E-4 2600 5.0 E-4 1.5 E-4 0.31 - -
V4 3.18 0.006 1.06 5.0 E-4 2200 5.0 E-4 9.4 E-5 0.19 - -
V5 3.13 0.004 1.04 3.5 E-4 2900 3.5 E-4 2.7 E-5 0.09 5.6 E-5 0.16
V6 3.36 0.005 1.12 5.0 E-4 2200 5.0 E-4 5.7 E-5 0.11 9.7 E-5 0.19
Mean 3.28 0.005 1.09 4.3 E-4 2550 4.3 E-4 8.1 E-5 0.19 6.4 E-5 0.15
St. Dev. 0.26 0.0017 0.08 7.5 E-5 345 7.5 E-5 4.1 E-5 0.08 2.2 E-5 0.03
C. o. V. 8% 34% 8% 17.4% 13.5% 17% 51% 42% 34% 18%
2.3.4 Diagonal Compression Test

Table 2.2 summarises the results of the whole diagonal compression test campaign according STM
E519-02 (2002) and Brignola et al. (2008).
Table 2.2. Results of the diagonal compression tests [Magenes et al., 2010].
𝐴𝑆𝑇𝑀
𝑃 𝑃 𝜏𝑚𝑎𝑥 𝜏𝑚𝑎𝑥
𝜏𝑚𝑎𝑥 = [𝑀𝑃𝑎] 𝑓𝑡 = [𝑀𝑃𝑎] [𝑀𝑃𝑎] 𝛾 [−]
2𝐴𝑛 2𝐴𝑛 3 3
D1 0.135 0.090 0.044 5.46 E-5
D2 0.183 0.128 0.061 8.78 E-5
D3 0.243 0.169 0.081 1.17 E-4
D4 0.197 0.137 0.065 6.72 E-5
D5 0.174 0.121 0.058 5.66 E-5
D6 0.249 0.174 0.083 9.49 E-5
Mean 0.197 0.137 0.065 7.96 E-5
St. Dev. 0.043 0.031 0.015 2.45 E-5
C. o. V. 21.8% 21.8% 21.8% 30.7%
Table 2.3. Summary of the results of experimental wall compression tests [Magenes et al., 2010].
fm [MPa] E [MPa] ft [MPa] G [MPa]
Mean 3.28 2550 0.137 840

St. Dev. 0.26 345 0.031 125
C. o. V. 8% 13.5% 21.8% 14.8%
The experimental tests provided average values of some basic mechanical parameters required for
modelling structural elements (e.g. the specimens of this investigation) and entire masonry buildings.
A summary of the results obtained for the main mechanical parameters is reported in Table 2.3.
2.4 Description of the Spandrel Specimens
The campaign consists of the test of four full-scale double-leaf stone spandrels. The test units follow
the specification of the test facility. They are made up of two piers and a spandrel supported by a
timber lintel.
Research Report
The dimensions of the first two specimens (S1 and S2) were identical and are reported in Figure 2.3.
In particular piers were 2400 mm-tall and 1400 mm-long; the spandrel was 1080 mm-tall, 1200 mm-
long and supported by a timber lintel 120 mm-tall and 1600 mm-long (thick as the masonry). The
thickness of the entire specimen as well as the lintel was 320 mm. The difference between the two
tests is the absence of a tensile resistant element coupled to the spandrel of the first specimen (S1).
1400 1200 1400
600
1080
2400
1200
1600
120
600
0
32
1400
Figure 2.3. Geometry of specimens S1 and S2 double-leaf stone masonry specimens (dimensions in mm).
The dimensions of the two slender specimens (S3 and S4) were identical and are reported in Figure
2.4. In detail, piers were 2400 mm-tall and 1400 mm-long; the spandrel was 780 mm-tall and 1200
mm-long. Specimen 3 was supported by a timber lintel 120 mm-tall and 1600 mm-long (thick as the
masonry) while the spandrel of specimen 4 was a simply supported thin timber lintel (with a thickness
of 25 mm). The thickness of the entire specimen as well as the lintel was 320 mm.
1400 1200 1400 1400 1200 1400
0 0
32 32
780
2400
780
2400
25
120
Figure 2.4. Geometry of specimens S3 (left) and S4 (right) double-leaf stone masonry specimens
(dimensions in mm).
Figure 2.5 shows pictures of the first specimen built on the concrete foundation ready to be
transported in the laboratory (for further detail about the transportation system see Section 2.5).
Figure 2.5. Picture of specimen S1 during and after construction.

2.5 Properties of the Test Setup
The EUCENTRE TREES Lab is equipped with two reaction walls both next to strong floor. The walls
and base of the reaction structures are constructed with prefabricated post-compressed reinforced
cement concrete blocks. The laboratory test setup, designed specifically for the scope, allows quasi-
static tests on full-scale masonry spandrels; it utilizes a strong wall and the strong floor of the
laboratory in Pavia.
In order to simulate the real behaviour of the spandrel it was necessary to realize a test configuration
that recreates the real boundary condition in the element. Particular attention was focused on the
interlocking between spandrel and piers due to the wall texture. A statically determined spandrel
element specimen was a fundamental binding in order to read the results in an easy and univocal
way (axial force, shear force and bending moment). Particular consideration was given on the
possibility to control each of the actions on the specimen (F, Ff, Veff), allowing the bending moment
and both axial and shear forces to be independent of each other. The static scheme of a masonry
specimen is shown in Figure 2.6.
2F 2F
Ff Ff
2F 2F 
Veff
Figure 2.6. Static scheme of the specimen during a test.
The shear force in the spandrel can be easily computed as the uplifting force in the actuators (minus
the weight of the pier and of the test equipment).
Vs  Veff (2.1)
ls
M s  Vs  (2.2)
2
where Vs is the shear in the spandrel, Veff the uplifting force in the actuators minus the weight of the
pier and of the test equipment, ls the spandrel length and Ms the moment in the spandrel.
The test setup developed for those tests is shown in Figure 2.7. The horizontal displacement of the
right pier was not restrained. The compression of the spandrel could be controlled by the tension of
an external rod. The vertical compression of the two piers was controlled by 4 rods each transferring
the forces of the actuators to the mobile foundation beam.
The actuators were set in displacement control. The shortening (or elongation) of the two actuators
must be the same in order to impose a movement of the beam parallel to the ground; the bases of
the two piers had to be parallel. Figure 2.8 shows that the deformation field corresponding to an uplift
of the right pier, i.e. the configuration adopted in the tests discussed in this study, is the same that
would be obtained by equal rotation of the two bases (with no restrained in horizontal direction).
Research Report
Actuators Actuators
F F
F F F F F F
Ff
Ff Ff Ff
Figure 2.7. Scheme of the test setup.
F F
F F F F F
F
lf
Ff
Ff Ff
Ff s el
lp
  
l1 l1
Figure 2.8. Comparison of the deformation fields corresponding to an uplift of the right pier (left) and a
rotation of the bases of the piers (right).
In particular, the rotation at the base of the piers, α, is given by:

  arctan( / l 1 )  arctan[ /( l s  l p )]   / l 1 (2.3)
with all the quantities in Equation (2.3) defined in Figure 2.8.
The compression forces on the piers should be constant and the rotation on the top of the two piers
should not be constrained. These forces have the double role of contributing to create the
interlocking of the spandrel/pier interface's bricks and increasing the strength of the piers to be sure
that the spandrel was the weakest part of the specimen.
The actuators were doubly pinned. This created a pendular system that allowed the piers to move
apart each other without restraining the compressed strut in the spandrel. The compression of the
spandrel could be controlled just by the tension of an external rod (in series with a hydraulic jack)
creating a force Ff, as shown in Figure 2.8.
The final geometry of the test apparatus is shown in Figure 2.9.
The reinforced concrete foundation beam built to raise the mobile pier was 400 mm thick, 1200 mm
large and 3 m long and was post-compressed by 3 internal rods. The other foundation, i.e. the one
anchored to the lab strong floor, is 2000 mm long and 500 mm thick. The high difference of the two
base beams allows cyclic testing; the mobile pier could be pulled up or pushed down compared to
the level of the other pier up to 100 mm (corresponding to an element drift approximately equal to
8%).
Each pier was connected to the foundation by 4 rods, which were tensioned by a single centred jack
in order to avoid restraining the top rotation of the piers.
More information about the test setup are reported in Graziotti et al. (2009).
Figure 2.9. Front view of the test apparatus.
2.6 Instrumentation of the Specimens
Different global and local deformation parameters as well as actuator forces were measured during
testing. Deformations were measured by means of two independent systems: traditional wired
displacement transducers and an optical system consisting of a grid of reflector markers monitored
by high-definition cameras.
Research Report
2.6.1 Actuators Force Transducers

Five load cells were used to measure the force given by the actuators and hydraulic jacks. The two
transducers measuring the force applied by the 500 kN vertical actuators were already installed
inside the body of the actuators. The other three actuators consisted of a digital manometer placed
in series with the jacks. Two of them measured the pressure in the vertical jack (compression in the
piers) and the other one in the jack that gives tension to the spandrel's horizontal rods (only in the
second test). Figure 2.10 depicts the jack and the load cell in series to three springs horizontally
compressing the spandrel and it shows the position of all the force measuring instruments.
Actuators
Load cells
F F
Ff Ff
Jacks
Figure 2.10. Position of the load cells measuring the forces acting on the specimen and on the structure (left)
and hydraulic jack and load cell of the spandrel’s horizontal rods (right).
2.6.2 Displacement Transducers (Potentiometers)

Potentiometers or displacement transducers were placed in convenient positions on the specimen,
in order to measure various differential displacements. 50 linear displacement transducers with ball
tip and with cylindrical case were used to measure relative displacements between the two ends of
the instruments, allowing the average deformation along the instrument length to be derived. A
scheme of the position of these instruments on the specimen is shown in Figure 2.11.
The acquisition system setup enables recording of multi-type synchronised signals. Instruments are
attached to the connection panels where the signals pass by the National Instruments (NI) signal
conditioner boards. By analogue to digital conversion, signals are finally conveyed to the NI SCXI
Multi-Plexer Chassis where they are stored and recorded.
FRONT REAR
34 35 37 38
33 36
47 48
OUT OF PLANE
4 6 22 20
5 32 21
28
15 30
12
7 8 9 10 26 25 24 23
13 14
11 16
31 29 27
1 2 3 18
19 17
49 50
44 45 46
39
40 41 42 43
Figure 2.11. Layout of the displacement transducers on the specimen (front/rear).
2.6.3 Advanced Video Acquisition

Information about the displacement response of specimens during the tests was provided by video
acquisition with high-definition cameras. The motion is detected through reflector markers, allowing
the evaluation of displacements in the vertical plane framed by the cameras.
Reflector markers were placed on the specimen in some specific locations to be monitored. In order
to have real values of displacements, cameras need to be calibrated using a pin-hole camera model.
The identification of each blob, i.e. the pixel representation of a marker in the scene, is performed
on groups of continuous pixels having the same colour. The blob centre is detected using the colour
intensity, with a theoretical precision of up to 0.02 sub-pixel using colour transitions, corresponding
to a precision in the order of 0.01 mm. To reduce issues concerning the interaction with the
environmental illumination, infrared lamps are adopted and visible light components are cut off with
appropriate filters on the camera lenses.
2.7 Test Procedure
As discussed in previous sections, four different specimens were tested in this campaign. The cyclic
tests on double-leaf stone masonry wall were carried out with displacement control (no force-
controlled load steps were conducted).
The first two specimens were identical, with the only difference consisting in the horizontal
compressive force in the spandrel: zero for specimen S1 and 75 kN, 38kN, 19kN for specimen S2.
The specimens S3 and S4 have identical geometry but different lintel configurations. The horizontal
compressive force was 28 kN. The vertical force in the piers was 75 kN, which is equivalent to a
compression of 0.17 MPa.
Following the specifications reported in Section 2.5, the base of the right pier was uplifted generating
a drift in the spandrel element. This element drift is nominally defined as:
drift el  el   s / l s (2.4)
In the preliminary phase, during the definition of the loading history, it was considered for simplicity
that the uplift of the base is equal to the difference in vertical displacement of the spandrel element
end sections:
  s (2.5)
The nominal element drift is then defined as:
n   / l s (2.6)
In each step the loading was stopped at different drift levels and subjected to three loading cycles.
Each cycle took approximately 120 seconds; the maximum velocity was 0.5 mm/s. At the end of
each load step, the loading was stopped, the cracks were marked and photos were taken.
2.7.1 Loading History of the First Spandrel Specimen (S1)

In this test there was no horizontal force applied to the spandrel, as there was no tie rod installed.
A problem in the mobile pier's support forced to begin the test with a monotonic ramp up to a
displacement of 6 mm (θn = 0.5%). After that, complete loading cycles were performed reaching
the specimen rupture at 18 mm (θn = 1.5%). The loading history followed during the test on
specimen S1 is summarised in Table 2.4.
.
2.7.2 Loading history of the Second Spandrel Specimen (S2)

In this test the horizontal force in the spandrel was constant in each cycle. In particular, three
decreasing values were used: 75 kN up to a drift of 9.6 mm (θn = 0.8%), 38 kN up to 36 mm (θn =
3%) and 19 kN repeated just for this last 3% drift cycle. The pre-tension of the tie rods was decreased
during the test because with the initial condition the resistance of the spandrel was too high do not
satisfying the strong piers-soft spandrel hypothesis.
The loading history followed during the test on specimen S2 is summarised in Table 2.5.
Research Report
Table 2.4. Load sequence followed for the test on spandrel specimen S1.
Nominal Cycle Total

Amplitude δ Velocity Ramp
N cycles element duration duration
[mm] [mm/s] duration [s]
drift θn [s] [s]
Monotonic 1 D 0.05% 0.6 0.025 24 96 288
Monotonic 2 D 0.08% 0.9 0.0375 24 96 576
Monotonic 3 D 0.10% 1.2 0.05 24 96 864
Monotonic 4 D 0.15% 1.8 0.0625 28.8 115.2 1209.6
Monotonic 5 D 0.20% 2.4 0.08 30 120 1569.6
Monotonic 6 D 0.25% 3 0.1 30 120 1929.6
Monotonic 7 D 0.30% 3.6 0.12 30 120 2289.6
Monotonic 8 D 0.40% 4.8 0.16 30 120 2649.6
Monotonic 9 D 0.50% 6 0.2 30 120 3009.6
zero
Cycles 3 D ±0.1% 1.2 0.05 24 96 288
Cycles 5 D ±0.2% 2.4 0.08 30 120 648
Cycles 7 D ±0.3% 3.6 0.12 30 120 1008
Cycles 8 D ±0.4% 4.8 0.16 30 120 1368
Cycles 9 D ±0.5% 6 0.2 30 120 1728
Cycles 10 D ±0.6% 7.2 0.24 30 120 3369.6
Cycles 12 D ±0.8% 9.6 0.32 30 120 4089.6
Cycles 13 D ±1% 12 0.4 30 120 4449.6
Cycles 15 D ±1.5% 18 0.5 36 144 5241.6
2.7.3 Loading History of the Third Spandrel Specimen (S3)

In this test the horizontal force in the spandrel was constant in each cycle and equal to 28 kN
(corresponding to a horizontal compressive stress in the spandrel σh = 0.11 MPa). Complete loading
cycles were performed and the specimen rupture was reached at a displacement of 60 mm (θn =
5%). The loading history followed during the test on specimen S3 is summarised in Table 2.6.
Nominal Cycle Total

Amplitude δ Velocity Ramp
N cycles element duration duration
[mm] [mm/s] duration [s]
drift θn [s] [s]
Cycles 3 D 75 ±0.1% 1.2 0.05 24 96 864
Cycles 4 D 75 ±0.15% 1.8 0.0625 28.8 115.2 1209.6
Cycles 5 D 75 ±0.2% 2.4 0.08 30 120 1569.6
Cycles 6 D 75 ±0.25% 3 0.1 30 120 1929.6
Cycles 7 D 75 ±0.3% 3.6 0.12 30 120 2289.6
Cycles 8 D 75 ±0.4% 4.8 0.16 30 120 2649.6
Cycles 9 D 75 ±0.5% 6 0.2 30 120 3009.6
Cycles 10 D 75 ±0.6% 7.2 0.24 30 120 3369.6
Cycles 11 D 75 ±0.7% 8.4 0.28 30 120 3729.6
Cycles 12 D 75 ±0.8% 9.6 0.32 30 120 4089.6
zero
Cycles 5 D 38 ±0.2% 2.4 0.08 30 120 360
Cycles 8 D 38 ±0.4% 4.8 0.16 30 120 360
Cycles 10 D 38 ±0.6% 7.2 0.24 30 120 720
Cycles 12 D 38 ±0.8% 9.6 0.32 30 120 1080
Cycles 13 D 38 ±1% 12 0.4 30 120 4449.6
Cycles 15 D 38 ±1.5% 18 0.5 36 144 5241.6
Cycles 17 D 38 ±2% 24 0.5 48 192 6321.6
Cycles 19 D 38 ±3% 36 0.5 72 288 7905.6
zero
Cycles 19 D 19 ±3% 36 0.5 72 288 864
2.7.4 Loading History of the Fourth Spandrel Specimen (S4)

In this test the horizontal force in the spandrel was constant in each cycle and equal to 28 kN
(corresponding to a horizontal compressive stress in the spandrel σh = 0.11 MPa). Complete loading
cycles were performed and the specimen rupture was reached at a displacement of 60 mm (θn =
5%). A problem in the actuators controller forced to begin the test with a fast monotonic ramp up to
a displacement of 7.5 mm (θn ≈ 0.65%). After that, complete loading cycles were performed and the
specimen rupture was reached at a displacement of 60 mm. The loading history followed during the
test on specimen S4 is summarised in Table 2.7.
Nominal Total
Amplitude Velocity Ramp Cycle duration
N cycles element duration
δ [mm] [mm/s] duration duration [s] [min]
drift θ n [s]
Cycles 2 A 0,10% 1,2 0,04 30 120 360 6,0
Cycles 3 A 0,15% 1,8 0,06 30 120 720 12,0
Cycles 4 A 0,20% 2,4 0,08 30 120 1080 18,0
Cycles 5 A 0,25% 3 0,1 30 120 1440 24,0
Cycles 6 A 0,30% 3,6 0,12 30 120 1800 30,0
Cycles 7 A 0,40% 4,8 0,16 30 120 2160 36,0
Cycles 8 A 0,50% 6 0,2 30 120 2520 42,0
Cycles 9 A 0,60% 7,2 0,24 30 120 2880 48,0
Cycles 10 A 0,70% 8,4 0,28 30 120 3240 54,0
Cycles 11 A 0,80% 9,6 0,32 30 120 3600 10s
Cycles 12 A 0,90% 10,8 0,36 30 120 3960 66,0
Cycles 13 A 1,00% 12 0,4 30 120 4320 72,0
Cycles 14 A 1,25% 15 0,5 30 120 4680 78,0
Cycles 15 A 1,50% 18 0,6 30 120 5040 84,0
Cycles 16 A 2,00% 24 0,8 30 120 5400 90,0
Cycles 17 A 3,00% 36 1,2 30 120 5760 96,0
Cycles 18 A 4,00% 48 1,6 30 120 6120 102,0
Cycles 19 A 5,00% 60 2 30 120 6480 108,0
Nominal Total
Amplitude Velocity Ramp Cycle
N cycles element duration
δ [mm] [mm/s] duration duration [s]
drift θ n [s]
Ramp 1 0,65% 7,8 0,78 10 10 20
Cycles 2 B 0,10% 1,2 0,04 30 120 360
Cycles 3 B 0,15% 1,8 0,06 30 120 720
Cycles 4 B 0,20% 2,4 0,08 30 120 1080
Cycles 5 B 0,25% 3 0,1 30 120 1440
Cycles 6 B 0,30% 3,6 0,12 30 120 1800
Cycles 7 B 0,40% 4,8 0,16 30 120 2160
Cycles 8 B 0,50% 6 0,2 30 120 2520
Cycles 9 B 0,60% 7,2 0,24 30 120 2880
Cycles 10 B 0,70% 8,4 0,28 30 120 3240
Cycles 11 B 0,80% 9,6 0,32 30 120 3600
Cycles 12 B 0,90% 10,8 0,36 30 120 3960
Cycles 13 B 1,00% 12 0,4 30 120 4320
Cycles 14 B 1,25% 15 0,5 30 120 4680
Cycles 15 B 1,50% 18 0,6 30 120 5040
Cycles 16 B 2,00% 24 0,8 30 120 5400
Cycles 17 B 3,00% 36 1,2 30 120 5760
Cycles 18 B 4,00% 48 1,6 30 120 6120
Cycles 19 B 5,00% 60 2 30 120 6480
2.8 Test Results for the First Specimen
This section summarises the main results of the test on specimen S1, which was characterised by
the absence of horizontal force on the spandrel. Particular attention was focused on the force–
displacement behaviour, deformation modes, cracks opening and dissipated energy. The test was
conducted on February 17, 2010.
Research Report
2.8.1 Test Observations

A problem in the piers support required to begin the test with a ramp of increasing displacement up
to an uplift of 6 mm (θn = 0.5%). The loading history and the instants of significant cracking are
represented in Figure 2.12.
24 2
Ff=0
18 1.5
12 1
M8
Nominal element drift θn [%]

C8
6 M7 0.5
Displacement δ [mm]
C7
M5
M4
0 0
C3
-6 -0.5
-12 -1
-18 -1.5
-24 -2
Time
Figure 2.12. Time history of monotonic and cyclic loading with significant cracking points for specimen S1.
After each test was performed under increasing levels of drift, cracks were detected and marked on
the specimen. A representation of the evolution of the crack pattern observed on specimen S1 is
reported in Figure 2.16.
The first crack occurred at step M4 (θn = 0.15%). This vertical flexural crack started in the upper right
corner of the spandrel and ended in the centre for a total length of about 400 mm. The test continued
with a monotonic ramp and a similar and opposite crack appeared at step M5 (θn = 0.2%), starting
from the lower left corner near the timber lintel. This monotonic ramp ended at a displacement of 6
mm equal to a nominal drift θn = 0.5%. At that stage the flexural cracks on the interfaces between
spandrel and piers became longer (approximately 700 mm). This indicates that the compressed tie
in the spandrel (from top left to bottom right) had a very small transversal section. At this stage the
lintel appeared to be well bonded to the piers and no relative displacement between the two
components was noticed. After that stage, the cyclic test begun.
The first cracks of the cyclic test were induced by flexure due to the downward movement of the right
pier and occurred at step C3 (θn = –0.1%). A comparison with the monotonic uplift ramp shows that
they were concentrated on the opposite corners. The lintel appeared to be bonded to the piers, as
evidenced from the picture in Figure 2.13.
The first sliding of the lintel on the piers took place during step C7 (θn = 0.3%). The left part of the
lintel was debonded from the pier due to the moment of the spandrel element; during the uplift of the
right pier that part of the lintel works as a tensile strut. At such a drift level, Coulomb friction between
masonry and lintel reached its limit, providing the maximum resisting moment of the element (see
Section 2.8.2).
Continuing the test, the spandrel element showed a rocking behaviour and no further cracks
appeared. During step C8 (θn = 0.4%) the cracks on the two interfaces started to open and close
under cycle reversal, allowing rigid rotations of the spandrel (Figure 2.13).
Figure 2.13. Crack pattern of spandrel at C3 (θn = –0.1%) (left) and C8 (θn = 0.4%) (right).
Figure 2.14. Crack pattern of spandrel and view of the lintel sliding at C10 (θ n = 0.6%).
During step C10 (θn = 0.6%) those cracks were opened more than 10 mm and, at the end of the
cycles, the gaps were not filled as shown in Figure 2.14.
The distance between the piers increased every cycle (Figure 2.14 and Figure 2.15), due to the
absence of a horizontal restoring force. In particular, the final relative displacements of the lintel
(after step C15, θn = 1.5%) were 38 mm on the left and 18 mm on the right, as shown in Figure 2.15
(right).
It was not easy to detect a real qualitative collapse point. With this crack configuration, the spandrel
element would have been able to couple the masonry piers up to higher drift levels.
Figure 2.15. View of the entire specimen at C10 (θn = 0.6%) and C15 (θn = 1.5%).
Research Report
(M4) (C3)
M0.15% C0.1%
1.8mm ±1.2mm
(M5) (C7)
M0.2% C0.3%
2.4mm ±3.6mm
(M7) (C8)
M0.3% C0.4%
3.6mm ±4.8mm
(M8) (C15)
M0.4% C1.5%
4.8mm ±18mm
Figure 2.16. Evolution of the crack pattern observed during the tests on specimen S1.
2.8.2 Force–Displacement Behaviour of the First Specimen

Figure 2.17 shows a plot of the global shear force versus the vertical displacement of the base of
the right pier. The light grey line represents the monotonic loading, whilst the thick black line
represents the envelope of all the cycles. Each plot also shows values of the equivalent base rotation
α, defined in accordance with Figure 2.8 and Equation (3.3).
The force was calculated considering the reading of two load cells in the vertical actuators, which
was adjusted considering the correct percentage of the weight of the specimen and the test facility,
as indicated in Equation (3.1). The displacement of the pier was obtained by monitoring a central
marker (representing the displacement of the base).
The light grey line represents the monotonic loading. The thick black line represents the envelope of
cycles.
Figure 2.17. Force–displacement cycles of specimen S1, considering the displacement of the base of the
right pier δ and the equivalent base rotation α.
Figure 2.18 shows the same hysteretic behaviour which is plotted considering the relative vertical
displacement of the two interfaces of the spandrel. That plot also reports values of the drift in the
spandrel θel, defined according to Figure 2.8 and Equation (1.3).
The maximum shear force reached during the test was about 30 kN both for positive and negative
loading. It occurred at a drift θel of about 1.5‰ (δ ≈ 2 mm). The maximum positive shear was reached
under the unexpected but necessary monotonic loading (grey line in Figure 2.17 and Figure 2.18;
for more details see Section 2.8.1). This contingency seems not to significantly change the behaviour
in terms of maximum positive/negative shear resistance and stiffness compared with the cyclic
behaviour that was normally imposed and monitored in negative direction.
A considerable drop in stiffness and maximum positive shear force was measured when the
specimen was reloaded up to a positive displacement δ = 5 mm. The decrease of the maximum
positive shear force was about 30% (from 32 kN to 20 kN).
The envelope of the maximum shear force dropped down to about 15 kN starting from an element
drift θel = 0.5%. Looking at the crack pattern evolution it is possible to notice that, from this point, the
spandrel exhibited a rocking mechanism, with two main opened cracks on the whole interfaces
between spandrel and piers. This phenomenon happened right after the first movement of the timber
lintel relatively to the stone masonry of the pier. Then the envelope of the maximum shear force
remained constant up to the end of the test.
If the shear–displacement plots reported in Figure 2.17 and Figure 2.18 are compared to each other,
it is possible to observe that they are substantially equal. This means that the median horizontal
sections of the piers did not rotate appreciably, indicating that the piers were not collaborating to the
deformation of the whole specimen.
Research Report
Figure 2.18. Force–displacement cycles of specimen S1, considering the relative displacement of the two
interfaces of the spandrel δs and the spandrel drift θel.
2.9 Test Results for the Second Specimen
the presence of a horizontal force on the spandrel. In particular attention was focused on the force-
displacement behaviour, deformation modes, cracks opening and dissipated energy. The test was
conducted on March 2, 2010.

The loading history and the instants of significant cracking are represented in Figure 2.19. Three
different levels of horizontal compressive force Ff were used (75 kN, 38kN and 19 kN).
After each test was performed under increasing levels of drift, cracks were detected and marked on
the specimen. A representation of the evolution of the crack pattern observed on specimen S2 is
reported in Figure 2.23.
The test started with a horizontal compression of 75 kN, equivalent to a compressive stress σh =
0.23 MPa. The first cracks were observed at step C6_75 (θn = 0.25%). A vertical small crazing in the
centre of the panel (one stone layer high) and a flexural crack in the left pier starting from the left
end of the timber lintel were observed (Figure 2.20). Another crack appeared on the right pier at step
C11 (θn = 0.7%). That crack was similar to the one observed in the other pier. This indicates that the
piers cracked before the spandrel, not satisfying the requirement of weak spandrel and strong piers
considered during the design of the test.
48 4
Ff=75kN Ff =38kN Ff =19kN
42
36 3
30
C17_38
24 2
C15_38
18 C14_38

C13_38
12 C11_75 1
6 C6_75
0 0
-6
-12 C11_75 -1
-18 C13_38
C15_38
-24 -2
C17_38
-30
-36 -3
-42
-48 -4
Time
Figure 2.19. Cyclic loading time history with significant cracking points for specimen S2.
Figure 2.20. Crack pattern of spandrel at C6_75 (θn = 0.25%) and C14_38 (θn = 1.25%).
In order to decrease the resistance of the spandrel, the horizontal compression was reduced by half
(down to 38 KN, corresponding to σh = 0.11 MPa).
The first cracks observed on specimen S2 were of shear type and formed in the spandrel. In
particular, when the positive displacement level C13_38 (θn = 1%) was attained, a diagonal crack
spread from the top left corner to the centre of the panel. When the load was reversed, a shear crack
propagated along the whole diagonal from the top right corner to the bottom left one. The cracks
closed at the end of the cycles due to the compressive horizontal force.
The positive displacement at C14_38 (θn =1.25%) caused the formation of a diagonal crack from the
centre of the panel to the bottom right of the spandrel (Figure 2.20, right). During the cycles C15_38
(θn =1.5%) other diffused cracks appeared in the spandrel; the most relevant crack was similar and
parallel to the fracture which formed at C14_38.
The last cycles that caused further cracks were those labelled as C17_38 (θn = 2%). In that case,
the openings consist of diagonal cracks near the left corners.
Two C19 (θn =3%) cycles were conducted at two different horizontal force levels: 38 kN and 19 kN.
The cracks occurred during the cycles C19_38 are represented in Figure 2.21 (left and right
respectively).
Research Report
A significant representation of the final crack pattern is reported in Figure 2.22. At the end of the test,
the cracks were all closed and they indicated a typical shear failure.
Figure 2.21. Crack pattern of spandrel corresponding to positive (left) and negative (right) displacements
C19_38 (θn = ±3%).
Figure 2.22. Crack pattern of the spandrel at the end of the test.
(C6_75) (C14_38)
C0.25% C1.25%
±3mm ±15mm
(C11_75) (C15_38)
C0.7% C1.5%
±8.4mm ±18mm
(C13_38) (C17_38)
C1% C2%
±12mm ±24mm
Figure 2.23. Evolution of the crack pattern observed during the test on specimen S2.
2.9.2 Force–Displacement Behaviour of the Second Specimen

Figure 2.24, Figure 2.25 and Figure 2.26 (Ff = 75kN, 38kN and 19 kN respectively) show plots of
the global shear force versus the vertical displacement of the base of the right pier. Each plot also
shows values of the equivalent base rotation α, defined in accordance with Figure 2.8 and Equation
(1.3).
The force was calculated on the basis of the reading of two load cells in the vertical actuators, which
was adjusted by considering the correct percentage of the weight of the specimen and the test facility
marker (representing the displacement of the base). The thick black line represents the envelope of
cycles.
Research Report
Figure 2.24. Force–displacement cycles of the second specimen (Ff = 75 kN), considering the displacement
of the base of the right pier δ and the equivalent base rotation α.
Figure 2.25. Force–displacement cycles of specimen S2 (Ff = 38 kN), considering the displacement of the
base of the right pier δ and the equivalent base rotation α.
Figure 2.26. Force–displacement cycles of specimen S2 (Ff = 19 kN), considering the displacement of the
base of the right pier δ and the equivalent base rotation α.
Figure 2.27, Figure 2.28 and Figure 2.29 show the same hysteretic behaviour plotted considering
the relative vertical displacement of the two interfaces of the spandrel. Each plot also reports values
of the drift in the spandrel θel, defined according to Figure 2.8 and Equation (3.3). If Figure 2.24 is
compared to Figure 2.25, it is possible to see that a change in the horizontal force Ff (from 75 kN to
38 kN) did not considerably influence the stiffness of the specimen.
The maximum shear force reached during the test was about 60 kN for both positive and negative
loading. It occurred at a drift θel of about ±2.1% (nominal drift θn=±3%, spandrel relative displacement
δs ≈ ±25mm, base displacement δ = ±36mm). In this test valid the approximation δ ≈ δs, was not valid
anymore because the piers were participating to the total specimen deformation with a rotation, and
hence θel < θn. From the previous plots it is possible to see that, for a given θn (e.g. δ=36mm, θn=3%),
a decrease of the applied horizontal force Ff corresponds to a larger element drift θel suggests that
the piers deformed more if well coupled to the spandrel.
The presence of the horizontal rods created a vertical force component opposed to the displacement
of the right pier and the consequent inclination of the rod. The magnitude of this vertical component
was less than 1% of Ff (that is 0.7 kN) at the maximum displacement level.
The observed behaviour is close to an elasto-plastic one with a well defined "yielding" point. The
plastic behaviour characterises the element after a drift θel ≈ 0.25%, up to the maximum imposed
drift without significant hardening or softening.
Looking at the crack pattern evolution, it is possible to notice that the elastic behaviour ended when
the first small vertical crack appeared in the centre of the panel, i.e. at C6_75 (θn = 0.25%).
Research Report
Figure 2.27. Force–displacement cycles of specimen S2 (Ff = 75 kN), considering the relative displacement
of the two interfaces of the spandrel δs and the spandrel drift θel.
2.10 Test Results for the Third Specimen
the presence of a horizontal force on the spandrel (that is 28 kN) and a well-connected timber lintel.
Particular attention was focused on the force–displacement behaviour. The test was conducted on
May 14, 2013.

The loading history and the instants of significant cracking are represented in Figure 2.30.
After each test performed under increasing levels of drift, the cracks in the specimen were detected
and marked on the specimen itself. A representation of the evolution of the crack pattern observed
for specimen S3 is reported in Figure 2.35.
The test started with a horizontal compression of 28 kN, equivalent to a stress σh = 0.11 MPa. The
first cracks were observed at step C5 (θn = 0.25%) which caused a vertical small crazing in the centre
of the panel (one stone layer high). As the test continued, shear cracks were observed on the
spandrel. In particular, at the positive displacement C6 (θn = 0.3%) a diagonal crack spread from the
top left corner to the centre of the panel (Figure 2.31).
At a negative displacement of 6 mm (point C8, θn = 0.5%) a shear crack propagated along the whole
diagonal from the top right corner to the bottom left one. The cracks resulted closed at the end of the
cycles due to the compressive horizontal force. A complete horizontal crack related to the relative
displacement between the timber lintel and the bottom part of the masonry spandrel was noticed at
C9 (θn = 0.5%).
Starting from cycles C12 (θn = 0.9%) other diffused cracks appeared on the spandrel. From C13 (θn
= 1%) the cracks started not to reclosing at the end of the cycles and their width increased during
following cycles.
Research Report
66 5,5
60
54 4,5
48
42 3,5
36
30 2,5
24

C13
Displacement δ [mm] 18 C12 1,5
C9
12
C5 C6
6 0,5
0
-6 -0,5
-12 C8
-18 -1,5
-24
-30 -2,5
-36
-42 -3,5
-48
-54 -4,5
-60
-66 -5,5
0 20 40 60 80 100
Time
Figure 2.31. Crack pattern of spandrel at C6 (θn = 0.3%) and C9 (θn = 0.6%).
Figure 2.32. Rear crack pattern of spandrel at C11 (θn = 0.8%).
The last cycle where new cracks formed was that labelled as C16 (θn = 2%). After that cycle, the
spandrel was heavily damaged and almost all stones suffered cracks on their entire perimeter
(Figure 2.32 and Figure 2.33). A significant representation of the final crack pattern is reported in
Figure 2.34.
Two C19 (θn = 5%) cycles were conducted at two different horizontal force levels: 28 kN and 0 kN.
Figure 2.33. Crack pattern of spandrel during positive (left) and negative (right) displacements C17 (θn =
±3%).
Research Report
Figure 2.34. Crack pattern of the S3 spandrel at the end of the test.
Figure 2.35. Evolution of crack pattern evolution observed during the different tests on the third specimen.
2.10.2 Force–Displacement Behaviour of the Third Specimen

the right pier. The plot also shows values of the equivalent base rotation α, defined in accordance
with Figure 2.8 and Equation (2.3).
was adjusted considering the correct percentage of the weight of the specimen and the test facility
all cycles.
The maximum shear force reached during the test was about 38 kN for both positive and negative
loading. It occurred at a nominal drift θn = ±5%, corresponding to a base displacement δ = ±60 mm.
The presence of the horizontal rods created a vertical component opposed to the displacement of
the right piers and the consequent inclination of the rod. The magnitude of that vertical component
was less than 2% of Ff (that is 0.5 kN) under the maximum displacement level.
The observed behaviour is close to an elasto-plastic one with a well defined "yielding" point. The
plastic behaviour characterises the element after θn ≈ 0.25%, up to the final imposed drift without
significant hardening or softening (similar to specimen S2).
Looking at the crack pattern evolution, it is possible to notice that the elastic behaviour ended as
soon as the first small vertical crack appeared in the centre of the panel, i.e. at C5 (θn = 0.25%).
Equivalent base rotation α [%]

-2.5 -2.08 -1.67 -1.25 -0.83 -0.42 0.42 0.83 1.25 1.67 2.08 2.5
40
30
20
10
Shear force [kN]
Fh=0 kN
0
-10
-20
Fh=28 kN
-30
-40
-70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70
2.11 Test Results for the Fourth Specimen
the presence of a horizontal force on the spandrel equal to 28 kN and a simply supported slender
timber lintel. Particular attention was focused on the force–displacement behaviour. The test was
conducted on June 13th, 2013.

The loading history and the instants of significant cracking are represented in Figure 2.37.
Research Report
66 5,5
60
54 4,5
48
42 3,5
36
30 2,5
24

C13
18 1,5
Displacement δ [mm] R1 C11
12
C5
6 0,5
0
-6 -0,5
-12 C10
-18 -1,5
-24
-30 -2,5
-36
-42 -3,5
-48
-54 -4,5
-60
-66 -5,5
0 20 40 60 80 100
Time
After each test was performed under increasing levels of drift, the cracks in the specimen were
detected and marked on the specimen itself. A representation of the evolution of the crack pattern
observed on specimen S4 is reported in Figure 2.41.
The test started with a horizontal compression of 28 kN, equivalent to a stress σh = 0.11 MPa.
A problem in the actuators controller caused an unexpected and very rapid positive displacement of
about 8 mm in the central actuator (R1). This caused the opening of a diagonal crack in the spandrel
(#2 on Figure 2.38) and a flexural crack in the left pier (#3 on Figure 2.38). After this problem the
test continued following the original program. In the following cycles all cracks were observed in the
same diagonal of the first crack (from top left to bottom right). The first crack on the other diagonal
was observed after step C10 (θn = –0.7%, δ = –8.4 mm), that is when displacement was
approximately the same of R1 in the other direction. From that point other diffused cracks appeared
in the spandrel. The slenderness of the lintel caused a more diffused cracking pattern due to the
relative higher displacement of the stones.
Figure 2.38. Crack pattern of spandrel at C5 (θn = 0.25%) and C11 (θn = 0.5%).
Figure 2.39. Rear crack pattern of spandrel at C11 (θn = 0.8%).
The last cycle with the formation of new cracks was that labelled as C16 (θn = 2%). After that cycle,
the spandrel was heavily damaged and almost all stones suffered cracks on their entire perimeter.
A significant representation of the final crack pattern is reported in Figure 2.42. Two C19 (θn = 5%)
cycles were conducted at two different horizontal force levels: 28 kN and 0 kN.
Figure 2.40. Crack pattern of spandrel during positive displacement at C17 (θn = 3%) and during negative
displacement at C18 (θn = –4%).
Research Report
Figure 2.41. Evolution of the crack pattern observed during the test on specimen S4.
Figure 2.42. Crack pattern of the S4 spandrel at the end of the test.
2.11.2 Force–Displacement Behaviour of the Fourth Specimen

the right pier. The plot also shows values of the equivalent base rotation α, defined in accordance
with Figure 2.8 and Equation (1.3).
was adjusted considering the correct percentage of the weight of the specimen and the test facility
all cycles.
The maximum shear force reached during the test was about 23 kN for positive displacement and
19 kN for the negative one. It occurred at a nominal drift θn = ±0.67%, corresponding to a base
displacement δ = ±8 mm.
The presence of the horizontal rods created a vertical component opposed to the displacement of
the right piers and the consequent inclination of the rod. The magnitude of that vertical component
was less than 2% of Ff (that is 0.5 kN) under the maximum displacement level.
A considerable drop in stiffness and maximum positive shear force was measured during reloading
after a positive displacement δ = 8 mm. In particular, the decrease of the maximum positive shear
force was about 70% (from 23 kN to 7 kN).
Looking at the crack pattern evolution, it is possible to notice that the maximum force (positive or
negative) corresponds to the opening of the two diagonal cracks at R1 and C10 (δ = ± 8 mm).

-2.5 -2.08 -1.67 -1.25 -0.83 -0.42 0.42 0.83 1.25 1.67 2.08 2.5
25
20
15
10
Fh=0 kN
Shear force [kN]
-5
-10
-15 Fh=28 kN
-20
-25
-70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70
2.12 Comparison of the Results of the Four Specimens
2.12.1 Force/Displacement
Figure 2.44 represents the shear force versus relative displacement δs (and spandrel drift θel) for
specimens S1 and S2.
The main difference consisted in a change of failure mechanism from a flexure/rocking mode
with added friction of the lintel to a shear one. This influenced the hysteretic behaviour
characterised by high strength and "thin" loops in case of horizontal compression, and low
strength and "fat" loops in case of absence of tie rod. Comparing the two curves it is evident that
the tie rod increased strength and strain capacity as well. In particular, the presence of a
horizontal compression amplified the maximum shear strength from 30 kN to 50 kN (plus 67%).
It is difficult to compare the levels of deformation capacity because the behaviour of the two
specimens was completely different and was not so straightforward to define an in-plane
Research Report
collapse point to compare. The initial stiffness was not influenced by the presence of a horizontal
compression.
Figure 2.44. Shear versus relative displacement curves obtained from the tests on specimens S1 (dashed
line) and S2 (Ff = 38 kN) (solid line).
Figure 2.45 represents the shear force versus base displacement δ (and equivalent base rotation α)
for specimens S3 and S4.

-2.5 -2.08 -1.67 -1.25 -0.83 -0.42 0.42 0.83 1.25 1.67 2.08 2.5
40
30
20
10
Shear force [kN]
-10
-20
-30
-40
-70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70
Figure 2.45. Shear versus base displacement curves obtained from the tests on specimens S4 (dashed line)
and S3 (solid line) (Ff = 28 kN).
The difference between the two configurations was the type of timber lintel. In particular,
specimen S3 had the same well-connected lintel of specimens S1 and S2, while the lintel in
specimen S4 was a simply-supported, very slender one.
The timber lintel produced two main effects on the spandrel. The first one was the increase in
strength due to the elastic stiffness of the lintel; the second one was to prevent relative
dislocation of stones, hence limiting cracking in the spandrel. Comparing the two curves it is
evident that the presence of a well-connected lintel increased the strength capacity. In particular,
it amplified the maximum shear strength from 22 kN to 35 kN (plus 60%). The lintel also
prevented a softening behaviour.
Figure 2.46 shows the backbone curves of the shear force versus base displacement δ (and
equivalent base rotation α) for all the specimens.
-1.25 -1.04 -0.83 -0.63 -0.42 -0.21 0.21 0.42 0.63 0.83 1.04 1.25
60
S1
S2
40
S3
S4
20
Shear force [kN]
-20
-40
-60
-30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30
Figure 2.46. Shear versus base displacement envelopes obtained from the tests on specimens S1 (grey
dashed line), S2 (grey solid line), S3 (black solid line) and S4 (black dashed line).
The difference between the grey lines and black ones is due to the geometry of the spandrel
(more slender than the others in specimens S3 and S4).
2.1.1. Dissipated Energy
Figure 2.47 represents the dissipated energy versus base displacement δ for the four specimens.
Comparing the four curves appears that for θel > 0.6% the presence of a tensile resistant element
increased the dissipated energy of the specimen (S1, S2). With the horizontal applied force the
curve resulted smoother and less influenced by local factors (e.g. the friction behaviour between
lintel and pier).
The difference between the grey solid line and black solid line is due to the geometry of the
spandrel (S3, more slender). The squatter a spandrel is and the more energy it dissipates.
Black lines represent S3 and S4 dissipated energy. The presence of a well-connected timber
lintel let the specimen to dissipate more energy for θel > 0.6%.
Research Report

0.21 0.42 0.63 0.83 1.04 1.25
800
S1
700
S2
S3
600
Dissipated Energy [J] S4
500
400
300
200
100
0
0 5 10 15 20 25 30
Figure 2.47. Dissipated energy versus base displacement curves obtained from the tests on specimens S1
(grey dashed line), S2 (grey solid line), S3 (black solid line) and S4 (black dashed line).
Comparing the four curves appears that for θel more than 0.6% (δ≈7.5mm) the presence of a tension
resistant element increased the dissipated energy of the specimen (S1, S2). With the horizontal
applied force the curve resulted smoother and less influenced by local factors (e.g. the friction
behaviour between lintel and pier).
The difference between the red solid line and black solid line is due to the geometry of the spandrel
(S3 more slender). The squatter a spandrel is and the more energy it dissipates.
Black lines represent S3 and S4 dissipated energy. The presence of a well-connected timber lintel
let the specimen to dissipate more energy for θel more than 0.6% (δ≈7.5mm).
A method initially used for the substitute structure analysis was proposed by Jacobsen [1960] in
order to calculate the energy absorbed by hysteretic steady-state cyclic response. The method can
be used in order to characterize the behaviour of the 4 specimens by comparing their cyclic
dissipated energy. In particular, the formulation proposed by Jacobsen is:
Ah
hyst  (2.7)
2  F  
where Ah is the area within one complete cycle force-displacement response (the dissipated energy),
Fm and Δm are the maximum force and displacement at that cycle.
Figure 2.48 represents the calculated ξhyst for the four specimens.

0.21 0.42 0.63 0.83 1.04 1.25
25
S1
S2
20 S3
S4
15
ξhyst [%]
10
0
0 5 10 15 20 25 30
Figure 2.48. Hysteretic damping ξhyst vs. base displacement curves obtained from the test on S1 (red
dashed), S2 (red solid), S3 (black solid) and S4 (black dashed).
Looking at the figure it is possible to notice how the specimen #1, that exhibits a flexural behaviour,
is characterized by a greater ξhyst. In particular a great increment of the hysteretic damping is due to
a decrease of the resisting force without a decreasing of the dissipated energy. Specimen #2 and
#3 that only differs for the height of the spandrel have ξhyst values that smoothly vary between 5%
and 12%. Specimen # 3 was characterized by a higher value of hysteretic damping compared to the
other two specimen with a shear behaviour. This is due to the absence of a rigid timber lintel. For a
given element drift, the consequence of the dislocation of the stones is a considerably lower
resistance force associated to an almost constant dissipated energy. In this case the spandrel reach
a very high value of ξhyst=20%, associated to a heavy damage of the specimen.
2.12.2 Idealization of the response

The force-displacement cyclic behaviour were idealized by means of simplified multi-linear
envelopes. The first step for the evaluation of the multi-linear curve is the construction of a cyclic
envelope of the hysteresis loops in order to evaluate the maximum shear force and its degradation.
Subsequently the initial stiffness of the spandrel ks can be obtained by drawing the secant to the
experimental envelope at 0.7.Vmax, where Vmax is the maximum shear of the envelope curve. Finally
the displacement δu of the envelope curve is evaluated at the displacement corresponding to the
strength degradation of 20% of Vmax or to the maximum displacement reached during the test (if not
degrading). The residual shear resistance was evaluated for δ>δu. The value of shear Vu
corresponding to the horizontal branch of the bilinear curve has been found by enduring that the
areas below the cyclic envelope curve and below the equivalent bilinear curve are equal. Knowing
the initial stiffness ks and the value of Vu it is possible to evaluate the elastic displacement δe as
Vu/ks.
Figure 2.49 ideally represents this process while Figure 2.50 shows the idealized backbone curves
for S1 (red dashed), S2 (red solid), S3 (black solid) and S4 (black dashed).
Research Report
k
s
Figure 2.49. Idealization of the cyclic response.

0.21 0.42 0.63 0.83 1.04 1.25 1.45
60
S1
50 S2
S3
S4
40
Shear force [kN]
30
20
10
0
0 5 10 15 20 25 30 35
Figure 2.50. Idealized backbone curves for S1 (red dashed), S2 (red solid), S3 (black solid) and S4 (black
dashed).
Table 3.9 summarize geometrical characteristics and idealized response of all the four specimens.
Table 2.8. Spandrel geometry and idealized response.
Specimen h s [cm] l s [cm] σ h [Mpa] ropture lintel h l [cm] k s [kN/mm] δ e [cm] δ u [cm] V u [kN] V res [kN]
S1 108 120 0 flex. 12 14.5 2.05 6.9 29.7 17
S2 108 120 0.12 shear 12 11.5 4.56 25* 52.2 -
S3 78 120 0.11 shear 12 12 2.64 60* 31.6 -
S4 78 120 0.11 shear 2.5 12.7 1.61 9.1 20.4 18
* end of the test
2.12.3 Stiffness
The initial stiffness ks of the 4 specimens varies between 11.5 kN/mm (S2) and 14.5 kN/mm (S1). All
the value are reported in summary Table 3.9.
An interesting comparison could be comparison between the initial stiffness and the perfectly elastic
one calculated by means of the elasticity theorem:
1
 ls 3   ls 
kel     (2.8)
 Emh  t  hs G  t  hs 
3
where ls is the spandrel length (1200 mm), hs is the height of the masonry in the spandrel (1080 mm
for S1, S2 and 780 mm for S3, S4), t is the masonry thickness (320 mm), χ is the shear correction
factor (1.2 for rectangular sections), G is the shear modulus of the masonry (840 MPa), and Emh is
the horizontal elastic modulus of the masonry (only information on the vertical one are available:
Emv=2550 MPa).
From this, the calculated elastic stiffness of the spandrel considered restrained on rigid piers (with
Emh=Emv) are kel=150 kN/mm for the squat ones and kel=88 kN/mm for the slender ones.
The fact that in reality Emh<<Emv could partially explain why the elastic stiffness overestimates the
initial stiffness of the spandrel. However, this does not explain why the stiffness for the two different
geometries are in reality so close. This could be due to the fact that the mortar in slender specimens
was stronger or, more in general, that the theory of elasticity is not able to interpret well the behaviour
of a portion of masonry with low compression as the spandrel.
2.12.4 Relevant deformation thresholds

On the approximation of the force-deformation envelopes could be defined some relevant
deformation thresholds such damage limitation αDL or near collapse points αNC. From the
observations during the test first crack deformations were identified αFC. All the quantities are
reported in Table 3.10 in terms of equivalent pier rotations in order to better compare these to the
piers limit states.
The quantities are defined as follows:
- αFC : opening of the first crack during the test;
- αDL: end of the elastic branch on the equivalent linearization;
- αNC: end of the test or decay of 20% of Vmax.
Table 2.9. Spandrel relevant deformation thresholds in terms of equivalent piers rotation.
Specimen αFC αDL αNC

S1 0.07 0.09 0.29
S2 0.11 0.19 1.04*
S3 0.11 0.11 2.5*
S4 - 0.07 0.38
* end of the test
A consideration about the near collapse point is necessary. For masonry piers the ultimate rotation
capacity, which corresponds to the limit state near collapse, is defined as the drift at which the
strength has dropped below 80% of the pier’s peak strength. αNC is calculated for specimen with
degrading (S1 and S4) in the same way. If this limit is applied in an assessment procedure, a residual
spandrel strength would be neglected. This could be over conservative. It should also be considered
that the spandrels do not contribute to transfer significant portions of vertical load to the ground. A
collapse of a spandrel (that surely could be dangerous for people) should be considered
nevertheless as a local damage. Consider also that the quasi-static cyclic tests on masonry
spandrels showed that the collapse of spandrels supported on timber lintels is caused by the collapse
of the lintel supports and that the collapse of spandrels supported on masonry arches starts with the
collapse of the arch [Beyer & Dazio, 2012]. It seems that, for realistic piers’ drift, if the spandrel is
supported by a lintel an in-plane collapse of the spandrel is unrealistic. Up to now, studies on the out
of plane behaviour of damaged masonry beams are not present in literature.
Research Report
2.12.5 Peak and residual strengths

Formulation for peak and residual strength of the test were proposed or verified. Two different
behaviour were identified: a flexural one in S1 and a shear behaviour in S2, S3 and S4.
Flexure (S1)
Looking at the behaviour of the first specimen it is possible to observe that two well defined vertical
almost straight cracks opened at both interfaces between spandrel and piers, depicting the
substantial ineffectiveness of the stone interlocking. Hence, the resistance is mainly provided by the
friction between timber lintel and masonry. The lintel worked as a tie while the masonry as a strut.
Figure 2.51 is a representation of the hypothesis on the force distribution.
Figure 2.51. Strut and tie hypothesis for the flexural behaviour on S1.
The flexural force Vf could be calculated as follows:

T  Tsup  Tinf (2.9)
Tsup   c     p   Alint (2.10)
Tinf   c     p   Alint   
Ws
(2.11)
2
hs
Vf  T  (2.12)
ls
where c is the cohesion and μ the friction coefficient between wood and masonry, σp is the vertical
stress in the pier/lintel interface, Alint is the lintel/masonry interface area, Ws is the weight of the
spandrel, Tsup and Tinf are the friction force between lintel and masonry.
Making a hypothesis on c=0.15 and μ=0.5 the peak force results Vf(peack)=32.4 kN that is very close
to the real value of Vmax=32kN and Vu=29.7kN.
The cohesion c is not able to give any contribution once the lintel is dislocated from its original
position. In order to compute the residual strength the hypothesis are c=0 and μ=0.5. The residual
force results Vf(residual)=13.2 kN that is very close to the real value of Vres≈15kN.
Shear (S2, S3, S4)
In the tests 2, 3 and 4 it was observed a shear behaviour. Hence, in case of horizontal compression
force, the strength prediction is given by Italian code [MIT, 2008] by:
ft 
Vs  hs  t  1 h (2.13)
b ft
where hs is the masonry spandrel height, t is the thickness of the specimen, ft is the tensile strength
associated to diagonal compression test, b is defined as 1<ls/hs<1.5, σh is the horizontal compression
in the spandrel.
In case of S2, the parameters in Equation 3.22 assumes following values: hs=108 cm, t= 32 cm, ft =
0.137 MPa, b= 1.11 and σh = 0.12 MPa. The result is Vs=58.4 kN that is close to the value of Vu= 52
kN.
In case of S3, and S4 the prediction formula uses the same values: hs=78 cm, t= 32 cm, ft = 0.137
MPa, b= 1.5 and σh = 0.11 MPa. The result is Vs=30.6 kN that is close to the value for S3 Vu= 31.6
kN while not predict with sufficient precision the strength for S4 (Vu= 20.4 kN).
This difference is due to the different timber lintel. A stiff lintel allows the exploitation of the entire
cross section in the shear resistance. Conversely, if the lintel is not sufficiently rigid, the relative
dislocation of the stones induces a reduction of the effective cross section reducing the shear
strength of the masonry spandrel.
2.13 Conclusions
The experimental study presented in this work aimed at investigating the seismic behaviour of stone
masonry spandrels, both in the presence and in the absence of a horizontal tie rod or timber lintel.
For this reason an experimental apparatus was designed in order to test full-scale masonry spandrel
specimens. The test setup allows to investigate the behaviour of spandrels recreating representative
and controlled boundary conditions in the element.
Quasi-static cyclic tests were performed under displacement control on double leaf stone masonry
specimens. In particular, the laboratory tests on four full-scale masonry spandrel specimens that
were carried out at the EUCENTRE TREES Lab in 2010 and in 2013. As described in Chapter 3,
studying the results of the tested specimens it is possible to observe that the masonry spandrels
maintain a non-negligible in-plane resistance and dissipative capacity even after their cracking (the
end of elastic range occurred at an element drift θel=0.25~0.4% for all the specimens). The ductility
capacity was greater than 6 for all the specimens. This behaviour is possible thanks to the presence
of a well anchored timber lintel that supports the spandrel even after the formation of large cracks.
The timber lintel produced two main effects on the spandrel. The first one is to increase the strength
due to its own elastic stiffness; the second one is to prevent relative dislocation of the stones limiting
the cracking in the spandrel.
The effect of a horizontal compression has been studied. T he main difference between a test with
or without a horizontal tie rod consisted in a change in the failure mode from flexure/rocking with
added friction of the lintel to a shear one. This influenced the hysteretic behaviour characterized by
higher strength and "thinner" loops (5%<ξhyst<12%) in case of horizontal compression, lower strength
and "fatter" loops (10%<ξhyst<18%) in case of absence of the tie rod.
A flexural strength criterion was formulated based on the experimental results. The shear strength
criterion proposed by Italian code was validated based on the response of the three specimens that
exhibited such a behaviour. The code formulation appears valid if a rigid lintel as well as a horizontal
compression force is present.
This information will be useful to perform a more realistic pushover (or time history) analysis of the
complete structural models modelling properly the spandrel element. The performed tests allowed
the identification of the in-plane element limit states and the creation of idealized multilinearization
of the seismic response of masonry beams. This may help to formulate method for identifying the
structural performance levels based on damage and/or displacement/deformation indicators (local
or global), which would support in the interpretation of the results of dynamic analyses or simply
identify the limit states for a performance-based assessment procedure of masonry buildings.
2.14 Future developments
A further study to propose simple mechanical models that describe the force transfer mechanisms
in the masonry spandrels at peak and residual strength (e.g. Beyer [2012]) will be an interesting
study field. The models should account for the presence of tie rods and timber lintel since their
presence resulted a determining factor to modify the cyclic response of the spandrel elements. The
interaction of masonry spandrel and timber lintel should be investigated in greater detail in order to
Research Report
reﬁne the proposed models. As also reported by Beyer [2012] it was found that the strength of
masonry spandrels is highly dependent on the axial force in the spandrels. Future research should
therefore also develop models which allow to estimate the axial force in the spandrel in URM wall
conﬁgurations.
Another development should come from the calibration of numerical models (e.g. Milani et al. [2009]
or Da Parè [2011]) in order to allow a parametric study that investigates, as an example, the role of
spandrel beams on the seismic response of mutistorey walls or of entire masonry buildings.
Finally, it is noticed that information on the out-of-plane behaviour of damaged spandrel elements is
still missing.
2.15 REFERENCES
ASTM E519-02 [2002] Standard Test Method for Diagonal Tension (Shear) in Masonry Assemblages, Annual
Book of ASTM Standards, American Society for Testing and Materials, Philadelphia, Pennsylvania, USA.
Beyer, K. (2012). Peak and residual strengths of brick masonry spandrels. Engineering Structures, 41: 533-
547.
Brignola, A., Frumento, S., Lagomarsino, S., Podestà, S. [2008] “Identification of shear parameters of masonry
panels through the in-situ diagonal compression test,” International Journal of Architectural Heritage, Vol.
3, No. 1, pp. 52-73.
CEN/TC 125 [1998] EN 1052–1: Methods of test for masonry - Part 1. Determination of compressive strength.
CEN/TC 125 [1999] EN 1015–11: Methods of test for mortar for masonry - Part 11. Determination of flexural
and compressive strength of hardened mortar.
Da Parè, M. (2011). The role of spandrel beams on the seismic response of masonry buildings. M.Sc.
Dissertation: MEEES-ROSE School, Pavia, Italy.
Galasco, A., Magenes, G., Penna, A., Da Paré, M. [2009] “In-plane cyclic shear tests of undressed double leaf
stone masonry panels,” Proceedings of the 14th European Conference on Earthquake Engineering, Ohrid,
FYROM.
Giuffrè, A. [1993] Sicurezza e conservazione dei centri storici ― Il caso Ortigia, Laterza, Bari, Italy [in Italian].
Graziotti, F., Magenes, G., Penna, A. [2009] “Progetto di una sperimentazione su elementi di fascia muraria,”
ReLUIS Project Deliverable 4.3-UR01-1, University of Pavia and EUCENTRE, Pavia, Italy [in Italian].
Magenes, G., Penna, A., Galasco, A., Rota, M. [2010] “Experimental characterisation of stone masonry
machanical properites,” Proceedings of the 8th International Masonry Conference, Dresden, Germany.
Milani, G., Beyer, K., & Dazio, A. (2009). Upper bound limit analysis of meso-mechanical spandrel models for
the pushover analysis of 2-D masonry frames. Engineering Structures, 31, 2696–271
Research Report
3 MONOTONIC AND CYCLIC TESTS ON TUFF STONE MASONRY SPANDRELS AT THE

UNIVERSITY OF NAPLES FEDERICO II
Fulvio Parisia, Nicola Augentia
3.1 SUMMARY
This chapter presents the main results of six in-plane lateral loading tests on full-scale unreinforced
masonry (URM) walls with single opening and different spandrel types. All tests were carried out in
a pseudo-static fashion with displacement control. The first specimen was subjected to monotonicl
loading whereas the others were tested under cyclic loading up to near-collapse conditions. All
specimens were made of tuff stone masonry which is widely present in earthquake-prone regions,
especially in Mediterranean countries. Some specimens were externally strengthened through
inorganic matrix-grid (IMG) composites, also named textile-reinforced mortar and fibre-reinforced
cementitious matrix composites in the literature. Testing entire perforated walls allowed the authors
to investigate the influence of spandrel type on their in-plane seismic capacity, including the
spandrel-pier interaction. The as-built specimens featured three types of spandrels: (i) URM spandrel
with wooden lintel above the opening, (ii) URM spandrel with shallow masonry arch above the
opening, and (iii) spandrel with shallow masonry arch above the opening and a reinforced concrete
(RC) bond beam running on top resulting in a composite URM-RC spandrel. The influence of
spandrel type is analysed in terms of observed damage, force–drift curves, displacement ductility
capacity, overstrength, hysteretic damping, and residual drift.
a Department of Structures for Engineering and Architecture, University of Naples Federico II, via Claudio 21, 80125 Naples,
Italy.
3.2 Introduction
Current seismic codes at both national and international levels [D.M. 14.01.2008, 2008; EN 1998-3,
2005; FEMA 356, 2000] do not give appropriate specifications for modelling and verification of
spandrels in URM walls with openings. This is because masonry research has been mainly focused
on the lateral behaviour of piers so far [Magenes and Calvi, 1997]. Nevertheless, past experimental
tests have shown a key role of spandrels before and after their strengthening with traditional systems
[Benedetti et al., 1998]. Those tests revealed that masonry buildings have the highest energy
dissipation capacity if earthquake damage develops within spandrels rather than piers. Such
preliminary outcomes motivated the authors to investigate the role of spandrels in the in-plane lateral
behaviour of URM walls with openings. Full-scale testing on spandrel-pier subassemblages was
believed to be the most effective solution to solve that issue. All tests were carried out in the
laboratory of the Department of Structures for Engineering and Architecture at University of Naples
Federico II, Italy. This research was sponsored by the Italian Department of Civil Protection and was
performed in the framework of two national earthquake engineering research projects: ReLUIS-DPC
2005–2008 and ReLUIS-DPC 2010–2013.
Different masonry types and strengthening systems were considered by other research groups
worldwide, as discussed in chapters 2, 4 and 5. The authors of this chapter focused on tuff stone
masonry spandrels with different configurations and IMG composite strengthening systems. This
research was motivated by two facts:
i) The large presence of tuff masonry constructions (including cultural heritage buildings and
infrastructures) in Mediterranean countries and other earthquake-prone countries such as Japan,
ii) IMG composites allow a better compatibility with porous masonry substrates (such as tuff stone
masonry) and full reversibility of retrofit systems in historical structures, in addition to benefits of
resin-based composites such as negligible increase in stiffness and inertia mass.
This chapter presents the main findings of the experimental tests which are compared each other in
terms of crack patterns, force–drift curves, displacement ductility capacity, overstrength, hysteretic
damping, and residual drift. Further details on each test can be found in [Augenti et al., 2011] and
[Parisi et al., 2014].
3.3 Experimental Programme
In-plane lateral loading tests were carried out on three tuff stone masonry walls with an opening. In
order to design the specimens and test setup, experimental tests were carried out to characterise
the main mechanical properties of the materials used for the fabrication of the specimens. Wall
specimen S1 was tested in as-built, pre-damaged, and repaired-strengthened conditions. Each
specimen condition is here referred to as ‘subassemblage’, so that the aforementioned conditions of
specimen S1 are respectively named as subassemblages S1a, S1b, and S1c. In the case of
specimen S1, the spandrel panel above the opening was supported by a wooden lintel. In the case
of specimen S2, that lintel was replaced by a masonry arch and a RC bond beam running at the top
of the spandrel. Finally, specimen S3 only had a masonry arch above the opening. Specimen S2
was tested in as-built conditions up to collapse. Specimen S3 was tested in both as-built and
repaired-strengthened conditions, herein referred to as subassemblages S3a and S3b, respectively.
Two different IMG strengthening systems were applied over the spandrels of subassemblages S1c
and S3b.
3.3.1 Material Properties

The tuff masonry used for the construction of the specimens was composed of yellow tuff stones
and hydraulic mortar. The latter was a mixture of natural sand and pozzolana-like reactive
aggregates with 1:6.25 water/sand ratio by weight. The stones and mortar had a mean unit weight
equal to 11.72 kN/m3 (with coefficient of variation CoV = 1.73%) and 16.92 kN/m3 (CoV = 0.65%).
The mechanical behaviour of tuff stones and pozzolana-like mortar was investigated through
compression, tensile and direct shear tests [Augenti and Parisi, 2010, 2011]. Tuff stones had mean
tensile strength ft = 0.23 MPa (CoV = 22.06%), mean compressive strength fc = 4.13 MPa (CoV =
Research Report
18.54%), mean Young’s modulus E = 1.54 GPa (CoV = 6.43%), and mean shear modulus G = 0.44
GPa (CoV = 25.78%). The pozzolana-like mortar had ft = 1.43 MPa (CoV = 6.23%), fc = 2.50 MPa
(CoV = 7.34%), E = 1.52 GPa (CoV = 22.12%) and G = 0.66 GPa (CoV = 10.66%). Direct shear
tests on two-layer masonry specimens provided shear sliding strength at zero confining stress 0,bj =
0.146 MPa and friction coefficient  = 0.287 for tuff stone-mortar interfaces (further mechanical
properties can be found in [Augenti and Parisi, 2011]).
Tuff stone masonry as a whole was characterised by means of both simple and diagonal
compression tests [Augenti and Parisi, 2010; Parisi et al., 2013] to estimate key parameters for both
macro-element and finite element models [Parisi and Augenti, 2013; Parisi et al., 2011]. Uniaxial
compression tests on masonry prisms allowed the authors to get the following mean properties: fc =
3.85 MPa, E = 2.07 GPa and G = 0.86 GPa in the case of compressive loading parallel to mortar
bed joints; fc = 3.96 MPa, E = 2.22 GPa and G = 0.92 GPa in the case of compressive loading
perpendicular to mortar bed joints. Diagonal compression tests provided mean diagonal shear
strength at zero confining stress 0,d = 0.22 MPa according to ASTM guidelines and 0,d = 0.28 MPa
according to RILEM guidelines [Parisi et al., 2013].
The RC bond beam which was cast in place on top of the spandrel of specimen S3 was composed
of type C20/25 concrete and type B450C reinforcing steel bars [EN 1993, 1993; D.M. 14.01.2008,
2008]. Such concrete and steel types are equivalent to concrete type Rck250 and steel type Feb44k
in the past Italian code on RC and metallic structures. Compressive and three-point bending tests
were carried out on concrete cubes and prismatic specimens, respectively. The concrete had the
following properties: ft = 2.82 MPa (CoV = 9.48%), fc = 35.29 MPa (CoV = 14.31%) and E = 27.86
GPa (CoV = 10.18%). The bond beam was reinforced with four longitudinal bars (14 mm in diameter)
and 2-leg stirrups (8 mm in diameter) with 200 mm spacing. Tensile tests were performed on
reinforcing steel bar specimens. The longitudinal bars had a mean yielding strength fy = 545.46 MPa
(CoV = 5.36%), mean ultimate strength ftu = 632.63 MPa (CoV = 2.43%), mean hardening ratio ftu/fy
= 1.16 (CoV = 3.18%) and mean elongation at rupture Agt = 11.73% (CoV = 9.63%). The following
values were found in the case of stirrups: fy = 510.22 MPa (CoV = 5.78%), ftu = 605.57 MPa (CoV =
3.74%), ftu/fy = 1.19 (CoV = 2.11%) and Agt = 15.59% (CoV = 6.35%).
The IMG strengthening systems which were applied over the spandrels of subassemblages S1c
(type A IMG system) and S3b (type B IMG system) were composed of an inorganic matrix and an
alkali-resistant grid. In both cases, the IMG system matrix was made of hydraulic lime and sand
(ratio by weight 1:3), added with glass fibres (ratio by total weight 1:10) and mixed with latex and
water (ratio by weight 2:1). That two-component inorganic matrix had ft = 6.00 MPa, fc = 16.00 MPa
and E = 8.00 GPa. Conversely, two different grids were used. In the case of subassemblage S1c,
the type A IMG system grid was a biaxial glass coated net with approximately 25 mm spacing, unit
weight w = 225 g/m2, tensile strength per unit length Ft = 45 kN/m, E = 72.00 GPa, and ultimate
strain u = 1.77%. In the case of specimen S3b, the type B IMG system grid was a biaxial basalt net
with approximately 6 mm spacing, w = 250 g/m2, Ft = 60 kN/m, E = 91.00 GPa, and u = 2.00%. Steel
fibre-reinforced polymer (SFRP) ties were also employed to strengthen subassemblage S3b.
3.3.2 Geometry and Construction Features of Specimens

The size of specimens and test setup were designed in a way to avoid any pre-defined boundary
condition for the spandrel panel above the opening and to develop most part of damage within it. To
that end, a trial-and-error design procedure based on static pushover analysis of macro-element
models of subassemblage S1a was used, as discussed in [Augenti et al., 2011]. That procedure
allowed the authors to predict the sequence of failure modes under increasing lateral drift and the
monotonic force–drift behaviour. The application of axial loads on piers (each of them equal to 200
kN) was found to be associated with spandrel failure under a given size of specimens. The latter
were single-leaf masonry walls with an opening and consisted of two 1.70 m-long URM piers, which
were connected by a spandrel panel with 1.70 m length and 1.00 m height.
Tuff stones were 100×150×300 mm3 in size and were arranged in a way to build up a running bond
masonry with mortar joints having thickness equal to 10 mm. The opening was also 1.70 m long and
its height was equal to 2.30 m. In order to apply axial loads at the top of the piers ensuring structural
continuity with an ideal overlying storey, three masonry layers were constructed over the piers.
Therefore, each specimen was globally 5.10 m-long, 3.62 m-high and 0.31 m-thick.
Figure 3.1a and Figure 3.1b show specimen S1 in as-built and repaired-strengthened conditions,
respectively. The wooden lintel above the opening had an anchorage length equal to 150 mm in both
piers (Figure 3.1a). After that subassemblage S1b was tested, some masonry units were replaced
and cracks were filled with mortar. Then, the type A IMG strengthening system was applied on both
sides of the spandrel in order to ensure a satisfactory anchorage to the piers, hence delineating
subassemblage S1c (Figure 3.1b).
Type A IMG strengthening system
320
320
1000
1100
3620
3620
wooden lintel
2300
2200
1700 1700 1700 310 1700 1700 1700
5100 5100
(a) (b)
(a) (b)
Figure 3.1. Geometry of (a) subassemblage S1a and (b) subassemblage S1c (dimensions in mm).
The realisation of type A system on each spandrel side consisted of the following steps:
i) Masonry pre-wetting and execution of first mortar layer having thickness of approximately 5 mm,
ii) Application of IMG system glass grid onto the wet binder with fibres aligned to mortar bed joints
of the spandrel (grid plies having an overlapping length approximately equal to 100 mm),
iii) Execution of second mortar layer having thickness of approximately 5 mm.
The nominal thickness of the type A IMG strengthening system was then approximately 10 mm on
each spandrel side.
Figure 3.2 shows specimen S2 where a shallow masonry arch, instead of a wooden lintel, was
realised above the opening and a RC bond beam was cast in place on top of the spandrel. The
thickness and rise of the arch were 300 mm and 330 mm, respectively. The bond beam had a cross
section size of 310×230 mm2.
Figure 3.3a and Figure 3.3b show specimen S3 in as-built and repaired-strengthened conditions,
respectively.
RC bond beam
320
230
770
3620
masonry arch
2300
1700 1700 1700 310

5100
Figure 3.2. Geometry of subassemblage S2 (dimensions in mm).

Research Report
Type B IMG strengthening system
320
320
1000
1160
3620
3620
3620
masonry arch SFRP ties
2300
2140
1700 1700 1700 310 1700 1700 1700
5100 5100
(a) (b)
(a) (b)
Figure 3.3. Geometry of (a) subassemblage S3a and (b) subassemblage S3b (dimensions in mm).
After that the spandrel panel was reconstructed and some cracks were filled with mortar, the type B
IMG strengthening system was applied to the spandrel as follows:
i) Masonry pre-wetting and central wrapping of the spandrel panel with a first mortar layer having
thickness of approximately 8 mm,
ii) Application of the IMG system basalt grid onto the central wet binder with fibres aligned to mortar
bed joints of the spandrel,
iii) Central wrapping of the spandrel panel with a second mortar layer having thickness of
approximately 5 mm,
iv) Wrapping of the left side of the spandrel panel with a first mortar layer having thickness of
approximately 8 mm,
v) Application of the IMG system basalt grid onto the left-side wet binder with fibres aligned to
mortar bed joints of the spandrel,
vi) Realization of squared pockets with side length equal to 200 mm, both above and below the
spandrel panel,
vii) Second wrapping of the left side of the spandrel panel a second mortar layer having thickness
of approximately 5 mm,
viii) Replication of steps (iv) to (vii) on the right side of the spandrel,
ix) Masonry pre-wetting and execution of first mortar layer having thickness of approximately 13 mm
on the remaining parts of the spandrel (front side),
x) Application of the IMG system basalt grid over the whole spandrel,
xi) Execution of second mortar layer having thickness of approximately 5 mm over the whole
spandrel,
xii) Replication of steps (ix) to (xii) on the back side of the spandrel,
xiii) Drilling and compressed air-based cleaning of vertical holes (24 mm in diameter) through the
whole height of the spandrel panel, at the centre of the left top squared pocket and a distance of
400 mm from piers,
xiv) Installation of 1200 mm-long SFRP ties in the holes and filling with epoxy resin,
xv) Filling of squared pockets with latex and second mortar layer having thickness of approximately
5 mm.
Therefore, the nominal thickness of the type B IMG strengthening system was approximately equal
to 18 mm.
3.3.3 Experimental Setup and Instrumentation

The piers of each specimen were built up over RC beams with size 200×310×1900 mm3, which were
cast in place over -shaped steel plates. The latter were bolted in turn to squared holes of the
laboratory strong floor. Figure 3.4 shows the experimental setup of the in-plane lateral loading tests.
Transverse frames were installed over the piers to provide reaction against vertical loading applied
at the piers’ centrelines by hydraulic jacks with nominal capacity equal to 500 kN. Rigid steel beams
were placed above the piers to ensure their uniform loading and two couples of
polytetrafluoroethylene (PTFE) layers were installed between the beams and the hydraulic jacks to
minimise friction at their interface. A non-prismatic steel wall was anchored to the laboratory strong
floor to provide reaction against lateral loading. The latter was applied through a servo-controlled
hydraulic actuator with nominal compressive capacity, nominal tensile capacity and stroke equal to
500 kN, 290 kN and ±250 mm, respectively. A suspension system was installed at the top of the
reaction wall in order to provide further support to the actuator. Six steel bars (18 mm in diameter)
were employed to allow lateral force reversals during cyclic tests. Finally, steel beams were bolted
to transverse reaction frames at both specimen sides to prevent potential out-of-plane failure modes.
A load cell with nominal safe capacity equal to 200 kN and maximum capacity approximately equal
to 250 kN was positioned at the actuator end in order to measure the lateral resisting force.
FRONT LATERAL
HEA300 beam
PTFE layers
Hydraulic jacks
Reaction frame HEA300 beam
Hydraulic actuator
RC bond beam
Rigid steel plate
HEA160 beams
Load cell
Reaction wall HEA140 column
HEA160 beams
Prestressed
M14 bars
RC beam
Strong floor Strong floor
Figure 3.4. Experimental setup.
Linear variable differential transformers (LVDTs) and potentiometer transducers (PTs) were
installed. In the case of specimen S1 (Figure 3.5), LVDTs were placed at the end sections of both
the spandrel and pier panels to obtain information about flexural deformations. PTs were mounted
along both pier and spandrel diagonals to capture shear deformations. Joint panels were not
instrumented as their behaviour was expected to be rigid owing to the specimen geometry
(assumption confirmed by tests). Lateral displacements were measured at the opposite side with
respect to the actuator through a potentiometer tagged as PT #1. Displacement readings at PT #1
were associated with lateral force readings at the load cell to plot force–drift diagrams. The
aforementioned readings were considered to be more accurate than load and stroke readings of the
horizontal actuator.
Figure 3.6 shows the instrumentation of specimen S2, which was rather different from that of
specimen S1 because the authors decided to investigate rocking-induced displacements of piers
and deformations of the RC bond beam. Rocking deformations of specimen S2 were measured
through four vertical LVDTs at the base of piers on the front side and four vertical PTs on the back
side; those PTs were fixed to RC beams to include the width of rocking-induced cracks at the base
of piers in displacement readings. LVDTs were installed in the proximity of the RC bond beam ends
(back side of the specimen), which were expected to suffer plastic hinging as a result of piers’
rocking. In addition, a horizontal LVDT was installed at the middle of the RC bond beam (back side)
to measure axial deformations. A similar instrumentation was installed over specimen S3 (Figure
3.7).
Research Report
FRONT BACK
PT #1 PT #1
LVDT #1 LVDT #3 LVDT #9 LVDT #11
PT #2 PT #3 PT #8 PT #9
LVDT #5 LVDT #8 LVDT #15
3250
3250
PT #4 PT #5 PT #6 PT #7 PT #10 PT #11 PT #12 PT #13
Figure 3.5. Instrumentation of specimen S1 (dimensions in mm).
FRONT BACK
PT #1 PT #1
LVDT #15
PT #2 PT #3
LVDT #1 LVDT #4
LVDT #9 LVDT #10
LVDT #2 LVDT #2
3400
3400
3250
3250
LVDT #6 LVDT #7
LVDT #5 LVDT #8
FRONT BACK
PT #1 PT #1
LVDT #15
PT #2 PT #3
LVDT #1 LVDT #4
LVDT #9 LVDT #10
LVDT #2 LVDT #2
3400
3400
3250
3250
LVDT #6 LVDT #7
LVDT #5 LVDT #8
3.3.4 Test Procedures

Lateral loading tests of specimen S1 were performed in as-built, pre-damaged and repaired-
strengthened conditions. The first lateral loading test was conducted on subassemblage S1a in order
to investigate nonlinear response up to the onset of significant cracking (i.e., permanent damage) in
the spandrel. A second test was carried out on subassemblage S1b, attempting to assess residual
response properties of pre-damaged URM buildings located in earthquake-prone regions. This is a
very important task because: (i) many buildings damaged by past earthquakes have not yet been
repaired until now so their seismic capacity should be estimated by considering potential effects of
previous damage, and (ii) cumulative damage should be considered in seismic capacity models of
URM buildings subjected to earthquake sequences. At the end of test on subassemblage S1b, the
specimen was first repaired by filling cracks with mortar and replacing some stones in piers, and
then it was strengthened with the type A IMG strengthening system on both sides of the spandrel.
That procedure led to obtain subassemblage S1c which was cyclically tested until a near-collapse
level was attained. That test was performed to assess the effectiveness of the proposed
strengthening technique for retrofit/upgrading of URM structures and fast remedial works during
seismic emergency scenarios.
Specimen S2 was subjected to in-plane lateral loading until a near-collapse state was reached,
meaning that a heavy damage to both piers and spandrel was observed.
Specimen S3 was cyclically tested up to a near-collapse state in both as-built and repaired-
strengthened conditions. Subassemblage S3a was repaired after testing and then the spandrel was
externally strengthened with the type B IMG system, resulting in subassemblage S3b.
All tests were carried out in a quasi-static fashion with displacement control to investigate nonlinear
behaviour under increasing deformation demands, including post-peak softening branch of force–
drift diagrams.
Each lateral loading test consisted of two stages: (i) application of vertical forces with magnitude
equal to 200 kN over each pier, and (ii) application of displacement-controlled horizontal force under
constant vertical forces. The lateral force was modulated by a computer program of the data
acquisition system so that the target displacement time-history at the actuator was obtained. All
measurements were recorded according to a sampling rate of 5 Hz.
Subassemblage S1a was tested under monotonically-increasing displacements, after that two initial
displacement cycles were applied to get a satisfactory reaction level between specimen and
actuator. Lateral loading was applied so that displacements were increased with a constant rate v =
0.01 mm/s up to a displacement reading of 28 mm in the actuator (corresponding to an interstorey
drift ratio  ≈ 1% and a first significant damage to the spandrel). It is noted that the interstorey drift
ratio was defined as  = d/y0 where: d is the lateral displacement at either the actuator or PT #1 side,
and y0 = 3050 mm was the height of the lateral loading line from piers’ base. Subassemblage S1b
was tested under cyclic displacements with v = 0.35 mm/s in a way to attain increasing amplitudes
in 7 different blocks and 3 times at each amplitude peak, up to almost the same target displacement
(d = 39.2 mm) and drift ( = 1.2%) reached in the monotonic test on subassemblage S1a. The
displacement increment between consecutive displacement groups of three displacement cycles
was set to 5.6 mm. Subassemblage S1c was tested under 15 cyclic displacement blocks up to d =
84 mm ( = 2.8%). In that case the displacement rate was set to 0.35 mm/s in the first 7 blocks and
0.70 mm/s in the last 8 blocks.
Figure 3.8a shows displacement time-histories employed for cyclic tests on subassemblages S1b
and S1c.
Subassemblage S2 was cyclically tested under 17 cyclic displacement blocks (i.e., 51 cycles) up to
d = 95.1 mm (namely,  = 3.1%). The displacement rate and increment were set to 0.70 mm/s and
5.6 mm, respectively, as shown in Figure 3.8b.
The same input displacement time-history was applied to subassemblages S3a and S3b, but the
tests were stopped at lower drift levels as a result of the attainment of near-collapse conditions. The
tests on subassemblages S3a and S3b were stopped at  = 1.1% after 18 cycles.
v = 0.35 mm/s v = 0.70 mm/s v = 0.70 mm/s
100 100
50 50
d [mm]
d [mm]
0 0
-50 -50
-100 -100
0 5000 10000 15000 0 5000 10000 15000
t [s] t [s]
(a) (b)
Figure 3.8. Input displacement time-histories for cyclic tests: (a) Subassemblages S1b and S1c and (b)
subassemblages S2, S3a and S3b.
Research Report
3.4 Crack Patterns
The monotonic test on subassemblage S1a revealed the main two failure modes which the
specimens were prone to suffer before strengthening: (i) flexural cracks due to rocking behavior of
piers and spandrel panel, and (ii) diagonal shear cracking of the spandrel panel (Figure 3.9a).
Such types of cracking were mainly governed by the specimen geometry and magnitude of axial
loads on piers. Nevertheless, most part of damage concentrated in the spandrel panel. In the case
of subassemblages S1a and S1b, the spandrel panel suffered diagonal shear cracking and small
flexural cracks at end sections at max equal to 0.9% and 1.1%, respectively (in general, max is herein
intended to be the maximum drift measured by PT #1 on the opposite side with respect to actuator
side). In the case of subassemblage S1, bed-joint sliding was also observed between the first two
upper masonry layers of the spandrel panel. The spandrel panel of subassemblage S1b suffered
roughly symmetric shear cracks depending on the cyclic nature of lateral loading (Figure 3.9b).
The application of the type A IMG strengthening system over the spandrel of subassemblage S1c
changed the failure mode from brittle diagonal shear cracking to more ductile horizontal cracking
due to bed-joint sliding. The latter failure mode may be associated with a dominant bending
behaviour of the spandrel panel (Figure 3.9c). In addition, masonry splitting was detected in the
proximity of the piers’ base and also on the spandrel panel at max = 2.5%. Based on experimental-
analytical comparisons, Parisi et al. [2011] showed that such a transition from shear to flexural failure
occurred because the IMG strengthening system caused an increase in shear strength significantly
greater than that in flexural strength. This suggests the use of transverse connections between IMG
system layers through the thickness of the spandrel panel, if the flexural strength of the spandrel
panel must be further increased. As a matter of fact, those connections would have avoided masonry
failure by transverse splitting, resulting in greater load-bearing capacity of subassemblage S1c.
The masonry arch and RC bond beam in the spandrel panel of specimen S2 induced a considerably
different damage at max = 3.1% (Figure 3.10), which was the result of a failure sequence that can be
summarised as follows:
i) Formation of axial cracks in the arch at  = 0.69%,
ii) Collapse of a little central piece of arch at  = 0.86%,
iii) Collapse of arch wedge at  = 1.03% after the piers begun to experience rocking rotations,
iv) Collapse of third piece of arch and early masonry crushing at the base of the piers at
 = 1.21%,
v) Collapse of fourth arch portion and early concrete cover spalling in the RC bond beam
(top of end section) at  = 1.72%,
vi) Complete concrete cover spalling at  = 2.58%,
vii) Collapse of first and second masonry layers above the arch (at the first and second
displacement cycles, respectively) and plastic hinging of the RC bond beam at
 = 2.76%.
The early occurrence of damage to the spandrel denoted the significant fragility of masonry arch to
different displacements at its supports. The difference in both horizontal and vertical displacements
was due to different rocking rotations of the piers and the associated drift demand on the spandrel
panel. It is underlined that the difference in pier drifts occurred, even though the piers were subjected
to the same boundary conditions and axial load magnitudes (see Sect. 6).
FRONT BACK
(a)
(a)
FRONT BACK
(b)
(b)
FRONT BACK
(c)
(c)
Figure 3.9. Crack patterns of (a) subassemblage S1a, (b) subassemblage S1b and (c) subassemblage S1c.
The rocking behaviour of the piers generated large drift demands on the RC bond beam, which
suffered plastic hinging even though shear cracks propagated from the masonry below within the
beam. This was clearly caused by the fact a 200 mm stirrup spacing was too large to favour full
plastic hinging without shear cracking in the RC beam. It is noted that the past Italian masonry code
[D.M. 20.11.1987] allowed the use of stirrups with 6 mm diameter and up to 300 mm spacing, so the
RC bond beam well represented the condition detected in many existing masonry buildings located
in Italy. Flexural cracking of the piers and shear cracking of the spandrel are depicted in Figure 3.10.
The spandrel-pier joint panels did not suffer any cracking, as in the case of subassemblages S1a,
S1b and S1c. This confirmed the assumption of the macro-element method used to design the
specimens. Subassemblage S3a experienced cracking in the masonry arch of the spandrel panel
even at small drift levels, with no damage to the spandrel-pier joint panels. The cracks observed on
subassemblage S3a are shown in Figure 3.11a.
Research Report
FRONT BACK
Figure 3.10. Crack patterns of subassemblage S2.
Diagonal shear cracking begun to develop at both lower ends of the spandrel panel under cyclic load
reversals, and then it propagated along diagonals involving the whole panel, as the lateral loading
test proceeded. The first portion of arch fell down at  = 0.37%, that is a drift level close to half the
drift measured in the case of subassemblage S2 at the first cracking of the masonry arch. The
collapse of a second part of arch was observed at  = 0.56%, together with horizontal flexural cracks
at the base of piers. Then, the spandrel panel totally failed in shear at  = 1.12% which produced the
sudden loss of connection between the piers, and hence a significant drop in the lateral resisting
force of the entire subassemblage. After the occurrence of that failure, the upper triangular fraction
of masonry above diagonal shear cracks did not provide any contribution to the lateral resistance
and the test was stopped.
The cyclic test on subassemblage S3b induced horizontal flexural cracks at the base of piers, even
at small drift levels. Those cracks developed at the height of the first and third masonry layers of the
left pier (front side), rather than at the interface between RC beam and pier base sections (Figure
3.11b). As the lateral drift imposed to the subassemblage was increased, the width of such cracks
enlarged whereas the spandrel panel did not suffer major cracking with the exception of a vertical
cracking which formed at the upper left corner (front side). On the same side, the right pier failed in
diagonal tension and the lower right masonry toe crushed at  = 0.56%. A rather symmetric diagonal
crack formed as the lateral drift reversed. As was increased, the vertical crack in the spandrel panel
and the diagonal crack in the right pier (front side) developed until masonry splitting was detected
close to the base of the right pier and slight debonding of the IMG composite took place over the
spandrel-pier joint panel at  = 0.93%. Finally, when  = 1.12% was reached, the lower left toe of the
left pier crushed and vertical masonry splitting was observed at the arch-right pier interface. The test
was stopped in that condition because a significant resistance drop occurred. In-plane lateral loading
mainly induced damage to the piers without causing major cracking in the spandrel nor in the
spandrel-pier joint panel.
FRONT BACK
(a)
(a)
FRONT BACK
(b)
(b)
Figure 3.11. Crack patterns of (a) subassemblage S3a and (b) subassemblage S3b.
Detailed pictures of crack patterns observed on the spandrel panels are provided in Figure 3.12a
through to Figure 3.12f. In the case of subassemblage S1a, vertical cracks at both ends were induced
by the rocking behaviour of the spandrel. On the contrary, the vertical crack close to the mid section
was caused by the flexural deformation of the lintel. Even though diagonal shear cracking developed
throughout the spandrel panel, the wooden lintel supported the masonry above. This action was
favoured by the fact that the wooden lintel did not lose anchorage within the piers. Figure 3.12b
shows similar cracks in addition to a horizontal sliding shear crack along the mortar bed joint between
the upper two masonry layers. Those types of cracks were almost fully replaced by horizontal cracks
which developed within the type A IMG strengthening system at the locations of mortar bed joints,
in the case of subassemblage S1c (Figure 3.12c).
The susceptibility of the spandrel panel to diagonal shear cracking was also a feature of
subassemblage S2, although the whole spandrel had a RC bond beam on top. Figure 3.12d shows
that almost the whole masonry arch above the opening fell down at large drift levels, because of
different displacements at its supports and large drift demands on the spandrel panel. Nevertheless,
two masonry layers did not collapse and did not lose their bond to the upper RC bond beam. The
latter did not behave as a simple rod able to transfer only axial loads, but it developed a coupling
beam action dissipating drift demands through plastic hinging at its ends. It can be seen in Figure
3.12d that the bond beam lost its upper concrete cover and suffered shear cracking as a result of a
partially effective transverse reinforcement.
Figure 3.12e reveals a perfect diagonal shear cracking of the spandrel panel without RC bond beam.
The deformation of the spandrel panel was well captured by diagonal PTs, because they were fixed
over the piers at approximately 230 mm from the opening. The same instrumentation scheme was
applied in the case of subassemblage S3b. Figure 3.12f shows that the IMG-strengthened spandrel
panel of subassemblage S3b did not suffer heavy damage. Indeed, only a vertical flexural crack was
detected at the upper part of the left end section (front side) of the spandrel panel. Almost the same
drift levels on subassemblages S1a, S1b and S3a (i.e., max equal to 0.9%, 1.1% and 0.9%,
respectively) generated moderate-to-heavy shear cracking in the spandrel panel. This means that
Research Report
the type B IMG strengthening system (including two vertical SFRP ties passing through the whole
spandrel) inhibited shear cracking.
(a) (b)
(c) (d)
(e) (f)
Figure 3.12. Damage to spandrel panels at the maximum drift levels: (a) Subassemblage S1a ( max = 0.9%),
(b) subassemblage S1b (max = 1.1%), (c) subassemblage S1c (max = 2.5%), (d) subassemblage S2 (max =
3.1%), (e) subassemblage S3a (max = 1.1%) and (f) subassemblage S3b (max = 0.9%).
Figure 3.13a through to Figure 3.13f show the front side of subassemblages at the end of in-plane
lateral loading tests. Damage is more evident in the last four figures where different grades of
masonry crushing at the base of the piers can be identified. That type of damage was more significant
in the case of subassemblages S1c and S2, as the latter were subjected to very large drift levels
(max equal to 2.5% and 3.1%, respectively) compared to the others.
(a) (b)
(c) (d)
(e) (f)
Figure 3.13. Overall damage at the maximum drift levels: (a) Subassemblage S1a (max = 0.9%), (b)
subassemblage S1b (max = 1.1%), (c) subassemblage S1c (max = 2.5%), (d) subassemblage S2 (max =
3.1%), (e) subassemblage S3a (max = 1.1%) and (f) subassemblage S3b (max = 0.9%).
Figure 3.13f shows large diagonal shear cracks in the right pier which caused a significant drop in
the lateral resisting force and hence the need to stop the test.
The in-plane lateral behaviour of subassemblages was significantly affected by large rocking
rotations of the piers. The crack width at the base of the piers exceeded 25 mm in the case of
subassemblage S1c (see also [Augenti et al., 2011]), 30 mm in the case of subassemblage S2
(Figure 3.14a and Figure 3.14b), and 10 mm in the case of subassemblage S3b (Figure 3.15a to
Figure 3.15d). Typically, the piers of each subassemblage experienced different rocking rotations as
a result of the coupling action provided by the spandrel panel. In any case, the larger was the base
rotation of the piers the greater was the drift demand on the spandrel panel.
Research Report
(a) (b)
Figure 3.14. Damage to the piers of subassemblage S2: (a) Rocking-induced crack of the right pier and (b)
crack width in the left pier.
(a) (b)
(c) (d)
Figure 3.15. Damage to the piers of subassemblage S3b: (a) Masonry crushing at the base of the left pier,
(b) diagonal shear cracking of the right pier, (c) rocking-induced crack width at the base of the left pier and
(d) crack width in the right pier.
The rocking behaviour of the piers was less significant in the case of subassemblages S3a and S3b
because of the fragility of the spandrel panel with masonry arch only. In fact, the lack of the RC bond
beam at the top of the spandrel induced a poor cooperation between the piers, resulting in the early
progressive collapse of the spandrel panel and hence a near-collapse state for subassemblages S3a
and S3b. The right pier of subassemblage S3b (front side) failed in flexure (see toe crushing in Figure
3.15a) whereas the left pier suffered diagonal shear cracking (Figure 3.15b). Nevertheless, those
different failure modes produced almost the same crack width, which exceeded 10 mm when the
maximum drift level was imposed to the subassemblage (Figure 3.15c and Figure 3.15d).
3.5 Experimental Force–Drift Curves
Force and displacement readings recorded by the data acquisition system were processed in order
to obtain force–drift curves. In the following, the lateral resisting force is denoted by H. A comparative
analysis of such curves is performed herein, considering drift measurements provided by PT #1 (that
is, the wire potentiometer placed at the height of the horizontal actuator). Figure 3.16a shows a
bilinear force–drift behaviour of subassemblage S1a up to Hmax = 184 kN and Hmax = 0.65%, which
caused diagonal shear cracking in the spandrel panel and a 15% drop in the lateral resisting force.
The latter increased again under increasing lateral drift, but the test was stopped at the attainment
of max = 0.9% to allow further cyclic testing in pre-damaged conditions on subassemblage S1b. That
subassemblage also experienced a bilinear force–drift behaviour with small residual drifts and low
energy dissipation capacity (Figure 3.16b). Backbone curves in both positive and negative loading
orientations are almost the same. The maximum lateral resisting force of subassemblage S1b in the
positive loading orientation was approximately equal to that of subassemblage S1a at the occurrence
of diagonal cracking in the spandrel panel (that is, 157 kN). This indicates that previous shear cracks
in existing URM walls with openings can considerably affect the residual load-bearing capacity.
Figure 3.16c shows that the type A IMG strengthening system did not significantly modify the lateral
stiffness of subassemblage S1c, allowing a bilinear behaviour with limited energy dissipation
capacity and satisfactory re-centring capacity (i.e., small residual drifts). The magnitude of lateral
drifts measured by PT #1 was higher in the negative loading orientation, even though the envelope
force–drift curves were very similar showing almost the same load-bearing capacity. The IMG
strengthening system allowed the load-bearing capacity of subassemblage S1a to be restored,
delaying strength degradation at larger drift levels and hence increasing displacement ductility. It is
emphasised that the lack of substantial changes in the stiffness ratio between the spandrel panel
and the piers had a favourable impact. Otherwise, a stiffness redistribution after strengthening could
have generated a change in failure modes, resulting in potential adverse effects on the force–drift
behaviour.
All lateral loading tests on subassemblages S1a, S1b and S1c evidenced that the force–drift
behaviour (be it monotonic or cyclic) was clearly affected by rocking of piers, which allowed both re-
centring capacity and large drift levels. It is worth noting to compare the main results of
subassemblages in terms of resisting force and drift at three performance levels: (i) cracking onset,
(ii) peak resistance, and (iii) maximum deformation (see the three couples of force and drift columns
in Table 3.1). The cracking point in the experimental force–drift diagrams (the envelopes for cyclic
tests) was defined at a lateral stiffness reduction of 10%. Such a computation was carried out by
monitoring the ratio between the secant lateral stiffness k and the simple moving average at each
force step. The secant stiffness was defined as the ratio between the measured resisting force H
and the corresponding displacement d. In that way, significant variations in stiffness were identified
and the secant lateral stiffness at cracking was defined as the ratio between the estimated cracking
force Hcr and the associated displacement dcr (corresponding to the cracking drift cr).
Research Report
250
200
150
H [kN]
100
50
0
0 1 2 3 4
 [%]
(a)
250
200
150
100
50
H [kN]
0
-50
-100
-150
-200
-250
-4 -3 -2 -1 0 1 2 3 4
 [%]
(b)
250
200
150
100
H [kN]
50
0
-50
-100
-150
-200
-250
-4 -3 -2 -1 0 1 2 3 4
 [%]
(c)
Figure 3.16. Experimental force–drift hysteretic curves and envelope curves of (a) subassemblage S1a, (b)
subassemblage S1b and (c) subassemblage S1c.
Table 3.1. Global experimental response parameters of subassemblages.
Hcr cr Hmax Hmax Hmax max

Subassemblage Loading orientation
[kN] [%] [kN] [%] [kN] [%]
positive (monotonic
S1a 99 0.06 184 0.65 172 0.89
test)
positive 97 0.07 155 0.57 146 1.10
S1b
negative 97 0.07 146 0.75 138 1.15
positive 92 0.07 189 1.11 159 2.46
S1c
negative 95 0.06 182 1.05 155 2.40
positive 132 0.09 224 0.52 126 2.67
S2
negative 185 0.16 224 0.88 109 3.13
positive 96 0.06 153 0.38 141 0.59
S3a
negative 73 0.07 143 0.62 84 1.10
positive 114 0.06 246 0.70 185 0.90
S3b
negative 117 0.05 231 0.37 206 0.85
The cracking resisting force of subassemblages S1a, S1b and S1c was respectively equal to 0.5,
0.6 and 0.5 times the peak resisting force, while the Italian Building Code (IBC) [D.M. 14.01.2008,
2008] assumes Hcr = 0.7Hmax. The cracking drift was instead between 0.06% and 0.07%. All those
subassemblages reached the peak resisting force in the positive loading orientation. In particular,
the peak resisting force of subassemblage S1b was found to be 16% and 21% lower than that of
subassemblage S1a, in the positive and negative force–drift quadrants, respectively. The lateral drift
corresponding to Hmax was denoted as Hmax. The average value of Hmax related to both loading
orientations is 0.66%, that is, approximately equal to that of subassemblage S1a. The cracking force
of subassemblage S1c was lower than those of subassemblages S1a and S1b, although no
significant changes in the relevant drift were found. The repair of subassemblage S1b and the
application of the type A IMG strengthening system over the spandrel (subassemblage S1c) induced
an increase in lateral load-bearing capacity, which was 22% and 25% higher than those computed
in the positive and negative loading orientations, respectively. Those resiting force levels were
reached at Hmax ranging between 1.05% and 1.10%.
The force–drift hysteretic behaviour of subassemblage S2 was bilinear as well, but the attainment of
peak resisting force was followed by a significant strength degradation up to the maximum drift
imposed during the test (Figure 3.17). The cracking force and the corresponding drift were
considerably higher than those estimated above. The ratio Hcr/Hmax was 0.6 in the positive loading
orientation and 0.8 in the negative loading orientation, and the presence of the RC bond beam on
top delayed the 10% stiffness reduction up to force and drift levels notably higher than those related
to subassemblages S1a, S1b and S1c. The peak resisting force was the same in both orientations,
but the relevant drifts were significantly different each other. The secant lateral stiffness at cracking
was approximately 10% lower than that estimated for subassemblage S1a; this can be motivated by
the presence of the masonry arch above the opening of subassemblage S2. It is emphasised that
Hmax was 22% higher than that measured for subassemblage S1a, in contrast to what was found in
terms of Hmax. Nevertheless, in the post-peak force–drift range, the gradual deterioration of the
spandrel panel progressively reduced the piers’ coupling under increasing lateral drift, causing a
rapid strength degradation of subassemblage S2.
Research Report
250
200
150
100
H [kN]
50
0
-50
-100
-150
-200
-250
-4 -3 -2 -1 0 1 2 3 4
 [%]
Figure 3.17. Experimental force–drift hysteretic curve and envelope curve of subassemblage S2.
250
200
150
100
H [kN]
50
0
-50
-100
-150
-200
-250
-4 -3 -2 -1 0 1 2 3 4
 [%]
(a)
250
200
150
100
H [kN]
50
0
-50
-100
-150
-200
-250
-4 -3 -2 -1 0 1 2 3 4
 [%]
(b)
Figure 3.18. Experimental force–drift hysteretic curves and envelope curves of (a) subassemblage S3a and
(b) subassemblage S3b.
The force–drift behaviour of subassemblages S3a and S3b is shown in Figure 3.18a and Figure
3.18b, respectively, where one can observe their low displacement capacity values compared to
those discussed for the aforementioned subassemblages. Conversely, large residual drifts were
measured especially in the case of subassemblage S3b as a consequence of the different behaviour
of the spandrel strengthened with the type B IMG system. No significant modifications in cracking
drifts can be observed in Table 3.1. Vice versa, peak resisting forces of subassemblages S1a, S2
and S3a were considerably different. In fact, subassemblage S3a was characterized by a peak
resistance 17% and 22% lower than that of subassemblage S1a in the positive and negative loading
orientations, respectively. The peak resisting force of subassemblage S3a was about the same of
subassemblage S1b in both loading orientations. The presence of the RC bond beam in
subassemblage S2 led to a peak resisting force 46% and 57% higher than those computed for
subassemblage S3a in the positive and negative orientations, respectively. Nonetheless, the
average values of cracking stiffness of subassemblages S2 and S3a were about the same.
If the positive loading orientation is considered, the cracking force of subassemblage S3b was 19%
higher than that of subassemblage S3a. The type B IMG strengthening system contributed to a
significant increase in load-bearing capacity, which was found to be 61% regardless of the loading
orientation. Therefore, the repair and strengthening interventions allowed not only to restore the peak
resistance of subassemblage S3a but also to considerably increase it. No improvements were
recorded in terms of both displacement and ductility capacities, probably because a change from
spandrel failure to left pier failure (front side) occurred in the case of subassemblage S3b.
Finally, the envelopes of hysteretic force–drift curves are compared in Figure 3.19. It can be
observed that:
i) initial stiffness of subassemblages was about the same and different from cracking stiffness
corresponding to a 10% reduction in secant stiffness on the rising branch of the force–drift
envelopes,
ii) the peak resisting force ranged approximately between 150 kN and 250 kN in both loading
orientations, that is, between 0.3 and 0.6 times the total vertical load on subassemblages (such
ratios are typically referred to as ‘seismic coefficients’),
iii) the spandrel with masonry arch only (subassemblages S3a and S3b) induced a very small
displacement capacity, which can be significantly increased by the addition of a RC bond beam
on top.
250
200
150
100
50
H [kN]
0
-50
-100 S1a S1b
-150 S1c S2
-200 S3a S3b
-250
-4 -2 0 2 4
 [%]
Figure 3.19. Comparison between experimental force–drift envelopes.
3.6 Ductility Capacity and Overstrength
The envelopes of the experimental force–drift curves were approximated as bilinear to characterise
the response of an idealised single-degree-of-freedom (SDOF) system. Idealised bilinear force–
deformation diagrams are extensively reported in literature as simplified models for evaluating the
in-plane seismic response of masonry walls, especially under cyclic loading [Magenes and Calvi,
1997; Shing et al., 1989; Vasconcelos and Lourenço, 2009].
Two bilinear idealisation procedures according to Tomaževič [2000] and both IBC [D.M. 14.01.2008,
2008] and Eurocode 8 (EC8) [EN 1998-3, 2005] were used to estimate the ultimate force Hu, yielding
drift e (corresponding to de), elastic stiffness ke, displacement ductility capacity , and overstrength
factor .
Research Report
The displacement ductility capacity is a useful characteristic that makes possible the reduction of
elastic seismic design actions, since it gives an indication of the ability of the structure to dissipate
the earthquake input energy. Such a ductility capacity was computed as:
u
 (3.1)
e
where u is the ultimate drift. The latter was assumed to be the experimental drift corresponding to
the following resisting force level:
H u  C Sd Hmax (3.2)
where CSd is the strength degradation factor that was assumed to be not lower than 0.8. Assuming
CSd ≥ 0.8 means that strength degradation on the post-peak falling branch of the force–drift envelope
is supposed to be not greater than 20%. This assumption derived from the fact that tests on
subassemblages were stopped at strength degradation levels usually lower than that typically
considered in seismic codes (i.e., 20%). Typically, the ultimate capacity point ( u,Hu) of the force–
drift curve was not equal to the experimental point at maximum deformation (Hmax,Hmax). It is also
underlined that the bilinear idealisation for subassemblages S1a and S1b was aimed at reaching
some hints on displacement ductility and strength reduction factors, being they related to moderate
damage levels. The computation of these seismic capacity parameters was useful especially to
compare the global performances of subassemblages.
The first bilinear idealisation procedure was based on the assumption of elastic stiffness ke equal to
the experimental secant stiffness at cracking kcr according to Tomaževič [2000]. The yielding drift
was computed by imposing equal areas below the actual and idealised curves. The ultimate force of
the idealised system was derived as the product of the assumed elastic stiffness for the yielding
displacement corresponding to the computed yielding drift. In this way, the ultimate resistance Hu of
the idealised envelope was evaluated by assuming equal energy dissipation capacity of the actual
and idealised structural systems.
The second bilinear idealisation procedure used in this study is reported in IBC [D.M. 14.01.2008,
2008] and EC8 [EN 1998-3, 2005]. Those building codes allow one to assume cracking at 0.7Hmax.
The elastic stiffness is then computed as the ratio between the idealised cracking force and the
corresponding displacement level on the actual curve. The yielding displacement is derived by
assuming equal areas below the actual and idealised force–drift diagrams.
The overstrength factor of each subassemblage was defined as:
Hu
 (3.3)
H cr
namely, the ratio between the ultimate and cracking forces. It is noted that IBC establishes to set 
= 1.4 in the case of single-storey URM buildings and  = 1.3 in the case of single-storey reinforced
masonry (RM) buildings.
Considering only displacement readings provided by PT #1 on the opposite side with respect to that
of the horizontal actuator, the properties of the idealised SDOF systems corresponding to the
subassemblages were obtained. Such properties are outlined in Table 3.2 and Table 3.3
for the bilinear idealisation procedures according to Tomaževič and to IBC and EC8, respectively.
If the estimates related to positive and negative orientations of lateral loading are averaged (Table
3.4 and Table 3.5), the differences between results of the two bilinear idealisation procedures tend
to vanish. This gives the chance to identify the main effects of the spandrel configuration on the in-
plane seismic capacity. Nevertheless, the bilinear idealisation procedure provided by IBC and EC8
was more sensitive to the post-cracking branch of force–drift envelopes, overestimating the yielding
drift and hence underestimating the ductility capacity. This was not the case of subassemblages S1b
and S2 for which capacity estimates were almost equal to those derived according to Tomaževič,
since elastic stiffness was approximately the same in both bilinear idealisation procedures due to
the high stiffness of post-cracking branch.
Table 3.2. Idealised SDOF system properties and overstrength factor according to Tomaževič.
Hu e u
Subassemblage Loading orientation  
[kN] [%] [%]
positive
S1a 162 0.10 0.89 8.51 1.64
(monotonic test)
positive 147 0.10 1.10 10.94 1.51
S1b
negative 137 0.10 1.15 11.19 1.39
positive 174 0.13 2.46 19.16 1.92
S1c
negative 169 0.09 2.40 27.75 1.70
positive 211 0.14 1.63 11.85 1.60
S2
negative 212 0.18 1.47 8.00 1.14
positive 147 0.09 0.59 6.44 1.53
S3a
negative 136 0.13 1.04 7.74 1.85
positive 229 0.11 0.89 7.93 2.01
S3b
negative 215 0.10 0.85 8.72 1.84
Based on such considerations, the estimates listed in Table 3.4 can be used to compare seismic
capacity properties of subassemblages. Figure 3.20 allows one to compare idealised force–drift
diagrams where the resisting force H is replaced by its ratio to the total vertical load (denoted as H/V).
First of all, the yielding drift was found to range between 0.10% and 0.11% in the case of
subassemblages S1a–c and S3a–b, whereas it was estimated in 0.16% in the case of
subassemblage S2. On the other hand, displacement capacity of subassemblage S1c was 1.6, 3.0
and 2.8 times those of subassemblages S2, S3a and S3b, respectively. The ultimate resisting force,
ultimate drift and displacement ductility of subassemblage S2 were correspondingly 49%, 90% and
40% greater than those of subassemblage S3a, as a result of the RC bond beam on top of the
spandrel. The overstrength factor of subassemblage S2 was 19% lower than that of subassemblage
S3a.
Table 3.3. Idealised SDOF system properties and overstrength factor according to IBC and EC8.
Hu e u
Subassemblage Loading orientation  
[kN] [%] [%]
positive
S1a 171 0.19 0.89 4.68 1.73
(monotonic test)
positive 152 0.16 1.10 6.77 1.51
S1b
negative 138 0.11 1.15 10.43 1.40
positive 180 0.28 2.46 8.67 1.98
S1c
negative 174 0.21 2.40 11.26 1.74
positive 212 0.15 1.63 10.76 1.61
S2
negative 210 0.16 1.47 8.91 1.13
positive 150 0.11 0.59 5.23 1.56
S3a
negative 135 0.12 1.04 8.51 1.84
positive 234 0.15 0.89 5.89 2.06
S3b
negative 220 0.13 0.85 6.36 1.88
Table 3.4. Average estimates of SDOF system properties and overstrength factor according to Tomaževič.
Research Report
Hu e u
Subassemblage  
[kN] [%] [%]
S1a 162 0.10 0.89 8.51 1.64
S1b 142 0.10 1.13 11.07 1.45
S1c 172 0.11 2.43 23.46 1.81
S2 212 0.16 1.55 9.93 1.37
S3a 142 0.11 0.82 7.09 1.69
S3b 222 0.11 0.87 8.33 1.93
Table 3.5. Average estimates of SDOF system properties and overstrength factor according to IBC and EC8.
Hu e u
Subassemblage  
[kN] [%] [%]
S1a 171 0.19 0.89 4.68 1.73
S1b 145 0.14 1.13 8.60 1.46
S1c 177 0.25 2.43 9.97 1.86
S2 211 0.16 1.55 9.84 1.37
S3a 143 0.12 0.82 6.87 1.70
S3b 227 0.14 0.87 6.13 1.97
0.6
0.4
H/V
0.2
S1a S1b
S1c S2
S3a S3b
0
0 0.5 1 1.5 2 2.5 3
 [%]
Figure 3.20. Idealised force–drift diagrams of subassemblages.
In the case of subassemblage S3b, strengthening of the spandrel with the type B IMG system
allowed a 57% increase in ultimate resisting force compared to subassemblage S3a. In addition,
since the yielding drift reduced of 5% and the ultimate drift increased of 7%, the ductility capacity
increased as well (approximately 17%). Finally, a 14% overstrength increase was also estimated. It
is worth noting that the improvement of the in-plane seismic capacity of subassemblage S3b was
reached even if near-collapse conditions were attained at the occurrence of diagonal shear cracking
in the right pier (front side) rather than in the spandrel panel.
3.7 Analysis of Pier Drifts
Displacement readings provided by the horizontal actuator and PT #1 at opposite edges of each
subassemblage allowed the estimation of lateral drifts of piers. The difference  between such drifts
can be used as a measure of the effectiveness of the spandrel as coupling element for the piers.
When such a difference was not zero, the spandrel shortened and hence was more able to transfer
a fraction of horizontal force from a pier to another. In Figure 3.21 the lateral resisting force is plotted
against the difference between pier drifts, based on the envelopes of experimental force–drift curves
in the positive loading orientation.
In the case of subassemblage S1a that was monotonically tested, the relationship between H and 
was rather linear up to H ≈ 120 kN, namely, the horizontal force level associated with the corner point
of the experimental force–drift curve shown in Figure 3.16a. In that condition,  was approximately
equal to 0.12% and was associated with drift levels of about 0.22% and 0.10% for the left and right
piers, respectively. It is noted that experimental force–drift curves are plotted in Figure 3.16a, Figure
3.16b and Figure 3.16c by considering the drift of the right pier as a reference. Therefore, the ratio
/ was about 1.2 at the corner point of the experimental force–drift curve of subassemblage S1a.
As H increased up to its maximum level and then decreased until the maximum drift was reached,
 gradually reduced as a result of spandrel elongation. This was also induced by the fact that rocking
rotations of the right pier (front side of subassemblage S1a) were delayed at larger drift levels.
300
S1a S1b
250 S1c S2
S3a S3b
200
H [kN]
150
100
50
0
0 0.1 0.2 0.3 0.4 0.5 0.6
 [%]
Figure 3.21. Lateral resisting force versus variation between pier drifts.
In the case of subassemblage S1b the trend of / was rather linear up to H ≈ 142 kN and  ≈
0.14% (corresponding to the second corner point in the force–drift envelope shown in (c)
Figure 3.16b). Nonetheless,  increased more quickly than in the case of subassemblage S1a under
increasing H. The corner point of the H– diagram related to such a subassemblage is associated
with  ≈ 0.21%, which outlines that / was about 1.5, namely greater than in the case of
subassemblage S1a. It is interesting that  increased to 0.20% when  reached 1.10%, namely,
their ratio reduced to 0.18 reflecting that damage to the spandrel caused a 18% variation in the pier
drifts.
In the case of subassemblage S1c, the relationship between H and was roughly linear up to Hmax =
189 kN and  = 0.18%, which was associated with Hmax = 1.11% (corresponding to / = 0.16). As
the lateral drift of that subassemblage increased,  linearly reduced up to 0.30% at max = 2.46%
corresponding to / = 0.12 in near-collapse conditions.
It is noteworthy that similar linear trends were found for subassemblages S2, S3a and S3b as the
lateral resisting force increased from zero to its peak level. In particular, in the case of
subassemblage S2 the pier drift variation reached 0.21% at Hmax = 224 kN, which in turn was
associated with Hmax = 0.52% (corresponding to / = 0.40). As the lateral drift increased up to  =
Research Report
1.63%,  reduced to 0.20% at H = 184 kN (peak resisting force of subassemblage S1a), resulting
in / = 0.12 at that stage of the test. As  was further increased up to 2.67%,  increased up to
0.44% resulting in / = 0.16 in near-collapse conditions.
In the case of subassemblage S3a the relationship between H and  was rather linear up to  =
0.15% and H = 147 kN, which is close to the maximum resisting force Hmax = 154 kN. That pier drift
variation was associated with the corner point of the force–drift envelope at  = 0.21%, resulting in
/ = 0.71. As  increased to 0.59% in the positive orientation, H linearly decreased and 
increased up to  = 0.54%, resulting in / = 2.57.
Finally, subassemblage S3b experienced a linear H– relationship fully overlapped to that of
subassemblage S3a, until it reached Hmax = 246 kN and  = 0.23% at Hmax = 0.70% (corresponding
to / = 0.33). As the lateral drift of that subassemblage increased up to max = 0.90%, the lateral
resisting force and  reduced to 185 kN and 0.22%, respectively, producing / = 0.24.
3.8 Hysteretic Damping
The energy dissipation capacity of subassemblages was also investigated and indirectly measured
through the hysteretic damping ratio, which is used in displacement-based design/assessment of
structures [Priestley et al., 2007].
When diagonal shear cracking of piers did not occur, the flag-shaped hysteretic response of
subassemblages produced a very limited dissipation capacity, which however is not the unique
capacity parameter providing earthquake resistance to URM structures. Hysteretic damping ratio
was not estimated for subassemblage S1a, because monotonic testing was carried out in that case.
It is underlined the equivalent damping ratio is typically regarded as the sum of two components, as
follows:
eq  eq ,e  eq ,h (3.4)
where eq,e is the equivalent damping ratio associated with elastic behaviour and eq,h is the equivalent
damping ratio associated with hysteretic behaviour. eq,h was computed by equating the energy Eh
dissipated in a hysteresis cycle of the subassemblage to the energy Es/4 dissipated by a viscous
damping cycle of an equivalent linear system [Chopra, 2006], where Es is the strain energy.
Therefore, eq,h was defined according to Equation (3.5):
1 Eh
eq ,h  (3.5)
4 Es
which turns out to be:
Eh
eq ,h  (3.6)
2 ke c y 0 
2
where ke is the elastic stiffness, c is the maximum drift of the subassemblage in a cycle, and y0 is the
height at which the lateral displacement d was measured.
Figure 3.22 shows the hysteretic damping ratio under varying drift of each subassemblage. The low
damping levels were basically due to the fact that almost all subassemblages experienced small
areas within cycles compared to elastic areas corresponding large drifts. In detail, it can be observed
that the hysteretic damping ratio was not greater than 4.5%.
5
4.5% S1b
S1c
4 3.9% S2
S3a
3 2.6% S3b
eq,h [%]
2.3%
2
1 1.3%
0
0 1 2 3 4
 [%]
Figure 3.22. Hysteretic damping ratio versus subassemblage drift.
3.9 Residual Drifts
Structures subjected to large inelastic deformations during severe ground shaking do not always
return to their initial ‘at-rest’ position, showing residual displacements. Even though collapse does
not occur, large residual drifts make the structure no longer usable or may require too expensive
repair measures. It is thus important to measure residual drifts after cyclic testing or real
earthquakes, in order to assess repairability conditions for different design/retrofit cases. Some
researchers have proposed the use of residual deformations as additional engineering demand
parameters together with maximum deformations, to quantify earthquake damage at different
seismic intensity levels [Pampanin et al., 2002]. In fact, residual interstorey drifts can have great
impact on economic losses, as shown in recent studies [Ramirez and Miranda, 2012].
In this section, residual drifts r of subassemblages (except for subassemblage S1a, which was
subjected to monotonic testing) are investigated under increasing maximum drift  imposed during
each loading cycle (Figure 3.23). Herein,  is associated with displacement readings at the horizontal
actuator, because they were less disturbed by progressive damage of subassemblages.
Nevertheless, the same findings could be drawn considering displacement readings of the wire
potentiometer PT #1.
Subassemblages S1b experienced zero residual drifts until  = 0.15% was reached; as  increased,
a logarithmic increase in r occurred until the residual-to-maximum drift ratio RMDR = r/ reached
11% at  = 1.26%. The same trend was found in the case of subassemblage S1c, which experienced
non-zero residual drifts for  ranging between 0.73% and 2.75%. In the latter condition RMDR was
found to be 8%.
Research Report
0.6
S1b
0.5 S1c
S2
S3a
0.4 S3b
r [%]
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5
 [%]
Figure 3.23. Residual drift versus subassemblage drift.
The presence of a masonry arch and RC bond beam in subassemblage S2 did not modify the onset
of residual drifts, which became different from zero at  = 0.18%. As  increased, the residual drift
increased until it reached 0.16% at  = 3.13%, resulting in RMDR = 5%. This indicates that the RC
bond beam was able to limit residual deformations, even if no significant improvements in terms of
hysteretic damping were found. This finding satisfactorily agrees with previous remarks, because
large drifts due to rocking behaviour of piers were observed during the test on subassemblage S2
and most part of permanent damage (and hence residual drifts) was caused by shearing
deformations.
Different trends in residual drifts can be observed for subassemblages S3a and S3b, which suffered
residual drifts larger than those discussed above. In both cases, r was zero until  reached 0.18%.
As the lateral drift increased, the residual drift of subassemblage S3a increased up to 0.26% at  =
1.12%, resulting in RMDR = 23%. In the case of subassemblage S3b, the residual drift increased
with the lateral drift, reaching r = 0.55% at  = 1.12%, resulting in RMDR = 49%.
3.10 Conclusions
The full-scale lateral loading tests on tuff masonry walls with a single opening provided useful
information in terms of crack patterns, force–drift behaviour, capacity properties of idealised SDOF
systems, overstrength factor, pier drifts, hysteretic damping, and residual drifts.
All tests evidenced the key role of rocking and spandrel configuration in the in-plane lateral behaviour
of subassemblages. The specimens tested in as-built and pre-damaged conditions (i.e.,
subassemblages S1a, S1b, S2 and S3a) suffered: (i) flexural cracks close to end sections of the
spandrel panel and base of the piers, and (ii) diagonal shear cracking of the spandrel panel. The
application of IMG strengthening systems caused a change in failure modes. Subassemblage S1c
experienced a transition from diagonal to extensive horizontal cracking of the spandrel panel,
whereas subassemblage S3b reached near-collapse conditions as a result of diagonal shear
cracking in a pier.
The geometry of subassemblages and the magnitude of axial loads on the piers influenced the rising
branch of force–drift envelopes, allowing a stable hysteretic rocking behaviour with large
displacements and small residual drifts. A crack width exceeding 25 mm and 30 mm was detected
on subassemblages S1c and S2, respectively, when near-collapse conditions were attained. Those
width levels reduced to about 10 mm in the case of subassemblage S3b. No significant damage to
spandrel-pier joint panels or debonding of IMG strengthening systems were observed.
Subassemblages S2 and S3a experienced a progressive collapse of the masonry arch above the
opening, even at small drift levels. When the rocking behaviour of piers was significant, large drift
demands on the spandrel panel were measured. This caused plastic hinging in the RC bond beam
running at the top of the spandrel in the case of subassemblage S2; in this way, the propagation of
shear cracks from the masonry below to the RC bond beam shows that 20 mm stirrup spacing should
be reduced to withstand plastic rotation demands generated by piers’ rocking. The presence of
masonry arch without RC bond beam in the spandrel of subassemblage S3a drastically reduced the
in-plane seismic capacity; it is worth noting the the first two parts of masonry arch fell down at drift
levels approximately equal to half the drift causing similar damage to subassemblage S2.
The type A IMG strengthening system applied over the spandrel of subassemblage S1c allowed the
restoration of the load-bearing capacity related to as-built conditions (subassemblage S1a), re-
centring capacity, and low strength degradation even at large drift levels. The increase in peak
resistance was estimated in 22% in that case. The same increase was found for subassemblage S2,
compared to subassemblage S1a, highlighting the great impact of the RC bond beam on top of the
spandrel. The bond beam also ensured a peak resistance 46% and 57% higher than that computed
for subassemblage S3a in the positive and negative orientations, respectively. Conversely, the load-
bearing capacity of subassemblage S3a was 17% and 22% lower than that of subassemblage S1a
in the positive and negative loading orientations, respectively. The type B IMG strengthening system
contributed to a significant increase in load-bearing capacity, which was found to be 61% regardless
of the loading orientation.
Bilinear idealisations of force–drift envelopes were derived to estimate seismic capacity properties.
Based on those computations, it can be outlined that:
 The peak lateral resistance can be supposed to be 35%, 30%, and 45% of the total gravity load
in the case of spandrel with wooden lintel, masonry arch, and both masonry arch and RC bond
beam, respectively,
 The yielding drift can be set to 0.1% in the case of spandrel with wooden lintel or masonry arch
and 0.16% in the presence of RC bond beam on top, either in as-built and IMG-strengthened
conditions,
 The ultimate drift can be assumed to be 0.8% in the case of spandrel with wooden lintel or
masonry arch and 1.5% in the presence of RC bond beam; this capacity feature can be set to
2.4% in the case of spandrel with wooden lintel and type A IMG strengthening and 0.8% in the
case of spandrel with masonry arch and type B IMG strengthening,
 The ductility capacity can be set to 7, 8.5, and 9.5 in the case of spandrel with masonry arch,
wooden lintel, and both masonry arch and RC bond beam, respectively; this capacity feature can
be assumed to be 2.5 times higher if the spandrel with wooden lintel is strengthened through the
type A IMG composite and 1.1 times higher if the spandrel with masonry arch is strengthened
through the type B IMG composite,
 The overstrength factor can be set to 1.6 in the case of spandrel with wooden lintel or masonry
arch and 1.3 in the presence of RC bond beam; this capacity feature can be increased to 1.8 in
the case of spandrel with wooden lintel and type A IMG strengthening and 1.9 in the case of
spandrel with masonry arch and type B IMG strengthening.
The analysis of pier drifts has shown a roughly linear relationship between resisting force and pier
drift variation until the peak resistance is reached. The discrepancy between pier drifts considerably
increased in the post-peak softening range, but less quickly than subassemblage drift. This was not
the case of subassemblage S3a (case of spandrel with both masonry arch and RC bond beam) for
which the pier drift variation was 2.6 times the subassemblage drift at near-collapse.
The energy dissipation capacity was investigated by computing the hysteretic damping ratio of each
subassemblage under varying drift up to near-collapse conditions. The authors recommends to
assume a hysteretic damping ratio equal to 2% in the case of spandrel with masonry arch (regardless
of the presence of RC bond beam) and 4% in the case of spandrel with wooden lintel.
Finally, residual drifts were also assessed in order to derive some hints on their relation with
observed damage. In detail, the ratio between residual and maximum drift at near-collapse was
computed as a measure of the ability of URM walls to be repaired after severe in-plane lateral
Research Report
loading. The residual-to-maximum drift ratio can be set to 10%, 20%, and 5% in the case of spandrel
with wooden lintel, masonry arch, and both masonry arch and RC bond beam, respectively. When
the spandrel with masonry arch is strengthened by the type B IMG composite, the residual-to-
maximum drift ratio is expected to increase up to approximately 50%, denoting the inability of the
perforated wall to be repaired after a strong level of in-plane lateral loading.
3.11 REFERENCES
Augenti, N., Parisi, F. [2010] “Constitutive models for tuff masonry under uniaxial compression,” Journal of
Materials in Civil Engineering, Vol. 22, No. 11, pp. 1102-1111.
Augenti, N., Parisi, F. [2011] “Constitutive modelling of tuff masonry in direct shear,” Construction and Building
Materials, Vol. 25, No. 4, pp. 1612-1620.
Augenti, N., Parisi, F., Prota, A., Manfredi, G. [2011] "In-plane lateral response of a full-scale masonry
subassemblage with and without an inorganic matrix-grid strengthening system," Journal of Composites
for Construction, Vol. 15, No. 4, pp. 578-590.
Benedetti, D., Carydis, P., Pezzoli, P. [1998] “Shaking table tests on 24 masonry buildings,” Earthquake
Engineering and Structural Dynamics, Vol. 27, No. 1, pp. 67-90.
Chopra, A.K. [2006] Dynamics of structures, Pearson Prentice-Hall, Upper Saddle River, New Jersey, USA.
DM 20.11.1987 [1987] Norme tecniche per la progettazione, esecuzione e collaudo degli edifici in muratura e
per il loro consolidamento, Ministero dei Lavori Pubblici, Rome, Italy [in Italian].
D.M. 14.01.2008 [2008] Decreto Ministeriale del 14 gennaio 2008: Approvazione delle nuove norme tecniche
per le costruzioni, Ministero delle Infrastrutture e dei Trasporti, Supplemento ordinario alla G.U. n. 29 del
4 febbraio 2008, Rome, Italy [in Italian].
EN 1993 [1993] Eurocode 2: Design of concrete structures. Comité Européen de Normalisation, Brussels,
Belgium.
EN 1998-3 [2005] Eurocode 8: Design of structures for earthquake resistance – Part 3: General rules, seismic
actions and rules for buildings, Comité Européen de Normalisation, Brussels, Belgium.
FEMA 356 [2000] Prestandard and Commentary for the Seismic Rehabilitation of Buildings, American
Technology Council, Washington, DC, USA.
Magenes, G., Calvi, G.M. [1997] "In-Plane seismic response of brick masonry walls," Earthquake Engineering
and Structural Dynamics, Vol. 26, No. 11, pp. 1091-1112.
Pampanin, S., Christopoulos, C., Priestley, M.J.N. [2002] Residual deformations in the performance-based
seismic assessment of frame structures, ROSE Research Report No. 2002/02, IUSS Press, Pavia, Italy.
Parisi, F., Augenti, N. [2013] “Seismic capacity of irregular unreinforced masonry walls with openings,”
Earthquake Engineering and Structural Dynamics, Vol. 42, No. 1, pp. 101-121.
Parisi, F., Augenti, N., Prota, A. [2014] “Implications of the spandrel type on the lateral behavior of unreinforced
masonry walls,” Earthquake Engineering and Structural Dynamics, Vol. 43, No. 12, pp. 1867-1887.
Parisi, F., Iovinella, I., Balsamo, A., Augenti, N., Prota, A. [2013] “In-plane behaviour of tuff masonry
strengthened with inorganic matrix-grid composites,” Composites Part B: Engineering, Vol. 45, No. 1, pp.
1657-1666.
Parisi, F., Lignola, G.P., Augenti, N., Prota, A., Manfredi, G. [2011] “Nonlinear behavior of a masonry
subassemblage before and after strengthening with inorganic matrix-grid composites,” Journal of
Composites for Construction, Vol. 15, No. 5, pp. 821-832.
Priestley, M.J.N., Calvi, G.M., Kowalsky, M.J. [2007] Displacement-Based Seismic Design of Structures, IUSS
Press, Pavia, Italy.
Research Report
Ramirez, C.M., Miranda, E. [2012] “Significance of residual drifts in building earthquake loss estimation,”
Earthquake Engineering and Structural Dynamics, Vol. 41, No. 11, pp. 1477-1493.
Shing, P.B., Noland, J.L., Klamerus, E., Spaeh, H. [1989] “Inelastic behavior of concrete masonry shear walls,”
Journal of Structural Engineering, Vol. 115, No. 9, pp. 2204-2225.
Tomaževič, M. [2000] “Earthquake-Resistant Design of Masonry Buildings” In: Elnashai, A.S., Dowling, P.J.
(eds.) Series on innovation in structures and construction, Vol. 1, Imperial College Press, London, UK.
Vasconcelos, G., Lourenço, P.B. [2009] “In-plane experimental behavior of stone masonry walls under cyclic
loading,” Journal of Structural Engineering, Vol. 135, No. 10, pp. 1269-1277.
Research Report
4 QUASI-STATIC CYCLIC AND MONOTONIC TESTS ON MASONRY SPANDRELS AND

COMPOSITE SPANDRELS AT ETH ZÜRICH
Katrin Beyera, Alessandro Daziob
4.1 SUMMARY
This chapter presents the results of an experimental campaign comprising nine quasi-static tests on
spandrel elements. Four of the tested spandrel elements were masonry spandrels, which featured
either a masonry arch or a timber lintel. These spandrel elements are representative of spandrels in
old masonry structures that were constructed using solid clay bricks and timber floors. The other five
spandrel elements represent spandrel elements in modern masonry buildings where the masonry
piers are connected at the storey levels either by reinforced concrete (RC) ring beams or slabs.
These spandrel elements were constructed using modern hollow core clay bricks and consisted of
a masonry spandrel supported by a RC beam. This chapter summarises the spandrel configuration,
the applied loading history as well as the failure modes and hysteretic response of the spandrel
elements.
a Earthquake Engineering and Structural Dynamics Laboratory, École Polytechnique Fédérale de Lausanne (EPFL), 1015
Lausanne, Switzerland (formerly ETH Zürich).
b G. Dazio & Associati SA, Switzerland (formerly ETH Zürich).
4.2 Introduction
In unreinforced masonry (URM) walls the vertical piers are connected by horizontal spandrel
elements. Numerical simulations have shown that spandrels influence significantly the global wall
behaviour under seismic loading [Magenes and Della Fontana, 2998; Magenes, 2000]. Despite their
importance, experimental data on the cyclic behaviour of these spandrels is very scarce. For this
reason a research programme was initiated in which the cyclic behaviour of masonry spandrels was
investigated experimentally and numerically. The tests were carried out in 2009 and 2010 at the
Swiss Federal Institute of Technology Zurich (ETHZ). Within this programme different configurations
of masonry spandrels were tested under quasi-static monotonic and cyclic loading. The spandrel
configurations that were investigated included pure masonry spandrels and masonry spandrels that
also comprised a reinforced concrete (RC) beam (“composite spandrels”). The former represent
masonry spandrels in existing masonry buildings with timber floors. Many of such buildings were
damaged during the 2009 earthquake in L’Aquila, Italy and several of these featured dominant
spandrel failure. Figure 4.1 and Figure 4.2 show two of such examples whereas in the first building
the spandrels failed due to shear and in the second building due to flexure. The composite spandrels
represent spandrels in newer masonry buildings with RC slabs or RC ring beams; examples of such
buildings are shown in Figure 4.3.
This chapter presents and overview on the results of nine spandrel tests and is a summary of two
papers published in Earthquake Spectra [Beyer and Dazio, 2012a, 2012b].
Figure 4.1. Shear failure of spandrels with shallow masonry arches in an old URM building after the L’Aquila
earthquake on April 6, 2009: Entire building (left) and detail of a spandrel (right).
Figure 4.2. Flexural failure of spandrels in an old URM building after the L’Aquila earthquake on April 6,
2009: Entire building (left) and detail of a spandrel (right).
Research Report
Figure 4.3. Modern URM buildings in Switzerland with composite spandrels (left) and RC slabs (right).
4.3 Experimental Programme
The experimental programme comprised four tests on masonry spandrels and five on composite
spandrels. The masonry spandrels were named TUA-TUD while the composite spandrels carried
the labels TU1-TU5. The four test units representing masonry spandrels consisted each of two piers
and the spandrel element. They were constructed using solid clay bricks featuring dimensions of
250×120×60 mm3. The span of the spandrel was 1.18 m, and the height of the spandrel was
approximately 1.26 m. The width of the masonry wall was 0.38 m, which corresponds to one and a
half times the brick length and included a 10 mm wide head joint. The bricks were laid in an English
bond pattern on both faces of the wall. The two first masonry spandrels (TUA and TUB) had masonry
spandrels supported on timber lintels while the other two masonry spandrels (TUC and TUD)
consisted of spandrels with shallow masonry arches. A drawing of the test units and photos of the
spandrel details are shown in Table 4.1.
All masonry spandrel test units were subjected to a quasi-static cyclic loading; the exact loading
history is outlined in detail in Section 4.4. The masonry spandrels were also subjected to an axial
force. In TUA and TUC the axial force in the spandrel was provided by means of two horizontal D13
mm high strength steel rods. The rods were post-tensioned at the beginning of the test, and the force
in the rods was kept constant during the test by means of hollow core jacks connected to a load
follower (a device which keeps the oild pressure in the hollow core jacks constant throughout the
test). In real buildings, the axial force in the spandrels will depend on the axial elongation of the
spandrel and on the boundary conditions provided by steel ties and adjacent piers. The axial force
in the spandrel will therefore vary during seismic loading. To study the effect of a varying axial force
in the spandrel on the force-deformation characteristic of the spandrel, the axial load in the spandrels
of TUB and TUD was provided by means of two rods that were slightly pre-tensioned at the beginning
and then locked-in. Fixing the length of the rods implied an axial force increase upon axial elongation
of the spandrel. In TUB the axial force was provided by means of two D10 mm threaded rods, in
TUD two 32 mm rods were used instead. To simulate the gravity loads in the piers, the piers were
post-tensioned by high strength steel rods; the axial stress applied to the piers is summarised in
Table 4.1 and was kept constant throughout the test.
The composite spandrel test units had slightly different dimensions than the masonry spandrel test
units. All composite spandrel test units consisted of an RC beam, a masonry spandrel and a masonry
pier on either side of the spandrel (Table 4.2). All composite spandrel test units had the same
geometry. The parameters investigated included (i) the type of brick, (ii) the loading regime and (iii)
the longitudinal reinforcement content of the RC beam. Two of the spandrels were subjected to
monotonic loading (TU1 and TU2) and three to cyclic loading (TU3-TU5, Table 4.2). TU1 and TU2
differed concerning the type of masonry that was used. For the construction of TU1, bricks with
staggered longitudinal webs were used (Figure 4.4). TU2-TU5, on the contrary, were constructed
using bricks with continuous longitudinal webs. TU2 and TU3 were identical test units but subjected
to different loading regimes (monotonic versus cyclic loading). TU3-TU5 were all subjected to cyclic
loading but differed with regard to the longitudinal reinforcement content of the RC beam (Table 4.2).
TU4 had a longitudinal reinforcement that was almost twice as large as the reinforcement content of
TU3. The reinforcement of TU4 satisfied the minimum reinforcement content requirements for RC
ring beams according to the past Italian seismic design code OPCM n. 343 [2005]. Only the shear
reinforcement of D6 mm @ 150 mm was stronger than the required D6 mm @ 250 mm. According
to the European seismic design code, Eurocode 8 [EN 1998-3, 2005], a ring beam must be fitted
with a minimum longitudinal reinforcement content area of 2 cm2, which is considerably smaller than
the reinforcement content of TU1-TU4. For this reason a fifth unit was tested, featuring a beam with
a smaller longitudinal reinforcement area (TU5). Table 4.2 also summarises the axial stress applied
to the piers. Unlike the masonry spandrel test units, the composite spandrel test units were not
subjected to an axial force applied to the spandrel. Standard material tests on mortar, bricks,
masonry, concrete and reinforcing bars were carried out. The results of these material tests are not
included in this paper but can be found in the original sources [Beyer and Dazio, 2012a, 2012b].
Table 4.1. Masonry spandrels: Loading scheme, spandrel type and details of the axial load application for
the four test units.
Axial stress in
Test Unit Loading Spandrel type Axial force in spandrel
piers
Constant, first 80 kN,
TUA Cyclic Timber lintel 0.33 MPa
then 40 kN
Variable, plain bar with
TUB Cyclic Timber lintel 0.33 MPa
low axial stiffness
TUC Cyclic Masonry arch 0.43 MPa Constant (80 kN)
Variable, plain bar with
TUD Cyclic Masonry arch 0.43 MPa
high axial stiffness
Research Report
Table 4.2. Composite spandrels: Loading scheme, spandrel type and details of the axial load application for
the four test units.
Longit.
Shear reinf.
Test Unit Loading reinf. of RC Axial stress in piers
of RC beam
beam
4 D12 mm
TU1 Monotonic D6@150 mm 0.4 MPa
(4.52 cm2)
4 D12 mm
TU2 Monotonic D6@150 mm 0.4 MPa
(4.52 cm2)
4 D12 mm
TU3 Cyclic D6@150 mm 0.4 MPa
(4.52 cm2)
4 D16 mm 0.4 MPa
TU4 Cyclic D6@150 mm
(8.04 cm2) (0.6 MPa1)
4 D10 mm
TU5 Cyclic D6@150 mm 0.4 MPa
(3.14 cm2)
1 Axial stress in piers was increased in final cycles.
Figure 4.4. Composite spandrels: Brick Type 1 with non-continuous webs (left) and Brick Type 2 with
continuous webs (right).
4.4 Test Setup and Loading History
In the test stand the test unit stood on two stiff steel beams (“lever beams”) that were supported on
hinges at the centre line of the piers and connected to servo-hydraulic actuators at their ends. During
testing the two servo-hydraulic actuators were moved with the same velocity in opposite directions.
As a result, the two horizontal lever beams rotated and the piers right and left to the spandrel were
subjected to the same drifts, which caused the demand on the spandrel. The support of the South
lever beam allowed next to a rotation also a sliding movement along the longitudinal axis of the
beam. Hence, the test stand did not restrain the axial elongation of the spandrel. The test stand
differed to those adopted by Gattesco et al. [2008] and Graziotti et al. [2009]. A comparison of the
different test stands can be found in [Beyer and Dazio, 2012b].
The test setup for the masonry and composite spandrel test units differed with respect to the span
between the two supports of the lever beam (3.28 m and 3.60 m, respectively). Drawings and photos
of the test setups are shown in Figure 4.5 to Figure 4.8. Different local and global quantities were
measured during testing by means of hard-wired and optical measurements. Details on the
instrumentation can be found in [Beyer and Dazio, 2012a, 2012b; Beyer et al., 2010].
Figure 4.5. Masonry spandrels: Drawing of the test setup. View from the East without side restraint
(dimensions in mm; LF = Hollow core jacks connected to load follower, which keeps the oil pressure
constant).
Figure 4.6. Composite spandrels: Drawing of the test setup. View from the East without side restraint
(dimensions in mm).
Both the monotonic and cyclic loading scheme followed different steps of storey drifts with
displacement control. The storey drift was defined as the average inclination of the two piers. When
Research Report
the monotonic loading scheme was applied, loading was stopped at the following storey drift levels
nom: 0.025%, 0.05%, 0.1%, 0.2%, 0.3%, 0.4%, 0.6%, 0.8%, 1.0%, 1.5%, 2.0%, 2.5%, 3.0%. The
same drift levels determined the amplitude of the cyclic loading scheme. At each drift level, the test
unit was subjected to two cycles (Figure 4.9).
Figure 4.7. Masonry spandrels: Photo of the test setup. View from the North-East.
Figure 4.8. Composite spandrel: Photo of the test setup. View from the East.
Figure 4.9. Loading history for quasi-static monotonic and cyclic spandrel tests.
4.5 Hysteretic Behaviour and Failure Modes
4.5.1 Masonry Spandrels

Figure 4.10 shows the crack pattern of the four masonry spandrels at failure. While TUA and TUD
developed mixed shear-flexural cracking patterns, the behaviour of TUB was governed by flexural
cracks at the spandrel ends and TUC developed mainly diagonal shear cracks.
For TUA the development of the crack patterns in the two loading directions differed. For the positive
direction of loading, the spandrel first developed a flexural mechanism with vertical cracks at both
ends of the spandrel, which reached around the timber lintel. For the negative direction of loading
(Figure 4.10a), most of the deformations were absorbed by a large stair-stepped shear crack, which
crossed the entire spandrel. As loading continued, such a diagonal crack developed also for the
positive direction of loading. The final crack pattern of the spandrel corresponded therefore to a
mixed flexure-shear failure with vertical cracks at the end of the spandrel and shear cracks in the
spandrel. The final failure of the test unit occurred for a drift of 3.0% and was associated with the
failure of the supports of the timber lintel in the piers, which rotated inwards due to the horizontal
shear stresses caused by the pull-out forces of the timber lintel.
TUB was the first test unit to be tested for which the axial load in the spandrel depended on the axial
elongation of the spandrel. Similar to TUA, the first cracks that developed were flexural cracks at
both ends of the spandrel reaching around the ends of the timber lintel (Figure 4.10b). The cracks
were rather vertical and passed through several bricks indicating a flexural failure of the spandrel.
As a result of the opening of the cracks and the elongation of the spandrel, the force in the horizontal
rods increased. The test unit failed just before reaching nom = –1.5% for the first time. Almost
simultaneously the North support of the lintel failed and the West horizontal rod fractured.
Research Report
TUC featured a shallow masonry arch instead of the timber lintel. Similar to TUA the axial force in
the spandrel was kept constant throughout the test but it was doubled to create a failure mode which
was different to that of TUA. For both directions of loading the diagonal shear cracks were the widest
cracks, hence indicating a shear failure of the spandrel (Figure 4.10c). Two hinges formed in the
arch, which were located at approximately 1/3 and 2/3 of the span. As loading continued the damage
to the arch increased with each cycle. The final failure occurred at a drift of nom = –3.5% due to the
collapse of the arch.
For TUD a new system concerning the horizontal rods was designed. In principle, the setup was
similar to that of TUB, i.e. the forces in the horizontal bars were not kept constant by means of a load
follower but varied according to the axial elongation of the spandrel. For TUD, however, the
horizontal bars consisted of D32 mm Dywidag bars, which were considerably stiffer than the D10
mm threaded rods used for TUB. For both directions of loading the damage to the spandrel was
concentrated in the upper half of the spandrel while the damage to the arch was at this point still
rather limited (Figure 4.10d). This was probably due to the eccentric horizontal rod, which for drift
demands larger than 1.5% led to a disaggregation of the top part of the spandrel. Starting with the
cycles with nom = ±1.0%, the spandrel and the arch were increasingly damaged and softened. The
integrity of the arch suffered in particular due to horizontal cracks running through the bricks of the
arch. As a consequence of these horizontal cracks, every time the loading direction was reversed
some lower parts of the bricks fell down. At the end of the test the front row of the central part of the
arch collapsed.
Figure 4.11 and Figure 4.12 show the hysteresis curves for the shear force and the axial force in the
spandrels, respectively. All masonry spandrels reached their peak resistance at about the same
rotation demand. Until the peak resistance was reached, the spandrel remained largely uncracked.
After cracking of the spandrel, the shear capacity of the spandrel dropped considerably due to the
formation of large shear or flexural cracks in the spandrel. Once the peak shear force was overcome,
the spandrel had the tendency to elongate and the axial restraint of the spandrel played an important
role on its overall behavior. The shape of the envelope of the shear force hysteresis depended on
the mode of application of the axial load in the spandrel. For TUA and TUC, for which the axial load
in the spandrel was kept constant by means of a load follower, the capacity of the spandrel dropped
with increasing drift demand.
For TUB and TUD, for which the axial load in the spandrel depended on its axial elongation, an
increase in capacity could be observed after the initial drop in strength due to the formation of the
first cracks in the spandrel. This increase in shear strength was directly related to the increase in
axial force in the spandrel (Figure 4.12a). As long as the spandrel remained uncracked, the force in
the horizontal rods did not change and was virtually equal to the initial post-tensioning force, which
was applied to straighten the horizontal bars. As soon as the spandrel cracked, the force in the
horizontal bars increased considerably. The thick grey lines in Figure 4.12b indicate the yield and
tensile strength of the two threaded D10 mm rods, which were used to apply the horizontal force to
TUB. For the cycles at nom = ±1.5% the yielding of the horizontal bars is clearly visible. At the end
of the test, one of the horizontal bars fractured and the horizontal force dropped to zero. The bars
used for applying an axial force to the spandrel of TUD were so strong that they remained elastic
throughout the test. The increase in axial force with rotation was almost linear up to a rotation of nom
= ±1.0% (Figure 4.12d). For larger rotations, the hysteresis loops of the axial load–rotation
relationship became fatter indicating a strong damage to the masonry spandrel. The axial force
dropped for cycles with rotations larger than ±1.5%. For these cycles the damage concentrated more
and more in the upper part of the spandrel and in the masonry arch below the horizontal rods. As a
consequence, the axial elongation at the height of the rod was smaller than in previous cycles.
Figure 4.10. Masonry spandrels at failure.

Research Report
Figure 4.11. Masonry spandrels: Spandrel shear–force deformation hysteresis.

Figure 4.12. Masonry spandrels: Hysteresis curves of the axial loads in the spandrels.
4.5.2 Composite Spandrels

Out of the five composite spandrels, TU1 showed a very different crack pattern than the others:
vertical cracks at both ends of the spandrel extended over the entire height of the spandrel and ran
almost straight through the bricks (Figure 4.14). Up to failure, the body of the masonry spandrel
remained virtually undamaged and all the deformation demand on the masonry spandrel was
absorbed by the two flexural cracks. Due to the staggered webs (Figure 4.4), the horizontal
compression strength and the tensile strength of the bricks was very small. Very early during the test
the masonry spandrel separated from the RC beam beneath it. Plastic hinges formed at both ends
of the beam, leading eventually to failure due to the rupture of longitudinal reinforcing bars.
The crack pattern of TU2 at failure was characterised by two major stair-stepped cracks (Figure
4.15a). At the beginning of the test, the bricks of TU2 remained largely undamaged and the cracks
followed the joints. As the deformation demand increased, crushing of the compression diagonal
Research Report
occurred, leading to a Southward shift of the positive plastic hinge. The final failure of the test unit
was caused by the rupture of longitudinal bars in both the negative and the positive plastic hinges.
TU3 was the first composite spandrel test unit subjected to cyclic loading. When compared to TU2,
which had been subjected to monotonic loading, it appeared that all bricks of the spandrel of TU3
were disjointed and the deformations more equally distributed over the spandrel (Figure 4.15b). For
the RC beam, the cyclic behavior led to a larger growth in length because the plastic strains in the
longitudinal reinforcing bars accumulated cycle after cycle. As a consequence, at the same drift
demand, the crack widths in the RC beam were larger for TU3 than for TU2. The shear reinforcement
in the RC beam was not sufficient to carry the entire shear force of the beam. As the contribution of
the concrete to the shear capacity decreased below a certain limit, a shear failure occurred within
the positive plastic hinge during the first cycle with nom = ±3%. When the loading was reversed
thereafter, a longitudinal bar fractured within the positive plastic hinge.
The RC beam of TU4 was considerably stiffer and stronger than that of TU3 and forced a gap to
open up between the bottom of the RC beam and the pier, which reduced the deformation demand
on the RC beam itself. Only one of the two plastic hinges formed for each loading direction. At the
end of the cycles with a rotation of nom = ±2.0%, the axial stress in the piers was increased from 0.4
MPa to 0.6 MPa. Consequently, the gap between the pier and RC beam did not open up as much
and the RC beam had to deform more. This increased deformation demand on the RC beam led to
the yielding of the positive plastic hinge, producing shear failure of the RC beam in the second cycle
(Figure 4.15c).
The RC beam of TU5 had a weaker longitudinal reinforcement than the first four test units. For TU5,
the horizontal joint between pier and RC beam hardly opened up and the extent of cracking of the
RC beam into the piers was also reduced. The test unit failed due to rupture of the longitudinal
reinforcement (Figure 4.15d).
Although TU1 and TU2 had different masonry properties and exhibited different masonry behaviours,
the force–rotation relationships of TU1 and TU2 were very similar (Figure 4.13a). For both test units,
the initial elastic phase was followed by a jagged branch, which was associated with the formation
of cracks in the masonry spandrel and the RC beam. The onset of yielding in the longitudinal
reinforcement of the RC beam caused the force–rotation curve to round off. It stayed level for a small
range of rotations and then began to drop. This reduction in capacity starting at a rotation of 1% to
1.5% was associated with the crushing of the compression diagonal. The peak capacities of TU3,
which had been subjected to cyclic loading, exceeded the curve of TU2 up to a drift of 1.5%, most
likely because of the slightly higher mortar strength of TU3. TU2 and TU3 were built at the same
time. Because TU2 was tested first, the mortar strength on the day of testing was slightly less than
the mortar strength of TU3 on the day of testing [Beyer and Dazio, 2012a]. For drift demands larger
than 1.5%, the reduction in capacity was stronger for TU3 than for TU2 – most likely due to a faster
degradation of the compression strut owing to the cyclic loading.
The force–deformation hysteresis of TU4 differed significantly from that of TU3. During the cycles
with nom = 0.4 MPa, the hysteresis loops were almost bilinear elastic (Figure 4.13c) due to the
rocking of the RC beam on the piers according to the mechanism described in the previous section.
When the axial stress in the piers was increased to  = 0.6 MPa, the maximum shear force in the
spandrel increased and the longitudinal reinforcement of the RC beam yielded leading to fatter
hysteresis loops.
The RC beam of TU5 had the smallest longitudinal reinforcement (4 D10 mm) of all the test units. It
is therefore likely that the stresses in the compression diagonal of the spandrel were smaller and the
crushing of the masonry played only a minor role. As a consequence, the drop in the capacity after
the peak was reached was less significant (Figure 4.13d). It is also likely that the relative contribution
of the masonry spandrel to the resistance of the spandrel element might therefore have been larger
leading to the more pinched behaviour of the hysteresis curves with stiffness recovering at the end
of the cycle, which is typical for a behaviour dominated by shear.
Figure 4.13. Composite spandrels: Spandrel shear–force deformation hysteresis.

Research Report
Figure 4.14. Composite spandrel TU1 at failure (Brick Type 1).
Figure 4.15. Composite spandrel TU2-5 at failure (Brick Type 2).

4.6 Conclusions
The four tests on masonry spandrel test units and the five tests on composite spandrel test units
yielded insights into the failure mechanisms of spandrels that are subjected to seismic loading. The
gained data allows one to develop and to validate numerical and mechanical model of such spandrel
elements. For the masonry spandrel, for example, mechanical strength models predicting the peak
and residual strength of masonry spandrels have been recently published [Beyer, 2012] and
numerical models have been validated [e.g. Mangalathu and Beyer, 2014]. Similar efforts that
address strength and deformation capacity of composite spandrels are underway. The final objective
is to define piece-wise linear envelopes of the force–rotation relationships of masonry and composite
spandrels that can be easily implemented in equivalent frame models of existing and new masonry
structures.
Research Report
4.7 REFERENCES
Beyer, K. [2012] “Peak and residual strengths of masonry spandrels,” Engineering Structures, Vol. 41, pp. 533-
547.
Beyer, K., Dazio, A. [2012a] “Quasi-static monotonic and cyclic tests on composite spandrels,” Earthquake
Spectra, Vol. 28, No. 3, pp. 885-906.
Beyer, K., Dazio, A. [2012b] “Quasi-static cyclic tests on masonry spandrels,” Earthquake Spectra, Vol. 28,
No. 3, pp. 907-929.
Beyer, K., Abo-El-Ezz, A., Dazio, A. [2010] “Quasi-static cyclic tests on different types of masonry spandrels,”
Report No. 327, Institute of Structural Engineering, ETH Zürich, Switzerland.
Gattesco, N., Clemente, I., Macorini, L., Noè, S. [2008] “Experimental investigation of the behavior of spandrels
in ancient masonry buildings,” Proceedings of the 14th World Conference on Earthquake Engineering,
Beijing, China.
Graziotti, F., Magenes, G., Penna, A. [2009] “Progetto di una sperimentazione su elementi di fascia muraria,”
ReLUIS Project Deliverable 4.3-UR01-1, University of Pavia and EUCENTRE, Pavia, Italy [in Italian].
OPCM n. 343 [2005] Ulteriori modifiche ed integrazioni all’OPCM n. 3274 del 20 marzo 2003 recante «Primi
elementi in materia di criteri generali per la classificazione sismica del territorio nazionale e di normative
tecniche per le costruzioni in zona sismica», Supplemento Ordinario n. 85 alla G.U. n. 107 del 10 maggio
2005, Ordinanza del Presidente del Consiglio dei Ministri del 3 maggio 2005, Rome, Italy [in Italian].
Magenes, G. [2000] “A method for pushover analysis in seismic assessment of masonry buildings,”
Proceedings of the 12th World Conference on Earthquake Engineering, Auckland, New Zealand.
Magenes, G., Della Fontana, A. [1998] “Simplified non-linear seismic analysis of masonry buildings,”
Proceedings of the British Masonry Society, No. 8, pp. 190-195.
Mangalathu, S., Beyer, K. [2014] “Numerical study on the force-deformation behaviour of masonry spandrels
with arches,” Journal of Earthquake Engineering, Vol. 18, No. 2, pp. 169-186.
Research Report
5 SPANDREL TESTING A THE UNIVERSITY OF AUCKLAND

Charlotte L. Knoxa, Jason M. Inghama
5.1 SUMMARY
Six full-scale unreinforced masonry (URM) substructures, constructed of two piers with a connecting
spandrel, were built in the laboratory and subjected to in-plane pseudo-static cyclic loading. The
objective of this testing was to acquire experimental data regarding the influence of the spandrel
geometry and relative strength on the seismic behaviour of coupled piers, in the context of a
perforated URM wall, with the design of the six pier-spandrel substructures selected to identify failure
patterns and key performance characteristics that define the seismic behaviour of in-plane perforated
URM walls. Subsequently a two-storey perforated unreinforced masonry wall was constructed at half
scale and subjected to in-plane pseudo-static cyclic loading. The purpose of this latter test was to
address the comparative absence of experimental results associated with system level response of
multi-storey perforated URM walls loaded in-plane, and more precisely, the influence of the
connecting spandrel on the behaviour of multi-storey piers.
a University of Auckland, 368 Khyber Pass Newmarke, Auckland 1023, New Zealand.
5.2 Introduction
Current assessment methods for establishing the in-plane strength of perforated URM walls can be
divided into two groupings. The first grouping involves delineation of a wall into two elements, being
the vertical piers which are the primary lateral load resisting elements, and the horizontal spandrels
which couple the piers and provide secondary lateral resistance [NZEE, 2006; ASCE/SEI41-06,
2007]. The second grouping involves treatment of the perforated wall as an equivalent frame with
pier, spandrel and joint elements [EN 1998-3, 2005; D.M. 14.01.2008, 2008], where the joints are
considered to be rigid elements and the spandrels are modelled as rotated piers. Whilst extensive
experimental research has been undertaken to investigate the in-plane seismic response of piers,
few researchers have considered the non-linear seismic response of spandrel panels. The behaviour
of the spandrel is fundamental in defining the boundary conditions of the piers, with these boundary
conditions in turn governing the available shear strength of the piers, and therefore the shear
strength of the masonry frame. Consequently the experimental programme reported herein aimed
to provide valuable experimental data on masonry spandrels having different configurations.
The 2011 Christchurch earthquake highlighted the problem with assuming that spandrels are
effectivly rigid, and that all the deformation in an unreinforced masonry frame occurs in the piers.
Figure 5.1a shows extensive spandrel failures in Avonmore House, Christchurch. The piers in the
façade of Avonmore House appear to have responded in a rocking mode, as the characteristic ‘X’
cracks of diagonal failure are evident in the spandrels and horizontal cracks were observed at the
base of the piers on the ground floor. Figure 5.1b shows the opposite occuring, where diagonal shear
failure occurred in the piers at levels one, two, three and four, located both in the interior piers which
had a low aspect ratio, as well as in the high aspect ratio piers that were located one column in from
the exterior.
(a) (b)
Figure 5.1. Failure patterns from the Canterbury earthquakes: (a) Spandrel failure in Avonmore House,
Christchurch and (b) pier failure in Kenton Chambers, Christchurch.
5.2.1 Substructure Testing

Six substructures were constructed with varying pier aspect ratios and with varying spandrel depths.
Each tested substructure consisted of two piers separated by an opening, with a spandrel laterally
connecting the top of the piers. These substructures were designated PS1-PS6 and the justification
for the selected geometric configurations of each substructure is presented below.
Research Report
Dimensions of the six substructures are listed in Table 5.1. Figure 5.2 shows the typical geometry
and layout of the substructure test units, with pier height defined as per an equivalent frame model
for the sole purpose of discussing geometry.
Table 5.1. Substructure dimensions.
Wall Thickness Pier A Pier B Spandrel Opening

t
[mm] Hp-A [mm] Lp-A [mm] Hp-B [mm] Lp-B [mm] Ls [mm] Ds [mm] Lo
[mm]
PS1 230 1795 1190 1795 1190 4420 590 1240

PS2 230 1795 1190 1795 710 4420 590 1740
PS3 230 1795 1190 1795 1190 4420 940 1240
PS4 230 1795 1190 1795 1190 4420 1260 1240
PS5 230 1795 1190 1795 1190 4420 940 1240
PS6 230 1280 1190 1280 1190 4420 1450 1240
Ls
t
380 Lp - A Lo Lp - B 426
Strong Wall
Ds
Spandrel
Hp - A
Hp - B
Pier A Pier B
812
Strong Floor Strong Floor
Figure 5.2. Typical geometry and layout of substructure for pseudo-static test: Side elevation (left) and end
elevation (right).
5.2.2 Testing Matrix

The test configuration was selected in order to experimentally investigate the influence of spandrel
geometry on the response of a pier/spandrel substructure. Three spandrel depths were investigated,
being 590 mm (shallow), 940 mm (average) and 1450 mm (deep). The effect of spandrel aspect ratio
was investigated by varying the spandrel depth between three otherwise identical substructures
(PS1, PS3 and PS4). Two forms of comparison were used to investigate the influence of pier aspect
ratio; firstly with one substructure having piers of different aspect ratio (PS2), and secondly by
comparing two substructures where the spandrel depths were different but the overall height of the
substructures was maintained, so that the pier height when measured to the bottom of the spandrel
differed between the two substructures (PS5 and PS6). The role of axial stress level was also
considered, with two stress levels applied, being approximately 0.2 MPa (low) and 0.5 MPa (high),
see Table 5.2. These axial stress levels represented piers on the top storey of a perforated wall (≈
0.2 MPa), and piers having two stories of masonry construction above (≈ 0.5 MPa). Details of the
pseudo-static testing matrix for the six substructures are presented in Table 5.2.
Table 5.2. Pseudo-static testing matrix.
Wall
Spandrel Axial Pier aspect ratio Spandrel Total height
depth stress Hp/L p aspect ratio [mm]
[MPa] Ds/Ls
PS1 Shallow 0.52 1.5 0.48 2385
PS2 Shallow 0.36/0.58 1.5/2.5 0.34 2385
PS3 Average 0.48 1.5 0.76 2740
PS4 Deep 0.50 1.5 1.17 3245
PS5 Average 0.22 1.5 0.76 2740
PS6 Deep 0.23 1.07 1.17 2740
5.2.3 Construction Details

The substructure test units were constructed by an experienced mason under supervision. Two-leaf
(wythe) walls are considered to be the most commonly occurring thickness for the bottom storey of
one and two storey New Zealand URM structures [Russell and Ingham, 2008], and consequently
two-leaf thick substructures were constructed. Common (American) bond employs header bricks
located in every 4th course, and has been found to be the most prevalent bond pattern used in New
Zealand [Russell and Ingham, 2008], hence the Common bond pattern was used to simulate a
section from the front façade of a multi-storey URM structure. The mortar was a 1:2:9
(cement:lime:sand) mix, corresponding to ASTM type ‘O’ mortar with nominally 10 mm thick mortar
joints. This mortar mix was found to be representative of the current strength of historical mortar
[Lumantarna, 2012].
The substructures were intentionally constructed in a way that replicated the observed, and often
deteriorated, finished quality of walls in real URM buildings. Testing of the specimens took place
approximately 28 days after construction of the specimen was complete. The curing time allowed
the mortar to reach its target compression strength.

All substructures were constructed using recycled vintage clay bricks obtained from a demolished
building. The original mortar was removed and the surfaces of the bricks were prepared for new
mortar before being reused in the test substructure. The bricks were estimated to be 80 years old as
the demolished building was known to have been constructed in the 1930’s. These vintage bricks
were required because the manufacturing processes for making new bricks are sufficiently different
to substantially alter the brick properties and characteristics. In particular, the difference in porosity
between currently manufactured bricks and old bricks results in the bond at the brick-mortar interface
being much weaker in new bricks using the mortar required for this test. Recognising that within a
building there is significant variability in brick properties, the reuse of vintage bricks introduced
realistic material variability into the test.
Random samples of bricks were taken during construction and tested in compression [ASTM C67-
11, 2011]. Prisms were also built during construction of the wall using randomly selected bricks, and
tested in compression and flexural tension. All material tests were conducted according to ASTM
standards [ASTM C1314-11, 2011; ASTM C109/C109M-11, 2011; ASTM E1111-04, 2004; ASTM
E519/E519M-10, 2010]. The results of these tests are shown in Table 5.3. A high coefficient of
variation for each property, compared to what is reasonable for concrete material tests for example
(10-15%), indicates a large variability in the materials used. This large variability was considered
acceptable because of the real variation in material properties in existing URM buildings.
Research Report
Table 5.3. Substructure material properties.
Value Mortar Brick Masonry Masonry Cohesion c Coeff.

compressive compressive compressive bond [MPa] of friction
strength fʹj strength fʹb strength fʹm strength fʹfb 
[MPa] [MPa] [MPa] [MPa]
Mean 2.9 25.4 9.2 0.08 0.3 0.7

COV 41% 19% 26% 29% 15% 16%
ASTM 2011 2011
Method ASTM ASTM
C109/C109M- ASTM C67-11 NZSEE NZSEE
of Test C1314-11 1072-05
11 guideline guideline
A weak mortar mix, being ASTM type ‘O’, was selected to simulate weather deteriorated mortar in
heritage URM buildings as recommended from research by Lumantarna [2012]. Portland cement
was becoming more widely available in the early part of the 20th Century when URM was the building
material of choice in New Zealand, as was hydrated lime, hence Standard Portland cement, hydrated
lime (Calcium Hydroxide) and river sand were used in the mortar.
5.2.5 Test Setup

All substructures were tested as shown in Figure 5.3. The bottom masonry course of each pier was
seated one course down into the large concrete bases, which were bolted onto the strong floor.
Setting the piers into the concrete bases prevented sliding along the artificially smooth strong floor
when horizontal load was applied to the substructure, but allowed for a sliding shear failure to
potentially occur in the bed joint between the first and second course of bricks.
Push Pull
Stressing Tendons
Load Cell Load Cell

Rollers
Box Beam
Steel reaction plate
Steel channel Hydraulic
actuator
Test Wall
Displacement
Measurement
Strong Wall
Ground
Restraint
Strong Floor
Strong Floor
Spring Spring
Loading Plan Side Elevation
Figure 5.3. Pseudo-static testing set-up: Side elevation (left) and end elevation (right).
The horizontal shear force was applied evenly over the vertical end of the spandrel using a hydraulic-
powered actuator, which reacted against the laboratory strong wall. Flat steel plates measuring
10 mm thick × 230 mm wide × 500 mm long were positioned on the top of the spandrel at the centre-
line of the piers, to provide a smooth horizontal rolling surface for a set of rollers which were then
placed on top of the steel plates. The rollers were positioned to reduce any friction transfer of the
horizontal load between the fabricated steel sectionb and the spandrel. A steel loading beam was
seated on the rollers. A 20 mm thick steel plate was welded on each end of the steel section so that
the horizontal load could be transferred from the actuator into the spandrel directly. Flybraces
mounted on the steel channel were designed to strengthen the connection between the endplate
b Fabricated section was an I section with d = 220 mm, bf = 250 mm, tf = 10 mm, tw = 8 mm.
and the beam so that the endplate did not deform, ensuring that the horizontal load was fully
transferred into the test substructure. The loading beam was 50 mm longer than the test unit, and
was not fixed to the test unit, and therefore contributed no strength to the spandrel.
The unbonded post-tensioning tendons positioned on the outside of the wall were used to apply the
axial load to the top of the wall. Springs were connected in series with the tendons to help eliminate
additional stress being applied to the test unit if rocking induced uplift of the piers was to occur. An
independent frame was positioned at one end of the wall, against which displacement of the spandrel
in the horizontal direction was measured, which eliminated any effects from flexing of the strong-
wall.
5.2.6 Testing Procedure

The cyclic loading history that was adopted for all substructure tests is shown in Figure 5.4. This
displacement-controlled pseudo-static procedure was employed to capture the non-uniformly
accumulated damage in the wall, and to enable observations of damage and failure mechanisms. It
is recognised that dynamic tests simulate with more accuracy the seismic actions on a structure,
however pseudo-static tests enable more accurate measurements of forces and displacements and
allow the damage evolution to be documented more easily [Calvi et al., 1996].
50 2.0%
45
40
1.5%
35
30
Actuator displacement (mm)
25 1.0%
20
15
0.5%
10
5
Drift
0 0.0%
-5
-10
-0.5%
-15
-20
-25 -1.0%
-30
-35
-1.5%
-40
-45
-50 -2.0%
0 5 10 15 20 25 30 35 40 45 50
Displacement cycles
Figure 5.4. Loading history for testing of the pier/spandrel substructures.
Failure was defined as a measured reduction in lateral strength of 20% or more when compared to
the maximum lateral strength previously recorded during testing. In substructures where a flexural
response dominated, a 20% strength reduction was typically not attained as a flexural response is
characterised by minimal strength loss with increasing displacement until ultimate drift levels are
developed, and therefore for this condition the testing was instead terminated when either the data
acquisition or the loading actuator reached their accuracy limits, or when stability of the substructures
was affected by twisting in the out-of-plane direction.

The initial response of all test units was almost identical, with cracks initiating in the top left corner
(Pier A side) of the opening for the pull cycle and in the top right corner (Pier B side) for the push
cycle. In all test units, a failure (either shear or mixed mode) of the spandrel element pre-empted
any discernible cracking within the pier element. In the test units where the pier aspect ratio was 1.5
and the spandrel depths were either average or deep, the spandrel failure mode was diagonal
tension, indicated by the characteristic ‘X’ cracks as shown in Figure 5.6b and Figure 5.7 for PS4
and PS5, respectively. Crack propagation generally followed the path of the mortar head and bed
joints, occurring along the brick-mortar interface, as is common in unreinforced masonry where the
ratio of the mortar compressive strength to the brick compressive strength is low.
A test error occurred prior to the loading of substructure PS4 due to a malfunction in the loading
actuator during set-up, causing a large push force to be applied which caused a diagonal crack to
Research Report
form in the spandrel. The crack immediately extended from the top of the spandrel above the joint-
spandrel interface on the Pier B side and stepped diagonally downward, extending along the brick-
mortar interface until it reached the top left corner of the opening. No crushing of the masonry at the
compressed toe of the rocking piers was noted.
Figure 5.8 shows the crack patterns that were evident at the conclusion of testing. The images all
show Pier A on the left and Pier B on the right, with the Push load (positive force) applied to the
centre of the right end of the spandrel.
(a) (b)
Figure 5.5. (a) Spandrel cracking in all bed joints of PS2 and (b) Rocking at base of Pier B in PS2.
(a) (b)
Figure 5.6. (a) Rocking at the base of pier B in PS3 and (b) Development of the diagonal cracks in the final
push cycle of PS4.
Figure 5.7. Rotation in spandrel of PS5.
General results are presented in Table 5.4, where Vmax is the maximum lateral force as recorded by
the load cell, DVmax is the lateral displacement (measured at the top of the spandrel) associated with
Vmax, max is the associated drift expressed as a percentage, Vcrack is the lateral force when cracking
was first observed, θcrack is the drift corresponding to Vcrack, and d80 and θ80 are the ultimate lateral
wall displacement and drift, respectively, corresponding to the point at which the lateral force had
degraded to 80% of Vmax. The effect of testing substructure PS4 in its pre-cracked condition in the
push cycle is evident by the approximately 30 kN difference between the maximum lateral force
recorded in the push and pull directions.
Table 5.4. Experimental results for pseudo-static testing of substructures.
Wall Vmax DVmax max Vcrack Dcrack crack D80 80

[kN] [mm] [%] [kN] [mm] [%] [mm] [%]
PS1(+) 88.9 14.94 0.626 34.5 0.95 0.04 17.3 0.73

PS1(-) -96.3 -15.0 0.629 -37.9 -1.02 0.042 >15 >0.63
PS2(+) 60.98 18.74 0.786 27.2 2.52 0.11 > 25 > 1.19
PS2(-) -55.52 -24.93 1.045 -25.86 -2.14 0.090 >25 >1.19
PS3(+) 102.8 12.15 0.443 44.4 1.95 0.071 > 24 > 0.87
PS3 (-) - -23.74 0.867 -59.0 -1.69 0.061 >23.7 > 0.87
104.90
PS4(+) 70.36 10.70 0.329 - - - > 26.4 > 0.81
PS4(-) -99.67 -25.81 0.795 -68.40 -4.208 0.129 > 25.8 > 0.80
PS5(+) 75.3 40.26 1.47 35.6 2.04 0.0745 > 40.2 > 1.47
PS5(-) -59.95 -38.18 1.39 -28.39 -2.39 0.087 > 38.2 > 1.39
PS6(+) 33.85 5.94 0.217 25.1 1.12 0.041 14.7 0.538
PS6(-) -26.88 -16.44 0.600 -24.30 -0.971 0.035 > 16.5 >
0.602
For the substructures tested, a limited elastic range of response was noted, with the average drift at
which cracking occurred being 0.09%. Cracking at the base of the piers was visible at an average
drift of 0.40% in all test units excluding PS6. For test unit PS6, initial cracking was observed at a
displacement of 1.12 mm (drift of 0.04%) with a corresponding force of 25.1 kN.
The ultimate drift was measured to be between 0.63% and >1.47% for the substructures with rocking
piers, and was approximately 0.6% for substructure PS6 where the overall response was governed
by shear failure of Pier A.
5.2.8 Deformation Modes

As discussed previously, extensive diagonal cracking occurred within the spandrel panel of those
substructures that had ‘average’ or ‘deep’ spandrel depth, whereas for the two substructures that
had a shallow spandrel depth the lateral loading resulted in a spread horizontal crack pattern in the
spandrel. Figure 5.9 shows photographically the two distinct crack patterns that were observed at
the final stages of testing. The failure modes of the spandrel can be described by three distinct
modes, as shown in Figure 5.10.
Research Report
(a) (b)
(c) (d)
(e) (f)
Figure 5.8. Crack patterns in substructures after cyclic pseudo-static testing (Pier A on the left): (a) PS1,
(b) PS2, (c) PS3, (d) PS4, (e) PS5 and (f) PS6.
Mode A (Figure 5.10a) is representative of a flexural response where the spandrel panel acts as
a rigid block effectively rocking between the two rigid joint panels. Cracking at the spandrel/joint
interface was noted as the initial crack location in the average and deep spandrels (substructures
PS3, PS4, PS5, and PS6), signifying that deformation Mode A occurred first. The initial flexural
crack only propagated vertically for the depth of one brick as the bond pattern resulted in a
discontinuous vertical mortar joint. This observation is in agreement with proposals for determining
the flexural strength of spandrels, which identify the interlocking phenomena as a source of flexural
strength. The interlocking originates at the interface between the end section of the spandrel and
the joint, due to the contiguous masonry [Cattari and Lagomarsino, 2008].
Mode B (Figure 5.10b) was the most frequently observed failure mode, with stair-stepped diagonal
cracking following the mortar joints at increased displacement amplitudes (Figure 5.9b).
(a) (b)
Figure 5.9. Spandrel failure modes from experimental testing: (a) Rotation in the spandrel of PS2 and (b)
diagonal cracking in the spandrel of PS5.
(a) (b) (c)
Figure 5.10. Spandrel deformation modes: (a) Mode A – Flexural cracking, (b) Mode B – Diagonal
cracking and (c) Mode C – Bending-bed joint sliding.
The proposed Mode C describes the failure pattern that was observed in substructures PS1 and
PS2 which both had shallow spandrel depths, and spandrel aspect ratios (Ds/Ls) of 0.48 and 0.34
respectively. Mode C (Figure 5.10c) is characterised by horizontal cracking that extends over the full
length of the spandrel and is evident in multiple courses. This type of failure has not previously been
reported as a failure mode for as-built URM spandrels, although Augenti et al. [2011] reported a
spread horizontal crack pattern as the failure pattern for a low aspect ratio spandrel panel (0.65)
which was retrofitted with an inorganic matrix-grid strengthening system, and subjected to cyclic
lateral loads. It is also noted that Mode C results in a high energy dissipation capacity.
The spandrel modes in Figure 5.10 were based on the crack patterns observed during laboratory
testing, but generally compare well with the damage sustained by perforated URM walls that was
reported following the 2010 and 2011 Christchurch earthquakes. Photographic evidence of failure
Mode A and failure Mode B can be identified in the spandrel elements shown in Figure 5.11. In
Figure 5.11a and Figure 5.11b vertical cracking at the ends of the spandrel element can be seen,
which is characteristic of a pure flexural response, and in Figure 5.11c ‘X’ cracks characteristic of
diagonal shear cracking are visible. During assessments of the damaged URM buildings no evidence
of Mode C spandrel failure was found.
In the substructures that were classified as having ‘average’ or deep’ spandrels, the crack pattern
within the spandrel region suggested a diagonal shear failure initiated by a flexural mechanism. The
initial flexural behaviour was characterised by the formation of vertical cracks that opened in the
head joints at the tensile zone of the interface between that spandrel and joint regions for a particular
direction of loading. This cracking was then followed by the formation of diagonal cracks that
extended from the original flexural cracks that occurred in the opposite loading cycle. This
observation suggests that the spandrel behaves as an equivalent strut. Cattari and Lagomarsino
[2008] found a similar crack progression and failure mode in their finite element model when it was
subjected to monotonic loading, but proposed that the diagonal cracking initiated from the centre of
the spandrel.
Research Report
In contrast to the modelling results, the cyclic loading in the experiments reported here caused a
flexural crack to initiate for a direction of loading, and in the opposite direction of loading the diagonal
crack continued to propagate from the initial flexural crack.
Adding to the visible evidence of failure modes within the spandrel of substructure PS5 and PS6,
three gauges were placed to measure the vertical displacement along the spandrel length. In Figure
5.12 the spandrel displacement profile is shown for PS3 for four horizontal displacement steps. The
deformed shape corresponds to that of a beam subjected to double bending at low displacement
amplitudes, and the sagging at the centre of the spandrel is a result of the extensive diagonal
cracking at high displacement amplitudes. The presence of the timber lintel had negligible influence
on the vertical displacements of the URM spandrel as the lintel was bending in its weak axis and
was only held in place by two nails embedded in the piers at each end, resulting in the ends being
effectively pinned.
(a)
(b) (c)
Figure 5.11. Photographic examples of spandrel deformation modes observed following the earthquakes in
Christchurch: (a) Flexural cracking in the central spandrel of a two storey perforated URM façade, (b)
flexural cracking in a spandrel element in Avonmore House and (c) diagonal shear cracking in the spandrel
above a high arched lintel in Avonmore House.
Figure 5.12. Vertical displacement of spandrel in PS3.
5.2.9 Spandrel Elongation

For testing of substructure PS5 and PS6 additional gauges were placed at the top and bottom of the
spandrel to measure elongation of the spandrel element length. Figure 5.13a shows spandrel
extension recorded, plotted against drift. The plot shows that there was an almost linear relationship
between increasing drift and spandrel extension. Figure 5.13b shows the same spandrel extension
recordings plotted against spandrel axial load, which was assumed to be equal to half of the applied
lateral load. The circular markers on Figure 5.13a mark the maximum extension for each
displacement cycle. The data suggests that spandrel extension initiated at an axial load of
approximately 18 kN in the push cycle and at 10 kN in the pull cycle, which correlates well with the
force at which cracking was first observed for each loading direction (Table 5.4).
3 50
40
2.5
30
Spandrel Extension (mm)
Spandrel Axial Load (kN)
2 20
10
1.5
0
1 -10
-20
0.5
-30
0 -40
-1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6 0 0.5 1 1.5 2 2.5 3
Drift (mm) Spandrel Extension (mm)
(a) (b)
Figure 5.13. Substructure PS5: (a) Spandrel extension against drift and (b) spandrel extension against
estimated spandrel axial load.
5.2.10 Force–Displacement Response

The force–displacement response of the six substructures is presented in Figure 5.14 to Figure 5.16.
The displacement plotted was measured at the top of the spandrel, and the corresponding drift
represents lateral displacement over the total height of the specimen, expressed as a percentage.
The force–displacement plot for specimen PS1 shows hysteretic loops having moderate energy
dissipation and moderate stiffness and strength degradation with increasing displacement. The
ultimate displacement was governed by the onset of twisting at the base of Pier B. The narrow
Research Report
hysteretic loops and the almost constant strength with increasing displacement that is exhibited in
the force–displacement response of specimens PS2, PS3, PS4 and PS5 indicates that their non-
linear response was governed by a flexural (rocking) failure mode. In specimens PS2, PS3, and
PS4, failure as defined in Section 5.2.6 could not be reached and therefore the tests were terminated
when the displacement had exceeded 1% drift. The testing of PS5 was terminated due to the stroke
capacity of the hydraulic actuator being reached. The force–displacement response of PS4 was
highly non-symmetrical (Figure 5.15b). This non-uniformity occurred because the wall was effectively
pre-damaged in the push direction due to a malfunction in the loading during set-up.
100 70
80
50
60
40 30
20
Force (kN)
Force (kN)
10
0
-10
-20
-40 -30
-60
-50
-80
-100 -70
-20 -10 0 10 20 -30 -20 -10 0 10 20 30
Displacement (mm) Displacement (mm)
(a) (b)
Figure 5.14. Bilinear idealisation and backbone curves of (a) PS1 and (b) PS2.
110 110
90 90
70 70
50 50
30 30
Force (kN)
Force (kN)
10 10
-10 -10
-30 -30
-50 -50
-70 -70
-90 -90
-110 -110
-35 -25 -15 -5 5 15 25 35 -30 -20 -10 0 10 20 30
(a) (b)
The force–displacement plot for specimen PS6, shown in Figure 5.16b, is strongly characteristic of
a shear failure mode. The large hysteretic loops, and severe strength and stiffness degradation after
peak lateral strength was reached, is consistent with the observed pier failure mode. The significant
loss in stiffness and lateral strength following the attainment of maximum lateral strength, as evident
in the positive quadrant of the force–displacement plot, was due to the diagonal shear failure of Pier
A. As diagonal shear failure was evident only in Pier A, the force–displacement plot was highly
unsymmetrical. As per the characteristics of the force–displacement relation of a rocking dominant
element, the measured force in the negative quadrant of the force–displacement plot for PS6
reached its maximum strength and then remained constant with increasing displacement. As the
stability of the wall and its ability to support gravity loads was reduced considerably after shear failure
of the pier, the ultimate displacement of PS6 was approximately half that of the specimens that
responded in flexure.
80 40
60
40 20
20
Force (kN)
Force (kN)
0 0
-20
-40 -20
-60
-80 -40
-50 -40 -30 -20 -10 0 10 20 30 40 50 -20 -10 0 10 20
(a) (b)
The six tested unreinforced masonry substructures generally exhibited highly non-linear
characteristics, with the measured hysteretic shape being dependent on the failure mode of the pier
element. The force–displacement response of test units where the piers exhibited a horizontal crack
at the base and proceeded to rock with increasing displacement, display no indication of strength
loss up to a drift of 1.0%. Where the piers exhibited a shear failure, significant strength and stiffness
losses resulted. Small strength losses are evident between cycles one and two at the same
displacement amplitude, indicating that in-plane unreinforced masonry component performance is
mildly affected when repeatedly loaded to the same displacement.
5.2.10.1 Bilinearisation
The envelope of the force–displacement relationship for the individual substructural responses was
idealised as an elastic-perfectly plastic relationship. Bilinear idealisation of the masonry substructure
in-plane cyclic response is a simplification tool for evaluating key global response parameters, that
has been extensively reported in past literature [Shing et al., 1989; Magenes and Calvi, 1997;
Vasconcelos and Lourenço, 2009, Augenti et al., 2011]. This idealisation method was chosen as it
captures the essential performance characteristics whilst remaining simple enough to apply to
desktop procedures. The bilinear idealisation parameters of the URM substructural performance
allowed comparison between the test units in terms of elastic stiffness, strength and ductility.
Following the Tomaževič [2000] method, the ratio between the ultimate and maximum lateral
strengths was found to be 0.89, which is in good agreement with past experimental results on URM
walls, where it has been suggested that Vu/Vmax is 0.9 for unreinforced masonry walls failing in a
shear governed mode [Sheppard and Lutman, 1988; Tomaževič, 2000].
5.2.10.2 Ductility Capacity

The general term ‘ductility’ is defined by Paulay and Priestley [1992] as: “the ability of a structure or
its components, or of the materials used to offer resistance to the inelastic domain of response ...
(and) includes the ability to sustain large deformations, and a capacity to absorb energy by hysteretic
behaviour.”
Unreinforced masonry walls subjected to in-plane loading are not typically thought of as ductile
structural elements and are therefore assigned a ductility factor of 1 in current seismic codes
[NZSEE, 2006]. The idealised bilinear models shown in Figure 5.14 to Figure 5.16a illustrate that
URM elements subjected to in-plane loading are capable of undergoing significant deformations with
minimal strength loss in the inelastic domain of response and therefore demonstrate some ductile
characteristics. Displacement ductility is typically computed as the ratio of the ultimate displacement
at a specified limit of strength degradation to the displacement at the end of the elastic region of
response [Paulay and Priestley, 1992]. For unreinforced masonry, a reduction in the resisting force
of 20% has been typified in seismic codes as the limit to ductility, but as noted in Section 5.2.6, a
rocking response is characterised by minimal strength degradation with increasing displacement and
therefore this assumption is not appropriate. For the calculation of displacement ductility as reported
Research Report
in Table 5.5, where strength degradation was minimal, the ultimate displacement was defined as the
experimental value corresponding to the maximum strength degradation observed during the tests,
and is reported in brackets. Although the attainment of this limit is sometimes termed ‘failure’, it is
important to note, especially in regards to a rocking response, that significant additional inelastic
deformations are still possible without structural collapse. As outlined in Table 5.5, despite using
conservative definitions, the ductility capacity of the six substructures was found to be between 4.6
and 16.0, which considerably exceeded typical values adopted in the current New Zealand seismic
assessment procedure [NZSEE, 2006]. The reported ductility for PS6 in Table 5.5 using the
Tomaževič [2000] method is artificially high due to the high initial elastic stiffness.
Table 5.5. Bilinear parameters for substructures when using the Tomaževič method c.
Wall Vu de du du/H  Vu/ Ke Vcrack Failure

[kN] [mm] [mm] [%] Vmax [kN/m] /Vmax mode
PS1(+) 86.36 4.25 16.75 0.702 3.94 0.97 20312 0.41 P-R
S-X
PS1(-) -83.63 -2.99 (-15.04) (0.63) 5.02 0.86 27901 0.45
PS2(+) 57.10 4.46 (24.54) (1.03) 5.50 0.94 12802 0.45 P-R
S-X
PS2(-) -49.30 -4.08 (-24.94) (1.04) 6.10 0.89 12070 0.47
PS3(+) 95.43 4.191 (28.20) (0.89) 6.73 0.93 22769 0.43 P-R
S-X
PS3(-) -83.24 -3.14 (-23.74) (0.87) 7.56 0.79 26511 0.54
PS4(+) 64.02 2.74 (26.48) (0.82) 9.66 0.91 23365 - P-R
S-X
PS4(-) -89.72 -5.52 (-25.81) (0.79) 4.67 0.90 16253 0.68
PS5(+) 64.64 3.71 (40.26) (1.47) 10.85 0.86 17450 0.47 P-R
S-X
PS5(-) -48.64 -4.09 (-38.18) (1.39) 9.33 0.81 11884 0.47
PS6(+) 26.67 0.20 14.70 0.54 73.5 0.88 147841 0.69 P-R/X
S-X
PS6(-) -24.89 -0.17 (-16.41) (0.600) 96.5 0.93 143933 0.90
Averag - - - 6.90e 0.89 - 0.49
e
5.2.10.3 Ultimate Drift

The dependency of displacement ductility on the adopted definition of the elastic displacement
creates a critical deficiency in the use of displacement ductility to provide a representative measure
of the non-linear behaviour of unreinforced masonry walls subjected to in-plane loading.
Furthermore, the growing trend towards displacement based design has moved the research focus
away from defining the performance of individual structural components in terms of strength, to
instead consider that of ultimate drift of individual structural components.
Substructures PS2, PS3, PS4 and PS5 exhibited high ultimate drift capacities with no corresponding
drop in lateral strength. For the purpose of limiting non-structural damage in a URM building current
structural assessment codes [EN 1998-3, 2005; NZSEE, 2006; D.M. 14.01.2008, 2008] suggest an
ultimate drift limit of 0.8% for a flexural response, and 0.4% for a shear response. Priestley et al.
[2007] also suggest a drift limit of 0.4% for URM walls failing in a shear dominated response. At a
drift of 0.8% substructures PS2, PS3, PS4 and PS5 were not in danger of collapse and had not lost
c Numbers in parenthesis are calculated from the maximum recorded displacement where a 20% reduction in the maximum
strength was not reached and therefore an ultimate displacement was not found.
their capacity to support gravity loads. Substructure PS6 exhibited diagonal shear failure in the piers
early in the test, and an ultimate drift of 0.56% was recorded. At this drift, large cracks had opened,
sliding was occurring on the bed joints, and some damage due to crushing of the masonry was
evident. To assess the current drift limits proposed by NSZEE [2006], the ratio of ultimate measured
lateral displacement (d80) to the code proposed lateral displacement limit for a flexural/shear
response is presented in Table 5.6.
Table 5.6. Assessment of drift levels achieved in the substructure testing.
Substructure 80 d80/dmax(NZSEE)

PS1(+) 0.73 1.20
PS1(-) > 0.63 1.04
PS2(+) > 1.19 1.74
PS2(-) > 1.19 1.74
PS3(+) > 0.87 1.67
PS3(-) > 0.87 1.65
PS4(+) > 0.81 1.84
PS4(-) > 0.80 1.80
PS5(+) > 1.47 2.80
PS5(-) > 1.39 2.66
PS6(+) 0.538 1.43
PS6(-) > 0.602 1.61
Average 1.77
The NZSEE assessment code adopts the ‘strong spandrel – weak pier’ simplified model for the
assessment of perforated walls, and therefore it follows that all the deformation is associated with
the pier elements. As such, the code proposed lateral displacement limit is calculated as the lateral
displacement for a pier with a height equal to the height of the adjacent opening. Table 5.6 shows
ratios all in excess of 1.0 with an average ratio of 1.77. The drift limits currently set by NZSEE [2006]
have been found in this investigation to be in the range of accurate to conservative, which is in
agreement with the conclusions of Russell [2010] for in-plane loaded URM walls and flanged URM
walls.
5.2.11 Conclusions for Substructure Testing

For all the substructures tested it was found that where the pier aspect ratio was high (greater than
1) and for both levels of axial stress tested, damage was initiated and concentrated in the spandrel
element, with subsequent rocking in the piers of each specimen evident during testing. In
substructures PS3, PS4 and PS5 the spandrels exhibited classic diagonal shear failure, effectively
uncoupling the piers and allowing pier rocking, with the effective pier height being equal to the total
height of the substructure. Whilst the pier rocking mechanism allowed for large displacement
capacities, self centering, and low residual displacement, the diagonal shear failure in the spandrel
element affected the maximum resistance of the substructure and the energy dissipation properties
of the substructure.
When characterising the response of the substructure, the spandrel behaviour dictated the boundary
conditions on the pier. For substructures where the spandrel element failed first, the piers were
considered to be unrestrained over the full height of the substructure, and therefore had a high aspect
ratio. In this case, the overall response of the substructure had strength and stiffness characteristics
similar to a pure rocking structure, where the lateral resistance reached its peak and then remained
reasonably constant with increasing displacement. Energy dissipation in this response mode was
minimal, which is typical for a rocking dominated response. For substructure PS6 where the pier
failed first in a diagonal shear mode, the substructure force–displacement response was
Research Report
characterised by large hysteretic loops and strength degradation with increased displacement,
typifying a pier shear response. Force–displacement plots indicated that the stiffness and energy
dissipating capacity of a substructure as a whole was largely dependent on the pier failure
mechanism.
Through this experimental programme, the occurrence of spandrel failure has been identified as a
possible response of perforated URM walls to lateral loading. For the one storey substructures
reported, where the demand/capacity ratio for the spandrel elements was higher than for the pier
elements, the spandrels failed first and the effective pier height was equated to the full storey height.
The implication of this finding for multi-storey structures is not yet fully understood. For this reason,
a testing programme to include a half-scale two storey perforated URM wall which was designed
based on the geometric qualities that were compatible with spandrel failure as shown herein.
5.3 Two Storey Perforated Wall
The two storey unreinforced masonry perforated wall was constructed at half scale. For ease of
construction and stability of the wall, the dimensions of the wall were half scale in length and width,
and full scale in thickness, using standard sized vintage clay bricks. The test wall form is typical of a
facade wall of a two storey stand-alone New Zealand URM building. Figure 5.17 illustrates the
geometry of the test wall. The locations of the openings provided regularity to the fenestration
pattern, with the ground floor windows and first floor windows aligned vertically. The doorway
opening was wider than the window openings (740 mm and 620 mm respectively), creating a small
offset of the interior upper and lower pier (piers B and C) centrelines (Table 5.7). The openings were
aligned horizontally at the top of the ground floor, and the base of the first floor, resulting in the
central spandrels having constant depth.
Arched window openings on the first floor were chosen to investigate their effect on spandrel
deformation mode, and therefore the depth of the spandrels between the ground and first floor
(central spandrels) was designed to be the same as the minimum depth of the top spandrel
measured from above the highest point of the arch. The central spandrels at mid wall height had
aspect ratios (depth/length) of between 0.82 and 0.85. For the spandrel components at the top of
the wall, the aspect ratios were between 0.84 and 0.87. As the depth varied over the length of the
spandrel due to the arched windows, the aspect ratios were calculated based on the minimum depth
above the peak of the arch. Based on the aspect ratios of the tested spandrels in the substructures,
the spandrels in the test wall are classified as having average depth as their aspect ratios were less
than 1, but also shallow when compared to the width of the adjacent piers.
The width of the wall was not scaled, with the ground floor being three leafs thick and the first floor
being two leafs thick. Full scale thickness was specified to ensure out-of-plane stability of the wall.
This decision was justified from the finding that there was no clear difference between the diagonal
shear strength of two leaf and three leaf wallettes as shown by Lin [2012] and by Dizhur & Ingham
[2013]. In addition, thickness is a known and accounted for variable in the NZSEE [2006] predictive
equations, and therefore represents a calculable effect on the in-plane lateral strength of a wall or
pier.
Previous system level experimental programmes [Yi et al., 2006; Thurston and Beattie, 2009] have
concluded that minimal coupling occurs between parallel walls, which supports the analysis of wall
behaviour using a two-dimensional approach. Based on these findings and laboratory space
limitations, the wall was constructed as an isolated wall, with applied vertical loads to account for
diaphragms, parapets and live loads.
4460 4460
360 600 360

890 600620 890890 620620 480
890 620 480 230 230
790
790
270
270
1850
1850
880
880
AT BTAT BT CT CTT
D DT
510
510
1210
1210
AG BGAG BG CG CG
D DG
2050
2050
510
510
360 600 360

830 600 830 830 620 480
830 620 480 350 350
(a) (b)
Figure 5.17. Two storey perforated test wall: (a) Side view and (b) end view (dimensions in mm).
Table 5.7. Pier geometry.
Location bp hp Aspect ratio

[mm] [mm] Pier A Pier B Pier C Pier D
Ground
350 1210 3.4 1.5 1.5 2.5
floor
First floor 230 880 2.4 1.0 1.0 1.8

A summary of the results from material testing is provided in Table 5.8. The brick compressive
strength was found to be low, indicating a soft brick as defined by NZSEE [2011], and this low brick
strength resulted in a low masonry compressive strength. Shear triplet tests were conducted at three
levels of axial compression force, being 4 kN, 8 kN, and 12 kN, as specified by RILEM MS-B.4
[1996]. The line of best fit determined that the coefficient of friction was approximately 0.4, and that
the cohesion was approximately 0.25. From analysis of the wallette testing results it was found that
the diagonal tensile strength was relatively consistent and had a value of 0.084 MPa, with a
coefficient of variation of 10% across the two leaf and three leaf wallettes. The shear modulus results
were highly variable, and this variability was reflected in the high coefficient of variation (40%).
Research Report
Table 5.8. Two storey wall material properties.
Mortar Brick Masonry Young’s Diag. Shear

comp. comp. comp. Cohesion Coeff. tensile
modulus modulus
Value strength fʹj strength strength c of friction strength
[MPa] fʹb fʹm E fʹdt G
[MPa] µ
[MPa] [MPa] [MPa] [MPa] [MPa]
Mean 3.2 16.28 5.74 0.25 0.40 1385 0.084 740
COV 12% 18% 25% - - 31% 10% 40%
ASTM ASTM ASTM

Method of ASTM ASTM RILEM MS- RILEM MS- ASTM
C109/C109 E519/E519 E519/E519
Test C67-11 C1314-11 B.4 B.4 E1111-04
M-11 M-10 M-10
5.3.2 Scaling
In support of the half scale adopted for the two storey wall test, a series of component and material
tests was completed. Scale was found to have minimal to no effect on the compressive strength
characteristics of masonry and brick, and no effect on stiffness of masonry in compression. The
effects of scale were found to be significant for diagonal shear strength and shear modulus, when
either the overall geometry were scaled by a factor of 2, or when the materials alone were scaled.
5.3.3 Test Setup

The two storey wall was tested as shown in Figure 5.18. Lateral force was applied at the centroid of
both spandrel cross-sections, using two hydraulic-powered actuators, which reacted against the
laboratory strong wall. Flat steel plates measuring 10 mm thick × 230 mm wide × 500 mm long were
positioned on the top of the spandrel at the centre-line of the piers, to provide a smooth horizontal
rolling surface for a set of rollers which were then placed on top of the steel plates. The rollers were
positioned to reduce any friction transfer of the horizontal load between the steel channel and the
spandrel. A steel loading beam was seated on the rollers.
A 20 mm thick steel plate was welded on each end of the steel channel so that the horizontal load
could be transferred from the actuator into the spandrel directly. Flybraces mounted on the steel
channel were designed to strengthen the connection between the endplate and the beam so that the
endplate did not deform, ensuring that the horizontal load was fully transferred into the test
substructure. The loading beam was 50 mm longer than the test unit, and was not fixed to the test
unit, and therefore contributed no strength to the spandrel. For load transfer to the central spandrel,
a 300 mm × 550 mm wide × 20 mm thick steel plate was clamped to each end of the wall using two
steel tendons located on each side of the wall.
Two sets of unbonded post-tensioning tendons positioned on the outside of the wall were used to
apply supplementary axial load to the top of the wall. Springs were connected in series with the
tendons to help reduce additional stress being applied to the test unit if rocking induced uplift of the
piers was to occur. In addition, load cells measuring the force transferred into the steel UB and
therefore into the test wall, from the stressing tendons, was constantly monitored and adjusted to
remain constant throughout the test.
To apply axial load to the central spandrel two steel UB sections were placed along the step created
in the spandrel cross-section due to reduction of the wall width from three leaf to two leaf. Similar to
the set-up at the top of the wall, a stressing tendon on each side of the wall was anchored to a steel
section running perpendicular to the wall and seated on the steel UB above the window base, with
a load cell measuring the total force transferred into the steel UB. A mortar pack was spread over a
width of approximately 100 mm at the centre of each pier to ensure that the axial load was transferred
vertically to the piers rather than onto the spandrel.
Push Pull Push Pull
Stressing Tendons Stressing Tendons
Load Cell LoadLoad

Cell Cell Load Cell
Rollers Rollers
Steel reaction plate Steel reaction plate Box Beam Box Beam
Top Hydraulic Top Hydraulic
Steel UB Steel UB actuator actuator
Load Cell Load Cell Central Hydraulic Central Hydraulic

actuator actuator
Mortar Pack Mortar Pack
Steel UB Steel UB
Steel end plates Steel end plates
Steel rods Steel rods

Steel UC Steel UC
Strong Wall Strong Wall
Spring Spring
Spring Spring
Loading Plan Loading Plan Side Elevation Side Elevation
(a) (b)
Figure 5.18. Two storey wall testing set-up: (a) Side view and (b) end view.
An independent frame was positioned at one end of the wall, against which displacement of the test
wall in the horizontal direction was measured, which eliminated any effects from flexing of the strong
wall. Additional axial loads were applied to the two storey perforated wall to simulate gravity loads.
The perforated wall was assumed to be part of a complete building having floor plan dimensions of
4.46 m × 5 m, and therefore simulated diaphragm, roof, parapet and live loads were applied based
on these dimensions. The axial load on the first floor and roof level diaphragm was 7.3 kN/m and 5.6
kN/m respectively.
5.3.4 Testing Procedure

A displacement-controlled pseudo-static procedure was employed to capture the non-uniformly
accumulated damage in the wall, and to enable observations of damage and failure mechanisms.
From dynamic testing of multi-storey unreinforced masonry buildings with similar perforations it has
been found that equal forces applied at each storey level provide a close approximation of typical
inertial forces recorded in shake table tests [Kingsley et al., 1996]. Ideally a force-controlled loading
procedure would have been utilised but the set-up and execution of force-controlled behaviour had
not yet been trialled with the hydraulic actuators in the University of Auckland laboratory, and
therefore displacement control was adopted.
The ratio of displacement applied to the top spandrel to the displacement applied at the central
spandrel was determined based on a first mode deformation shape. This relationship was assumed
constant at all times and therefore the displacement amplitude at any time applied at the central
spandrel was 0.57 of the displacement amplitude applied to the top spandrel as shown in Figure
5.19.
Research Report
80
70 Top Displacement
60 Mid Displacement
50
40
Displacement (mm)
30
20
10
0
-10
-20
-30
-40
-50
-60
-70
-80
0 0.5 1 1.5 2 2.5 3
Time (sec) 4
x 10
Figure 5.19. Loading history for testing of the two storey wall.
During testing the top spandrel was subjected to an initial displacement of 0.5 mm in each direction,
with displacements then increased by increments of 0.5 mm in every second cycle until 2 mm was
reached, after which the increments were increased to 2 mm in every second cycle, measured at
the top course of bricks in the spandrel. This displacement history was applied to both the top and
central spandrels simultaneously, with the displacements at the central spandrel being 0.57 of the
absolute displacement of the top spandrel.
In the elastic state of the structure, when a displacement is applied to the top spandrel, a resulting
displacement is also induced at the central spandrel height in proportion to the relative stiffness of
the ground and top stories. It also follows that an applied displacement at the mid spandrel will induce
a displacement at the top of the wall. Therefore the force demand at each loading position is a
function of the applied displacement at both loading positions.

Initial cracking in the test wall occurred on the pull cycle at a drift of 0.16% with a corresponding base
shear of 37.5 kN. Horizontal cracks propagated along the pier/joint interface at the top and base of
Pier AT, suggesting that Pier AT was rocking (Figure 5.20a). This crack pattern was mirrored in Pier
DT at a drift of 0.22%. At the same drift level, horizontal cracks formed in the arch and extended
horizontally along the pier/joint interface above piers BT and CT (Figure 5.20b). In the central spandrel
components horizontal cracks spread along the mortar-brick interface and were observed in all
layers of the bed joints through the depth of the spandrels, creating five parallel horizontal cracks
(Figure 5.20c). At a drift of 1.28% (corresponding to 50 mm top lateral displacement) wide cracks
had developed in the arched lintels and resulted in bricks dislodging and falling out. During the push
cycle to a top displacement of 55 mm complete detachment of the top of Pier DT from the joint was
visible, suggesting that the top section of the wall, both top spandrel and joint components, were
rocking as one panel. Figure 5.21 shows the crack pattern at the conclusion of testing. The image
shows Pier A on the left and Pier D on the far right, with the push load (positive force) applied to the
centre of the top and central spandrels adjacent to Pier D, directed towards the left.
5.3.6 Force–Displacement Response

The force–displacement response of the test wall is presented in Figure 5.22a through to Figure
5.23b. The displacement plotted was measured by the actuators and therefore represents the
applied displacement, and the corresponding drift, which is calculated as the lateral displacement
over the height measured from the base to the application of load, expressed as a percentage.
(a) (b)
(c)
Figure 5.20. Photos of two storey wall testing: (a) Horizontal crack at the top of Pier A T, (b) crack
propagating from arches horizontally across top of Pier BT and (c) cracking through all bed-joints in central
spandrel component (photos taken from the reverse side of the wall, so that Pier A is the far right pier, and
Pier D is the far left pier).
The response recorded by the top actuator, shown in the force-displacement plot in Figure 5.22a,
has characteristics of both a flexural and shear response mode. In the positive quadrant, preceding
a drift level of 1.2%, there is a dominant bilinear trend in the loading profile. The response has a
constant linear elastic stiffness, with stiffness decreasing in the second branch resulting in strength
degradation with increasing lateral displacement. The strength reached a maximum at about 0.2-
0.4% drift and then gradually decreased. Following a drift of 1.2% the loading and unloading
stiffness decreases rapidly, which is characteristic of a shear failure. In the negative quadrant a
flag shaped hysteretic loop is visible, which typifies a rocking response. In the final displacement
cycles, to 55 mm and 60 mm lateral displacement, the loading and unloading stiffness decreased
rapidly.
The form of the force–displacement plot for the response recorded by the actuator at mid wall
height, shown in Figure 5.22b, is considerably different to the behaviour recorded by the top
actuator. Notably the elastic stiffness decreased with each cycle, and the response more closely
resembled a degrading linear elastic model. A change in response occurred in the displacement
Research Report
cycle to a top displacement of 50 mm, where the strength increased and the unloading stiffness
reduced to almost zero, resulting in a narrow hysteretic loop.
Figure 5.23a shows both the top and mid response together, and Figure 5.23b shows the total
base shear plotted against the top lateral displacement. From the plot of base shear against top
lateral displacement it can been seen that the total response of the perforated wall was fairly
symmetric in hysteretic shape, with a higher total strength in the positive loading direction (average
of 44 kN), compared to the total strength in the negative loading direction (average of 38 kN). With
increasing displacement, the base shear/lateral strength remains constant, indicating that the
strength degradation evident in Figure 5.22a was balanced by the increase in strength with drift
as shown in Figure 5.22b.
4460
790
1850
880
510
1210
2050
510
360 600 830 830 620 480
Figure 5.21. Crack pattern for the as-built two storey perforated wall after cyclic pseudo-static testing.
5.3.7 Global Wall Behaviour

The extensive cracking in the spandrel panels at low drift levels and the cracking in the piers localised
to the base and top of the piers was indicative of a weak spandrel response where the overall wall
behaviour is governed by the pier response. The damage to the spandrel panels at mid wall height
and the lack of cracking at the top of the ground floor interior piers, and at the base of the top floor
interior piers suggested that the interior piers deformed as a double storey height pier. This failure
mode was confirmed by the null readings of the portal gauges that were positioned to measure
relative vertical displacement between the joint and the piers at mid wall height.
Drift  (rads) Drift  (rads)

-1.2 -0.8 -0.4 0 0.4 0.8 1.2 -1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6
40 40
30
20 20
Top Lateral Force (kN)
Mid Force (kN)

10
0 0
-10
-20 -20
-30
-40 -40
-60 -40 -20 0 20 40 60 -40 -30 -20 -10 0 10 20 30 40
Top Lateral Displacement (mm) Mid Displacement (mm)
(a) (b)
Figure 5.22. Force–displacement response recorded from (a) top actuator and (b) mid-height actuator.
Drift  (rads)
40 -1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6
60
40
20
Lateral Force (kN)
20
Base Shear (kN)
0
0
-20
-20
Top Response -40

Mid Response
-40
-60 -40 -20 0 20 40 60 -60
Lateral Displacement (mm) -60 -40 -20 0 20 40 60
Top Lateral Displacement (mm)
(a) (b)
Figure 5.23. (a) Mid and top force–displacement response and (b) force–displacement response for total
force against top displacement.
No vertical movement of the base spandrel was recorded and therefore confirmed that no global
rocking of the wall occurred. Figure 5.24a and Figure 5.24b show an idealised representation of the
wall response based on the displacement gauges that reported relative vertical displacements
between pier/spandrel/joint components and absolute horizontal displacements.
The graphical representation of the overall global response highlights the central spandrel
component’s coupling beam type response, and the dominance of the pier flexural response. The
vertical gauge readings suggested that the rotation of the ground floor piers was not equal across
the four piers. Also noteworthy is the distinction between the grouping of components within the
external pier adjacent to the point of loading, and in the external pier on the opposite side to loading.
It is shown in Figure 5.24a that external pier D separated at the interface between the top of Pier DG
and the joint, whereas external Pier A separates at the interface between the top of the joint and the
base of Pier AT. The figures show that the response was mirrored in the push and pull cycles.
Research Report
(a) (b)
Figure 5.24. Simplified wall deformation: (a) Push cycle and (b) pull cycle.
At a top displacement of 50 mm (corresponding to a drift of 1.28%) a sudden loss in stiffness

occurred within the top floor. For clarity of explanation, the top row of joint and spandrel components
are referred to as the ‘top spandrel’. The loss in stiffness is attributed in part to the complete
disconnection of the top spandrel from the top of Pier D (as shown in Figure 5.25a), which resulted
in the load from the top actuator being distributed between three piers rather than four. At this point
in the test damage to the joint regions was extensive, including the visible opening of diagonal cracks,
and the interaction between the top of Pier A and the top spandrel changed from rocking to sliding
(as shown in Figure 5.25b). Because the applied displacement of the mid and the top actuators were
slaved, the loss in stiffness within the top spandrel led to a reduction in the induced displacement at
mid height and hence an increase in lateral force demand on the first floor actuator, which is seen in
the strength growth.
(a) (b)
Figure 5.25. (a) Separation of Pier D and top spandrel and (b) sliding between Pier A and top spandrel.
5.3.8 Spandrel Behaviour

The spandrel response mode is critical for defining the pier boundary conditions, which in turn is a
key variable when determining the pier deformation mode and lateral strength. The deformation
modes of the central spandrels and arched top spandrels were different and are therefore discussed
separately below.
5.3.8.1 Central Spandrel Deformations

The formation of horizontal cracks along the brick-mortar interface between all the brick courses was
visible in the three central spandrel components. This deformation mode (Mode C in Figure 5.10c)
was previously encountered in substructures PS1 and PS2, which also had shallow spandrels. Mode
C is associated with spandrels that have a relatively shallow depth when compared with the length
of the adjoining piers. With increasing rotation of the piers, the differential lateral displacement over
the height of spandrel results in deformation of the spandrel panel in the form of a parallelogram as
shown graphically in Figure 5.26a. As the spandrel deforms, the portal gauges that were placed
diagonally across the spandrel recorded the diagonal shortening and elongation. The crack pattern
is explained by the bed joint shear stresses induced by the displacement incompatibility exceeding
the shear strength between the brick courses within the spandrel, resulting in sliding between the
courses, shown graphically in Figure 5.26b and Figure 5.26c.
The shear force that was transferred by the spandrel between the adjacent piers could not be
measured directly due to the method of applying horizontal displacement to the test wall, and hence
the force-displacement response and energy dissipation capacities of the spandrel when deforming
in this sliding mode cannot be established.
5.3.8.2 Arched Spandrel Damage

The arched spandrels experienced extensive damage. Diagonal stepped cracking propagated
throughout the spandrel panel (Figure 5.27a), not limited to the spandrel, but also extended into the
joint region. Lateral movement within the arch region was suggested by identifying that arched lintel
bricks had become loose and in some cases had dropped out (Figure 5.27b).
(a)
(b) (c)
Figure 5.26. Mode C spandrel deformation: (a) Deformation of the coupling spandrel between two rocking
piers, (b) rotation of spandrel stepped over brick courses and (c) sliding between brick courses due to
displacement incompatability (see also Figure 5.10).
Research Report
(a) (b)
Figure 5.27. Crack patterns above arched lintels: (a) Cracking above arched lintel and (b) Arched lintel
with bricks dislodging .
5.3.9 Bilinear Modelling

The envelope of the force–displacement relationship for the top-level, mid-level and total response
was idealised as an elastic-perfectly plastic relationship. The bilinear idealisation parameters for the
multi-storey URM perforated wall performance allowed comparison between the storey responses
in terms of elastic stiffness, strength and ductility. The parameters for the bilinear idealisation for
each storey, for loading in both directions, are reported in Table 5.9 according to the Tomaževič
[2000] procedure and in Table 5.10 according to the Magenes and Calvi [1997] procedure. The
bilinear idealisation is plotted against the force–displacement response in Figure 5.28a for the total
response, in Figure 5.28b for the top floor response, and in Figure 5.29 for the first floor response,
with the first cycle peaks shown in blue.
The graphical presentation of the bilinear idealisation indicates that the Tomaževič and the Magenes
and Calvi procedures result in similar backbone curves (both ultimate shear strength and elastic
stiffness) where the experimental force post ‘yielding’ is fairly constant. For the top floor response,
where the positive experimental backbone includes strength degradation, the Tomaževič procedure
results in a lower ultimate shear force (Vu) as the procedure takes into consideration the behaviour
post yield, whereas the Magenes and Calvi ultimate shear force parameter is purely based on the
peak response. The spike in the experimental peak force in the final cycles of the first floor response
strongly influenced the bilinear idealisation parameters according to the Magenes and Calvi method
and therefore was unrepresentative of the experimental backbone curve. The shape of the
experimental backbone curve for the first floor response more closely resembles multi-linear
behaviour with positive second branch stiffness (strain hardening), rather than being an elastic
perfectly-plastic model.
Table 5.9. Bilinear parameters in accordance with the Tomaževič [2000] procedure d.
Vu de du du/H [%] d Vu/Vmax Ke Vcrack/

[kN] [mm] [mm] [kN/m] Vmax
Total (+) 46.85 2.81 (54.75) (1.40) (19.5) 0.94 16654 0.62
Total (-) 38.28 2.52 (55.08) (1.41) (21.9) 0.92 15179 0.41
Top (+) 27.27 2.85 52.50 1.35 18.4 0.88 9566 0.58
Top (-) 16.16 2.46 (-35.90) (0.92) 14.6 0.84 6573 0.69
Mid (+) 20.14 1.85 27.35 1.33 14.8 0.73 10905 0.48
Mid (-) 23.19 2.80 (-32.10) (1.57) 11.5 0.89 8294 0.33
d Numbers in parenthesis are calculated from the maximum recorded displacement where a 20% reduction in the maximum
Table 5.10. Bilinear parameters according to the Magenes and Calvi [1997] procedure e.
Vu de du Ke Ductility
[kN] [mm] [mm] [kN/m] d
Total (+) 44.84 6.69 (54.75) 14025 (8.2)
Total (-) -37.43 -7.93 (-55.01) 8717 (6.9)
Top (+) 27.90 3.68 51.96 7576 14.1
Top (-) -17.38 -2.57 -31.97 6757 12.4
Mid (+) 24.98 15.44 (32.01) 1618 (2.1)
Mid (-) -23.50 -5.64 (-32.07) 4168 (5.7)
5.3.10 Ductility Capacity

URM is generally not thought of as being ductile in the typical definition of the concept, but the
idealised bilinear models shown in Figure 5.28a to Figure 5.29 illustrate that URM components
subjected to in-plane loading are capable of undergoing significant deformations with minimal
strength loss in the inelastic domain of response and therefore demonstrate some ductile
characteristics. As discussed previously with reference to substructure response, in a typical rocking
response the structure can achieve high drift ratios with minimal strength degradation, and therefore
the codified definition of failure that determines the ultimate drift is difficult to apply. For this reason,
where strength degradation was minimal, the ultimate displacement was defined as the experimental
value corresponding to the maximum strength degradation observed during the tests, and is reported
in brackets.
Drift  (rads) Drift  (rads)

-1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2
60 40
30
40
20
20
Base Shear (kN)
Top Force (kN)
10
0 0
-10
-20
-20
-40
Magenes & Calvi -30 Magenes & Calvi
Tomazevic Tomazevic
-60 -40
-60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60
Top Lateral Displacement (mm) Top Lateral Displacement (mm)
(a) (b)
Figure 5.28. Bilinear idealisation of (a) the total response and (b) top floor response.
e Numbers in parenthesis are calculated from the maximum recorded displacement where a 20% reduction in the maximum
Research Report
Drift  (rads)
-1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6
40
30
20
Mid Force (kN)

10
-10
-20
-30 Magenes & Calvi

Tomazevic
-40
-40 -20 0 20 40
Mid Lateral Displacement (mm)
Figure 5.29. Bilinear idealisation of the first floor response.
The calculated ductility defined by du/ds 𝑑𝑢 /𝑑𝑒 is shown in Table 5.9 and Table 5.10 for the two
respective idealisation procedures. The adoption of the Magenes and Calvi procedure resulted in a
reduced elastic stiffness and hence a reduction in the calculated drift by 50% on average.
5.3.11 Ultimate Drift

The high elastic stiffness of unreinforced masonry walls when subjected to loading in-plane, coupled
with its capacity for inelastic deformation in a flexure dominated response, correspond to
unrepresentative high ductility values. For this reason, ultimate drift provides a more demonstrative
measure of the inelastic response of unreinforced masonry walls.
The average ultimate drift recorded for the wall response was 1.40%, at which point there was no
strength degradation or loss of stability. The positive displacement cycle for the top floor response,
which failed due to shear cracking, reached an ultimate drift of 0.92%. As such, the drift limits
currently set by NZSEE [2006] have been found to be in the range of accurate to conservative.
5.4 Conclusions
Pseudo-static testing of six full-scale pier-spandrel URM substructures was conducted to investigate
the effect of spandrel depth on the performance of a perforated wall. In addition, the key variables
that control the deformation mode of piers were assessed by testing geometrically identical test units
with different pier axial stress levels, and constructing a test unit with adjacent piers of different
length.
Pseudo-static testing of a half-scale two storey perforated URM wall representative of a typical
Typology C New Zealand URM building was presented. The geometry of the wall was designed to
investigate the effect of shallow spandrels on multi-storey response. In addition, the damaged wall
was repaired with a textile reinforced mortar strategically applied to control the failure mode and to
improve performance. The following conclusions are drawn from the laboratory study:
 The shallow central spandrels behaved as coupling beams, allowing multi-storey pier rocking to
occur in the two internal piers.
 The external piers rocked over their inter-storey effective height.
 The total response (top displacement against base shear) was found to approximate elastic
perfectly-plastic behaviour, with a non-negligible non-linear response.
 The calculated level of displacement ductility was shown to be misrepresentative of the non-
linear behaviour, with a calculated average displacement ductility of 20 for the total response.
Ultimate drift was confirmed to be a representative measure of the deformation capacity of
unreinforced masonry walls when subjected to in-plane seismic loading. The as-built wall
reached an ultimate drift of 1.4% with minimal strength loss, but with extensive damage to the
top spandrel.
 Pier effective height for a flexure response was found to be dependent on loading direction for
cases where the adjacent opening heights were not identical.
 Repair of the damaged wall with strategically placed textile reinforced mortar resulted in an
increase in hysteretic energy dissipation, a reduced elastic stiffness, an average 15% increase
in lateral strength, and a similar drift capacity.
Research Report
5.5 REFERENCES
ASCE/SEI 41-06 [2007] Seismic Rehabilitation of Existing Buildings, American Society of Civil Engineering.
ASTM E1111-04 [2004] Standard Test Method for Young's Modulus, Tangent Modulus, and Chord Modulus,
American Society for Testing and Materials, Philidelphia, Pennsylvania, USA.
ASTM 1072-05 [2005] Test Method for Measurement of Masonry Flexural Bond Strength, American Society
for Testing and Materials, Philidelphia, Pennsylvania, USA.
ASTM E519/E519M-10 [2010] Standard Test Method for Diagonal Tension (Shear) in Masonry Assemblages,
American Society for Testing and Materials, Philidelphia, Pennsylvania, USA.
ASTM C1314-11 [2011] Standard Test Method for Compressive Strength of Masonry Prisms, American
Society for Testing and Materials, Philidelphia, Pennsylvania, USA.
ASTM C67-11 [2011] Standard test methods for sampling and testing brick and structural clay tile, American
Society for Testing and Materials, Philidelphia, Pennsylvania, USA.
ASTM C109/C109M-11 [2011] Standard Test Method for Compressive Strength of Hyraulic Cement Mortars
(Using 2-in or 50-mm Cube Specimens), American Society for Testing and Materials, Philidelphia,
Pennsylvania, USA.
Augenti, N., Parisi, F., Prota, A., Manfredi, G. [2011] "In-plane lateral response of a full-scale masonry
subassemblage with and without an inorganic matrix-grid strengthening system," Journal of Composites
for Construction, Vol. 15, No. 4, pp. 578-590.
Calvi, G.M., Kingsley, G.R., Magenes, G. [1996] "Testing masonry structures for seismic assessment,"
Earthquake Spectra, Vol. 12, No. 1, pp. 145-162.
Cattari, S., Lagomarsino, S. [2008] "A strength criterion for the flexural behaviour of spandrels in unreinforced
masonry walls," Proceedings of the 14th World Conference on Earthquake Engineering, Beijing, China.
Dizhur, D., and J. M. Ingham. [2013] "Diagonal tension strength of vintage unreinforced clay brick masonry
wall panels." Construction and Building Materials 43, pp. 418-427.
D.M. 14.01.2008 [2008] Decreto Ministeriale del 14 gennaio 2008: Approvazione delle nuove norme tecniche
per le costruzioni, Ministero delle Infrastrutture e dei Trasporti, Supplemento ordinario alla G.U. n. 29 del
4 febbraio 2008, Rome, Italy [in Italian].
Kingsley, G.R., Magenes, G., Calvi, G.M. [1996] "Measured seismic behaviour of a two-storey masonry
building," Worldwide Advances in Structural Concrete and Masonry, Proceedings of the Committee on
Concrete and Masonry Symposium, ASCE Structures Congress XIV, Chicago, Illinois, USA.
Lin, Y. [2012] Strengthening of Structures using ECC Shotcrete, PhD Thesis, University of Auckland, New
Zealand.
Lumantarna, R. [2012] Material Characterisation of New Zealand Unreinforced Masonry Buildings, PhD
Thesis, University of Auckland, New Zealand.
Magenes, G., Calvi, G.M. [1997] "In-Plane seismic response of brick masonry walls," Earthquake Engineering
and Structural Dynamics, Vol. 26, No. 11, pp. 1091-1112.
NZSEE [2006] Assessment and Improvement of the Structural Performance of Buildings and Earthquakes -
Recommendations of a NZSEE Study Group on Earthquake Risk Buildings, New Zealand Society for
Earthquake Engineering, Wellington, New Zealand.
NZSEE [2011] Assessment and Improvement of Unreinforced Masonry Buildings for Earthquake Resistance,
New Zealand Society for Earthquake Engineering, Wellington, New Zealand.
Paulay, T., Priestley, M.J.N. [1992] Seismic Design of Reinforced Concrete and Masonry Buildings, John Wiley
and Sons, New York, USA.
Priestley, M.J.N., Calvi, G.M., Kowalsky, M.J. [2007] Displacement-Based Seismic Design of Structures, IUSS
Press, Pavia, Italy.
RILEM MS-B.4 [1996] Determination of shear strength index for masonry unit/mortar junction, International
Union of Laboratories and Experts in Construction Materials, Systems and Structures, Bagneux, France.
Russell, A.P., Ingham, J.M. [2008] "Trends in the architectural characterisation of unreinforced masonry in
New Zealand," Proceedings of the 14th International Brick and Block Masonry Conference, Bondi,
Australia.
Russell, A. P. [2010] Characterisation and seismic assessment of unreinforced masonry buildings, Ph.D.
Thesis, University of Aukland, New Zealand.
Sheppard, P., Lutman, M. [1988] "Seismic risk assessment and design of building structures," in Koridze, A.
(ed.) Estimation of expected seismic vulnerability: A simple methodology for medium sized groups of older
buildings, Omega Scientific, Wallingford, UK, pp. 47-62.
Shing, P.B., Noland, J.L., Klamerus, E., Spaeh, H. [1989] "Inelastic behaviour of concrete masonry shear
walls," Journal of Structural Engineering, Vol. 115, No. 9, pp. 2204-2225.
Thurston, S.J., Beattie, G.J. [2009] "Seismic performance of New Zealand two-storey brick veneer houses,"
NZSEE Conference – Why do we still tolerate buildings that are unsafe in earthquakes, Christchurch,
New Zealand.
Tomaževič, M. [2000] “Earthquake-Resistant Design of Masonry Buildings” In: Elnashai, A.S., Dowling, P.J.
(eds.) Series on innovation in structures and construction, Vol. 1, Imperial College Press, London, UK.
Vasconcelos, G., Lourenço, P.B. [2009] "In-plane experimental behaviour of stone masonry walls under cyclic
loading," Journal of Structural Engineering, Vol. 135, No. 10, pp. 1269-1277.
Yi, T., Moon, F.L., Leon, R., Kahn, L. [2006] "Lateral Load Tests on a Two-Storey Unreinforced Masonry
Building," Journal of Structural Engineering, Vol. 132, No. 5, pp. 643-652.
Research Report
6 CONCLUSIONS
6.1 Summary of Experimental Results
The full-scale experimental tests presented in previous chapters were performed on spandrel-pier
specimens with different features in terms of geometry, masonry type, masonry supporting elements
above openings (i.e. lintels, shallow masonry arches or reinforced concrete (RC) bond beams
running at bottom of the spandrel), presence/absence of tensile resistant elements (i.e. tie rods or
RC bond beams running on top of the spandrel), magnitude of axial loads on piers, and presence of
strengthening systems over the spandrel.
The experimental tests on double-leaf stone masonry spandrel elements allow the following
conclusions to be drawn:
i) The strength and the energy dissipation capacities clearly increase as the height of the spandrel
increases while the span is kept constant.
ii) The presence of an axial force in the spandrel (as in the presence of steel ties) can influence the
hysteretic behaviour and failure mode. The former was characterised by low strength and "fat"
loops in case of spandrel without tie, and high strength and thinner loops in case of spandrel with
horizontal compression imposed by a tie rod. The failure mode consisted of vertical flexural
cracks at spandrel-pier intersections in case of spandrel without tie and diagonal shear cracking
in presence of a tie. The tie rod increased both strength and deformation capacity of spandrels
without influencing significantly the initial stiffness.
iii) The presence of timber lintels well connected/engaged to adjacent piers produces a strength
increase because the lintels first behave elastically, and subsequently, even when their
connection to the masonry has subjected to cracking and slippage, the still induce a considerable
stiffening effect on the spandrel. Furthermore, well-connected lintels prevent relative dislocation
of stones, hence limiting damage in the spandrel. In case of very flexible, simply supported lintels,
the spandrel behaviour is completely different: a lower strength is attained and a softened
response takes place.
iv) If a lintel prevents the the falling of stones, large angular deformations can be attained in a
spandrel maintaining a residual shear transfer capacity.
Experimental tests on unreinforced masonry (URM) spandrels made of solid clay bricks and
composite URM-RC spandrel made of perforated clay bricks provided the following results:
i) The level of axial force in spandrel can have an influence on the failure mode of the spandrel
panel. URM spandrel elements with timber lintel experienced mixed shear-flexural cracking and
flexural cracks at spandrel ends under constant and variable axial force in spandrel, respectively.
URM spandrel elements with shallow masonry arch suffered mixed shear-flexural cracking and
mainly diagonal shear cracks in case of constant and variable axial force in spandrel,
respectively.
ii) The peak resistance of URM spandrels can be approximately the same, but the presence of a
tie rod (that provides a variable compressive load in the spandrel according to its elongation)
significantly reduces the post-peak softening behaviour.
iii) Composite URM-RC spandrels can suffer either vertical flexural cracks (running along the entire
height of the spandrel element at its ends) or stair-stepped cracks in the masonry and plastic
hinging in the RC bond beam at spandrel-pier intersections, even until rupture of longitudinal
reinforcing bars. As a result and at a given drift demand, cyclic loading in the beam induced
cracks wider than those observed in case of monotonic loading, as a result of plastic strain
accumulation in longitudinal bars from a cycle to another.
iv) Increasing the longitudinal reinforcement area in the RC bond beam and the axial stress level in
piers can increase the peak resistance of the composite URM-RC spandrel. Besides, the
increase in longitudinal reinforcement area provides a stiffer and stronger RC beam that tends
to detach from the pier at the bottom, resulting in a lower deformation demand on the beam itself
and a flexural hysteretic behaviour of the spandrel element. If also the compression level in piers
increases, the gap between the bottom of the beam and piers does not open up as much and
the deformation demand on the beam increases, producing a change from flexural failure to
shear failure.
v) The force–displacement relationship of composite URM-RC spandrels is characterised by
moderate strength degradation, "fat" loops in case of relatively flexible/strong RC beams, and
"thin" loops in case of stiffer/stronger RC beams. As the compression level in piers increases,
the hysteresis loops become fatter.
As far as the impact of spandrel type on perforated masonry walls is concerned, the full-scale lateral
loading tests on tuff stone masonry walls with single opening produced the following major findings:
i) The in-plane nonlinear response of masonry walls with openings to lateral loads is strongly
influenced by the spandrel configuration; this is particularly evident for the peak resisting base
shear, ductility capacity, ultimate drift, overstrength, and residual drift, reflecting different features
in terms of equivalent bilinear capacity systems and repairability after severe earthquake events.
ii) Low levels of axial load on piers can induce them to experience rocking behaviour with "thin"
loops, large drift capacity and small residual drifts, resulting in large drift demands on spandrel
panels above openings.
iii) In case of spandrel panels with large aspect ratio, the spandrel tends to suffer vertical flexural
cracks that can be eventually followed by diagonal shear cracking, depending on the horizontal
compression level associated with the lateral force at the height of the spandrel in a perforated
masonry wall.
iv) Timber lintels above openings that are well connected to piers play a fundamental role in
supporting the masonry of spandrel panels, forcing the latter to distribute lateral actions between
piers up to large drift levels.
v) Shallow masonry arches above openings drastically reduce the in-plane seismic capacity of
masonry walls with openings, as they can suffer progressive collapse even at small levels of
lateral drift.
vi) RC bond beams running on top of spandrels provide a coupling action to piers and are subjected
to large inelastic rotation demands imposed by piers’ rocking, hence experiencing plastic hinging
at spandrel-pier intersections. In that context, the propagation of shear cracks from the masonry
below to the RC bond beam shows that a 20 mm stirrup spacing (alike those detected in existing
masonry buildings) in the beam can limit the plastic rotation capacity.
vii) The application of inorganic matrix-grid (IMG) composite strengthening systems externally
bonded to spandrels can cause a change in the failure mode, e.g. from diagonal shear cracking
to horizontal cracking in case of spandrel with timber lintel
viii) IMG composites allow a significant increase in lateral load-bearing capacity, potentially providing
peak resistance improvements comparable to those favoured by the realization of RC bond
beams.
The lateral loading tests on vintage clay brick masonry walls with single and multiple openings and
storeys led to the following conclusions:
i) Shallow spandrel panels behaved as coupling beams, allowing multi-storey pier rocking to occur
in the two internal piers; external piers rocked over their inter-storey effective height.
ii) Pier effective height for a flexure response was found to be dependent on loading orientation for
cases where the adjacent opening heights were not identical.
Research Report
iii) The total response of perforated masonry walls (in the form of top displacement versus base
shear) was found to approximate elastic perfectly-plastic behaviour, with a non-negligible non-
linear response.
iv) The calculated level of displacement ductility was shown to be misrepresentative of the non-
linear behaviour, with a calculated average displacement ductility of 20 for the total response.
Ultimate drift was confirmed to be a representative measure of the deformation capacity of
unreinforced masonry walls when subjected to in-plane seismic loading.
v) Repair of the damaged wall with strategically placed textile reinforced mortar resulted in a
reduced elastic stiffness, an increase in lateral strength and hysteretic energy dissipation
capacity, and a similar drift capacity.
6.2 Potential for Future Research Developments
The research presented in this report can be further extended to perforated walls with spandrel types
other than those studied so far.
A first line of research development is deemed to be related to the experimental derivation of ad hoc
force–drift diagrams for spandrel elements made masonry types different from those tested so far.
Further tests could be envisaged on tuff stone masonry, concrete masonry and rubble stone
masonry, either before or after strengthening, and with or without ties, RC bond beams, timber lintels
or shallow masonry arches. Those diagrams can be either bilinear or multilinear approximations of
actual experimental envelope curves of hysteresis loops provided by cyclic tests. The effects of
several types of traditional strengthening techniques on spandrel elements, such as masonry
repointing, mortar injections and reinforced plaster, need also to be investigated and properly
characterised from the viewpoint of capacity modelling.
A second line of research development can address the implications of the additional spandrel types
to be studied on simple walls with single openings. The experimental results presented in this report
have highlighted a significant influence of the spandrel type on the in-plane seismic capacity of
perforated walls, especially in terms of peak resistance, lateral drift capacity, hysteretic damping and
residual drift.
More in general, regardless of the spandrel type being considered, further research is required to
assess the in-plane response of perforated walls with multiple storeys and openings. The simplest
case is that of “regular” perforated walls, namely those having openings with the same size and
aligned in both vertical and horizontal directions. Nevertheless, misaligned or differently-sized
openings can delineate the so-called “irregular” perforated walls which are frequently detected in
existing buildings, particularly those not designed for earthquake resistance. Therefore, the role of
spandrels in the seismic capacity of irregular walls with openings can be the subject of future
experimental programmes.
Finally, how the level of axial stress in adjacent piers can affect on one hand the spandrel behaviour
and on the other the overall seismic capacity of perforated walls is definitely a matter of further
investigation. This is the case of the spandrel types considered in this report and other types
identified above.
Seismic response of unreinforced masonry buildings is strongly influenced by in-plane behaviour and damage of spandrels.
In the last decade, this issue has been one of the most investigated topics in earthquake engineering research programmes
dealing with masonry structures. Several working groups in different countries have focused on numerical simulation and
experimental testing of spandrels, considering their variability in terms of masonry type, presence of tensile-resistant
elements such as steel ties and reinforced concrete bond beams, magnitude of gravity loads, and spandrel geometry.
Some research groups have also explored the role of innovative strengthening systems aimed at increasing strength and/
or deformation capacity of spandrel panels above openings.
This report provides a comprehensive discussion of experimental researches carried out by four research groups in Italy,
Switzerland and New Zealand. In all cases, most of experimental tests were carried out on full-scale masonry specimens,
either focusing on spandrel panels or addressing the pier-spandrel interaction within in-plane laterally loaded walls. More
in detail, this report is asimed at reflecting and presenting the complementary nature of recent researches on spandrels.
Special emphasis is given to observed damage, force–displacement behaviour and nonlinear capacity measures of
spandrels. Valuable data on energy dissipation capacity and ultimate drift of spandrel panels and perforated masonry
walls are also reported and discussed. In-plane overstrength and displacement ductility capacity of masonry walls with
single openings are characterised through bilinear idealisation of experimental force–displacement diagrams. All testing
programmes show that the geometrical and construction features of spandrels significantly influence the in-plane seismic
capacity of perforated masonry walls and their repairability after cyclic loading. The latter feature is measured through the
ratio of residual drift to the maximum drift of each wall specimen. Rocking behaviour of piers notably increases demand on
spandrel panels, inducing the formation of plastic hinges in reinforced concrete bond beams at spandrel-pier intersections.
Finally, seismic capacity of perforated walls is also found to depend on the spandrel-pier connection, highlighting the
influence of boundary conditions on nonlinear behaviour of spandrel panels.
The type and amount of experimental data collected in this report can support the improvement of macro-element capacity
models and building codes for seismic performance assessment of masonry buildings. Several issues require further
numerical and experimental investigation and are identified in each section of the report. The extension of experimental
testing to full-scale perforated masonry walls with multiple storeys and openings is one of those research needs, starting
from preliminary findings on half-scale specimens presented herein.

Research Report 01 - 2016

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Research Report 01 - 2016

Transféré par

Droits d'auteur :

Formats disponibles

EXPERIMENTAL RESEARCHES ON THE SEISMIC BEHAVIOUR

OF MASONRY SPANDRELS: AN INTERNATIONAL PERSPECTIVE

According to law, EUCENTRE Foundation trademark cannot be reproduced, copied or utilized,

© Copyright 2016 - EUCENTRE

prodotto e distribuito da:

The research described in this report was partially funded by:

TABLE OF CONTENTS ................................................................................................................ III

LIST OF TABLES .........................................................................................................................VII

LIST OF FIGURES ......................................................................................................................VIII

2 EXPERIMENTAL CAMPAIGN ON DOUBLE-LEAF STONE MASONRY SPECIMENS AT THE

3 MONOTONIC AND CYCLIC TESTS ON TUFF STONE MASONRY SPANDRELS AT THE

4 QUASI-STATIC CYCLIC AND MONOTONIC TESTS ON MASONRY SPANDRELS AND

5 SPANDREL TESTING A THE UNIVERSITY OF AUCKLAND ................................................ 97

6 CONCLUSIONS .................................................................................................................... 129

1.1 Motivation for the Report

1.2 Aim and Objectives

1.3 Organisation of the Report

2 EXPERIMENTAL CAMPAIGN ON DOUBLE-LEAF STONE MASONRY SPECIMENS AT

Seismic response of unreinforced masonry (URM) structures is influenced by the behaviour of

2.2 Description of the Experimental Campaign

2.3 Double-Leaf Stone Masonry Walls

2.1. Compression Tests on Stone Masonry Specimens

2.3.3 Vertical Compression Test

2.3.4 Diagonal Compression Test

fm [MPa] E [MPa] ft [MPa] G [MPa]

Mean 3.28 2550 0.137 840

2.4 Description of the Spandrel Specimens

Figure 2.5. Picture of specimen S1 during and after construction.

2.5 Properties of the Test Setup

Figure 2.6. Static scheme of the specimen during a test.

Figure 2.7. Scheme of the test setup.

In particular, the rotation at the base of the piers, α, is given by:

Figure 2.9. Front view of the test apparatus.

2.6 Instrumentation of the Specimens

2.6.1 Actuators Force Transducers

2.6.2 Displacement Transducers (Potentiometers)

Figure 2.11. Layout of the displacement transducers on the specimen (front/rear).

2.6.3 Advanced Video Acquisition

2.7 Test Procedure

2.7.1 Loading History of the First Spandrel Specimen (S1)

2.7.2 Loading history of the Second Spandrel Specimen (S2)

Nominal Cycle Total

2.7.3 Loading History of the Third Spandrel Specimen (S3)

Nominal Cycle Total

2.7.4 Loading History of the Fourth Spandrel Specimen (S4)

2.8 Test Results for the First Specimen

2.8.1 Test Observations

Nominal element drift θn [%]

2.8.2 Force–Displacement Behaviour of the First Specimen

2.9 Test Results for the Second Specimen

2.9.1 Test Observations

Nominal element drift θn [%]

2.9.2 Force–Displacement Behaviour of the Second Specimen

2.10 Test Results for the Third Specimen

2.10.1 Test Observations

Nominal element drift θn [%]

Figure 2.32. Rear crack pattern of spandrel at C11 (θn = 0.8%).

2.10.2 Force–Displacement Behaviour of the Third Specimen

Equivalent base rotation α [%]

2.11 Test Results for the Fourth Specimen

2.11.1 Test Observations

Nominal element drift θn [%]

Figure 2.39. Rear crack pattern of spandrel at C11 (θn = 0.8%).