Académique Documents
Professionnel Documents
Culture Documents
INTRODUCTION
Loads due to heavy vehicles are often a critical design case for buried arch structures,
especially for those subject to very high loads from off-road vehicles, or for long span structures
subject to highway loading. The typical design procedures used for these structures employ a
2D plane strain finite element analysis, which models the distribution of the wheel loads through
the fill in the plane of the analysis, but not in the transverse direction. Provisions for distribution
of wheel loads through fill in the Australian Standard Bridge Design Code, AS 5100(1), are
based on classical elastic solutions for stress distribution in uniform materials, which do not take
account of non-linear soil behaviour, or of 3D interaction effects between the soil and the
structure.
This paper presents a brief review of previous research, and the results of 3D finite element
analyses of two typical arch structures, with three different longitudinal configurations, and 3
different fill heights; a total of 18 separate analyses. For each analysis the maximum
increments in bending moments are compared with the results of a 2D plane strain analysis with
wheel loads distributed in accordance with AS 5100.
The arch structures analysed were:
1. A typical mine haul road structure with loads due to maximum current weight earth moving
trucks (635 tonne)
2. A typical road structure, carrying a multi-lane road or rail line under a road with M1600
loading to AS 5100.
th
Internal span, m
10.6
18.6
Internal height, m
7.0
8.0
Thickness, mm
300
400
CAT 797
M1600
6237
1440
4158
120
Axle width, m
9.53
2.4
Load type
PREVIOUS RESEARCH
The majority of previous research on the distribution of road vehicle wheel loads through fill
relates to loads on rectangular section structures, such as box culverts, or loads on pipes,
where the ratio of fill height to structure span is typically much greater than for the structures
considered in this paper. Soil-structure interaction effects are much less significant in these
structures, and a simple 2D analysis, with distribution through the fill based on elastic behaviour,
is appropriate. In addition to the Australian Bridge Design Code similar provisions are found in
most international codes, such as the AASHTO bridge code (2), (3).
Research considering longitudinal soil-structure interaction effects is much more limited.
Examples include a report from the University of Kansas (Han, Acharya et al. (4)) primarily
concerned with the effect of the road pavement on load distribution, and detailed analysis and
testing of wheel load effects on corrugated steel culverts by Moore, Brachman et al. (5). These
structure types have significantly different behaviour from reinforced concrete arch structures,
and neither of these reports considered the effect of different longitudinal connectivity
arrangements on the structural behaviour.
Research into appropriate 2D analysis procedures is much more extensive, including research
related to the CANDE analysis software by The Transportation Research Board (6), and a
number of papers by Jenkins (711). Research work includes independent analyses using
different finite element programs and monitoring of structures during construction. Procedures
for analysis of buried arch structures under uniform soil loading are therefore well established
and verified, but this level of verification has not yet been applied to the effects of wheel loads.
th
The model included the arch structure, surrounding fill, and foundation soils, with
boundaries sufficiently far from the structure to have negligible effect on the structure
behavior.
Fill soils were modelled as non-linear materials with a Mohr-Coulomb failure criterion,
using typical short-term stiffness properties.
Arch and footing concrete, and foundation soils were modelled as linear-elastic in
accordance with usual design practice.
The analysis modelled the construction sequence, with fill materials placed in layers on
alternate sides of the arch, with a top of fill level difference of 1 metre.
Soil elements were connected to the arch with friction elements allowing sliding of the
soil around the arch.
After completion of the construction stage modelling to finished ground level a uniform
load of 30 kPa was applied to one half of the model, to check that the 2D and 3D models
gave similar results under uniform loading.
Point loads representing the design vehicles were then applied in the position found to
generate the maximum absolute moment increment in the arch structure. Further details
of the loads applied are given below.
No attempt was made to model the effect of concrete or asphalt paving, but the thin top soil
layer was given a small cohesion (compared with the strength of typical paving materials) of 20
kPa to help convergence of the model under high wheel loading.
Materials properties used in the analyses are shown in Table 2, and typical model layouts for the
2D and 3D analyses are shown in Figures 1 and 2 respectively, with details showing the use of
friction elements in Figure 3. The fill elastic modulus values were relatively high values,
intended to model typical behavior for good quality granular fill under short term loading. For
ultimate design analyses much lower values are appropriate, to model lower bound behavior
under long term loading. Reasons for the use of different friction angles of the fill materials
between the 2D and 3D models are discussed below.
2D Analysis
For the 2D analysis the soil was modelled with 8 node 2D plane strain plate elements, and the
arch with a series of short beam elements, connected to the soil with friction elements with a
length of half the arch section thickness. The arch elements were modelled as pinned at the
base and the crown. Boundary nodes were restrained in the X and Y directions at the base,
and in the X direction only at the sides.
A non-linear static analysis was carried out on each construction stage, with both material and
geometric non-linearity included. A 10 kPa compaction load was applied to the top of each fill
layer.
For the final fill stage the 10 kPa compaction load was applied and removed, then the following
live loads were applied:
Point loads representing the design heavy vehicle load, distributed over the fill depth in
accordance with AS 5100 requirements.
The vehicle loads were applied in the position that resulted in the greatest bending moment
increment in the 3D analysis.
th
3D Analysis
20,000
20,000
125
125
50
50
30
25
90
90
34
30
0.3
0.3
0.5
0.5
th
The 2D model used for comparison with the 3D results had a mesh with the same element sizes
as a transverse plane through the 3D model. As a check on the sensitivity of the model to the
size of the mesh the 2D mesh was subdivided, increasing the total number of plate elements
from 239 to 669. Figure 3 shows a detail of the subdivided mesh. The maximum bending
moment under completed fill plus truck loading increased by only 2% in the finer mesh.
3D analysis
For the 3D analysis the soil was modelled with 8 node brick elements, and the arch with 8 node
plate shell elements, connected to the soil with friction elements in the same way as the 2D
model. The 3D model was generated by extruding the 2D model in the Z direction, with an
overall length of 20 metres. Base nodes were restrained in all three directions, the ends parallel
to the YZ plane were restrained in the X direction, and the sides parallel to the XY plane in the Z
direction. The XY plane was treated as a plane of symmetry, with the main vehicle path centre
line lying on this plane.
The precast arch elements were modelled as 1 metre wide plates, using rigid links to join plates
to model the three different longitudinal configurations.
The 3D analysis was carried out in layers, with 10 kPa compaction load, followed by a 30 kPa
load over half the model, as for the 2D analysis. Point loads representing the vehicle loading
were generated by the Strand7 moving load analysis, and the load position that generated the
greatest absolute moment increment was extracted for replication in the 2D analysis. For the
mining truck loading the applied loads were equivalent to a single truck. For the M1600 loading
the applied loads were equivalent to 3 loaded lanes. Since the loading was applied as a
symmetrical load the full load was applied to the central lane, and the adjacent lane had a
reduction factor of 0.8. The loading was therefore equivalent to one lane loaded with a factor of
1.0, and both adjacent lanes with a factor of 0.8, rather than the 0.8 and 0.4 factors for the
second and third loaded lanes, as specified in AS 5100.
ANALYSIS RESULTS
The analysis results are summarised in Figures 4 to 17. Figures 4 to 11 show results under
dead load or uniform live load, showing that the 2D and 3D analyses produced consistent
results under these conditions. Figures 12 to 17 show the bending moment increment due to
heavy vehicle loading, and show significantly different results in the 2D and 3D analyses. In
each of the graphs:
The X axis shows distance from the arch crown, measured around the arch.
Each graph shows results (deflections or bending moments) around the arch for a
specified arch type (Type 1 or Type 2) and fill height (1.5 m, 3 m or 6 m).
The 2D results are not affected by the longitudinal configuration. Results are shown as
2D-1 or 2D-2.
Each graph shows 3D results for each of the longitudinal configurations (1 = segmental, 2
= offset segmental, 3 = continuous), shown as 3D1-1, 3D1-2, 3D1-3 for the Type 1 arch
(10.6 m span), or 3D2-1, 3D2-2, 3D2-3 for the Type 2 arch (18.6 m span).
th
stiffness of the 3D model is due to the use of 8 node brick elements. Better results may be
achievable with 16 or 20 node bricks, at the expense of considerably increased run-time.
The bending moment increment due to the uniform live load of 30 kPa is shown in Figures 10
and 11, for Arch Type 1 and 2 respectively. The bending moments for the Type 2 Arch showed a
good match between the 2D and 3D analyses, and results were acceptably close for Type 1.
Deflection, mm
10
-10
5
0
-5
10
-5
-10
-15
Distance From Crown, m
2D-1
3D-1-1
3D-1-2
3D-1-3
Deflection, mm
-10
0
-5
10
-5
-10
-15
Distance From Crown, m
2D-1
3D-1-1
3D-1-2
3D-1-3
th
70
60
50
40
30
20
10
-10
-5
0
0
-10
Distance From Crown, m
2D-1
3D-1-2
3D-1-1
10
3D-1-3
Deflection, mm
-14
2
0
-9
-4
11
-2
-4
Distance From Crown, m
2D-2
3D-2-1
3D-2-2
3D-2-3
-9
-4
11
Deflection, mm
-5
-10
-15
-20
-25
Distance From Crown, m
2D-2
3D-2-1
3D-2-2
3D-2-3
th
100
-14
50
0
-9
-4
11
-50
-100
-150
Distance From Crown, m
2D-2
3D-2-1
3D-2-2
3D-2-3
20
10
0
-10
-5
-10
10
-20
-30
-40
-50
Distance From Crown, m
2D-1
3D-1-1
3D-1-2
3D-1-3
Figure 10: Bending Moments Increment for Uniform 30 kPa Load; Arch Type 1.
Bending Moment; Type 2 Arch, Inc. 12-11, 1.5 m Cover
60
40
20
-14
0
-9
-4
-20
11
-40
-60
-80
-100
-120
2D-2
3D-2-1
3D-2-2
3D-2-3
Figure 11: Bending Moments Increment for Uniform 30 kPa Load; Arch Type 2.
th
50
-10
0
-5
10
-50
-100
-150
-200
-250
Distance From Crown, m
2D-1
3D-1-1
3D-1-2
3D-1-3
Figure 12: Bending Moments Increment for CAT 797 Load; Arch Type 1; 1.5 m Cover.
Bending Moment; Type 1 Arch, Inc. 16 - 12, 3.0 m Cover
60
40
-10.0
20
0
-5.0
-20
0.0
5.0
10.0
-40
-60
-80
-100
-120
Distance From Crown, m
2D-1
3D-1-1
3D-1-2
3D-1-3
Figure 13: Bending Moments Increment for CAT 797 Load; Arch Type 1; 3.0 m Cover.
Bending Moment; Type 1 Arch, Inc. 18 - 16, 6.0 m Cover
10
5
-10
0
-5
-5
10
-10
-15
-20
-25
-30
-35
Distance From Crown, m
2D-1
3D-1-1
3D-1-2
3D-1-3
Figure 14: Bending Moments Increment for CAT 797 Load; Arch Type 1; 6.0 m Cover.
th
20
-14
0
-9
-4
-20
11
-40
-60
-80
-100
-120
Distance From Crown, m
2D-2
3D-2-1
3D-2-2
3D-2-3
Figure 15: Bending Moments Increment for M1600 Load; Arch Type 2; 1.5 m Cover.
Bending Moment; Type 2 Arch, Inc. 18-14, 3.0 m Cover
30
20
10
-14
0
-9
-4
-10
11
-20
-30
-40
-50
-60
-70
Distance From Crown, m
2D-2
3D-2-1
3D-2-2
3D-2-3
Figure 16: Bending Moments Increment for M1600 Load; Arch Type 2; 3.0 m Cover.
Bending Moment; Type 2 Arch, Inc. 19-16, 6.0 m Cover
15
10
-14
5
0
-9
-4
11
-5
-10
-15
-20
Distance From Crown, m
2D-2
3D-2-1
3D-2-2
3D-2-3
Figure 17: Bending Moments Increment for M1600 Load; Arch Type 2; 6.0 m Cover.
With 3.0 m fill over the crown the 2D analysis had greatly reduced moment increments for the
Type 1 section, with bending moments approximately equal to the 3D results for longitudinal
configurations 2 and 3. The Type 1 longitudinal configuration gave the highest absolute
moment increment, being about 20% higher than the other analyses. For the Type 2 section the
10
th
pattern seen with 1.5 m cover was repeated, with the 2D results being much more conservative
than any of the 3D analyses, but all bending moments were reduced by about 50%, compared
with the 1.5 m cover results.
With 6 metres cover the moment increment from the 2D analysis for Section Type 1 was
significantly less than that for any of the 3D analyses; however in this case the live loading
reduced the maximum absolute bending, so none of the results would be critical. For Section
Type 2 the 2D maximum moment increment results were similar to the 3D results for Type 2-1,
with the maximum moment from Types 2-2 and 2-3 being about 30% lower.
DISCUSSION
3D soil-structure interaction effects
For loads uniform along the length of the structure it would be expected that a 2D plane strain
analysis would give results very close to those from a 3D analysis with equivalent boundary
conditions, since strains in the Z direction will be zero, in both the 2D and 3D analyses.
Preliminary analyses for this paper found that the 3D analyses tended to be significantly stiffer
when using identical soil properties. This is believed to be because of the different behaviour of
the 8 node 2D plane strain plate element and the 8 node 3D brick element, particularly with
regard to Mohr-Coulomb stresses under low confining pressure. Fair to good agreement
between the 2D and 3D analyses was achieved by reducing the soil friction angle for the 3D
analyses.
For non-uniform loads soil-structure interaction effects can be expected to significantly change
the behaviour of the structure, when analysed in 3D:
The structure immediately under the point of loading will deflect downwards, tending to
increase the lateral spread of the applied load
Continuous structures will distribute loads to the base through bending along the length of
the structure, as well as in-plane bending.
Resisting soil pressures will be distributed along the structure, particularly for continuous
structures, or precast structures with joints longitudinally offset at the crown.
These effects would be expected to be greatest with shallow cover over the arch, where
longitudinal distribution of pressures through the soil will be restricted.
The results of the analyses were broadly in line with the expected behaviour, with all 3D
analyses with 1.5 metres cover showing large reductions in the bending moment increment due
to live load, and a greater reduction with longitudinal configuration Types 2 and 3. With
increased cover the difference between the 2D and 3D analyses reduced as expected, however
there were significant differences in the behaviour of the Type 1 and Type 2 sections, with no
obvious means of predicting this behaviour, other than carrying out a detailed 3D analysis. For
the structures examined in this report the 2D analysis results ranged between conservative and
approximately equal to the 3D results, however it is possible that in some cases a 2D analysis
with high cover will give unconservative results.
CONCLUSIONS
The major findings from the comparison of 2D and 3D finite element analyses were:
The bending moments in the structures under uniform load were sensitive to the fill friction
angle. Due to the different behavior of the 8 node brick elements, compared with the 2D 8
node plane strain elements, it was necessary to reduce the fill friction angle in the 3D
analyses to get consistent results under uniform loads.
With 1.5m fill the bending moment increment due to heavy vehicle loading was
significantly less in the 3D analyses than in the 2D analyses. The moment reduction was
11
th
greatest for the continuous structures (longitudinal configuration Type 3), and least for
precast structures with fully independent segments (Type 1), with intermediate results for
longitudinal configuration Type 2 structures.
The difference between the 2D and 3D analyses reduced with increasing fill height.
The relative difference in results between the 2D and 3D analyses was different for the two
arch sections, with no obvious simple method to predict the 3D results, other than carrying
out the full 3D analysis.
The increase in bending moment due to live load found in the 2D analyses was greater
than or approximately equal to the results from the corresponding 3D analyses for all
cases examined in this study, but it is possible that the 2D analysis might be
unconservative for high fills in some cases.
Based on the analyses described in this paper the following recommendations are made for the
analysis and specification of buried arch structures with heavy vehicle loading and/or large
spans:
Use a good quality granular select fill around the arch, with design based on long term
lower bound values for the fill stiffness and friction angle. A typical fill elastic modulus
value used for design purposes is 30-50 MPa, compared with the values of 90 MPa used
in this study.
For a cover of 1.5 m over the arch crown the 2D analysis results were found to be
conservative. Considering the uncertainties in the number and effect of long term
repetitions of maximum loads, a conservative approach is considered appropriate, and
use of 3D analysis for design purposes is not recommended without further investigation
and verification of the results.
For higher covers the 2D analysis results were found to be less conservative. A 2D
analysis in conjunction with high estimated friction angles may in some circumstances be
unconservative, but use of lower bound values for the fill friction angle and stiffness will
ensure conservative results overall.
Use of a longitudinal configuration with precast element joints offset at the crown was
found to significantly improve the distribution of wheel loads, with maximum bending
moments being reduced to close to the value achieved with a fully continuous structure.
This configuration is recommended for use where heavy vehicle loading is the critical
design load, and/or there are significant uncertainties in the magnitude or number of
repetitions of maximum vehicle loads.
Heavy vehicle loads in conjunction with shallow cover will often result in large variations in
reinforcement stresses. AS 5100 provisions for fatigue stress range limits must be
observed.
REFERENCES
1. Standards Australia, AS 5100-2004, Australian Standard Bridge Code, SAI Global,
Sydney.
2. AASHTO, AASHTO LRFD Bridge Construction Specifications, AASHTO, Washington,
USA
3. Beakley, J (2010), The Use of AASHTO LRFD Bridge Design Specifications with
Culverts, American Concrete Pipe Association, Irving, USA
4. Han, J; Acharta, R. et al. (2013), Improved Load Distribution for Load Rating of Low-Fill
Box Structures, The University of Kansas
12
th
AUTHOR BIOGRAPHIES
Doug Jenkins: Doug Jenkins is an independent consultant, specialising in the analysis and
design of bridges, abutments, buried arch structures and reinforced soil retaining walls,
including the application of advanced analysis techniques to the design of these structures. He
has 40 years experience in the design of bridges, including major projects in Australia and
many overseas countries, including UK, Middle East, Turkey, Japan and Malaysia. In addition
to design and verification of bridges and related structures his current activities include condition
surveys and reports, forensic investigations, software development, and training and lecturing.
He is currently National President of the Concrete Institute of Australia.
Chris Lawson: Chris Lawson is the Engineering Manager at Reinforced Earth Pty Ltd in
Australia. He is a civil engineer and specialises in retaining wall and bridge design, buried arch
structures and construction and maintenance for over 25 years. Chris is also a member of the
Reinforced Earth group international technical committee and is responsible for engineering and
R&D activity within the Asia Pacific region.
Copyright Licence Agreement
th
The Author allows ARRB Group Ltd to publish the work/s submitted for the 9 Austroads Bridge
Conference, granting ARRB the non-exclusive right to:
publish the work in printed format
publish the work in electronic format
publish the work online.
The Author retains the right to use their work, illustrations (line art, photographs, figures, plates) and
research data in their own future works
The Author warrants that they are entitled to deal with the Intellectual Property Rights in the works
submitted, including clearing all third party intellectual property rights and obtaining formal permission from
their respective institutions or employers before submission, where necessary.
13