Vous êtes sur la page 1sur 9

366

NOTE / NOTE

Arching effect in fine sand due to base yielding


Jamshid Sadrekarimi and Alireza Abbasnejad

Abstract: This paper presents results of an experimental work on the arching effect in loose and dense sand. The appara-
tus comprises concentric circular trapdoors with different diameters that can yield downward while stresses and deforma-
tions are recorded simultaneously. As the trapdoor starts to yield, the whole mass of soil deforms elastically. However,
after a specified displacement that depends on the trapdoor diameter and soil relative density, the soil mass behaves plasti-
cally. This behavior, which is due to flow phenomenon, continues until the stress applied onto the trapdoor decreases to a
minimum value. Then the stress carried by the trapdoor shows an ascending trend. This indicates the gradual separation of
the yielding mass from the whole soil body. Finally, the flow process creates a stable arch of sand. This process is called
the arching mechanism. Depending on the trapdoor diameter, there is a critical relative density at and beyond which the
test leads to the formation of a stable arch. The results are also compared with Terzaghi’s theory and the assumption of an
upper boundary solution is discussed.
Key words: arching effect, embankment, relative displacement, stress distribution.
Résumé : Cet article présente les résultats de travaux expérimentaux sur les effets d’arche dans des sables lâches et den-
ses. L’appareil est fait de portes circulaires et concentriques avec des diamètres différents qui peuvent atteindre leur limite
élastique en direction verticale pendant que les contraintes et déformations sont enregistrées simultanément. Lorsque la
porte atteint sa limite élastique, la masse entière du sol se déforme élastiquement. Cependant, suite à un déplacement spé-
cifique qui dépend du diamètre de la porte et de la densité relative du sol, la masse de sol se comporte de façon plastique.
Ce comportement, qui est dû au phénomène d’écoulement, continue jusqu’à ce que la contrainte appliquée sur la porte di-
minue à sa valeur minimale. Ensuite, la contrainte portée par la porte démontre une tendance à la hausse. Ceci indique
une séparation graduelle de la masse ayant atteint sa limite élastique de l’échantillon total de sol. Finalement, le processus
d’écoulement fait en sorte qu’une arche stable de sable est établie, ce qu’on appelle « mécanisme d’arche ». Selon le dia-
mètre de la porte, à une valeur égale ou au-dessus d’une densité relative critique il y a formation d’une arche stable. Les
résultats de cette étude sont aussi comparés à la théorie de Terzaghi, et les suppositions pour obtenir la solution à la fron-
tière supérieure sont discutées.
Mots-clés : effet d’arche, remblai, déplacement relatif, distribution des contraintes.
[Traduit par la Rédaction]

Introduction Bg
½3 sv ¼ ð1  eK tan4 z=B Þ for c ¼ 0; q ¼ 0
The essential features of arching were demonstrated by K tan4
Terzaghi (1943) performing experiments on sand with a
yielding trapdoor. The equations he proposed for stress due where B is the trapdoor radius, g is the soil density, c is the
to arching effect considering plastic behavior are as follows: coefficient of cohesion, K is the lateral earth pressure coeffi-
cient, 4 is internal friction angle, z is depth from the soil
B½g  ðc=BÞ surface, and q is the surcharge pressure.
½1 sv ¼ ð1  eK tan4 z=B Þ
K tan4 Finn (1963) modified the plastic behavior assumption of
for c > 0 ; q ¼ 0 Terzaghi (1943) and developed experiments assuming the
soil is linearly elastic. Burghignoli (1981) studied the arch-
Bg ing effect in the soil over a flexible trapdoor and obtained
½2 sv ¼ ð1  eK tan4 z=B Þ þ qeK tan4 z=B expressions based on relative stiffness between the trapdoor
K tan4 and the soil. Getzler et al. (1968) used models composed of
for c ¼ 0; q > 0 aluminum blocks with various shapes that were buried in
Received 4 July 2007. Accepted 8 September 2009. Published on the NRC Research Press Web site at cgj.nrc.ca on 8 March 2010.
J. Sadrekarimi and A. Abbasnejad.1 Faculty of Civil Engineering, University of Tabriz, Tabriz, East Azerbaijan, Iran.
1Corresponding author (e-mail: abbasnejad_ar@yahoo.com).

Can. Geotech. J. 47: 366–374 (2010) doi:10.1139/T09-107 Published by NRC Research Press
Sadrekarimi and Abbasnejad 367

coarse sand and a uniform pressure was applied to the sand the soil mass was in a surface dry state. The soil was classi-
surface. Hoeg (1968) designed a model tunnel with a rigid fied as SP-SM in accordance with the Unified Soil Classifi-
steel cylinder split longitudinally into 12 segments. Clough cation System (ASTM 2006). The maximum and minimum
and Duncan (1971) applied the finite element method to an- dry densities were 16.77 and 12.26 kN/m3, respectively. To
alyze retaining wall behavior using various assumptions re- determine shear strength parameters corresponding to the
garding characteristics of the wall–backfill interface. relevant stress levels, direct shear tests under 1.62 and
Atkinson and Potts (1977) examined the behavior of shal- 11.23 kPa surcharges were carried out. The results are illus-
low tunnels in dry sand using a tunnel model lined with a trated in Figs. 1 and 2. The magnitude of the internal fric-
thin rubber membrane in two laboratory situations: under tion angle, 4, depends on the magnitude of the state of
gravity and in a centrifuge with 75g acceleration. Then they stress for a particular soil (Atkinson and Potts 1977). The
compared the experimental results with the upper and lower lower the normal load, the higher the 4 angle.
boundary theory. The results showed that the upper and
lower boundary theory has proper accuracy to estimate the Model properties
loads applied to the lining of a tunnel. Koutsabeloulis and A device comprising a sand container with circular trap-
Griffiths (1989) studied the stress distribution around a trap- doors at its base was designed and constructed by the au-
door due to the arching effect using the finite element thors. The whole system is shown schematically in Fig. 3.
method. Ono and Yamada (1993) studied the arching effect The sand container was 0.358 m3 in volume and 60 cm in
in a soil mass around a tunnel and behind a retaining wall height with an octagonal horizontal cross section with a cir-
using a rubber membrane filled with air under pressure cumferential circle of 98 cm diameter. The container was
equal to that applied from the soil, then the pressure was de- made of a 4 mm thick steel plate strengthened with stiff-
creased until the air was extracted. eners. As shown in Figs. 3 and 4, three concentric circular
Hashash and Whittle (2002) presented a detailed interpre- trapdoors were mounted under the base of the container.
tation of the evolution of stresses around a braced excavation The trapdoors — 10, 20, and 30 cm in diameter — could
in a deep layer of soft clay. The boundaries of the arching yield downward separately by a very sophisticated compu-
zones for both single tunneling and parallel tunneling were terized system as shown in Figs. 3–5. The magnitude of the
determined by Lee et al. (2004). The prediction of load supported by each trapdoor was measured using a load
tunneling-induced ground movements during excavation of a cell. Displacement of the trapdoors as well as the surface of
soft-ground tunnel has been carried out using various meth- the soil mass due to trapdoor yielding were all monitored
ods, including empirical methods derived from field observa- using a linear variable differential transformer (LVDT) in-
tions (Clough and Schmidt 1981) and centrifuge modeling stalled under the platform and over the soil surface. The sys-
(Mair 1979; Wu and Lee 2003) or numerical methods (Lee tem was designed to allow each trapdoor to yield separately
and Rowe 1991). Park and Adachi (2002) performed model by unscrewing the holding nuts.
tests under 1g acceleration conditions to simulate tunneling
events in unconsolidated ground with various levels of in- Test procedure
clined layers. Stone and Newson (2002) presented the results The container was filled with loose sand using a sand
of a series of centrifuge tests designed to investigate the ef- raining device. To produce dense sand, the container was
fects of arching on the soil–structure interaction. filled with layers of sand while they were compacted evenly
Continuum approaches describing the stress dip under with a 4.54 kg rammer. Each layer of soil was 5 cm in
sand piles can be found in Wittmer et al. (1997), Savage thickness. The falling height of the rammer and number of
(1997), and Didwania et al. (2000); and a comprehensive re- blows varied depending on the expected densities. This stage
view of continuum efforts was presented by Savage (1998). was very time-consuming and several tests were carried out
Michalowski and Park (2004) focused on the limit analysis to make sure the soil density was identical throughout the
approach and admissible radial stress fields in prismatic sand mass. After filling the container with sand, the nuts
sand piles. To the best of the authors’ knowledge, in spite holding the trapdoor were unscrewed while the upward pres-
of the presence of a vast amount investigations on soil arch- sure on the trapdoor was being adjusted to ensure that the
ing, there is no ample study into the parameters affecting the trapdoor did not displace. This was a final stage of the test.
arching mechanism in a soil mass and the stress and strain At this stage the recorded stress was very close to the geo-
path. This paper presents results of an experimental work in static pressure, s0 = gh, where g is the density of the sand
which variations of the normal stress in a soil mass carried and h is the height of the sand mass in the container. Fol-
out using circular trapdoors were monitored while the trap- lowing this stage, the trapdoor was slowly yielded down-
doors were yielding gradually. ward by loosening the load cell major screw. This trend
was continued until the load displayed by the load cell
Procedure tended towards a constant value.
Soil properties
The test soil was natural cohesionless silty sand with Results
100% and 9% passing the No. 10 and No. 200 sieves, re- The test results with 10, 20, and 30 cm diameter trapdoors
spectively. The coefficient of curvature, Cc, coefficient of are depicted in Figs. 6–8, respectively. In these figures, s/s0
uniformity, Cu, and specific gravity of the solid particles is the ratio of normal stress that is applied onto the trapdoor
were 1.1, 5.3, and 2.61, respectively. The natural moisture during any stage of yielding to the same stress at the initial
content of soil mass was 3% and at this moisture content stationary state. This stress ratio defines the stress reduction

Published by NRC Research Press


368 Can. Geotech. J. Vol. 47, 2010

Fig. 1. Internal friction angle, 4, versus relative density, Dr. R2, regression squared; sn, normal stress.

Fig. 2. Dilation angle, n, versus relative density, Dr.

Fig. 3. Schematic diagram of apparatus. All dimensions in centimetres.

Published by NRC Research Press


Sadrekarimi and Abbasnejad 369

Fig. 4. Detail of trapdoors, load cell, and displacement gauge.

Fig. 5. General view of the test system.

level due to the arching effect. DH is the trapdoor down- creases sharply as the trapdoor yields. At this stage the
ward yield value. whole mass of sand behaves mostly elastically. As the trap-
door yield continues, the stress ratio decreases and tends to-
Discussion ward a minimum value and then increases again until it
reaches an ultimate constant level. This trend is true for all
Main aspects of the phenomenon the trapdoors. However, as the diameter of the trapdoor in-
Referring to Figs. 6–8, it is observed that at the early creases and (or) the relative density of the test soil de-
stage of the tests the stress applied against the trapdoor de- creases, the minimum and ultimate stress ratios both

Published by NRC Research Press


370 Can. Geotech. J. Vol. 47, 2010

Fig. 6. Stress ratio–yield plots for 10 cm diameter trapdoor.

Fig. 7. Stress ratio–yield plots for 20 cm diameter trapdoor.

increase. This behavior may be justified as follows. As the Then, continuing the downward displacement and decreas-
trapdoor yield starts, the overlying soil weight that is carried ing the stress ratio to a minimum value, failure occurs. At
by the trapdoor is transmitted gradually onto the container failure state, depending on the trapdoor diameter, relative
base surrounding the trapdoor. For this reason at initial stage density, and dilation angle of the sand, the failing sand
of the test, in which the sand mass behaves mostly elasti- mass dilates and imposes further stress onto the trapdoor.
cally, a minute yield is followed by a sharp decrease in the This phenomenon continues until the failure surface is de-
stress carried by the trapdoor. As the trapdoor yield pro- veloped and the yielded mass of sand departs from the
ceeds, random plastic points in the sand mass are estab- whole mass. Following this stage there is no longer any
lished. At this stage, stress adjustment due to trapdoor stress or mass exchange between two parts. Accordingly,
yielding is not immediate and occurs with some amount of the load cell displays a constant value. It was observed that
time lag. This is attributed to the flow phenomenon that oc- as the relative density of the test soil was decreased to a
curs due to the plastic behavior of the yielding sand mass. specified value, no stable arch was established and the fail-

Published by NRC Research Press


Sadrekarimi and Abbasnejad 371

Fig. 8. Stress ratio–yield plots for 30 cm diameter trapdoor.

Fig. 9. Stable arch for 20 cm diameter trapdoor.

ure surface developed up to the surface of the sand mass. A longer causes any effect on the stress applied onto the trap-
representative stable arch for the 20 cm diameter trapdoor door. Line A may be accepted as a line that defines the min-
and relative density Dr = 38.60% is shown in Fig. 9. Fig- imum stress ratio. It shows that at Dr = 39% the stress ratio
ure 10 shows the collapse of soil mass caused by an unsta- tends to be a minimum value. This is probably due to the
ble arch for the same trapdoor diameter and Dr = 16.52%. fact that as the Dr value exceeds 39%, the soil density in-
Referring to Figs. 6–8, the ultimate stresses applied onto creases and causes a heavier mass of soil to collapse. Fig-
the trapdoor are determined and plotted against relative den- ure 12 shows the pressure imposed on the trapdoor versus
sity in Fig. 11. It is observed that the ultimate stress ratio the relative density of the test soil. In this case, similar to
generally decreases as the relative density increases. How- the ultimate stress case, the critical and minimum relative
ever, up to a certain Dr value, say 28% for the 20 cm diam- densities may be defined.
eter trapdoor, no stable arch is established. As the relative
density increases the failure mode changes from sand flow Comparison of experimental results with Terzaghi’s
to arch formation. Line B may be adopted as the stable and theory
unstable arch boundary. As the relative density increases the Referring to the Terzaghi’s (1943) procedure explained
stress approaches a minimum value. At this relative density above, the stress ratios applied onto the trapdoors were cal-
the pressure applied against the buried trapdoor is at a mini- culated for different relative densities and the results are
mum. Then, further increase in soil relative density increases shown in Fig. 13. Equation [3] was employed to carry out
the stress carried by the trapdoor until it tends toward an calculations; and, for each test, sv was obtained from this
asymptote. In other words, the relative density increment no equation. The coefficient of cohesion, c, was set to zero and

Published by NRC Research Press


372 Can. Geotech. J. Vol. 47, 2010

Fig. 10. Unstable arch for 20 cm diameter trapdoor.

Fig. 11. Ultimate stress (sUltimate) ratio – relative density plots.

Fig. 12. Minimum stress (sminimum) ratio – relative density plots.

Published by NRC Research Press


Sadrekarimi and Abbasnejad 373

Fig. 13. Stress applied onto the trapdoor obtained from Terzaghi’s (1943) theory.

Fig. 14. Comparison of cones with 2n and 24 vertex angles with test results.

there was no surcharge pressure on the sand, therefore q = 0. Comparison of results with the upper boundary theory
Note that according to Trezaghi’s method, the soil mass Lower boundary theory assumes that the soil behaves
resting above 2.5 times the trapdoor diameter was assumed elastically and there is no failure zone in the soil mass.
as surcharge. Referring to Fig. 13, it is seen that the stress However, in upper boundary theory a complete failure in
ratio s/s0 increases as the relative density increases. Also, the soil mass is assumed so that the failure surface above
comparing Fig. 13 with Figs. 6–8 reveals that the stress val- tunnel shapes a cone with 2n vertex angle, in which n is the
ues obtained from Terzaghi’s theory correspond to the stress dilation angle of the soil (Atkinson and Potts 1977). In
values measured at low strains, say DH << 0.1 mm. Fig. 14, the separated soil cones with 2n and 24 vertex an-

Published by NRC Research Press


374 Can. Geotech. J. Vol. 47, 2010

gles and those obtained from the current experiment are de- Clough, G.W., and Schmidt, B. 1981. Design and performance of
picted, in which 4 is internal friction angle. It is seen that excavation and tunnels in soft clay. In Soft clay engineering.
the theoretical cones with a 2n vertex angle are very conser- Elsevier, Amsterdam. pp. 600–634.
vative, as they are quite distinct from the experimental Didwania, A.K., Cantelaube, F., and Goddard, J.D. 2000. Static
vaults. It appears that if the vertex angle is taken as 24 multiplicity of stress states in granular heaps. Proceedings of
rather than 2n it would be more justifiable, although with a the Royal Society of London, Series A, 456: 2569–2588.
24 angle the load due to arching is underestimated some- Getzler, Z., Komornik, A., and Maturik, A. 1968. Model study on
what. arching above buried structures. Journal of the Soil Mechanics
and Foundations Division, ASCE, 94(5): 1123–1141.
Hashash, Y.M.A., and Whittle, A.J. 2002. Mechanism of load trans-
fer and arching for braced excavation in clay. Journal of Geo-
Summary and conclusions technical and Geoenvironmental Engineering, ASCE, 128(3):
The relative density of soil and trapdoor diameter are both 187–197. doi:10.1061/(ASCE)1090-0241(2002)128:3(187).
dominant factors affecting the formation of a stable arch. As Hoeg, K. 1968. Stresses against underground structural cylinders.
the trapdoor yields, following a small initial mostly elastic Journal of the Soil Mechanics and Foundations Division, ASCE,
strain, the soil mass deforms plastically with larger strain 94(4): 833–858.
rates and pressure applied onto the trapdoor decreases to a Koutsabeloulis, N.C., and Griffiths, D.V. 1989. Numerical model-
minimum value. Then, as the trapdoor yield continues, de- ing of the trapdoor problem. Géotechnique, 39(1): 77–89.
pending on the dilation angle and relative density of the test doi:10.1680/geot.1989.39.1.77.
soil, the stress level on the trapdoor increases gently and fi- Lee, K.M., and Rowe, R.K. 1991. An analysis of three-dimensional
nally tends toward a constant value. At this stage the yield- ground movements: the thunder bay tunnel. Canadian Geotech-
ing sand mass separates from the whole mass. nical Journal, 28(1): 25–41. doi:10.1139/t91-004.
Lee, C.J., Chiang, K.H., and Kou, C.M. 2004. Ground movement
There is a critical relative density below which no stable
and tunnel stability when tunneling in sandy ground. Journal of
arch is formed. Also, there is an optimum relative density
the Chinese Institute of Engineers, 27(7): 1021–1032.
at which the stress applied onto the trapdoor is at a mini-
Finn, W.D. 1963. Boundary value problems of soil mechanics.
mum value. Beyond this density, the stress level gently in- Journal of the Soil Mechanics and Foundations Division, ASCE,
creases and ultimately tends toward a constant value 89(SM5): 39–72.
(Figs. 11, 12). Mair, R.J. 1979. Centrifugal modeling of tunnel construction in soft
As the trapdoor yields, a lower and upper boundary for clay. Ph.D. thesis, University of Cambridge, Cambridge, UK.
the stress applied onto the trapdoor can be defined. Michalowski, R.L., and Park, N. 2004. Admissible stress fields and
From Terzaghi’s (1943) theory, the stress ratio s/s0 in- arching in piles of sand. Géotechnique, 54(8): 529–538. doi:10.
creases as the relative density increases. However, the cur- 1680/geot.54.8.529.52008.
rent work proves that the ultimate stress decreases as the Ono, K., and Yamada, M. 1993. Analysis of arching action in gran-
soil relative density increases. This may be attributed to the ular mass. Géotechnique, 43(1): 105–120.
fact that the stress values obtained from Terzaghi’s theory Park, S.H., and Adachi, T. 2002. Laboratory model tests and FE
correspond to very low displacements. analyses on tunneling in the unconsolidated ground with in-
Comparing the results with the upper boundary theory, it clined layers. Tunnelling and Underground Space Technology,
is seen that the theoretical cones with 2n vertex angle are 17(2): 181–193. doi:10.1016/S0886-7798(02)00003-2.
conservative, as they are quite distinct from the experimen- Savage, S.B. 1997. Problems in the statics and dynamics of granu-
tal vaults. It appears that it would be more justifiable if the lar materials. In Powders and Grains ’97, Proceedings of the
vertex angle was taken as 24 rather than 2n. Third International Conference, Durham, N.C., 18–23 May
1997. Edited by R.P. Behringer and J.T. Jenkins. A.A. Balkema,
Rotterdam, the Netherlands. pp. 185–194.
Savage, S.B. 1998. Modeling and granular material boundary value
References problems. In Physics of dry granular media. Edited by H.J. Herr-
ASTM. 2006. Standard practice for classification of soils for engi- mann, J.-P. Hovi, and S. Luding. Kluwer Academic Press, Inc.,
neering purposes (Unified Soil Classification System). ASTM Dordrecht, the Netherlands. pp. 25–95.
standard D2487. American Society for Testing and Materials, Stone, K.J.L., and Newson, T.A. 2002. Arching effects in soil–structure
West Conshohocken, Pa. interaction. In Proceedings of the 4th International Conference on
Atkinson, J.H., and Potts, D.M. 1977. Stability of a shallow circu- Physical Modeling in Geotechnics – ICPMG 02, St. John’s,
lar tunnel in cohesionless soil. Géotechnique, 27(2): 203–215. Nfld., 10–12 July 2002. Edited by R. Phillips, P.J. Guo, and
doi:10.1680/geot.1977.27.2.203. R. Popescu. A.A. Balkema, Lisse, the Netherlands. pp. 935–939.
Burghignoli, A. 1981. Soil interaction in buried structures. In Pro- Terzaghi, K. 1943. Theoretical soil mechanics. John Wiley and
ceedings of the Tenth International Conference on Soil Me- Sons, New York.
chanics and Foundation Engineering, Stockholm, Sweden, 15– Wittmer, J.P., Cates, M.E., and Claudin, P. 1997. Stress propaga-
19 June 1981. Balkema, Rotterdam, the Netherlands. Vol. 2, tion and arching in static sand piles. Journal de Physique. I,
pp. 69–74. 7(1): 39–80. doi:10.1051/jp1:1997126.
Clough, G.W., and Duncan, J.M. 1971. Finite element analyses of Wu, B.R., and Lee, C.J. 2003. Ground movements and collapse
retaining wall behavior. Journal of the Soil Mechanics and mechanisms induced by tunneling in clayey soil. International
Foundations Division, ASCE, 97(12): 1657–1673. Journal of Physical Modelling in Geotechnics, 3(4): 13–27.

Published by NRC Research Press

Vous aimerez peut-être aussi