Geotechnical Performance of Qanatsduring the 2003 Bam, Iran, Earthquake
Frederic Pellet,a… Kambod Amini Hosseini,b… Mohammad Kazem Jafari,b…
Fatma Zohra Zerfa,a… Mohammad Reza Mahdavifar,b… andMohammad Keshavarz Bakhshayeshb…
The Bam, Iran, earthquake caused ground failure in several locations inand around Bam and Baravat, including collapse of existing qanats �traditionalunderground irrigation tunnels�, local soil block falls, and landslides.However, no evidence of liquefaction was observed in the area. Groundfailures were located and characterized during the site investigations carriedout a few days after the earthquake. In addition, aerial photos taken two daysafter the event were evaluated and digitized in GIS. Based on these studies,land subsidence due to collapse of qanats, local toppling, and sliding of soilblocks were mapped. Instabilities of qanats were evaluated using 2-D and 3-Dmodels, and the results were compared with existing conditions. Goodagreement was observed between the results of the numerical modeling andcondition of the qanats. �DOI: 10.1193/1.2123247�
INTRODUCTION
On 26 December 2003 at 01:56:56 GMT, a destructive earthquake struck southeastof Iran and destroyed the city of Bam, Baravat, and some small villages at the area.Based on official data, more than 26,000 persons died during this event and many morewere injured. The Mw was 6.5, with a focal depth of about 8 km. Based on the recordedmotion at Bam station, the PGA values are 0.98 g, 0.78 g, and 0.62 g in the verticalcomponent, normal, and parallel to the fault direction, respectively �Eshghi and Zare2003�.
Due to this event, several ground failures were identified, including land subsidenceand landslides. No evidence of liquefaction was observed. Collapse of the qanats wasthe most dominant type of ground failure in the area, and several sinkholes were alsoinduced during the main shock and aftershocks. This damage interrupted the irrigationsystem in the area and increased damage to lifelines and properties.
GEOGRAPHY AND GEOLOGICAL SETTING
Bam is located 175 km southeast of Kerman City. Bam has a total area of about5,400 hectares. Topographically, most parts of the city are flat, with volcanic hills in thenorth and southwest parts of the Bam area. The average altitude of the city is about
a� Laboratory 3S, University Joseph Fourier, 38041, Grenoble, Cedex 9, Franceb� International Institute of Earthquake Engineering and Seismology, 26, Arghavan St. N. Dibaji, Farmanieh,
1,050 m above sea level. Due to the dry climatic condition, qanats and deep wells arethe main sources of water for drinking and irrigation. The groundwater level is estimatedto be 30 m below ground surface in most parts of Bam and its vicinity.
Different types of lithology can be observed at the area: recent and late Quaternaryalluvium; Paleocene sedimentary rocks; Eocene volcanic rocks; and intrusive igneousrocks �granodiorite�. Most parts of Bam and Baravat are covered by Quaternary depos-its, including: yellow to brown sand and silt �Qm1�; brown gravel, sand and silt depositsdue to seasonal flooding �Qm2�; coarse grain gravel of alluvial fans �Qf2� and coarsegrain riverine deposits. The deposit �Qm1� along the Bam fault has about five degrees ofdip toward the east. The compaction of this layer is lower than that of the other deposits,although it is older, and frequent deep erosion of this layer can be observed. Distributionof Qm2 deposits �including gravel, sand, and silt, including some thin layers of fine grainsediments as lenses� is widespread at Bam and Baravat. The thickness of these denselayers ranges from a few meters to more than 50 meters. At some locations, there isweak cementation due to infiltration of surface water among these deposits. Shear wavevelocity at this layer is about 600 m/s at a depth of 5 m based on in situ measurementscarried out by IIEES.
The next Quaternary deposits in the eastern parts of Bam are alluvial fans �Qf2�, in-cluding coarse grain sediments. The thickness of these deposits at some places is about100 m. Alluvial fans and terraces of �Qal� are the youngest deposits in the area that ex-tends along the route of the existing seasonal river �Posht-e-Rood�. These sediments arequite loose, without cohesion and cementation.
Bam Citadel, located in the northwest of the city, is the only site where a rock out-crop can be observed. This outcrop consists of andesite and basalt without considerableweathering.
Figure 1 shows the geological map of the area prepared by IIEES. This effort wasbased on the existing geological map of the Geological Survey of Iran �GSI�. Figures 2and 3 depict two schematic geological sections. These figures are based on the data col-lected during site investigation and geophysical explorations. As Figures 2 and 3 illus-trate, the geological conditions at the site are complex and could affect site response.
LANDSLIDES AND SOIL BLOCK FALLS
Based on evaluation of aerial photos taken by the Iranian National Cartography Cen-ter �NCC� two days after the earthquake, and complementary site investigations carriedout by IIEES some days after the event, it is estimated that more than 3000 small andlarge soil block falls and landslides were triggered by the Bam earthquake over an areaof 61 km2, mostly east of Bam and west of Baravat. Figures 4 and 5 show aerial photosof the block falls and Figure 6 illustrates the distribution of block slides and falls. As canbe observed, most ground failures were located along or very close to the Bam fault,reducing in density as distance to fault increases. The susceptibility of this area for slid-ing can be related to the existing geological formation �Qm1� and tectonic activities ofthe Bam fault that caused a hilly morphology that has been eroded, producing deep
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S139
channels. In addition, it seems that strong ground motion at this area is much higher thanthe rest of the city of Bam and Baravat. Two major types of landsides triggered by theBam earthquake can be categorized as follows:
1. Highly disrupted shallow falls that were concentrated west of Baravat and eastof Bam. These slopes are extremely susceptible to failure and it is estimated thatmore than 75% were affected by the earthquake. The main characteristics oflandslides in this area were their small sizes and shallow depths �Figure 5�.
2. Deep coherent landslides with volumes of several hundreds to thousands of cu-bic meters. Such landslides were observed west of Baravat and north of Rah-mani village �Figure 7�.
Fortunately, these landslides have no considerable direct effects on the life and prop-erties, as no important lifelines or structures have been constructed at the landslide ar-eas. The only case of damage to a building caused by landslide or block falls was inEsfikan �Figure 8�.
Figure 1. Geological map of Bam.
S140 F. PELLET ET AL.
GROUND SUBSIDENCE DUE TO COLLAPSE OF QANATS
The most important geotechnical instabilities observed during Bam earthquake wasrelated to the existing qanats and land subsidence above underground galleries.
Figure 2. Geological profile along line A-A� at Figure 1.
Figure 3. Geological profile along line B-B� at Figure 1.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S141
GENERAL DESCRIPTION OF QANATS
Qanats, also known as karez, is a traditional system to bring groundwater fromhigher grounds to the plains. This system was invented in Iran about 3,000 years agoand spread to other cultures, especially along the Silk Road. Although qanat technologydates back thousands of years, it is still one of the most economical methods to transportunderground water to the surface in arid and semiarid regions, and for a large share ofthe supply in such areas comes from qanats.
In general, qanats are gently sloping underground tunnels dug into alluvium orwater-bearing sedimentary rock to pierce the underground water table and penetrate theaquifer beneath. Water from the aquifer filters into these channels, flows down theirgentle slope, and emerges as a surface stream of water at or near a village or town �Fig-ure 9�.
Most of these gravity-flow tunnel-wells are relatively short, some 5 km or less inlength �Beaumont 1989�. The longest, however, extend 40 or 50 km beneath groundlevel before surfacing at a residential area �English 1966�.
The cross-section of a qanat tunnel is roughly 1.5 m high and 1 m wide, largeenough to accommodate men working. Every 30 to 100 m, vertical shafts are dug downto a depth of anywhere from 10 to 100 m to the water-bearing tunnels. These shafts pro-vide air to qanat diggers working beneath the surface and enable excavated soil to beremoved from the tunnel and lifted to the surface. The soil is dumped around the open-
Figure 4. Locations of some soil block falls and landslides at Baravat.
S142 F. PELLET ET AL.
Figure 5. Soil block falls; �a� west of Baravat and �b� east of Bam.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S143
ing of the shaft to form a small mound to keep surface runoff from entering the shaft andbringing silt and other debris with it. The shafts also provide repair teams with relativelyeasy access to tunnels when blockages occur. The donut-shaped spoil heaps around thetops of these vertical shafts appear on the surface as a chain-of-wells, a distinctive fea-ture of landscapes in qanat-watered regions �Figure 10�.
If the soil is firm, no lining is required for the tunnels and shafts but in loose soil,reinforcing rings are installed at intervals in the tunnel to prevent cave-ins. These ringsare usually made of sun-dried clay brick. Figures 11 and 12 show the method of exca-vation of a qanat.
SPECIFICATIONS OF QANATS
Bam is located in an arid region with low annual rainfall, so underground water isthe main source of water in the area for both consumption and irrigation. Based on theavailable data, before the Bam earthquake about 50% of the water supply was providedby 67 qanats in the city of Bam and 370 qanats in the region �PCI and OYO 2004�. Inaddition, there are several old qanats, now dry, in the area dating from hundreds of yearsago. Some of these qanats can be traced in aerial photos �Figure 13�.
Most of the underground galleries of the qanats in Bam are shallow, less than 10 min depth. But there are also some deep qanats in the area for transferring water to adja-cent villages. The diameter of these galleries varies between 80 cm to 1 m. At locationswith soft and collapsible layers, the galleries have been supported by handmade ringscalled kaval �Figure 14�.
Figure 6. Earthquake-triggered landslide and block fall distribution map; base map preparedusing GIS.
S144 F. PELLET ET AL.
QANATS BEHAVIOR DURING HISTORICAL EARTHQUAKES OF IRAN
It is a general perception that underground openings perform much better duringearthquake shaking when compared with surface structures, but this is not the case forqanats. There were numerous cases of damage to qanats due to collapse of shafts or gal-leries during both the historical and recent earthquakes in Iran. The damage varies frompartial to complete collapse of galleries and temporary to permanent blockage of waterflow. The collapse of a qanat gallery usually causes land subsidence or sinkholes at theground surface. Table 1 presents a summary of damage to qanats due to historical andrecent earthquakes in Iran. As shown, there has been considerable damage to qanat net-works due to strong ground shaking.
DAMAGE TO THE QANATS OF BAM DURING THE EARTHQUAKE
Most of the active qanat networks at Bam and Baravat were damaged by the Bamearthquake: about 40% severely damaged or completely collapsed. Due to blockage ofwater in the collapsed qanats, the water supply for irrigation of palm and citrus farms inthe area has been reduced significantly, causing a major economic loss in agriculturalproductivity. The Bam region was one of the main agricultural areas for palms and citrusin Iran before the event. It is estimated that agricultural products in the year followingthe earthquake were lower than half the annual average.
Different types of damage to qanats can be distinguished: damage to the shafts andunderground shallow-to-deep galleries, or partial to complete collapse of the galleries
Figure 7. Deep tensional crack and soil block falls south of Bam.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S145
causing sinkholes at the ground surface level. Most of the sinkholes were related to thecollapse of shallow qanats of 3–10 m in depth, but a few cases of deep sinkholes canalso be observed in the area. The size of sinkholes varied with the dimensions of under-ground openings, any enlargement process that may have occurred, and erosion of shaftand tunnels. At some locations, because of existing close parallel galleries, very largesinkholes were formed at the ground surface.
Figure 8. Damage to a building due to landslide at Esfikan, northeast of Bam.
Figure 9. Schematic views of different parts of a qanat �Wulff 1968�.
S146 F. PELLET ET AL.
Near the Bam fault, the damage was more severe and several small and large sink-holes were observed. Far from the fault, the effects of earthquake on qanats were lessintense, and only some fissures and cracks were observed along the galleries and shaftsthat were related to the small settlement or partial collapse of some parts of the qanats.Figures 15–20 depict a few sinkholes induced by Bam earthquake.
These sinkholes also caused some damage to the structures and lifelines especially inBaravat and south of Bam. Several buildings and roads were damaged due to collapse ofold qanats. The most visible sinkholes were related to the access roads at Baravat shownin Figures 21 and 22. Even the main road from Kerman to Zahedan, which passes southof Bam, was damaged severely because of sinkholes. This damage caused delay for res-cue and relief teams needing access to the damaged area.
In addition, damage to buildings due to collapse of qanats have been reported. Figure23 shows how the collapse of a previously unknown qanat caused damage to a buildingin Baravat.
DISTRIBUTION MAP OF SINKHOLES IN BAM AND BARAVAT
Using the collected data during site visits and study of aerial photos, a distributionmap of sinkholes was prepared using GIS. This map is shown in Figures 24 and 25. Themap also provides insight to the intensity of shaking. It is evident that the damage zoneis concentrated along the Bam fault.
Figure 10. Aerial photo of a qanat in Bam region.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S147
NUMERICAL MODELING OF QANAT BEHAVIOR
In order to study the mechanism of collapse of the qanats and to determine the maincause of collapse, a series of numerical analysis were carried out. For this purpose thefinite element code CAST3M developed in CEA �Centre de l’Energie Atomique, France�was used �Le Fichoux 1998�. This program allows three-dimensional modeling that isquite useful for modeling of qanats.
CONSTITUTIVE MODEL
It was supposed that the soil behavior follows Cambridge model �Cam-Clay�. Thismodel was developed typically for soils by Burland and Roscoe �1968� and Schofieldand Wroth �1968�. The Cam-Clay model is based on the concept of critical state, whichallows taking into account the two elastoplastic mechanisms that govern the behavior ofgeomaterials. The first mechanism, purely contractant, is associated with the sphericalcomponent of the stress tensor. The second mechanism is governed by internal frictionand associated with the deviatoric stress tensor. The latter is commonly contractant, thenpossibly dilatant, just before macroscopic failure. The advantage of the Cam-Clay modelresides in its successful coupling between these two mechanisms, which allows a correct
Figure 11. Schematic view of excavating a qanat �Wulff 1968�.
S148 F. PELLET ET AL.
Figure 12. Removing excavated material through the shafts, Bam area.
Figure 13. The location of qanats south of Bam are shown in this aerial photo of the area.
has been blocked by earthquake shaking at the depth of 20 m along the gallery.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S149
Table 1. Damage to qanats during earthquakes in Iran �Ambraseys and Melville 1982�
Damage to Qanat Location Date
Collapse Shiraz 1853Collapse and long ground fissures�5–8 m�
Ghochan 1893
Collapse in 5 km of qanat length Ghochan 1895Collapse Laleh Zar 1923Collapse Kopeh Dagh 1929Collapse Ah-Mobarak Abad 1930Crack along an old qanat North Behabad 1933Collapse and being dried Doost Abad 1947Collapse Dashte Bayaz 1968Collapse Ferdoos 1968Collapse of 5 km qanat tunnel and 180 accesswells
Ghir-Kazerin 1972
Collapse Karizan-Khavaf 1979
Figure 14. A qanat with supporting rings, south of Bam. The workers try to open this qanat that
S150 F. PELLET ET AL.
Figure 15. Location of some sinkholes west of Baravat.
Figure 16. Large sinkhole east of Bam.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S151
evaluation of permanent strains induced by the combination of spherical and deviatoricstresses expected during seismic loading. Hence, by using the Cam-Clay model, a quali-tative assessment of collapse of qanats can be made. In fact, during seismic loading, per-manent strains accumulate throughout cycles. These irreversible strains develop duringthe unloading-reloading cycles due to the hardening and softening of soil, depending onthe change in the elastic zone. Nevertheless, in the modified Cam-Clay model �Roscoeand Burland 1968�, this kind of response cannot be reproduced because the yield surfaceis unaffected by the unloading-reloading that occurs in the elastic zone. However, it isthe most suitable model implemented in CAST3M to reproduce the behavior of qanatsand shafts during an earthquake. The modified Cam-Clay yield function, which is alsothe flow rule, is defined as:
f = g = q2 − M2�p��pc� − p��� �1�
where:
p� is the mean effective stress
q is the equivalent shear stress
pc� is the preconsolidation stress which acts as a hardening parameter
M is the stress ratio q /p� at critical state related to the frictional angle through thefollowing relationship:
Figure 17. Sinkhole induced due to collapse of a shallow qanat, west of Baravat.
S152 F. PELLET ET AL.
M =6 sin �
3 − sin ��2�
The yield surface is described by an ellipse in q-p plane. The top of the ellipse islocated on the critical line state, which describes the stress states at post-peak failure. Inthe q-p plane this function takes the shape presented in Figure 26.
When a stress state reaches the yield surface, plastic strain takes place, which is gov-erned through an associated flow rule. The permanent change in incremental volumetricplastic strain is given as:
d�vp = �� − �
1 + e�dpc�
pc��3�
The direction of the plastic strain rate is specified by the stress gradient of a flowfunction �f /��ij. If the yield function and flow function are chosen to be the same, theplasticity is called associative.
Hence, the Modified Cam-Clay model requires the specification of five parameters,values that can be derived from standard odometer or triaxial compression tests. Theseparameters are as follows:
• M, the stress ratio q /p� at critical state
• �, the gradient of the normal consolidation line in e-ln�p� space
Figure 18. Large sinkhole, south of Bam.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S153
• �, the gradient of the swelling and recompression lines in e-ln�p� space is givenby:
� =�1 + e�p
K�4�
where K is the elastic bulk modulus and e the void ratio
• e, the critical voids ration which locates the consolidation lines in e-ln�p�� space.This is taken as the voids ratio at unit p� on the critical state line.
NUMERICAL ANALYSES
A parametric study was performed to analyze the effect of the material propertiesand the effect of the depth for both the three- and the two-dimensional models. Table 2presents the cases studied. The parameters were chosen for gravelly sand, where c is thecohesion, � is the frictional angle, and h the depth of the qanat. The other Cam-Clayparameters were fixed as follows:
• The young modulus E=50 MPa
• The Poisson ratio �=0.27
• The over consolidation ratio OCR=1.2
• The gradient of the normal consolidation line �=0.15
Figure 19. Large sinkhole observed, west of Baravat.
S154 F. PELLET ET AL.
• The gradient of the swelling and recompression lines �=0.002
• The void ratio e=0.45
In this analysis, circular sections of shaft and the gallery were considered with di-ameters equal to 1 m. The model was prepared based on shaft depths of between 5 and15 m, with a gallery length of 30 m between two shafts. Finally, a three-dimensionalmesh constructed by cubic elements with 8 nodes was used �Figure 27�. Some two-dimensional computations were also performed using quadrilateral elements with 4nodes, shown in Figure 28. Since the qanats and shafts were built hundreds of years ago,we have assumed that their excavation has no effect on the seismic response, and con-sequently, the excavation was not simulated. The weight of the soil was applied first, andin the second stage, the structure was subjected to seismic loading. No lining was con-sidered for the tunnels and shafts, and the damping was assumed to be zero.
The mesh and boundary conditions for the model were defined based on the geo-metric parameters of qanats. The seismic load was applied at the base of the structurerepresented in Figures 27 and 28 in the o-x or in the o-y direction. The displacement ofthe base in the vertical direction is canceled. To reproduce the infinite and periodic char-acter of the structure in the lateral directions, the parallel vertical boundaries of thestructure were assumed to move similarly.
Figure 20. Collapse of shafts and tensional cracks along the galleries of old qanats, west ofBaravat.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S155
For all these examples, the Bam accelerogram �Figure 29� was applied to the meshesat the base of the model only in o-x direction �transverse to the tunnel� �Figures 27 and28�. For a depth equal to 10 m, the case where the accelerogram is applied only in theo-y direction �parallel to the tunnel� was also analyzed.
RESULTS
The soil strength is essentially governed by the amplitudes of the mean effectivestress and the von Mises stress. The variations of these quantities throughout the struc-tures are given in Figures 30 and 31.
Based on the results, large variations of mean stress �Figure 30� and von Mises stress�Figure 31� induced at the top of the shafts allowed large plastic strain to be developed�Figure 32�. Knowing that the strength of soil decrease with the mean stress, it can bededuced that close to the ground surface the strength is very low and may also be af-fected by the alteration due to the cyclic loading. This consideration explains why seri-ous damages and collapse of the top of the shaft were observed during Bam earthquake.This phenomenon is also well illustrated by the performed computations �Figure 32�,which show that large strains were developed at the tops of shafts.
Around the qanat surface, Figure 30 shows that the mean stress is a compressivestress. Consequently, this area will be less sensitive to the loss of strength, and it couldbe more stable than the top of the shaft. This explains why the plastic strains developedin the tunnel are smaller than those on the top of the shafts �Figure 32�. In addition, the
Figure 21. Sinkhole in one of the by-passes at Baravat due to collapse of a qanat.
S156 F. PELLET ET AL.
von Mises stress along the tunnel will not significantly change due to the significant ini-tial shear stress around the tunnel relative to the shear stress induced by the earthquake.Thus, considering the cyclic behavior of soils �Luong 1980�, the strength loss during theearthquake will be small in the tunnel if an adequate constitutive model is used.
We have also noticed that plastic strains do not develop uniformly along the tunnel.The maximum plastic strain occurs far from the shaft, as indicated in Figure 32. Thismay be due to the tunnel length that induces a bending effect, with large strains concen-trated in the middle of the tunnel. This result is in good agreement with the observedsinkholes at the soil surface between shafts during Bam earthquake.
The results of two-dimensional modeling show smaller plastic strains compared tothree-dimensional calculations �Figure 33�. In fact, as indicated previously, the plasticstrain is not distributed uniformly along the qanat and the large strain concentration dueto the qanat length cannot be taken into account in two-dimensional calculations.
Figure 22. Damage to one of the main roads and a religious arc due to the collapse of a hiddenqanat in Baravat �photo by M. Zare�.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S157
Figure 23. Building damage due to collapse of an old hidden qanat in Baravat.
Figure 24. Photo mosaic used for evaluation of the sinkholes in Bam and Baravat.
Figure 26. Yield surface of the modified Cam-Clay model in q-p plane.
S158 F. PELLET ET AL.
Table 2. Model parameters
c=10 KPa; �=35° c=20 KPa; �=35° c=10 KPa; �=38°
h=5 m; — —h=10 m; h=10 m; h=10 m;h=15 m; — —
Figure 25. Distribution of sinkholes in Bam and Baravat.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S159
Figure 27. Three-dimensional mesh: seismic load and boundary conditions of the model.
Figure 28. Two-dimensional mesh: seismic load and boundary conditions of the model.
S160 F. PELLET ET AL.
Figure 29. Bam accelerogram.
Figure 30. Mean stress distribution at the end of the earthquake �h=10 m, c=10 KPa, �
=35°�.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S161
Figure 31. Von Mises stress distribution at the end of the earthquake �h=10 m, c=10 KPa,�=35°�.
Figure 32. Plastic strain distribution at the end of the earthquake �h=10 m, c=10 KPa, �
=35°�.
S162 F. PELLET ET AL.
Table 3 presents the principal results obtained from the parametric analysis. The re-sults show that if the seismic acceleration is applied in o-x direction, it results in largerstrains than if it is applied in the o-y direction. Table 3 also shows that when the soilcohesion or the frictional angle decreases, the plastic strains increase. The effect of thedepth increase is clearly seen by the expansion of the plastic strains in the shafts and inthe tunnels.
CONCLUSION
The Bam earthquake caused ground failure in the general area of Bam and Baravat.Qanat failures were most significant. In addition several landslides and soil block fallswere observed along the riversides and a manmade channel. No evidence of liquefactionwas observed in the area. These ground failures were recorded and mapped in GIS, anddamage distribution was evaluated. Based on the results, both instabilities of qanats andlandslides were concentrated along the Bam fault. It is shown that strong shaking nearthe fault was considerable in comparison to other areas.
The qanat failures, in turn, caused severe damage to the existing underground watersupply network in the area. The sinkholes induced at top of the shafts and along the gal-leries of existing qanats were studied. The results of numerical modeling of qanats usingthe Bam motion show collapse due to strong shaking.
Figure 33. Plastic strain distribution at the end of the earthquake �h=10 m, c=10 KPa, �=35°�.
GEOTECHNICAL PERFORMANCE OF QANATS DURING THE 2003 BAM, IRAN, EARTHQUAKE S163
REFERENCES1
Ambraseys, N. N., and Melville, C. P., 1982. A History of Persian Earthquakes, CambridgeUniversity Press, London.
Beaumont, P., 1989. “The Qanat: A means of water provision from groundwater sources,”Qanat, Kariz, and Khattara: Traditional Water Systems in the Middle East and North Africa,edited by P. Beaumont, M. Bonine, and K. McLachlan, MENAS Press, London.
Burland, I. B., and Roscoe, K. H., 1968. On the generalized stress-strain behaviour of wet clay,Engineering Plasticity, edited by J. Heyman and F. A. Leckie, Cambridge University Press,Cambridge.
English, P. W., 1966. City and Village in Iran, University of Wisconsin Press, Madison, WI.
Eshghi, S., and Zare, M., 2003. Bam �SE Iran� earthquake of 26 December 2003, Mw 6.5: APreliminary Reconnaissance Report. �http://www.iiees.ac.ir/English/bam_report_english_recc.html�.
Le Fichoux, E., 1998. Présentation et Utilisation de CASTEM2000, Rapport ENSTA-LME. �inFrench�.
1 Publication of this special issue on the Bam, Iran, earthquake was supported by the Learning from EarthquakesProgram of the Earthquake Engineering Research Institute, with funding from the National Science Foundationunder grant CMS-0131895. Any opinions, findings, conclusions, or recommendations expressed herein are theauthors’ and do not necessarily reflect the views of the National Science Foundation, the Earthquake Engineer-
Table 3. Maximum plastic strains computed for a range of soil parameters and qanat condition
Model condition
c=10 KPa�=35°
c=20 KPa�=35°
c=10 KPa�=38°
h=5 m; h=10 m; h=15 m; h=10 m; h=10 m;Plastic Strains
ing Research Institute, or the authors’ organizations.
S164 F. PELLET ET AL.
Luong, M. P., 1980. Phénomènes cycliques dans les sols pulvérulents, Revue Française de Géo-technique. No. 10, 39–53.
PCI and OYO, 2004. The Bam Earthquake Study, report prepared for Japan InternationalCooperation Agency.
Schofield, A. N., and Wroth, C. P., 1968. Critical State Soil Mechanics, McGraw-Hill, London.Wulff, H. E., 1968. The Qanats of Iran, Scientific American, April, 94–105. Available online at
