Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.

Your browser appears to have cookies disabled. For the best experience of this website, please enable cookies in your browser

We'll assume we have your consent to use cookies, for example so you won't need to log in each time you visit our site.
Learn more

Performance of an articulated culvert under an earth dam on a compressible foundation


This paper presents the design and performance of an articulated concrete culvert under the 63m high Jacarei earthfill dam which sits on a compressible foundation. Design details comprised partial replacement of compressible foundation material with compacted soil, choice of conduit geometry to improve soil compaction and reduce stress redistribution around the conduit and the selection of culvert camber and joint design to minimise joint openings. Finally, instrumentation data of culvert joint openings due to foundation settlements with time are presented.


Articulated concrete culverts near the base of earth dams have been extensively used as part of diversion systems and spillways, especially in small flood control dams. Rutledge and Gould (1973), summarised data of vertical and horizontal movements of concrete culverts under 20 small earth dams with height varying between 6m and 16m. All these dams rested on compressible foundations and their measured final settlements were in the range of 27mm to 520mm.

The major design concern in such cases, when settlements are so high, is the possibility of opening of joints or cracking of pipe section which could endanger the stability of the dam. Therefore design aspects which would induce differential movements in culverts must be thoroughly investigated to minimise the possibility of dam failure. Because of the high stress distribution around the culvert, in most cases settlement calculation using classical methods does not give good results and because there is no generally accepted guidelines concerning relations between maximum differential movements and culvert damage, design decisions must be based on experience and be guided preferably by finite element analysis.

A more recent study (Rizzoli et al, 1991) assesses data from six French earth dams with heights ranging from 16.5m to 25m, also on compressible foundations and incorporating articulated pipes as part of spillways. Observed data of pipe settlements showed maximum total settlement varying from 150mm to 460mm and maximum differential settlement in the range of 16mm to 38mm.

Settlement data from both these studies complements each other in the sense that they cover dam heights from 6m to 25m, the limits of dam height in which articulated underground rigid culverts are mostly used.

Settlement profiles in all these dams increased towards the centre of the dam and showed a maximum near the dam axis. Rizzoli's data also shows that most settlement occurred during construction and reservoir filling. Associated with vertical dam settlements were horizontal movements which could induce lateral spreading of the dam. This increases the risk of failure and therefore requires ways of keeping the risk of failure below safe limits.

The use of articulated culverts under high earth dams with compressible foundations is quite rare because culvert settlement would inevitably be large as would be the risk of dam failure. However, exceptions exist and involve using special design details such as partial replacement of the compressible foundation material with compacted soil, the choice of ideal conduit geometry to improve soil compaction and to reduce stress redistribution around the conduit and the selection of conduit camber and joint design to minimise joint opening (Gaioto, 1992).

Classical examples of dam failures incorporating articulated culverts are reported by ASCE/USCOLD (1975). One case is the failure of Meeks Cabin dam, a 53m high earthfill-rockfill embankment built on a compressible foundation. The major cause of the accident was excessive foundation settlements which resulted in the spreading of the conduit foundation which caused the conduit joints to open by up to 22.86mm.

This paper presents the finite element analysis and instrumentation results of a rigid concrete culvert placed under Jacarei dam on the Jacarei river, in Sao Paulo State, south east Brazil. The dam is a maximum of 63m high and sits on a thick compressible foundation. It is probably one of highest dams in the world using this type of design still performing to design stage predictions.

Project description

The Jacarei and Jaguari dam reservoirs are connected by an earth slope channel and form the main storage reservoir of the SABESP (Cia de Saneamento Basico de Sao Paulo) Cantareira System, which supplies 22m3/s of untreated water to Sao Paulo city area (Gaioto et al, 1981). The system was completed in 1982 and both dams incorporate concrete culverts as part of the spillways and diversion systems. However, as only the Jacarei dam was built on a compressible foundation it is on this structure that the paper will focus.

The Jacarei dam structure mainly consists of an earthfill embankment, the Morning Glory spillway - connected to a rigid culvert with trapezoidal cross-section which also serves as diversion structure - and a bottom outlet conduit (Figure 1).

The homogeneous earthfill embankment, 63m high and 1,200m long at the crest, has a longitudinal axis with a smooth S shape. Its internal drainage comprises a vertical filter slightly downstream from the dam axis connected to the downstream toe by a horizontal drainage blanket. The volume of the compacted earthfill embankment is 7M.m3 (Figure 2).

The Morning Glory spillway is set on the top of the upstream segment of the outlet culvert. Its crest lip has an internal diameter of 6.3m. Its main function during dam construction was to control the partial reservoir level so that it was possible to supply water before the dams were completed. When both dams were finished, flood water was channelled through the Jaguari dam superficial spillway since both reservoirs are connected.

For most of its length, the bottom outlet is composed of two rectangular reinforced concrete cells: the lower, with an internal cross- section 6.6m wide by 5.8m high and the upper, 6.6m wide by 2m high and is 70m long. It is connected to the upstream part of the diversion culvert and its water intake with a prismatic shape (5m high and 6.4m wide) has grades in all its faces. Flow is controlled by two Howell-Bunger dispersing valves, which allow water to flow into a blinder chamber. The top cell of the bottom outlet allows air flow for the dispersion process and gives access for operation and maintenance.

The diversion culvert also consists of two rectangular cells: the bottom one with internal dimensions of 4m by 4.8m and the upper 4m by 2.2m. Its external trapezoidal cross-section is 11.7m high, 7.6m wide at the bottom and 5.8m at the top. For 240m of its 340m length it runs under the dam embankment and is split into 20m sections separated by keyed joints. The culvert stilling basin has 36m long impact teeth.

Local geology

The Jacarei dam is underlain by the Crystalline Complex, composed mainly of migmatitic rocks cut by aplitic and pegmatitic veins and basic and granitic intrusions. The rock mass discontinuities at the dam site are related to transcurrent and normal tectonism.

Intense weathering at the site has resulted in the deposition of thick soil layers. Major erosion processes also occurred at the highest slopes which has led to a high concentration of boulders on the ground surface and at shallow depth within the weathered layers.

The geology directly beneath the dam is mixed, with quartz-diorites under the left embankment, migmatites under the right embankment and alluvial soils in the central section. At the right edge, where several geological faults were detected, the rock outcrops upstream from the dam axis and strongly dips downstream where, under the dam toe, it is covered by a 20m thick soil layer.

At the dam site the river floodplain is 500m wide and consists of recent alluvial layers of low strength up to 4m thick. Occasionally it is also composed of shallow layers of silty colluvium, underlying residual soil layers which rarely display relic structures, in turn overlying residual soils retaining the original characteristics of the parent rock.

At the left embankment the overburden is very thick and has very little colluvium and residual soil layers. The main feature observed at this location is a discontinuity (fault zone) that crosses the dam axis making an angle of 45degrees between the centre of the floodplain and the left embankment.

Culvert placement

Considering culvert dimensions and the magnitude of the imposed loads from the 45m high cover, the ideal foundation material would naturally be a stratum of very low compressibility. However, because bedrock occurs at great depth and also because of hydraulic and construction requirements, such ideal conditions could not be found. Therefore a solution in which the rigid culvert could be safely placed on the top of compressible stratum was sought.

The overall response of the culvert was investigated for five different options, in which culvert position varied along the central dam zone. Comparative analyses of such responses were carried out in order to choose the best solution. First, compressibility contours including the bedrock and the residual soil layers were prepared. For the selected culvert positions, preliminary culvert settlement profiles were drawn based on the geotechnical parameters of foundation materials and height of cover along the culvert axis. Some of these analyses considered the possibility of partial substitution of foundation soil by compacted clay or granular soils.

Additional investigations were carried out for the preferred option, located at the left embankment, to obtain more information on the geotechnical parameters of the foundation materials.

This information was used to decide the excavation depths and levels of filling with compacted materials. Figure 3 shows this excavation and backfill work. Downstream of the dam's vertical chimney filter, the incompetent foundation soil (with SPT N-value less than 10) was removed and replaced by a coarse sand filter layer up to 1m thick around the culvert.

The maximum calculated culvert settlement reached 500mm. The main actions taken to mitigate possible problems related to excessive foundation deformations and to reduce compressive stress on the culvert walls were:

Widening the culvert trench to minimise arching effect and therefore reduce the overburden stress redistribution which would increase stress on the walls.

Splitting the culvert into 20m long segments to provide articulation between adjacent segments. These joints were keyed and fastened by double waterstop joints.

Building the culvert with a camber, corresponding to 30% of the calculated settlement at each culvert joint.

Designing joint faces such that internal opening was always greater than bottom one, to minimise the consequences of joint opening and closure due to differential settlement, such as excessive joint opening and slab concrete crushing. The critical section was built with 25mm at the top and 75mm at the base (see detail, Figure 3).

Provision for installing additional neoprene waterstop joints after the embankment construction. These watertight joints were installed in two steps after most of the culvert settlement had occurred.

Designing the external culvert walls with a 1:10 slope to improve soil- concrete interface.

Cut-off collars were not included, to avoid stress concentration on the culvert walls due to the very high levels of predicted soil deformation and to avoid poor compaction next to the walls.

The stretch of the culvert downstream of the dam's vertical filter was surrounded by a 1m thick sand filter.

Installation of instrumentation which allowed the measurement of culvert stress and deformation after completion of the initial construction stages and that would supply sufficient information to explain culvert behaviour over time.

Assessment of culvert stresses and deformations

Stresses and deformations on the culvert walls were calculated at each joint for the selected option. Analysis using Marston-Spangler theory for positive projecting conduits and the finite element method considering the soil as a linear-elastic material were carried

out. At the soil-culvert contact, joint elements were introduced in the FEM analysis considering high values of joint stiffness coefficients (Kt and Kn), in order to get soil yielding at the contact at the first loading level.

Instituto de Pesquisas Tenologicas (IPT) pneumatic and Maihak's vibrating wire earth pressure transducers and piezometers were installed around the culvert to measure soil stresses and pore pressures in order to obtain the stress distribution on the culvert. The instrumented sections were located at the centre of the dam and 30m upstream of the dam axis, Figure 4.

The measurement of culvert settlement was made using two pins placed at 2m from the centre of each culvert joint in the bottom slab. Two telescopic KM type devices were also installed next to the right edge of the culvert on the instrumented sections. The first plate of this system was fixed to the culvert base. The relative movements between culvert members were recorded by triorthogonal instruments, one at each culvert joint. This allowed the measurement of differential settlements of the culvert between joints, horizontal differential movement between joint sections and opening and closing joint movements. The location of culvert instrumentation is also shown in Figure 4.

Stress and deformation data

Comparisons of stress variations on the top of the culvert with the increases of height of cover using Marson-Spangler classical theory and the FEM are presented in Figure 5.

After certain stress levels both methods predict stresses higher the geostatic stress (H) and well above the measured stresses. For Marston- Spangler method this critical level, from where predicted stresses become higher than geostatic stresses, is 5m and for the FEM, 12m. For cover higher than 15m the difference between the two methods is nearly constant.

However, the first method always predicts the highest stresses. This pattern of behaviour varies from site to site because it depends on the relative compressibility between the pipe and the surrounding soil, as shown by Silveira et al, 1982.

Figure 6 shows the stress data obtained from FEM analysis and instrumentation for the two instrumented culvert sections. As can be seen, measured stresses are generally smaller than stresses predicted by FEM. Although obtained from sections with heights of cover of 45m and 35m, FEM stress data for both sections are not only of similar magnitude but also the pattern of stress distribution is similar. The lack of symmetry noticed in both sections might be caused by the foundation compressibility which differs from point to point. This fact was observed not only with FEM data but also with data recorded by the instrumentation.

The difference between predicted and measured settlement results is thought to be mainly due to the softening of residual soil caused by improper dewatering during culvert construction.

Figure 7 shows the variation of settlement and horizontal movement profiles with time. As can be seen, most settlement occurred in the first three years and the settlement rate decreases substantially with time. From June 1990 to July 1991, partial settlements varied from 0 to 6mm (IPT, 1993), showing that secondary consolidation of the foundation is still occurring in the central culvert segments. As a result of changes in the reading system, from July 1991 instrumentation is producing scattered data which is why no additional data is included in this analysis.


The authors would like to express their gratitude to the Department of Geotechnical Engineering, University of Sao Paulo, Sao Carlos and to SABESP.


ASCE/USCOLD (1975) Lessons from Dams Incidents, USA, New York, 387p.

Gaioto, N, (1992), Systems to Control Water Flow in Earth Dam Designs, Dept. of Geotechnical Engineering, USP/Sao Carlos, 267p. (In Portuguese).

Gaioto, N, Pinca, RL, Martins, A, Pacheco, JG & Cipparrone, M, (1981) Jacarei Dam Diversion Culvert: a Planned Design to Tolerate Large Deformations, XIV CBGB, Recife, Brazil, vol. I, pp. 129-145. (In Portuguese).

IPT - Instituto de Pesquisas Tenologicas (1993) Report No. 31.051/93, Sao Paulo, Brazil. (In Portuguese).

Rizzoli, JL, Moullec, D, Reverchon, B, Isambert, F & Dupas, JM, (1991) Observations on Compressible Foundation Behavior and Very Compressible Design: Dams on Siene, Aube and Marne reservoirs, Lazer Dams, 17th Congress on Large Dams, Vienna, Austria, 12-17 June, vol. III, pp. 105-134.

Rutledge, PC and Gould, JP (1973) Movements of Articulated Conduits under Earth Dams on Compressible Foundations, in Embankment Dam Engineering, edited by Hirschfeld, RC & Poulos, SJ, New York, John Wiley & Sons, pp. 209-238.

Silveira, JFA, Martins, A, Pinca, RL (1982) Jacarei and Jaguari Dam Diversion Culverts: Stress Analysis of Soil-Concrete Interface, VII CBMSEF, Olinda, Brazil, vol. 6, pp.133-153. (In Portuguese).

Have your say

You must sign in to make a comment

Please remember that the submission of any material is governed by our Terms and Conditions and by submitting material you confirm your agreement to these Terms and Conditions. Please note comments made online may also be published in the print edition of New Civil Engineer. Links may be included in your comments but HTML is not permitted.