TRAFFIC LOAD INDUCED PERMANENT DEFORMATION OF LOW ROAD EMBANKMENT ON SOFT SUBSOIL

Transcription

1 TRAFFIC LOAD INDUCD PRMANNT DFORMATION OF LOW ROAD MBANKMNT ON SOFT SUBSOIL Jin-Chun Chai 1 and Norihiko Miura 2 ABSTRACT A simple method for predicting the traffic load induced settlement of low road embankment on soft subsoil is proposed. The traffic load induced dynamic stress in the subsoil is calculated by multi-layer elastic theory. Then the plastic vertical strain in subsoil is calculated by an empirical euation, in which the constants are related to the physical and mechanical properties of subsoil. The method was applied to analyze a 1. km long test section of low road embankment on the soft Ariake clay deposit, Saga prefecture, Japan. The test section had 11 subsections with different construction methods. Four (4) of the subsections were analyzed in this study. It shows that the prediction agreed with the field data reasonably well, and the proposed method is useful for design the road with low embankment on soft subsoil. For the case studied, the depth significantly influenced by traffic load is about 6 m below the base of the embankment. The analyses together with field data indicate that increasing the thickness and stiffness of the subgrade is very effective for reducing the traffic load induced permanent deformation of soft subsoil. INTRODUCTION For low road embankments (less than 3. m high) constructed on soft subsoil, the permanent deformation of subsoil due to traffic load controls the design life as well as the maintenance costs. To select a cost effective design, it is desirable to predict the traffic load induced time dependent settlement. Although some methods have been proposed to calculate the traffic load induced settlement, a rational and simple method is not available yet. Several factors affect the traffic load induced deformation, namely, (a) the strength and deformation characteristics of the subsoil, (b) the properties of the pavement and subgrade of road, and (c) the magnitude and number of applications of traffic load. Any useful settlement prediction method should consider these factors directly or indirectly. The existing methods can be divided into 3 groups: (a) numerical methods; (b) euivalent static loading methods; and (c) empirical euations. Theoretically, explicit simulation of the response of subsoil under repeated load is preferable. A method proposed by Hyodo et al. (1996) combines dynamic numerical analysis (two-dimensional) and dynamic triaxial test results to predict the traffic load induced deformation. The method of explicit simulation is difficult to use in engineering practice. The response of subsoil under traffic load is a three-dimensional (3D) problem and the number of load applications is extremely large. Furthermore, a considerable number of soil parameters are needed in the modeling. With a time consuming calculation, the accuracy of the results may not be guaranteed. Kutara et al. (198) proposed a design method, in which an euivalent static load represents the traffic load, and the one-dimensional consolidation theory is used to predict the settlement. Although this method is simple, both the 3D traffic loading transfer mechanism and the mechanism of repeated load induced settlement are not taken into consideration. Based on field measurement, Fujikawa et al. (1996) proposed a method to estimate the distribution of traffic load induced 1 Associate Professor, Department of Civil ngineering, Saga University, 1 Honjo, Saga 84-82, Japan 2 Professor/Director, the Institute of Lowland Technology, Saga University, 1 Honjo, Saga 84-82, Japan.

2 consolidation settlement in subsoil. In Fujikawa et al.'s method, for the traffic load induced stress increment, a triangular distribution pattern is assumed, i.e. maximum at ground surface and linearly decreased to zero at the depth of influence. This method is better than Kutara et al.'s proposal. However, traffic load induced stress is not calculated explicitly, and the behavior of subsoil under repeated load is also not considered. A number of empirical euations have been proposed to predict the permanent deformation of cohesive soil under repeated load. Among them, the power euation proposed by Monismith et al. (1975) has been widely used. Li and Selig (1996) developed a method to determine the constants in the power euation, in which the magnitude and the number of applications of traffic load, and the strength of subsoil are directly included in the euation, and the physical state of subsoil is considered indirectly. Li and Selig (1998) showed several successful applications of the euation to predict the settlement of cohesive subsoil under train loading. The values of the constants suggested by Li and Selig seem applicable to remolded (compacted) cohesive soil but not suitable for natural clay deposit. Also, the effect of initial static deviator stress was not considered. A new rational empirical euation is proposed in this study to calculate the permanent settlement of low road embankment on soft subsoil. The method considers the effects of (a) initial static deviator stress, (b) the magnitude and the number of applications of traffic load, and (c) the strength and compression characteristics of subsoil. The constants used in the euation are related to physical and mechanical properties of subsoil, such as plasticity index and compression index. Application of the proposed euation to a test section (with 11 subsections) at the Saga prefecture road (Soh-Ohzi Morodomi), Japan, is described. The calculated values are compared with measured data. Discussions are presented on the validity of the proposed method as well as the effect of subgrade properties on permanent deformation of subsoil. A NW MPIRICAL QUATION A Brief Review of Li and Selig's Method A commonly used power euation for calculating the cumulative plastic strain of cohesive soils under repeated loading is as follows (Monismith et al., 1975): b ε p = A N (1) Where ε p = cumulative plastic strain (%), N = the number of repeated load applications, A and b are constants. Li and Selig (1996) proposed an euation for calculating A as: A a( ) d m = (2) f Where d = traffic load induced dynamic deviator stress, f = static failure deviator stress of soil, a and m are constants. Three constants a, b and m are related to plasticity index of subsoil and the suggested values are as in Table 1. Table 1 Constants suggested by Li and Selig (1996) Soil type a b m CH(High plasticity clay) CL(Low plasticity clay) MH(elastic silt) ML(silt) Newly Proposed mpirical uation It is commonly accepted that when the dynamic deviator stress exceeds a critical level (dynamic strength), the plastic strain increases rapidly with repeated loading. Also, under repeated load, the response of the normally consolidated soil is different from that of overconsolidated soil. Assuming that the shear stress acting in soil is below dynamic strength of soil (not a failure problem), and the subsoil is in a normally to slightly overconsolidated state. Then, considering the effect of initial static deviator stress, a new empirical euation for calculating the cumulative plastic strain of soft subsoil under traffic loading is proposed as:

3 d m is n b ε p = a ( ) (1 + ) N (3) f f where is = initial static deviator stress, and n is a constant. The meanings of other parameters are as defined previously. In total,. 3 has 4 constants: a, b, m, and n. Note, when N=1,. 3 calculates the plastic strain for first repeated load application. Dynamic deviator stress d. To estimate the value of d, the 3D traffic load transfer mechanism and the characteristics of multi-layer foundation must be considered. Burmister's multi-layer elastic solution (Burmister, 1945) is considered suitable for this purpose. Using the numerical techniues, the non-linearity of the subsoil behavior can be considered. Constants b and m. At present, the values suggested by Li and Selig (1996) are adopted (Table 1). The parameter b controls the increment rate of plastic strain with the number of repeated load applications. Li and Selig (1996) showed that b is not sensitive to the magnitude of dynamic deviator stress. It is mainly influenced by soil type. The parameter m influences both the magnitude and distribution of plastic strain with depth. Since d / f is less than 1., the larger the m value, the faster decrease of ε p with depth. Constant a. The parameter a controls the magnitude of plastic strain. The traffic load induced deformation mainly consists two parts: dynamic consolidation and shear deformation. The amount of dynamic consolidation deformation is directly related to the compression index (C c ) of soil. Also, C c is one of the factors affecting the magnitude of shear deformation. Therefore, it is rational to relate a with the compression index (C c ) of subsoil. a = αc c (4) For the case studied in this paper, the value of C c is about.8 to1., and a α value of 8. can be evaluated. The values of a suggested by Li and selig (Table 1) are.64 to 1.2. If using α of 8., a C c of.9 to.15 can be estimated. Although Li and Selig (1996) did not report the value of C c, for compacted clayey soil, a C c of.9 to.15 seems reasonable. Constant n. The constant n controls the effect of initial static deviator stress on traffic load induced plastic deformation. Most laboratory triaxial tests on the effect of initial static deviator stress were conducted for the purpose of evaluating earthuake response. Therefore, a relatively larger dynamic deviator stress had been used. Samang (1997) conducted a series of dynamic triaxial tests on undisturbed Ariake clays under undrained conditions. Ariake clay No. 1 had a liuid limit of about 58 to 76% and plasticity index of 32 to 45. The clay content (<5µm) was 41 to 64%. Ariake clay No. 2 had a liuid limit of 86% and plasticity index of 48. The clay content was 61%. Samang s data are re-plotted in peak strain ε p versus initial deviator stress ratio is /p co in Fig. 1 for 2 load application cycles. In this figure, is =σ 1o -σ 3o and p co =(σ 1o +2σ 3o )/3 andσ 1o and σ 3o are initial major and minor principal stresses. The data points in Fig. 1 correspond to σ 3o /σ 1o of.39 to.75. For both clay samples, the ε p increased almost linearly with is /p co. Non-linearity may be expected for a higher dynamic deviator stress and a larger number of loading applications. Also, the undrained triaxial test may not represent the field condition, where the partial consolidation occurs. Nevertheless, at present, based on the limited test data, it is proposed that the magnitude of plastic strain be assumed to linearly increase with initial static deviator stress, so that n=1. is adopted. ANALYSIS OF A TST SCTION AT TH SAGA PRFCTUR ROAD Initial deviator stress ration is /p' c In 1995, a 1. km long test section was constructed at the Saga prefecture road, Soh-Ohzi Morodomi line, Saga P eak axial strain (% ) Ariake clay No. 1 d /P' c =.374 Ariaka clay No. 2 d /P' c =.385 N=2 cycles (Data from L. Samang, 1997) Fig. 1 Relationship of peak axial strain/initial

4 Japan. The average embankment height at the test section was about.75 m (Civil ngineering Department of Saga Prefecture, 1996). The test section was divided into 11 subsections and several construction methods were tested. The methods include (a) surface lime treatment, (b) surface cement treatment, (c) light-weighted fill material, (d) geogrid reinforcement, and (e) cement column improvement of subsoil. Among 11 subsections, 4 of them are discussed in this paper. The subgrade treatments of these 4 subsections are given in Fig. 2. A typical subsoil profile at the site is shown in Fig. 3 together with physical properties. In the figure, A c means alluvial clay, A s means alluvial sand, and H c means Hasuike non-marine clay deposit. The soft deposit is about 2 m thick with 3 clay layers sandwiched 2 thin sand layers. About 1. m thick surface crust is in an overconsolidated state with an overconsolidation ratio (OCR) of 2. to 4. (mainly due to weathering and aging). The deposit below 1. m depth is in a normally to slightly overconsolidated state. The compression index of the clay layers is about.8 to 1. for upper layer (A c1 ) and 1. to 1.5 for lower soft layers (A c2 and A c3 ). Ascon Crushed Stones Cement Treated ( u=1.4 MPa) Cement Treatment (CM) Ascon Cement Treated (u=2.44 MPa) Soil-cement Column (Area replacement ratio: 36%) Cement Treatment & Soil-cement Column (CS) Ascon xpansive Cement Crushed Stones Treatment Glass-grid Cement Treated (u=1.98mpa) xpansive Cement Treatment & Glass-grid (CM) Ascon Crushed Stones Glass-grid CementTreated (u=.58mpa) Unit: mm Cement Treatment & Glass-grid (CMG) Fig. 2 Subgrade treatments of 4 test subsections (Saga Prefecture road) The construction period was about 2 months for the sections considered. Four (4) months after completion, the road was opened for traffic. The settlements of the road after opening for traffic were monitored. Since the upper subsoil was in a lightly overconsolidated state, field measurement together with finite element analysis confirmed that the settlement due to.75 m embankment loading (about 7 mm) was almost finished before the road was opened for traffic. In following analysis, it is considered that the measured settlement after the road was opened to traffic was caused by traffic load only. The total traffic intensity (two ways) was about 2 car/day and most of them were family cars. The trucks were about 6% of total traffic (1 trucks/day). The T.P. (m ) G.L. (m ) G rain Size Soil Clay Silt Llayer (% ) Sand H c1 A c1 A s1 A c2 A c3 H c2 (% ) 1 1 w p w l w n w p : p la stic lim it w l : li u id lim it w n : natural water content T o ta l U n it W e ig ht : γ t (kn /m 3 ) 5 1 Fig. 3 A typical soil profile at test section of Saga prefecture road Table 2 Adopted Young s modulus Sub- Pavement Subgrade Upper Subsoil Lower Subsoil Section H H H H CM CS CM CMG

5 analysis showed that trucks caused almost all the plastic deformation of the road. In the analysis, it was assumed that the average weight of a truck was 1 kn (back axis 8% and front axis 2 %), the traffic load amplification factor was 1., and radius of tire/road contact area was 2 mm for back tires (double) and 1 mm for front tire (Utida, 1988). The adopted Young s modulus () and thickness (H) of each layer are summarized in Table 2. The modulus for cement treated layers was estimated as 1 times the unconfined compression strength ( u ). The modulus for natural subsoil was assumed to be 2 times the undrained shear strength (S u ). The calculated dynamic deviator stress was 3 to 5 kpa just below the subgrade, and reduced to about 1. kpa at 6 m depth. Table 3 gives calculated results of CM subsection (see Fig. 2) as an example. The static undrained shear strength (S u ) of the subsoil is calculated by the following empirical euation (Ladd et al., 1991). S Sσ ' ( OCR) m u = v (5) Where σ v = effective vertical stress. Values of S=.35 and OCR=1 (m=.8) were used. Table 3 Calculated value for CM subsection Depth* Thickness d s S u.15.3 Cement mixing treated *Depth below the base of embankment Variation of Traffic Load Induced Settlement with Time The adopted values of constants in. 3 wereα = 8., C c =.8 to 1., b =.18, m = 2. and n = 1.. The calculated traffic load induced settlements are compared with measured data in Figs. 4 and 5. The calculation predicted the field data reasonably well, which indicates that the proposed method is a useful tool for designing the low road embankment on soft subsoil. In Fig. 4, CM and CS subsections (see Fig. 2) are compared, which had different depth of improvement. For CS subsection, soil-cement columns (36% replacement ratio by area) improved the upper subsoil. The dynamic deviator stress below the columns was small and the traffic load induced settlement was much smaller than CM subsection. In Fig. 5, CM and CMG subsections (see Fig. 2) are compared. For these two subsections, the strength and stiffness of the cement treated layer was different, and CMG subsection had a lower strength and stiffness. The settlement of CMG subsection was more than twice of CM subsection. This case history indicates that an increase in the thickness and stiffness of subgrade is efficient for reducing the traffic load induced permanent settlement. Traffic load induced settlement (mm) CS subsection CM subsection lapsed time after open to traffic (days) Fig. 4 Settlements of CM and CS subsections Traffic load induced settlement (mm) CM subsection CMG subsection lapsed time after open to traffic (days) Fig. 5 Settlement of CM and CMG subsections Distribution of Traffic Load Induced Plastic Strain with Depth The calculated plastic strain distribution with depth is depicted in Fig. 6 for 1 year after opening to traffic. It indicates that the significant influence depth of traffic load is about 6 m below the base of the embankment. The plastic strain reduced very uickly within upper 2 m. It also indicates that the weaker the subgrade (CMG subsection), the larger the plastic strain in upper subsoil. Fujikawa et al. (1996) reported a case history for the traffic load induced settlement of a low road embankment. The site is at national road No. 34 in Saga prefecture, Japan. After 2 years traffic loading, the subsoils

6 under the road, and 2 m away from the center of the road, were sampled and the physical and mechanical properties were investigated. Comparing the results of the soil samples under the road and away from the road revealed that within about 7 m depth, the traffic load had caused a reduction in the void ratio. This observed results is in agreement with the calculated results in Fig. 6. CONCLUSIONS An empirical method for predicting the traffic load induced permanent deformation of low road embankment on Vertical strain (%) soft subsoil is proposed. The method considers the 3D traffic Fig. 6 Calculated vertical strain variation with depth load transfer mechanism, the magnitude and the number of traffic load applications, multi-layer ground conditions, and the strength and compression characteristics of soft subsoil. The method was applied to analyze a 1. km long test section at the Saga prefecture road, Saga, Japan. Comparing the calculated results with the field measurements, it indicates that the proposed method predicted the measured data reasonably well. The method can be used for calibrating the design of low road embankment on soft subsoil. The analysis also revealed following two pints. (a) For the case investigated, the significant influence depth of traffic load is about 6 m below the base of the embankment. (b) Increasing the thickness and stiffness of subgrade is an effective way of reducing the traffic load induced permanent deformation of soft subsoil. RFRNCS Depth below the base of embankment Burmister, D.M. (1945). The general theory of stresses and displacements in layered system I. Journal of Applied Physics, Vol. 16, pp Civil ngineering Department, Saga Prefecture (1996). A study on predicting the differential settlement and rational design method for road on soft subsoil. Research Report, Saga Prefecture, Japan, p 77 (in Japanese). Fijikawa, K., Miura, N., and Beppu, I. (1996). Field investigation on the settlement of low embankment road due to traffic load and its prediction. Soils and Foundations, Vol. 36, No. 4, pp (in Japanese). Hyodo, M., Yasuhara, K., and Murata, H. (1996). Deformation analysis of the soft clay foundation of low embankment road under traffic loading. Proc. 31st Symp. of Japanese Society of Soil Mechanics and Foundation ngineering, pp (in Japanese). Kutara, K., Miki, H., Mashita, Y., and Seki, K. (198). Settlement and countermeasures of the road with low embankment on soft ground. Tech. Report of Civil ng., JSC, Vol. 22, No. 8, pp (in Japanese). Li, D. And Selig,.T. (1996). Cumulative plastic deformation for fine-grained subgrade soils. Journal of Geotechnical ngineering, ASC, Vol. 122, No. 12, pp Li, D. And Selig,.T. (1998). Method for railway track foundation design. II: application. Journal of Geotechnical and Geoenvironmental ngineering, ASC, Vol. 124, No. 4, pp Ladd, C. C. (1991). Stability evaluation during stage construction. Journal of Geotechnical ngineering, ASC, Vol. 117, No. 4, pp Monismith, C.L., Ogawa, N., and Freeme, C.R. (1975). Permanent deformation characteristics of subsoil due to repeated loading. Transp. Res. Rec. No. 537, Transportation Research Board, Washington, D.C., pp Samang, L. (1997). Settlement of soft cohesive deposit induced by cyclic loading. Dr. of ngineering Dissertation, Saga University, p. 24. Uchida, I. (1988). Road ngineering. Morikita Publisher, Tokyo, Japan (in Japanese) CM CM CMG 1 year after open to traffic