Technical Paper by D. Leshchinsky, M. Dechasakulsom, V. Kaliakin and H. Ling

CREEP AND STRESS RELAXATION OF GEOGRIDS

¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾

ABSTRACT: Most research on the long-term behavior of geosynthetics has focused on creep. Very little data is available for the stress relaxation behavior of geogrids. Consideration of both creep and stress relaxation in design may ultimately produce a more efficient structure while allowing for the use of smaller creep reduction factors. This paper reflects an initial effort to identify experimentally the stress relaxation behavior of some typical geogrids used in soil reinforcing. A simple test technique has been developed in which the stress relaxation can be measured directly. Tests were conducted on three types of geogrids, subjected to initial loads of 40, 60 and 80% of their ultimate short term strengths. Each test was carried out for a period of one month or until creep rupture occurred, whichever shorter. The measured amount of maximum potential stress relaxation for polyester grids was about 30% of their initial load; for polyethylene grids this potential was about 50%.

KEYWORDS: Geogrids, stress relaxation, creep, polyethylene, polyester.

AUTHORS: D. Leshchinsky, Professor, Department of Civil Engineering, University of Delaware, Newark, DE 19716, USA, Telephone: 1/302-831-2446, Telefax: 1/302-831-3640, M. Dechasakulsom, Graduate Student, Department of Civil Engineering, University of Delaware, Newark, DE 19716, USA, Telephone: 1/302-831-6650, Telefax: 1/302-831-3640, V. kaliakin, Associate Professor, Department of Civil Engineering, University of Delaware, Newark, DE 19716, USA, Telephone: 1/302-831-2409, Telefax: 1/302-831-3640,and H. Ling, Research Assistant Professor, Department of Civil Engineering, University of Delaware, Newark, DE 19716, USA, Telephone: 1/302-831-4952, Telefax: 1/302-831-3640

PUBLICATION: Geosynthetics International is published by the Industrial Fabrics Association International, 345 Cedar St., Suite 800, St. Paul, MN, 55101, USA, Telephone: 1/612-222-2508, Telefax: 1/612-222-8215. Geosynthetics International is registered under ISSN 1072-6349.

DATES: Original manuscript received 25 March 1997, accepted 1 July 1997. Discussion open until 1 May 1998.

REFERENCE: D. Leshchinsky, M. Dechasakulsom, V. Kaliakin and H. Ling, 1997, "Creep and Stress Relaxation of Geogrids", Geosynthetics International, Vol. 4, No. 5, pp. 463-479.

¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾ ¾

1 INTRODUCTION

When embedded in soil and subjected to load, all polymeric materials exhibit time-dependent response; i.e., creep and stress relaxation. Lacking a suitable analytical procedure to quantify such time dependent response, current design procedures employ a large reduction factor, ranging from about 2 to 5 (depending upon the type of polymer), applied to the short term ultimate strength of the material (Koerner, 1994). Clearly, such an approach is straightforward; it does not, however, necessarily reflect reality. For example, no link is made between the time-dependent behavior of the reinforcement and its interaction with the backfill soil along common interfaces. Thus, with respect to time dependent behavior, a soil structure reinforced with multiple layers of geosynthetic is treated in the same manner as a structure reinforced with a single layer. Yet the greater redistribution of load associated with the former case is likely to lead to a different type of time dependent response than the single layer system. That is, as some layers are stressed more than others, they will tend to creep at a faster rate, thus allowing the deforming soil to transfer load to adjacent layers (load shedding). The end result is a load reduction (stress relaxation) of highly stressed layers. Consequently, with respect to time dependent behavior, current design procedures may be overly conservative. However, the number and spacing of reinforcing layers at which this potential over-conservatism evolves is not known. Furthermore, while this conservatism is true for a free-draining soil, it may not be for cohesive soil that itself exhibits time dependent behavior.

The problem of time dependent response of geosynthetically reinforced soil structures is thus interesting and challenging : (1) From the standpoint of fundamental mechanics; (2) In light of current design procedures; and, more importantly, (3) With reference to future applications where, for economic reasons, low quality backfill is likely to be used.

Standing in the way of a thorough understanding of the time dependent behavior of geosynthetically reinforced soil structures are two items: 1. The lack of stress relaxation experimental data; and 2. The lack of a robust constitutive model for the reinforcement that will account for both creep and stress relaxation in a rational manner. The objective of this paper is to present creep and, more importantly, stress relaxation data measured for some typical geogrids used in reinforced walls and slopes.

2 BACKGROUND

A comprehensive review of research on creep of geosynthetics is given by Koerner et al. (1993). Numerous creep tests on geosynthetics (e.g., Finnigan 1977; Allen et al. 1982, to name a few) were performed in-isolation; i.e., unconfined. In an effort to better simulate the effects of soil confinement on creep, in-soil tests have also been performed (e.g., McGown et al. 1982; Holtz et al. 1982). However, the results of such tests were somewhat inconclusive. For example, McGown et al. (1982) showed a drastic effect of soil confinement on the creep of geotextiles. Conversely, Matichard et al. (1990) and Blivet et al. (1992) demonstrated insignificant effects. Furthermore, the interpretation of in-soil test results is complicated by the presence of soil-reinforcement interfaces. For example, Kutara et al. (1988) used a pullout test apparatus to show that creep of geogrids, rather than slip along the soil-reinforcement interface, is the factor controlling long term response. Observations of creep behavior of geosynthetics embedded in sand were reported by Min et al. (1995). Wu (1994) and Wu and Helwany (1996) demonstrated that creep deformations greatly depend on the type of confining soil. In sand, the interface was shown to have a restraining effect on creep. However, in clay the creep rate of the confining soil was faster than that of the geosynthetic, thus inducing an accelerating creep in the reinforcement. Clearly, these tests demonstrate that creep of embedded polymers depends also on the properties of the confining soil and on the interface characteristics between the two.

Comparatively little experimental work has been conducted with respect to polymer relaxation in conjunction with soil reinforcement (Koerner et al. 1993). Limited studies of stress relaxation in geosynthetics were conducted by Krupin et al. (1982) and by Greenwood and Myles (1986). These experiments were extended to geogrids by Greenwood (1990). Koerner et al. (1993) presented some test results on high density polyethylene geomembranes, and attempted to develop a generalized relationship between creep and stress relaxation. Stress relaxation behavior of geomembranes has also been reported by Soong et al. (1994).

From the available literature it is thus evident that while a large body of data exists for creep tests, very little information is available for stress relaxation. This imbalance is explained in part by the fact that current simplified design procedures are based on simple interpretation of creep test data, and thus do not require stress relaxation results. More precisely, current design guidelines (e.g., Leshchinsky and Perry, 1987; Christopher et al. 1989) bypass the creep problem by specifying an allowable tensile strength that is a fraction of the ultimate short term value. Though such an approach is simple to apply, it completely ignores the interaction between the soil and the reinforcement.

Another reason for the scarcity of stress relaxation data is the perceived complexity of the test. In creep, the tensile load is maintained constant and strains are measured with time. However, in a stress relaxation test, the strain developed immediately after load application is maintained constant via reduction of tensile load with time. Such a test inevitably requires a sophisticated device with feedback control.

A reduction factor applied to the short term ultimate strength of the reinforcement represents the most practical design approach. However, to better simulate reality, this factor should be determined in a more rational manner; namely, from a thorough numerical parametric study that accounts for different types of backfills and reinforcement layouts, different structural geometries, as well as different types of geosynthetics. Requisite for such a study is the existence of: (1) Realistic constitutive models for the backfill soil; (2) A method to account for the interaction between the reinforcement and the soil along their common interfaces; (3) A robust constitutive model that can simulate the time dependent behavior of geosynthetics; and (4) A computer program that has been verified against case histories and in which items 1 through 3 have been efficiently implemented. With respect to the reinforcement, input to such a program should include only in-isolation material properties; the effect of confinement and soil-reinforcement interaction along common interfaces will be accounted for in the course of the analysis via items 1 through 3.

However, before such a comprehensive study can be conducted, the stress relaxation behavior of geosynthetics used in soil reinforcing must be studied experimentally. This paper presents results of stress relaxation tests on some typical geogrids. These results require relatively little interpretation and can be used independently by other researchers to conduct a numerical study of the long term behavior of geosynthetic reinforced soil structures.

3 TEST PROCEDURE

Stress Relaxation Testing

Figure 1 shows the schematics of the test setup. It consists of a set of roller grips, load cell, weights and a reed relay. The geogrid specimen is attached at both ends to roller grip clamps. The upper clamp is attached by a cable to a stationary frame. The lower clamp is connected to a load cell. The other end of the load cell is subjected to a static load. The load is comprised of both metal weights and the weight produced by water contained in a 120 liter plastic tank. The combined load produces the tensile force in the geogrid specimen and the load cell measures this force.

Before the load is applied to the specimen, the entire loading fixture is supported by a hydraulic jack. The jack is released quickly and smoothly, and the amount of displacement of the bottom grip is measured. Since there is negligible slippage between the roller grips and the specimen (the grips was more of a clamp than a friction-type grip), this displacement is used to calculate the immediate strain due to the total load. In about a minute after load application, the position of the stationary relay is adjusted relative to the magnet. Since this relay is sensitive to a magnetic field (reed relay), as the specimen deforms with time (i.e., creep strain occurs), the electrical points of the relay close. As a result, the submerged pump, having a discharge capacity of 20 liters per minute, is activated. Through removal of water from the plastic tank, the tensile force applied to the specimen is reduced. This stress relaxation causes the specimen to quickly return to its original position after producing very small creep that is assumed to be negligible. The reed relay is adjusted to respond once a displacement of 0.5 mm occurs over a specimen length of about 500 mm (i.e., a feedback process responding with relaxation to a creep strain of 0.1%). Water pumping to achieve relaxation lasts for a period of less than 15 seconds at a time. The load cell indicates the actual load applied to the specimen with time so that only negligible creep would develop.

In performing the stress relaxation tests, it was observed that the experimental apparatus stabilized very quickly, thus allowing reliable results to be measured. Consequently, in the subsequent presentation of stress relaxation results, the force in the geogrids is, for convenience, normalized by the value measured at 0.1 minute.

Creep Testing

The test setup as described above was also used to conduct creep tests. However, in these tests the load was maintained constant. The reed relay-submerged pump system was replace by two dial gages; one attached to measure the displacement of a fixed point near the top of the specimen and the second to measure the displacement of a fixed point near the bottom. The difference in measured values represents the displacement developed over a known length. Hence, the creep strain could be calculated.

All tests were conducted in a temperature-controlled room. The temperature was maintained between 20 to 24° C and the relative humidity varied between 30 and 50%. For practical reasons (i.e., time constraints and total number of tests), it was decided that each test would be conducted over a period of 30 days.

Scope of Tests

A total of nine different geogrids were tested for stress relaxation and creep. Table 1 provides their designation (type A, B, and C), polymer type, coating type, and manufacturing process. Of each geogrid family, three different types were tested: high, medium and low strength. Table 1 also provides the ultimate short term strength of each geogrid as determined following the ASTM standard for wide-width tensile strength test (D 4595-94). Roller grips were used to clamp the specimens in these tests. At least three ultimate strength tests were conducted on each type of geogrid to ensure consistent results. Test results were compared with strength values reported in the manufacturer's literature; generally the agreement was within less than 5%.

In the stress relaxation and creep tests, each geogrid type was subjected to three levels of load: 40, 60, and 80% of the ultimate load. In the creep tests, these loads were maintained constant. In the stress relaxation tests, these were the initial loads. However, in tests on geogrids type C1, C2 and C3, the 80% value required a stress relaxation rate in excess of the capacity of the testing system. Consequently, these specimen were tested only for relaxation at initial loads of 40 and 60% of their ultimate strength. To ensure repeatability of the results, more than one test was performed on a given geogrid at a given level of load. The results shown in Figure 2 are typical.

It should be noted that all tests were conducted at or above currently allowed load values for design (current creep reduction factors do not allow for tension exceeding 60% of ultimate strength for polyesters and 40% for HDPE). However, current design values are selected such as to produce tolerable creep strain at the end of the life of the structure (typically 100 years); to observe the phenomenon of stress relaxation (or creep) within the test duration (1 month), higher loads are thus necessary. Furthermore, since a major objective of this project is to determine whether the current reduction factors for creep can be further lowered if one accounts for stress relaxation, it is in fact necessary to conduct tests at loads exceeding current recommended values.

4 TEST RESULTS

The stress relaxation and creep test results for the polyester geogrids A1, A2 and A3 are presented in Figures 3, 4 and 5, respectively. In the stress relaxation tests conducted at loads of 80, 60 and 40% of ultimate, the force decreased by an average of 31, 32 and 32%, respectively. The largest variation in measured values was noted for the 60% specimens, where in one case the initial force was reduced by approximately 50%. After one month duration, the rate of relaxation in the A1, A2 and A3 specimens was negligible. In the creep tests performed at approximately 80% of ultimate, all three geogrids exhibited brief periods of tertiary creep that culminated with the rupture of the specimens. Because of the scale used in Figure 3 to 5, the periods of tertiary creep are not readily evident. When subjected to loads that were 60% of ultimate, the three geogrids exhibited rather varied levels of creep strain, and did not enter the tertiary regime. At a load of 40% of ultimate, the strain rate was somewhat lower than for the 60% specimens.

The stress relaxation and creep test results for the polyester geogrids B1, B2 and B3 are presented in Figures 6, 7 and 8, respectively. In the stress relaxation tests conducted at loads of 80, 60 and 40% of ultimate, the force decreased by an average of 23, 29 and 30%, respectively. The variation in values for the B1, B2 and B3 geogrids was much less than for the A1, A2 and A3 grids. When subjected to loads at 80% of ultimate, in the creep tests geogrids B1, B2 and B3 all ruptured (again, due to the scale used in Figures 6 to 8, the tertiary phase is not readily evident). At loads of 60 and 40% of ultimate, the specimens all exhibited a relatively low strain rate and total creep strain.

The stress relaxation and creep test results for the polyethylene geogrids C1, C2 and C3 are presented in Figures 9, 10 and 11, respectively. In the stress relaxation tests, these geogrids exhibited substantial decreases in force. In particular, at a load of 60% of ultimate, the force decreased by an average of 53%; under 40% of ultimate, the average decrease was 58%. In the creep tests, all specimens subjected to loads of 80 and 60% of ultimate exhibited periods of tertiary creep followed by rupture. In the case of the 80% specimens, this rupture was particularly rapid, occurring 100 to 200 minutes after initiation of creep. Under loads of 40% of ultimate, the polyethylene geogrids seemed to have reached a plateau after a few days of loading, thus exhibiting a very low creep strain rate.

To summarize the creep performance of the geogrids tested, note that, with the exception of polyester geogrid A3 (Figure 5), all specimens ruptured when loaded to 80% of the ultimate. As shown in Figure 9 to 11, for the polyethylene geogrids rupture was preceded by the onset of rather pronounced tertiary creep. At 60% load level, all of the polyethylene specimens again exhibited pronounced tertiary creep followed by rupture. Conversely, none of the polyester geogrids ruptured. At a given load level, the polyethylene geogrids exhibited higher creep strain and higher total strain than the polyester geogrids. The initial strains in the polyester geogrids were, however, higher than in the polyethylene geogrids. The above observations are consistent with the findings of Greenwood and Myles (1986), who noted that upon initial loading, polyester yarns exhibit relatively high elongation; the subsequent (creep) strains are, however, quite low.

From a micromechanical point of view, it would be expected that geogrids that exhibit large creep strains should also exhibit large amounts of force reduction under conditions of stress relaxation. This is explained by the observation that in both cases the long-chain molecules comprising the material must somehow re-arrange themselves to eventually come into equilibrium with the applied load or deformation. For the geogrids tested, the above hypothesis was indeed supported. The polyester geogrids exhibited relatively low creep strains and force relaxation. The polyethylene geogrids, on the other hand, exhibited rather large creep strains and relaxation in force.

5 IMPLICATIONS OF RESULTS

Creep tests represent an idealized situation in which the load carried by the geosynthetic is constant with time. Since creep is associated with the deformation of the geosynthetic, soil deformations must occur as well and hence, contrary to the testing procedure, the load in the reinforcement must change as the creep deformation increases. Stress relaxation, on the other hand, requires no movement of the soil and may therefore proceed independently of the soil. Consequently, current design practice, which is based only on creep test results and ignores stress relaxation, might be overly conservative.

Stress relaxation tests represent the opposite extreme of creep tests. That is, the geosynthetic is restrained from creep deformation via reduction of tensile load. In reality this situation represents a case in which load is transferred to the adjacent soil (or adjacent reinforcement layers) as the geosynthetic tends to creep, and thus creep is arrested. Test results presented in this paper imply that as the tendency for creep increases, the potential for stress relaxation increases. In some geosynthetics this relaxation can be as much as about 50% of the initial load (e.g., see results for polyethylene where the initial load of 40% of the ultimate strength has dropped to about 18% after a month). However, without proper verification, it might be overly optimistic to assume that full relaxation will indeed occur for a particular structure.

Real structures exist in an interactive environment in which soil deformation occurs in response to creep, thus causing the load carried by the creeping geosynthetic to decrease due to load redistribution. At a final and stationary state of equilibrium, a more efficient state of stress may exist. For example, in a multiple-layer system, understressed geosynthetic layers may carry higher load whereas overstressed layers may carry lower load. If a single-layer system (or a multiple-layer system with large spacing in between layers) is considered, the deforming soil will start bearing higher load, thus relieving the stress off the creeping layer. In either case, all components of the structure (i.e., all geosynthetic layers and the reinforced soil mass) are more evenly "taxed" relative to their ability to contribute, eventually producing a more efficient structure. The end result could be a design process in which some creep is desirable. This means that the creep reduction factors currently used in design could be reduced. The present work implies that the potential upper limit to this reduction (i.e., full relaxation) is about 50%.

6 CONCLUSION

A test device was devised to measure stress relaxation and creep of geogrids. The test results for three types of geogrid have been presented. All tests were conducted either for one month duration or until rupture occurred. Initial loads of 40, 60 and 80% of the ultimate short term strength were applied. The measured relaxation for polyester grids was about 30% of the initial load; for polyethylene grids it was about 50%.

The present work implies that the potential upper limit to this reduction (i.e., full relaxation) is about 50%. However, additional work needs to be performed in order to assess the validity of this value. In particular, a through numerical parametric study that accounts for the different types of backfills, reinforcement types and layouts, and structural geometries needs to be performed. Requisite to such a study is the existence of realistic constitutive models for the backfill soil and geosynthetic, a method to account for the interaction along soil-geosynthetic interfaces, and a robust computer program in which the above items are incorporated. The scope of this paper has been limited to presenting the results of creep and, more importantly, stress relaxation tests on some typical geogrids. The different aspects related to the aforementioned parametric study will be addressed in future publications.

ACKNOWLEDGMENT

This project is supported by the National Science Foundation under grant CMS-9523141. This support is gratefully acknowledged.

REFERENCES