Introduction Geosynthetics are increasingly being used for engineering applications.They are typically used in conjunction with soil and other types of geosynthetics and have a number of roles such as reinforcement, separation, drainage and as barriers. They are employed in a range of diverse applications such as landfill barriers and construction of steep soil slopes.
The use of geosynthetics in a structure introduces potential planes of weakness, resulting in a requirement to assess the stability along interfaces between soil and geosynthetic and between geosynthetic and geosynthetic. Figure 1 shows a schematic of a landfill containment system. It demonstrates the range of materials that can be used and emphasises the importance of the soil/geosynthetic systems in controlling slope stability of both the capping system and the side slopes during construction (ie prior to waste placement).
Shear behaviour of side slope interfaces also influences post-waste placement barrier performance.
Settlement of the waste can result in overstressing of the geomembrane via transfer of load through shear at the interface and this can lead to loss of barrier integrity.
Figure 2 shows a relatively steep landfill side slope being lined with a geomembrane and illustrates the importance of interface shear strength in ensuring the constructability of such structures.
Assessment of stability requires detailed knowledge of the stress/strain behaviour of interfaces, as post peak shear strengths are often mobilised resulting from the strain incompatibility of soils, geosynthetics and waste materials.
It is common practice to measure interface shear strength in a direct shear apparatus (DSA) as used in soil mechanics but with a much larger shear plane. Design parameters are obtained by carrying out performance tests (ie using site specific materials and relevant boundary conditions).
There are three standards in common use that provide guidance on testing procedures; BS 6906:1991, ASTM D5321-92 and a German recommendation for landfill design GDA E 3-8 of 1998 (Gartung and Neff, 1998). The final version of a preliminary European standard (prEN WI 00189015) is imminent. In addition, a significant number of research papers have been published on this topic in the past 15 years.
It would appear therefore that there is adequate information and guidance to ensure high quality testing is carried out.However this is not the case.There is growing evidence that tests specified to obtain parameters for design, and those reported in the literature, often lack sufficient control on the key factors affecting the measured values. This results in uncertainty regarding the likely variability of measured shear strengths, and in some instances is leading to the use of over estimated interface strengths in design.
This paper provides a summary of the key factors influencing measured shear strength behaviour. It gives guidance, references key publications on the main issues controlling the measurements and includes test results that illustrate typical observed behaviour.
A companion paper (Dixon et al, 2002, to be published in Ground Engineering March) discusses the selection of characteristic strength parameters from laboratory results for use in design.
While the issues included here are important for the assessment of all geosynthetics, there are specific additional considerations for the testing of geogrids, geonets and geosynthetic clay liners (GCL) that are not covered. Some of the recommendations given are relevant for direct shear tests entirely on soil, although care should be exercised in applying the findings of this study as important issues specific to soil tests have not been included.
Variability of measured interface strengths In recent years, research has been conducted to quantify the likely variability of test results and to identify key factors that control measured strengths. As part of the development of the new European geosynthetic test standard, inter-laboratory comparison tests were conducted in an effort to quantify the likely scatter in measured strengths resulting from the use of different operators and test equipment (Gourc & Lalarakotoson, 1997).
Tests were carried out in seven commercial and research laboratories - two each in France, Germany and UK and one in Italy - using geosynthetic materials supplied by the co-ordinator and obtained from one source. The interface shear strengths between a range of geosynthetic materials and standard sand were measured.
Two similar, and complementary, inter-laboratory comparison test programmes were conducted by a working group of the German Society for Geotechnical Engineering in 1995 and 1996, as part of their response to development of the European standard (Blümel and Stoewahse, 1998). The latter programme incorporated a more detailed specification of the testing procedure. These programmes, each involving approximately 20 laboratories, produced a range of measured strengths that is similar to the European study. Figures 3 and 4 show test results from the German studies for a non-woven geotextile v sand interface.
The significant variability of the shear stress v displacement curves in Figure 3 is typical. The different laboratories produced a range of peak and large displacement shear strengths, and widely varying stress vs displacement relationships. Figure 4 shows the distribution of peak failure envelopes obtained by the laboratories.
In addition to the large variation of results, of particular concern is that some laboratories produced high, and hence unsafe, shear strengths. Inspection of the data in Figure 3 shows clearly that some of the results are significantly in error (indicated by the shape of the shear stress v displacement curves). An experienced engineer would not use these results and would require repeat tests to be carried out. By removing this spurious data the variability could be reduced significantly.
However, it is worth noting that all tests were conducted by laboratories experienced in measuring geosynthetic interface shear strength, and that the laboratories knew their results would be compared with those from a large number of other laboratories (ie their competitors). It can only be assumed that those who submitted the spurious test results must have considered them to be correct. This indicates that some laboratories lack the experience to interpret the results they obtain.
There are three categories of factors that lead to variability of measured interface shear strength:
a) test apparatus design b) operator/test procedure c) variability of both geosynthetic and soil materials.
Both the European and German test programmes used clearly defined common test standards and samples from a common source, but involved different operators and a range of different DSA designs.
Repeatability can be improved by using one design of DSA and one operator, although the results may have a consistent error. Test programmes have been carried out under these conditions at Hannover University (Blümel et al, 1996) and Loughborough University (Dixon et al, 2000).
Scatter of results from these tests 'under conditions of repeatability' would be primarily due to variation in the geosynthetic and soil test material. Some results of these studies are shown in Figure 5, together with the results of inter-laboratory tests, as coefficient of variation (standard deviation/mean) v normal stress for interfaces between sand and a geotextile as well as between a geotextile and a geomembrane.
Each point represents a number of tests on materials from the same source conducted at the same normal stress.
The two important trends that can be observed are the reduced scatter of data obtained if tests are carried out in one laboratory (not surprisingly), and an increase in the coefficient of variation with decreasing normal stress for all repeatability testing. The latter is of practical importance to landfill cover system design due to the increased uncertainty in measured interface strengths at low normal stresses.
Unfortunately, rather than provide confidence in the ability to undertake reproducible tests, the results of these inter-comparison test programmes cast doubt on the applicability of aspects of current test procedures. The main factors that result in variability of test data are discussed in subsequent sections.
Detailed evaluation of the results from the inter-laboratory comparison test programmes and discussions with participants indicates that one of the main reasons for the scatter of measured strength data is the different DSA devices used for test performance. Blümel and Stoewahse (1998) investigated some effects of design of DSA device on test results and this work has been extended by Stoewahse (2001).
Most of the DSA used for shear testing with geosynthetics has a top box or frame with a test area of about 300mm by 300mm. The lower part has an equal test area or a box that is longer than the upper one in the direction of shear movement so that the shear area is kept constant during the test. Stoewahse (2001) used four types of DSA with different top box supports and with different loading and load controlling systems (Figure 6).
The floating top box designed by Casagrande is supported at one point only and is able to rotate around this support (Figure 6a).The design of the top box is well known from soil testing and is in accordance with BS 1377: Part 7 and ASTM D 3080. However, there are only a few devices of this type with sufficiently large test areas for interface friction testing.
The type of DSA mainly used for geosynthetic interface testing is constructed in a manner such that the top box is fixed (Figure 6c).
This device was designed specifically to measure interface shear behaviour.The fixed top box was introduced on the premise that a pre-formed shear plane (ie the interface) is located between the bottom and top boxes, and hence formation of a shear plane with associated volumetric straining of material in the top box does not occur.
However, this assumption is wrong for testing of many combinations of materials and hence there are concerns regarding the magnitude and time dependent variation of the vertical stress on the interface during shearing.
Vertical stress is usually applied by air or water pressure via a membrane, and hence is known only on the top of the sample. Friction between the test material and internal walls of the top box both during application of normal stress and shearing will alter the actual vertical stress acting in the shear plane by an unknown amount.
Five of these devices are in use in the UK and more than forty in the US. Approximately 25 of these fixed top box devices are in use in Germany.
A device was modified in co-operation with manufacturer Wille GeoTechnik in Goettingen, Germany to overcome the problem of unknown vertical stress in the shear plane with the fixed-box-DSA (Figure 6d).
The modification allows the average vertical stress acting on the interface to be determined by measuring the vertical support forces to the top box. The pressure applied to the top of the sample is then regulated to keep the resulting vertical force on the interface at a constant value.
Another approach to improve the DSA was made by separating the loading system from the upper box and allowing it to move vertically, but not to rotate during the test (Figure 6b). The vertically moveable top box together with a control system ensures that the vertical stress applied to the interface remains constant during the testing process (Stoewahse, 2001).
This configuration was selected as the standard DSA design incorporated in the German DIN 18 137-3.
Variation of results caused by different types of test apparatus Results of friction tests on the sand-geotextile interface conducted in different DSA-types, all with a shear plane of 300mm square, are presented below. The normal stresses applied were between 20kPa and 200kPa. At each normal stress at least two individual tests with three different densities of dry sand were conducted.
Additionally, direct shear tests on the sand were performed in all devices. Reference values for the internal shear strength of the sand were obtained in triaxial tests. The sand used for the test is the standard sand according to EN 196-1. This sand was also used in the European and German inter-laboratory comparison test programmes.
In Figure 7 the peak shear strength vs normal stress measured with different types of DSA is plotted for the interface between very dense sand and a geotextile. It is obvious that the fixed box device gives considerably higher peak shear strength values than the other DSA. The results obtained with the other three devices seem to be comparable. As discussed above, using the fixed box gave higher values caused by restraint forces (Stoewahse, 2001).
In Figure 8, the friction values d, derived by linear regression, for the sand and geotextile interface are plotted v void ratio of the sand. For comparison, the results of direct shear tests with this sand are shown in a similar diagram together with some data from triaxial tests.These were performed to obtain calibration data for the angle of internal friction f.
It can be seen that for all void ratios investigated the friction parameters measured with both fixed box devices are somewhat higher than those from the other DSA types. The values of f obtained with the vertically movable box are in good accordance with the results of the triaxial tests.
In Figure 9 shear force T and normal force N v displacement s developed during the test are plotted for the DSA types with fixed and with vertically movable top box. In the upper graph the shear force v displacement curves are shown. The applied load was P = 9kN which is equal to a normal stress of 100kPa. This load value P is marked in the lower diagram by a full line.
The observed variation of N in the fixed top box is due to friction forces acting at its internal wall. During shear these forces are increasing and this affects the related shear force T. In the vertically movable box the force N acting normal to the interface remains nearly constant during the whole test and closely matches the applied load P. It can be concluded that the fixed top box design should not be used for tests wholly on soil.
This research has led to the DSA type with vertically movable top box being recommended for shear testing on soil and geosynthetic interfaces in Germany.
Test procedure The four standards listed in the introduction provide useful guidance on testing procedures and evaluation of measured data for both specifier and operator. ASTM gives guidance for performance testing of soil v geosynthetic and geosynthetic v geosynthetic interfaces. BS6906 Part 8 essentially covers only index tests on these two types of interface, although some limited guidance on performance testing is provided in Appendix A.
The proposed European standard is restricted to index tests on standard sand v geosynthetic interfaces.The BS and ASTM are about 10 years old and therefore do not include recent developments, and the proposed CEN document is of limited use for designers, as it only covers index testing. GDA E3-8 is specifically devoted to landfill design and gives detailed recommendations for performance testing of all kinds of interfaces for liner systems and covers, although it is not available in English.
Table 1 summarises the scope and guidance provided by the test standards. This section of the paper provides a summary of the guidance given by each of these standards and comments on key elements of the test procedure, including references to papers detailing recent research.
Example results are given showing the influence of selected aspects of the test procedure.
None of the guidelines specify the construction of the testing device although detailed specification of the DSA exists in all the standards for direct shear tests on soils. As there are many DSA with a fixed top box or similar specifications in use, it is important that a full description of the testing equipment is provided with the test results.The investigating laboratory should comment on the key question of how the effective normal stress on the interface is calculated or measured during shearing (Blümel and Stoewahse, 1998, and Blümel et al, 2000).
In fixed top box DSA, the gap between the top and bottom boxes must be set prior to shearing. Advice from the test standards is both ambiguous and outdated. The gap size must be so small that no soil particles can migrate out of the box but it must also be large enough so that no constraints are induced. This is nearly impossible to achieve in a fixed box DSA as shown above.
Bemben and Schulze (1998) demonstrated that the gap size has a significant effect on the measured strength. Unfortunately, they did not describe the type of DSA they used in the study. The use of a gap size value in accordance with ASTM D5321 and BS6906 can lead to significant errors if additional considerations on the materials to be tested are not made.
Practical experience of the authors indicates that the accuracy with which the gap can be adjusted is not less than ±0.5mm in a 300mm square DSA. This is in the typical range of grain size d 85 for sands, which is recommended as the gap size in both ASTM D5321 and BS6906.
For friction tests with geosynthetics, their compressibility has to be considered. In tests with vertically movable and tilting top boxes the initial gap size is not as important as there is an immediate relief of constraints if the gap chosen at the start of the test is too small.
The thickness of a soil layer placed in the top box and the surface roughness of the load plate can also affect the test results. For a sandgeotextile interface Stoewahse (2001) varied the thickness of the sand layer in the top box and used two load plates, one with a smooth and the other with a rough surface.The tests were performed in a 300mm square DSA with a fixed top box. Results are shown in Figure 10.
It can be concluded that in a 300mm square box a sample thickness of at least 50mm is sufficient for sandy materials. For fine-grained cohesive soils it was found that the sample thickness could be reduced to about 30mm to shorten the time required for consolidation before starting the test. Generally it is recommended that a rough load plate be used for these tests.
Stretching of geosynthetics during friction testing may occur and has to be prevented by modified fixing techniques. If the stress-displacement curves or the geosynthetic samples indicate that stretching has occurred, the test should be omitted and repeated with a sufficient fixing technique.
Tests should be carried out in a temperature controlled environment (20infinityC ±2infinityC) and using materials conditioned to this temperature.
Pasqualini et al (1993) documented temperature effects on friction behaviour of geosynthetic interfaces and showed that they can be significant.
As water is likely to be present in most of the applications of geosynthetics in landfill design it is recommend that friction tests be performed with submerged materials, unless special conditions have to be taken into consideration by the design engineers.
For tests with clayey soils this is very important as such material can absorb water. Swelling leads to a reduction in interface shear strength.
Consideration should be given to whether the softened state is relevant for stability analyses of liner systems. In performance testing of geosynthetic v geosynthetic interfaces it is also important to use site-specific soils in the top box (ie overlying the upper geosynthetic). Jones and Dixon (1998) showed that grading, particle size and particle shape have a direct influence on the shear strength of a geomembrane v non-woven geotextile interface. In a liner system the soil above a geotextile affects the micro scale distribution and magnitude of normal stress at the friction interface.
The shearing rates specified in the standards for geosynthetic v geosynthetic and sand v geosynthetic tests are appropriate. Stark et al (1996) and Stoewahse (2001) showed that peak shear strength values are independent of shearing rate for the range 0.03mm to 40mm per minute.
However, for performance testing, appropriate shearing rates must be specified according to the critical conditions expected on site (ie drained or undrained). Drained tests can take many hours or even days when involving cohesive soils and therefore are seldom carried out, Reporting of results The evaluation of different landfill slope failures has shown that a contributing factor is a lack of communication between the participating parties. It is important to inform the testing engineer about the project details and the site-specific conditions to ensure correct testing conditions are applied and thus appropriate results obtained.
In addition, the test method and results must be documented in a manner such that design engineers - who might not be knowledgeable about the testing practice - are able to interpret the results and use them in stability calculations to produce a rigorous design.
From the inter-laboratory comparison test programmes it has been found that even apparently minor changes in testing conditions can affect the results significantly. Therefore, a detailed test report is necessary and must include information on the following:
l Description of test device: Including support of the top box, load application, etc.
lTest set up and boundary conditions: Shearing rate, samples tested dry or submerged, consolidation time, method of fixing the geosynthetics to resist stretching, exact location of interface in relation to top and bottom boxes, gap between top and bottom boxes (if relevant), placement method of soils (eg compaction effort and layer thickness) and density and water content before and after the test.
lFull material descriptions: Geosynthetics: manufacturer, mass per area, thickness, polymer, description of the structure, etc; soils: origin, soil mechanical classification, other mechanical parameters.
lDescription of sampling methods employed: Geosynthetics: sample preparation, eg pre-soaking; soils: any form of pretreatment like crushing of aggregates, drying, adding of water.
lTest results: Shear stress v displacement curves, peak shear stress v normal stress plots, large displacement shear stress v normal stress plots, volumetric changes v displacement if relevant, soil mechanics parameters at beginning and end of test, shear strength parameters d and a and the method of derivation (eg linear regression).
lState of the materials after the test: Stretching of the geosynthetics, abrasion of geomembrane textures, orientation of geotextile fibres, post-shearing damage such as tearing of stitch bonding or welding points, development of additional shear zones in geocomposites and also in soils, changes of water content etc.
The list above is necessarily incomplete. With the development of new geosynthetics other aspects might become important. For example, creep aspects are not usually considered in short-term friction tests but can occur under compression during application of the normal stress (eg when testing geocomposite drains made from polymer foam pieces).
Further details are given in the standards summarised in Table 1, especially in GDA E3-8. Procedures for evaluation of test results and derivation of characteristic values are presented in the companion paper 'Geosynthetic interface shear behaviour: Part 2 Characteristic values for use in design' (to be published in GE March 2002).
Summary Interface shear strengths involving geosynthetics are often measured incorrectly. At present, available test standards provide limited guidance and this is leading to inconsistencies. Results of inter-laboratory comparison testing programmes in Europe and Germany have shown a large scattering of friction test results.
A significant element of the data scattering is believed to be due to the construction of the testing devices.
Additional research on the influence of several testing boundary conditions has been undertaken at Loughborough University, UK, and Hannover University, Germany. Experimental and numerical investigations have shown a significant effect of the support of the upper box on the test results. Kinematical restrictions of the upper box cause constraint forces in the system and this leads to high values of measured interface shear strengths.
A direct shear apparatus with a vertically movable top box has been developed at Hannover University.
Friction tests on different geosynthetic interfaces gave reliable values of shear strength. Results of direct shear tests in this device on sand are in good agreement with triaxial test results. The direct shear device with a vertically movable box is recommended as a standard device for interface testing.
Moisture conditions, temperature, particle size and grading of the soils involved are some of the many additional factors influencing test results.This paper gives advice on the selection of testing conditions and other factors that must be reported with the results and considered in the application of the test data.
Information on the variation of test data and guidance on obtaining characteristic values of interface strength parameters for use in design calculations is given in the companion paper.