Geotechnical Characteristics of Soft Lake Marl with Preload Test Results
Mark R. Muszynski, P.E.
Geotechnical Engineer, Gosling Czubak Engineering Sciences, Inc.,
Charles D. Brumbaugh, P.E.
Manager of Geotechnical and Materials Testing Services,
This paper describes the compressibility and consolidation characteristics of lake marl using both laboratory testing and a field preload test to predict the short term and long term settlement behavior on a residential development site. An estimate of primary consolidation settlement was made based on laboratory test results and later compared with the field preload test settlements. Secondary compression settlement predictions were made for the site based on the laboratory test results. Practical engineering assumptions about the lake marl and a reliability analysis, using the laboratory and field data gained to help predict settlement and strength, are discussed. Considerations related to developing grading and preloading recommendations are identified.
KEYWORDS: lake marl; consolidation; compressibility; soft soil; soft ground engineering.
Lake marl is common in many areas of the world and is sometimes found around existing inland lakes or older lake sites that have since completed eutrophication and are now hidden. This soil is characterized by its light color, low plasticity, high silt fraction, and variable shell content. Lake marl is typically very soft with relatively low undrained shear strengths, modest preconsolidation pressures at shallow depths, and high compressibilities under load. A pronounced secondary compression component of the total settlement is also commonly observed. For these reasons, lake marl is considered problematic for the purposes of foundation and embankment support.
The main purpose of this paper is to show the results of laboratory soil testing and a field preload test on a site underlain by lake marl. Laboratory testing including strength tests, consolidation tests, and consistency tests are described. Attention is paid to how we maximized the information gained in the laboratory and field testing and to arrive at the final grading and preloading recommendations.
A residential development project was planned for Harbor Springs; a small Great Lakes town in the northern, lower-peninsula area of Michigan in the United States. The proposed development consisted of the construction of seven residences on an approximate 0.8 ha (2 acre) site. Two general sections of the site were identified. The first section involves the addition of soil on an existing slope from a road to the south onto the subject site towards the north. The project included grade changes on the site using imported fill. First, the embankment slope would be shifted to the north as a result of these grading plans. In general, up to about 2.7 m (9 ft) of imported fill was necessary in this area to move the embankment to the north and increase the flat area. The second section of the site included the addition of approximately 0.6 m (2 feet) of imported fill to raise the grade over the relatively flat area of the site on the north side. This second section of the site is larger in areal extent than that of the first section as shown by the approximate contour lines on the site map in Fig. 1.
Figure 1. Overall approximate site plan prior to grading and construction, N.T.S. (North is at the top of the Figure)
Nine soil borings were undertaken on the site to help define the soil conditions. The soil types varied with depth, while the stratification was relatively uniform. The site generally contained about 12 inches of black topsoil underlain by soft light gray to white lake marl with occasional tiny shells to depths of about 4 m (13 ft) below grade, underlain by an approximate 1.2 m (4 ft) layer of clayey silt, which was followed by medium dense sand and gravel, and dense silt soils to depths of up to 16.7 m (55 feet). A generalized cross section of the soils encountered is shown in Fig. 2. Groundwater depth was about 0.9 m (3 ft) below grade in most areas of the site at the time of the original site exploration. The groundwater was deeper at other areas where the surface elevation was greater.
Figure 2. Generalized cross section of the site, N.T.S.
A pile foundation system was selected for support of the residences to bypass the soft existing soils. The client preferred this approach because bearing strata existed at relatively shallow depths, materials and labor were readily available for that particular foundation type, and piles have been used successfully on other sites with similar underlying granular bearing soils.
However, another problem existed related to the soft soils: In placing the proposed imported fill, we estimated that several inches of settlement in would occur in some cases due to consolidation and long term compression of the lake marl. Concerns about pile dragdown related to this settlement, long term integrity of the underground utilities, and maintaining grades were also raised in light of the grading plans. The strength of the soft soils, when loaded with fill, was also in question. One other related concern of the client was the slight “jelly” sensation detected as a large vehicle would pass on the nearby street.
By the instruction of our client, we began a more detailed study on the soft lake marl. The study included both laboratory testing and field testing. The objective of the study was to determine the settlement potential of these soils over time, and provide recommendations regarding site work and how to bypass these potential problems.
The lake marl specimen that was subjected to consolidation and laboratory testing was obtained from a depth of 1.2 m (4 ft) below grade near the proposed field preload test area using a Shelby Tube. The sample was transported to the laboratory being careful not to disturb or subject the soil to excessive vibration. Three specimens were obtained from this Shelby tube. Each of the specimens was subjected to 1-D consolidation tests in order to obtain compressibility parameters in the overconsolidated (OC) and normally consolidated (NC) stress ranges, secondary compression behavior, and quasi-preconsolidation behavior. This quasi-preconsolidation condition becomes apparent after a period of secondary compression (aging), when an otherwise NC specimen is loaded to a greater pressure level than secondary compression stress level. The result is a curve in e-log p’ space, similar to what would be noted in an OC sample. The behavior has been documented by several researchers including, Schmertmann, (1991), Mesri (2005), and others.
The initial consolidation test was treated as a trial run in that the results were not weighted as heavily as subsequent tests. In our experience, it is a good idea to conduct a preliminary test in order to get a feel for the specimen’s consolidation characteristics. Reasons for this are to identify issues during specimen trimming, to obtain general specimen deformations during the test, and to determine the general time to reach 90% consolidation. This pre-test ultimately helps to obtain higher quality data from the remaining tests to be used for computations. Other laboratory tests performed included Atterberg limits, specific gravity of solids, Torvane measurements, reaction to dilute HCL, dry strength evaluation, and natural moisture contents.
The present effective overburden pressure, svo’, was estimated to be 15.6 kPa (325 psf) at the depth the lake marl sample was obtained. The specific gravity of solids, Gs, of the lake marl is 2.68 based on results using a helium pycnometer. Table 1 shows some of the geotechnical characteristics of the lake marl. Figure 3 shows a grain size distribution curve for a representative lake marl specimen. As expected, the lake marl had a very high CaCO3 content as evidenced by its strong reaction to dilute HCl. The oven dry specimens were friable and had low strengths. The natural water contents of the lake marl specimens obtained in the soil borings were between 40% and 130%, with most between 50-70%. Using the apparent specific gravity obtained on the solids, and assuming near-saturated conditions, the void ratio of the specimens ranged from about 1.1 to 3.5. During the Atterberg limits testing, we noticed that the intact specimen became slightly softer and lost some amount of strength during reworking. This behavior would suggest that the lake marl is sensitive to some degree, and this is expected of this soil type.
Figure 3. Grain size distribution of the lake marl
Table 1. Geotechnical characteristics of the lake marl
Figure 4 shows the results of a consolidation test on the lake marl. This test was performed using a Brainard Killman pneumatic odemeter. Several items of interest are apparent in these test results:
Note the relatively well defined curve near the preconsolidation pressure and the low strain to reach svo’ (15.6 kPa (325 psf)), both of which indicate that obtaining a reasonably undisturbed sample of the lake marl is possible using conventional techniques, combined with due care during sampling and transporting. We used a small stress increases (low load increment ratio) to better define the preconsolidation stress on the specimen.
The recompression curve formed a well defined break at the preconsolidation stress created on the NC curve.
During test #2, the specimen was subjected to a sustained effective stress of 60 kPa (1,250 psf) overnight. This short aging duration gave rise to a decrease in void ratio as shown in Fig. 4 at Point “A.” When reloaded with a small stress increase, an indication of a quasi-preconsolidation stress was observed. This behavior has been observed by the author on other fine-grained and coarse-grained soils and is described by Schmertmann (1991) as well. This quasi-preconsolidation effect is significant because it implies that with aging, an increase in strength and a decrease in compressibility occur.
Figure 4. Consolidation test results of lake marl 1.2m bgs
After observing the effect of aging on the specimen in Fig. 4, we conducted another consolidation test for the purpose of subjecting the specimen to a longer duration aging cycle. A new specimen was trimmed from the same Shelby tube and subjected to a consolidation test with the same load increments. This time, as the preconsolidation stress was exceeded, the specimen remained with an effective stress of 60 kPa (1,250 psf) over the course of about 20 days (Pt. “B” in Fig. 4). This pressure was a conscious decision because of the site plans, and is greater than the field preconsolidation pressure of 33.5 kPa (700 psf). The grading plan called for as much as 2.7 m (9 ft) of imported fill in some areas of the site. This amount of new fill would deliver a total stress increase of approximately 57.5 kPa (1,200 psf) to the underlying soils. With constant pressure used, this test was used to estimate the secondary compression coefficient, Ca for the lake marl. Terzaghi, et. al. (1996) explained that the secondary compression value is greater in the NC range than under effective pressures less than the preconsolidation pressure. Mesri and Vardhanabhuti (2005) also discuss this consideration for secondary compression.
Fig. 5 shows the results of the secondary compression test on the lake marl. The expected curve, from first loading at this stress level, is apparent. The slope of the curve becomes flatter, as expected, when the theoretical 100% consolidation stage has been attained at Point “C.” The theoretical 100% consolidation point by the Casagrande method required about 115 minutes for an approximate 2.3 cm thick specimen under single drainage conditions.
Figure 5. Secondary compression test on lake marl 1.2m bgs, 60kPa
An example of the Taylor chart obtained during the testing is shown in Fig. 6. Based on the Taylor charts obtained during consolidation testing, the average time to reach 90% consolidation was about 5 minutes for an approximate 2.3 cm (0.92 in) thick specimen under single drainage conditions (Pt. “D” in Fig. 6). The average coefficient of consolidation, cv, based on multiple tests in the NC range was 0.012 cm2/s. The large value of cv is probably due to the rather high permeability of the lake marl arising from the substantial silt and sand content, and low (under 5% by mass) clay size fraction. Table 2 shows the results of the some of the laboratory testing performed.
Table 2. Average Consolidation/strength characteristics of
the lake marl specimens at 1.2m bgs
Figure 6. Taylor chart-lake marl specimen at constant effective stress of 60 kPa
Shear strength of the lake marl was evaluated using the SHANSEP (stress history and normalized soil engineering properties) relationship discussed in Ladd (1991), A.K.A., “the c/p ratio modified to include OCR” as shown in Eqn. (1). Torvane tests were also conducted directly on representative samples to observe as another indication of the su/svo’ ratio at the test depth.
|su/svo¢ = (S)(OCRm)||(1)|
For soils such as silts, organic soils (excluding peats), and clays with shells, S = 0.25 and m = 0.8 were selected as starting parameters (Ladd, 1991 & 2003). The lake marl on our site appears to fit this description, and the results of su seemed to be reasonable and relatively consistent with the shear strengths obtained using the Torvane device at the test depth. Using these tools, the undrained shear strength at the sampling depth (1.2m) is estimated to be about 8.4kPa (175psf). Equation 1 also provides the means to estimate su at other depths below grade.
PRELIMINARY FIELD TESTING: SURCHARGE PAD AND PRELOAD TEST
The preload test was planned for the lower (flat) area of the site near the northeast corner as shown in Fig. 1 and Fig. 2. This area was more accessible at the early stages of this work and since the soil strata were relatively consistent across the site, this area was judged to be representative for our purposes. This is also the area of the site where the Shelby tube sample used for the majority of the laboratory testing was obtained.
The original plans called for the surcharge pad to be 1.2 m (4 ft) thick by 9 m (30 ft) by 9 m (30 ft) at the crest. We calculated that about 10 cm (4 in) of primary settlement would occur within 1 month with the preloading test program planned. The height and width of the surcharge test pad was selected for several reasons: First, the total stress added to the underlying lake marl soils would be near or they would exceed the preconsolidation pressure at many depths within the lake marl stratum assuming an OCR attenuation with depth. This is important as the settlement within the NC range was desired to help predict how the thicker imported fills on the south side of the site would react. Second, the pressure added throughout the day would not exceed the strength of the soil, especially since the surcharge sand was placed as described in the following paragraphs. Third, the width of the pad would stress the lake marl as uniformly throughout the stratum as practical with a field test of this type. It was recognized that there would indeed be some stress attenuation with depth, however. Fourth, the size of the surcharge pad would allow complete placement within a single day.
To help monitor the amount of pore pressure generation within the lake marl, an open standpipe piezometer was installed on the site, at the center of the proposed surcharge pad. The piezometer tip was installed at the center of the proposed test pad several days before sand placement. The piezometer was installed to a depth below grade of 2 m (6.6 feet), and consisted of a PVC tube with a perforated end of about 15 cm (6 in) long. The piezometer was placed at an elevation known to be the approximate center of the lake marl stratum. At 2.54 cm in diameter, it was recognized that the water elevation increase or decrease would be rather unresponsive and demonstrate a lag time from loading, but it would give a general indication of the pore pressure during loading. The piezometer was intended to be used as an indication of pore pressure generation (increase or decrease) within the underlying lake marl soils.
Five settlement monitor plates were fabricated and placed within the proposed surcharge pad as shown in Fig. 7. The monitor points were intended to provide settlement information at the ground level beneath the surcharge pad. An additional four monitor plates were placed about 1.8 m (6 ft) outside of the surcharge pad at the corners as shown in Figure 7. The purpose of these monitor points was to supplement the information gained using the piezometer, and help provide a warning of soil yielding, in the form of upward movement at these locations. These monitor points outside the test area were referred to as “ground monitor” plates during the course of the field preload test.
Figure 7. Surcharge pad monitor point locations, N.T.S.
After placement of the monitor points and piezometers, the field preload test began mid-morning on June 13, 2004. Imported sand fill was delivered and placed using a small skid steer front end loader. The sand fill was tracked in, i.e., without formal compaction techniques. During placement of the fill, density tests were conducted in order to help determine the actual moist unit weight of the fill being placed. The average unit weight was about 17.28 kN/m3 (110 pcf), with a moisture content of about 3%. The final height of the fill was about 1.16 m (3.8 ft), with a pressure at the base near the center of 20 kPa (420 psf) as intended. Photos of the monitoring points, surcharge fill placement, and the completed surcharge pad are shown in Figs. 8, 9, and 10.
Figure 8. Photo taken looking west showing the monitor points and settlement plates
prior to surcharge sand placement. The trees to the left (south) are blocking the view of the existing embankment.
Figure 9. Photo taken looking towards the northeast during surcharge placement
Figure 10. Completed surcharge test pad. Photo taken looking northwest.
The surcharge pad was placed over the course of about eight hours. No indication of soil yielding was observed during the placement. We came to this conclusion because the settlement appeared to be proceeding at a rate that was expected, no surface cracking (indicating excessive lateral spreading) was observed, the surrounding “ground monitor” or “heave” plates did not register any movement during the placement, and the water level did not rise to a height that would suggest the soil was nearing the yield stress of the soil based on preliminary estimates.
PRELOAD TEST RESULTS
The settlement results of the preload test are shown in Fig. 11. All five settlement plates moved downward over time, at an ever-decreasing rate as expected. We observed up to about 7.9 cm (3.1 inches) of settlement over the course of a month for the center settlement plate. This settlement plate is the closest approximation to 1-D consolidation behavior because of its location within the surcharge pad. This rate was relatively consistent with our time rate estimates. We estimated that approximately 30 days would be required for 90% of the consolidation to occur in the field for the preload test. The settlement, as measured by the center plate, appeared to be nearing the end of primary consolidation as 30 days approached. According to a Taylor curve constructed using the center plate data, T90 appeared to have occurred at about 20 days after initial loading. Our predicted settlement magnitude was within about 30% of our estimate, with our settlement estimate near 100% consolidation being 10 cm (4 in). There was a dramatic difference in the settlement displayed on the western plates as shown in Fig. 11. We believe that this is due less to the variations in the soils, and more to those plates being too close to the edge of the surcharge pad. The center plate is the closest representation of the 1-D consolidation condition.
Figure 11. Settlement plate elevation chart for surcharge pad
Fig. 12 shows the center settlement plate, piezometer tip elevation, and water elevation within the piezometer. The water elevation was corrected for vertical movement of the piezometer. The elevation of the piezometer tip indicates that it moved about half the amount of the center settlement plate placed at the original surface. This behavior was anticipated since the settlement at the original surface should be greater than at other elevations within the stratum because of the direction of settlement with respect to the surface elevation. However, since the piezometer tip may have been subject to dragdown from the lake marl soils above, using the tip movement as an indication of settlement at that discrete elevation is dubious at best.
The tip of the piezometer was installed to a depth of about 2 m (6.6 ft) below original grade. The groundwater levels recorded in the piezometer showed an increase in pore water pressure near the tip during loading, and a subsequent lowering as consolidation continued, as expected. The lag time of the water level within the piezometer, as mentioned before, precludes the direct use of these measurements in determining time for consolidation to occur. Other soil mechanisms, such as capillarity, and the notion of wetting and drying cycles, and movement of the piezometer combined with the additional localized pore pressures created may also contribute to the fact that the pore pressure did not fully return to the original groundwater level.
The most important information gleaned from the use of the piezometer was that the lake marl appeared to be on the wet side of critical state. That is, when sheared, the lake marl tends to contract, leading to elevated pore water pressures within the soil matrix immediately after loading. This would lead to lower shear strengths during loading initially. Because of this, the short term (undrained) case is the most critical to analyze for stability considerations.
Figure 12. Center settlement plate with piezometer
PRELOADING PLAN CONSIDERATIONS
Based on the results of the borings, the laboratory testing, and field testing, the site required a preloading program to reduce post grading settlements. We anticipated settlements over time would be up to 2.5 cm to 5 cm (1 to 2 in) in the areas where 0.61 m (2 ft) of new permanent fill would be added and up to about 45 cm (18 in) in the area where about 2.7 m (9 ft) of new permanent fill was planned. These estimates include settlement arising from primary consolidation and secondary compression over about 100 years.
A reliability analysis was undertaken using the method described by Duncan et al. (1999). This is a first order second moment (FOSM) method, and it uses the Taylor series to estimate the coefficient of variation (COV), and ultimately the probability of exceeding an end result with a given set of parameters, each with defined lower and upper bounds. These analyses indicated that our primary consolidation settlement estimates had an approximate 30 to 40% chance of being met or exceeded, and a 10% chance that the actual primary settlements would be about double of what we estimated.
Probable time for 90% consolidation of the approximate 4.2 m (14 ft) thick lake marl stratum would be on the order of 20 to 30 days. Time to reach 50% consolidation was anticipated to be about 7 days, with about 2 days necessary to attain 30% consolidation. Based on examination of the silty soil underlying the lake marl, we judged that it would not tend to impede the drainage of the lake marl appreciably. Because of this, we anticipated drainage at the top and the bottom, for a total drainage path of about 7 feet beginning at the center of the lake marl stratum. It should be noted that the silt stratum was included as additional lake marl thickness in the time-rate estimates to provide an additional “margin of comfort” in our estimates.
In a similar fashion to the settlement predictions, the same type of reliability analysis was conducted to gain a better understanding of how reliable our consolidation rate estimates were. We found that the rate estimate pertaining to 90% consolidation occurring with 30 days was about 60 to 70% reliable. In other words, there was a 30 to 40% chance that it would take longer to arrive at 90% consolidation. If the silt underlying the lake marl did completely impede drainage during consolidation, the single drainage conditions revealed that about 4 months would be necessary for 90% consolidation. There was a 10% chance that up to about 8 months would be required to attain 90% consolidation. This was viewed as a worst case scenario concerning consolidation rate.
These reliability analyses are intended to be used as general indicators concerning the suitability of our estimates with the information we had available. They were used to gain an understanding of the anticipated behavior of the lake marl under loading, and to help temper our overall judgment of soil behavior on the site.
The estimated time required to reach varying degrees of consolidation of the lake marl, along with grading schedule, were important considerations to this project. For example, based on our initial discussions with the earthwork contractor, we learned that no more than about 0.6m (2 feet) could be placed over the entire site in a normal work week. With this in mind, we decided that the lake marl would be able to consolidate at least 50% prior to placement of subsequent lifts. This consolidation would give rise to increased strength of the lake marl during earthwork, and the possibility of a shear failure of the lake marl would be lower.
Based on the laboratory testing, and engineering judgment, the OCR of the lake marl soil directly below or very near to the man-made embankment was likely just above unity. The OCR values obtained on the relatively flat areas away from the road embankment were approximately 2.2 at 1.2 m deep, and then they seemed to decrease to a value closer to unity at lower depths within the lake marl stratum based on moisture contents and compressive strength estimates. This degree of overconsolidation at the specimen depth considered may have been due to the effects of aging, groundwater fluctuations, chemical changes, past loading or other factors. Because of the moderate OCR values near the surface, and their attenuation with depth, past desiccation events may have been the primary cause of the overconsolidation of the lake marl observed on the site. These overconsolidation mechanisms are discussed by Schmertmann (1991) and Terzaghi et al. (1996).
A creep component (unrelated to that of secondary compression) of the long term settlement was considered. To lower the potential for creep-related movement, it was our opinion that the total stress placed on the soil should be limited to between 50% and 75% of the undrained shear strength of the soil. Because of this, a staged loading program pertaining to the grading plan was recommended. It was necessary to load the lake marl in such a way that the soil gains strength, in stages, as load is applied. This method, the undrained strength analysis, USA, is discussed by Ladd (1991 and 2003). The USA is a method that is useful for evaluating the strength of saturated soils where the preconsolidation stress will be exceeded during grading. The USA takes into account the increase in soil strength as consolidation occurs, relative to the load caused by subsequent lifts placed over the site. In this manner, the undrained shear strength of the lake marl at a given depth remained well above the increase in total stress arising from the load applied at the surface level during grading.
Fortunately, the client’s schedule allowed for placement of the imported fill and surcharge fill over the course of the fall, and the surcharge was allowed to remain on the site for a period of over four months. This amount of time allowed the preloading program to progress in such a manner that the objectives set forth during the planning stage were achieved. Also, if double-drained conditions existed, we judged that the grading schedule allowed ample time to reach about 90 to 95% consolidation and some subsequent secondary compression. If single-drained conditions existed, greater than 90% consolidation would likely occur over the course of the earthwork activities (about 4 months).
A summary of the grading recommendations based on the laboratory and field testing completed are as follows: The flat area to the north of the site, which would be raised permanently by up to 0.6 m (2 ft) would be preloaded by adding an additional 2 feet of surcharge fill across the area. This surcharge would remain on the site for period of about 1 to 2 months. The area to the south containing the existing embankment, the site of the proposed 2.7 m (9 ft) of new soil, would also be preloaded. During earthwork in this area, the contractor would place no more than 3 feet of soil in a given week on this area to raise grade and the fill would be placed uniformly and compacted to a specified degree. When the proposed grade is reached, an additional 1.2 m (4 ft) of surcharge would be placed. The surcharge would also remain in place over the winter for a period of about 4 months since the schedule allowed it. During the preloading program, the proposed 1(v):3(h) embankment slope would be temporarily decreased to about 1(v):4(h), and the surcharge over the relatively level area to the north would be slightly thicker at the toe of this proposed slope. This would help reduce the chances for slope instability during the preloading period. We conducted slope stability analyses for cases before, during, and after preloading. With the recommendations given, we predicted the slope would be stable with a factor of safety against a rotational slope failure of about 2 immediately after loading during each of the grading stages planned.
Since the construction schedule allowed for preloading durations up to 2 months longer than necessary to reach the desired degree of primary consolidation and secondary compression it was decided that an exhaustive monitoring plan would not be necessary. Instead, we observed grading activities to be certain that grading was progressing in general conformance with our recommendations. Some limited monitoring near the toe of the slope and near the crest was undertaken. Movement of the plates installed near the toe of the slope was not detected during grading. The settlement monitor points near the crest of the embankment were damaged early on by construction traffic. Perhaps we made these points look too much like targets.
A site containing soft lake marl soil was improved using a preloading program prior to final grading. Laboratory and field tests were completed to help provide grading and preloading recommendations for the site. After the final preloading program was complete, the client reported that the effects of passing traffic were less pronounced than before treatment of the lake marl soils. This was due to a strength gain of the soils during preloading. Although the settlement of the imported fill and surcharge fill was not extensively monitored during the main preloading period, the results of the laboratory and field testing helped to understand how the lake marl would respond to the applied load, and ample time was allowed for preloading the site.
The first residence was constructed in 2006, approximately 1 year after completion of the preloading program in 2005. The remaining residences will be constructed in the future. A permanent road was installed on the site immediately after surcharging was complete. It has been reported that the road is performing well and holding the proposed grades.
The residences within this development will be established on pile foundation systems bearing on the competent sand and gravel soils below the soft lake marl soils. The utility connections and foundation system for the first residence constructed, permanent access road, and general site grades all appear to be performing as intended as a result of the preloading program undertaken on this site.
We wish to thank the Cottage Company of Harbor Springs, Michigan for allowing us to share this example of a successful preloading program. We're grateful to Ralph Hodek and Stan Vitton for the use of their laboratory facilities at Michigan Technological University in Houghton, Michigan, and Torsten Mayrberger, also at Michigan Tech, for his assistance in obtaining the long-term compression test data.
|© 2007 ejge|