Frost Park 2003
Frost Park 2003
Frost Park 2003
1
Paper ID GTJ20039850_261
Available online at: www.astm.org
ABSTRACT: Current field sampling techniques are not readily able to produce high-quality undisturbed granular soil specimens for laboratory
testing at an affordable cost. Accordingly, numerous specimen reconstitution methods have been developed for use in the laboratory. Among these
methods, moist tamping has the advantage that it is relatively easy to control the global specimen density achieved, even for loose specimens, al-
though unfortunately the method has also been qualitatively shown to yield less uniform specimens. This paper describes the findings of a study that
critically assessed the moist tamping method by measuring the forces applied during the tamping process and investigated the uniformity of speci-
mens prepared with the method by quantitatively analyzing both X-ray and optical images.
Copyright © 2003 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. 1
2 GEOTECHNICAL TESTING JOURNAL
B
FIG. 1—(a) Real image of ASTM graded sand; (b) particle-size distribution curve of ASTM graded sand.
FIG. 2—(a) Schematic view of moist tamping system; (b) moist-tamped specimen with under-compaction ratio of 3%.
FROST AND PARK ON MOIST TAMPING TECHNIQUE 3
Moist Tamping Equipment ments that a moisture content of 5% was adequate for moist tamp-
ing the ASTM graded sand (Jang 1997). To minimize the evapora-
To achieve a specific density in each layer, predetermined tion of water during compaction, predetermined weights of dry
amounts of soil were compacted into seven layers of equal thick- sand and water for each layer were placed in labeled containers and
ness. Figure 2 shows a schematic of the moist tamping system and covered with a lid before being mixed. The sand in the containers
the target layer densities for a moist-tamped specimen with an un- was transferred into the mold by spooning about 15 20 g each
dercompaction ratio of 3%. A collar attached to the tamping rod time while ensuring as uniform spreading as possible. Each sand
was adjusted prior to the compaction of each layer to ensure that the layer was then compacted to the specified density.
tamping disc or shoe could not be advanced beyond the desired
elevation during the compaction of that layer. For example, when Forces and Stresses Applied to Moist Tamped Specimens
the second layer was being compacted, the position of the reference
collar was adjusted to make the distance between the tamping disc The force applied during each tamping effort was measured with
and the top surface of the bottom platen twice the distance used in the load cell integrated within the tamping rod as shown in Fig. 2a.
preparing the previous layer. To measure the compaction force ap- The time sequence of loads applied on one layer (fourth layer of
plied in each tamping effort without being influenced by factors 10% relative density specimen) is shown in Fig. 3 for illustrative
such as friction between the tamping rod and the linear bearing, or purposes. The recorded data from the load cell is shown in Fig. 3a.
possible eccentric loading on the load cell, a load cell was located By discarding tamps and/or time periods for which negligible force
between the bottom of the tamping rod and the tamping disc. The was recorded (less than 2.25 N) and identifying the peak force in
load cell used in this study was a Futek L1650 that had a diameter each tamping effort, the load cell readings were processed as shown
of 25.4 mm, a height of 12.7 mm, and a capacity of 45.3 kg. The di- in Fig. 3b, where the horizontal axis is the tamp number. The cu-
ameter of the load cell was selected to be smaller than that of the mulative force transferred to a layer was obtained by summing these
tamping disc (35.5 mm) to avoid imposing limitations with suc- peak forces as shown in Fig. 3c. Generally, the cumulative com-
cessful compaction at the edge of the specimen mold. pactive effort applied to one layer can be divided into three stages:
Each specimen was 71.1 mm in diameter and 142.2 mm in height initial compaction, main compaction, and final compaction (Fig.
with an initial moisture content of 5% and was compacted in seven 3c). These three portions can be identified by comparing the slopes
layers in a split compaction mold. It was found through experi- of the cumulative force plot. Prior to the beginning of compaction,
FIG. 3—Example data of moist tamping test: (a) original recorded force data; (b) peak forces from recorded data; (c) cumulative forces in a layer.
4 GEOTECHNICAL TESTING JOURNAL
the soil is in a very loose condition because it is placed into the mold during initial tamping. Nevertheless, during this stage, the surface
by spooning, and the capillary forces due to the added moisture pre- becomes approximately level. In the last stage of compaction, i.e.,
vents the collapse of the soil. Therefore, it is relatively easy to push final compaction, forces transferred to the soil are relatively small;
the tamping rod into the soil during the early stages. This stage is re- however, the predetermined thickness for the layer was typically not
ferred to as initial compaction. In this stage, the full area of the tamp- achieved until this stage. The compaction of a particular layer was
ing foot surface is typically in contact with the soil since the soil is determined to be complete once negligible forces were transferred
loose and deforms freely by a kneading action. During the next to the soil and the surface of the layer appeared to be level. A total
stage, called main compaction, a greater compactive force is re- of 13 moist-tamped specimens for which force data was acquired
quired in each tamping effort, especially for dense specimens. The were reconstituted during this study. Examples of incremental and
maximum force transfer to the soil occurs during this second stage. cumulative forces transferred to each of the seven layers in speci-
The contact area between the tamping disc surface and the soil be- mens with global relative densities of 10, 40, and 75% are shown in
ing compacted in this stage is not necessarily the full area of the disc Figs. 4, 5, and 6, respectively. Cumulative forces for these speci-
surface because of some irregularities in the layer surface formed mens are summarized in Table 1. It can be seen that both the indi-
TABLE 1—Cumulative force data from load cell in moist tamped specimens.
Layer
Density Specimen Average
% ID 1 2 3 4 5 6 7 (kN)
Layer
Density Specimen Average
% ID 1 2 3 4 5 6 7 (kPa)
the average peak stress and the average cumulative force increase
exponentially, reflecting that the total energy required to prepare
dense moist-tamped specimens is significantly greater than that
required to prepare loose specimens. The results in Fig. 10 can also
be used to indicate what the initial confining pressure should be
greater than for triaxial testing of moist-tamped specimens of
ASTM graded sands prepared at any given density if the specimens
are to be tested in a virgin stress state.
FIG. 10—Average values of cumulative force and peak stress for differ-
ent relative densities. FIG. 11—Expanded view of coupon prepared for X-ray image.
10 GEOTECHNICAL TESTING JOURNAL
Developed X-ray films were placed on a back-illuminated stage, loose material appears to exist immediately below the layer also re-
and images were captured using a digital camera and an image flects the effects of the scarifying of the upper surface of each layer
analysis system. The uniformity of the specimens was analyzed by prior to placement of the next layer. The darkest zone at the top of
investigating the gray scale variations of the resulting image. Black the particular specimen shown in Fig. 12 is due to the presence of
was assigned a value of zero, while white was assigned a value of some air bubbles in the resin following the curing process. Glob-
255 on a gray scale. In the present study, a lower (darker) gray scale ally, the gray scale is incrementally decreasing (becoming looser)
value implied a looser sand state because more X-rays were able to from top to bottom, indicating that the undercompaction ratio used
penetrate through the specimen and the X-ray film was exposed to was marginally high. However, the loosest portion of the specimen
a greater amount of radiation. is the second layer from the bottom and not the first (bottom) layer
Figure 12a illustrates the X-ray image of a coupon sectioned of the specimen because of the interaction with the rigid bottom
from a moist-tamped specimen (Dr 40%). A total of 29 vertical platen. These quantitative measurements are thus consistent with
measurement profiles spaced six pixels apart were selected for the the results of the cumulative force and peak stress analysis previ-
analysis of the gray scale variations. The average of the 29 gray ously discussed.
scale profiles is illustrated in Fig. 12b. Gray scale variations can be An air-pluviated specimen (Dr 45%) and a wet pluviated spec-
seen in each layer as well as throughout the entire specimen. Since imen (Dr 43%) were also prepared and preserved using the same
the specimen was compacted layer by layer, the gray scale varies epoxy resin impregnation procedure. X-ray images of sectioned
more near the layer boundaries than inside each layer with a sig- coupons from these specimens were captured and are illustrated in
nificant change as one moves from the top of one layer to the bot- Fig. 13a and Fig. 14a for the air- and wet-pluviated specimens, re-
tom of the overlaying layer. The darkest (loosest) part of each layer spectively. Gray scale variations of the X-ray images along the
is at the bottom of the layer. Less force is transferred to this part of specimen heights were obtained by the same method used for the
the layer since it is further from the tamping disc. The fact that moist-tamped specimen and are shown in Figs. 13b and 14b. It can
FIG. 12—Image analysis on X-ray image: (a) X-ray image of sectioned specimen; (b) gray scale analysis on X-ray image.
FROST AND PARK ON MOIST TAMPING TECHNIQUE 11
FIG. 13—Image analysis on X-ray image: (a) X-ray image of air-pluviated specimen; (b) gray scale analysis on X-ray image.
be seen that density variations exist in both the air-pluviated and This type of analysis is relatively easy to conduct. However, infor-
wet-pluviated specimens also; however, they are more gradual in mation about the actual density within the specimen cannot be di-
nature. More significant variations were noted near the top and bot- rectly obtained. To directly study variations in the density of a
tom of these specimens. The top of both of the pluviated specimens specimen, analysis of optical images is required (Kuo 1994; Jang
was denser than the other portions of the specimens. This is con- 1997; Park 1999; Frost et. al. 1999; Chen 2000). This technique can
sidered to reflect the additional force applied to the soil when the provide quantitative measures of parameters such as the local void
top cap was put in place. It is further anticipated that the extent of ratio distribution and particle orientations.
this near surface densification would be less for denser specimens. The sectioned coupon from which the X-ray image of the moist-
The bottom portion of the air-pluviated specimen is also dense be- tamped specimen was captured was cut into three smaller coupons
cause of the different pluviation mechanism at the bottom of spec- so that the surface could be subjected to a polishing and grinding
imen than at other locations within the specimen. At the beginning process to allow the sand particles to be distinguished from the
of air pluviation, sand particles impact the rigid base, whereas the surrounding epoxy matrix (Jang et. al. 1999). A total of 72 optical
energy from particles later in the process is dissipated into the soil images were obtained at the locations shown in Fig. 15a. For the
already in the mold. Wet pluviation reduces this effect because the uniformity analysis, the void ratio of each image was obtained us-
particles pluviate slowly and have a lower impact energy. Notwith- ing image analysis techniques. Figure 15b shows the density mea-
standing these various factors, this gray scale analysis of the X-ray surements from the 72 images. Each point in this figure represents
images shows that the global densities in the air- and wet-pluviated the average of the three densities measured from the three images
specimens are more uniform than in the moist-tamped specimens. at that elevation within the specimen. The average relative density
of the specimen was 40% as indicated in Fig. 15b. The solid verti-
Density Measurement Using Optical Image Analysis cal line in each layer indicates the average density for that layer.
The dashed line in each layer indicates the density trend in that
The gray scale analysis of X-ray images described above pro- layer. It can be seen that the relative density within a layer typically
vides an overview of the density variations within the specimens. increases by about 15% from bottom to top. Within the zone out-
12 GEOTECHNICAL TESTING JOURNAL
FIG. 14—Image analysis on X-ray image: (a) X-ray image of wet-pluviated specimen; (b) gray scale analysis on X-ray image.
side the bottom and top layers, the mean layer relative density in- stresses used in conventional triaxial tests. For example, peak ver-
creases by about 11% from about 34% in Layer 2 to about 45% in tical stresses of 95 and 184 kPa were measured in this study for
Layer 6, indicating the effect of the 3% undercompaction ratio used specimens with target relative densities of 50 and 75%, respec-
in preparing this specimen. Local variations in relative density of tively. These stresses applied during specimen preparation should
up to 15% are noted in the top and bottom layers, reflecting the in- be carefully considered in interpreting the results of shear tests per-
fluence of the boundary constraints. The standard deviation of the formed on specimens under initial effective confining stresses less
relative density within this moist- tamped specimen was found to than these values. Ignoring the fact that higher stress conditions are
be 6%. For a dense moist-tamped specimen with a relative density applied during the specimen preparation stage than during subse-
of 75%, the standard deviation was also found to be about 6%. A quent shear testing can lead to material behavior reflecting the
comparison using the same method of analysis showed that the cor- effects of prior stress history being incorrectly interpreted as the
responding standard deviations within air-pluviated specimens behavior of material under a virgin stress state.
with mean relative densities of 50 and 75% were about 3%, indi- The measurements made in this study have shown that the bot-
cating consistently less variable conditions. tom layer experiences larger forces during compaction than the
second layer due to the presence of the rigid base platen. The sec-
Conclusions ond layer in moist-tamped specimens was quantitatively shown to
be the loosest layer based on analysis of both the X-ray film and the
The critical assessment of the moist tamping method described optical images. Quantitative analysis of the X-ray images provided
herein has provided some unique insight into this method of prepa- information about the global density variation of specimens. Wet-
ration and raises significant concerns about the suitability of the pluviated and air-pluviated specimens are more uniform than
method for future use. moist-tamped specimens.
The vertical stresses applied during preparation of moist-tamped Quantitative measures of relative density were obtained using
specimens are sometimes higher than the typical initial confining analysis of optical images. Intra-layer relative density variations
FROST AND PARK ON MOIST TAMPING TECHNIQUE 13
FIG. 15—Image analysis on sectioned specimen: (a) positions of images taken; (b) result of image analysis on the sectioned specimen.
of up to 15% were noted while inter-layer mean relative density Frost, J. D., Jang, D. J., Park, J. Y., and Chen, C. C., 1999, “Quan-
variations of about 10% were measured. The intra-layer variations titative Characterization of Microstructure Evolution,” Proceed-
are believed to reflect the inherent limitations in the method, ings of International Workshop on Physics and Mechanics of
whereas the inter-layer variations reflect the specific under-com- Soil Liquefaction, pp. 169–177.
paction ratio selected and thus can be reduced by selection of a Gilbert, P. A. and Marcuson III, W. F., 1988, “Density Variation in
different ratio. The standard deviation of local relative density in Specimens Subjected to Cyclic and Monotonic Loading,” Jour-
air-pluviated specimens was about half that of moist-tamped nal of Geotechnical Engineering Division, ASCE, Vol. 114, No.
specimens. 1, pp. 1–20.
Jang, D. J., 1997, “Quantification of Sand Structure and its Evolu-
Acknowledgment tion during Shearing Using Image Analysis,” partial fulfillment
of the requirements for the degree of Doctor of Philosophy,
The research described in this paper was supported by NSF con- Georgia Institute of Technology.
tract number CMS 9457549. This support is gratefully acknowl- Jang, D. J., Frost, J. D., and Park, J. Y., 1999, “Preparation
edged. of Epoxy Impregnated Sand Coupons for Image Analysis,”
ASTM, Geotechnical Testing Journal, GTJODJ, Vol. 22, No. 2,
References pp. 147–158.
Chen, C. C., 2000, “Shear Induced Evolution of Structure in Water Kuo, C. Y., 1994, “Quantifying the Fabric of Granular Materials—
Deposited Specimens,” partial fulfillment of the requirements Image Analysis Approach,” partial fulfillment of the require-
for the degree of Doctor of Philosophy, Georgia Institute of ments for the degree of Doctor of Philosophy, Georgia Institute
Technology. of Technology.
Frost, J. D., 1989, “Studies on the Monotonic and Cyclic Behavior Ladd, R. S., 1974, “Specimen Preparation and Liquefaction of
of Sands,” partial fulfillment of the requirements for the degree Sands,” Journal of Geotechnical Engineering Division, ASCE,
of Doctor of Philosophy, Purdue University. Vol. 100, No. 10, pp. 118–184.
14 GEOTECHNICAL TESTING JOURNAL
Ladd, R. S., 1978, “Preparing Specimens Using Undercom- lanandan, K., 1977, “Effects of Sample Preparation on Sand Liq-
paction,” ASTM, Geotechnical Testing Journal, GTJODJ, Vol. uefaction,” Journal of Geotechnical Engineering Division,
1, No. 1, pp. 16–23. ASCE, Vol. 103, No. GT2, pp. 91–108.
Mulilis, J. P., Chan, C. K., and Seed, H. B., 1975, “The Effects of Park, J. Y., 1999, “A Critical Assessment of Moist Tamping and Its
Method of Sample Preparation on the Cyclic Stress Strain Be- Effect on the Initial and Evolving Structure of Dilatant Triaxial
havior of Sands,” EERC Report 75-18. Specimens,” partial fulfillment of the requirements for the de-
Mulilis, J. P., Seed, H. B., Chan, C. K., Mitchell J. K., and Aru- gree of Doctor of Philosophy, Georgia Institute of Technology.