Construction of geogrid encased stone columns
summer project pal|
Active In SP
Joined: Jan 2011
05-02-2011, 07:51 AM
ajmal seminar.doc (Size: 3.48 MB / Downloads: 148)
Conventional vibrated stone columns are typically used to improve the engineering properties of soft soils for the support of lightly and moderately loaded structures such as motorway embankments and large diameter storage tanks. The technique generally comprises replacing soft soil with columns of crushed rock gravel that act to stiffen the soil mass, increasing stability and reducing the time rate of consolidation. Although generally considered to be suited to soils with undrained shear strengths (su) above 15 kPa, columns have been used in soils with su as low as 6 kPa. However, below this strength, slumping of the column (where conical shaped columns consistent with the friction angle of the stone aggregate are formed) may occur during installation or the column may undergo excessive radial expansion during loading.
Geotextile encased columns (GECs) have recently been used to extend column use to very soft and extremely soft soils. The technique has been adopted successfully in Europe and more recently in South America. GECs comprise a fine granular aggregate encased with a seamless geotextile sock. The sock provides the required radial confinement in very soft and extremely soft soils.
Although design and implementation of GECs is well established, its performance may be limited in some conditions by the high strains and relative fragility of the geotextile. Alexiew et al.(2005) reports that circumferential strains in the order of 1–4% are generally required to mobilise hoop forces in the geotextile, which may result in significant radial expansion and therefore settlement during loading (which is often compensated for during the construction period). Coarse aggregates like crushed rock may cause damage to the geotextile. Furthermore, the column generally receives little compaction during installation to limit the damage caused to the geotextile by vibration.
To provide a more robust and perhaps stiffer alternative to geotextile and to broaden the appeal of geosynthetics in stone column ground improvement, the use of geogrid encasement has recently been investigated.Trunk et al. (2004) reported on unconfined compression testing of stone aggregate columns measuring 1.88 m in height and 0.6 m in diameter. The columns were encased with geogrid sleeves constructed by welding geogrid into a cylinder. The geogrid encasement was constructed by rolling geogrid into a cylinder and mechanically welding a narrow section of overlap (about 30 mm wide) to form the encasement sleeve. A photograph of a section of welded encasement is presented in Fig.1. The encasement was then filled with crushed rock aggregate, compacted and loaded in unconfined compression. Failure generally comprised column buckling (most likely associated with the unconfined conditions), with little damage caused to the welded encasement. Pressures of about 1400 kPa were measured prior to failure, with hoop strains of about 2% measured in the horizontal ribs. The work was accompanied by successful site installation trials of large-scale geogrid encased stone columns constructed in a similar manner.
Although the technique of welding provides an effective method for constructing geogrid encasement sleeves, it is unlikely to be used in practice. The technique requires either the shipment of a large welding frame to site or prefabrication of geogrid encasement sleeves away from site. Both methods are unlikely to be cost effective. On this basis, an alternative method of encasement construction was investigated. The impact of different geogrid properties and column aggregates on encased column behaviour was also investigated.
Fig. 1 . Welded geogrid encasement
2. METHOD OF OVERLAP
The ‘‘method of overlap’’ was considered likely to provide an efficient and cost-effective alternative to welding. Using this technique, an encasement sleeve was constructed from geogrid with a nominal amount of circumferential overlap. Once the sleeve was filled with aggregate and loaded, it was anticipated that interlock from the aggregate protruding between the geogrid apertures and the overlapped section of geogrid may provide a level of encasement fixity similar to welding.
A program of testing was used to investigate the performance of encased columns constructed using the method of overlap. This comprised small-scale testing of model columns encased with mesh (simulating geogrid in small-scale) and installed in very soft clay. This testing was then supplemented by unconfined testing of medium-scale columns constructed from typical stone column aggregate and encased with geogrid. The medium-scale testing was also used to investigate the impact of different stone aggregates and geogrid properties on encased column behaviour.
3. SMALL-SCALE LABORATORY TESTS
The authors reported on a series of small-scale tests undertaken to investigate several aspects of encased column behaviour (Gniel and Bouazza, 2009). These included investigating the impact of encasing only the upper section of a column compared to full length encasement and the difference between isolated and group column behaviour. In addition to this testing, small-scale tests have recently been undertaken to investigate the method of overlap used for encasement construction. A summary of the previous testing reported by the authors is set out below, providing a brief background to the new scope of testing.
3.1. SUMMARY OF PREVIOUS TESTS
The small-scale testing comprised staged loading of simulated group and isolated columns measuring about 51 mm in diameter and 310 mm in height. The columns were installed in consolidation cells measuring about 150 mm in diameter and containing very soft clay with su 5 kPa. Columns were installed using a replacement technique, where a clay cavity was excavated and replaced with a frozen sand column that was thawed prior to loading. The dimensions of the columns and properties of the sand were considered to provide a reasonable small-scale representation of a typical stone column. Geogrid was simulated using fibreglass and aluminium mesh, with geometric properties including aperture shape and size scaled down from full-size. Encasement welding was simulated by bonding a narrow section of overlapped mesh with an epoxy-resin adhesive. Where mesh encasement was used in testing, it was placed within a mould and filled with sand prior to freezing. For group column tests, the clay and column were loaded, creating unit-cell boundary conditions that simulated group column loading. For isolated column tests, only the column was loaded using a small footing.
3.2. SCOPE OF NEW TESTING
To incorporate the method of overlap into the scope of small scale testing, interlock between the sand and mesh was required. Both the sand and mesh were considered to provide a reasonable small-scale representation of typical stone column aggregate and geogrid, respectively. Furthermore, the angular sand particles were observed to protrude from between the mesh apertures during welded encasement tests, behaviour considered necessary to achieve interlock between the two materials. On this basis, several small-scale tests were undertaken to investigate the method of overlap as an alternative method of encasement construction.
3.3. TEST PROGRAM
Two group column tests and one isolated column test were undertaken on columns with full-length fibreglass mesh encasement constructed using the method of overlap. In each test, the clay and sand column were prepared using the same technique as for welded columns. Columns constructed with overlapped encasement were also loaded in the same manner as welded columns. A description of the three tests is provided in Table 1.
In test SS-1, the overlapped encasement sleeve was held in position by stitching it with cotton thread at approximately 30 mm intervals along the length of the sleeve. This was used to prevent the encasement from unravelling during column installation. However, the stitching appeared to introduce areas of stress concentration in the mesh during column loading. When the column was extruded, the mesh had failed, with tears propagating from stitched locations. Based on this observation, the mesh encasement used in the two subsequent tests (SS-2 and SS-3) was not stitched. The encasement did not unravel during installation provided that the column remained fully frozen during handling.
Table 1: Description of small-scale tests.
Test no. Column type Encasement type
SS-1 Group column Full-length fibreglass mesh encasement. 0.5_ (50%)
Circumferential overlap and stitched in position
SS-2 Group column Full-length fibreglass mesh encasement. 0.5_ (50%)
circumferential overlap, no stitching
SS-3 Isolated column Full-length fibreglass mesh encasement. 1_ (100%)
circumferential overlap, no stitching
3.4. GROUP COLUMN TEST RESULTS
The axial stress–strain behaviour of the two group column tests with overlapped encasement (encasement was overlapped by a half circumference) are compared to the behaviour of a welded fibreglass encased column test in Fig. 2. Results indicate a similar compressibility and therefore stiffness in all three tests prior to failure. The mesh that was overlapped without stitching (SS-2) failed at a similar cell pressure to the welded encasement test. Extrusion of this sample indicated that the mesh had failed by breakage of horizontal ribs in a diagonal plane stretching about 40 mm from the top of the column. This mode of failure was similar to the failure mode of the welded column and indicated that a similar level of fixity had been achieved using the method of overlap.
The results for the mesh that was overlapped and stitched (SS-1) are less clear. As the second load stage for this test was relatively large, it is difficult to assess at what column stress the failure occurred. Also, loading was continued after the mesh had failed to assess the effect of this failure on encased column performance. This additional compression significantly distorted the shape of the encased column. However, as described earlier, examination of the mesh following extrusion indicated that encasement failure had propagated from the stitched locations in the lower half of the column. As the cell pressure at failure was significantly less than for the other two tests, the stitching was considered to have contributed to the failure.
3.5. ISOLATED COLUMN TEST RESULTS
For the isolated column test (SS-3), the encasement was overlapped by a full circumference, as it was considered that small footing loading would result in less confinement being provided to the column when compared to unit-cell loading. The axial stress– strain relationship for the column with overlapped encasement is compared to a column with welded encasement in Fig. 3. In both tests, failure of the mesh did not occur due to the high pressures required to do so. The behaviour of both columns was similar, with a slightly lower compressibility measured for the column with overlapped encasement. This was probably a result of the section of overlap increasing the amount of mesh encasement in the column cross-section, thereby increasing column stiffness. The test also indicated that a similar level of encasement fixity to welding had been achieved using the method of overlap.
For encasement construction, the method of overlap appears to provide an effective alternative to welding, at least in small-scale. Small-scale tests indicated that the performance of columns constructed using the method of overlap was similar to welded columns. There was little difference in the behaviour of columns encased with 50% circumferential overlap and 100% circumferential overlap, although stitching may have had a detrimental effect on encasement performance. Based on these findings, medium-scale column tests were undertaken to investigate whether the method of overlap was effective with full-scale materials.
4. MEDIUM-SCALE LABORATORY TESTS
Following the positive results from small-scale testing, a program of medium-scale testing was undertaken to investigate the effectiveness of the method of overlap with full-scale materials including typical stone column aggregates and geogrid. A range of different materials was tested to investigate their effect on encased column behaviour.
Medium-scale tests were generally undertaken on columns measuring 0.86 m in height and about 0.24 m in diameter. Using methods similar to those adopted by Trunk et al. (2004), columns were loaded in unconfined compression. A constant displacement technique of loading was adopted rather than static loading due to time constraints. Several different geogrids were used to construct geogrids
Several different geogrids were used to construct geogrid encasement. Four biaxial geogrids with roughly square apertures and two uniaxial geogrids with rectangular apertures were supplied by NAUE GmbH for testing. The six geogrids comprise monolithic polyester ribs that are mechanically welded to form a grid, and are summarised in Table 2. The strength and geometric properties of the geogrids are presented in Table 3. The properties of geogrid B-4 are not included because it comprises a composite material manufactured from geogrid and geotextile. The composite material is manufactured by placing a layer of thin geotextile between the machine direction and cross-machine direction ribs (aligned perpendicularly) prior to welding. For encasement applications, the composite material derives its strength from the geogrid. The properties of the geogrid component are consistent with geogrid B-1
4.2. COLUMN AGGREGATES
Two types of crushed rock aggregate were used to test encased column behaviour, ‘‘20/50 mm rubble’’ and ‘‘14/10 mm gravel’’. The 20/50 mm rubble is considered representative of typical conventional stone column aggregate and was therefore used in the majority of stone column tests. The smaller 14/10 mm gravel was used to test whether sufficient interlock (using the method of overlap) could be achieved using smaller aggregates. Its use was therefore confined to a limited number of tests. Rounded gravels such as river gravels were not investigated as part of this study.
The 20/50 mm rubble comprised angular crushed basalt rock graded between 20 mm and 50 mm. Relative density testing was undertaken on the 20/50 mm rubble to measure the minimum and maximum dry density. For encased column testing, a relative density of between about 60% and 70% was targeted. This range is considered realistic for encased columns in stalled using a replacement technique. Shear box testing was also undertaken to measure the internal angle of friction (at a confining pressure of 250 kPa), although this was not used as part of any further analysis. The properties of the 20/50 mm rubble measured in this testing are presented in Table 4. As the 14/10 mm gravel was only used for testing the interlock in a limited number of columns, no strength or density testing of this aggregate was undertaken.
4.3. COLUMN CONSTRUCTION
Geogrid encasement was constructed by rolling a flat section of geogrid into an encasement sleeve with a diameter of about 0.24 m, although this varied slightly for different geogrids. The height of the encasement was kept constant at 0.86 m in all tests. Initially, about 50% of the column circumference was overlapped. However, testing indicated that this amount was insufficient in some cases and overlap was increased to 100% in later tests. For uniaxial and biaxial geogrids, encasement was initially constructed with horizontal ribs located on the inside of the sleeve. However, a greater capacity was measured in early tests when horizontal ribs were located on the outside of the column. Although this aspect of encasement construction was not further investigated or assessed, encasement was constructed with horizontal ribs located on the outside in later tests.
Cable ties were used to fix the encasement sleeve in position and prevent it from unravelling during column construction and initial loading. The tensile strength of the cable ties was measured to be significantly lower than the strength of the geogrid ribs and therefore provided little circumferential support to the encasement in the latter stages of loading. At this stage, interlock between the aggregate and overlapped section of the encasement was generally required to continue loading. Cable ties were used at six evenly spaced locations along the length of the encasement,corresponding to an average spacing of about 170 mm. A geogrids plate was also attached to the base of the column to retain the gravel during handling. The plate was fixed to the encasement sleeve with 8 evenly spaced cable ties placed around its perimeter.A photograph of a typical encasement sleeve is presented in Fig.4.
Fig. 4: Typical encasement sleeve used for medium-scale testing
The completed encasement sleeve was filled with aggregate in layers of about 200 mm. After the placement of each layer, the aggregate was compacted by shaking the encasement by hand for about 1 minute. Upon completion, the column was weighed to measure an approximate column density (volume was calculated using the internal diameter and height of the encasement). The adopted method of compaction generally resulted in a relative density between about 60% and 70%, which was the targeted level of compaction. The relative density of each column prepared for testing is presented in Table 5 (the relative density of columns constructed using 14/10 mm gravel was not measured). A photograph of compacted columns prepared using different types of geogrid encasement is presented in Fig. 5.
Fig. 5. Columns prepared for testing with encasement constructed from different geogrids.
4.4. COLUMN LOADING
Columns were tested using an Amsler compression test machine, located in the Civil Engineering laboratories at Monash University, Clayton. The compression test machine comprises a base plate which is raised by hydraulic piston (with a maximum travel of about 80 mm). For column testing, the sample column was placed on the base plate and forced upwards against a fixed crosshead. The distance between the base plate and cross-head, and therefore column height, was measured using string potentiometers (stringpots). A data acquisition system was used to record load and displacement at user-defined intervals. Column loading was generally undertaken at a displacement rate of about 1–2 mm/min. A rate of 0.3 mm/min was adopted in one test to investigate the effect of slower loading on column behaviour. No significant difference in behaviour was observed for this test. A photograph of a column prepared for testing in the Amsler is presented in Fig. 6.
Fig. 6. Column being loaded in unconfined compression
General purpose strain gauges were attached to horizontal ribs to measure geogrid hoop strains. Four strain gauges were generally attached to each encased column in the same plane vertically, but at evenly spaced locations along the length of the column. Where 50% overlap was adopted in testing, strain gauges were attached to the non-overlapped section of the encasement. Where 100% overlap was adopted in testing, the strain gauges were located at the beginning of the overlapped section.
4.6. TYPICAL COLUMN BEHAVIOUR
Sixteen unconfined column tests were undertaken to investigate the method of overlap and impact of material properties on encased column behaviour. The properties of the columns used in testing are summarised in Table 5.
In most column tests, loading initially caused axial compression of the stone aggregates, with relatively little circumferential strain measured during this stage. As loading continued, aggregate particles began to protrude from the geogrid apertures, indicating that the column was expanding in the radial direction. This coincided with a significant increase in measured circumferential strain, indicating that the encasement was working to confine the aggregate. This can be considered as the point where geogrid hoop strains are mobilised and this typically occurred at hoop strains of less than 0.5%, significantly less than the range of 1–4% reported by Alexiew et al. (2005) for geotextile encasement.
Axial compression and radial straining continued to occur as the column was further loaded, often punctuated by the sound of rock particles breaking and crushing. The circumferential (hoop) strains measured at different locations on the geogrid encasement are plotted against axial column strain in Fig. 7. The initial stage of the test where axial compression occurred without significant mobilisation of geogrid hoop forces is highlighted. The results presented are for test MS-8, where encasement was constructed from geogrid B-2. The results presented comprise strain measurements taken prior to the unloading stage (less than about 80mm compression) and are considered representative of the behaviour of most column tests.
Cable ties often broke in the early stages of column loading, resulting in slight slipping or unravelling of the overlapped section of geogrid (and a reduction in measured geogrid hoop strain) before the protruding rock interlocked with the geogrid, providing fixity. Prior to column failure, the sound and frequency of rock crushing intensified. Circumferential strains generally increased roughly linearly with axial strain. Failure was observed to occur in one of three ways, described below.
Fig. 7. Circumferential strain behavior for a typical column test.
4.6.1. MODE 1 – INSUFFICIENT OVERLAP, SUFFICIENT INTERLOCK
When an insufficient amount of overlap was used to construct the geogrid encasement but the interlock between the aggregate and the geogrid ribs was adequate, the geogrid junctions (generally welded) sheared in the section of overlap. Vertical ribs generallytore away from horizontal ribs or vice-versa, resulting in a loss of circumferential confinement and failure generally in the upper or mid section of the column.
4.6.2. MODE 2 – INSUFFICIENT INTERLOCK
When interlock was insufficient, the overlapped section of geogrid encasement tended to slip untilmost of the circumferential loadwas supported by the cable ties. When these failed, the geogrid encasement generally unravelled with little damage occurring to the geogrid. This mode of failure generally occurred when 14/10mm gravel was used rather than the 20/50mm rubble.
4.6.3. MODE 3 – SUFFICIENT OVERLAP, SUFFICIENT INTERLOCK
When both overlap and interlock were sufficient, tension failure occurred in the geogrid ribs that provided circumferential support to the column. This was considered the desired failure mode because it indicated that the tensile capacity of the geogrid was being used. One or two ribs generally failed simultaneously, with ribs above and below this location failing with ongoing loading. The ultimate tensile strain of the geogrid (and therefore tensile strength) was generally reduced in testing due to partial cutting of the ribs from the angular-shaped crushed rock. This aspect is discussed in more detail in later sections.
The results of unconfined medium-scale column tests are presented in Table 6. Data presented comprises the failure mode (1, 2 or 3, as set out earlier), maximum column load, average and maximum hoop strains measured in the four strain gauges and the average Young’s modulus for columns. Young’s modulus was calculated by dividing the column pressure (measured by dividing the column load by the column cross-sectional area) by the axial strain. The column Young’s modulus was averaged over the first 80 mm of axial displacement, as the load-deformation behaviour was roughly linear to this point. Beyond this point, testing often incorporated an unload-reload cycle due to limitations in the travel of the piston and base plate of the compression test machine. This made calculation of Young’s modulus and meaningful comparison between tests difficult for axial displacements greater than about 80 mm. Test results are discussed in more detail below.
4.7.1. BIAXIAL GEOGRID TESTS
Initial tests were undertaken on encasement constructed from B-1 geogrid (biaxial geogrid with the lowest stiffness and strength of those tested) with approximately 50% overlap. Three tests were undertaken adopting this configuration. For the two tests where horizontal ribs were located on the inside of the encasement (MS-1 and MS-2), failure comprised tearing of junctions in the section of overlap, indicating insufficient overlap. For the third test (MS-3), the horizontal ribs were located on the outside of the encasement. In this case, failure occurred by breakage of horizontal ribs in tension (the desired failure mode). The maximum load of 84 kN was also higher than for the first two tests, although there was no significant change in column stiffness or measured geogrid hoop strain. Based on these results, encasement in subsequent tests was constructed with horizontal ribs located on the outside of the column. Overlap was also increased to 100% to reduce the likelihood of overlap failure.
Two tests were undertaken on B-1 geogrid encased columns constructed with 100% overlap. Results indicate that the desired failure mode was achieved in both cases. Maximum column loads of about 95 kN were higher than for the 50% encased tests. However, geogrid hoop strains and Young’s modulus did not vary significantly between columns with 50% and 100% overlapped encasement. Furthermore, the average geogrid hoop strains at failure of about 3% were about half the tensile strain of 6.2%, measured by the manufacturer in laboratory uniaxial tensile tests. This may have been (at least in part) due to the strain being measured in the section of overlap for columns with 100% circumferential encasement, thereby underestimating the actual maximum strain. However, partial cutting of the horizontal geogrid ribs from angular crushed rock was observed during loading in a number of tests. A photograph of partial cutting identified in a section of geogrid from test MS-1 is presented in Fig.9. Cutting of the geogrid may have significantly reduced the strength of the ribs and may account for the reduction in measured tensile strain.
Following the five B-1 geogrid tests, two tests (MS-6 and MS-7) were undertaken on columns constructed from B-2 geogrid encasement with 100% overlap. The higher strength geogrid encasement produced encased columns with greater stiffness (an average Young’s modulus of about 18 MPa compared to 14 MPa) and a maximum load of 113 kN compared to 95 kN, as set out in Table 6. Hoop strains were also significantly higher than for B-1 geogrid encasement, with average and maximum strains of 4.0 and 5.2, respectively. This was most likely a result of the increased rib cross-sectional area (as indicated by the smaller apertures presented in Table 3) providing greater resistance to cutting from angular pieces of crushed rock.
Fig. 8. Partial cutting of horizontal ribs
Two final biaxial encasement tests (MS-8 and MS-9) were undertaken using encasement constructed with B-3 geogrid (100% overlap), the highest strength biaxial geogrid used in testing. The maximum loads and Young’s moduli presented inTable 6 are greater than measured in earlier tests and are consistent with the higher strength encasement. The average and maximum hoop strains of 4.2% and 5.6% are also slightly higher than for earlier tests, further supporting the contention that higher strength geogrids provide greater resistance to cutting. The maximum loads and hoop strains measured in columns constructed from biaxial geogrid encasement with 100% circumferential overlap are compared in Fig. 9.
Fig. 9.Maximum load and hoop strain measurements for biaxial geogrid encased columns.
4.7.2. UNIAXIAL GEOGRID TESTS
Three columns constructed from uniaxial geogrid encasement were tested in axial compression. The first test (MS-10) comprised U-1 geogrid encasement with about 60% overlap. As indicated in Table 6, failure comprised tearing of the geogrid welds in the section of overlap, indicating that the amount of overlap was insufficient. As a result, failure occurred at hoop strains averaging about 1.5%, significantly less than the tensile strength of the georid ribs and less than the average hoop strains measured for biaxial encasement. On this basis, encasement was increased to 100% for test MS-11. However, results of this test also indicated that the amount of overlap was insufficient. Although the maximum load increased from 56 kN to 99 kN, the average hoop strains were still significantly less than measured in earlier biaxial tests.
The third and final uniaxial geogrid encased column (test MS-12) was constructed from U-2 geogrid, with 100% overlap. Although the encasement comprised geogrid with about double the strength (and stiffness) of U-1 geogrid in the hoop direction, it failed at a lower load than in test MS-11. Failure generally comprised insufficient overlap, at hoop strains of about 1%. This performance when compared to the lower strength uniaxial geogrid was attributed to geogrid stiffness. As the U-2 geogrid was significantly stiffer than the U-1 geogrid, it was more difficult to roll into a sleeve and therefore more support was required to prevent it from unravelling in unconfined loading.
In all three tests, the vertical ribs in the encasement were observed to deform (buckle, twist and fold) to a greater extent than in biaxial encasement tests. The lower stiffness of the vertical ribs in the uniaxial encasement and the greater spacing of these ribs compared to biaxial geogrid may account for the observed deformation. As the strength and stiffness of vertical ribs are considered to play an important role in providing interlock between the overlap and stone aggregate, deformation of these ribs is likely to result in increased radial expansion and therefore increased axial deformation when compared to biaxial geogrids. The lower strength and increased spacing of vertical ribs in uniaxial geogrid may also account for the observed mode of failure and lower measured hoop strains in uniaxial encasement tests. On this basis, further testing may indicate that uniaxial geogrids are not suited to encasement using the method of overlap. Further testing of larger diameter columns (where more welded junction is present in the section of overlap) is recommended in confined conditions to assess the suitability of uniaxial geogrids to encasement applications.
4.7.3. GEOTEXTILE AND SMALLER AGGREGATE TESTS
Several tests were undertaken to investigate the ability of smaller aggregate to interlock with overlapped geogrid encasement. The encasement for the first test (MS-13) was constructed from B-1geogrid with about 50% overlap. As the
14/10 mm gravel could not be retained in the encasement because of the geogrid aperture size; a thin geotextile sleeve was placed inside the encasement. Gravel aggregate was then placed and compacted in a similar manner to earlier tests. However, during loading, this geotextile sleeve prevented the aggregate from adequately interlocking with the section of overlap and the encasement unravelled immediately following the failure of cable ties. A maximum load of 26 kN and average hoop strains of about 0.5% were measured at failure.
To investigate whether geotextile also prevented larger aggregates from interlocking with the geogrid, a test was undertaken using 20/50 mm rubble encased with B-4 geogrid, which comprised a geogrid/geotextile composite. Although a slightly higher load was measured, the interlock and overlap were also inadequate. On this basis, the use of geotextile was not considered compatible with the method of overlap and no further testing of geotextile was undertaken.
In place of geotextile, a thin plastic film was wrapped around the outside of two columns (constructed from B-1 geogrid encasement with 100% overlap) to investigate the impact of smaller aggregate. The plastic film comprised a single layer of ‘‘cling wrap’’, a low strength, ductile film typically used in food preparation. The film was considered to provide little confinement to the column (at least when compared to soil) but enough to retain 14/10 mm gravel in the geogrid encasement sleeve. Using encasement wrapped with the plastic film, the aggregate was observed to protrude between ribs in the section of overlap to a greater extent than in geotextile tests.
The results of tests MS-15 and MS-16, where 14/10 mm gravel columns encased with geogrid and plastic film were loaded, are similar and provide contrast to the results of equivalent 20/50 mm rubble tests. Column capacity and Young’s modulus were significantly higher than the 20/50 mm rubble tests, possibly due to the smaller aggregate behaving with greater stiffness. Even though the columns failed by insufficient interlock, strains of up to 3.5% were measured, higher than the average 3% strain measured in equivalent0/50 mm rubble tests. Little damage or cutting of the geogrid encasement was observed in 14/10 mm gravel tests, unlike 20/50 mm rubble tests. Based on these results, it was concluded that although the smaller aggregate did not interlock as well with the geogrid, it did not damage the geogrid ribs like the coarser (sharper) aggregates. Further testing is recommended to investigate the performance of smaller aggregates in confined conditions.
Based on the results of small-scale and medium-scale testing, the method of overlap appears to provide an effective technique of encasement construction, providing a level of fixity similar to welding. The results of this testing, particularly the medium-scale study where real geogrid and typical stone column aggregates were used, are discussed below. Discussion focuses on practical aspects of encasement construction and performance, with consideration also given to further testing and potential full-scale site trials.
5.1. BIAXIAL GEOGRIDS
Biaxial geogrids appear best suited to the method of overlap, partly due to the higher stiffness and closer spacing of vertical ribs that facilitate interlock in the section of overlap. Based on testing of medium-scale columns, 50% overlap was not sufficient in some cases. This is unlikely to be the case for full-scale (larger diameter) columns, where a greater number of vertical ribs and therefore more junctions will be present in the section of overlap. Confinement from surrounding soil is also likely to benefit the performance of overlapped encasement. Nevertheless, 100% overlap is recommended for further testing to prevent overlap failure. Investigation into the minimum number of junctions needed in the section of overlap to achieve adequate encasement fixity is also required.
Testing indicates that column stiffness increases with geogrid strength (and stiffness). For encasement constructed from B-1 geogrid, the average column Young’s modulus was about 14 MPa. This increased to about 20 MPa for encasement constructed from B-3 geogrid. This range compared favourably to the 15 MPa to 18 MPa range reported by Trunk et al. (2004), where similar aggregates and geogrid strengths (at least in the hoop direction) were used in welded encasement tests.
Higher strength geogrids also appear to be more robust, most likely due to the greater cross-sectional area of geogrid ribs providing greater resistance to cutting. Higher strength geogrid (like the B-2 or B-3 geogrid used in this study) is therefore recommended for further testing, particularly when used in combination with coarser crushed rock aggregates. In medium-scale tests, a 35% reduction in the strength of B-3 geogrid was measured when compared to the strength measured by the manufacturer in the laboratory. Further testing is therefore required to measure the strength reduction associated with different aggregates and to assess a suitable factor of safety.
5.2. UNIAXIAL GEOGRIDS AND COMPOSITE GEOGRIDS
The vertical ribs of uniaxial geogrids were observed to deform more significantly than in biaxial encasement tests, probably due to their lower stiffness and greater spacing. This resulted in the encasement overlap slipping and tearing, even when 100% circumferential overlap was adopted. As testing of columns with greater than 100% overlap was not considered practical, overlapped uniaxial geogrid encasement is probably not suited to unconfined testing. To further investigate the suitability of uniaxial geogrid to the method of overlap, confined testing is recommended, preferably with large diameter columns where more junctions will be present in the section of overlap. As uniaxial geogrids are generally of higher strength in one direction than biaxial geogrids, successful testing may provide a high strength, high stiffness option for encasement construction.
Geotextile and geogrid composites, like geogrid B-4, are not suited to the method of overlap because the stone aggregate cannot interlock sufficiently with the geogrid ribs.
The geogrids and stone aggregates adopted in medium-scale testing are considered representative of the types of material to be used for encased column construction. Testing indicates that the coarser 20/50 mm rubble provides better interlock with geogrid encasement and is probably best suited to column construction using the method of overlap. However, the coarse aggregate was observed to cut the geogrid in several instances and therefore its use with more robust geogrids like geogrid B-3 is recommended. Careful consideration would also need to be given to the selection of an appropriate strength reduction factor. In contrast, smaller aggregates such as the 14/10 mm gravel appear to cause little damage to the geogrid but do not achieve the same level of interlock as coarser aggregates, at least in unconfined conditions. Confined testing of smaller aggregates is recommended to investigate whether interlock can be increased through confinement. In this event, smaller aggregates may be better suited to the method of overlap.
5.4. PRACTICAL APPLICATIONS
As cable ties were shown to be capable of supporting columns during handling and loading, they are likely to provide a simple method for encasement construction on site. The cable ties used in medium-scale testing were required to support the column during handling and in the early stages of unconfined compression. For site construction, soil will provide confinement during the early stages of loading and therefore cable ties are only required during handling. On this basis, it is envisaged that cable ties used at a spacing of about 0.5 m along the length of the column will be adequate. Encasement may be constructed by rolling flat sheets of geogrid into cylinders with the required diameter and overlap. Cable ties may then be used to fix the encasement in position.
Geogrid encasement is likely to be adapted to existing techniques used to install conventional stone columns. A possible replacement technique is illustrated in Fig. 10, where an encasement sleeve would need to be constructed with a slightly smaller diameter than the internal diameter of the casing. Displacement installation may comprise wrapping the encasement sleeve around the outside of the vibroflot, as adopted by Trunk et al. (2004) in field trials, or using a system similar to the method adopted for displacement installation of GECs (Alexiew et al., 2005).
The technique of geogrid encasement is likely to extend the use of stone columns to very soft and extremely soft soils, in the same way GECs have been used in the past. The technique may provide a stiffer alternative to GECs, as the geogrid is typically more robust and can therefore be used with coarser aggregates and receive greater compaction during installation. The technique could also be adopted on conventional stone column project and implimentations to improve the performance of the foundation and possibly reduce the number of stone columns required, depending on settlement criteria. Site installation and full-scale load trials are required for final verification of the technique.
Fig. 11. Proposed method of replacement installation
The research presented in this paper was used to investigate an efficient and effective method of geogrid encasement construction for use with stone column ground improvement. Small-scale and medium-scale tests were undertaken. Based on the findings of this research, the following conclusions are made.
• The method of overlap, where interlock between stone aggregate and an overlapped section of geogrid encasement is used to prevent the encasement from unravelling, provides a level of fixity similar to welding.
• The method of overlap appears likely to provide a simple method of encasement construction that is suited to use on site and provides an effective alternative to welding. During installation, encasement may be temporarily fixed in position using cable ties, probably at spacing of about 0.5 m along the length of the encasement (although this would need to be refined in testing).
• Biaxial geogrids are best suited to encased column construction, with higher strength geogrids like the B-3 geogrid providing the stiffest column response and greatest robustness of the different encasement materials tested.
• To achieve a sufficient level of fixity, columns should generally be constructed with 100% circumferential overlap. Further research is recommended to refine this and to determine the minimum number of junctions required in the section of overlap.
• 20/50 mm rubble, considered typical of conventional stone column aggregate, provides the greatest level of interlock but may reduce geogrid strength through cutting. Higher strength geogrids provide greater resistance to cutting.
• The smaller (and less sharp) 14/10 mm gravel reduced the damage caused to the geogrid but did not provide the same level of interlock. Further confined testing of columns is recommended to investigate whether smaller aggregates are better suited to the method of overlap.
• Uniaxial encasement constructed with 100% overlap was not sufficient in medium-scale tests due to the lower stiffness and greater spacing of vertical ribs, combined with the relatively small column diameters. Confined testing of uniaxial geogrids is required to determine their suitability to the method of overlap.
• Geotextile encasement and geotextile/geogrid composites are not suited to the method of overlap.
1. Gniel, J., Bouazza, A., 2009. Improvement of soft soils using geogrid encased stone columns. Geotextiles and Geomembranes 27 (3), 167–175.
2. Construction of geogrid encased stone columns: A new proposal based on laboratory testingJoel Gniel a,1, Abdelmalek Bouazza b,*