Long-term filtration performance of nonwoven geotextile-sludge systems

Transcription

1 Geosynthetics International,,, No. Long-term filtration performance of nonwoven geotextile-sludge systems A. H. Aydilek and T. B. Edil Assistant Professor, Department of Civil and Environmental Engineering, University of Maryland, 6 Glenn Martin Hall, College Park, Maryland 7, USA, Telephone: + 69, Telefax: + 8, aydilek@eng.umd.edu Professor & Chair Geological Engineering Program, and Professor, Department of Civil and Environmental Engineering, University of Wisconsin-Madison, Madison, Wisconsin, 76, USA, Telephone: , Telefax: , edil@engr.wisc.edu Received January, revised May, accepted June ABSTRACT: A study of the filtration behaviour of contaminated wastewater treatment sludges was conducted. The laboratory portion of the study included a series of filtration tests with different nonwoven geotextiles. Filtration performance of the sludge nonwoven-geotextile systems was also observed in field test cells by exhuming geotextile samples from the cells after exposure followed by analysis. The results indicated that the sludges could be filtered with nonwoven geotextiles selected on the basis of geotextile permittivity. The standard gradient ratio test did not always reflect the filtration performance, and therefore other clogging ratios should be considered. Two commonly used geotextile constriction sizes, O 9 and O, could not be related to either clogging or piping. Additionally, ratios of geotextile constriction size to soil particle size in the existing filter criteria did not always predict the observed filtration performance. KEYWORDS: Geosynthetics, Capping, Filtration, Gradient ratio test, Image analysis, Nonwoven geotextile, Pore constriction size, Sludge REFERENCE: Aydilek, A. H. & Edil, T. B. (). Long-term filtration performance of nonwoven geotextile-sludge systems. Geosynthetics International,, No.,. INTRODUCTION The retirement of large industrial waste storage facilities in accordance with environmental regulations has become a critical cost issue for industry and a challenge to the geotechnical community. Many facilities were constructed prior to the emergence of modern environmental regulations and contain a variety of contaminated high water content materials. Some examples are PCB (polychlorinated biphenyls) containing wastewater treatment sludges, contaminated harbour dredgings, waste pickle liquor sludges, asbestos-containing sediments, and contaminated river bottom sediments. These materials are typically contained in surface impoundments such as lagoons, ponds or old quarries. One of the least costly remediation alternatives is capping (Grefe 989; Zeman 99). Among various types of capping options, composite caps are usually preferred, in which a geotextile component serves as a reinforcement, separation, and filtration layer. The emphasis in this paper is on the filtration performance of nonwoven geotextiles in such a composite cap. This study was undertaken as part of the development of a closure system for the PCBcontaminated wastewater treatment sludges in Madison, Wisconsin. For the evaluation of the filtration performance, both clogging and retention behaviour of these filters had to be investigated. The low water solubility of PCBs in contaminated sludges and their partition on the solid phase requires that the retention performance of these filters be analysed considering more stringent piping rate limits than the existing criteria. On the other hand, high organic content in sludge promotes clogging, which requires an assessment of the applicability of existing anti-clogging criteria to sludge. To respond to this need, laboratory soil filtration tests (gradient ratio and filter press tests) and geotextile permittivity tests were conducted with various sludge geotextile systems. The same tests were also performed on a reference silty sand having the same particle size distribution as the sludge. The long-term filtration performance of sludge geotextile systems was also observed in four field test cells capped using a lightweight fill. Permittivity tests and image analyses were performed to quantify the degree of clogging observed in the laboratory and in the field. Finally, comparisons were made between the laboratory and field results. This paper presents the results of analyses performed using nonwoven geotextiles; woven geotextiles were previously presented by Aydilek and Edil () # Thomas Telford Ltd

2 Long-term filtration performance of nonwoven geotextile-sludge systems. LABORATORY TESTS.. Materials... Sludge and soil The physical and chemical properties of the PCBcontaminated wastewater treatment sludge are listed in Table. The sludge is classified as organic silty sand (SM) according to the Unified Soil Classification System. Attempts were made to determine the liquid and plastic limits of the sludge samples; however, such measurements were not possible owing to its non-plastic nature. The mean specific gravity of the sludge was.8. The reference soil used in this study was a silty sand and was prepared as a mixture of Portage sand and Madison silt to approximate the particle size distribution of the sludge. The mean specific gravity of the soil was.67, and it lacked any organic matter.... s In the laboratory testing programme, six nonwoven geotextiles were used. Field test cells were constructed using four of these geotextiles. The geotextiles were selected from the ones most often used in filter applications and had a wide range of apparent opening size (AOS or O 9 ), and permittivity. The physical and Table. Properties of wastewater treatment plant sludge used in the study Property Value Water content (%) 7 Solids content by weight (%) 8 6 Specific gravity, G s.8 Organic content (%) Atterberg limits (%) Non-plastic Fines content (particles <.7 mm) (%) D (mm).8 D (mm). D 6 (mm). Coefficient of uniformity, C u Coefficient of curvature, C c PCB content (mg/kg) to 9 In-situ undrained shear strength (kpa)..6 In-situ hydraulic conductivity (m/s) Notes: Determined from plate load tests. Determined from two-stage borehole permeameters. hydraulic properties of the geotextiles are given in Table... Methods Three different types of soil geotextile filtration test were conducted in this study. Among the various test methods, two methods standardised by the American Society of Testing and Materials (ASTM) were used: the gradient ratio test (ASTM D ) and permittivity test (ASTM D 9). A third test, a modified version of the American Petroleum Institute (API) filter press test (American Petroleum Institute 98), was also used. Gradient ratio tests were conducted to determine the clogging performance of soil geotextile combinations. Contrary to the h procedure prescribed in ASTM D, the tests were continued for more than months to understand the long-term clogging performance of these systems. Hydraulic gradients of,., and 7. were used in the tests. Methods described by Aydilek and Edil () were followed for the specimen preparation. In all of the tests, a fully automated water de-airing system continuously supplied the test water. The dissolved oxygen content of the water was regularly checked and maintained at between. and mg/l, less than the limit of 6 mg/l set by ASTM D. In order to quantify the change in the hydraulic performance of geotextiles after subjecting them to filtration with sludge or silty sand in the gradient ratio test or in the field, laboratory permittivity tests were conducted on the geotextiles following the procedure described in ASTM D 9. The API filter press was used to determine the retention characteristics of the sludge geotextile systems. A pressure of 7 kpa was applied during the test to simulate the expected lower field gradients, and the tests were continued for h to observe the long-term retention performance. The procedure described by Aydilek and Edil () was followed for the filter press tests. It is well known that the filtration performance of a soil geotextile system is directly related to its constriction size distribution (CSD), i.e. distribution of the minimum opening sizes of the flow channels in a geotextile. A probabilistic model coupled with image analysis, named CONS, was developed to determine the constriction sizes of the geotextiles. Thin sections of epoxy-resin-impregnated geotextiles were prepared, and Table. Physical and hydraulic properties of nonwoven geotextiles used in the study Structure, polymer type Mass/unit area (g/m ) Thickness (mm) Apparent opening size, AOS (mm) O 9 constriction size Porosity (%) Permittivity (s ) N NW, NP, HB, PP 6..8 NA L NW, NP, STF, PP P NW, NP, STF, PP I NW, NP, STF, PP M NW, NP, CF, PP K NW, NP, CF, PP Notes: NW, nonwoven; NP, Needle-punched; STF, staple fibre; CF, continuous filament; HB, heat-bonded; PP, polypropylene; NA, Not analysed. Mass/unit area and thickness are the manufacturer s reported values. Permittivities were measured in the laboratory per ASTM D 9, and porosity values were determined using the method described by Wayne and Koerner (99). The lower bound of AOS values are reported according to personal communication with the manufacturers. O 9 constriction sizes were determined using image-based code CONS.

3 Aydilek and Edil image processing operations and probabilistic analyses were performed on the captured images of these thin sections. Aydilek et al. (, ) provides detailed description of the method. Analyses were conducted on virgin geotextiles as well as on the ones subjected to filtration in the laboratory and in the field. Owing to current limitations, the methodology is applicable only to needle-punched nonwoven geotextiles; therefore the pore structure of N (heat-bonded nonwoven) was not analysed.. FIELD STUDY: CONSTRUCTION OF CELLS AND FIELD INVESTIGATIONS A field study was conducted to investigate the filtration performance of sludge geotextile systems. Four sludge lagoon test cells were capped using a pervious lightweight fill consisting of a mixture of wood chip and soil separated from the sludge by four different nonwoven geotextiles, referred to as s I, L, P, and N. The thickness of the sludge and the fill ranged from. m to. m and from. m to m, respectively. The physical and hydraulic properties of the geotextiles used in the field study are given in Table. The cells were instrumented with piezometers, settlement plates and surface survey blocks. Piezometers installed at different depths in the sludge provided information about the pore water heads and, therefore, a measure of the clogging performance. Figure shows the layout of the test cells. Samples were exhumed at months after the construction of the test cells. Sludge and geotextile samples were collected and transported to the laboratory for further analysis. Samples of the cap material were also collected to investigate the intrusion of sludge solids into the cap due to consolidation of the sludge under the weight of the cap (i.e. separation performance of geotextiles). Laboratory tests (i.e. permittivity tests, image analyses, filter press tests) were performed on the exhumed geotextile samples to assess their clogging performance in the field application. Aydilek () and Aydilek and Edil () provide detailed information about the construction of the test cells, field exhumation procedure, and methodology employed in the laboratory tests. Pore water heads and the surface water regime in the field test cells were monitored during and after cap construction. Excess pore water heads, which occurred because of loading of the sludge by the cap, dissipated approximately within years after the construction, mostly by upward flow, and followed the trend observed in the settlement data for all of the cells. RESULTS AND ANALYSIS.. Clogging behaviour of geotextiles... Specimens exposed to filtration in the laboratory For the analysis of gradient ratio test results, two different ratios were used: gradient ratio (GR) and permeability ratio (K R ). ASTM D defines GR as the ratio of hydraulic gradient in the contact zone to hydraulic gradient in the soil, whereas K R is the ratio of the stabilised hydraulic conductivity of the soil to that of the stabilised system hydraulic conductivity: GR ¼ i soilngeoxtile interface i soil K R ¼ k soil k system ðþ ðþ NOT TO SCALE N m N Surface survey block. m.6 m.6 m Figure. Layout of field test cells L P I 8 m Settlement plate Sample piezometer configuration Locations of the piezometer tips Bottom of sludge layer Mid-depth of sludge layer Sludge _ geotextile interface Above geotextile (to determine groundwater fluctuations)

4 Long-term filtration performance of nonwoven geotextile-sludge systems where i soil geotextile interface is the hydraulic gradient in the contact zone and i soil is the hydraulic gradient in the soil. k soil and k system are the hydraulic conductivities in the soil and the entire system respectively. The hydraulic conductivity of the entire system, k system, is determined using the applied hydraulic gradient on the soil geotextile system (i.e.,.,, 7.). For k soil calculations, i soil values were calculated using the readings registered by manometers located mm and 7 mm from the top of the middle section of the permeameter. For both of the hydraulic conductivities (i.e. k soil and k system ) stabilised flow rates were used (determined by taking the average of the last five stabilised values for each test). Another ratio, called the permittivity ratio (C R ), was defined as the ratio of the geotextile permittivity after the gradient ratio test (i.e. after filtration) to the virgin geotextile permittivity. This ratio should be equal to for an unclogged geotextile, indicating no change in flow capacity of the geotextile. Considering the criticality of the applications and other factors (e.g. chemical and biological clogging), a flow reduction of up to % (this corresponds to a permittivity ratio of.8) is allowed when using filters for sludges. As explained in Aydilek and Edil (), K R ¼ andc R ¼.8 are set as the limits for acceptable clogging of sludge geotextile systems. The values of GR, K R and C R for the sludge geotextile and silty sand geotextile systems exposed to filtration in the laboratory study are given in Table. A review of the data in Table shows that all of the geotextiles tested with sludge would be considered clogged based on the criterion that sets GR of as the limit, and only one of them ( I) would be considered unclogged when the US Army Corps of Engineers limit of is used (Haliburton and Wood 98). However, analysis of the K R ratios, which are based on the measured hydraulic conductivities at different locations in the soil, does not support these conclusions. Under the initial hydraulic gradient applied, some of the K R values were slightly above the suggested limit of (not shown in the table) (Aydilek ). As the hydraulic gradient was increased, the ratios decreased, possibly as a result of relocation and redeposition of the sludge particles. Additionally, the C R ratios show that the flow capacity of the geotextiles remained unchanged in most cases during laboratory filtration, and are comparable with the K R ratios. There are various problems associated with the testing of sludges in the standard gradient ratio test, such as clogging of manometer ports and possible biological and chemical clogging due to organic matter and contaminants. These problems, coupled with the results found in this study, suggest that K R and C R rather than GR appear to be more consistent indicators of clogging behaviour of sludge geotextile systems. s (GR and K R ) for sludge and silty sand are plotted against virgin geotextile permittivity in Figure. K R decreases with increasing permittivity for both sludge and silty sand, showing the effect of geotextile permittivity on clogging behaviour. The change in GR is small for permittivity values greater than. s, and the ratios stabilise at about C ¼. s. All the geotextiles, except N, indicated satisfactory performance based on their K R ratios. Furthermore, most C R values are between.8 and. (Table ), indicating that the flow capacity of the geotextiles remained practically unchanged. The exception to that is N, which is a heatbonded nonwoven geotextile and is occasionally not successful in filtration applications (Haliburton and Wood 98). The results indicated that permittivity is a good indicator of geotextile filtration performance with sludge. For instance, a permittivity ratio of.79 was measured for L (thick nonwoven needlepunched geotextile), which is below the limit of.8 set as the permittivity ratio criterion for clogging (Aydilek and Edil ), i.e. the minimum permittivity ratio allowed. This finding is consistent with the K R of this geotextile, which is., indicating that the geotextile is close to clogging based on a maximum allowable K R (Aydilek and Edil ). These trends are consistent with the findings of Faure et al. () and Krug et al. (). Both studies indicated that permittivity is the main pore structure parameter affecting the clogging performance of nonwoven geotextiles. Table. Summary of laboratory test results Sludge Inorganic silty sand Virgin geotextile permittivity (s ) Results from gradient ratio tests Stabilised gradient ratio, GR Permeability ratio, K R Results from permittivity tests Permittivity ratio, C R Results from gradient ratio tests Stabilised gradient ratio, GR Permeability ratio, K R Results from permittivity tests Permittivity ratio, C R N L P I M K Note: Reported clogging values are the stabilised values at the end of testing.

5 Aydilek and Edil Virgin geotextile permittivity (s! ) Gradient ratio Virgin geotextile permittivity (s! ) Change in O after filtration tests (%) Change in O 9 after filtration tests (%)!!!!!!!!!! Tested with sludge Tested with silty sand Virgin geotextile permittivity (s! ) Tested with sludge Tested with silty sand Virgin geotextile permittivity (s! ) Figure. Relationship of permittivity to clogging performance of: sludge; silty sand The geotextiles subjected to gradient ratio tests with sludge were analysed to determine the degree of change in their CSDs using the probabilistic model CONS. The exception to that was N. Its fibre structure caused problems in image processing, and therefore it was eliminated from the image analysis programme. Figure shows the changes in two characteristic constriction sizes, O 9 and O, after the laboratory tests conducted with sludge and silty sand. The decrease in the O 9 constriction sizes is less than % in all cases. However, the geotextiles experienced greater changes in their median pore sizes, O, a decrease in the range 8 %. This is in agreement with the findings of the previous research indicating that smaller constriction sizes (i.e. O to O ) may be more prone to changes during filtration (Fischer 99; Fischer et al. 99; Millar et al. 98). Figures and show the planar and crosssectional view images of two geotextiles with two different permittivities after testing with sludge in the gradient ratio test (s L and K, respectively). For comparison purposes, the images of virgin specimens of those geotextiles are also given in the same figures. The geotextile with the larger permittivity ( K in this case) clogged less, and therefore experienced smaller amounts of decrease in its constriction size. Generally, Figure. Decrease in: O 9 pore opening sizes; O pore opening sizes after laboratory gradient ratio tests this was the case when it was tested either with silty sand or with sludge. In all cases, the decrease in the constriction sizes was less for geotextiles tested with silty sand. The CSD curves of most geotextiles shifted to the left after the gradient ratio tests (Aydilek ), indicating that clogging affected not only a single characteristic constriction size (i.e. O 9 or O ) but rather the whole pore structure. The change in CSD was greater for the geotextiles having lower permittivity values. Figure 6 describes such a change in CSD for s L and K with very different permittivities.... Laboratory tests on specimens exposed to filtration in the field Figure 7 shows micrographs of the exhumed sample of L. The image of a virgin sample of the same geotextile is also given in Figure 7. The sludge particles were attached onto the fibres, possibly because of the biological activity. Figure 7c shows that some of the fibres were reoriented, probably because of the stresses imposed by the cap and the construction equipment. The permittivity of the exhumed geotextiles was determined using the methods described in ASTM D 9. Four permittivity tests were conducted on each sample, and permittivity ratios (C R ) were calculated based on the

6 Long-term filtration performance of nonwoven geotextile-sludge systems Sludge particles Sludge particles (c) (d) Figure. L (C ¼.7 s ). Planar and cross-sectional images of virgin geotextile; (c) planar and (d) cross-sectional images of post-gradient ratio test specimens Sludge particles Sludge particles attached to the fibres (c) (d) Figure. K (C ¼. s ). Planar and cross-sectional images of virgin geotextile; (c) planar and (d) cross-sectional images of post-gradient ratio test specimens

7 6 Aydilek and Edil Percent finer by number (%) Constriction diameter (mm) Virgin geotextile Post-gradient ratio test Percent finer by number (%) 8 6 Virgin geotextile Post-gradient ratio test... Constriction diameter (mm) Figure 6. Changes in constriction size distribution (CSD) of: L (C ¼.7 s ); K (C ¼. s ) Figure 7. L: cross-sectional images of: virgin geotextile;, (c) exhumed specimens (magnification ¼ ) (c) mean value. Table summarises the changes in permittivity ratios for each geotextile, exhumed year after construction. The table shows that permittivity ratios are usually within the limits.8. for the field specimens, with an exception in N. For the exhumed geotextiles, the decrease in the O 9 constriction size ranged from 8.% to %; however, somewhat greater changes in the median constriction size, O, were observed, as shown in Table (i.e. as large as %). This trend is similar to the observations made in the laboratory tests that smaller pores clog more. In general, geotextiles with relatively higher permittivities experience less clogging. Smaller geotextile constriction sizes (i.e. O ) can be related to permittivities as they are the controlling sizes for clogging. As a result, smaller changes in O values were observed for the geotextiles with relatively higher permittivities.... Comparison of laboratory and field results Both field and laboratory results indicate that the geotextiles, in general, were not clogged when tested with the sludge. The exception to that was N. A relatively larger permeability ratio (K R ) was obtained for N when it was tested with sludge in the laboratory. The permittivity values obtained from the field specimens were also comparable to the measured permittivities from the laboratory tests obtained prior to the field study (i.e. C R values of. and. for the laboratory and field specimens, respectively). C R :8 was measured for s L, P, and I; they were considered unclogged because of K R values of less than during the laboratory tests. Further observations on the clogging behaviour were made through analysing the changes in the two constriction sizes. The decrease in O constriction sizes of the exhumed geotextiles compares well with the O of the ones exposed to filtration in the laboratory (Table ). Additionally, the percentage reductions in O 9 values are comparable. The geotextiles experienced slightly larger reductions in their O constriction sizes than their O 9 values.

8 Long-term filtration performance of nonwoven geotextile-sludge systems 7 Table. Changes in permittivity, and two characteristic constriction sizes of nonwoven geotextiles exposed to filtration with sludge in the laboratory and in the field Permittivity ratio, C R Decrease in O 9 constriction size (%) Decrease in O constriction size (%) Virgin geotextile permittivity (s ) s exposed to filtration in the laboratory s exposed to filtration in the field s exposed to filtration in the laboratory s exposed to filtration in the field s exposed to filtration in the laboratory s exposed to filtration in the field N.7.. NA NA NA NA L P I M..9 NA.6 NA NA K.. NA 8. NA 8.6 NA Note: NA, not analysed... Retention behaviour of the geotextiles... Specimens exposed to filtration in the laboratory Gradient ratio tests provided valuable information about the retention performance of the geotextiles, as the material that piped through was continually monitored. The amount of sludge piped through the geotextiles in all tests was in the range of 7 8 g/m, but less than g/m, a limit suggested by Lafleur et al. (989) for internal stability of soils (Figure 8a). A lower limit of 9 g/m was set for the retention of sludge solids above the geotextile. This limit was established by considering the environmental regulations that set a limit for the maximum concentration of PCBs in the environment and the amount of excess pore water discharged during consolidation of sludge in the lagoons (Aydilek ). The measured piping rates were still lower than this new lower limit adopted. Further analysis of the retention performance was performed in calculating the amount of piped sludge solids as an equivalent amount of silty sand (taking into account their different specific gravities), and the values are plotted against the geotextile permittivities in Figure 8b. The figure provides a direct comparison between the piped amounts of sludge solids and silty sand, and suggests that K(C ¼. s ) is the only one exhibiting a piping rate above the limit of 9 g/m. Piping rates were higher for the silty sand in all tests, and K (C ¼. s ) had a piped amount above the limit of g/m when it was tested with silty sand. Similar observations were made by Aydilek and Edil () for woven geotextiles, and could be attributed to the clod-like nature of sludge, which prevents the movement of solid particles freely since the particle size distributions of both materials were essentially the same. Figure 8 also suggests that, in general, piping rate increased with increasing permittivity for both soils when needle-punched geotextiles were tested, consistent with the findings of Krug et al. (). For the two geotextiles having the same permittivities, the heatbonded one ( N) experienced less piping than its companion ( L). This is probably due to the lower constriction sizes of heat-bonded geotextiles. Overall, permittivity is an important nonwoven geotextile pore structure parameter that controls retention performance. Retention performance was also evaluated by measuring the amount of piped sludge solids in the filter press tests, and the results are given in Figure 8a. The Piping rate (g/m ) Silty sand equivalent piping rate (g/m ) GR test - sludge GR test - silty sand FP test - sludge Mass limit of 9 g/m for sludge M K I L P N Virgin geotextile permittivity (s! ) GR test - sludge GR test - silty sand FP test - sludge Mass limit of 9 g/m for sludge L P N M I K Virgin geotextile permittivity: s! Figure 8. Retention performance of geotextiles based on laboratory tests: piping rate against permittivity; piping rates with an equivalent amount of silty sand against permittivity. (FP, filter press; GR, gradient ratio)

9 8 Aydilek and Edil percentage of retained sludge solids was in the range 97 99% in the filter press tests, and the measured piping rates were insignificant ( g/m ). No significant discoloration of the outflow was observed in any of the tests. A filter cake with a thickness of.. mm was formed on the surface of the geotextiles. The formation of a low hydraulic conductivity filter cake is usually desirable as long as it does not cause clogging, as it contributes to retention of solids. The filter press test is a quick and simple test method, and can be effective for comparative evaluation of short-term retention of sludge by geotextiles. However, long-term piping performance is specifically important in the case of contaminated materials such as sludges, and can be determined more definitively by collecting the fines at the bottom part of the permeameter after the gradient ratio tests. Post-gradient ratio test sieve analyses were performed on the sludge and silty sand samples taken from different depths in the permeameters, and they were compared with the particle size distributions (PSD) determined prior to testing (Aydilek ). Some of the PSDs were indicative of widespread piping consistent with the measured piped material collected at the bottom of the permeameter. Two plots showing the PSDs in different parts of the permeameter for the sludge geotextile testing are given in Figure 9 for demonstration purposes. For N (C ¼.7 s ), relatively less deviation is observed from the original sludge PSD when compared with the PSD of K (C ¼. s ) (i.e. up to % compared with up to 7%). The piping rates for those geotextiles were 7 g/m and 8 g/m respectively.... Laboratory tests on specimens exposed to filtration in the field Field observations during the exhumation process indicated that no significant piping of sludge through the geotextile and intrusion into the overlying cap was occurring. Similar observations were made when the specimens were transported to the laboratory for more accurate analyses (Aydilek and Edil ). The weight of the intruded solids was insignificant, in each case being less than g for g of collected cap material. This corresponded to a piping rate of g/m, lower than the limit of g/m set by Lafleur et al. (989) for laboratory tests and the limit of 9 g/m set considering the environmental regulations. All of the geotextiles were considered to have performed satisfactorily in terms both of separation of the sludge from the overlaying cap and of retention of the sludge solids by means of preventing excessive piping. Filter press tests were conducted on the exhumed and virgin samples of the geotextiles using the sludge, and the changes in effluent amount (also called filtrate loss) were calculated. The results are shown in Figure along with the changes in permittivity. The figure shows that the change in the filtrate loss values year after the field placement are, in general, comparable to the findings obtained from the permittivity tests. For instance, a % decrease in the filtrate loss is obtained for N, Percent finer by weight (%) Percent finer by weight (%) 8 6 PSD of sludge PSD - mm PSD - mm PSD - 7 mm Particle size, D (mm) Water inlet mm mm 7 mm Water outlet PSD of sludge PSD - mm PSD - mm PSD - 7 mm... Particle size, D (mm) Water inlet mm mm 7 mm Water outlet Figure 9. Changes in particle size distribution (PSD) of sludge exposed to filtration in the laboratory: N (C ¼.7 s ); K (C ¼. s ) Change in filtrate loss or permittivity of exhumed nonwoven geotextiles (%)!!!6!8 N L Geotextlie P I Change in permittivity Change in filtrate loss Virgin geotextile permittivity (s! ) Figure. Retention performance of geotextiles exposed to filtration in the field which is comparable to a 7% decrease in the flow capacity of the same geotextile as indicated by the permittivity test (i.e. C R ¼. would indicate a 7% decrease in flow capacity).... Comparison of laboratory and field results The geotextiles performed well both in the laboratory and in the field in terms of their retention performance

10 Long-term filtration performance of nonwoven geotextile-sludge systems 9 when tested with sludge. The piping rates were below the set limit, which is based on consideration both of the internal stability of the soil and of environmental protection from release of contaminated solids. The only exception to that was K, as discussed above. Field piping rates were generally comparable to those observed in the laboratory tests (i.e. 7 8 g/m as against g/m ), even though occasionally some of the geotextiles experienced higher piping rates in the field. This was attributed to additional dynamic loads exerted on the sludge by the trucks during construction. Piping rates generally increased with increasing permittivity in both the gradient ratio and the filter press tests. As mentioned above, the filter press test is a simple and quick test for comparing piping rates; however, the specimen size required for the test is small and the test duration is relatively short. The gradient ratio test provides the piping rates more effectively... Definition of an acceptable zone for filtration After combining the retention and clogging performances observed in the laboratory tests, an acceptable zone is defined for the filter behaviour of sludge geotextile and silty sand geotextile systems in Figure. For sludge only permeability ratios, and for silty sand both permeability and gradient ratios were employed, as those were the discriminating indicators Acceptable zone based on retention and clogging limits Piping rate Virgin geotextile permittivity (s! ) Gradient ratio Acceptable zone based on retention and clogging limits Virgin geotextile permittivity (s! ) Piping rate Figure. Acceptable filter zone for: sludge geotextile systems; silty sand geotextile systems Piping rate (g/m ) Piping rate (g/m ) of the clogging behaviour for each material. A new piping rate limit of 9 g/m is set as the acceptable limit for retention performance of contaminated sludge, whereas a limit of g/m is used for silty sand. An upper limit of is used for the clogging ratios with both materials. Also, a minimum permittivity limit of.7 s is set for sludge and silty sand, because the trends in the plots indicate that the geotextiles with permittivities smaller than.7 s may have clogging ratios greater than... Applicability of the existing filter criteria Existing geotextile filter selection criteria use size ratios in the form of O x /D x, where O refers to a characteristic constriction size of the geotextile and D refers to a characteristic particle size of the soil. In the analysis herein, O x was used to designate the constriction sizes determined from the probabilistic model CONS. Use of constriction size is important, because the minimum diameter of pore flow channel controls the filtration process. Two constriction sizes commonly used as part of both clogging and retention criteria, O 9 and O, were selected and correlated to the degree of clogging observed in the laboratory tests. Figure shows that, for either sludge or silty sand, O 9 and O were not related to the clogging ratios. Piping rates are plotted against the two characteristic sizes in Figure : as for the trends for clogging, no clear-cut trends could be observed between the piping rates and O constriction sizes. On the other hand, piping rates show a generally increasing trend with increasing O 9 constriction size. An extensive analysis was conducted by considering the commonly used geotextile constriction size-to-soil particle size ratios in the existing published filter criteria. The results of the long-term gradient ratio tests and the CONS-based geotextile constriction sizes were used to determine the applicability of widely used geotextile filtration criteria (Table ) to the nonwoven geotextiles tested with silty sand and sludge. Figure compares the actual performance observed for these systems in the laboratory tests with the predictions of the existing filter criteria. N was eliminated from the comparisons since, as mentioned before, at its current status CONS is capable of analysing only the needle-punched nonwovens. Five out of the six geotextiles tested did not experience significant excessive piping in the laboratory tests. K (a continuous filament nonwoven needlepunched geotextile) was the only geotextile that resulted in a relatively large piping rate. Figure shows that the existing empirical criteria predict the retention performance observed in the laboratory tests in most cases. The prediction percentage was in the range 6 8% for both silty sand and sludge (in one case the prediction was as low as %). For instance, a prediction percentage of 8% means that the particular retention criterion was able to predict the performance of four of the five geotextiles analysed. All the geotextiles except N, a heat-bonded nonwoven geotextile, performed successfully in the

11 Aydilek and Edil O (mm) O 9 (mm) Gradient ratio Gradient ratio O (mm) O 9 (mm) Figure. Relationship of clogging to O and O 9 for: sludge; silty sand Piping rate (g/m ) O (mm) Piping rate (g/m ) O 9 (mm) Piping rate (g/m ) O (mm) Piping rate (g/m ) O 9 (mm) Figure. Relationship of retention to O and O 9 for: sludge; silty sand

12 Long-term filtration performance of nonwoven geotextile-sludge systems Table. Existing nonwoven geotextile filter selection criteria used for comparisons Existing retention criteria Criterion Ogink (97) Schober and Teindl (979) Millar et al. (98) Carroll (98) Christopher and Holtz (98) Giroud (988) Fischer (99) Suggested ratio O 9 /D 9.8 O 9 /D. 7. O /D 8 O 9 /D 8 O 9 /D 8 O 9 /D 9/C u O 8 /D 7 Existing clogging criteria Criterion Millar et al. (98); Fischer (99) Christopher and Holtz (98) Christopher and Holtz (98); Koerner (997) French Committee on s and Geomembranes (986) Fischer et al. (99) Suggested ratio O =D O 9 =D Porosity % O 9 =D O =D.8. laboratory tests in terms of clogging performance. For sludge, four of the existing filter criteria and for silty sand, five of the existing filter criteria accurately predicted the clogging performance observed in the laboratory tests. The existing filter criteria were, in general, successful; however, they did not predict the observed filtration performance in all cases. Considering the criticality of the filter design, the selection of geotextile is crucial in contaminated materials, such as sludges. Therefore a parametric study is required to evaluate O x /D x ratios and find the best ratio that clearly discriminates clogging and retention performance for such geomaterials. % Success in prediction of retention performance % Sucess in prediction of clogging performance Carroll (98) Millar et al. (98) Millar et al. (98) Koerner (997) Christopher and Holtz (98) Fischer (99) Ogink (97) Christopher and Holtz (98) Schoeber and Teindl (979) French Committee on GT & GM (986) Fischer (99) Silty sand Sludge Giroud (988) Silty sand Sludge Figure. Comparisons with existing filter selection criteria: retention; clogging Fischer et al. (99). CONCLUSIONS Filtration performance of six different nonwoven geotextiles with contaminated organic wastewater treatment sludge was evaluated in the laboratory. For comparison, these geotextiles were also tested with an inorganic silty sand having the same particle size distribution as the sludge. Long-term gradient ratio tests and filter press tests were conducted with these two materials. Permittivity tests and image analysis were also performed on the geotextile samples before and after the gradient ratio tests. Filtration performance of geotextiles with sludge was also observed in field test cells. s placed over the sludge under a cap were exhumed after a period of months. Permittivity tests, filter press tests and image analysis were performed on the exhumed geotextiles. The following conclusions are advanced as a result of the laboratory and field studies:. The filtration characteristics of sludge are different from those of silty sand even though they may have the same particle size distribution. The presence of organic mass and the clod-like structure of the sludge results in more complicated filtration phenomena. This structure reduces piping of contaminated sludge solids but also promotes clogging of geotextiles.. The gradient ratio test (ASTM D ) has certain limitations when used with sludge. The gradient ratio, GR, as calculated does not necessarily reflect the actual clogging behaviour for sludge owing to the issues associated with testing organic sludge. However, another clogging ratio, i.e. permeability ratio (K R ), allows a clearer definition of clogging. A third ratio, permittivity ratio (C R ), introduced to quantify changes in flow capacity of geotextiles subjected to filtration, supports K R determined from gradient ratio tests, and provides an additional check on the results.. The modified filter press test is a quick and simple test method, and can be effective for comparative evaluation of short-term retention of sludge by geotextiles. Long-term piping performance, specifically important in the case of contaminated materials, can be determined more definitively by collecting the fines at the bottom part of the permeameter after the gradient ratio tests.

13 Aydilek and Edil. Permittivity is the most important nonwoven geotextile pore structure parameter defining filtration. When tested with sludges or silty sands, less clogging and more piping was observed with increasing permittivity. An acceptable filter zone is defined based on acceptable clogging and retention limits and permittivity values.. Clear trends could not be observed between the two commonly used geotextile constriction sizes, O 9 and O, and clogging or piping. Only O 9 sizes showed some correlation with piping. Additionally, ratios of geotextile constriction size to soil particle size in the existing filter criteria did not always predict the observed filtration performance. It appears that a parametric study is needed for evaluating various O x /D x ratios and finding the most discriminating ratio for clogging and retention performance.. s subjected to gradient ratio tests were analysed to determine the reduction in their two characteristic constriction sizes, O 9, and O, using an image analysis method. The decrease was less than % for most of the geotextiles. Larger reductions were observed for the geotextiles with relatively smaller permittivities, indicating that permittivity has a direct effect on the clogging performance of nonwoven geotextiles.. The field performance of geotextiles was generally consistent with that observed in the laboratory tests. In general, comparable C R values were obtained for the specimens exposed to filtration in the field or in the laboratory. The field geotextiles experienced slightly larger reductions in their O constriction sizes than in O 9 (up to % as against %). These reductions were, in general, comparable to those observed for the geotextiles exposed to filtration in the laboratory. Field piping rates were generally comparable to those observed in the laboratory tests. ACKNOWLEDGEMENTS The authors would like to express their appreciation to the Madison Metropolitan Sewerage District for its cooperation in the field tests and for partially funding the study; in particular, Mr David Taylor s assistance is gratefully acknowledged. The anonymous reviewers are also thanked for their valuable input. Mr Mark J. Stephani assisted in the laboratory and the field study. The opinions expressed in this paper are solely those of the authors and do not necessarily reflect the opinions of the Madison Metropolitan Sewerage District. NOTATIONS Basic SI units are given in parentheses. AOS C c CSD apparent opening size (m) coefficient of curvature (dimensionless) constriction size distribution (dimensionless) i soil geotextile interface C u coefficient of uniformity (dimensionless) D x soil particle size that x% of particles are smaller than (m) G s specific gravity (dimensionless) GR gradient ratio (dimensionless) i soil hydraulic gradient in soil tested (dimensionless) hydraulic gradient at soil geotextile interface (dimensionless) permeability ratio (dimensionless) hydraulic conductivity of soil tested (m/s) hydraulic conductivity of entire soil geotextile system (m/s) O x geotextile constriction size that x% of pores are smaller than (m) K R k soil k system REFERENCES C geotextile permittivity (s ) C R permittivity ratio (dimensionless) ASTM D 9, Test Methods for Water Permeability of s by Permittivity, American Society for Testing and Materials, West Conshohocken, Pennsylvania, USA. ASTM D, Standard Test Method for Measuring the Soil geotextile System Clogging Potential by the Gradient Ratio, American Society for Testing and Materials, West Conshohocken, Pennsylvania, USA. American Petroleum Institute (98). Standard Test Procedure for Field Testing Drilling Fluids, American Petroleum Institute Recommended Practice B, th edn, Washington, USA, pp. 9. Aydilek, A. H. (). Filtration Performance of -Wastewater Sludge Systems, PhD dissertation, University of Wisconsin- Madison, Madison, Wisconsin, USA, 6 pp. Aydilek A. H. (). Field performance of geotextile reinforced sludge caps. Proceedings of Geosynthetics, IFAI, Portland, Oregon, USA, February, pp Aydilek, A. H. & Edil, T. B. (). Filtration performance of woven geotextiles with wastewater treatment sludge. Geosynthetics International, 9, No., 69. Aydilek, A. H., Oguz, S. H. & Edil, T. B. (). Digital image analysis to determine pore opening size distribution of nonwoven geotextiles. ASCE Journal of Computing in Civil Engineering, 6, No., 8 9. Aydilek, A. H., Oguz, S. H. & Edil, T. B. (). Constriction size of geosynthetic filters. ASCE Journal of Geotechnical and Geoenvironmental Engineering, submitted. Carroll, R. G. (98). Filter Criteria. Transportation Research Record 96, Engineering Fabrics in Transportation Construction, D.C. p. Christopher, B. R. & Holtz, R. D. (98). Engineering Manual, Report FHWA-TS-86/, US Federal Highway Administration, pp. Faure, Y. H., Kehila, Y., Olivier, F. & Paillez, S. (). Behavior of nonwoven geotextiles for filtering particles in suspension. Filters and Drainage in Geotechnical and Geoenvironmental Engineering, Wolski, W. & Mlynarek, J., Editors, Balkema, Rotterdam, pp Fischer, G. R (99). The Influence of Fabric Pore Structure on the Behavior of Filters, PhD dissertation, University of Washington, Seattle, Washington, USA, pp. Fischer, G. R., Holtz, R. D. & Christopher, B.R. (99). Filter criteria based on pore size distribution. Proceedings of the Fourth International Conference on s, The Hague, The Netherlands,, pp French Committee on s and Geomembranes (986). Recommendations for the use of geotextiles in drainage and filtration systems. Manual, Institut Textile de France, Boulogne- Billincourt, France, pp. (as cited by Fischer 99). Giroud, J. P. (988). Review of geotextile filter criteria. Proceedings of the First Indian s Conference on Reinforced Soil and s, Bombay, India, pp. 6. Grefe, R. P. (989). Closure of papermill sludge lagoons using geosynthetics and subsequent performance. Proceedings of the th

14 Long-term filtration performance of nonwoven geotextile-sludge systems Annual Madison Waste Conference, University of Wisconsin- Madison, Madison, Wisconsin, USA, pp. 6. Haliburton, T. A. & Wood, P. D. (98). Evaluation of the US Army Corps of Engineers gradient ratio test for geotextile performance. Proceedings of the Second International Conference on s, Las Vegas, Nevada, USA,, pp. 97. Koerner, R. M. (997). Designing with Geosynthetics, th edn, Prentice Hall, Englewood Cliffs, New Jersey, USA, 76 pp. Krug, M., Heyer, D. & Floss, R. (). Filtration effectiveness of geotextile in cover sealing systems of landfills. Filters and Drainage in Geotechnical and Geoenvironmental Engineering, Wolski, W. & Mlynarek, J., Editors, Balkema, Rotterdam, The Netherlands, pp Lafleur, J., Mlynarek, J. & Rollin, A. L. (989). Filtration of broadly graded cohesionless soils. ASCE Journal of Geotechnical Engineering,, No., Millar, P. J., Ho, K. W. & Turnbull, H. R. (98). A Study of Filter Fabrics for Geotechnical Applications in New Zealand, Central Laboratories Report No.-8/, Ministry of Works and Development, New Zealand. Ogink, H. J. M. (97). Investigations on the Hydraulic Characteristics of Synthetic Fabrics. Publication No. 6, Delft Hydraulics Laboratory (as cited by Fischer 99). Schober, W. & Teindl, H. (979). Filter criteria for geotextiles. Proceedings of Seventh European Conference on Soil Mechanics and Foundation Engineering, Brighton, England,, pp. 9. Wayne, M. H. & Koerner, R. M. (99). Correlation between longterm flow testing and current geotextile filtration design practice. Proceedings of Geosynthetics 9, IFAI, Vancouver, British Columbia, Canada,, pp. 7. Zeman, A. J. (99). Subaqueous capping of very soft contaminated sediments. Canadian Geotechnical Journal,, No., The Editors welcome discussion on all papers published in Geosynthetics International. Please your contribution to discussion@geosynthetics-international.com by August.