Clogging of Gravel Drainage Layers Permeated with Landfill Leachate

Transcription

1 Clogging of Gravel Drainage Layers Permeated with Landfill Leachate Reagan McIsaac 1 and R. Kerry Rowe, F.ASCE 2 Abstract: Ten flow cells, called mesocosms, are used to investigate the effect of different gravel sizes 38 and 19 mm and operating conditions on clogging of leachate collection systems. These mesocosms simulated in real time and real scale the two-dimensional leachate flow conditions representative of a section of a continuous 300-mm-thick gravel drainage blanket adjacent to a leachate collection pipe in a primary leachate collection system. The tests were terminated after 6 12 years of operation. In some mesocosms the full 300 mm of gravel was. In others, the leachate level was initially set at 100 mm and the upper 200 mm were un. Although the flow through all mesocosms was similar, the clogging in the fully gravel was substantially more than in the partially gravel. After 6 years of operation, typically, less than 10% of the initial pore space was filled with clog material in the un gravel. For the zone, 45% of the initial pore space was filled with clog material in the fully design as compared to only 31% in the partially design. The 38 mm gravel performed much better than the 19 mm gravel. For example, it maintained a hydraulic conductivity that was higher than the 19 mm gravel even after operating for twice as long. Up to four mesocosms were placed in series, with the effluent from one mesocosm being the influent for another. The reduction in mass loading within the first mesocosm reduced the amount of clogging within the mesocosm later in series. There was a clear progression of decreasing amounts of initial pore space filled with clog material in the last mesocosm in series, and most of the clogging was due to the vertically percolating leachate. DOI: / ASCE : CE Database subject headings: Municipal wastes; Landfills; Hydraulic conductivity; Service life; Biofilm; Calcium carbonate; Drainage; Clogging. Introduction Leachate collection systems are a critical component of barrier systems in today s landfills. They control the leachate head acting on the landfill base liner and collect and remove contaminants, hence, minimizing contaminant impact on the environment. Modern leachate collection systems typically are comprised of a network of perforated high-density polyethylene HDPE leachate collection pipes embedded in a continuous drainage blanket of uniformly graded granular material covering the landfill base liner. Field evidence has shown that voids within the granular leachate collection layer become filled with clog material as a result of the growth of biomass, the bio-induced chemical precipitation of inorganic matter predominantly calcium carbonate and the accumulation of particulate matter Brune et al. 1994; Fleming et al. 1999; Maliva et al. 2000; Bouchez et al. 2003; Levine et al As the leachate collection layer clogs, the porosity 1 Ph.D. student, Dept. of Civil Engineering, Univ. of Western Ontario, London, Ontario, Canada N6A 5B9 2 Professor, Geoengineering Centre at Queen s RMC, Queen s Univ., Ellis Hall, Kingston, Ontario, Canada, K7L 3N6 corresponding author. kerry@civil.queensu.ca Note. Discussion open until January 1, Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on September 11, 2006; approved on February 12, This paper is part of the Journal of Geotechnical and Geoenvironmental Engineering, Vol. 133, No. 8, August 1, ASCE, ISSN /2007/ /$ and the hydraulic conductivity can be reduced to the point where the leachate head on the liner can no longer be controlled to the design level typically, 0.3 m Rowe et al. 2004, thus shortening the period of effective functioning of the leachate collection system. Since leachate collection systems may be required to collect and remove leachate for extended periods of time, it is important to be able to design them to minimize the clogging process and prolong their long-term performance and service life. The potential for clogging of many different design configurations used in practice is not fully understood. To investigate the effect of different-sized drainage materials and operating conditions on clogging of the gravel drainage material, experiments were initiated Fleming 1999; Fleming and Rowe 2004 to examine in real time and real scale the two-dimensional leachate flow conditions representative of a section of a continuous granular blanket adjacent to a leachate collection pipe in a primary leachate collection system. Details regarding the design and early operation of these cells called mesocosms are given by Fleming and Rowe Relevant to this paper are four different design configurations Table 1 involving a 300-mm-thick drainage. For all cases examined in this paper there was no filter separator between the waste and the drainage gravel. The effect of a filter has been reported by McIsaac and Rowe The objective of this paper is to examine the effect of grain size, un versus conditions, and mass loading on the clogging of the gravel at the time of the termination of these tests. Particular attention will be paid to the distribution of clog material in the gravel especially over the last year of operations / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007

2 Table 1. Summary of Mesocosm Design Variable and Duration of Operation. All Mesocosms Had a 300 mm Gravel Drainage Layer and a Waste Layer Directly over the Gravel No Separator Filter Mesocosm Gravel size mm Leachate level mm Leachate influent Leachate effluent Days Test duration Years C PS Fresh To C-23 4, C PS Fresh To C-26 2, C PS Fresh Discharged 2, C PS Fresh Discharged 2, C PS Second in series To C-24 2, C PS Second in series Discharged 2, C PS Third in series To C-25 2, C PS Forth in series Discharged 2, C S Fresh Discharged 2, C S Fresh Discharged Note: C S-1 was disassembled by Fleming 1999 and details are reported by Fleming and Rowe Materials and Methods Mesocosm Fabrication and Materials The mesocosms Fig. 1 and Table 1 were fabricated Fleming and Rowe 2004 from welded 9-mm-thick PVC sheeting and were built at a large enough scale internal dimensions measuring 565 mm in length, 235 mm in width, and 574 mm in height to simulate at full scale, in real time, and with materials typically used in practice the last 0.5 m of a continuous granular blanket adjacent to a leachate collection pipe in a primary leachate collection system. The collection system design generally consisted of waste material overlying a 300-mm-thick gravel drainage layer of crushed dolomitic limestone overlying a nonwoven geotextile/ sand cushion graded at 1.5% to a half section of PVC perforated pipe. The waste material was a mixture of refuse and cover soil taken from auger boreholes in an area of the City of London W12A Landfill Site and was 5 10 years in age at the time of sampling. The 38 mm gravel had a D 10 =20 mm, D 60 =27 mm, and D 85 =33 mm, an average initial porosity of 0.43, and an initial hydraulic conductivity of 0.78 m/ s. The 19 mm gravel had a D 10 =10 mm, D 60 =16 mm, and D 85 =19 mm, and an average initial porosity of The PVC perforated pipe had an internal pipe diameter of 102 mm, two rows of perforations, perforation diameter of 15.9 mm, and perforation spacing along the pipe of 127 mm. The mesocosm nomenclature Table 1 summarizes the case. For example, C PS-3 corresponds to mesocosm C-24 with 38 mm gravel, was partially PS, and was the third mesocosm 3 in series. Eight mesocosms used 38 mm gravel, two used 19 mm gravel, two operated with the 300 mm gravel layer fully S, and eight PS with the bottom 100 mm of the gravel and the top 200 mm un. Some mesocosms were placed in series with the effluent from one mesocosm being the influent for another with a maximum of four in series Table 1. The mesocosms in series were placed in line with a short length of tubing connecting the effluent port from one mesocosm to the influent port of the next in series. Four mesocosms C PS-1, C PS-2, C PS-3, C PS-4 comprised one series and two mesocosms C PS-1, C PS-2 comprised the other series. Fig. 1. Schematic of experimental mesocosm cells and the prescribed interval spacing and location over which wet mass measurements were made within the mesocosms at termination adapted from Fleming and Rowe 2004 JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007 / 1027

3 Mesocosm Operation The leachate used in the mesocosms was collected from the Keele Valley Landfill KVL for details, see Rowe The leachate was collected from a manhole on the main header line at the downstream end of the leachate collection system. Keele Valley Landfill leachate was introduced at rates representative of field conditions in two dimensions. The vertical rate of approximately 73 ml/ day corresponded to an infiltration of about 200 mm/ year. The horizontal flow of 3,456 ml/ day 1.26 m 3 /yr was selected to simulate the average horizontal flow in the drainage layer near the collection pipe corresponding to the same vertical infiltration rate over a 25 m drainage path to the collection pipe. Tests were conducted at 27±2 C to simulate the conditions anticipated in an active leachate collection system. One mesocosm C S-1 was terminated after 1.6 years of operation and results were reported by Fleming and Rowe Nine of the mesocosms were terminated after 6 years of operation and one C PS-1 after 12 years of operation. Experimental Analysis Operational Testing A testing program was implemented to monitor and quantify the amount of clogging and changes in leachate composition both temporally and spatially. Water quality testing was performed on leachate samples collected from before the influent valve and after the effluent valve using test methodologies described by McIsaac These samples were tested immediately to obtain chemical oxygen demand COD, calcium Ca 2+ concentration, and ph. Tests were performed to follow the change in drainable porosity, and hence, the change in void volume, with time as clogging developed. The measured drainable porosity is the ratio of the volume of the leachate removed to the total volume of the drained interval. The drainable porosity will be lower than the actual porosity because of incomplete draining of the leachate under gravity due to fluid adhering to the drainage medium and clog material. Drainable porosities were measured over the discrete intervals shown in Fig. 2. Termination Testing Termination of the mesocosms allowed for the inspection of the degree of clogging that occurred over their operational lifespan. The clogged drainage gravel was removed and stripped of clog material over the intervals shown in Fig. 1 and the total mass of wet clog material was measured for each section. This mass includes biological matter, chemical precipitates, and fines that had accumulated to form the clog material. The bulk density of the clog material was measured using ASTM D 854 ASTM Based on the mass of clog removed from the disassembled mesocosms, the bulk density of the clog material, and the initial void volume in each sample section, the volume of clog within each section and the resulting void volume occupancy VVO was calculated. The VVO is the ratio of the volume of pore space occupied by clog material to the initial void volume. Thus, a VVO of 100% would indicate that the initial void volume is completely filled with clog material. Clog samples were sent for elemental analysis. Hydraulic conductivity testing was performed on the clogged gravel from C PS-1 and C PS-1. At termination, the lid of the mesocosm was removed from C PS-1 and the un was removed by hand. Saran Wrap was placed over the surface of the gravel and wax was poured over the Saran Wrap to fill in the open void space at Fig. 2. Vertical profiles through the mesocosms showing the interval spacing and location over which drainable porosities were measured within the mesocosms: a 38 mm gravel C PS-1, C PS-1 ; 19 mm gravel C PS-1, C PS-1 ; second in series C PS-2, C PS-2 ; third in series C PS-3 ; and fourth in series C PS-4 ; b fully C S-1, C S-1 the gravel lid contact and to build up a surface that was level on which to place the lid. The lid was modified to fit inside the mesocosm. The lid was placed into the mesocosm while the wax was still warm so it would cure with the lid creating a tight seal with the lid. The hydraulic conductivity was obtained knowing the specified flow and the measured head difference between piezometers spaced at 100 mm intervals along the mesocosm. A solid piece of clogged 19 mm gravel was removed from the first 120 mm of the layer of C PS-1 and placed into a permeameter for hydraulic conductivity testing. Wax was used to seal the sample to the inside of the permeameter and to prevent short circuiting between the wall of the permeameter and the clogged sample. Influent and Effluent Leachate Characteristics The leachate from Keele Valley was highly variable during the period of testing. The change in the influent COD, Ca 2+ concentration, and ph was monitored and is shown in Fig. 3, and reflects the natural variability of leachate characteristics with time. The strength of the leachate from samples collected from the constantly circulated storage tanks, the supply manifold, and from the ends of the pump lines feeding the mesocosms are representative of the influent strength entering the mesocosms. From 0 to 675 days laboratory feedstock data used to represent the mesocosm influent strength were obtained from Fleming From 675 to 1,500 days data were obtained from Armstrong Subsequent data were obtained from McIsaac The leachate supplied to each operating mesocosm originated from the same source. During the operation of the majority of the mesocosms 0 2,250 days, the leachate had a high organic strength COD values indicative of a young leachate. The organic strength was mainly in the form of volatile fatty acids McIsaac After 2,775 days, when mesocosm C PS-1 was the only one in 1028 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007

4 Fig. 3. COD and Ca concentrations and ph values within influent leachate to the mesocosms data up to 675 days from Fleming 1999 ; data from 675 to 1,500 days from Armstrong 1998 operation, the concentration of key raw leachate constituents from the Keele Valley Landfill was lower than in previous years likely due to treatment of the leachate in the KVL leachate collection system before it reaches the collection sump Rowe and Van- Gulck Since the leachate collected had been subjected to significant treatment it was not representative of leachate entering the collection system. In order to have a composition especially in key components such as COD and Ca 2+ similar to earlier Keele Valley Landfill leachate, the leachate feedstock for Mesocosm C PS-1 was spiked between 2,826 and 3,950 days McIsaac After April ,950 days nonaugmented raw Keele Valley Landfill leachate was delivered to Mesocosm C PS-1. In general, the effluent ph values remained relatively constant at about 7.5 despite variations in the influent ph Fig. 3. The organics in the leachate degraded quickly as it passed through the mesocosms. Only a short length of gravel 565 mm partially occluded with biofilm was required to cause a significant reduction in the COD and Ca concentrations in the effluent leachate. In Mesocosms C PS-1 and C PS-1, with 38 mm gravel, the average COD and Ca concentrations in the effluent were 20 and 17% of that in the influent, respectively. The drop was higher for the mesocosms filled with 19 mm gravel C PS-1 and C PS-1 with the average COD and Ca concentrations in the effluent being 17 and 9% of that in the influent, respectively. After the first 100 days this reduction in COD and Ca concentration in the effluent leachate remained at these levels except as noted below for the remainder of the tests. An exception to this generalization occurred at times when the partially columns had much higher than normal influent COD concentrations Fig. 3. At these times there was less relative change in leachate concentration with the effluent concentration for Mesocosms C PS-1 and C PS-1 being 41 65% of the influent concentration. This is likely the result of the inability of the mixed population of bacteria to adapt to the increased concentrations within a short period of time. In contrast, the fully of C S-1 was better able to cope with the large fluctuations in influent COD concentrations, and the ratio of effluent to influent COD typically remained below about 30% during the period of high influent concentration. This is likely due to the much greater amount of biomass per unit volume of leachate and the longer retention times for leachate within a fully that had the same flow rate as the partially mesocosm. This also corresponded to substantially more clog mass being accumulated within the fully gravel drainage layer. Similarly, Mesocosms C PS-1 and C PS-1 that were filled with the 19 mm gravel had normalized COD values that were less variable than in the mesocosms filled with 38 mm gravel C PS-1 and C PS-1 with average ratios of effluent to influent concentration of 16% for COD during periods of high influent concentration. The 19 mm gravel has a much greater surface area per unit volume than the 38 mm gravel and this provides a much greater surface area for biofilm growth. This allows more exposure of the leachate to active biomass, and JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007 / 1029

5 Fig. 4. Distribution of wet solids within the mesocosms for partially 19 mm gravel C PS-1 after 6 years, 38 mm gravel C S-1 after 6 years, and partially 38 mm gravel after 6 years C PS-1 and 12 years C PS-1 hence more leachate treatment and greater clogging. An increase in the variability of the normalized values was observed near the end of the tests for the 19 mm gravel and this was hypothesized to be because, at this time, the majority of the void volume was filled with dense inorganic clog and the shearing of biofilm reduced the amount of active biomass in the voids. This hypothesis was confirmed at the time of disassembly where the clog mass was found to be substantially drier and denser with substantially less soft biofilm filling the voids than observed for the mesocosms filled with 38 mm gravel. The variations in the water quality of the leachate as it passed through the mesocosms are consistent with the leachate chemistry study by Rittmann et al An environment conducive to clog development or the precipitation of CaCO 3 is established within the leachate as it passes through the drainage material. Rittmann et al showed that the loss in COD was primarily due to fermentation of acetic acid to carbonic acid. This resulted in a shift in ph to higher values, which together with the increased carbonate concentration, promoted the development of inorganic clog material. Fully Saturated Conditions Fig. 4 shows the mass of wet solids for four mesocosms and the drainable porosities for all mesocosm are given in Figs. 5 and 6. The VVO within the mesocosms is given in Table 2. A comparison of the results for Mesocosms C PS-1 and C S-1 shows that the fully gravel yielded much greater clog development and lower porosity than in the partially mesocosm. For example, for the and mm intervals, 29 and 26% respectively of the voids were filled with clog material for the gravel C S-1 as compared to only 7 and 9% for the un gravel C PS-1. Higher VVOs were also measured within the mm interval of the fully design than in the corresponding zone of the mesocosms that were operated partially. The VVO was 87 and 55% within the 0 50 and mm intervals, respectively, for Mesocosm C S-1 versus 76 and 40% within Mesocosm C PS-1. Fully Mesocosms C S-1 and C S-1 were identical except that times of termination were 1.6 and 6.1 years, respectively. Within 1.6 years the gravel section from 0 to 70 mm was largely filled with clog material with a VVO of 92%. An additional 4.5 years of operation resulted in considerably higher amounts of clog throughout the remainder of the gravel thickness see C S-1 in Table 2. The VVO within the middle of the mesocosms from 50 to 260 mm ranged from 12 to 17% after 1.6 years of operation to 26 to 55% after 6.1 years of operation. The total flow was the same through the 100 mm zone in C PS-1 and the fully 300 mm in C S-1 while the volume of pores through which the leachate originally flowed was about three times larger for the fully case. This appeared to result in less localized solid cemented clogging but more soft clog in the fully mesocosm. For the fully gravel the voids in the bottom 75 mm visibly appeared to be completely filled 100% visual VVO with predominantly gelatin-like soft clog. However, this soft clog had a porous structure that explains the calculated VVO of 87% within the 0 50 mm interval. This was attributed to less competition from the development of inorganic clog for void space along with lower induced shear stresses on the active soft biofilm from the flowing leachate than for the 100-mm-thick mesocosm. Thus while the clog in the fully gravel was softer and porous, there was much more clogging than for the mesocosms run with only 100 mm. From 25 to 75 mm in the fully the clog 1030 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007

6 Fig. 5. Drainable porosities in the mesocosms with a 38 mm drainage gravel: a C PS-1; b C PS-1; and 19 mm drainage gravel c C PS-1; and d C PS-1 material was predominately black, gelatin like, and soft. From 0 to 25 mm the clog material was a light tan color and was drier and less viscous than the black biofilm Fig. 7. The light tan color and change in texture of the clog material are likely a result of the accumulation of material and siltation from the waste layer due to the absence of an effective filter between the waste and the gravel as indicated by the high silicon and aluminum content. The amount of Si was 12.89%/dry and Al was 3.00%/dry in the 0 50 mm interval versus concentrations less than Si= 1.22%/dry and Al= 0.25%/dry when a nonwoven geotextile filter was used between the waste and the gravel McIsaac and Rowe The Si and Al content of the fully Mesocosm C S-1 was also significantly higher than in Mesocosm C PS-1 Table 3. This is likely due to a periodic increase in the leachate level during normal operation from a gas lock or periodic clogging of the effluent valve for example, that would cause leachate to enter into the waste layer and likely facilitate the rinsing of particulate matter from the waste material and the accumulation within the base of the drainage. Greater amounts of biofilm growth occurred within the mm interval of the drainage gravel in the mesocosms operating under fully conditions than compared to the same interval when maintained un. From 100 to 300 mm a 3 5-mm-thick layer of biofilm growth localized on the top lateral surface of all the gravel particles for the mesocosm as shown in Fig. 8 a. This accounted for a large portion of the clog occluding the voids VVO=29%. In contrast, for the mesocosm where this zone was un, there was only a thin, typically, less than 2 mm, film and a VVO of less than 9% with a sporadic distribution of biofilm growth on the gravel Fig. 8 b. This is considered to be because conditions are more conducive to biofilm growth and accumulation than un conditions. When the gravel is, the organic components in the leachate are distributed throughout the mm intervals and there are sufficient nutrients to support the growth of biofilms throughout the gravel and the contact time between the bacteria and the leachate is far greater than when un. These results highlight the improved performance of the gravel drainage layer in terms of reduced clogging when the layer is not allowed to saturate and hence the importance of pumping leachate regularly rather than allow it to build up in the gravel drainage layer. Un Conditions The amount and distribution of clog material within the un gravel in designs operating partially was significantly different than in the s discussed above. Very little clogging occurred within the un gravel layers of the mesocosms run for 6 years and the VVO, was typi- JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007 / 1031

7 Fig. 6. Drainable porosities in the mesocosm operating fully : a C S-1, second in series; b C PS-2; c C PS-2, third in series; d C PS-3, fourth in series; and e C PS-4 cally, less than 10% within the un layers from elevation 100 to 260 mm. Even after 12.6 years of operation, very little clog had developed in the un gravel and the VVO was 12% in the un interval between 180 and 260 mm. The VVO of 32% in the mm interval for Mesocosm C PS-1 arose because clogging of the layer from 0 to 100 mm had caused the leachate level to rise and the interval mm was partially at later times giving rise to the substantially greater clogging of this zone in the test run for 12.6 years than for C PS-1, which was run for 6.1 years. Biologically induced clogging on the un gravel was not uniform. Very little solid inorganic clog was observed on the un gravel and the clog material was predominantly biofilm. The black biofilm growth was localized to the top of relatively horizontal surfaces of the un gravel while the bottom surfaces remained relatively pristine. Clog did not develop on the actual particle-to-particle contacts but black biofilm did grow where the interface gap between two particles was between about 1 and 3 mm. This is attributed to the retention of leachate due to a capillary fringe between the particles providing an environment in which biological growth could occur. Once the void distance between two particles was greater than 2 3 mm the capillary action ceased. This highlights the benefit of using large uniform particles such that the amount of retained moisture in the unsat / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007

8 Table 2. Void Volume Occupancy within the Mesocosms Mesocosm Time at termination years Interval mm C PS-1 19 mm gravel C S-1 fully C PS-1 first in series C PS-1 first in series C PS-2 second in series C PS-2 second in series C PS-3 third in series C PS-4 fourth in series C S-1 fully Note: C S-1 was disassembled after 1.6 years by Fleming 1999 and details are reported by Fleming and Rowe Unless otherwise noted, all mesocosms had 300 mm of 38 mm gravel in the drainage layer and the design allowed the lowest 100 mm to be and the remainder was un. Fig. 7. Accumulation of soft clog material surrounding a gravel particle removed from the base of the fully mesocosm 0 50 mm sample interval. Light brown discoloration of the typically black biofilm is due to particulate matter that was transported in from the waste layer. urated zone is minimized by minimizing the zones where a capillary fringe can develop. Only a fraction of the total surface area of the un was available for leachate retention and biofilm growth. The percentage of the surface area of the un gravel covered with biofilm increased from approximately 30±10% coverage within the upper mm to 50±10% coverage in the lower mm sections of the un. This is much lower than the 100% biofilm coverage within the. The increase in the amount of biofilm on the un gravel with depth coincides with a greater distribution of leachate and nutrients observed on the gravel with depth. The sporadic distribution of active biofilm in the un gravel limits the degree of contact between the bacteria and the leachate as the leachate flows through the un gravel and thus limits biologically induced clogging under un conditions. The amount of calcium in the clog material in the upper un gravel 13.1%/dry was substantially less than in the lower un gravel 27.8%/dry, indicating Table 3. Composition of Clog Removed from the Mesocosms with 19 and 38 mm Gravel after 6 Years Parameter 38 mm gravel C PS-1 lower mid 19 mm gravel C PS-1 upper 19 mm gravel C PS-1 lower Fully C S-1 lower Fully C S-1 upper 2nd in series C PS-2 lower 3rd in series C PS-3 lower 4th in series C PS-4 lower Water content %/wet Organic matter TVS;%/dry Carbonate as CO 3 %/dry Calcium, Ca %/dry Magnesium, Mg %/dry Silicon, Si %/dry Iron, Fe %/dry Sodium, Na %/dry Aluminum, Al %/dry Potassium, K %/dry Phosphorus, P %/dry Titanium, Ti %/dry Manganese, Mn %/dry Strontium, Sr mg/kg 1,000 1,230 1, Barium, Ba mg/kg Ca/CO Note: TVS total volatile solids. JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007 / 1033

9 Fig. 8. Gravel after 6 years operation from 100 to 180 mm interval for: a fully ; b un conditions. Note that the fully conditions created an environment conducive to biological growth on the gravel and the development of thick biofilms evident in a. also that very little biologically induced clogging was likely occurring in the upper un s. The clog material removed from the upper un of C-03 had substantially more Si 17.1%/dry and Al 3.5%/dry than in the lower un layer Si=4.32%/dry and Al=1%/dry, which is likely from the accumulation of fines rinsed from the waste material on the top lateral surfaces of the gravel. The composition of the clog removed from the lower un of C PS-1 Table 4 does indicate that the clog that has developed is similar in composition as that in the gravel indicating that similar biologically induced clogging is occurring in the lower un gravel as is in the gravel. It was also observed at termination that some of the gravel particles were covered with a thin layer of leachate but not a layer of biofilm. This suggests that the leachate flow pattern through the un is not constant with time. The low flow rate combined with a transient flow pattern does not provide a constant supply of leachate nutrients to sustain active biological growth over the entire surface area of the particles in the un. As identified in column studies by Rowe et al. 2000, leachate contaminant mass loading has a significant impact on the rate and extent of clogging. Reducing the mass of nutrients and inorganic material for biological activity and precipitation reduced clogging. Compared to the the un of a leachate collection system operates under lower flow rates and as a result the mass loading in terms of flow rate are lower than in the s adjacent to a leachate collection pipe. Thus, the relative lack of clog material within the un gravel of the mesocosms versus the large amount of Table 4. Composition of Clog Removed from the Mesocosms C PS-1 with 38 mm Gravel after 12.6 years Parameter Upper un Lower un Upper middle Lower influent Lower middle Lower effluent Pipe left-hand corner Pipe right-hand corner Water content %/wet Organic matter TVS;%/dry Carbonate as CO 3 %/dry Calcium, Ca %/dry Magnesium, Mg %/dry Silicon, Si %/dry Iron, Fe %/dry Sodium, Na %/dry Aluminum, Al %/dry Potassium, K %/dry Phosphorus, P %/dry Titanium, Ti %/dry Manganese, Mn %/dry Strontium, Sr mg/kg ,200 1,030 1, ,200 Barium, Ba mg/kg Ca/CO Note: TVS total volatile solids / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007

10 Table 5. Clog Densities Mg/m 3 within the 19 and 38 mm Gravel 19 mm gravel 38 mm gravel Interval mm Influent Middle Effluent Influent Middle Effluent ,709 1,799 1,483 1,300 1,281 1, ,483 1,468 1,438 1,574 1,164 1,235 clogging found within the gravel is attributed to the lower leachate mass loading to the un. Also, McIsaac and Rowe 2006 observed different degrees of clogging, with the use of different separator-filters and this will also result in different degrees of leachate treatment before the leachate enters the un. Therefore, the mass loading in terms of leachate strength organic and inorganic concentration entering the un s will differ for different filter separator designs. Effect of Particle Size The un 19 mm gravel clogged more rapidly than the 38 mm gravel. The 19 mm gravel resulted in more clog mass Fig. 4, higher VVO values Table 4, and lower drainable porosities Fig. 5 within the upper and lower sections of the un than the 38 mm gravel. Due to the smaller size of the gravel there were more particle to particle contacts where capillary action could retain the leachate and there were more flat lateral surfaces within the layer where leachate could pool on the surface of the gravel. The result was a higher volume of retained leachate in the un layer of the 19 mm gravel that allowed for the formation of clog material, predominantly the growth of biofilm. A similar amount of clog mass was removed from the 50 to 100 mm interval of the layer Fig. 4 and the drainable porosities were similar Fig. 5 for both the 19 and 38 mm gravel. However, the density of the clog was, typically, higher in the 19 mm gravel Table 5 due to the greater abundance of cementatious clog material. As a result the VVO value for the 19 mm gravel was marginally less than for the 38 mm gravel although the mass of clog per unit initial clean void volume was higher. Within the 0 50 mm interval of the gravel, significantly less clog mass was removed for the 19 mm gravel than the 38 mm Fig. 4 and the corresponding VVO values were 62 and 76% for the 19 and 38 mm gravel, respectively. However, at disassembly of the mesocosms the clog material from the 19 mm gravel appeared to be more mature than that from the 38 mm gravel. Column studies performed by VanGulck and Rowe 2004, which were terminated at different elapsed times showed clog progressed through different stages. As observed in their study, initially, biofilm developed quickly on the drainage material, then changed with time to a soft slime then to a slime with hard particles sand-size solid material in a soft matrix, and then to a solid porous concretion of coral like biorock structure. At disassembly of the mesocosms filled with 19 mm gravel the clog material appeared and felt harder and was absent of a thick layer of soft active biofilm. The entire surface area of the particles throughout the gravel for both the upper and lower layers and influent, middle, and effluent sections were completely covered with hard inorganic clog and the entire was cemented together like concrete. Most of the constrictions within the 19 mm gravel were filled with hard chemical clog and the voids between constrictions were partially filled. Some of the voids were filled with hard clog material having its own pore structure and hence secondary porosity. The 19 mm gravel had smaller openings to its remaining voids than the 38 mm gravel. The accumulation of soft clog was abundant in the larger voids of the 38 mm gravel. The lower seepage velocities and corresponding lower induced shear stresses due to the larger constriction openings in the 38 mm gravel allowed for the accumulation of softer clog than in the 19 mm. At the effluent end of the 38 mm gravel mesocosms, some voids were visually filled with a thick viscous clog however, unlike the 19 mm gravel, the particles were only lightly cemented together. The clog material within the first 100 mm of the gravel at the influent end of the 38 mm gravel mesocosms was more cemented than at the effluent end but had not reached the same level of solid cementation as observed in the 19 mm gravel. Not all of the gravel particles in the influent end of the 38 mm gravel were completely covered with hard inorganic clog after 6 years. Some of the voids were filled with mm clog material. Thick layers of soft biofilm were present. The cemented 38 mm gravel was separated by gentle prying with a screwdriver with more effort required in the influent end. Excessive force was not required to breakup the clog for the 38 mm gravel, in distinct contrast to the 19 mm gravel as discussed below. For the mesocosm tests using 19 mm diameter gravel, excessive force applied by a hammer to a chisel was required to break up the entire cemented for sampling Fig. 9 a. Even after 25 min exposure to mechanical agitation in a sieve shaker, chunks of concreted gravel remained intact. Cementation within the un 19 mm gravel was observed whereas none was observed for the un 38 mm gravel Fig. 9 c. Also, for the 19 mm gravel, less clog mass was required to give rise to substantial cementation than for the larger 38 mm gravel. Thus the high surface area, abundant particle-to-particle contact points, smaller void openings, and shorter distances between individual particles resulted in a highly cemented gravel with the same or even less clog mass for the 19 mm gravel than for the 38 mm gravel within a 6 year period. Within the gravel, lower VVO Table 2 were measured within the 19 mm gravel than for the 38 mm gravel yet the average measured hydraulic conductivity through the first 120 mm of clogged 19 mm gravel was m/s after only 6 years of operation. This was more than 47% lower than the m/s measured for the 38 mm gravel after 12.6 years of operation Fig. 10. Thus, due to the pore structure and the preferential development of clog at void openings, less clog was required to cause substantial reductions in hydraulic conductivity for the 19 mm gravel than the 38 mm. Composition of the clog in the 19 mm gravel was similar to that in the 38 mm gravel except that more magnesium was measured in the 19 mm gravel clog material Table 3. The amount of organic matter in the 19 mm clog material was the lowest of any mesocosm and is due to the relatively uniform intense mature clog formation throughout the 19 mm gravel discussed previously. JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007 / 1035

11 Fig. 9. Cemented mass of gravel removed adjacent to the influent port for: a 19 mm gravel C PS-1 ; b 38 mm gravel C PS-1 after 6 years. Note the larger size of the cemented clump under otherwise similar conditions for the 19 mm over than 38 mm gravel. Cementation of the un 19 mm gravel was also observed c. Based on the foregoing discussion, it can be concluded that the 38 mm gravel performed much better over a 12.6 year period than the 19 mm gravel did over a 6 year period and, hence, should be preferred for use as a drainage material for leachate collections systems that require a long service life. Mesocosms in Series Fig. 10. Measured hydraulic conductivities of the clogged gravel: a 38 mm gravel C PS-1 after 12 years; b 19 mm gravel C PS-1 after 6 years Fig. 11 shows the distribution of wet mass with distance from the inlet of the mesocosm for the mesocosms that were in series. Most of the mass was accumulated within the first 200 mm from where the raw leachate entered the system. For the mm interval, the amount of clog mass steadily decreased until it reached a relatively constant value at about 900 mm from the inlet. For the 0 50 mm layer most of the mass was contained in the first 1,000 mm from the inlet of the fresh leachate, but there was generally a gradual decline in mass along the entire 2,200 mm flow path. This was expected given the findings from Rowe et al that showed that a reduction in mass loading reduces clogging. Due to the reduction in mass loading with distance from the initial inlet, there was a clear progression of decreasing VVO values in the 100 mm from 58 to to 16 to 12% moving from the first to last mesocosm in series. Very little inorganic hard clog or cementation of the gravel Fig. 11. Distribution of wet solids in the mesocosms in series C PS-1, C PS-2, C PS-3, and C PS / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007

12 was observed after the first mesocosm indicating very little inorganic mass was available to precipitate on the granular medium for the later series mesocosms. Moving from the first to last mesocosm in series, the percentage of calcium in the clog material decreased, with percentages ranging from 25.6 to 21.3 to 20.9 to 20.0%/dry for C PS-1, C PS-2, C PS-3, and C PS-4, respectively, but had a higher percentage magnesium 5.2, 4.9, and 4.8%/dry for C PS-2, C PS-3, and C PS-4, respectively compared to the mesocosm first in series Ca=25.6%/dry, Mg=2.2%/dry in C PS-1. The biofilm was less viscous and thinner in the layer as the distance from the inlet increased. This is likely due to a decrease in the concentration of organic constituents in the leachate with distance due to microbial activity in the first 570 mm from the inlet. As discussed earlier, at exit from the first mesocosm, the organic concentrations in the leachate had dropped to 18% of the initial influent values. Drainable porosities in the increased from 0.2 in the first mesocosm in series, to approximately 0.35 in the second mesocosm and greater than 0.40 in the third and fourth in series Fig. 6. The leachate organic load in the mesocosms in series was sufficient to maintain some growth of biofilm that generally resulted in more clog mass than in the un gravel. In the last mesocosm most of the clogging was due to the vertically percolating leachate. The third and fourth mesocosms in series had a thin approximately 9 mm thick layer of gritty ooze an accumulation of fines and biofilm on the top surface of the base geotextile and is likely due to the washing of fines from the waste layer in the absence of a separator-filter between the waste layer and the drainage gravel. As would be expected, the clogging of the un zone was the same in all mesocosms since being in series had no effect on the leachate reaching the un zone. This illustrates the repeatability of the experiments, as does the similarity of results for duplicated Mesocosms C PS-2 and C PS-2 Table 2. Biofilm development within the layers of the mesocosm in series occurred predominantly on the top of the gravel and at particle-to-particle contacts. This indicates that clog development within the layer is not initially uniform over the entire surface area of the gravel particles and that the accumulation of fines on the top lateral surfaces of the gravel may promote accelerated clog development initially. However, eventually clogging extended fully around the gravel in the zone. It is noted that the clogging observed in these mesocosm tests over 6 years was less than observed in the field after 4 5 years by Fleming et al This suggests that the Keele Valley leachate used in the mesocosm tests was considerably lower in its concentration of fatty acids and calcium than the leachate that must have been flowing into the Keele Valley collection system from the waste. This hypothesis is consistent with the findings from the mesocosms in series, which show that there is considerable depletion of fatty acids and calcium with distance as leachate passes through the drainage layer. This indicates that the use of the end-of-pipe leachate concentration for predicting collection system performance may not be conservative and more research is required to characterize leachate as it enters leachate collection systems in the field. Effect of Time Fig. 12. Clog in pipe after: a 6 years C PS-1 ; b 12.6 years C PS-1 of operation Mesocosm C PS-1 operated for twice as long as its duplicate C PS-1 and experienced significantly more clog development within certain areas of the drainage gravel. As might be expected, the additional 6 years of operation had no real affect on the amount of intruded waste material at the waste/gravel interface in the interval mm Fig. 4 and similar VVO values of 56 and 51% were measured in this interval for C PS-1 and C PS-1, respectively. Fig. 5 shows a general slow decreasing trend in the drainable porosity values with time due to the relatively slow continuous clog development in the interval of mm of gravel that remained un for the entire 12 years of operation. For this un zone there was a 33% increase in VVO, from 9% after 6 years to 12% after 12 years, however the clogging in the un zone remained well below that in the gravel where the VVO was 98 and 63% in the lower and upper gravel of C PS-1. Thus, the additional 6 years of operation resulted in the 0 50 mm interval becoming essentially totally occluded with clog with a VVO of 98% after 12 years compared to 75% after 6 years. An additional 23% of the void volume was filled with clog in the mm interval and an additional 25% in the initially un mm layer between 6 and 12 years. The additional 6 years of operation also resulted in significantly more clog mass within the pipe as shown in Fig. 12. After 12 years, 75% of the pipe volume was filled with clog material. Clog material did not develop on the portion of the pipe above the top perforations Fig. 12 and it did not occlude the perforations. Between the top perforation and the deposit of clog material in the pipe, the pipe wall was coated with an approximately 4-mm-thick layer of very hard cemented clog material. The majority of the clog material in the pipe was calcium and carbonate Table 4. Although the top half of the accumulated clog material in the pipe appeared to be somewhat homogeneous, the clog material in the bottom half of the pipe was not. In the left-hand corner of the pipe Fig. 12 the clog material was sand sized material in a soft matrix and was denser, drier, and flat black compared to the other corner right-hand side of Fig. 12, which was a soft, gelatin, and very glossy black slime. Hydraulic conductivity measurements were made along the JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007 / 1037

13 length of C PS-1 in the just prior to termination. The average measured hydraulic conductivities from Section 1 to 6 were , , , , , and m/s Fig. 10. The lowest hydraulic conductivity in the 38 mm gravel was measured at the influent Section 1 and increased through to Section 4. Sections 5 and 6 had values less than Section 4. The high value obtained in Section 4 coincides with the lower amount of mass measured within the middle of the mesocosm Fig. 10 within the 0 50 mm interval. The clog that was deposited in the pipe was sufficient to cause a reduction in hydraulic conductivity to values similar to that measured in the clogged gravel directly adjacent to the pipe. The hydraulic conductivity for the clog material in the pipe Section 6 was m/s compared to m/s in the clogged gravel adjacent to the pipe Section 5. Total clogging of the base of the drainage gravel in C PS-1 and the bottom portion of the pipe prevented significant flow of leachate through the bottom perforations in the pipe and resulted in an increase in the leachate level such that it could flow into the pipe through the top perforations. This change in flow with the consequent increase in mass loading resulted in the previously un gravel intervals from 0 to 20 and 20 to 60 mm Fig. 5 to become and led to an increase in the clog formation in this zone compared to that at 6 years. The drainable porosity results for C PS-1 Fig. 5 b show that by approximately 2,500 days 6.8 years the gravel intervals 20 to 0 mm and 40 to 20 mm had reached drainable porosity values of less than 0.1. Beginning around 2,100 days 5.7 years the measured drainable porosity in the initially un zone from 0 to 20 mm, just above the initially zone, decreased from about 0.35 to 0.20 in roughly 500 days 1.4 years as the zone became and mass loading increased due to the preferential flow of leachate through this more permeable gravel. As further clogging developed in the zone the leachate level continues to rise and so the initially un zone from 20 to 60 mm experience a rapid decrease in drainable porosity over a period of 2.6 years from a relatively unclogged value of approximately 0.40 at 2,550 days 7 years to 0.20 at 3,500 days 9.6 years. Very little clog developed within the drainage gravel after approximately 10.3 years 3,750 days in Fig. 5 due to the low strength of the leachate supplied to C PS-1 after this time. Clog samples were retrieved from many locations within Mesocosm C PS-1 to allow an identification of the spatial distribution of clog composition with the mesocosm. The upper layer had percentages of Si and Al of 4.5 and 0.9%/dry, respectively, that were higher than the corresponding values of 1.4 and 0.4%/dry in the lower layer. This indicates that, in the absence of a filter separator, the fines transported from the waste material above the drainage gravel could potentially be trapped in the soft biofilm in the upper layers. It was also found that the percentages of Si and Al of 17.0 and 3.5%/dry, respectively, in the upper un gravel was much higher than the lower un gravel where the corresponding values were 4.3 and 1%/dry, respectively. At the influent end of the gravel of C PS-1, where the cementation of the gravel was more intense and the clog was more mature than in the middle or effluent sections of the mesocosm, calcium and carbonate 30.5 and 52.4%/dry represented almost 83% of the clog while the biofilm only represented 6.9%/dry. This clog was also had a moisture content of 33.3%/wet, which was much lower than 58 60%/wet in the lower middle and effluent sections of the gravel. These sections also had lower calcium carbonate with Ca representing %/dry and carbonate %/dry for a total of 75.8 and 71.5% of the clog. In contrast there was more biofilm that at the inlet, with biofilm representing 10.0% of the dry mass. Conclusions Mesocosm experiments which simulate the last 500 mm of the drainage layer closest to the leachate collection pipe in a landfill under field conditions were terminated after 1.6, 6, and 12 years. The mesocosms discussed herein examine the effect of versus un conditions, grain size, and mass loading on the clogging process and the extent of clog development. This work has shown that: 1. Maintaining the 300 mm drainage systems fully resulted in greater overall clogging, with a VVO of 45%, than was observed for mesocosms where the zone was confined to only 100 mm and the VVO was 31% after 6 years. Operating the full height of the gravel drainage layer submerged increased the retention time of the leachate within the gravel and created an environment more conducive to microbial growth on the gravel throughout. The development of thick biofilms and accumulation of soft clog was the dominant clog mechanism over the entire drainage layer thickness for the fully layers. Periodic increases in the leachate level into the waste layer during the operation of the fully designed mesocosm resulted in more clogging due to siltation and the rinsing of particulate matter into the base of the drainage layer. In a field case, leachate collection systems operating fully would offer even less control of leachate levels and this phenomena may be expected to further increase clogging within the, especially in designs with no filter separators between the waste and the gravel. Designing leachate collection systems that maintain low leachate heights would reduce the potential for clogging due to fines rinsing from the waste. 2. Very little clogging occurred within the un gravel layers. VVO were, typically, less than 10% over 6 years and 12% over 12 years in the un gravel. The clog material that did develop within the un gravel was predominantly biological and was limited to areas on the gravel where leachate could be retained for instance on the top lateral surfaces of the gravel and near particle-to-particle contacts. Short leachate retention times, low leachate contaminant mass loading in terms of flow rate and concentration, and a sporadic distribution of biofilm limit biologically induced clogging within the un gravel. Leachate drainage systems should be designed to operate with a minimum drainage height, for example, by keeping them pumped and not allowing them to remain in a state, to reduce the impact of clogging and to extend the service life of the leachate collection system. 3. The 38 mm gravel performed much better over a 12.6 year period than the 19 mm gravel did over a 6 year period. The 38 mm s were not as severely cemented with dense precipitated clog as the 19 mm gravel and there was considerably less soft biofilm within the un 38 mm s than the 19 mm gravel. A hydraulic conductivity of m/s for the 38 mm gravel after 12.6 years was higher than the m/s measured for the 19 mm gravel after 6 years. Less clog was required to cause the reduction in hydraulic conductivity for the 19 mm 1038 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / AUGUST 2007