Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Numerical Simulation of Fly-Ash Transport in Three Sands of Different Particle-Size Distributions Using Hydrus-1d

Numerical Simulation of Fly-Ash Transport in Three Sands of Different Particle-Size Distributions... J. Hydrol. Hydromech., 59, 2011, 3, 206­216 DOI: 10.2478/v10098-011-0017-5 RADKA KODESOVÁ1), ALES KAPICKA2), JAKUB LEBEDA1), HANA GRISON2), MARTIN KOCÁREK1), EDUARD PETROVSKÝ2) 1) Department of Soil Science and Soil Protection, Czech University of Life Sciences Prague, Kamýcká 129, 165 21 Prague 6, Czech Republic; Mailto: kodesova@af.czu.cz 2) Institute of Geophysics, Academy of Sciences of the Czech Republic, Bocní 1401, 141 31 Prague 4, Czech Republic. Study is focused on the numerical modeling of fly-ash transport in three sands, which was experimentally studied in the laboratory. Sands were packed in glass cylinders with diameter of 5.52 cm and height of 18 cm. Sands were also packed in plastic cylinders with diameter of 30 cm and height of 40 cm. The fly-ash and pulse infiltrations were applied on the top of all cylinders. Visually observed and gravimetrically evaluated fly-ash migration in small cylinders corresponded to fly-ash mobility in large columns detected using the SM400 Kappameter. The HYDRUS-1D code was used to simulate observed fly-ash transport. Parameters of soil hydraulic functions were either obtained using the Tempe cells and the RETC program or estimated using numerical inversion of transient water flow data measured in both types of columns using HYDRUS-1D. Parameters characterizing colloid transport in sands were then estimated from the final fly-ash distribution in sandy columns using attachment/detachment concept in HYDRUS-1D. Fly-ash mobility increased with increasing sand particle sizes, e.g. pore sizes. Particle sizes and pore water velocity influenced the attachment coefficient, which was calculated assuming filtration theory. The same longitudinal dispersivity, sticking efficiency and detachment coefficient sufficiently characterized fly-ash behavior in all sands. KEY WORDS: Sand, Fly-ash Migration, Magnetic Attachment/Detachment Concept, Filtration Theory. Susceptibility, Numerical Simulation, Radka Kodesová, Ales Kapicka, Jakub Lebeda, Hana Grison, Martin Kocárek, Eduard Petrovský: NUMERICKÁ SIMULACE TRANSPORTU ÚLETOVÉHO POPÍLKU VE TECH PÍSCÍCH RZNÉHO ZRNITOSTNÍHO SLOZENÍ POMOCÍ HYDRUS-1D. J. Hydrol. Hydromech., 59, 3; 45 lit., 4 obr., 4 tab. Studie je zamena na numerické modelování transportu úletového popílku ve tech píscích, který byl experimentáln studován v laboratoi. Písky byly nahutnny ve sklenných válcích o prmru 5,52 cm a výsce 18 cm. Písky byly také nahutnny v plastových válcích o prmru 30 cm a výsce 40 cm. Na povrchu válc byly aplikovány jednorázové infiltrace vody s popílkem. Migrace úletového popílku pozorovaná vizuáln a zjistná gravimetricky v malých válcích odpovídala mobilit úletového popílku detekované Kappametrem SM400 ve velkých válcích. Pozorovaný transport úletového popílku byl simulován programem HYDRUS-1D. Parametry hydraulických funkcí byly získány bu pomocí Tempských cel a programu RETC nebo odhadovány numerickou inverzí transientních data mených na obou typech válc programem HYDRUS-1D. Parametry charakterizující transport koloid v píscích byly potom odhadovány z konecné distribuce úletového popílku v píscitých sloupcích uzitím konceptu attachment/detachment (pipojení/odpojení) v programu HYDRUS-1D. Mobilita úletového popílku se zvysovala se zvysující se velikostí píscitých zrn, tj. s velikostí pór. Velikost zrn a pórová rychlost ovlivnila depozicní (attachment) koeficient, který byl pocítán na základ filtracní teorie. Stejné hodnoty podélné disperze, efektivity blokování (sticking efficiency) a mobilizacního (detachement) koeficientu charakterizovaly chování úletového popílku ve vsech píscích. KLÍCOVÁ SLOVA: písek, migrace úletového popílku, magnetická susceptibilita, numerická simulace, attachment/detachment koncept, filtracní teorie. Introduction Soil and water contamination due to atmospheric deposition is a serious environmental problem especially in the industrial regions. It was shown that magnetic measurements can serve as proxy of industrial immisions as well as heavy-metal contamination (Petrovský and Elwood, 1999; Hanesch et al., 2003; Boyko et al., 2004). The method is based on the fact that industrial fly-ashes contain about 10% of ferrimagnetic particles, namely Fe-oxides (Flanders, 1994; Kapicka et al., 2001). Magnetic susceptibility of soils was mapped in surroundings of local sources of pollution (Kapicka et al., 1999; Hanesch and Scholger, 2002; Spiteri et al., 2005; Sharma and Tripathi, 2007), or at regional to national scales (Hay et al., 1997; Heller et al., 1998; Magiera and Strzyszcz, 2000; Kapicka et al., 2008). Ferrimagnetic particles are usually accumulated at the soil surface. However, they may be transferred into grater depths due to the soil tillage, erosion etc. In some soils (like sandy soils or soils with well developed and stable aggregates) particles may migrate within the soil profile. The soil and groundwater contamination associated with the presence of suspended particles has recently attracted significant attention because of either harmfulness of colloid particles themselves or due to toxic substances adsorbed onto the colloidal particles. Colloid transport is studied in both the macroscopic and microscopic scales. Many studies have been done in microscale using the artificial micromodels to show single particles behavior in pores of specific shapes and under different pore saturation by water via monitoring of particle trajectories or their attachment to solid particles due to the small pore radius, pore blocking by colloids, attachments to other colloid particles, water film covered solid particles or water-air interfacial tension (Sirivithayapakorn and Keller, 2003; Keller et al., 2004; Auset and Keller, 2004). Colloid transport was also experimentally studied at macroscale, such as in laboratory studies of Bolster et al. (1999), Bradford et al. (2002), Gargiulo et al. (2007, 2008), Torkzaban et al. (2008), Foppen et al. (2010) and Schinner et al. (2010) who investigated bacteria mobility in porous columns, or in the field study Wossner et al. (2001) who evaluated the viral transport in sand and gravel aquifer. Colloid transport may be mathematically simulated in microscale (Gao et al., 2010). Several mathematical models describing colloid transport in macroscale have been also designed (Sim and Chrysikopulos, 2000; Bradford et al., 2003; Kim, 2006). The conceptual model for colloid transport included in HYDRUS-1D code (Simnek et al., 2008) accounts for colloid attachment, straining and exclusion. The HYDRUS-1D code was successfully tested and results were presented by Bradford et al. (2003, 2004 and 2005), Jiang et al. (2010) or Gargiulo et al. (2007, 2008). Fly-ash transport in porous media has not been previously studied. Fly-ash was usually used to change properties of original porous material. Among others Zhang et al. (2010) used fly-ash to increase phosphorus sorption in sand. In experimental and mathematical studies mentioned above the maximum colloid-size was 1 m that is 10 times smaller then the average size of fly-ash particles. As showed by Gargiulo et al. (2007), bacteria (R. rhodochrous), transport of which was observed and simulated in tree sands, may create threedimensional aggregates that can reach the size of 10­15 m, while single cells have a diameter of only 0.8­1 m. They suggested that tendency of this bacteria to form aggregates (similar size as the average size of fly-ash particle) enhanced their deposition in the porous material. The main goals of this study were: 1. experimental evaluation of fly-ash mobility in three types of sands, 2. numerical evaluation of such size particles transport within the sand columns using the HYDRUS-1D code, 3. specification of sand vulnerability to fly-ash contamination depending on soil particle-size distribution. Material and methods Experimental data Small columns Three technical quartz sands (composition, is documented in thin section images, Fig. 1) from the sand mine Sklopísek Stelec were used to study flyash mobility in porous media: ST8 (particle sizes 0.10­0.63 mm) ­ fine sand, ST03/08 (particle sizes 0.315­0.80 mm) ­ medium sand, and ST 06/12 (particle sizes 0.63­1.25) ­ coarse sand. Sands were packed in glass cylinders with diameter of 5.52 cm and heighs of 18 cm. Two grams of fly-ash and 100 cm3 of distilled water were applied on the top of the sand columns (SGC1 run). Suspension infiltration and outflow at the column bottom was monitored. The typical breakthrough curve was not measured. The small column experiments were performed to 207 validate fly-ash distribution in large soil columns described using the magnetic susceptibility data (see below) and to obtain hydraulic and transport parameters under similar conditions. The fly-ash migration was assessed visually. Then the soil columns were divided into the 2 cm thick layers. Flyash was washed out from each sand layer by 100 cm3 of distilled water and after that water was extracted using the evaporation table. The amount of fly-ash in each sand layer was determined gravimetrically. To eliminate possible impact of very fine particles presented in the sands, approximately the same amounts (3 replicates) as in each layer of clean sands were also washed by 100 cm3 of distilled water and water was again extracted using the evaporation table. Detected amount of fine particles were subtracted from values determined for column layers. Fly ash content in soil column was expressed as a total fly-ash mass per unit of dry sand, mt [M M-1]. Experiments were repeated with the same amount of fly-ash, and 200 or 300 cm3 of distilled water (SGC2 and SGC2 runs) (Tab. 1). Fig. 1. Sand morphology and pore structure studied on micro-morphological images of fine (top left), medium (top right) and coarse (bottom) sands. T a b l e 1. Characteristics of experiments performed in sands packed in small glass columns (SGC): inflow (Qinf) and outflow (Qout) volume, inflow (cinf) and outflow (cout) concentration, time of infiltration pulse (tinf), time of outflow beginning (tout,ini), total simulated time (ttot). Sand Fine ST 06/12 Exp. run SGC1 SGC2 SGC3 SGC1 SGC2 SGC3 SGC1 SGC2 SGC3 Qinf [cm3] 100 200 300 100 200 300 100 200 300 cinf [mg cm-3] 20 10 6.7 20 10 6.7 20 10 6.7 tinf [min] 0.40 1.37 2.78 0.19 0.52 1.12 0.17 0.29 0.42 Qout [cm3] 0 45.9 145.4 0 67.0 167.7 0 83.7 183.0 cout [mg cm-3] ­ 0 0.123 ­ 0.111 0.077 ­ 3.80 2.77 tout,ini [min] ­ 1.07 1.05 ­ 0.47 0.43 ­ 0.25 0.22 ttot [min] 5 5 5 3 3 3 2 2 2 Large columns Sands were also packed in plastic cylinders with diameter of 30 cm and height of 40 cm. The 5 cm coarse sand drainage layers were formed below the studied sand layers. The sensors SM200 (Delta-T Devices Ltd., 2005) for soil-water content measurements and micro-tensiometers Tensior 5 (UMS GmbH, Munich, 2005) for pressure head measurements were placed 10, 20 and 30 cm below the sand surface to monitor water regime within the sand columns. The plastic tube for inserting the Kappameter SM400 (Petrovský at al., 2004) was positioned in the column center. The SM400 device, which measures vertical distribution of magnetic susceptibility, was used to monitor migration of the fly-ash ferrimagnetic particles. The water regime was initially studied on each soil column using the ponding infiltration (maximum ponding depth of 5 cm) on the column top. Three different fresh water infiltration pulses were applied in different experimental runs (LPC1, LPC2, LPC3) (Tab. 2). Next the fly-ash (emitted from Porici power plant) and fresh water was applied (LPC-FA run). Time intervals between experiments were 2 to 4 weeks to insure similar initial conditions. The applied fly-ash doses varied for different sands due to the sensitivity of the SM400 device. The applied fly-ash amount increased with expected increase of fly-ash mobility in each sand material: 50 g on the fine, 100 g on the medium and 150 g on the coarse sand. Water regime was monitored using the SM200 sensors and micro-tensiometers Tensior 5. The final magnetic susceptibility within the soil profile and consequently fly-ash distribution was measured when no significant water flow and consequently no fly-ash transport was observed. T a b l e 2. Characteristics of experiments perform in sands packed in large plastic columns (LPC): inflow volume (Qinf), inflow concentration (cinf), time of infiltration pulse (tinf), total simulated time (ttot). Sand Fine ST 06/12 Exp. run LPC1 LPC2 LPC3 LPC-FA LPC1 LPC2 LPC3 LPC-FA LPC1 LPC2 LPC3 LPC-FA Qinf [cm3] 4700 8910 15850 3700 5350 9780 10430 3120 4950 16750 17110 2100 cinf [mg cm-3] ­ ­ ­ 13.51 ­ ­ ­ 32.05 ­ ­ ­ 71.43 tinf [min] 1.67 3.88 7.50 1.31 0.68 1.28 1.42 0.40 0.35 1.22 1.25 0.15 ttot [min] 40 60 70 20 50 60 70 10 30 40 75 10 Numerical simulations Theory The HYDRUS-1D program (Simnek et al., 2008a,b) was applied to simulate water flow and fly-ash transport in sands. The Richards equation, which describes the one-dimensional isothermal Darcian flow in a variably saturated rigid porous medium, is used in this model. To solve this equation, the soil hydraulic properties must be specified. The van Genuchten (1980) analytical expressions are used to describe soil hydraulic functions, the soil water retention curve, (h), and the hydraulic conductivity function, K(): e= (h) - r = s -r ( 1+| h| ) , h<0 e = 1, h0 (1) m K( ) = K s el 1 - ( 1 - 1/ m ) , e h<0 (2) K( ) = K s, h0 where is the soil water content [L3 L-3], e ­ the effective soil water content [­], r and s are the residual and saturated soil water contents [L3 L-3], respectively, h ­ the pressure head [L], l ­ the poreconnectivity parameter [­], ­ the reciprocal of the air-entry pressure head [L-1], n [­] is related to the 209 slope of the retention curve at the inflection point, and m = 1 - 1/n, K ­ the hydraulic conductivity [L T-1], Ks ­ the saturated hydraulic conductivity [L T-1]. Parameters r, s, , n and Ks must be known when simulating water regime. A modified form of the advection-dispersion equation is employed in HYDRUS-1D to simulate virus, colloid, and bacteria transport assuming attachment/detachment concept. Mass transfer between the aqueous and solid phase is described using following equation: s = ka c - kd s, t (3) where c is the (colloid, virus, bacteria) concentration in the aqueous phase [M L-3], s ­ the solid phase (colloid, virus, bacteria) concentration [M M-1], d ­ the bulk density [M L-3], ka ­ the first-order deposition (attachment) coefficient [T-1], kd ­ the first-order entrainment (detachment) coefficient [T-1], and is a colloid retention function [­], which is function of the diameter of the sand grain, dc [L], and coordinate. The attachment coefficient may be calculated using filtration theory as a function of the soil water content, diameter of the sand grains, dc [L], diameter of the colloid particles, dco [L], sticking efficiency, [­], pore-water velocity v [L T-1], and the single-collector efficiency, [­]. Theory is in greater detail described in HYDRUS1D manual (Simnek et al., 2008a). Parameters d, dc, dco, , and kd must be known when simulating colloid transport. In addition the longitudinal dispersivity, DL [L], must be defined to solve the advection-dispersion equation. Small columns The water flow within the sand columns was initially numerically analyzed. Soil water retention curves were measured using the Tempe cells. Sands were packed into 100-cm3 cylinders (soil core height of 5.1 cm and diameter of 5.0 cm) placed into the cells (3 replicates for each sand). Initially, fully saturated soil samples were slowly drained using six pressure head steps (­10, ­30, ­50, ­100, ­200 and -350 cm) during a 2-week period. Data points of soil water retention curve were evaluated using the final soil water content and water outflow monitored using byrettes. The RETC program (van Genuchten et al., 2005) was used to obtain r, s, and n parameters of van Genuchten function (Tab. 3) by fitting data of all 3 replicates (which were 210 very similar). Parameters r were set at 0.01 cm3 cm-3 in all cases. Parameter n was restricted to be maximally 3 (larger values caused computational instabilities when simulating some of following water and flay-ash transport scenarios). The saturated hydraulic conductivities, Ks, were estimated using numerical inversion of outflow data measured on small glass columns (SGC3 ­ infiltration pulse of 300 cm3) using HYDRUS-1D. Initial conditions were set at low pressure head values of ­500 cm corresponding to very dry conditions. Observed infiltration rates were used to specify the top boundary conditions. Seepage face boundary conditions were defined at the bottom. Next, the fly-ash transport was simulated. The bulk density, d, diameters of the sand grains, dc, and diameter of the colloid particles, dco, were set to measured values (Tab. 4). The longitudinal dispersivity, DL, sticking efficiency, , and the detachment coefficient, kd, were calibrated manually from the final fly-ash content, mt, in soil columns. Since mt values expressed the average total fly ash content in 2 cm thick sand layers, the corresponding information had to be computed from the simulated data (using final c, s and and d) to find the best fits. Concentration flux and zero concentration gradient were set at the top and bottom, respectively. Obtained soil hydraulic and fly-ash transport parameters were used to simulate water flow and flyash transport when infiltration pulses of 100 or 200 cm3 were applied, to validate obtained results. Large columns The water flow within the sand columns was first analyzed numerically using the data from previous experiments. Since simulated water regime did not closely correspond to the monitored one, a new parameter optimization was performed. Parameter r, s and n were set to values obtained from the Tempe cells experiment (Tab. 3). The and Ks values were estimated using numerical inversion of the soil water contents and pressure heads measured in the large plastic columns using HYDRUS-1D. Initial conditions were set at values measured at the beginning of each test. Observed infiltration rates were used to specify the top boundary conditions. Seepage face boundary conditions were defined at the bottom. Optimization was performed for three infiltration pulses (LPC1, LPC2, and LPC3). Finally, parameters (Tab. 4), which were obtained on small columns to characterize fly-ash transport, were used to simulate fly-ash behavior in the large columns under condition of LPC-FA run. A new optimization was not performed, because measured magnetic susceptibility data could not be directly related to fly-ash content within the sand columns (measurements were also impacted by water content, temperature etc., Kapicka et al., 2011). In addition, measured values represented magnetic susceptibility integrated within the depth of approximately 2.5 cm, which made difficult to correctly analyze magnetic susceptibility at the column top. Two simulations were performed for each sand using either the soil hydraulic parameters used for simulating the small column scenarios or the average soil hydraulic parameters obtained from the large column data. T a b l e 3. Van Genuchten soil hydraulic parameters obtained using the Tempe cells and the RETC program (small glass column ­ SGC), numerical inversion using the HYDRUS-1D program from small glass (SGC) or large plastic column (LPC). Sand Fine ST 06/12 Cylinder/Exp. run SGC3 LPC1 LPC2 LPC3 SGC3 LPC1 LPC2 LPC3 SGC3 LPC1 LPC2 LPC3 s [cm3 cm-3] 0.400 0.400 0.400 0.400 0.399 0.399 0.399 0.399 0.399 0.399 0.399 0.399 r [cm3 cm-3] 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 [cm-1] 0.047 0.063 0.055 0.071 0.077 0.116 0.123 0.185 0.094 0.190 0.178 0.203 n [­] 3 3 3 3 3 3 3 3 3 3 3 3 Ks [cm min-1] 5.30 2.89 10.63 3.43 12.77 20.00 22.88 17.45 29.95 31.91 24.85 30.45 T a b l e 4. Parameters used to simulate fly-ash migration in sands. Sand Bulk density [g cm-3] 1.53 1.54 1.56 Longitudinal dispersivity [cm] 1.5 1.5 1.5 Sand particle diameter [cm] 0.035 0.056 0.093 Fly-ash particle diameter [cm] 0.001 0.001 0.001 Sticking efficiency [­] 5 5 5 First-order detachment coefficient [min-1] 1.5 1.5 1.5 Fine ST 06/12 Results and discussion Small columns The measured final fly-ash distributions in the sand columns packed in the small glass cylinders are shown in Fig. 2. The fly-ash recovery (sum of fly-ash content in sand and suspension discharge at the bottom) was 99.5, 99.7 and 94.9 % (fine sand), 95.2, 99.7 and 99.4 % (medium sand), and 92.5, 92.1 and 95.2 % (coarse sand) of applied dose, for 100, 200 and 300 cm3 of distilled water, respectively. The fly-ash discharge at the bottom was 0, 0 and 0.9 % (fine sand), 0, 0.37 and 0.65 % (fine sand), and 0, 15.9 and 31.9 % (coarse sand) for 100, 200 and 300 cm3 of distilled water, respectively. Results indicated different fly-ash behavior in various sands, as expected. While the fly-ash migrated freely thought the course sand, in the other two sands the fly-ash mobility decreased with decreasing grain sizes, e.g. pore sizes, which are documented on micromorphological images (Fig. 1). This is in agreement with the previous studies of Gargiulo et al. (2007), Torkzaban et al. (2008) and others. The simulated final fly-ash distributions (Fig. 3) satisfactorily fitted observed data for all scenarios. Soil hydraulic parameters obtained from SGC3 scenario (Tab. 3) were sufficient to describe water flow from scenarios SGC1 and SGC2 for each sand. Interestingly the same longitudinal dispersivity, DL, sticking efficiency, , and the detachment coefficient, kd, (Tab. 4) could be used not only for all 3 scenarios but also for all sands. Simulations for various sands differed only in the bulk density, d, and the diameters of the sand grains, dc, (Tab. 4) which affected the attachment coefficients. 211 Fine sand Fly-ash content in 2 cm layer / simulated [mg g ] -1 Medium sand Fly-ash content in 2 cm layer / simulated [mg g-1] Coarse sand Fly-ash content in 2 cm layer / simulated [mg g -1] Fig. 2. Final fly-ash distribution measured in 2 cm thick sand layers and simulated using HYDRUS-1D in fine (top), medium (bottom left) and coarse (bottom right) sands packed in small glass cylinders. Volume magnetic susceptibility [10 -3 SI] -0.25 0 -5 -10 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 Fly-Ash Content [mg g-1] FS depth MS depth -5 -10 -15 -20 -25 -30 -35 -40 CS depth Fine sand Medium sand Coarse sand -15 Fine sand -20 -25 -30 -35 -40 Coarse sand Medium sand Fly-ash content [mg g-1] 0 0 -5 -10 5 10 15 20 25 30 35 40 45 -15 -20 -25 -30 -35 -40 Fine sand Medium sand Coarse sand Fig. 3. Measured final volume magnetic susceptibilities (top left) indicating the maximum fly-ash concentrations (the peak values) and the fly-ash depths, and the final fly-ash distributions simulated using HYDRUS-1D in sands packed in large plastic cylinders using soil hydraulic parameters from the Tempe cells and small glass cylinder data (top right) and from the large plastic cylinder flow data (bottom). Large columns The final distributions of magnetic susceptibility within the sand columns packed in the large plastic cylinders are shown in Fig. 3. Measured magnetic susceptibility is in all cases highest at the top of the sand column due to the high content of fly-ash. As mentioned above, the magnetic susceptibilities were affected not only by the ferrimagnetic particles, but also by higher water content and temperature. However, despite the coincidental impact of all phenomena, the magnetic particles migration is indicated into the following depths: 7 cm in the fine sand, 12 cm in medium and 22 cm in coarse sand. The peak values indicated the depth of the highest fly-ash content within the sand columns. The lowest peak value of magnetic susceptibility was measured for the coarse sand due to the intensive migration of magnetic particles within the sand column. Fly-ash behavior is also documented in the micromorphological images (Fig. 4). Thin sections were made from undisturbed sand samples polluted by fly-ash taken at the columns top after the experiments. While fly-ash migrated without significant restriction through the coarse sandy material (photo not shown), in the other two sands fly-ash is accumulated in few bottle neck pores. However, the flyash mobility was documented in both cases. Fig. 4. Fly-ash distribution in fine (left) and medium (right) sands. The final fly-ash distributions (Fig. 3) simulated using the average soil hydraulic parameters obtained from the large column data corresponded better to the indicated (by magnetic susceptibility) fly-ash distribution compare to that simulated using the parameters from the small column scenarios. Assuming SM400 sensor sensitivity, simulated data satisfactorily approximate the indicated peak values and the depths of fly-ash contamination. In contrast to findings of other authors (summarized in Gargiulo et al., 2008), the model based on the filtration theory satisfactorily approximated observed fly-ash distribution in all cases. According to Gargiulo et al. (2008), recent studies have reported that the classical filtration theory is inadequate to describe colloid and bacteria deposition profiles when an energy barrier exists (unfavorable deposition conditions), and for larger sized colloids and finer texture media. The colloid behavior in porous media strongly depends of on porous media and colloid surface properties. The -potential, which is the most im- portant, was measured for fly-ash suspension using ZetaPALS (Brookhaven Instrument Corporation, USA) and equaled to ­10.08 mV. This value was considerably higher (closer to zero) compare to values measured for ultra fine fly-ash (varied between ­15 and ­40 mV and was lower for higher pH) by Potgieter-Vermaak et al. (2005) and indicates an incipient instability of the colloidal dispersion (for lower absolute value of -potential attraction exceeds repulsion and dispersion flocculates). However, the fly-ash flocculation was not observed during the experiments probably mainly because all experiments were relatively short and suspension flow was fast. The -potential of quartz grains was not measured because its measurement requires a different technique, which was not available. There are several studies, in which the -potential of quartz grains was measured. Johnsor (1999) documented that the -potential (­48 to ­16 mV) increased with the increasing electrolyte ionic strength (10-5 to 10-1 M). Elimelech et al. (2000) showed that the -potential (­24 to ­80 mV) con213 siderably decreased with increasing pH (3 to 9) for 10-3 M NaCl used as a background electrolyte. Saprykina et al. (2010) cited considerably lower absolute value of the -potential (­25.4 mV at pH 7.2). Grain coatings may significantly increase the -potential value (even to the positive values) (Elimelech et al., 2000) and considerably influence (increase) colloid (which is negatively charged) attachment when studying colloid transport in sandy materials (Shani et al., 2008). No quartz grain coatings were expected in our study. The pH of water with sands was measured (7.25) and increased with fly-ash content (11.66, 12.4, 12.28, 12.34 and 12.38 for 10, 25, 50, 75 and 100 mg cm-3, respectively). The other factor influencing the colloid attachment is a colloid and sand grain surface roughness and specific area. Sand grains were not round and smooth as is visible in Fig. 1. All these data and information indicated that conditions were slightly unfavorable for the fly-ash attachment to the sand grain. The numerical model success may be probably again contributed to the relatively simple course of the experiment (fast and short). Flyash was transported mainly during the sand sample saturation. No additional water was used to mobilize fly-ash particles. Conclusions The fly-ash migration within the sand columns of various particle size distributions was studied using different techniques. The experiments proved flyash mobility in studied sands. Visually observed and gravimetrically evaluated fly-ash migration in the small glass cylinders corresponded to the flyash mobility in the large columns detected magnetically using the SM400. Fly-ash mobility strongly depended on the sand particle size distribution. While the fly-ash migrated freely through the coarse sand, in the other two sands the fly-ash migration decreased with the decreasing sand particle sizes. Water flow and fly-ash transport was simulated using the HYDRUS-1D code. Sand particle distribution influenced porous system, hydraulic properties and consequently water flow intensity. Particle sizes and pore water velocity influenced the first-order deposition (attachment) coefficient, which was calculated assuming filtration theory. The other parameters (longitudinal dispersivity, sticking efficiency and the detachment coefficient) characterizing fly-ash transport were the same for all three sands. Simulated fly-ash distribution with214 in the sands columns satisfactorily approximated observed and indicated data. Acknowledgement: Authors acknowledge the financial support of the Grant Agency of Academy of Sciences of the Czech Republic grant No. A300120701, and the Ministry of Education, Youth and Sports grant No. MSM 6046070901. We also thank to Anna Zigová for taking micromorphological images. Our thanks are also to Vladana Ciprýnová for her help with the sand selection and to sand mine Sklopísek Stelec that generously supplied us by 500 kilograms of technical sands. List of symbols e r s h l n m K Ks c s ­ water content [L3 L-3], ­ effective soil water content [­], ­ residual soil water contents [L3 L-3], ­ saturated soil water contents [L3 L-3], ­ pressure head [L], ­ pore-connectivity parameter [­], ­ parameter [L-1], ­ parameter [­], ­ parameter [­], ­ hydraulic conductivity [L T-1], ­ saturated hydraulic conductivity [L T-1], ­ (colloid, virus, bacteria) concentration in the aqueous phase [M L-3], ­ solid phase (colloid, virus, bacteria) concentration [M M-1], ­ bulk density [M L-3], ­ first-order deposition (attachment) coefficient [T-1], ­ first-order entrainment (detachment) coefficient [T-1], ­ colloid retention function [­], ­ diameter of the sand grains [L], ­ diameter of the colloid particles [L], ­ sticking efficiency [­], ­ pore-water velocity [L T-1], ­ single-collector efficiency [­], ­ longitudinal dispersivity [L], ­ final fly-ash content in soil columns [M M-1]. d ka kd dc dco v DL mt http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Hydrology and Hydromechanics de Gruyter

Numerical Simulation of Fly-Ash Transport in Three Sands of Different Particle-Size Distributions Using Hydrus-1d

Loading next page...
 
/lp/de-gruyter/numerical-simulation-of-fly-ash-transport-in-three-sands-of-different-oSAy7qrRNj

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
de Gruyter
Copyright
Copyright © 2011 by the
ISSN
0042-790X
DOI
10.2478/v10098-011-0017-5
Publisher site
See Article on Publisher Site

Abstract

J. Hydrol. Hydromech., 59, 2011, 3, 206­216 DOI: 10.2478/v10098-011-0017-5 RADKA KODESOVÁ1), ALES KAPICKA2), JAKUB LEBEDA1), HANA GRISON2), MARTIN KOCÁREK1), EDUARD PETROVSKÝ2) 1) Department of Soil Science and Soil Protection, Czech University of Life Sciences Prague, Kamýcká 129, 165 21 Prague 6, Czech Republic; Mailto: kodesova@af.czu.cz 2) Institute of Geophysics, Academy of Sciences of the Czech Republic, Bocní 1401, 141 31 Prague 4, Czech Republic. Study is focused on the numerical modeling of fly-ash transport in three sands, which was experimentally studied in the laboratory. Sands were packed in glass cylinders with diameter of 5.52 cm and height of 18 cm. Sands were also packed in plastic cylinders with diameter of 30 cm and height of 40 cm. The fly-ash and pulse infiltrations were applied on the top of all cylinders. Visually observed and gravimetrically evaluated fly-ash migration in small cylinders corresponded to fly-ash mobility in large columns detected using the SM400 Kappameter. The HYDRUS-1D code was used to simulate observed fly-ash transport. Parameters of soil hydraulic functions were either obtained using the Tempe cells and the RETC program or estimated using numerical inversion of transient water flow data measured in both types of columns using HYDRUS-1D. Parameters characterizing colloid transport in sands were then estimated from the final fly-ash distribution in sandy columns using attachment/detachment concept in HYDRUS-1D. Fly-ash mobility increased with increasing sand particle sizes, e.g. pore sizes. Particle sizes and pore water velocity influenced the attachment coefficient, which was calculated assuming filtration theory. The same longitudinal dispersivity, sticking efficiency and detachment coefficient sufficiently characterized fly-ash behavior in all sands. KEY WORDS: Sand, Fly-ash Migration, Magnetic Attachment/Detachment Concept, Filtration Theory. Susceptibility, Numerical Simulation, Radka Kodesová, Ales Kapicka, Jakub Lebeda, Hana Grison, Martin Kocárek, Eduard Petrovský: NUMERICKÁ SIMULACE TRANSPORTU ÚLETOVÉHO POPÍLKU VE TECH PÍSCÍCH RZNÉHO ZRNITOSTNÍHO SLOZENÍ POMOCÍ HYDRUS-1D. J. Hydrol. Hydromech., 59, 3; 45 lit., 4 obr., 4 tab. Studie je zamena na numerické modelování transportu úletového popílku ve tech píscích, který byl experimentáln studován v laboratoi. Písky byly nahutnny ve sklenných válcích o prmru 5,52 cm a výsce 18 cm. Písky byly také nahutnny v plastových válcích o prmru 30 cm a výsce 40 cm. Na povrchu válc byly aplikovány jednorázové infiltrace vody s popílkem. Migrace úletového popílku pozorovaná vizuáln a zjistná gravimetricky v malých válcích odpovídala mobilit úletového popílku detekované Kappametrem SM400 ve velkých válcích. Pozorovaný transport úletového popílku byl simulován programem HYDRUS-1D. Parametry hydraulických funkcí byly získány bu pomocí Tempských cel a programu RETC nebo odhadovány numerickou inverzí transientních data mených na obou typech válc programem HYDRUS-1D. Parametry charakterizující transport koloid v píscích byly potom odhadovány z konecné distribuce úletového popílku v píscitých sloupcích uzitím konceptu attachment/detachment (pipojení/odpojení) v programu HYDRUS-1D. Mobilita úletového popílku se zvysovala se zvysující se velikostí píscitých zrn, tj. s velikostí pór. Velikost zrn a pórová rychlost ovlivnila depozicní (attachment) koeficient, který byl pocítán na základ filtracní teorie. Stejné hodnoty podélné disperze, efektivity blokování (sticking efficiency) a mobilizacního (detachement) koeficientu charakterizovaly chování úletového popílku ve vsech píscích. KLÍCOVÁ SLOVA: písek, migrace úletového popílku, magnetická susceptibilita, numerická simulace, attachment/detachment koncept, filtracní teorie. Introduction Soil and water contamination due to atmospheric deposition is a serious environmental problem especially in the industrial regions. It was shown that magnetic measurements can serve as proxy of industrial immisions as well as heavy-metal contamination (Petrovský and Elwood, 1999; Hanesch et al., 2003; Boyko et al., 2004). The method is based on the fact that industrial fly-ashes contain about 10% of ferrimagnetic particles, namely Fe-oxides (Flanders, 1994; Kapicka et al., 2001). Magnetic susceptibility of soils was mapped in surroundings of local sources of pollution (Kapicka et al., 1999; Hanesch and Scholger, 2002; Spiteri et al., 2005; Sharma and Tripathi, 2007), or at regional to national scales (Hay et al., 1997; Heller et al., 1998; Magiera and Strzyszcz, 2000; Kapicka et al., 2008). Ferrimagnetic particles are usually accumulated at the soil surface. However, they may be transferred into grater depths due to the soil tillage, erosion etc. In some soils (like sandy soils or soils with well developed and stable aggregates) particles may migrate within the soil profile. The soil and groundwater contamination associated with the presence of suspended particles has recently attracted significant attention because of either harmfulness of colloid particles themselves or due to toxic substances adsorbed onto the colloidal particles. Colloid transport is studied in both the macroscopic and microscopic scales. Many studies have been done in microscale using the artificial micromodels to show single particles behavior in pores of specific shapes and under different pore saturation by water via monitoring of particle trajectories or their attachment to solid particles due to the small pore radius, pore blocking by colloids, attachments to other colloid particles, water film covered solid particles or water-air interfacial tension (Sirivithayapakorn and Keller, 2003; Keller et al., 2004; Auset and Keller, 2004). Colloid transport was also experimentally studied at macroscale, such as in laboratory studies of Bolster et al. (1999), Bradford et al. (2002), Gargiulo et al. (2007, 2008), Torkzaban et al. (2008), Foppen et al. (2010) and Schinner et al. (2010) who investigated bacteria mobility in porous columns, or in the field study Wossner et al. (2001) who evaluated the viral transport in sand and gravel aquifer. Colloid transport may be mathematically simulated in microscale (Gao et al., 2010). Several mathematical models describing colloid transport in macroscale have been also designed (Sim and Chrysikopulos, 2000; Bradford et al., 2003; Kim, 2006). The conceptual model for colloid transport included in HYDRUS-1D code (Simnek et al., 2008) accounts for colloid attachment, straining and exclusion. The HYDRUS-1D code was successfully tested and results were presented by Bradford et al. (2003, 2004 and 2005), Jiang et al. (2010) or Gargiulo et al. (2007, 2008). Fly-ash transport in porous media has not been previously studied. Fly-ash was usually used to change properties of original porous material. Among others Zhang et al. (2010) used fly-ash to increase phosphorus sorption in sand. In experimental and mathematical studies mentioned above the maximum colloid-size was 1 m that is 10 times smaller then the average size of fly-ash particles. As showed by Gargiulo et al. (2007), bacteria (R. rhodochrous), transport of which was observed and simulated in tree sands, may create threedimensional aggregates that can reach the size of 10­15 m, while single cells have a diameter of only 0.8­1 m. They suggested that tendency of this bacteria to form aggregates (similar size as the average size of fly-ash particle) enhanced their deposition in the porous material. The main goals of this study were: 1. experimental evaluation of fly-ash mobility in three types of sands, 2. numerical evaluation of such size particles transport within the sand columns using the HYDRUS-1D code, 3. specification of sand vulnerability to fly-ash contamination depending on soil particle-size distribution. Material and methods Experimental data Small columns Three technical quartz sands (composition, is documented in thin section images, Fig. 1) from the sand mine Sklopísek Stelec were used to study flyash mobility in porous media: ST8 (particle sizes 0.10­0.63 mm) ­ fine sand, ST03/08 (particle sizes 0.315­0.80 mm) ­ medium sand, and ST 06/12 (particle sizes 0.63­1.25) ­ coarse sand. Sands were packed in glass cylinders with diameter of 5.52 cm and heighs of 18 cm. Two grams of fly-ash and 100 cm3 of distilled water were applied on the top of the sand columns (SGC1 run). Suspension infiltration and outflow at the column bottom was monitored. The typical breakthrough curve was not measured. The small column experiments were performed to 207 validate fly-ash distribution in large soil columns described using the magnetic susceptibility data (see below) and to obtain hydraulic and transport parameters under similar conditions. The fly-ash migration was assessed visually. Then the soil columns were divided into the 2 cm thick layers. Flyash was washed out from each sand layer by 100 cm3 of distilled water and after that water was extracted using the evaporation table. The amount of fly-ash in each sand layer was determined gravimetrically. To eliminate possible impact of very fine particles presented in the sands, approximately the same amounts (3 replicates) as in each layer of clean sands were also washed by 100 cm3 of distilled water and water was again extracted using the evaporation table. Detected amount of fine particles were subtracted from values determined for column layers. Fly ash content in soil column was expressed as a total fly-ash mass per unit of dry sand, mt [M M-1]. Experiments were repeated with the same amount of fly-ash, and 200 or 300 cm3 of distilled water (SGC2 and SGC2 runs) (Tab. 1). Fig. 1. Sand morphology and pore structure studied on micro-morphological images of fine (top left), medium (top right) and coarse (bottom) sands. T a b l e 1. Characteristics of experiments performed in sands packed in small glass columns (SGC): inflow (Qinf) and outflow (Qout) volume, inflow (cinf) and outflow (cout) concentration, time of infiltration pulse (tinf), time of outflow beginning (tout,ini), total simulated time (ttot). Sand Fine ST 06/12 Exp. run SGC1 SGC2 SGC3 SGC1 SGC2 SGC3 SGC1 SGC2 SGC3 Qinf [cm3] 100 200 300 100 200 300 100 200 300 cinf [mg cm-3] 20 10 6.7 20 10 6.7 20 10 6.7 tinf [min] 0.40 1.37 2.78 0.19 0.52 1.12 0.17 0.29 0.42 Qout [cm3] 0 45.9 145.4 0 67.0 167.7 0 83.7 183.0 cout [mg cm-3] ­ 0 0.123 ­ 0.111 0.077 ­ 3.80 2.77 tout,ini [min] ­ 1.07 1.05 ­ 0.47 0.43 ­ 0.25 0.22 ttot [min] 5 5 5 3 3 3 2 2 2 Large columns Sands were also packed in plastic cylinders with diameter of 30 cm and height of 40 cm. The 5 cm coarse sand drainage layers were formed below the studied sand layers. The sensors SM200 (Delta-T Devices Ltd., 2005) for soil-water content measurements and micro-tensiometers Tensior 5 (UMS GmbH, Munich, 2005) for pressure head measurements were placed 10, 20 and 30 cm below the sand surface to monitor water regime within the sand columns. The plastic tube for inserting the Kappameter SM400 (Petrovský at al., 2004) was positioned in the column center. The SM400 device, which measures vertical distribution of magnetic susceptibility, was used to monitor migration of the fly-ash ferrimagnetic particles. The water regime was initially studied on each soil column using the ponding infiltration (maximum ponding depth of 5 cm) on the column top. Three different fresh water infiltration pulses were applied in different experimental runs (LPC1, LPC2, LPC3) (Tab. 2). Next the fly-ash (emitted from Porici power plant) and fresh water was applied (LPC-FA run). Time intervals between experiments were 2 to 4 weeks to insure similar initial conditions. The applied fly-ash doses varied for different sands due to the sensitivity of the SM400 device. The applied fly-ash amount increased with expected increase of fly-ash mobility in each sand material: 50 g on the fine, 100 g on the medium and 150 g on the coarse sand. Water regime was monitored using the SM200 sensors and micro-tensiometers Tensior 5. The final magnetic susceptibility within the soil profile and consequently fly-ash distribution was measured when no significant water flow and consequently no fly-ash transport was observed. T a b l e 2. Characteristics of experiments perform in sands packed in large plastic columns (LPC): inflow volume (Qinf), inflow concentration (cinf), time of infiltration pulse (tinf), total simulated time (ttot). Sand Fine ST 06/12 Exp. run LPC1 LPC2 LPC3 LPC-FA LPC1 LPC2 LPC3 LPC-FA LPC1 LPC2 LPC3 LPC-FA Qinf [cm3] 4700 8910 15850 3700 5350 9780 10430 3120 4950 16750 17110 2100 cinf [mg cm-3] ­ ­ ­ 13.51 ­ ­ ­ 32.05 ­ ­ ­ 71.43 tinf [min] 1.67 3.88 7.50 1.31 0.68 1.28 1.42 0.40 0.35 1.22 1.25 0.15 ttot [min] 40 60 70 20 50 60 70 10 30 40 75 10 Numerical simulations Theory The HYDRUS-1D program (Simnek et al., 2008a,b) was applied to simulate water flow and fly-ash transport in sands. The Richards equation, which describes the one-dimensional isothermal Darcian flow in a variably saturated rigid porous medium, is used in this model. To solve this equation, the soil hydraulic properties must be specified. The van Genuchten (1980) analytical expressions are used to describe soil hydraulic functions, the soil water retention curve, (h), and the hydraulic conductivity function, K(): e= (h) - r = s -r ( 1+| h| ) , h<0 e = 1, h0 (1) m K( ) = K s el 1 - ( 1 - 1/ m ) , e h<0 (2) K( ) = K s, h0 where is the soil water content [L3 L-3], e ­ the effective soil water content [­], r and s are the residual and saturated soil water contents [L3 L-3], respectively, h ­ the pressure head [L], l ­ the poreconnectivity parameter [­], ­ the reciprocal of the air-entry pressure head [L-1], n [­] is related to the 209 slope of the retention curve at the inflection point, and m = 1 - 1/n, K ­ the hydraulic conductivity [L T-1], Ks ­ the saturated hydraulic conductivity [L T-1]. Parameters r, s, , n and Ks must be known when simulating water regime. A modified form of the advection-dispersion equation is employed in HYDRUS-1D to simulate virus, colloid, and bacteria transport assuming attachment/detachment concept. Mass transfer between the aqueous and solid phase is described using following equation: s = ka c - kd s, t (3) where c is the (colloid, virus, bacteria) concentration in the aqueous phase [M L-3], s ­ the solid phase (colloid, virus, bacteria) concentration [M M-1], d ­ the bulk density [M L-3], ka ­ the first-order deposition (attachment) coefficient [T-1], kd ­ the first-order entrainment (detachment) coefficient [T-1], and is a colloid retention function [­], which is function of the diameter of the sand grain, dc [L], and coordinate. The attachment coefficient may be calculated using filtration theory as a function of the soil water content, diameter of the sand grains, dc [L], diameter of the colloid particles, dco [L], sticking efficiency, [­], pore-water velocity v [L T-1], and the single-collector efficiency, [­]. Theory is in greater detail described in HYDRUS1D manual (Simnek et al., 2008a). Parameters d, dc, dco, , and kd must be known when simulating colloid transport. In addition the longitudinal dispersivity, DL [L], must be defined to solve the advection-dispersion equation. Small columns The water flow within the sand columns was initially numerically analyzed. Soil water retention curves were measured using the Tempe cells. Sands were packed into 100-cm3 cylinders (soil core height of 5.1 cm and diameter of 5.0 cm) placed into the cells (3 replicates for each sand). Initially, fully saturated soil samples were slowly drained using six pressure head steps (­10, ­30, ­50, ­100, ­200 and -350 cm) during a 2-week period. Data points of soil water retention curve were evaluated using the final soil water content and water outflow monitored using byrettes. The RETC program (van Genuchten et al., 2005) was used to obtain r, s, and n parameters of van Genuchten function (Tab. 3) by fitting data of all 3 replicates (which were 210 very similar). Parameters r were set at 0.01 cm3 cm-3 in all cases. Parameter n was restricted to be maximally 3 (larger values caused computational instabilities when simulating some of following water and flay-ash transport scenarios). The saturated hydraulic conductivities, Ks, were estimated using numerical inversion of outflow data measured on small glass columns (SGC3 ­ infiltration pulse of 300 cm3) using HYDRUS-1D. Initial conditions were set at low pressure head values of ­500 cm corresponding to very dry conditions. Observed infiltration rates were used to specify the top boundary conditions. Seepage face boundary conditions were defined at the bottom. Next, the fly-ash transport was simulated. The bulk density, d, diameters of the sand grains, dc, and diameter of the colloid particles, dco, were set to measured values (Tab. 4). The longitudinal dispersivity, DL, sticking efficiency, , and the detachment coefficient, kd, were calibrated manually from the final fly-ash content, mt, in soil columns. Since mt values expressed the average total fly ash content in 2 cm thick sand layers, the corresponding information had to be computed from the simulated data (using final c, s and and d) to find the best fits. Concentration flux and zero concentration gradient were set at the top and bottom, respectively. Obtained soil hydraulic and fly-ash transport parameters were used to simulate water flow and flyash transport when infiltration pulses of 100 or 200 cm3 were applied, to validate obtained results. Large columns The water flow within the sand columns was first analyzed numerically using the data from previous experiments. Since simulated water regime did not closely correspond to the monitored one, a new parameter optimization was performed. Parameter r, s and n were set to values obtained from the Tempe cells experiment (Tab. 3). The and Ks values were estimated using numerical inversion of the soil water contents and pressure heads measured in the large plastic columns using HYDRUS-1D. Initial conditions were set at values measured at the beginning of each test. Observed infiltration rates were used to specify the top boundary conditions. Seepage face boundary conditions were defined at the bottom. Optimization was performed for three infiltration pulses (LPC1, LPC2, and LPC3). Finally, parameters (Tab. 4), which were obtained on small columns to characterize fly-ash transport, were used to simulate fly-ash behavior in the large columns under condition of LPC-FA run. A new optimization was not performed, because measured magnetic susceptibility data could not be directly related to fly-ash content within the sand columns (measurements were also impacted by water content, temperature etc., Kapicka et al., 2011). In addition, measured values represented magnetic susceptibility integrated within the depth of approximately 2.5 cm, which made difficult to correctly analyze magnetic susceptibility at the column top. Two simulations were performed for each sand using either the soil hydraulic parameters used for simulating the small column scenarios or the average soil hydraulic parameters obtained from the large column data. T a b l e 3. Van Genuchten soil hydraulic parameters obtained using the Tempe cells and the RETC program (small glass column ­ SGC), numerical inversion using the HYDRUS-1D program from small glass (SGC) or large plastic column (LPC). Sand Fine ST 06/12 Cylinder/Exp. run SGC3 LPC1 LPC2 LPC3 SGC3 LPC1 LPC2 LPC3 SGC3 LPC1 LPC2 LPC3 s [cm3 cm-3] 0.400 0.400 0.400 0.400 0.399 0.399 0.399 0.399 0.399 0.399 0.399 0.399 r [cm3 cm-3] 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 [cm-1] 0.047 0.063 0.055 0.071 0.077 0.116 0.123 0.185 0.094 0.190 0.178 0.203 n [­] 3 3 3 3 3 3 3 3 3 3 3 3 Ks [cm min-1] 5.30 2.89 10.63 3.43 12.77 20.00 22.88 17.45 29.95 31.91 24.85 30.45 T a b l e 4. Parameters used to simulate fly-ash migration in sands. Sand Bulk density [g cm-3] 1.53 1.54 1.56 Longitudinal dispersivity [cm] 1.5 1.5 1.5 Sand particle diameter [cm] 0.035 0.056 0.093 Fly-ash particle diameter [cm] 0.001 0.001 0.001 Sticking efficiency [­] 5 5 5 First-order detachment coefficient [min-1] 1.5 1.5 1.5 Fine ST 06/12 Results and discussion Small columns The measured final fly-ash distributions in the sand columns packed in the small glass cylinders are shown in Fig. 2. The fly-ash recovery (sum of fly-ash content in sand and suspension discharge at the bottom) was 99.5, 99.7 and 94.9 % (fine sand), 95.2, 99.7 and 99.4 % (medium sand), and 92.5, 92.1 and 95.2 % (coarse sand) of applied dose, for 100, 200 and 300 cm3 of distilled water, respectively. The fly-ash discharge at the bottom was 0, 0 and 0.9 % (fine sand), 0, 0.37 and 0.65 % (fine sand), and 0, 15.9 and 31.9 % (coarse sand) for 100, 200 and 300 cm3 of distilled water, respectively. Results indicated different fly-ash behavior in various sands, as expected. While the fly-ash migrated freely thought the course sand, in the other two sands the fly-ash mobility decreased with decreasing grain sizes, e.g. pore sizes, which are documented on micromorphological images (Fig. 1). This is in agreement with the previous studies of Gargiulo et al. (2007), Torkzaban et al. (2008) and others. The simulated final fly-ash distributions (Fig. 3) satisfactorily fitted observed data for all scenarios. Soil hydraulic parameters obtained from SGC3 scenario (Tab. 3) were sufficient to describe water flow from scenarios SGC1 and SGC2 for each sand. Interestingly the same longitudinal dispersivity, DL, sticking efficiency, , and the detachment coefficient, kd, (Tab. 4) could be used not only for all 3 scenarios but also for all sands. Simulations for various sands differed only in the bulk density, d, and the diameters of the sand grains, dc, (Tab. 4) which affected the attachment coefficients. 211 Fine sand Fly-ash content in 2 cm layer / simulated [mg g ] -1 Medium sand Fly-ash content in 2 cm layer / simulated [mg g-1] Coarse sand Fly-ash content in 2 cm layer / simulated [mg g -1] Fig. 2. Final fly-ash distribution measured in 2 cm thick sand layers and simulated using HYDRUS-1D in fine (top), medium (bottom left) and coarse (bottom right) sands packed in small glass cylinders. Volume magnetic susceptibility [10 -3 SI] -0.25 0 -5 -10 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 Fly-Ash Content [mg g-1] FS depth MS depth -5 -10 -15 -20 -25 -30 -35 -40 CS depth Fine sand Medium sand Coarse sand -15 Fine sand -20 -25 -30 -35 -40 Coarse sand Medium sand Fly-ash content [mg g-1] 0 0 -5 -10 5 10 15 20 25 30 35 40 45 -15 -20 -25 -30 -35 -40 Fine sand Medium sand Coarse sand Fig. 3. Measured final volume magnetic susceptibilities (top left) indicating the maximum fly-ash concentrations (the peak values) and the fly-ash depths, and the final fly-ash distributions simulated using HYDRUS-1D in sands packed in large plastic cylinders using soil hydraulic parameters from the Tempe cells and small glass cylinder data (top right) and from the large plastic cylinder flow data (bottom). Large columns The final distributions of magnetic susceptibility within the sand columns packed in the large plastic cylinders are shown in Fig. 3. Measured magnetic susceptibility is in all cases highest at the top of the sand column due to the high content of fly-ash. As mentioned above, the magnetic susceptibilities were affected not only by the ferrimagnetic particles, but also by higher water content and temperature. However, despite the coincidental impact of all phenomena, the magnetic particles migration is indicated into the following depths: 7 cm in the fine sand, 12 cm in medium and 22 cm in coarse sand. The peak values indicated the depth of the highest fly-ash content within the sand columns. The lowest peak value of magnetic susceptibility was measured for the coarse sand due to the intensive migration of magnetic particles within the sand column. Fly-ash behavior is also documented in the micromorphological images (Fig. 4). Thin sections were made from undisturbed sand samples polluted by fly-ash taken at the columns top after the experiments. While fly-ash migrated without significant restriction through the coarse sandy material (photo not shown), in the other two sands fly-ash is accumulated in few bottle neck pores. However, the flyash mobility was documented in both cases. Fig. 4. Fly-ash distribution in fine (left) and medium (right) sands. The final fly-ash distributions (Fig. 3) simulated using the average soil hydraulic parameters obtained from the large column data corresponded better to the indicated (by magnetic susceptibility) fly-ash distribution compare to that simulated using the parameters from the small column scenarios. Assuming SM400 sensor sensitivity, simulated data satisfactorily approximate the indicated peak values and the depths of fly-ash contamination. In contrast to findings of other authors (summarized in Gargiulo et al., 2008), the model based on the filtration theory satisfactorily approximated observed fly-ash distribution in all cases. According to Gargiulo et al. (2008), recent studies have reported that the classical filtration theory is inadequate to describe colloid and bacteria deposition profiles when an energy barrier exists (unfavorable deposition conditions), and for larger sized colloids and finer texture media. The colloid behavior in porous media strongly depends of on porous media and colloid surface properties. The -potential, which is the most im- portant, was measured for fly-ash suspension using ZetaPALS (Brookhaven Instrument Corporation, USA) and equaled to ­10.08 mV. This value was considerably higher (closer to zero) compare to values measured for ultra fine fly-ash (varied between ­15 and ­40 mV and was lower for higher pH) by Potgieter-Vermaak et al. (2005) and indicates an incipient instability of the colloidal dispersion (for lower absolute value of -potential attraction exceeds repulsion and dispersion flocculates). However, the fly-ash flocculation was not observed during the experiments probably mainly because all experiments were relatively short and suspension flow was fast. The -potential of quartz grains was not measured because its measurement requires a different technique, which was not available. There are several studies, in which the -potential of quartz grains was measured. Johnsor (1999) documented that the -potential (­48 to ­16 mV) increased with the increasing electrolyte ionic strength (10-5 to 10-1 M). Elimelech et al. (2000) showed that the -potential (­24 to ­80 mV) con213 siderably decreased with increasing pH (3 to 9) for 10-3 M NaCl used as a background electrolyte. Saprykina et al. (2010) cited considerably lower absolute value of the -potential (­25.4 mV at pH 7.2). Grain coatings may significantly increase the -potential value (even to the positive values) (Elimelech et al., 2000) and considerably influence (increase) colloid (which is negatively charged) attachment when studying colloid transport in sandy materials (Shani et al., 2008). No quartz grain coatings were expected in our study. The pH of water with sands was measured (7.25) and increased with fly-ash content (11.66, 12.4, 12.28, 12.34 and 12.38 for 10, 25, 50, 75 and 100 mg cm-3, respectively). The other factor influencing the colloid attachment is a colloid and sand grain surface roughness and specific area. Sand grains were not round and smooth as is visible in Fig. 1. All these data and information indicated that conditions were slightly unfavorable for the fly-ash attachment to the sand grain. The numerical model success may be probably again contributed to the relatively simple course of the experiment (fast and short). Flyash was transported mainly during the sand sample saturation. No additional water was used to mobilize fly-ash particles. Conclusions The fly-ash migration within the sand columns of various particle size distributions was studied using different techniques. The experiments proved flyash mobility in studied sands. Visually observed and gravimetrically evaluated fly-ash migration in the small glass cylinders corresponded to the flyash mobility in the large columns detected magnetically using the SM400. Fly-ash mobility strongly depended on the sand particle size distribution. While the fly-ash migrated freely through the coarse sand, in the other two sands the fly-ash migration decreased with the decreasing sand particle sizes. Water flow and fly-ash transport was simulated using the HYDRUS-1D code. Sand particle distribution influenced porous system, hydraulic properties and consequently water flow intensity. Particle sizes and pore water velocity influenced the first-order deposition (attachment) coefficient, which was calculated assuming filtration theory. The other parameters (longitudinal dispersivity, sticking efficiency and the detachment coefficient) characterizing fly-ash transport were the same for all three sands. Simulated fly-ash distribution with214 in the sands columns satisfactorily approximated observed and indicated data. Acknowledgement: Authors acknowledge the financial support of the Grant Agency of Academy of Sciences of the Czech Republic grant No. A300120701, and the Ministry of Education, Youth and Sports grant No. MSM 6046070901. We also thank to Anna Zigová for taking micromorphological images. Our thanks are also to Vladana Ciprýnová for her help with the sand selection and to sand mine Sklopísek Stelec that generously supplied us by 500 kilograms of technical sands. List of symbols e r s h l n m K Ks c s ­ water content [L3 L-3], ­ effective soil water content [­], ­ residual soil water contents [L3 L-3], ­ saturated soil water contents [L3 L-3], ­ pressure head [L], ­ pore-connectivity parameter [­], ­ parameter [L-1], ­ parameter [­], ­ parameter [­], ­ hydraulic conductivity [L T-1], ­ saturated hydraulic conductivity [L T-1], ­ (colloid, virus, bacteria) concentration in the aqueous phase [M L-3], ­ solid phase (colloid, virus, bacteria) concentration [M M-1], ­ bulk density [M L-3], ­ first-order deposition (attachment) coefficient [T-1], ­ first-order entrainment (detachment) coefficient [T-1], ­ colloid retention function [­], ­ diameter of the sand grains [L], ­ diameter of the colloid particles [L], ­ sticking efficiency [­], ­ pore-water velocity [L T-1], ­ single-collector efficiency [­], ­ longitudinal dispersivity [L], ­ final fly-ash content in soil columns [M M-1]. d ka kd dc dco v DL mt

Journal

Journal of Hydrology and Hydromechanicsde Gruyter

Published: Sep 1, 2011

References