Particles, Pathogens, and Porous Media: Physics, Chemistry, and Engineering Desmond F. Lawler University of Texas Assessing Pathogen Fate, Transport, and Risk in Natural and Engineered Water Treatment Banff Centre, Banff, Alberta September 23, 2012 [email protected] 512 471 4595 Conventional Drinking Water Treatment Plant Effluent Concentration Typical Filter Run Maximum allowable concentration Head Loss B c Max. allowable head loss B h Time or Cumulative Volume Throughput Filtration Complexity • Two dependent variables of importance – Head Loss – Effluent Quality • Never at Steady State • Two different modes of operation (filtration and backwashing) • Numerous Independent Variables Independent Variables in Filtration • Time (or Cumulative Vol. Throughput since last backwash) • Particles: size distribution, concentration, density, shape, surface charge, surface chemistry • Water (physical): Temp, and therefore viscosity and density • Water (chemical): effects on surface chemistry of media and particles • Media: Size, size distribution, density, shape, porosity, depth, and surface chemistry • Operation: filtration velocity (Q/A) Take Home Message (THM) #1 Chemistry matters (a lot) O’Melia, C.R. (1974) “The Role of Polyelectrolytes in Filtration Processes,” Report # EPA-670/274-032, U.S. Environmental Protection Agency, Cincinnati, OH. Alum Dose (mg/L) 50 2.5 2.5 0 1 3 Turbidity (NTU) 40 30 20 10 0 0 60 120 180 Time (min) 240 300 THM #2 Particle size matters Flow Sedimentation Interception Brownian Motion Media Grain Streamlines Yao, K.M., Habibian, M.T., and O'Melia, C.R. (1971) "Water and Wastewater Filtration: Concepts and Applications," Env. Sci. and Tech., 5, 11, 1105-1112 Yao, Habibian, and O'Melia Model Sedimentation Interception Brownian Motion Interception : 3 dp ηI = 2 dc Se dim entation : ηG = 2 (ρp − ρl ) g d p 2 18 µ v Media Grain Streamlines kT Brownian Motion : ηB = 0.9 µd d v p c η = ηI + ηG + ηB N out 3(1 − ε )αηL ln =− 2d c N in 2 3 Tot Single collector removal efficiency log η -1 -2 -3 Conventional rapid sand filter Yao et al. model -4 -5 -6 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (d in µm) p 1.0 1.5 Tufenkji and Elimelech Model η Br = 2.4 As Pe −0.715 N R −0.081 N vdW 0.052 1 3 η I = 0.55 As N R 1.675 N A 0.125 η g = 0.22 N R NR = NA = −0.24 NG 1.11 dp As dc AH 2 3πµd p vo NG = N vdW = ( ρ p − ρ L ) gd p 18µv o AH k BT N vdw 0.053 Tufenkji, N. and Elimelech, M. (2004) “Correlation Equation for Predicting Single-Collector Efficiency in Physicochemical Filtration in Saturated Porous Media,” Env. Sci. and Tech., 38, 2, 529-536. 2(1 − p 5 ) = 2 − 3 p + 3 p5 − 2 p6 1 a = (1 − ε ) 3 b vd Pe = o c Dp p = 2 Dp = k BT 3πµd p Tot Single collector removal efficiency log η -1 -2 -3 Conventional rapid sand filter Tufenkji & Elimelech model -4 -5 -6 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (dp in µm) 1.0 1.5 Overall filter removal efficiency Conventional rapid sand filter Tufenkji & Elimelech model 1.0 0.8 α =1 0.6 α = 0.1 0.4 0.2 α = 0.01 0.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (d in µm) p 1.0 1.5 Removal Efficiency, η filter 1.0 d = 0.6 mm c 0.8 v = 0.14 cm/s o 0.6 ρp = 1.08 g/cm3 0.4 T = 25 C α=1 L = 60 cm o -13 H = 1 x 10 a Yao Rajagopalan & Tien Tufenkji & Elimelech 0.2 0.0 -1.0 erg -0.5 0.0 0.5 1.0 1.5 log of Particle Diameter (d in µm) p 2.0 Assumptions about Pathogens • In terms of their behavior, pathogens are like any other particles in a filter, with the critical characteristics being size and surface chemistry. • Surface Chemistry: • Cysts of Giardia and Cryptosporidium are coated with Natural Organic Matter and are therefore negatively charged from that adsorbed material. • Bacteria and viruses are intrinsically negatively charged due to functional groups that are part of their surfaces. Pathogen size • Viruses: 5 – 100 nm; choose 10 nm as most common; log dp = -2.0 • Bacteria: 0.7 to 2.5 µm; choose 1 µm; log dp = 0.0 • Cryptosporidium: 2 – 6 µm; choose 3 µm; log dp = 0.5 • Giardia: 7 to 14 µm; choose 10 µm; log dp = 1.0 Pathogen Shapes • Viruses: Often highly non-spherical, with varying types of appendages • Bacteria: Sometimes nearly spherical, but a variety of shapes (and appendages) • Cysts and Oocysts: Reasonably spherical 10 Influent -3 -1 (∆N/∆dp in cm µm ) log of particle size distributiojn funciton Unripened sand filter 8 6 Effluent 4 2 0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (dp in µm) 1.0 1.5 THMs #3 and #4 • Ripening happens • Detachment happens Moran, D.C., Moran, M.C., Cushing, R.S., and Lawler, D.F. (1993) “Particle Behavior in Deep-Bed Filtration: Part 1-Ripening and Breakthrough,” J. Amer. Water Works Assoc., 85, 12, 69-81. Moran, M.C., Moran, D.C., Cushing, R.S., and Lawler, D.F. (1993) “Particle Behavior in Deep Bed Filtration: Part 2-Particle Detachment,” J. Amer. Water Works Assoc., 85, 12, 82-93. Gear Pump Fluid S ampling Ports A Data Acquisition S ystem C F Peristaltic Pump G Pressure S ampling Ports Filter Effluent 3 (∆N/∆d in #/cm -µm) Exp. SM4 v = 1.8 mm/s d = 1.85 mm 5.0 4.0 p c Time = 17 min. Port A Influent 3.0 Port C 193 mm 2.0 1.0 p LOG(∆N/∆d ) PARTICLE SIZE DISTRIBUTION FUNCTION 6.0 Port G 746 mm 0.0 -1.0 A -2.0 0.0 0.4 0.8 1.2 Log of Particle Diameter (d in µm) p 1.6 REMOVAL EFFICIENCY (Number Basis) Effects of Media Size: Initial Removal 1.0 0.8 0.6 Depth = 746 mm dc=1.85 mm; dc=1.85 mm; dc=0.78 mm; dc=0.78 mm; 0.4 0.2 Exp. SM4 Model Exp. SM5 Model 0.0 0.0 0.2 0.4 0.6 0.8 1.0 in LOG OF PARTICLE DIAMETER (dµm) p 1.2 Effects of Media Size Depth = 746 mm REMOVAL EFFICIENCY (NUMBER BASIS) Depth = 193 mm 1.0 A2 A1 0.8 0.6 dc = 1.85 mm; Exp. SM4 dc = 0.78 mm; Exp. SM5 dc = 1.85 mm; Exp. SM4 dc = 0.78 mm; Exp. SM5 0.4 0.2 Window 1 (1.1 µm < d p < 1.8 µm) Window 1 0.0 0 600 1200 1800 TIME (minutes) 2400 3000 0 600 1200 1800 2400 TIME (minutes) 3000 REMOVAL EFFICIENCY (Numer Basis) Depth = 193 mm Depth = 746 mm 1.0 B2 B1 0.8 0.6 0.4 0.2 Window 3 (3.0 µm< dp < 4.8 µm) Window 3 0.0 0 600 1200 1800 TIME (minutes) 2400 30000 600 1200 1800 TIME (minutes) 2400 3000 According to Yao, for equal removal in two filters, when removal is dominated by sedimentation: L1 L2 = v1 d c1 v 2 d c2 When the two fillters have the same filtration velocity but different media sizes: L1 L2 = d c1 d c2 1.0 REMOVAL EFFICIENCY (Number Basis) A B 0.8 0.6 Window 1 (1.1 µm < d p < 1.8 µm) 0.4 dc = 1.85 mm; 416 mm; Exp. SM4 dc = 0.78 mm; 193 mm; Exp. SM5 0.2 Window 3 (3.0 µm < d p < 4.8 µm) 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 CUM. SURFACE AREA CAPTURED (10 12 0.0 0.5 1.0 1.5 2.5 3.0 CUM. SURFACE AREA CAPTURED 2 µm ) 2.0 (10 12 2 µm ) ADDITIONAL HEAD LOSS (cm) 50 dc =1.85; 416 mm; SM4 dc =0.78; 193 mm; SM5 40 30 20 10 0 -10 0 0.5 1 1.5 2 2.5 3 CUMULATIVE SURFACE AREA CAPTURED µm ( 2) 3 (∆N/∆d in #/cm -µm) 5 4 p p LOG(∆N/∆d ) PARTICLE SIZE DISTRIBUTION FUNCTIO 6 3 2 Port A Influent Port G 17 min 1 0 Port G 66 min -1 -2 0.0 0.4 0.8 1.2 Log of Particle Diameter (d in µm) p 1.6 p 3 (∆N/∆d in #/cm -µm) 5 4 3 2 1 p LOG(∆N/∆d ) PARTICLE SIZE DISTRIBUTION FUNCTION 6 0 Port A Influent Port G 66 min Port G 2682 min -1 -2 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 LOG OF PARTICLE DIAMETER (d in µm) p p 3 (∆N/∆d in #/cm -µm) 6.0 5.0 A 4.0 Exp. SM3 v = 5.5 mm/s d = 0.78 mm 3.0 2.0 1.0 p LOG(∆N/∆d ) PARTICLE SIZE DISTRIBUTION FUNCTIO Special Detachment Experiments c Port A Influent (Filtered effluent) 0.0 -1.0 -2.0 -0.2 0.0 Time = 17 min. Port C 193 mm (16 min. after influent switch) 0.2 0.4 0.6 0.8 1.0 1.2 Log of Particle Diameter (d in µm) p 1.4 EFFLUENT/INFLUENT (Number Basis) 30 06 min. 29 min. 43 min. 78 min. 25 20 Top 193 mm 15 10 5 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 of Particle Diameter (din µm) Log ofLog Particle Diameter (d p inµm) p 1.4 Application of Characteristic Reaction Time Concept to Filtration Lawler, D.F., and Nason, J.A., “Granular Media Filtration: Old Process, New Thoughts,” Water Science and Technology, 53, 7, 1-7, 2006. • In batch system, time that it would take for a constituent to reach its ultimate value, if the reaction continued at its initial rate. • For negative reaction rate: c(0) tchar = − rc (0 ) Concentration Definition: Characteristic Reaction Time Concentration Slope = Initial Rate t char Time Characteristic Reaction Time for First Order Reactions dc rc = = −k1c dt c(t ) = c(0) exp( −k1t ) c0 Concentration • In batch system, time for a constituent to reach 1/e of its original value. Concentration c0 e tchar tchar 1 = k1 Time Filtration “Reaction” • Removal occurs over space, not time • Earliest model from Iwasaki, 1937: dN = −λN dx N (x ) = N in exp(− λx ) or, in full filter : N out L = N in exp(− λL ) = N in exp − Lchar • Characteristic Removal Length = Lchar = 1/λ Definition: Characteristic Removal Length 3 (1 − ε )αη N out = exp − L N in 2 dc Lchar dc 2 = 3 (1 − ε )αη 3 (1 − ε )αη N out L L = exp − = exp − N in dc 2 Lchar Decisions • Characteristic removal length: Lchar dc 2 = 3 (1 − ε )αη • Choice of particle size (η = f ( d p ) ) – For monomedia wastewater filters: dp = 7 µm – For monomedia drinking water filters: dp = 1 µm – For dual media filters: • anthracite dp = 7 µm • sand dp = 1 µm • For each particle size, choice of N out N in • For dual media filters, relative sizes of the two media Removal Efficiency, η filter 1.0 d = 0.6 mm c 0.8 v = 5 m/h 0 0.6 ρp = 1.08 g/cm3 0.4 T = 298 K α=1 L = 60 cm 0.2 0.0 -1.0 Yao Rajagopalan & Tien Tufenkji & Elimelech -0.5 0.0 0.5 A = 1 x 10 -13 H 1.0 1.5 log of Particle Diameter (d in µm) p erg 2.0 10 Volume Distribution (∆V/∆log dp in 106 µm3 cm-3) Alum Sweep Floc 8 0 min 14 min 6 4 2 0 -0.25 0 0.25 0.5 0.75 1 1.25 1.5 Log of Particle Diameter (dp in µm) Data from Multimedia Alum Plants Plant Glendale, AZ California 1 Corvallis, OR Spartanburg, SC Phoenix, AZ Oceanside, CA Oakland, CA Media Type Sand Anthracite Sand Anthracite Sand Anthracite Sand Anthracite Sand Anthracite Sand Anthracite Sand Anthracite Effective Size (mm) 0.50 0.90 0.60 1.15 0.40 1.05 0.50 1.00 0.46 0.95 0.40 1.00 0.50 0.90 Depth (cm) 35.6 61.0 30.5 76.2 26.7 49.5 33.0 33.0 25.4 50.8 25.4 45.7 30.5 45.7 L/dc Filtration ηsand Total Velocity (1 µm) (--) (m/h) ηanth (7 µm) 1389 11.0 26% 35% 1171 14.7 14% 24% 1138 18.3 26% 19% 991 14.7 22% 16% 1087 14.7 20% 26% 1092 12.2 28% 22% 1118 12.2 22% 27% Dual media, choose Media Size (mm) d c ,anth = 2d c ,sand Settling velocity (cm/s) Sand Anthracite Sand Anthracite 0.4 0.5 0.8 1.0 5.89 7.59 5.52 7.04 0.6 0.7 1.2 1.4 9.28 10.94 8.51 9.93 g cm 3 g = 1.50 3 cm ρ sand = 2.65 ρ anth (cm) req Required Filter Depth, L 250 d c,anth = 2d v = 10 m/h c,sand 0 200 150 Anthracite 100 Sand 50 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Media Size, d (mm) c 1.6 (cm) req Required Filter Depth, L 250 d c,anth 200 = 2d v = 10 m/h c,sand 0 Total 150 Anthracite 100 Sand 50 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Media Size, d (mm) c 1.6 Required Filter Depth, L req (cm) Dual Media Filter Design: Velocity 250 200 20 15 10 150 5 d c,anth = 2d c,sand v (m/h) 0 100 50 0 0.3 0.4 0.5 0.6 0.7 Sand Size, d (mm) c,sand 0.8 Required Filter Depth, L req (cm) Dual Media Filter Design 250 200 20 15 10 150 5 d c,anth = 2d c,sand v (m/h) 0 100 L/d = 1000 c 50 0 0.3 0.4 0.5 0.6 0.7 Sand Size, d (mm) c,sand 0.8 Required Filter Depth, L req (cm) Current Design Guidelines 500 400 300 200 1500 L/d = 2000 1300 1000 c 100 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Media Size, d (mm) c Summary of This Lchar Proposal • Know particle characteristics, primarily particle density • Choose media size – For dual media, danth = 2 dsand • Choose filtration velocity • Calculate Lchar for dp = 1 µm and 7 µm from Tufenkji and Elimelech model • Choose L = 0.29 Lchar Advantages of This Lchar Proposal • Much better grounded in filtration theory than current design guidelines • More reasonable in accounting for variations in filtration velocity (and all media and suspension characteristics) • Should save on pilot scale tests • Remarkably, it seems more consistent with current practice than the simple design guidelines currently used Application to slow sand filters Location Filtration velocity (m/h) 0.05 Media depth (m) 0.30-1.06 Media size (mm) 0.3 West Hartford, CT1 0.21 0.61-0.69 0.25-0.35 New Haven, CT1 0.03 0.41-0.56 0.3 Salem, OR2 0.2 0.92 0.3 WWTP Ruhleben, Berlin, Germany3 0.2-0.25 0.8 1-2 0.2 1.0-1.3 0.15-0.30 0.1-0.3 1 0.15-0.35 Springland, MA1 Kosrae, Micronesia4 Traditional design5 Applications to slow sand filters Characteristics of slow sand filters: • • • • • Far slower velocity ( 0.1 m/h instead of 5 m/h) Relatively similar depth Small media size and relatively low porosity No chemical addition prior to filter Assume alpha = 0.001 Total filter removal efficiency 1.0 Slow sand filter (Salem OR design) 0.8 0.6 0.4 Yao Yao α = 0.01 T&E α = 0.001 Yao T&E 0.2 0.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (dp in µm) 1.0 1.5 Total filter removal efficiency 1.0 Extremes of slow sand filter design 0.8 New Haven α = 0.001 0.6 0.4 0.2 Berlin α = 0.01 0.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (dp in µm) 1.0 1.5 Applications to slow sand filters Results • Very poor removal with low alpha, especially of small particles • What is wrong? Alpha is perhaps much higher • Why? Conjecture: MOs excrete extracellular polymers that cover the media and increase alpha Applications to riverbank filtration Location Seepage velocity (m/h) Media depth (m) Pembrok, NH6 0.46 54.9 Milford, NH6 0.95 22.9 Louisville, KY6 0.51 12.2 Cedar Rapids, IA6 0.16 19.5 Common design7 0.03-1.00 10-600 Media size (mm) 0.3-0.8 Total filter removal efficiency Application to riverbank filtration 1.0 0.8 d , v, L, α c 0.5, 0.5, 10 m, 0.001 or 0.5, 0.5, 100m, 0.0001 0.6 0.4 0.2 0.5, 1.0, 20, 0.0001 0.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log of particle diameter (dp in µm) 1.0 1.5 Riverbank Filtration (more conjecture) • These systems behave like gravel flocculators: in the long run, they don’t remove many small particles, but they attach small particles that break off as flocs and therefore show great turbidity reduction. • They also act as concentration equalization units—holding up particles in the bank when the concentration is high and releasing particles when the concentration is low. • They also are biologically active and release EPS that contribute to a higher alpha than would otherwise be present. THMs for filtration systems • Chemistry matters (a lot) (and may be influenced by microbiology) • Particle size matters (but maybe not quite like the models suggest) • Ripening and detachment happen • Filtration analysis and design should be done rationally using the characteristic removal length • Filter systems have more similarities than differences Thanks to my Teachers • Phil Singer • Charlie O’Melia • • • • • • • • • Sarah Clark Bob Cushing Jeannie Darby Sara Kau Ijung Kim JinKeun Kim Dan Moran Melissa Moran Jeff Nason
© Copyright 2026 Paperzz