Arne Verlefde, Bas Hejman, Emle Cornelssen, Bart Van der Bruggen, Hans van Djk Partner for progress Modellng of removal of organc mcropollutants by nanofltraton Workshop Developments n drnkng water Organc mcropollutants Kwa 2005 2 1
Organc mcropollutants MW: often around 200300 g/mol NFmembranes: MWCO around 200300 g/mol TS80 TSF (Trsep Corp.) UTC20 (Toray) NTR 7250 (NttoDenko) SC3100 (Toray) Desal HL 51 (OsmoncsDesal) MWCO (Da) 200 350 300400 200300 150300 Smallest pollutants (e.g. MTBE, NDMA): problem? Kwa 2005 3 Rejecton modellng Rejecton (full scale) Dfferent steps: Identfcaton of mechansms of rejecton Identfcaton of parameters nfluencng rejectonmechansms Relatng parameters to structure (QSAR) Structure parameters rejecton Kwa 2005 4 2
Rejecton mechansms for organc pollutants n NF Membrane Sterc hndrance (=sevng mechansm) Kwa 2005 5 Mechansms n NF Membrane Electrostatc nteractons Kwa 2005 6 3
Mechansms n NF Membrane Hydrophobc nteractons (& adsorpton) Kwa 2005 7 Mechansms parameters Solute and membrane parameters nfluencng rejecton Solute parameters Membrane parameters Sterc hndrance Adsorpton/hydrophobc nteractons Charge nteractons molar mass (MW) molecular sze log K ow (dpole moment) pk a MWCO pore sze (& dstrbuton) hydrophobcty (e.g. contact angle) Membrane charge (e.g. Zetapotentaal) Other nfluences on rejecton: Operatonal parameters (flux, pressure, crossflow velocty (concentraton polarsaton)) Water qualty: DOC, IS (onc strength), temperature, turbdty (suspended solds) Kwa 2005 8 4
Rejecton of organc pollutants Benchscale crossflow experments Trsep TS80 TSF membrane Characterzaton (500 ppm MgSO 4 ) for 2 hours Fltraton experments (4 days) Benchscale test (5,5 bars; 10% recovery) Feed concentraton ( m g /l) 6 5 4 3 2 1 0 0 1 2 3 4 5 6 Tme (days) Atrazne Smazne Duron Metamtron Kwa 2005 9 Rejecton of uncharged organcs Dfferent ranges of hydrophobcty: log K ow < 1 (hydrophlc) log Kow> 3 (hydrophobc) 1 < log K ow < 2 (ntermedate) Kwa 2005 10 5
Rejecton of uncharged organcs Dfferent ranges of hydrophobcty: log K ow < 1 (hydrophlc) log Kow> 3 (hydrophobc) 1 < log K ow < 2 (ntermedate) Rejecton modellng: 2 ( ln(mw) ln(mwco) 0.56s ) MW* 1 1 R(MW*) = exp 2 s 2 MW 2s 0 MW π MW MW dmw (Van der Bruggen and Vandecasteele, 2002) Kwa 2005 11 Uncharged organc compounds Trsep TS80 100 90 80 70 Rejecton (%) 60 50 40 30 20 10 0 MWCO 0 50 100 150 200 250 300 350 400 Molecular weght (g/mol) log Kow>3 log Kow<1 1< log Kow< 1,5 Kwa 2005 12 6
Uncharged organc compounds Trsep TS80 100 90 80 70 Rejecton (%) 60 50 40 30 20 log Kow<1 1<log Kow<2 3<log Kow MWCO log Kow>3 (g/mol) log Kow<1100 1< log Kow< 1,5 180 250 10 0 MWCO 0 50 100 150 200 250 300 350 400 Molecular weght (g/mol) Kwa 2005 13 Uncharged compounds: concluson cutoffs for 1 membrane MWCO: not always a useful parameter Hydrophlc pollutants: sevng curve Hydrophobc pollutants: wdespread cutoff (hydrophobc nteractons) Hydrophlc pollutants: MWCO= 80 g/mol 200 g/mol Rejecton of hydrophlc pollutants better than expected Rejectons hydrophobc pollutants s lower Kwa 2005 14 7
Uncharged compounds Good news for hydrophlc (polar) compounds, BUT: 100 MTBE 90 90 80 80 70 70 Rejecton (%) 60 60 50 50 40 40 30 30 20 20 10 10 NDMA 0 0 00 50 50 100 100 150 150 200 250 250 300 300 350 350 400 400 Molecular weght (g/mol) log Kow>3 log Kow>3 log Kow<1 log Kow<1 1< log Kow< 1,5 MTBE: MW=80 g/mol: on top of cutoff: rejecton vares (090%) (small varatons n membranes?) Kwa 2005 15 Charge effects Membrane surface: negatvely charged Hgh rejectons for negatvely charged compound (charge repulsons) Expected: lower rejectons for postvely charged compounds Incorporaton of extra CPfactor concentraton c b GouyChapman method: calculate concentraton profles Kwa 2005 16 8
Charge nfluence on rejecton 100,0 Retente (% ) 80,0 60,0 40,0 20,0 Negat ef geladen Post ef geladen 0,0 0 100 200 300 400 Molecuulgew cht (g/mol) Kwa 2005 17 Fullscale plant: nfluence of recovery on rejecton 1,00 0,95 Retente () 0,90 0,85 0,80 0,75 0,70 0,65 0,60 0,55 Merenzuur Glycolzuur Azjnzuur Melkzuur Malonzuur Benzoëzuur Phenylazjnzuur 0,50 0 10 20 30 40 50 60 70 80 90 100 Recovery (%) Hogere recovery lagere retente Kwa 2005 18 9
Fullscale plant: nfluence of charge effect Rejecton of charged and uncharged acds vs recovery Rejecton () 1,00 0,90 0,80 0,70 0,60 0,50 0,40 0,30 0,20 0,10 0,00 Recovery (%) Formc acd ph 8 (charged) Acetc acd ph 8 (charged) Formc acd ph 5 (uncharged) Acetc acd ph 5 (uncharged) Kwa 2005 19 Fullscale plant: modellng of rejecton Convectondffuson model J w =K w.(δpδπ) J s =J w.c p =K s.δck c c b J w Mass balances R=1(c p /c f ) R = 1 K w, βk c, ( ΔP Δπ ) 2( 1 S ) 2( 1 S ) s, 1 β K c, 2 S βk S = 1 Z R tot n K A w, = 1 = 1 n ( Δ P Δπ ) K A ( Δ P Δπ ) w, = 1 Kwa 2005 20 Z 1 j= 0 1 S Z 1 S j j j 10
Fullscale modellng Dffusonconvecton model J = K Δc K c J s s c b w Rejecton vs recovery duron Rejecton vs recovery metamtron 100% 100% Rej ecton ( ) 80% 60% 40% 20% Model data Real data Rej ecton () 80% 60% 40% 20% Model data Real data 0% Recovery () Maxmum element rejecton=78.2% Mnmum element rejecton=74.0 % 0% Recovery () Maxmum element rejecton=84.2% Mnmum element rejecton=82.6 % Kwa 2005 21 Fullscale modellng Convecton model J =K c J Rejecton vs recovery duron 100% s c b w (Rejecton constant) 100% 1 βkc, R = S 1 βkc, 2 1 ( S ) Rejecton vs recovery metamtron = 1 Z Rej ecton () 80% 60% 40% 20% Model data Real data Rej ecton () 80% 60% 40% 20% Model data Real data 0% Recovery () Maxmum element rejecton=78.0% Mnmum element rejecton=75.1 % 0% Recovery () Maxmum element rejecton=82.3% Mnmum element rejecton=79.8 % Kwa 2005 22 11
Fullscale modellng Dffuson model J s = K Δc Rejecton vs recovery duron 100% s (Rejecton dependent on flux) R = 1 K w, 1 βk ( ΔP Δπ ) 2( 1 S ) Rejecton vs recovery metamtron 100% s, = 1 Z 2 S 80% 80% Rej ecton ( ) 60% 40% 20% Model data Real data Rej ecton () 60% 40% 20% Model data Real data 0% Recovery () Maxmum element rejecton=79.6% Mnmum element rejecton=73.1 % 0% Recovery () Maxmum element rejecton=82.8% Mnmum element rejecton=78.0 % Kwa 2005 23 Flux dependency of rejecton Test on benchscale, feed water of plot 1 0,9 Rejecton () 0,8 0,7 Flux range n plot Atrazne Smazne Duron Metamtron 0,6 0,5 0 0,000002 0,000004 0,000006 0,000008 Permeate flux (m/s) Flux dependency neglgeble Kwa 2005 24 12
Fullscale predcton concluson Hghest flux (=best) elements n frst stages, lower flux elements n last stages Therefore underpredcton n frst stage, overpredcton n last stage One parametermodellng s feasble Fluxdepency of rejecton not mportant for compounds modelled Measurement of rejecton on benchscale predcton on fullscale Kwa 2005 25 Partner for progress Thank you for your attenton! Questons? 13