22 august, QSPR prediction of pharmaceutical removal with GAC Delft David de Ridder Carbamazepine
22 august, Problem statement micropollutants different micropollutants in surface water Variable mixture RIWA Rijn measures about 250 org. micropollutants Analysis programs costly and time consuming QSAR can be used as a screening tool
22 august, QSAR werk David de Ridder Singular processes Activated carbon (AC) Aeration Ozone UV Combinations of processes AC-UV/H2O2-AC (RBF)-NF-AC
22 august, Activated carbon Research objectives Estimate: Influence of turbulence on adsorption kinetics Influence of water quality Influence of preloading carbon Construction of QSPR model Accuracy of QSPR prediction Ce, qe
22 august, Experiment set-up pharmaceutical selection
22 august, Experiment set-up matrix Demi waterSurface waterWaste water Regenerated carbon Kinetic (car, mix) Kinetic low turbulence (mix) Equilibrium (car, met, car+met, mix) Preloaded carbon Equilibrium (mix)Kinetic (mix) Equilibrium (mix)
22 august, Experiment set-up 2 litre solution 200 mg carbon Adsorption time 1 day – 6 weeks mg carbon Adsorption time 8 weeks Carbon 0,63-0,71 mm
22 august, Results Kinetic experiments
22 august, Results Kinetic experiments
22 august, Results Kinetic experiments
22 august, Results Kinetic experiments Lower turbulence decreases adsorption significantly Preloading large influence on negatives, and no significant effect on positives.
22 august, Results - Equilibrium
22 august, Results - Equilibrium
22 august, Results - Equilibrium
22 august, Results initial remarks MW in range D no significant influence Log D has higher impact on removal of negatives At similar pKa, higher log D yields higher removal At similar log D, positives are removed 1,2-2 times more effective than negatives In wastewater: Positives comparable removal as surface water Neutrals better removal than in surface water Significant removal of 4 negatives in blank!
22 august, Results hypotheses Preloading creates a negatively charged layer onto the carbon, rejecting negatives and attracting positives In the MW range of D, probably most carbon micropores will be available for adsorption In wastewater, (bio)degradation of negatives is preferred.
22 august, Results - QSPR construction 4 (out of 21) compounds excluded for verification MLR (multivariable linear regression) prediction model
22 august, Results - QSPR prediction
22 august, Results - QSPR validation
22 august, Results – model accuracy Specific mispredictions Consequent over/underprediction Less data available
22 august, Results – Freundlich parameters
22 august, Initial conclusions Model prediction Applied carbon dose too high -> ultrapure models inaccurate Initial degradation negatives wastewater -> negatives not taken properly into account Surface water: general underprediction Ce at higher carbon dose Wastewater: Specific overprediction Ce (Terbutaline, Salbutamol, metropolol, Clenbuterol, Aminopyrine) Demiwater PL: consequent over/underprediction
22 august, Further research Dataset Larger variation MW (if relevant) Process conditions Carbon type Change preparation preloaded carbon Lower carbon doses pH variation at same water quality (decrease charge negatives, lower log D negatives & higher log D positives) Analysis ATP to check for biological activity NOM characterisation (blank & adsorbed (?)) Carbon characterisation (PSD, hydrophobicity, carbon pKa)
22 august, ¿Questions/remarks?
22 august,
22 august,
22 august,
22 august,