Development and application of a predictive model for advanced wastewater treatment by adsorption onto powdered activated carbon

Highlights

• IAST-TRM enables modeling of multi-component adsorption within wastewater matrices

• Model calibration requires a simple set of batch tests to reflect matrix effects

• Application tested for seven organic micropollutants in presence of competing DOC

• Model accuracy depends mainly on different adsorbate and adsorbent characteristics

• Dynamic system simulation assists to make a choice about PAC, dose and contact time

Abstract

This work presents a mathematical method to describe adsorptive removal of organic micropollutants (OMPs) and dissolved organic carbon (DOC) from wastewater treatment plant effluent using powdered activated carbon (PAC). The developed model is based on the tracer model (TRM) as a modification of the ideal adsorbed solution theory (IAST) and uses the fictive component approach for organic matter fractionation. It enables the simulation of multisolute adsorption of OMPs considering competitive adsorption behavior of organic background compounds (OBC). Adsorption equilibrium data for DOC and seven different OMPs as well as kinetic data for DOC were derived from batch experiments performed with secondary clarifier effluent of two municipal wastewater treatment plants (WWTP 1 and WWTP 2). Two conventional PAC products were investigated as well as one biogenic PAC (BioPAC). Verification and validation of the fitting results based on operational data of WWTP 1 showed promising prediction of DOC and OMP removal efficiency. However, when applied to a static simulation of a full-scale PAC adsorption stage, the model overpredicts the removal efficiency of sulfamethoxazole and candesartan. For benzotriazole, carbamazepine or hydrochlorothiazide, predicted removal falls below operational removal. The model can be used to predict removals of good adsorbable OMPs but fails to accurately predict the removals of OMPs with variable or low PAC affinity. The model was further used for a dynamic simulation of DOC and diclofenac effluent concentrations of a full-scale PAC adsorption stage with varying operating conditions and influent concentrations. Results show that the hydraulic retention time (HRT) in the contact reactor is a decisive operational parameter for OMP removal efficiency besides the PAC dose.

Introduction

Municipal wastewater treatment plants (WWTPs) are considered to be a major point source for the emission of anthropogenic organic micropollutants (OMPs) - such as pharmaceutical residues, hormones, personal care products, detergents into the aquatic environment. In recent years there is growing concern and interest in the installation of advanced treatment technologies for the removal of OMPs at WWTPs (Loos et al., 2013). Currently, ozonation and adsorption using powdered activated carbon (PAC) or granular activated carbon (GAC) are the technologies that are widely studied and implemented in pilot and full-scale plants (Boehler et al., 2012; Hollender et al., 2009). The process selection mainly depends on local conditions involving consideration of multiple factors in a cost-benefit analysis (Margot et al., 2013). To select the most suitable treatment technology it is necessary to estimate the potential OMP removal as precisely as possible. In order to do that, predictive mathematical adsorption models are needed.

Mathematical models are able to represent the essential aspects of the adsorptive removal processes via relationships and variables, which assists a better understanding of processes and enables to predict process behavior. In literature, different mathematical models for adsorption of OMPs and dissolved organic matter onto PAC in wastewater matrices are studied in order to gather information about adsorption equilibrium of different adsorbates or to qualify interactions between different adsorbates. Some of the most important models for multisolute adsorption are summarized in the following passages.

Competitive adsorption can be described by the ideal adsorbed solution theory (IAST) which allows predicting the multicomponent adsorption from single-solute isotherm parameters (Worch 2021). The drawbacks of IAST approach is that by using single-solute isotherm parameters it fails to model the interactions and competition between OMPs and organic background concentration (OBC). Multisolute adsorption effects, such as interactions between adsorbate and adsorbent, different OMP and OBC accessibilities to adsorbent micropores or effects of pore blockage, are not considered. This leads to a general overestimation of competitive adsorption effects of DOC (Worch, 2021). Furthermore, the IAST requires detailed knowledge of solution composition and high experimental effort in terms of single-solute isotherm parameters of all OMPs and OBCs. In order to overcome the described problems, two IAST modifications were developed, which are Equivalent Background Compound Model (EBCM) (Najm et al., 1991) and Tracer model (TRM) (Burwig et al., 1995; Rabolt et al., 1998). EBCM assumes one equivalent background compound (EBC) as competing adsorbate for each OMP, leaving all OMP data (concentration, single-solute isotherm parameter) unchanged. The TRM modification enables the correction of single-solute isotherm parameters of the OMP in the presence of OBC for further application in the IAST. However, since the different constituents of the organic matter present in water or wastewater are very hard to be identified and only total isotherms can be measured experimentally using collective parameters such as DOC, IAST precondition of known solution composition is not fulfilled. One solution to the issue is to use what is known as the fictive component approach or adsorption analysis (Crittenden et al., 1985; Worch, 1989). This approach allows the characterization of DOC composition based on adsorbability and the transformation of the unknown mixture into a certain number of fictive components.

Zietzschmann et al. (2016) modeled the competitive adsorption between OMPs and background organic matter by modifying IAST using the EBCM. For each OMP, an individual EBC was set considering all observed competition effects. A simplified approach of IAST-EBC model was also used by Shimabuku et al. (2017) to compare competitive adsorption in deionized and stormwater on pulverized GAC.

TRM replaces single-solute OMP isotherm parameters in IAST with the corrected parameters in the presence of OBC after determination of different fictive OBC fractions using the fictive component approach. The main advantage of TRM compared to EBCM is its simplicity, as there is no need to add new EBC fractions for each OMP. Furthermore, adsorption of OBC and OMP can also be modeled parallelly (which is not possible in EBCM). Worch (2010) compared TRM and EBCM approaches and reported that TRM is advantageous due to these reasons.

However, only few attempts of simulation using IAST modified by TRM were found in literature. Viegas et al. (2020) predicted the adsorption of carbamazepine, diclofenac and sulfamethoxazole on a carob waste-derived activated powdered biochar. Modeling was based on a four-component approach consisting of three OMP and total DOC without fractionation. However, Viegas et al. (2020) studied the adsorption equilibrium and isotherm of the OMPs without considering adsorption kinetics. Therefore, their model could not perform dynamic simulation with varying operational parameters such as PAC dose and hydraulic retention time (HRT). Nowotny et al. (2007) modeled adsorption competition between organic matter and OMP onto powdered and granulated activated carbon in terms of isotherm parameter fitting using the tracer model (TRM). The single-solute Freundlich parameters K and n were fitted for each OMP and DOC fraction leading to an overdetermined system, in which several combinations of parameters might give the same goodness of fit and therefore reduce equifinality. Acevedo et al. (2021) focused on adsorption and biodegradation of organic matter inside GAC filtration and used the fictive component approach together with an IAST-based data fitting routine to determine adsorption dynamics of DOC with GAC. However, modeling of OMP removal was not part of the study of Acevedo et al. (2021).

To the best of the authors’ knowledge, an IAST-TRM model which combines both adsorption isotherms and adsorption kinetics to simulate removal of OMPs in wastewater matrix has not been developed yet. This paper presents thus for the first time a simplified model variation of IAST-TRM that considers not only the removal efficiency for different OMPs but also the significance of adsorption kinetics by PAC in real wastewater. It investigates the adsorption performance of PAC by applying a mathematical model to steady state and dynamic simulation of a full-scale PAC treatment stage. With use of multisolute experimental data for the correction of single-solute DOC and OMP isotherms the model requires only a limited amount of data sets respectively experimental effort in terms of model calibration. The kinetic simulation step in the model allows dynamic simulation of different operational scenarios. The developed adsorption model fosters thus better understanding of OMP removal by adsorption onto PAC.