SYSTEM OBSERVABILITY AND NONLINEAR PARAMETER IDENTIFICATION OF NONYLPHENOL BIODEGRADATION KINETICS zyxw Vishal Shah', Aditya Chaubap, Ravi zyxwvu I? Ramachandran3, R a d Orddiiez4 and Kauser Jahan5 zy 1. zyxwvutsrqp California Institute of Technology, shah@caltech.edu 2. Virginia Polytechnic Institute, adchauba@vt.edu 3. Rowan University, ravi@rowan.edu zyxw 4. University of Dayton, Raul.Ordonez@notes.udayton.edu 5. Rowan University, jahan@rowan.edu ABSTRACT Addition of surfactants, detergents and emulsifiers has been suc- cessfully applied for cleanup of petroleum-contaminated sites. How- ever, a certain groupof widely used alkylphenolethoxylates(APEs) surfactants was recently banned in Europe because scientists dis- covered that APE breakdown products are estrogenic and highly toxic to aquatic organisms. Nonylphenol is one of the very toxic breakdown products. The process of nonylphenol biodegradation is very important to many scientists because of its potential effec- tiveness as a treatment tool for pollution. However, very little in- formation is available on the biodegradation kinetics of nonylphe- nol. Kinetic information is necessary for predicting the fate of pol- lutants. We start with Monod's model for nonylphenol biodegra- dation which is based on a coupled system of nonlinear differen- tial equations. We prove that the states of the system and the pa- rameters of Monod's model are locally observable. This enables us to perform a meaningful parameter estimation analysis. By using nonlinear least-squares optimization, we obtain the biodegradation kinetics and verify physical feasibility on independent datasets. 1. INTRODUCTION Researchers have indicated that solubilization followed by micro- bial metabolism of organic contaminants is technically feasible and has potential as aremedial technology [I]. Addition of surfactants, detergents andemulsifiers has been successfully applied for cleanup of petroleum-contaminated sites. However, acertain group of widely used alkylphenolethoxylates (APEs) surfactants were banned in Eu- rope because scientists discovered that APE breakdown products are highly toxic to aquatic organisms. Recent evidence that some APE breakdown products are estrogenic has intensified concern over their environmental and human health effects [2]. APEs are nonionic surfactants made up of a branched chain ethylene oxide to produce an ethoxylate chain. Most APEs enter the aquatic environment from wastewater treatment plant discharges. Nonylphenol, one of the breakdown products of APEs, is known to be extremely toxic. Nonylphenol, adsorbs to soils and sludges, tends to bioaccumulate, and has been shown to be mildly estro- genic. Sewage sludge thus applied to agricultural land may con- tain nonylphenol. European studies indicate high concentrations of nonylphenol in treated sewage sludge thereby indicating sludge disposal as a source of aquatic contamination. Nonylphenol biodegra- dation is therefore a significant environmental issue [31. However very little information is available on the biodegradation kinetics of nonylphenol. Kinetic information is important for predicting the fate of pollutants. It has been demonstrated that it is possible to determine intrinsic kinetics of single organic compounds by using oxygen uptake data from electrolytic respirometry zyxw 141. The purpose of this paper is to use nonlinear least-squares op- timization to obtain the intrinsic kinetic parameters of nonylphe- no1 biodegradation from oxygen uptake data. These parameters are part of an existing mathematical biodegradation model of coupled nonlinear differential equations known as Monad's model [SI. The model has three states, namely, oxygen uptake. substrate (or pol- lutant) removal and bacterial cell growth. Only theoxygen uptake is experimentally measurable. We perform an analysis of Monod's model to show that its states and parameters are locally observable about the entire state space except for some isolated points. The model parameters obtained by nonlinear optimization are in agree- ment with the experimental measurements for all three states. This not only makes oxygen uptake data very attractive as it can elimi- nate monitoring of substrate removal and bacterial cell growth but also verifies the feasibility of using nonlinear optimization. The oxygen uptakedatais obtainedusing electrolytic respirometry. This is relatively inexpensive and gives fast. accurate and reliable data. zyx 2. MODELING OF THE NONYLPHENOL BIODEGRADATION PROCESS The process of nonylphenol biodegradation is given by a set of non- linear differential equations that describe its kinetic behavior. This is known as Monod's model of bacterial kinetics and is given by 151 zyxwvutsr dS 1 umsx . - -- dt zyxwvuts Y K,+S This system has three states: S is theconcentrationof substrate. X is the active cell mass concentration, and 0, is the cumulative oxygen consumption in the reactor (oxygen uptake which we mea- sure). The time evolution of the system is also determined by the several system parameters or biodegradation coefficients involved pm is the specific growth rate of substrate; hh is the endogenous decay coefficient: K, is the substrate utilization coefficient; Y is 0-7803-7761-31031Sl7.0082003 IEEE IU-24