Comparison of a type curve and a least-squared errors method to estimate biofilm kinetic parameters

• Published in: Water research (Oxford) Vol. 37; no. 13; pp. 3279 - 3285
• Main Authors: Riefler, R.Guy, Smets, Barth F.
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.07.2003; Elsevier Science
• Subjects: Applied sciences; Bacteria; Biofilm; Biofilms; Biological and medical sciences; Biological treatment of waters; Bioreactors; Biotechnology; Diffusion; Dinitrotoluene; Environment and pollution; Exact sciences and technology; Fluidized bed reactor; Fundamental and applied biological sciences. Psychology; General purification processes; Industrial applications and implications. Economical aspects; Kinetics; Least-Squares Analysis; Parameter estimation; Pollution; Population Dynamics; Reproducibility of Results; Wastewaters; Water treatment and pollution; Fluidized bed reactor; Parameter estimation; 2,6-DNT; 2,4-DNT; 2,6-dinitrotoluene; FBBR; fluidized bed biofilm reactor; 2,4-dinitrotoluene; Dinitrotoluene; Biofilm; Kinetics; least-squared errors; LSE; Bioreactor; Least squares method; Type curve; Kinetic parameter; Waste water purification; Biological purification
• ISSN: 0043-1354; 1879-2448
• DOI: 10.1016/S0043-1354(03)00139-8

Because of the kinetics of diffusion limitation and the difficulty of replicating biofilm structure in a test vessel, biokinetic parameters of substrate consumption (maximum specific substrate removal rate, q m and half-maximum removal coefficient, K s) are particularly difficult to measure in biofilm reactors. In this research, a type curve method using load-shift experiments was compared to a method using a least-squared errors (LSE) routine. More accurate and precise estimates were obtained with the LSE routine than with the type curve method by removing subjectivity from the estimation process. In addition, the LSE estimation process allowed a more rigorous evaluation of the adequacy of the data fit, permitted estimation of approximate confidence intervals of parameters, and identified flaws in the data set. As a result, we advocate the use of the recently developed LSE based estimation routine to estimate parameters from such experiments.