The increasing knowledge of anaerobic processes has led to the development and implementation of sophisticated and complex mathematical models. The quantification of the models' ability in describing the anaerobic demanding processes becomes fundamental for a realistic estimate of the model parameters. However, there is little consensus about the model's structure and parametric identifiability questions. Those questions are not yet sufficiently elucidated in the reported anaerobic digestion modeling studies. In this paper, a complex numerical model is proposed for simulating the anaerobic biogas production from the co-digestion of organic waste material. An innovative complex dynamic model structure is proposed to support full-scale anaerobic plant design and operation decisions and to assist laboratory scale and pilot codigestion research. The model facilitates the understanding of the co-digestion effects and therefore discards any potential negative impacts from mixing based on random or heuristic decisions. This paper introduces an innovative modeling procedure, including the application of uncertainty and global sensitivity analysis (LHS/PRCC/eFAST), which allows the study of a multi-dimensional parameter space globally so all uncertainties can be identified among the parameters; a multi-steps approach that gives a clear picture of the main sensitive model parameter. Among parameters, special concerns will be given to those identified as sensitive conducting to the digester methane optimization. Sophisticated and stable algorithms are designed for the model cost function minimization criteria and result in an increased accuracy. The parameter uncertainty and sensitivity analysis revealed that the hydrolysis and acidogenesis phase are the most affecting phases of the methane production. Parameter such the polymer hydrolysis rate (Kh), the specific acidogens maximum growth rate (µ_MAX1), the saturation constant for acidogens (K_(S1 )), the saturation constant for acetoclastic methanogens (K_(S5 )), and the gas liquid mass transfer coefficient for CH4 (Kla7), contribute the most to the variance of the complex model estimate of methane.
|Publication status||Published - 7 Mar 2018|
- anaerobic process modelling
- Latin hypercube sampling
- Partial rank correlation, (PRCC)
- Extended Fourier amplitude sensitivity (eFAST)