Skip to Main content Skip to Navigation

Nonlinear Modeling, Identification and Control of Membrane Bioreactors

Abstract : This thesis proposes a simple submerge membrane bioreactor (sMBR) dynamic model that comprises physical and biological process behaviors. The filtration, physical aspect, is a resistance-in-series model that is composed with reversible resistance, linked to sludge cake formation process that can be detached by air scouring, and the irreversible fouling resistance. The biological feature is implemented extending the simple chemostat model to the filtration mechanism. The model asymptotic analysis, observability, controllability and fast and slow dynamic study are carried out. The latter, based on the Tikhonov's theorem, reveals the possibility to simplify model dynamics by decoupling the process in three time scales, i.e. long-term fouling evolution (slow dynamic), biological degradation (fast dynamic) and fouling cake formation (ultrafast dynamic). As sMBR processes are relativity new, real process data are scarce. Thus, a recirculating aquaculture system pilot plant with an sMBR is design, build and automated. Process online measurements such as: temperature, total suspended solids (TSS), ammonia and nitrate effluent concentrations, air cross- and effluent flow rates and trans-membrane pressure are gathered in other to validate the proposed model. To evidence the model general framework the same model is confronted with real data sets obtained from an sMBR wastewater treatment plant. Therefore, a parameter identification is organized in three steps corresponding to the three time scales obtained from the analytical analysis. The parameter identification is implemented using a weighted least-squares cost function and the inverse of the Fisher Information Matrix (FIM), which is used to obtain the parameters confidence intervals, is computed by a lower bound on the covariance matrix of the parameter estimates. The model capacity to predict trans-membrane pressure and biological degradation is proved by model validation and cross-validation results, in which an accurate correlation coefficients (R^2) of approximately 0.83 are obtained. Concerning the process control, two different approach are used: a partial-linearizing feedback Lyapunov controller is designed in order to stabilize the fouling production by actuating in the air cross- and effluent flows; and a nonlinear model predictive control (NMPC) is implemented in other to optimize the effluent production rate and maximize the period between two chemical cleaning procedures. The results included in this thesis show the importance of analytical model studies in order to process cognition and model simplification. Another important point is the simple dynamic model structure, with a small quantity of the parameters, which is adequate to implement advanced control strategies on sMBR processes and, similarly, to predict biological degradation and fouling build-up dynamics.
Keywords : Control Mbr Modeling
Document type :
Complete list of metadata

Cited literature [98 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Tuesday, March 5, 2019 - 2:03:09 PM
Last modification on : Monday, November 23, 2020 - 10:07:08 AM
Long-term archiving on: : Thursday, June 6, 2019 - 2:59:52 PM


Files produced by the author(s)


  • HAL Id : tel-01130312, version 2
  • PRODINRA : 285990


Guilherme Araujo Pimentel. Nonlinear Modeling, Identification and Control of Membrane Bioreactors. General Mathematics [math.GM]. Université Montpellier; Université de Mons, 2015. English. ⟨NNT : 2015MONTS219⟩. ⟨tel-01130312v2⟩



Record views


Files downloads