Modelling and identification of a climbing film evaporator.
Thesis DisciplineChemical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The objective of this work is the characterisation of a model to describe the dynamics of a climbing film evaporator over a wide operating range. The model should be sufficiently simple so that the calculation of control action is straight-forward. The hypothesis of this thesis is that such a model may be constructed to describe the dynamics of a climbing film evaporator well. The first chapter introduces the topics of climbing film evaporation, two-phase flow and modelling and identification. The second chapter describes the climbing film evaporator used in this study. For this purpose, the evaporator was fully instrumented with temperature sensors, conductivity cells and flow-meters. The instrumentation was commissioned and calibrated. In the next chapter a distributed parameter model of the climbing film evaporator is derived from the one-dimensional homogeneous two-phase flow equations, the parameters of which were to be determined by identification. The next two chapters summarise the field of identification. The fourth chapter presents and develops identification techniques for lumped parameter models, and the fifth chapter describes and constructs identification methods for distributed parameter models. The methods described in the lumped parameter identification section were recursive, so that they could be used in real-time to track time-varying parameters – a feature that is useful in the design of self tuning regulators. These identification methods used a UD factorisation algorithm and were found to be robust for inappropriate choices of system dead-time. Accurate estimates of dead-time were obtained from either method. The distributed parameter identification methods investigated were optimisation schemes to minimise an output least square error criterion. Methods for solving the distributed parameter identification problem using the method of characteristics were investigated and developed. Identification using the method of characteristics is appropriate as the partial differential equations describing the climbing film evaporator are hyperbolic in nature. Chapter six presents the identification strategy adopted to model the evaporator using the techniques described and developed in chapters four and five. The experiments for the collection of data to be used in the various models are designed. A range of models of the climbing film evaporator were identified. The simplest models for the evaporator were global black-box linear models. Gain-scheduled linear models were identified to attempt compensation for system non-linearity. Finally the parameters of distributed parameter models for the climbing film evaporator were investigated. These models are presented and discussed in chapters seven, eight and nine respectively. The thesis is organised so that pages are numbered within chapters, with nomenclature and references listed at the end of each chapter. The paper entitled "Multi-input, multi-output identification of a pilot-plant climbing film evaporator" is based upon this work (Appendix VI). The paper has been accepted for presentation at the 12th World Congress of the International Federation of Automatic Control, Sydney Convention and Exhibition Centre, Darling Harbour, Sydney, Australia, 19th-23rd July 1993.