关键词:
Structural analysis
Algorithms
Reaction products
Differentiation
Chemical engineering and industrial chemistry
摘要:
Mathematical modeling of dynamic chemical engineering systems such as flow-sheet simulation, transient analysis of reaction and separation networks, computational fluid dynamic, or control problems often leads to systems of differential-algebraic equations (DAEs). Knowing the index of a DAE is an important prerequisite for its consistent initialization and its numerical solution. For large sets of DAEs occurring in chemical engineering practice, index analysis requires computational procedures. This paper presents a new approach for automatic index analysis. Earlier index analysis algorithms were based on purely structural calculus. Unfortunately, structural methods may actually arbitrarily over- or underestimate the correct DAE index. To overcome the shortcomings of existing approaches, this paper proposes a novel algorithm expanding DAE analysis to include both symbolic and numerical information. The novel symbolic numerical analysis accounts for variable cancellation and processes linear expressions accurately. For the case of purely linear DAE, the new method is rigorous. For the general nonlinear DAE, the novel approach gives a more precise picture in most cases. The paper introduces the theory supporting the methodology and provides the proof of concept with small illustrative examples.