Reimplementation of MMS in Dymola Using Bond Graphs

Abstract

Modeling and simulation have become indispensable design tools since they permit the engineer to predict the behavior of a system before it is actually built. In fact, modeling and simulation are the only techniques available that make it possible to analyze arbitrarily nonlinear systems accurately even under varying experimental conditions.

Computer simulation has become more and more common in the past 30 years due to the increased availability of more and more powerful digital computers. Several modeling techniques have been developed for different purposes, e.g. the Finite Element and Finite Difference Methods or multibody dynamic system simulation techniques in mechanical engineering or simulation, and design systems for integrated circuit design in electrical engineering.

In this text, a special topic in continuous systems simulation is addressed, namely power plant dynamic performance simulation. In this field, simulation becomes extremely important due to the importance and the impact of recent decisions for the present and for the future and because of the huge investments that are necessary in this engineering field. For power plant simulation, a simulation system was developed by EPRI in the early 1980's, called the Modular Modeling System or simply MMS. MMS consists of a set of pre-programmed modules representing various power plant components like pumps, pipes, heat exchangers or turbines. The modules are combined in a module library that is the kernel of the MMS. An actual power plant model is built by selecting the desired modules, supplying specific parameterization data to turn the generic modules in actual representations of power plant components and finally combining the modules in a simulation program defining the module interconnections.

Once a model of a power plant is built, it needs to be executed in order to determine the desired transients of plant or single component performance. This is done using a so called simulation language. MMS was and still is based on the simulation language ACSL. ACSL is an excellent software product to deal with the numerical solution of systems of ordinary differential and algebraic equations and provides some useful options like several integration algorithms and the eigenvalue determination of a given system. The language is not so well suited, however, to deal with large, hierarchically structured models itself. For this purpose, several so called modeling languages like Dymola were introduced. Dymola has a very sophisticated system to build hierarchically structured models, and it provides for a very easy interconnectability of the single model parts to a larger model. Dymola is unable to execute a model, therefore it provides for interfaces for simulation languages, e.g. ACSL. This enables Dymola to turn a model into a specific simulation program. Dymola can be regarded as a very powerful macro handler for simulation languages. Large scale models like powerplant models become much easier to handle if they are coded in Dymola rather than in ACSL.

Yet another part of this thesis expands on modeling techniques by themselves rather than on the modeling tools mentioned so far. The MMS model equations are derived from the basic laws of conservation, like conservation of mass, energy and momentum. In this thesis, the bond graph modeling technique is applied to derive models for power plant simulation. The bond graph modeling technique is based on power conservation rather than on energy conservation. It supports the modeler in building more structured and error-free models, because it is based on a restricted set of well defined and very simple model elements and because the bond graph represents the topological structure of a model as well as its computational structure. Also, the bond graph is a mainly graphical technique and therefore the model structure becomes much clearer from a bond graph than from a set of conservation equations.