The use of the computer as an aid to the difficult process of abstracting a physical reality to a mathematical model is a must if simulation is to have a prosperous future. However, the effective use of and development of reusable models relies on a consistent framework for model-representation to be successful.
The central theme of this paper is to present a model library for simulation of diesel engine systems using the bond graph method as the unifying and consistent framework for model representation.
2. MODEL LIBRARY
The models developed are all added into a component model library supported by the modeling environment MSI - Modeling System 1 [3]. This general modeling environment supports multiple input/output formalism, making it ideally suited for modeling of complex systems. Also its large flexibility in integrating models from different sources are emphasized. Complete models or submodels can be prepared for different simulation environments such as ACSL [1], Matlab/Simulink [2] and several others. A description of a complete modeling and simulation environment established by the authors are given in [4, and the structure of the library is further discussed in [5]. The great flexibility and modularity of the model representation makes it ideal for representing a modelers model library.
In this work the diesel engine models developed are all entered into this modeling environment and as such creating a model library for development of diesel engine models. Models and model skeletons currently available are:
Cylinder:
・4-stroke cylinder
・2-stroke cylinder with ports
・Fake cylinder (copies master cylinder)
・Air/Exhaust valves
・Heat transfer
・Combustion model
・NOx models
・Slider-Crank mechanism
・Engine friction model
Turbocharger:
・Axial flow turbine
・Radial flow turbine
・Radial compressor
・Turbocharger rotor with inertia
・Bypass valve
Air and exhaust system:
・Air receiver
・Exhaust receiver
・Air cooler
・Wave action pipe models
・Air filter
Shafting system:
・ Stiff single inertia shaft
・ Lumped model for torsional vibration calculations
・ Normal mode model
・ Gear
Auxiliaries:
・RPM Controller
・Load (Propeller, Generator)
Utilizing the model library together with efficient modeling software, has reduced model development time by a large margin. Now complex models can be developed within hours without sacrificing the flexibility needed to approach new problems.
3. MODELING FRAMEWORK
Bond graphs are a unified graphical notation for the representation of physical systems [6] . The basic concept is that power is the fundamental constraint in a physical system and is the one variable that is common to the whole system.
A bond is considered to conduct power between the ports instantaneously and without loss. Each bond has an effort e, and flow f, variable attached, and in true bond graphs the product of the two variables is the instantaneous power exchange between two systems.
Two other types of variables which are important are momentum p, and displacemen, q, in generalized notation. Momentum is defined as the time integral of an effort and the displacement variable is the time integral of a flow variable. Momentum and displacement variables can be used to represent system energy and thus state variables.
Using the classification of power and energy variables presented previously, it turns out that only nine basic types of multiparty elements are required in order to represent models in a variety of energy domains. These multi-ports function as components of subsystem and system models. They are, in many cases, idealized mathematical version of real components, and in other cases they are used to model physical effects. The nine basic multi-ports are classified in Fig. I according to the way they process energy. Roughly speaking, these elements account for energy supply, dissipation, storage and transduction from one model to another (i.e. the conversion of energy from one form to another).
Using these basic elements, a graphical model can be constructed of a system. It is a remarkable fact that models based on apparent diverse branches of engineering science all can be expressed using the notation of bond graphs. This allows one to study the structure of a system model. The nature of the parts of the model and the manner in which the parts interact can be made evident in a graphical format.
Using bond graphs, the differential equations can be unambiguously constructed in an algorithmic manner directly from the graph, preferably using a computer program. However, before the equations can be written from the bond graph, causal information must be added to the graph. Causality means obtaining an explicit indication of which variables for an element is to be considered independent and which are to be considered dependent.
