The objective of this report is to set forth a group of time-domain models for the early-stage design study of shipboard power systems, and to demonstrate their use on various system architectures. The effort stemmed out of an earlier effort in which waveform-level models of three notional architectures – a Medium Voltage AC System, a High-Frequency AC System, and a Medium Voltage DC System were partially developed. Unfortunately, these codes were extremely computationally intense, limiting their usefulness for early design studies in which large numbers of runs, and a degree of user interactiveness, is required.
This effort again considered three systems - a Medium Voltage AC System, a High-Frequency AC System, and a Medium Voltage DC System. However, this effort was focused on simplified models to serve the needs of interactive early stage design. Within this context, this work focused on two distinct aspects. The first aspect was the development of a set of component models in mathematical form. Such a description is advantageous in that it is language independent. The second aspect of this effort was the use of the component models to study the three aforementioned systems.
With regard to the first aspect of this work, that is the component modeling, fundamental component models (in mathematical form) were defined in sufficient detail to represent a notional system using the three aforementioned architectures. The component models are highly simplified abstractions of shipboard power system components. The motivation for the simplification is two-fold. First, at an early design stage it is doubtful if the parameters needed for a more detailed system representation would be available. A highly detailed simulation would be based on many assumptions leading to results which are no more indicative of actual performance than a highly simplified simulation. The second reason for the creation of highly simplified model is for the sake of computational speed, so that system simulations based on the component models will run at speeds compatible with the needs of exploring the system behavior under a large variety of conditions.
The types of model simplifications used are three-fold. First, throughout this report average-value models are used. In particular, the switching of the power semiconductors is only represented on an average-value basis. Secondly, reduced-order models are typically used. Thus, high-frequency dynamics have been neglected. Simulation based on these models cannot be used to predict behaviors such as the initial response to a fault. In general, temporal predictions of features on a time scale of ~100 ms or less will not be reliable. The third simplification that has been made is that many components are represented in the abstract based on the operation goals of the component rather than on the details of what might physically be present.
The set of models provided herein is fairly extensive and adequate to serve as a basis for studying a variety of power system architectures. The models set forth include: turbines, turbine governors, wound-rotor synchronous machine based ac generators, generator paralleling controls, rectified wound-rotor synchronous machine based dc generation systems, ac input permanent magnet synchronous machine based propulsion drives, dc input permanent magnet synchronous machine based propulsion drives, hydrodynamic models, ac and dc pulsed load models, isolated dc/dc conversion models, dc loads, non-isolated dc/ac inverter modules, ac loads, active zonal rectifiers, circuit breakers and controls, as well as a variety of supporting components.
For the purposes of brevity and because of the resources available, model validation results are not presented herein. However, comments on model maturity have been included with each component to provide the reader with a sense of the degree of model confidence for each component.
The component models developed under this effort were set forth in a previous report, “Notional System Report,” but are included again herein in Appendix A. The remainder of this effort focused on applying the component models in Appendix A to MVAC, HFAC, and MVDC instantiations of a notional architecture. Unfortunately, this aspect of the effort was only partially successful. It is demonstrated herein that the models developed are computationally effective; however, in the case of all three systems either only partial system models are used or there are remaining simulation issues to be resolved.
The primary reason for this failure to completely model the systems was related to the choice of simulation engine. In particular, Simscape was used. This proved to be rather trying on the part of the simulationists involved in this effort. The reason that Simscape was chosen were the results of a study of a small notional system also considered by the group, and set forth in . Therein, it was shown that Simscape yielded superior performance over a number of simulation implementations. Unfortunately, as the system scope grew, this language proved problematic. In retrospect, it is recommended that Simulink be used instead, preferably with a user defined solver of the network interconnection equations, also as described in . Such an approach should be much more robust, and more open to modification if needed.
- M. Andrus
- M. Bosworth
- J. Crider
- H. Ouroua
- E. Santi
- S. Sudhoff