Extending simplified large-scale power system models

A method for predicting the transient and steady-state characteristics of electrical power systems on Spice is presented. The analysis method presented significantly reduces computer memory requirements and run-time. The results correlate with more detailed models that require extensive computer resources and significant run-time. This Paper extends the original work presented in a 1992 IECEC paper written by the authors [6]. Specifically, the negative impedance characteristic of constant power loads has been added to the technique. This provides a more realistic simulation for electrical power systems that are dominated by load power converters. When simulating an electrical power system, each component must be carefully considered. This includes parameters describing cable characteristics (e.g., line inductance, capacitance, and series resistance) as well as detailed descriptions of all load and source input filters, and transfer functions (characteristics). In most circumstances, each of these power system elements is described as a sub-circuit. These sub-circuits are then connected together in a single macro-model. This single model may contain over 1000 nodes. A method to simplify this model can reduce the number of nodes by a factor of 10 or more. The method presented is relatively simple and allows the user to obtain system level characterization with reduced turn-around time (from hours to minutes). The major advantage of this technique is that the simulator itself can be used as a pre-processor to aid the engineer. Classical circuit synthesis techniques with extensions are then applied to reduce the model. Examples will be presented which illustrate this technique. Comparisons in results and run-time will be included. By using this technique, system level performance prediction can be obtained in a much more efficient manner while still providing the required accuracy. This extends the capability and utility of both simulators and hardware platforms.