A medium voltage DC (MVDC) shipboard power system (SPS) architecture is a notional candidate design for future “all-electric” ships, as it is expected to increase power density and distribution efficiency. However, the lack of experience in proven MVDC technology has inherent risks of substantial delays during the development process of controllers is larger than the risk with AC system designs. Model based system engineering methods, specifically controller hardware in the loop simulations, provide powerful means to lower the risk of a control development cycle. The ability to incrementally increase the fidelity of the model during the development and integration of a controller towards a more accurate representations of the power system network in question is expected to reduce the frequency and severity of potential problems. In the early stages of the control development, a control developer’s main concern is to verify that algorithm implementation meets the functional requirements to satisfy the shipboard power system’s needs. Performance of the implementation is at this stage is of less concern. There may also be limited knowledge about the specific power system that eventually will communicate with the controller. Hence, initially a reduced fidelity model may suffice. As the development process progresses, the model fidelity will evolve to ever increasing fidelity and accuracy. Eventually, the model will have to include all the relevant details and execute in real time in order to test the controller’s real time performance before field deployment. A low-fidelity discrete event simulation first provides insight into the confidence of the algorithm’s ability to satisfy its functions by leveraging its faster than real time nature to perform more experiments than possible in real-time. Subsequently, using a low-fidelity real-time simulation, the developer could capture the elapsed time of characteristic events performed by the controller, characterizing both the algorithm’s functionality and performance with respect to wall-clock time. A first-order dynamic simulation contains a subset of the transient information in a real power system, where a developer can again test both functional and performance requirements in a more relevant environment. Finally, a high-fidelity real-time simulation is the closest model to the real world, permitting a developer to vet their implementation on the most relevant environment that is not a real power system. In order to facilitate such progressive increase in fidelity and real-time capability, an adaptable signal interface between controller and model has been developed and tested. In particular, this thesis presents design criteria and requirements for such an interface focusing on the development of a fault management approach for a breakerless (unit based) MVDC system. Each model must capture the power system characteristics such as network connectivity, power flows, generator outputs, load demands, and electrical switch states, in a monotonic fashion. That is information gradually progresses from the low-fidelity discrete event simulation to the high-fidelity real-time simulation. Each power system model can be on a different operational hardware or run with a different software, requiring an adaptable signal interface to manage the flow of information from simulation environment to the controls. Such an interface must adapt to changing communication behaviors, signal descriptions, and simulator and controller endpoints. Experimental results validate the interface design and control development process. The thesis concludes that the proposed approach is feasible while acknowledging the additional work required to continually refine the process for a more generic use in the future.
