Pareto Optimization to enable set-based Designs of ship systems By Researchers Rounak Siddaiah, Marzieh Karami, Graduate Students-University of Wisconsin-Milwaukee Dr. Rob Cuzner, Assistant Professor-University of Wisconsin-Milwaukee

As a result of the ongoing improvements in the area of power electronics, the penetration of power electronic converter systems into numerous traditional and modern applications has continuously been increasing. An example is the growing utilization of electric propulsion in automobiles, ships and aircraft. These above examples are driven by the increased environmental awareness and the demand for higher functionality of the modern society. Based on the observed trends in state-of-the-art applications, it can be expected that the need for enhanced power electronic converter systems with a higher efficiency, reliability and functionality at concurrently lower costs, weight, volume and development time will further increase in the near future. The analysis of the relevant literature shows that the dimensioning and optimization of converter systems is often carried out with a limited scope and focusing on only one or a few performance criteria. This is often a consequence of a missing systematic approach which enables a comprehensive consideration of the inherent trade-offs between different performance measures (e.g. converter volume vs. losses).

Abstractions, measurements and empirical research Abstractions, measurements and empirical research. The primary aim of the proposed design method is the precise prediction of the converter characteristics which enables an effective and efficient system optimization with fewer hardware prototype iterations and a shorter time to market. The proposed methodology is applied to the multi-objective optimization and comparison of various DC/DC and DC/AC converter systems under consideration of the latest available SiC semiconductors. The case studies on the one hand highlight the capability of the employed methodology to handle complex optimization problems and to find the corresponding solutions with a comparably low computational effort and time need. Experimental measurements on an implemented DC/DC hardware prototype on the other hand verify the high accuracy of the employed models. The results of the two case studies reveal that the employment of SiC semiconductors in the considered applications is attractive despite the higher costs of SiC semiconductors in contrast to conventional Si components. In both examples, the superior performance of SiC enables considerable improvements regarding efficiency, power density and functionality of the converter systems at comparable overall system costs.

Pareto Optimization Objective Function Constraints Design Variables Size Weight Losses Reliability Constraints Ripple current in the capacitors Maximum voltage on the capacitors Losses and thermal constraints Design Variables PEBB types Electrical frequency Output DC voltage level Ambient temperature Output Variables Arm inductor current Submodule capacitor Arm inductor

Accepted design variables Accepted Vaules Module Types PEBB 1000, PEBB 6000 DC Voltage Values (kV) 12, 18, 24, 30, 36 Generator Frequency (Hz) 60, 120, 240, 360 Inlet Cooling Water Temperature (°C) 5-35 The possibility of using diesel generators with flexibility in the nominal electric frequency and voltage levels provide an opportunity to decrease the size of the conversion topology. Therefore, selecting the generator frequency and also the rated voltage of the generator are two other input arguments to make the MMLC design optimal. The other design variable is DC rated voltage. This variable will affect the other aspects of microgrid as well so it can be set as an input value or a variable

Algorithm for designing an optimal MMCs

Different controllers Controller with the goal to keep the DC voltage of submodule capacitors constant: Suitable for Inverters Closed loop scheme that needs measuring capacitor voltage Generate the control signal with nominal average voltage of capacitor

Virtual Prototyping Routine

Modular multilevel convertor

PEBB layout

Conclusion Besides innovation and the employment of MMC technologies, system optimization is a main enabler of improved performance. In this context, the methodology of virtual prototyping is identified to be a powerful tool for systematic optimizations of the converter systems’ primary performance measures, i.e. the efficiency, volume, weight, component costs and the reliability.