Power Density Optimization of a DC/DC Converter for an Aircraft Supercapacitors Energy Storage Niklas Fritz, Mohamed Rashed and Christian Klumpner The Department of Electrical and Electronic Engineering The University of Nottingham, United Kingdom niklas.fritz@rwth-aachen.de, mohamed.rashed@nottingham.ac.uk, christian.klumpner@nottingham.ac.uk Abstract—This paper presents a computationally efficient de- sign algorithm for a DC/DC converter, optimized for power density. Interleaving and series-connecting of several converter cells is included in the analysis. The choice of the converter design parameters, such as the semiconductor devices or the number of interleaved converter cells may have unpredictable impacts on the resulting power density. Therefore, fully analytical models of the operating waveforms, the converter losses and also the converter weight are introduced. Using those models, an algorithm is developed to optimize a 2 kW DC/DC converter for an aircraft supercapacitors energy storage system. Index Terms—Synchronous Buck Converter, Modelling, Opti- mization, Power density, Supercapacitor, More Electric Aircraft I. I NTRODUCTION The More Electric Aircraft (MEA) is a new concept in the development of next-generation aircrafts [1, 2]. Traditional hydraulic and pneumatic actuators used to drive critical fly- ing surfaces are increasingly replaced by electrical systems, which have the potential to be significantly more efficient and light-weight, which would reflect in lower fuel consumption. However, adopting the MEA concept imposes challenges to the design of the electrical power system as its power rating significantly increases. Therefore, it is crucial that the electri- cal components are optimized for power density. This work is related to the design of a supercapacitor energy storage system which is needed to smoothen the ripple of the power system, deal with regeneration and enhance the power surge capability. The system will consist of a supercapacitor stack, interfaced with the aircraft’s 270 V DC bus via a bidi- rectional non-isolated DC/DC converter and the synchronous buck converter has been chosen as the preferred topology for this study. In addition to the generic topology, more complex variants such as interleaved (parallel channels) and cascaded (series connection of cells) converters are considered. Deciding on design parameters of the converter is a complex task. The choice of semiconductor devices, the number of in- terleaved half bridges, the amount of inductance, the switching frequency, all those parameters have great influence on power density which are hard to predict. The approach in this paper helps to establish a link of analytical equations between the design variables and the power density. The synchronous buck converter, although being a hard- switched topology, has been shown to achieve attractive power densities [3, 4] when using wide-bandgap (WBG) power semiconductors operating at hundreds of kHz which enables minimization of magnetics whilst the cooling requirements remain reasonable due to the lower switching losses. Analytical models providing an evaluation of current and voltage waveforms in this topology and the evaluation of the losses have been investigated in [5–7], also weight minimiza- tions have been proposed in [7, 8]. The benefits of interleaving have been analyzed in [6, 9–11]. This paper adds the analysis of cascaded converters and further modelling of the converter weight. Moreover, the ripple charge of the input capacitor is analyzed for interleaved converters. The optimization procedure in this work is based on fully analytical modelling and consists of three layers, as proposed in [12]: First, the converter current waveforms are analyzed and key variables, such as for example RMS or ripple currents, are derived. Based on these, an analytical loss model of both active and passive components is developed. A model to estimate the weight of the main components is added, which, for example, links the converter losses to the weight of the required cooling system. As the adopted models are fully analytical, there is no need to run time-consuming simulations in time domain. Hence, the major advantage of this technique is its speed in calculation whilst being able to assess a wide range of design configurations. This approach has some limitations, for example the high-frequency losses in the magnetics, which are difficult to assess analytically. The paper is structured as follows: In section II, the syn- chronous buck converter topology and its variations are intro- duced. Sections III, IV and V demonstrate the modelling of the operating waveforms, the losses and the weight, respectively. Section VI presents the optimization procedure. In section VII, the results are summarized and the conclusions are presented in section VIII. Figure 1: The generic synchronous buck converter topology