Comparative study of modeling methods of bidirectional isolated DC-DC converter for DC grid applications

To simulate a network with high number of appliances, having a good model for each device is of absolute necessity. Dc grids are a type of network which consist of many devices. One of the devices which is commonly used in Dc grid applications is Three Phase Dual Active Bridges (DAB3) converter. This research work investigates three possible approaches to model the DAB3, with the aim of comparing the modeling methods. First, the operating principle of DAB3 is presented. Following this, the DAB3 is modeled using three approaches; namely: Simplified reduced-order state-space averaging (or, State Space Averaging (SSA)) Full Order State Space Averaging (FOSSA) Switching Function (SF) In SSA only the dc component of state variables is considered and since transformer currents have a null dc component, the SSA modeling approach neglects their dynamic. Consequently, output voltage would be the only state variable, resulting in first-order state-space equations, which give the average values of output voltage. In the other words, SSA models DAB3 as a capacitor to which a current is injected. The advantage of SSA is that the DAB3 is modeled with first-order state space equations, making it simple and fast. However, the dynamics of the transformer are not included in the model. To counter this problem, in the FOSSA approach, more terms in the Fourier series of state variables are considered, however this will increase the complexity of model. Therefore, a compromise between accuracy and complexity of the model should be achieved. In DAB3, the transformer currents are quasi-sinusoids consisting of six steps and therefore the fundamental component is the main one. Hence, considering the fundamental component will be a good trade-off between accuracy and complexity of the model as it can model dynamics of the transformer current with reduced numerical calculations. In this work, dc and fundamental components of the output voltages and transformer currents are used to model DAB3 which leads to a seventh-order state space equation. Derivation of state equations, in FOSSA approach, is more complex than SSA, but gives a better description of DAB3 operation. Considering higher-order harmonics of the Fourier series will further increase the accuracy and robustness of the model, but is accompanied with certain drawbacks. For example, consideration of higher order harmonics, makes the derivation and simulation of the model more complex. The main disadvantages of state space averaging, which is a motivation to apply another approach for modeling DAB3, is its sensitivity respect to the circuit configuration. In the other words, whenever there is a minor change in the circuit configuration new state equations should be obtained for describing the new circuit. Therefore, choosing an approach which is simple and easily reconfigurable whenever the circuit configuration change is of great interest to resolve this problem. The modeling approach based on switching function concept has such a characteristic because it model the power converters based on their function rather than their topology. SF models the converter based on its operation principle which make it so flexible. In fact, the main advantage of SF model is its flexibility which enables it to model the converter in different situations and for varied purposes. For example, one may be interested in realizing dead time in the model which is possible only by SF and not SSA/FOSSA. Accuracy of SF model beside its capability to be implemented on FPGA, give a fast and accurate models which is excellent for real time simulation Finally, SSA and FOSSA are suitable for processor based simulation. On the other hand SF is suitable for FPGA based simulations. Detailed comparison of these modeling approaches is presented in this work for which a circuit simulation using PLECS is conducted.