Operation and Control of Triple Active Bridge DC-DC Converter for Remote DC Microgrids

Abstract

In recent years, DC microgrids (MGs) have emerged as a practical solution for providing power to remote areas that lack access to the utility grid. These systems rely on the effective integration of renewable energy sources, such as photovoltaic (PV) systems, with energy storage systems (ESS) to ensure a continuous power supply. The performance of the MG is heavily reliant on the power electronic converters that interface these components. This work focuses on the modeling and control of a DC MG that utilizes the Triple Active Bridge (TAB) converter. The TAB is a multi-port topology that can integrate and manage power flow between a PV source, an ESS, and the DC bus through a single, isolated conversion stage.

In grid-connected systems, a grid-tie converter typically regulates the DC bus voltage. However, because the remote microgrids in this work operate in islanded mode without external grid support, the TAB converter must autonomously regulate its own output voltage to ensure stability. To achieve this autonomous regulation, a cascaded control architecture is implemented as a baseline. The primary objective of this control is to maintain a stable DC bus while maximizing the energy harvested from the PV array. Therefore, the control is structured to operate the PV port in Maximum Power Point Tracking (MPPT) mode, while the ESS is tasked with dynamically compensating for power mismatches to regulate the DC bus voltage. A fundamental challenge in this system is the inherent cross-coupling between the ports of the TAB converter. To address this, this work details the derivation of the converter's average and small-signal models, which provide the basis for control analysis. A decoupling matrix, derived from these models, is implemented to enable independent control of the PV and ESS port currents.

A significant operational challenge in an islanded MG occurs when the ESS reaches its state-of-charge (SOC) limits and can no longer absorb excess power. This condition can lead to DC bus overvoltage if the PV system continues to operate at its MPP. To solve this, a supervisory coordination control strategy is proposed. This controller monitors the system's status and, upon detection of a full ESS, seamlessly transitions the PV from MPP mode to a non-MPP mod. In this mode, the PV system curtails its output to balance the power and maintain the stability of the DC bus. The performance of this coordination strategy and the mode transition are confirmed through hardware experiments.

Furthermore, the intermittent nature of solar irradiance causes rapid PV power fluctuations that induce transient voltage deviations on the DC bus, which conventional feedback loops cannot fully suppress. To address this, a proactive feedforward control method is investigated. After showing that a simple constant-gain feedforward is insufficient for systems with unequal controller dynamics, the primary contribution of this work is presented: the complete analytical derivation of a dynamic feedforward gain. The derived gain's action is adjusted based on the different response characteristics of the PV and ESS current controllers allowing for more effective disturbance cancellation. The effectiveness of this method in reducing voltage transients is validated through simulations using The National Renewable Energy Laboratory (NREL) solar data and extensive hardware experiments.

Finally, to address the need for higher power capacity and enhanced reliability beyond what a single converter can offer, a scalable architecture based on parallel-connected TAB converters is presented. This modular architecture consists of multiple TAB units connected to a common DC bus, which provides redundancy and allows the system's capacity to be expanded. A higher-order model for the complete parallel system is developed, and a centralized control strategy is implemented to ensure both the regulation of the common DC bus voltage and proportional current sharing among the parallel modules. The performance of this scalable architecture is validated on a hardware prototype.