This paper proposes an energy management system (EMS) for a fuel cell hybrid electric vehicle (FCHEV). The EMS is based on nonlinear model predictive control (NMPC) and employs a recurrent neural network (RNN) for modeling a proton exchange membrane (PEM) fuel cell (FC). The NMPC makes it possible the formulation of control objectives not allowed by a linear model predictive control (MPC), such as maximum efficiency point tracking (MEPT) of the FC, while the RNN can accurately predict the FC nonlinear dynamics. The EMS was implemented on a low-cost development board, and the experiments were performed in real-time on a hardware-in-the-loop (HiL) test bench equipped with a real 3 kW FC stack. The experimental results demonstrate that the NMPC EMS is able to meet the vehicle's energy demand, as well as to operate the FC in its most efficient region. Moreover, a comparative study is performed between the proposed NMPC, a linear MPC and hysteresis band control (HBC). The results of this comparative study demonstrate that the NMPC provides a better fuel economy and can reduce FC degradation.

The main objective of this paper is to select the optimal configuration of a ship’s power system, considering the use of fuel cells and batteries, that would achieve the lowest CO₂ emissions also taking into consideration the number of battery cycles. The ship analyzed in this work is a Platform Supply Vessel (PSV) used to support oil and gas offshore platforms transporting goods, equipment, and personnel. The proposed scheme considers the ship’s retrofitting. The ship’s original main generators are maintained, and the fuel cell and batteries are installed as complementary sources. Moreover, a sensitivity analysis is pursued on the ship’s demand curve. The simulations used to calculate the CO₂ emissions for each of the new hybrid configurations were developed using HOMER software. The proposed solutions are auxiliary generators, three types of batteries, and a proton-exchange membrane fuel cell (PEMFC) with different sizes of hydrogen tanks. The PEMFC and batteries were sized as containerized solutions, and the sizing of the auxiliary engines was based on previous works. Each configuration consists of a combination of these solutions. The selection of the best configuration is one contribution of this paper. The new configurations are classified according to the reduction of CO₂ emitted in comparison to the original system. For different demand levels, the results indicate that the configuration classification may vary. Another valuable contribution of this work is the sizing of the battery and hydrogen storage systems. They were installed in 20 ft containers, since the installation of batteries, fuel cells and hydrogen tanks in containers is widely used for ship retrofit. As a result, the most significant reduction of CO₂ emissions is 10.69%. This is achieved when the configuration includes main generators, auxiliary generators, a 3,119 kW lithium nickel manganese cobalt (LNMC) battery, a 250 kW PEMFC, and 581 kg of stored hydrogen.

Towards a future with more renewable energy, oil and gas (O&G) will still play a major role in the energy and mobility sectors. Therefore, scientists must also investigate ways to mitigate carbon emissions in O&G production. In this sense, a power hub with local generation can be employed in offshore production sites to allow the adoption of more efficient power generation technologies without the weight and space constraints that exist in usual Floating Production, Storage and Offloading (FPSO) platforms. This power hub can be connected to the FPSOs forming an isolated offshore power grid that requires further study. This work investigates stability issues of such offshore power networks. Simulations of a system composed by the power hub and three identical FPSO units were performed in PSCAD, and the stability of the system was validated according to the IEC 61892-1 standard. Results demonstrate that it is possible to operate such system with a stable and secure supply. The main contributions of this work are the electrical modeling of the power hub and of the resulting isolated offshore electrical grid, and a detailed discussion of the rising challenges and the required models for dynamic electrical studies.

This work proposes a Neural Generalized Predictive Control (NGPC) for tracking the maximum efficiency or the maximum power point of proton exchange membrane fuel cell (PEMFC) stacks. The NGPC consists of a nonlinear type of model predictive control (MPC) in which the prediction model is a neural network (NN). Such control strategy combines the advantages of PEMFC modeling using NNs and the MPC family of controllers. NNs are capable of modeling the whole nonlinear dynamics and the time-varying behavior of PEMFC stacks, while the MPC can easily handle system constraints and it manipulates multiple control variables of complex systems. Through its optimization algorithm, the NGPC determines an optimal current necessary for the stack to operate at maximum efficiency or maximum power. This paper demonstrates the operation of the proposed control technique through a simulation study of a grid-connected PEMFC power generation system.

Hydrogen-related technologies are a promising solution for energy storage systems. The energy surplus from renewables or the off-peak power in a centralized electric system can be used to feed an electrolyzer to produce hydrogen. Then, the stored hydrogen can be converted to electric energy using proton exchange membrane fuel cells (PEMFCs), which have higher efficiency than other conventional energy conversion technologies. However, a PEMFC stack might be subject to variations in its energy conversion efficiency due to undesirable changes in the internal operating conditions, which may lead to a relatively high loss of generation capacity and an increase in the hydrogen consumption. A system identification method using recurrent neural networks (RNNs) able to build prediction models for the efficiency is proposed in this paper aiming at a further use in a control application to maximize the stack's efficiency in real-time. A 3 kW commercial PEMFC stack operating under a typical residential load profile is employed in this work to obtain experimental data. Three models are developed with the aforementioned method using data sets with different sizes for training the RNNs. The results show good agreement between the predicted and the measured efficiency, with a coefficient of determination (R² coefficient) higher than 0.999 and a validation mean squared error (MSE) lower than 0.02 for most of the models. The results also reveal that, for a fixed prediction horizon of 6.4 h, all the three models present similar performance, with the size of the training data sets making little difference in the performance indicators.