1. Introduction
International maritime transport contributed about 2.89% of the total world anthropogenic greenhouse gas (GHG) emissions in 2018, according to the International Maritime Organization (IMO) [
1]. To contextualize this data, this means that “if global shipping were a country, it would be the sixth largest producer of greenhouse gas emissions. Only the United States, China, Russia, India and Japan emit more carbon dioxide than the world’s shipping fleet” [
2]. Furthermore, if the regulations and agreements on GHG emissions from ships are analyzed, it is found that: “at least until the fall of 2020, the only mandatory regulatory action limiting GHG emissions from ships has been the adoption of the so-called Energy Efficiency Design Index (EEDI) by the IMO, which is an index that measures CO
2 emissions per tonne-mile” [
3]. An intermediate target (in terms of ambition) proposed in the Initial IMO Strategy [
4] is to reduce these emissions by at least 40% by 2030 (i.e., returning to 2008 levels). In this context, it is obvious that developing solutions to reduce the fuel consumption of ships is essential to meet this target.
Road transport is leading the transition to decarbonization. There are many possible pathways to achieving this aim, including strategies such as improvements to aerodynamics, improvements to propulsion efficiency, waste heat recovery, alternative fuels and power sources (hydrogen, biofuels, etc.), hybridization of drive trains and electrification, among many others [
5]. In [
6], it was shown that electrification of road freight is feasible in developed countries. Although electrification of the transport system undoubtedly faces serious challenges, the technologies currently available are mature (e.g., hybrid vehicles and electric vehicles), and developments in the electrification of road transport are attracting attention in other sectors, such as maritime transport. Hybridization, for example, is regarded as a viable and transitory step toward powertrain electrification [
7], with significant potential for power-saving in ships [
8]. Hybrid propulsion systems make it possible to operate the combustion engines within high-efficiency operating points (avoiding low-efficiency operating points by using the battery), thus resulting in energy savings, depending on the characteristics of the driving cycle and the kind of ship (e.g., a range of 9–13% of energy consumption reduction in the case of inland transport vessels [
9] and 2.91–7.48% fuel consumption reduction in short-haul RORO ferries [
10]).
The hybridization of ships requires the appropriate design of integrated electric power systems, which in turn, require the presence of complex control systems. Simulation and hardware-in-the-loop (HIL) testing make it possible to assess how the system responds to perturbations, as well as its functional integration [
11]. Since these tools are well-known in the academic field, it should be fruitful to analyze the recent literature (i.e., published within the last five years and providing sufficient methodological information regarding modeling and simulation/experimentation) to look for the advances made in this area of research.
Focusing first on simulation-based approaches, there has been a large number of such studies [
12,
13,
14,
15,
16,
17,
18,
19], one of the most common objectives of which has been to investigate the performance of hybrid storage systems that can add capabilities to the ship (e.g., mitigating load fluctuations) [
12,
13,
14]. In [
12], equivalent circuit models of the hybrid storage systems under study were constructed using specialized libraries of MATLAB/Simulink. The simulation response of these models was compared to that obtained from an experimental setup composed of real lead-acid and lithium-ion batteries, supercapacitors, a power converter to supply the loads and energy-storage system controllers to control the operation modes of each energy-storage unit. The proper interaction between batteries and ultracapacitors for shipboard electric propulsion systems was studied in [
13], where the large power and torque fluctuations due to propeller rotational motion and waves were solved using two novel energy management strategies (EMSs) based on model predictive control (MPC). Simulation results show that the coordination within HESS provides substantial benefits in terms of reducing fluctuations and losses. Although the software used for the simulations was not specified in the paper, all the system parameters were provided together with the sensitivity analysis of the parameters of the controllers. The research presented in [
13] was continued in [
14], where the proposed control was improved using an adaptive MPC with online identification of the parameters of the hybrid energy storage system. Both a simulation and experiments were performed to show its effectiveness, resulting in a power-loss reduction as high as 15% in the experiments compared to the MPC without online parameter identification. Another common objective of the studies in this category is the optimal scheduling of the power generation of the hybrid ship [
15,
16]. This approach involves complex energy management strategies for optimizing control using different algorithms. However, because their objective is long-term, requiring test periods of several hours, there is a tendency for the models to be simplifications, and consequently, some system variables, such as the duty cycles of the power converters or the current and voltage waveforms of the electrical machines, are not taken into consideration. In [
15], the sampling rate the data loggers used to obtain the loading profile used for the simulations was set to 40 s and the durations of the simulations were 12 and 24 h with time-steps too long to consider detailed models, but this allowed the researchers to demonstrate that the ECMS offered notable fuel savings. Instead of fuel savings, the objective of [
16] was to minimize the operation cost of a full electric-propulsion ship and the GHG emissions. In this case, a sampling time of 30 min and a load profile of 9 h were considered. The proposed algorithm, based on the particle swarm optimization (PSO) method and fuzzy logic, was validated against the results provided by applying dynamic programming to the minimization problem formulated. When the objective of the study is to evaluate a novel control strategy in detail, or to conduct a comparison between different controls, it becomes necessary to develop detailed models of the system components. These models enable high-fidelity simulations to be carried out that test the performance of the control strategies proposed, and which, in some cases, are also validated by experimental laboratory tests [
18]. A meta-heuristic algorithm, specifically a grey-wolf optimization algorithm for power management, was proposed in [
10] for a direct current (DC)-based electric-propulsion ship. The performance of the proposed algorithm was assessed through simulations using MATLAB/Simulink software with detailed models that included power converters, diesel gensets and batteries, and simplified models of the loads (represented by variable resistances). Fuel consumption and emission indicators were compared with the results of classical power management via rule-based control for the same DC system, and of a homologous, conventional alternating-current (AC) propulsion system, showing a 7.48% fuel-consumption reduction. Another DC topology was considered in [
17], but including a fuel cell and photovoltaic panels, to propose a power-management strategy based on decentralized MPC. The MATLAB/Simulink environment was used to study the proposed system under various operational modes. The modeling of the system included nonlinear average switching models of the converters, simplified models for the energy sources where the parameters were taken from commercial components and simplified electric loads that were proposed to be replaced by electric motors with their dynamics in future works. In [
18], a simple and accurate control scheme based on the stabilization of the DC-link was proposed for an electric-propulsion excursion ship with a diesel generator and hybrid energy-storage system composed of batteries and supercapacitors. The model of the system was developed in the MATLAB/Simulink environment using simplified models (e.g., models of power converters based on average behavior) to avoid unnecessary complexity. Despite the simplification, the results provided by an experimental platform supported the satisfactory performance of the proposed control. Finally, inspired by previous automotive research in EMS, two ECMS-based controllers were applied to a ship powered by a hybrid propulsion plant in [
19]. The proposed EMSs had the additional challenge of determining the optimal power-split between different power sources in real-time. The real-time EMSs were compared to a rule-based EMS and dynamic programming was used to benchmark the results. The model of the system was based on a modular, hierarchical and causal modeling approach that allowed for investigating fuel consumption estimation, engine thermal loading and maneuverability, among other variables. The complexity of the model could be summarized as follows: the model consisted of five sub-models with a system of differential and algebraic equations and two sub-models of algebraic equations. MATLAB/Simulink software was used to simulate the experiments for the different control strategies. Simulation results demonstrated that fuel savings of 5% to 10% can be achieved with the proposed methods, within 1–2% of the global optimum solution.
Moving on now to HIL testing, because this technology is relatively new, there are fewer, albeit more recent, studies [
20,
21,
22,
23,
24,
25,
26,
27]. HIL testing is commonly applied in prototyping new components/subsystems to demonstrate their applicability to real environments [
11]. In the case of ships, it allows short circuits and other transients to be tested without the dangers and cost involved in testing real high-power hardware. In [
20,
21], HIL was used to validate the power systems and EMSs of electric ships based on MVDC. In the case of [
20], a novel EMS based on fuzzy logic was compared with a classic control system based on PI controllers by real-time HIL simulations along with experimental validation using controller hardware-in-the-loop (CHIL). A different approach was carried out in [
21], where the power components were emulated using reduced-scale components (propulsion motor, propeller machine and power converters), and the control algorithm was implemented in a dSPACE HIL platform communicating with them. A different ship topology was studied in [
22], in which an AC microgrid was used for the integration of different renewable sources and storage systems in the ship’s power system. HIL simulations using the OPAL RT-Lab platform were carried out to validate the proposed control algorithm (sine cosine algorithm based on wavelet mutation, SCAWM), taking into account that the HIL method provides real-time analysis and considers errors and delays that do not exist in offline simulations. The results showed that the proposed EMS was able to offer an effective trade-off between power generation and load, and thereby maintain the quality of power and frequency deviation within the desired limits. Less detailed models were considered in [
23] for a hybrid-electric ship consisting of a diesel engine and a battery on an AC network. Given that the objective of the EMS was to optimize fuel consumption and GHG emissions in the long term (the study analyzed these variables over a year of operation), the power system models needed to be simplified. The HIL experiment used the dSPACE PX10 platform for emulating the ship’s power system (modeled using MATLAB/Simulink), a dSPACE MicroAutobox for running the EMS and a desktop for monitoring the experimental variables. In a similar vein, when the objective of the research is to find an optimal power flow over a long period, the level of detail required of the models for the real-time simulations is reduced [
24,
25]. In [
24], the objective was to demonstrate the operability of an emission-free ferry using a propulsion system based on fuel cells and batteries. In this case, real sailing routes were used to perform the HIL simulations. A similar study for another fuel cell-battery ship was carried out in [
25]. Another kind of DC-based full-electric ship supplied by batteries and supercapacitors was considered in [
26] for testing a power-distribution strategy in DC networks based on the virtual impedance method. In this case, the HIL experiment was performed in a StarSim HIL platform, in which the power system was emulated with detailed models, and digital signal processors were employed to implement the proposed control. Finally, [
27] is given as an example of the power hardware-in-the-loop (PHIL) approach. This approach combines HIL technology with high-power components, and can be considered as the step after HIL testing along the path to prototyping. The laboratory described in [
27] was rated as 5 MW at DC voltages from 6 to 24 kV and was used for the development of projects related to naval research such as MVDC systems for all-electric ships.
Following a review of the literature on HIL real-time simulations of electric-propulsion systems for ships, the main contributions of this paper are the following:
The development of a high-fidelity benchmark for hybrid-electric vessels using diesel generators and batteries. This benchmark consists of detailed models, for which the parameters are provided, thus enabling the models to be reproduced on other platforms.
The validation of the proposed hybrid-electric ship topology and control system using real-time HIL simulations on a Typhoon HIL402 platform.
The incorporation of the maximum degree of complexity allowed by the HIL platform. The electrical topology of the proposed benchmark has not been commonly employed in previous studies using HIL technology, probably because the complexity of the models has stretched the computational capacity of commercially available HIL platforms. To circumvent this limitation, previous studies have often simplified parts of the model (e.g., reducing the system from three-phase to one-phase, averaging power converters, etc.), to allow the inclusion of more elements in the system while neglecting some phenomena that can appear in the real world.
The application of the EMS to maritime transport. Although the proposed system itself is not novel (in that it follows the rule-based strategy used in commercial hybrid-electric vehicles), it has never before been implemented in ships, nor has its performance been validated by HIL technology.
The main aim of this study was to validate the proposed hybrid-electric topology with real-time HIL simulations and high-fidelity models. The results provided by the HIL simulations show the operation of the different components of the system in detail, and will be valuable should these systems reach the prototyping stage. Furthermore, the study also evaluated the response of the system to potential disturbances resulting from whatever EMS is used. The results show that the system performed correctly during these transitory events with no undesirable responses.
The paper is organized as follows. The introduction above locates the research in the context of the current literature on hybrid ships and sets out its main contributions, together with the aim of the study.
Section 2 below provides all the information relating to the models developed in this work, including parameters, mathematical expressions and block diagrams, to allow the results to be replicated. In addition, the energy management system (EMS) used for the validation is described in-depth.
Section 3 presents the results obtained from the real-time HIL simulations. It describes two case studies, one analyzing the operation of the system under steady-state conditions and the other analyzing the transients resulting from changes in the different modes of operation controlled by the EMS. These results are discussed in
Section 4. Finally, the last section presents the conclusions drawn from this work.
4. Discussion
Here, analysis of the results follows the same pattern as in the last section, taking the stationary operation first and then transient.
Concerning stability, it can be seen that operation of the battery system for both load support and power recovery affected the harmonic distortion of the currents supplied by the generator. However, the system was able to supply the required torque to the propulsion system under all circumstances. Regarding the harmonics observed in the generator currents, since this is the main power source of the system when the battery system is disabled, the harmonic components that appear, in order of importance, are the 5th, 7th, 11th and 13th. They are injected by the electrical drive of the propulsion system. The same harmonic components appear when the battery provides support to the diesel system, but as magnitudes increase, so too does the harmonic distortion. This is an effect that must be taken into account when hybrid systems are proposed. Conversely, when the battery is being charged, the same harmonic components appear, but with a reduction in magnitude. Nonetheless, in all cases, these harmonic currents can be reduced by increasing the switching frequency. In this study, the switching frequency was set to 10 kHz, but it may be increased up to 16 kHz without incurring increased switching losses.
The results of the scenarios studied for transient operation showed that the electrical power system performed appropriately when the EMS commands changed the mode of operation of the system components. Before the development of the model and its EMS, it might have been hypothesized that sudden disturbances—such as connection/disconnection of the battery system, or a sudden propulsion power increment—could have resulted in undesirable spikes, power drops or power unbalances, among others. However, after a review of the results for each of the scenarios studied (sudden increments of speed command, connection of the battery system for charging, and transition from a battery as the main power source to a diesel generator), it can be concluded that the behavior of the system did not show an anomaly. The increments of the current magnitudes during these transitions were within adequate limits, the duration of the transients short and no oscillations observed.
In future studies, the parallelization of HIL devices will be considered to increase the real-time simulation capacity and increase the switching converters in the model. This will enable the passive rectifier of the propulsion system to be replaced by an active rectifier, and regenerative braking could also be studied in detail. Furthermore, once the model has been validated, we intend to develop more complex EMSs and test them using control hardware-in-the-loop (C HIL), then implementing the EMSs in external controllers interacting with the HIL device.