Introduction
According to recent data [1], the electrification of transportation is key to reduce the greenhouse gas emissions. Some of the main challenges for the wide adoption of Electric Vehicles (EVs) are to extend the range of the vehicle and reduce the time required to fully charge the battery. This implies the development of battery systems which present at the same time an increased energy density, specific energy, and power capability [2].
To achieve such objectives, the global research in battery systems focuses on several paths. On the one hand, new battery chemistries are being developed trying to reach both energy and power requirements at cell level [3]. On the other hand, Hybrid Energy Storage Systems (HESSs) typically combining two cell technologies are being proposed, allowing to achieve energy and power requirements at system level [4], [5]. Although several cell technologies are already commercially available to reach high power capabilities (e.g. supercapacitors [6] and Lithium-titanate-oxide batteries [7]), the development of cell chemistries, presenting increased energy density and specific energy, remains challenging. Among the different possible solutions, only Li-metal anode technologies seem to be able to provide a satisfying improvement in the short- middle-term [8]. Particularly, Lithium-sulfur (Li-S) technology is considered as a promising candidate due to its outstanding theoretical gravimetric capacity [9]. Such technology, which typically combines Lithium-metal anode with a sulfur-based cathode [10], is gaining attention within the battery research community [11], [12] and early commercial products were already introduced in the market in the past few years.
An additional challenge in HESS solutions is that power electronics interfaces are typically required to connect both battery technologies, and implement energy management strategies to ensure the optimal use of both battery technologies [5], [13], [14]. Such additional circuitry could compromise the energy density and specific energy improvements achieved through battery chemistry hybridization. Therefore, the development of reduced volume and weight power converters is of high importance [5], [15]–[17]. Within this context, the use of wide-bandgap semiconductors such as Gallium Nitride (GaN) devices presents an increased interest. In fact, their low conduction resistance and high-frequency operation characteristics allows to derive high efficiency and reduced size power converters [15], [17]–[19], improving in turn the efficiency, energy density and specific energy at system level.
Several HESS solutions were already proposed in the literature [5], [20]–[22]. Nevertheless, most of them were focused on improving the power capability of conventional Li-ion technology through the incorporation of supercapacitors, neglecting the need to improve the energy requirements of state-of-the-art systems. Moreover, most papers only involved simulation results of system behavior, while experimental results are unusual in the literature. Finally, no special attention was paid on the impact of the power converters themselves on system energy density and specific energy.
The present study is based on the analysis of the experimental results of a HESS prototype with high energy and power performance under different automotive profiles. The HESS system is composed of a LTO anode Lithium battery that guarantees high power requirements and a module of the novel Li-S technology with the aim of increasing the specific energy. A bidirectional power converter based on GaN semiconductors was developed to connect both batteries.
Therefore, the present paper aims to bring several contributions to the literature:
The proposed HESS integrates the latest Li-S technology, which enables to increase the energy density and specific energy of the system.
The impact of the power converter stage is considered and minimized through the development of high-efficiency and reduced size GaN-based power converter.
The developed HESS prototype is experimentally validated under different EV profiles, and the obtained results are thoroughly evaluated.
This paper is structured as follows, Section II describes the main characteristics of the selected Li-ion and Li-S cell and modules, as well as the adopted hybridization topology. Section III emphasizes on the performance evaluation and experimental validation of the developed GaN-based power converter. Section IV details the control of the system, including the power converter control loop and the high-level energy management strategy. Section V shows the experimental results obtained with the developed HESS prototype, submitted to different automotive power profiles. Section VI discusses the obtained results and the main limitations and further works of the study are highlighted. Finally, Section VII draws the main conclusions of the study.
Hybrid Energy Storage System
This section aims at describing the different components of the developed HESS prototype, specifying successively the main characteristics of the Li-ion battery module, Li-S battery module, and selected coupling topology. Figure 1 shows the battery cells used for the development of both battery modules, while the Table 1 summarizes the characteristics at cell and module level.
Commercial battery cells used for the HESS: (a) Li-S pouch cells, and (b) Prismatic LTO Li-ion cell.
A. Li-S Battery Module
As previously mentioned, the Li-S technology which combines Li-metal anode with a sulfur-based cathode, was selected to increase the energy density and specific energy of the HESS. Commercial pouch cells [see Fig. 1(a)], were used to develop the Li-S module. Although such cells are early commercial products of a technology under development, they already present increased energy characteristics compared to state-of-the-art commercial Li-ion technology. It is noteworthy that the nominal capacity of 14.7Ah indicated in Table 1 corresponds to the capacity extracted through 0.2 C discharge. Discharging curves obtained with these cells (from laboratory tests and represented in datasheet) reveal important capacity decreases for increasing C-rates.
The developed Li-S module consisted of 24 cells connected in series, incorporated within a house-made mechanical packaging frame for laboratory use (see Table 1).
B. Li-Ion Battery Module
The role of the Li-ion component within the developed HESS prototype is to provide power capability to the battery-pack. For this reason, the selected Li-ion cell technology should be oriented to power application. Among the different Li-ion chemistries, the LTO anode is particularly suited for high-power applications [23], due to the reduced formation of Solid Electrolyte Interface (SEI) and dendritic Lithium, which enhance the potential for high rate capability [10].
For this reason, a 23 Ah prismatic cell [see Fig. 1(b)], composed of a LTO anode and Nickel-Manganese-Cobalt (NMC) cathode, was selected for the development of the Li-ion module to be included in the HESS battery-pack. In order to compose the battery module, 24 cells were connected in series (see Table 1).
C. Hybridization Topology
Several hybridization topologies were proposed in the literature to combine energy storage systems, involving different coupling methods with the load system and between both storage systems [4], [24]. In EV applications, the HESS is typically connected with the electric engine by a DC-AC converter [25]. Moreover, the connection between both storage systems could be broadly classified between passive, active or semi-active (cascade) coupling architectures [4].
On the one hand, passive architectures imply direct connection between the storage systems, without any conversion stage. This allows to save the cost and power losses related to power converters. For EV application, removing power converters also implies to increase the energy density and specific energy of the HESS. Nevertheless, passive architectures do not permit to control the power demand for each storage system, leading to non-optimal operation and accelerated degradation of each individual battery system. The failure of one battery could also affect the other one and compromise the performance of the whole system [4], [24].
On the other hand, active architectures typically include DC-DC power converters associated to each individual storage system before the DC-AC conversion stage to the engine. This makes the implementation of advanced energy management strategies possible, as the power demand of each storage system could be actively controlled in accordance with its intrinsic characteristics and optimal operation boundaries. Nevertheless, this is achieved at the expense of increased system costs, weight, volume, and power losses. Finally, semi-active (cascade) architectures consist on coupling both storage systems in series, introducing a DC-DC power conversion stage between them. Such an approach represents a suitable trade-off for EV applications, providing enough flexibility to directly control the power demand of one storage system, and ensuring at the same time reduced system costs, weight, volume, and power losses compared to active solutions [26].
Within the context of this research, a semi-active topology was designed, as illustrated in Fig. 2. In fact, the early maturity stage of the Li-S cell technology requires careful operation to minimize degradation [27]. This makes the use of a power converter necessary for optimal control of the Li-S system. In order to minimize the weight, volume and power losses of the system, and do not compromise the high energy density and specific energy obtained through Li-S cells integration, a high-performance power converter is required.
Semi-active HESS for EV: (a) EV power distribution system, (b) interleaved bidirectional buck-boost converter based on GaN devices for Li-S and Li-ion semi-active hybridization and (c) buck-boost key waveforms, interleaved carriers, branches duty cycles (
Bidirectional Dc-Dc Converter
Considering characteristics of batteries presented in Table 1, specifications of the power converter are derived in Table 2. The non-inverting buck-boost converter topology was considered in this work, due to its simplicity and similar battery voltage levels. Fig. 2(b) shows the power converter which is composed of two half-bridge branches interconnected by an inductor. These branches are named as primary and secondary considering a power flow from Li-ion battery to Li-S battery. The non-inverting buck-boost converter can operate in different operation modes [15], [28]. Operation modes of the power converter are related to voltage relation, along with primary (
This converter is operated in a Continuous Conduction Mode (CCM), avoiding high current stresses and allowing simple linear control [see Fig. 2(c)]. As a result of the required current ripple and power level, interleaved solution is presented as the preferred choice. Fig. 2(c) illustrates the key waveforms under CCM operation in one-switching period. With the interleaving solution the current ripple is reduced
Thus, the aim of this section is to evaluate the performance of the designed power converter in comparison to Si-based solution. Besides, experimental measurements are included in order to validate the performance of the presented solution for the HESS requirements.
A. Power Converter Performance
When analyzing the performance of a power converter, semiconductors and inductive components are the major contributors of power losses, volume and weight. Thus, the impact on the performance of these components was modeled, estimating power losses of semiconductors along with weight and volume models. These models are based on updated commercial database for inductive components and heatsinks [15]. The presented approximation models are suitable for general optimization routines, reducing long convergence times, as it is presented in [29] for medium frequency high-power converters.
The performance of the power converter was evaluated considering the following characteristics:
Cooling system is based on natural convection;
ceramic capacitors with negligible power losses and volume impact as input and output capacitors;
commercial 200 V Si OptiMOS 3 device (Infineon);
low-voltage GaN power semiconductors (EPC2034);
converter output power is rated at 1000W;
volume and weight estimation without housing.
Fig. 3 presents power losses, volume and weight comparison for Si-based and GaN-based solutions operating at high-switching frequency (300 kHz). GaN devices present lower power losses than Si counterparts. Losses of Si semiconductors begin to be relevant at high-switching frequency and hard-switching operation mode. This power losses increment has a direct impact on the volume and weight of the power converter. This is mainly due to cooling system volume impact, as it is shown in Fig. 3. The increase of switching frequency results on a reduction of the whole GaN-based power converter volume and weight. This is mainly due to low power losses of GaN devices, resulting on a negligible requirement of cooling system (see Fig. 3).
Theoretical power converter performance analysis, comparing GaN-based and Si-based solutions in terms of power losses, volume and weight (without housing).
GaN-based power converter achieves a clear improvement in terms of power losses, volume and weight in comparison to conventional Si-based solution, with a reduction of 45% and 51% in volume and weight, respectively.
Regarding the performance of GaN semiconductors, Fig. 4 shows the analytical power losses distribution of every switch of the power converter. For this analysis, similar duty cycles for both branches were assumed (
Analytical power losses distribution of one power cell based on GaN devices, differing between primary and secondary branches, and considering a power delivery from Li-S to Li-ion.
B. Experimental Evaluation
Considering the theoretical performance results, a GaN-based experimental prototype was built. Figure 5(a) shows the DC-DC converter that connects the two battery technologies. This converter is comprised by two interleaved power modules, resulting on four half-bridge cells and two inductors, within an aluminum housing, which is used to cool-down the system. Regarding the half-bridge power cell, it is composed of a power board and a conditioning board, as it is depicted in Fig. 5(b). The power converter is based on low-voltage GaN semiconductors, resulting on a design with a reduced size and weight (see Fig. 5).
Multi-cell bidirectional buck-boost converter for HESS. (a) DC-DC prototype composed by four power cells and two inductors. (b) GaN-based half-bridge power cell conformed by power and conditioning boards.
Additionally, Fig. 6 shows the performance of the power converter in terms of efficiency for bidirectional operation, which was measured with Yokogawa WT500. With this analysis low power losses of the GaN-based power converter is demonstrated, achieving high efficiency even at 300 kHz switching frequency (see Fig. 6). A peak efficiency of 96.5% for the nominal operation is measured, i.e., Li-S 0.2C current rate. Considering
Experimental evaluation of the efficiency of the DC-DC power converter working at nominal operating conditions, i.e., Li-S 0.2 C current rate.
Control Strategy
The control strategy depicted in Fig. 7 was implemented in the HESS prototype, with a bidirectional power delivery control. This control strategy was composed of two main stages: the energy management strategy and the power converter control loop.
HESS control strategy. (a) Flowchart representing the rule-based energy management strategy, being
On the one hand, for the energy management strategy, a rule-based method was proposed. The current required by the Li-S battery module was defined according to each battery State-of-Charge (SOC), the required output current and current thresholds, resulting from laboratory results in order to optimize the efficiency and lifetime of batteries. On the other hand, for the power converter control loop, a buck-boost control strategy was chosen with an interleaving Pulse Width Modulation (PWM).
A. Energy Management Strategy
The main goal of the energy management strategy was to decide about how to dispatch the current demand toward the Li-S and Li-ion battery modules, maximizing performances of the HESS and ensuring at the same time a safe and long-lasting operation of both modules [see Fig. 7(a)]. For that purpose, the load current (
The current demand was always primarily oriented to the Li-S module, and the maximum current provided by the Li-S system (
Additionally, simple module balancing measures were included in the energy management strategy, stating the following: i) if the whole current demand was lower than the maximum current limit set to the Li-S, and if the SOC of the Li-ion module was lower than the SOC of the Li-S module, the Li-S system was pushed to provide the maximum allowed current (0.2 C) to charge the Li-ion system; ii) during the regenerative braking phases, the collected energy was oriented to the system presenting the lowest SOC.
B. Converter Control Loop
Regarding the power converter control, the implemented control strategy was comprised as follows: i) Li-S battery current control obtaining inductor current reference (\begin{align*} \delta _{\mathrm {1,i}}=&\frac {V_{\mathrm {Li-S}} + v_{\mathrm {L,i}}^{\ast}}{V_{\mathrm {Li-ion}}}, \tag{1}\\ \delta _{\mathrm {2,i}}=&\frac {V_{\mathrm {Li-ion}} - v_{\mathrm {L,i}}^{\ast}}{V_{\mathrm {Li-S}}}. \tag{2}\end{align*}
Besides, if the converter is operated in buck regime, meaning that
Control loop for a bidirectional operating condition. Generated duty cycles (
Hess Experimental Evaluation
This section describes and discusses the experimental results achieved with the developed HESS prototype submitted to realistic EV driving profiles. Figure 9 shows the general schematic of the test bench. For the driving cycle, a battery tester has been used, which can be used as a load or a charger. All the tests have been done at a controlled temperature (20°C), in a climatic chamber.
Experimental set-up. (a) DBS prototype with control and auxiliary circuits location. (b) General schematic of the laboratory test bench.
The main goals of the experimental study were: i) to analyze the behavior of the HESS prototype under several EV driving scenarios; ii) to verify the increase of energy density and specific energy achieved by combining Li-S and Li-ion technologies under realistic profiles; iii) to identify system limitations and assess how far such HESS prototype would be from a potential use in a commercial EV system.
For this purpose, several experiments were conducted in laboratory. First, the prototype was submitted to a Worldwide Harmonized Test Cycle in Charge Depletion (WLTC-CD) driving profile, which is widely accepted to be representative of an EV motorway driving scenario [30]. Such profile was scaled-down to the developed HESS prototype size, and repeatedly applied until the system was fully discharged. In Section V-A, the profile was first applied to the single Li-ion module, for further comparison purposes. Then, in Section V-B the same profile was applied to the full HESS prototype to analyze the benefits of hybridizing Li-S and Li-ion technologies for EV application. Nevertheless, real-world driving scenarios does not typically imply consecutive sequences of driving cycles. For this reason, in Section V-C, the motorway profile was slightly modified including idle periods between each sequence of WLTC-CD driving cycle. Finally, in Section V-D, the WLTC-CD profile was modified to simulate lower speed and energy-oriented automotive profiles, more representative of urban driving scenarios. All the tests were carried out until the minimum voltage of a battery cell was achieved.
A. Single Li-Ion System: High-Speed Driving Scenario
For comparison purposes, the scaled-down WLTC-CD profile was first applied to the single Li-ion battery module used in the HESS prototype. The cells of the Li-ion module were initialized at 95% SOC.
Figure 10 depicts the results of the experiment, indicating the scaled-down WLTC-CD power profile which was repeatedly applied to the Li-ion module (upper graph), and the subsequent evolution of the SOC of the module (bottom graph).
Scaled-down WLTC-CD power profile, repeatedly applied to the single Li-ion module (upper graph). Negative values of the current indicate that the battery is discharging, and positive values indicates the battery is charging through regenerative braking. Evolution of the SOC of the Li-ion module is shown in bottom graph.
The Li-ion module was able to supply 112.94 Wh.L−1 and 72.67 Wh.kg−1. The WLTC-CD profile was supplied 3.32 times, calculated as the ratio between the total energy delivered during the test, with respect to the energy corresponding to a single WLTC-CD profile completion. The test was stopped when the module voltage reached the limit discharge voltage value.
B. Hess Prototype: High-Speed Driving Scenario
The high-speed WLTC-CD driving profile was then applied to the whole HESS prototype in the laboratory environment. The cells of both Li-ion and Li-S modules were initialized at 95% SOC.
Under the high-speed driving scenario, the HESS prototype was able to supply 1248.9 Wh. The WLTC-CD profile was supplied 3.67 times, as it is depicted in Fig. 11. In terms of energy density and specific energy, 82.49 Wh.L−1 and 64.5 Wh.kg−1 were respectively extracted from the HESS, which corresponds to a reduction of −26.96% and −11.25% compared to the results obtained in Section V-A. Such a performance reduction is due to the additional volume and weight of the Li-S module, while the corresponding energy was not fully extractable in this scenario. In this case, the test was stopped because one cell reached the discharge voltage limitation. Notice that, because of cells imbalance, energy remained in both Li-ion and Li-S modules. At the end of the test, 68.57% SOC and 25.81% SOC was remaining respectively in the Li-S and the Li-ion module.
Behavior of the HESS under high-speed driving profile. Upper graph: current profile provided by the Li-ion module (black curve) and the Li-S module (blue curve). Bottom graph: SOC evolution of the Li-ion and Li-S cells (respectively in black and blue curves).
C. Hess Prototype: High-Speed Driving Scenario With Idle Periods
Although the scenario implemented in Section V-B is useful to unveil the limitations of the developed HESS prototype, it is noteworthy that in a typical real-world automotive application, driving cycles are not successively applied. In fact, a more realistic operation would involve idle period or pauses between each driving cycle, corresponding to the periods in which the EV would be parked.
For this reason, an additional experiment was conducted intercalating such pauses between each WLTC-CD driving profile. In total, the duration of the idle periods in which the HESS system was not electrically excited was approximately one hour. The results of the applied driving cycle are depicted in Fig. 12.
Behavior of the HESS under high-speed driving profile with idle periods. Upper graph: current profile provided by the Li-ion module (black curve) and the Li-S module (blue curve). Bottom graph: SOC evolution of the Li-ion and Li-S cells (respectively in black and blue curves).
The WLTC-CD profile was supplied 4.71 times with a total energy of 1603.4 Wh. In terms of energy density and specific energy, 105.9 Wh.L−1 and 82.79 Wh.kg−1 were respectively extracted from the HESS, which corresponds to a reduction of −6.23% and an increase of +13.91% compared to the results obtained in Section V-A, for the single Li-ion system. The test was stopped when the Li-ion module was completely discharged. At the end of the test, 37.11% SOC remained in the Li-S module.
D. Hess Prototype: Urban Driving Scenario
The WLTC-CD automotive profile considered in the previous experiments was originally designed to represent high-speed driving (~120 km.h−1) typically related to motorway-type roads. In this section, such power-profile was modified to represent a less power-demanding and smoother conduction, more representative of urban driving scenarios. For this purpose, the automotive power profile used in Section V-B was reduced by a factor of two. In order to maintain an equivalent energy demand, the duration of the automotive profile was also multiplied by a factor of two. Five-minutes short pauses were intercalated between each driving cycle, for cell relaxation purposes.
The HESS prototype was able to supply 1730.2 Wh. The WLTC-CD profile was supplied 5.09 times. The extracted specific energy and energy density were 89.36 Wh.kg−1 and 114.28 Wh. L−1, respectively, which represents a +22.96% and +1.19% increase compared to the specific energy and energy densities extracted from the Li-ion system in Section V-A. In this experiment, the test was stopped when one Li-S cell of the module reached the discharging voltage limitation (see Fig. 13 and Fig. 14). At the end of the test, the SOC of the Li-ion and Li-S modules were respectively 5.65% and 24.23% SOC.
Behavior of the HESS under urban driving profile. Upper graph: current profile provided by the Li-ion module (black curve) and the Li-S module (blue curve). Bottom graph: SOC evolution of the Li-ion and Li-S cells (respectively in black and blue curves).
Voltage curves obtained during the experiment for the 24 Li-S cells implemented in the module. Each curve corresponds to a specific Li-S cell (
Discussion Limitations and Further Works
The main goal of the study was to develop a HESS combining the high-power capability of LTO Li-ion batteries and the high energy density and specific energy of Li-S technology. Table 3 depicts the available energy, energy density and specific energy for the single Li-ion battery module and for the developed HESS prototype. The nominal values of the systems are depicted, as well as the energy extracted for the different driving scenarios.
Regarding the energy density and specific energy, two values are differentiated in the Table 3. The first value was calculated considering the weight/volume of the prototype modules and DC-DC converter. Nevertheless, as the developed HESS was a laboratory prototype, the packaging of the Li-ion and Li-S modules was not optimized and could easily be reduced, although this was out of the scope of the study. Therefore, the second value indicated in Table 3 corresponds to the energy density and specific energy calculated considering only the weight/volume of the cells themselves and the DC-DC converter.
As expected, the hybridization of both technologies provided respectively an increase of energy density and specific energy of +5.56% and +28.21%, compared to the single Li-ion system. Besides, discarding the influence of non-optimal packaging of the prototype +8.50% and +32.93% improvement is expected.
Interestingly, the achieved energy density and specific energy values could be relatively reduced compared to energy-oriented Li-ion chemistries, e.g. NMC622 cathodes [31]. Nevertheless, this could potentially be compensated by increasing the gravimetric and volumetric Cell-to-Pack ratios, as demonstrated in [31], which is conceivable for the developed HESS due to GaN-based power converter. Therefore, it is reasonable to state that, optimizing the housing to industrial production quality, the developed HESS concept could be a suitable solution to achieve at the same time high energy density and specific energy at pack level, and an increased power capability provided by LTO anode technology.
Moreover, it is noteworthy that not all the nominal energy is necessarily extractable, and the range of an EV integrating such a HESS would strongly depend on the driving profile. Indeed, applying the WLTC-CD high-speed driving profile (see Section V-B), the experiment was interrupted when the Li-ion module reached the lower voltage limitation, and 68.57% SOC remained in the Li-S module after the end of the experiment. This could be explained by the power restriction applied to the Li-S in the energy management strategy (see Fig. 7), which did not allow the Li-S cell to provide more than a C-rate value of 0.2 C in normal operation. The reason to impose such a restrictive limitation was that the Li-S technology used in the HESS prototype suffers drastic capacity reduction when cycled at higher C-rates. Consequently, due to the power-demanding character of the profile, the Li-ion module was forced to cover most of the demand of the WLTC-CD profile and was discharging faster than the Li-S module (see Fig. 13).
The results from this experiment suggest that the limited power capability of the Li-S technology could potentially be an obstacle for EV application, particularly for a usage involving high-speed and extended range. For such EV usages, it would be necessary to improve the power capability of the Li-S technology. Alternatively, other hybridization approaches could also be considered, involving energy-oriented Li-ion cells and for example future generations of supercapacitors [5].
In a more frequent real-world automotive scenario, when idle periods were intercalated between each driving cycle (see Section V-C), more energy was extracted from the HESS system compared to the results achieved in Section V-B (see Table 3). The reason for that resides in the energy management strategy depicted in Fig. 7 and implemented in the DC-DC of the HESS prototype: during the pauses incorporated in between each automotive driving cycles, the Li-S module was used to charge the Li-ion module, as long as the Li-S module showed a higher SOC state than the Li-ion module. In such a way, the HESS was able to supply respectively an increase of energy density and specific energy of −6.23% and +13.91%, compared to the single Li-ion system (−3.57% and +18.14% discarding the influence of non-optimal packaging of the prototype). It is noteworthy that the energy deliverable by the HESS is strongly related to the duration of the idle period. Although this experiment considered idle periods of approximately 15 minutes, each driving session could be separated by several hours in real-world scenario (e.g., 8 hours during the night and working times). Therefore, the improvement of the delivered energy density and specific energy could be easily increased considering larger pauses, surrounded this way the +5.56% and +28.21% increase obtained with nominal energy values of the modules. This experiment highlights the importance of the energy management strategies regarding the HESS performance. Further works include research on improved energy management strategies based on optimization methods.
Moreover, during the experiment corresponding to the urban driving scenario (Section V-D), the HESS was able to supply additional energy, providing 1730.2 Wh. The experiment was interrupted because one Li-S cell of the module reached the discharging voltage limitation, while 5.65% and 24.23% SOC was remaining in the Li-ion and Li-S modules, respectively. Fig. 14 depicts the voltage curves obtained during the experiment for each individual Li-S cell.
The Li-S cell which reached the voltage limitation is clearly recognizable in Fig. 14, and the corresponding voltage curve suggests that the SOC of such cell was lower than the remaining cells in the module. This could be caused by i) a progressive unbalancing trend between the Li-S cells during the experiment, possibly due to intrinsic cell-to-cell discrepancies regarding coulombic efficiency and/or self-discharge rates [32], or ii) a premature aging of such cell compared to the others, leading to a relatively reduced capacity or internal resistance increase [27]. As Li-S cell chemistry is a recent technology which is still under continuous development, there is a clear lack of background on Li-S state estimation methods and analysis tools. This hinders, for now, to clarify the precise reasons behind the accelerated voltage drop of this cell. This should be analyzed in further works.
The performance of the battery at module-level is always limited by the weakest cell, and this is clearly corroborated in this experiment. This highlights the paramount importance of fostering further research on the development of reliable and accurate SOC and State-of-Health (SOH) algorithms for Li-S technology, which in turns could allow making progress on effective balancing strategies tailored to Li-S battery modules.
As energy was remaining in the Li-ion module after the experiment was stopped by Li-S voltage limitation, the profile was applied again to the single Li-ion module, in order to push the HESS prototype until its broadest energy performance limitation. This way, the HESS prototype was able to supply 1781.7Wh in total. The extracted energy density and specific energy were 117.68 Wh.L−1 and 92.02 Wh.kg−1 respectively, which represents a +4.21% and +26.62% increase with respect to the energy density and specific energy extracted from the single Li-ion system experiment.
Finally, the low cycle life of Li-S technology reported in literature [2], would also have a significant impact on system behavior. Nevertheless, aging studies remained out of the scope this research and should be investigated in further works.
Conclusion
In this study, the development of a hybrid energy storage system was investigated for EV application, with the aim of achieving at the same time an increased energy density, specific energy, and power capability. The system was composed of three key components, namely i) a Li-ion battery module to fulfill high-power requirements; ii) a Li-S battery module to increase the energy density and specific energy of the system, and iii) a GaN-based DC-DC power converter to control and protect the sensitive Li-S system, maintaining at the same time a reduced size and high-efficiency at system level. The GaN-based power converter achieved a clear improvement in terms of power losses, volume and weight in comparison to conventional Si-based solutions, with a reduction of 45% and 51% in volume and weight, respectively.
The developed HESS prototype achieved respectively an increase of energy density and specific energy of +5.56% and +28.21%, compared to a single Li-ion system. Moreover, in order to assess how far such HESS technology would be from potential use in an EV application, the prototype was submitted to several power profiles representing different real-world driving scenarios. The obtained results allowed to identify the main paths for improvement and suggests focusing research efforts on i) working on next-generation Li-S cell technology, especially improving the power capability, ii) further investigations on developing battery models, state algorithms and analysis tools tailored to Li-S chemistry, and iii) continue developing application-oriented energy management strategies to maximize the performances of current Li-S technology.