*3.3. Advanced Battery Management Systems*

matches the parallel/series structure of the battery cells [49,65].

technology advantages [60].

*3.3. Advanced Battery Management Systems* As pointed out in Section 3.1, the BMS plays a fundamental role in every energy storage system [61], particularly those based on Li-ion technology. Besides the basic monitoring and management functions mentioned above, there are advanced functions that updated BMS is required to perform. First of all, the BMS architecture must be designed to fit to different configurations of the battery, that usually consists of cells, modules (usually series connected), and strings (usually parallel connected). The BMS functions may be spread among the various layers of the BMS architecture [62]. Typical architectures span from a single master BMS that controls the entire battery, to hierarchical distributed ones, by which the BMS hardware is distributed over the battery modules and even down the individual battery cells [63,64]. Figure 5 shows a typical BMS hierarchical architecture that As pointed out in Section 3.1, the BMS plays a fundamental role in every energy storage system [61], particularly those based on Li-ion technology. Besides the basic monitoring and management functions mentioned above, there are advanced functions that updated BMS is required to perform. First of all, the BMS architecture must be designed to fit to different configurations of the battery, that usually consists of cells, modules (usually series connected), and strings (usually parallel connected). The BMS functions may be spread among the various layers of the BMS architecture [62]. Typical architectures span from a single master BMS that controls the entire battery, to hierarchical distributed ones, by which the BMS hardware is distributed over the battery modules and even down the individual battery cells [63,64]. Figure 5 shows a typical BMS hierarchical architecture that matches the parallel/series structure of the battery cells [49,65].

*Appl. Sci.* **2019**, *11*, x FOR PEER REVIEW 9 of 18

**Figure 5.** Hierarchical architecture of a BMS matching the battery structure consisting of four paralleled strings of 6 series connected modules each. The Pack Management Unit (PMU) is the master BMS, connected to the lower layer slave BMSs (MMUs). [49]. **Figure 5.** Hierarchical architecture of a BMS matching the battery structure consisting of four paralleled strings of 6 series connected modules each. The Pack Management Unit (PMU) is the master BMS, connected to the lower layer slave BMSs (MMUs) [49].

The State of Charge (SoC) of a battery is a fundamental quantity as it indicates the residual charge remaining in the battery, and thus, gives indications on the residual support provided by the ESS to the application. Moreover, the SoC estimate must be carried out on every cell of the battery, as the battery cells may be mismatched, and that mismatch may increase with ageing. Unfortunately, SoC is a quantity not directly measurable, but it must be calculated from other quantities, e.g., by integrating, in time, the current flowing in the battery. As in every non-ideal integration process, offsets or inaccuracies in the current sensor readings rapidly lead to a calculation of the integral that is affected by increasing and unacceptable errors. Many alternative approaches to the so-called Coulomb counting technique have been proposed [66,67]. Most of them are based on a model of the battery that is subjected to the same load and operating conditions of the real battery. The status of the battery, estimated from the model calculation, resembles the status of the real battery. Therefore, the SoC and other battery parameters can be estimated by the model. The most popular models are electrical models. They provide acceptable accuracy, together with affordable computation complexity. The battery is represented by an electromotive force (the so-called Open Circuit Voltage) that varies as a function of SoC, a series resistance, that models the internal losses, as well as a series of some resistor-capacitor parallel groups, which model the relaxation phenomena in the battery. The BMS is, thus, required to solve the model-based circuit to obtain an accurate estimate of the SoC. A further effect, that makes the SoC estimate tougher, is the variation of the model parameters. In fact, the values of the electrical components, which make up the battery model, vary in time, according to the state of charge, temperature, the environmental conditions, and the health status of the battery. A further challenge for an advanced BMS is, thus, to identify and track the parameter changes, only by observing the current, voltages, and temperatures of the battery cells. The State of Charge (SoC) of a battery is a fundamental quantity as it indicates the residual charge remaining in the battery, and thus, gives indications on the residual support provided by the ESS to the application. Moreover, the SoC estimate must be carried out on every cell of the battery, as the battery cells may be mismatched, and that mismatch may increase with ageing. Unfortunately, SoC is a quantity not directly measurable, but it must be calculated from other quantities, e.g., by integrating, in time, the current flowing in the battery. As in every non-ideal integration process, offsets or inaccuracies in the current sensor readings rapidly lead to a calculation of the integral that is affected by increasing and unacceptable errors. Many alternative approaches to the so-called Coulomb counting technique have been proposed [66,67]. Most of them are based on a model of the battery that is subjected to the same load and operating conditions of the real battery. The status of the battery, estimated from the model calculation, resembles the status of the real battery. Therefore, the SoC and other battery parameters can be estimated by the model. The most popular models are electrical models. They provide acceptable accuracy, together with affordable computation complexity. The battery is represented by an electromotive force (the so-called Open Circuit Voltage) that varies as a function of SoC, a series resistance, that models the internal losses, as well as a series of some resistor-capacitor parallel groups, which model the relaxation phenomena in the battery. The BMS is, thus, required to solve the model-based circuit to obtain an accurate estimate of the SoC. A further effect, that makes the SoC estimate tougher, is the variation of the model parameters. In fact, the values of the electrical components, which make up the battery model, vary in time, according to the state of charge, temperature, the environmental conditions, and the health status of the battery. A further challenge for an advanced BMS is, thus, to identify and track the parameter changes, only by observing the current, voltages, and temperatures of the battery cells.

Several state estimation and parameter identification techniques have been applied in the literature, starting from the moving window least square methods, to filtering, such as Extended Kalman Filters and Particle Filters. In any case the choice of the estimation technique must be traded off between the computation complexity required and the accuracy needed. Several state estimation and parameter identification techniques have been applied in the literature, starting from the moving window least square methods, to filtering, such as Extended Kalman Filters and Particle Filters. In any case the choice of the estimation technique must be traded off between the computation complexity required and the accuracy needed.

An even tougher task for a BMS is the evaluation of the health status of a battery [68,69]. The

An even tougher task for a BMS is the evaluation of the health status of a battery [68,69]. The battery ageing is a complex process that depends on the utilization of the battery in term of operating temperatures, load profiles, and the number of charge/discharge cycles. It typically results in the progressive fading of the maximum capacity of the battery and the increase of the series resistance, which determine the increased internal losses and reduced power capability. The knowledge of the battery health status, together with the expected operational profile, may lead the BMS to estimate the Residual Useful Life of the battery. progressive fading of the maximum capacity of the battery and the increase of the series resistance, which determine the increased internal losses and reduced power capability. The knowledge of the battery health status, together with the expected operational profile, may lead the BMS to estimate the Residual Useful Life of the battery. *3.4. Battery Charging Process* Battery cell manufacturers usually provide the user with constraints about the appropriate

*Appl. Sci.* **2019**, *11*, x FOR PEER REVIEW 10 of 18

temperatures, load profiles, and the number of charge/discharge cycles. It typically results in the

### *3.4. Battery Charging Process* recharging process of the battery. The suggested charging profile is the Constant-Current Constant-

Battery cell manufacturers usually provide the user with constraints about the appropriate recharging process of the battery. The suggested charging profile is the Constant-Current Constant-Voltage (CC-CV) one [70]. It consists of a first phase in which the battery is charged with a constant current. The current value depends on the cell technology, and it is in the order of the battery capacity *C* expressed in ampere. Fast charging usually exceeds 1 C, whereas slow charging spans from 0.1 C to 0.5 C. The faster the charging, the higher the internal cell temperature and thus the possible stress that may determine ageing. The second phase starts when the battery voltage reaches the float value. The charger keeps the voltage constant since then, to gradually fill the battery up the maximum charge level. The process ends when the decreasing current reaches a threshold, usually a rather low fraction of the battery capacity expressed in ampere. The battery inserted in the multi-source/sink micro-grid scenario, depicted in Figure 6, cannot fully adhere to the above described charging process. Instead, as the charging and discharging phases are defined according to the overall management of the energy fluxes, the BMS should verify the manufacturer constraints in terms of current, temperature, and individual cell voltages. Should it happen, the bi-directional DC/DC converter, that connects the battery to the power bus (see detailed circuit schematic in Figure 7), must be driven in such a way as to reduce the charging current and to withstand the manufacturer's limits. As the battery is not subjected to full discharge/charge cycles, the accurate knowledge of the battery SoC is mandatory to fully exploit the established energy flux policy mentioned in Section 2.3. Voltage (CC-CV) one [70]. It consists of a first phase in which the battery is charged with a constant current. The current value depends on the cell technology, and it is in the order of the battery capacity *C* expressed in ampere. Fast charging usually exceeds 1 *C*, whereas slow charging spans from 0.1 *C* to 0.5 *C*. The faster the charging, the higher the internal cell temperature and thus the possible stress that may determine ageing. The second phase starts when the battery voltage reaches the float value. The charger keeps the voltage constant since then, to gradually fill the battery up the maximum charge level. The process ends when the decreasing current reaches a threshold, usually a rather low fraction of the battery capacity expressed in ampere. The battery inserted in the multi-source/sink micro-grid scenario, depicted in Figure 6, cannot fully adhere to the above described charging process. Instead, as the charging and discharging phases are defined according to the overall management of the energy fluxes, the BMS should verify the manufacturer constraints in terms of current, temperature, and individual cell voltages. Should it happen, the bi-directional DC/DC converter, that connects the battery to the power bus (see detailed circuit schematic in Figure 7), must be driven in such a way as to reduce the charging current and to withstand the manufacturer's limits. As the battery is not subjected to full discharge/charge cycles, the accurate knowledge of the battery SoC is mandatory to fully exploit the established energy flux policy mentioned in Section 2.3.

**Figure 6.** New hybrid micro-grid with full connectivity among RES node (DC), EV node (DC) and **Figure 6.** New hybrid micro-grid with full connectivity among RES node (DC), EV node (DC) and main grid node (AC) and relevant bi-directional flows. EV: Electric Vehicle.

As anticipated above, a new micro-grid architecture is described in this Section (see Figure 6), which exploits bi-directional EV charger plus RES to overcome the flexibility limit of the architecture, as shown in Figure 1. Section 3 has shown why Li-ion batteries are being widely used for the high power and energy densities. Thus, they serve as energy storage unit connecting to the smart grid and to a RES such as PV panels in our focused application. This leads to a hub connection discussion: The current industry devices lack an efficient integrated connection system, since PV panels are not directly connected to the EV charger. Typically, they are first connected to the grid via a

main grid node (AC) and relevant bi-directional flows. EV: Electric Vehicle

**4. Innovative Micro-Grid with Bi-Directional Flows for RES and EV Charging**

higher system cost. Instead, a novel bidirectional EV charger system is proposed in Fig. 6 to build a direct connection between PV panels and the EV, and to create a V2G path. Therefore, reduced power Mosfets.

phase 380Vac one.

with high efficiency and high-power density.

converter to adapt the output solar PV level to the DC power bus level.

loss and lower system cost features are achieved from a highly integrated power electronics system

The idea of the system scheme in Fig. 6, as further detailed in the circuit scheme of Figure 7, is having a central DC power bus, plus a bidirectional flow between the AC grid and the power DC bus thanks to a 3-phase bidirectional converter, and a bidirectional flow between the DC battery on-board the EV and the DC power bus thanks to a bidirectional DC/DC converter. A unidirectional flow is still foreseen from the PV solar panel sub-system towards the DC power bus, using a boost DC/DC

The DC power bus voltage is typically in the range from 250 V to 600 V, e.g. it has been sized at 400 V in [3] in case of a 10 kW bidirectional EV charger and at 450 V in case of a 1.65 kW bidirectional EV charger in [71]. In [72–74], as special optimization case, a bidirectional EV charger is proposed where the value of the DC power bus can be sized from 500V to 840V. The AC grid is typically a 3-

In the detailed circuit schematic of the EV bidirectional charger in Figure. 7 the bidirectional 3 phase AC/DC converter and the boost DC/DC converter follow classic circuit solutions. For the isolated bi-directional DC/DC converter in Fig. 7, instead of using a dual-active bridge topology as in [71–74] (see Figure 8B), a half-bridge series resonant LLC topology is proposed (that of Figure 8A). This approach allows reducing the number of active switches to be used and hence it makes more convenient the adoption of SiC power Mosfets (e.g. Cree C2M0040120D adopted in [3]) instead of classic Si power Mosfets (e.g. Infineon IPW60R045CP adopted for the primary stage in [71]). Indeed, SiC power Mosfets are more expensive than Si power Mosfets: a market analysis on stocks from 100 to 1000 devices for the power mosfets in Table 1 has shown that the selling price for each SiC power device is 1.7 to 3.5 times higher than that of Si power device. Hence, a circuit solution like that in

**Figure 7.** Circuit schematic of the bi-directional converter (1 AC grid port and 1 DC EV battery bidirectional ports; 1 DC unidirectional solar power port). **Figure 7.** Circuit schematic of the bi-directional converter (1 AC grid port and 1 DC EV battery bidirectional ports; 1 DC unidirectional solar power port).
