**1. Introduction**

Recent developments in energy storage systems (ESS) and fast charging technologies extend the range of electric vehicles and their increasing market share are reducing prices [1–3].

The European Union set the target of 40% reduction of greenhouse gas (GHG) emissions and of 27% share of renewable energy by 2030 [4], with a potential reduction of 80%–95% of GHG and 55%–75% of gross final energy consumption from renewable sources [5] by 2050.

The transport sector contributes almost a quarter of Europe's GHG emissions and buses are responsible for 8% of transport emissions. In 2019, electric buses all over Europe count 2200 units [6], less than 1% of European bus fleet (about 770,000 units [7]). A study forecasts that electric buses will reach more than 23,000 units in 2025 [8].

An opportunity to shift towards electric transportation is the retrofit [9]. This was fostered in Italy by a recent national policy initiative. The Italian Ministry Economic Development (MISE) issued the order no. 219 of 1 December 2015 [10] to allow this procedure.

A retrofit replaces an Internal Combustion Engine (ICE) with an electric kit (composed by a motor, a battery and some electronics). Today, it is applicable only to M1 and N1 vehicle categories (cars weighing up to 3.5 tons), however there are future possibilities to extend this opportunity to larger vehicles.

This study focuses on small public transport vehicles; those minibuses are maximum 6 m long with a passenger capacity of 30 people. It shows a comparison of the data gathered by three consecutive projects all founded by MISE in the last four years. Project partners were ENEA and four Italian Universities (University of Firenze, Sapienza of Rome, Roma Tre and Pisa).

Each project used the same bus model, a Tecnobus Gulliver ESP500, equipped with different prototypes of energy storage systems (ESS). Figure 1 shows the buses of the three projects:


**Figure 1.** Project prototypes: P1 (**a**), P2 (**b**) and P3 (**c**).

Each project had its own experimental testing campaign, each taking place into the ENEA campus. The campus has about ten kilometers of internal roads, so there are many workers moving during the day. This allowed us simulating a common transport service.

Preliminary results have shown that each bus has same performance and the energy consumption of the vehicle is not influenced by the energy storage. Moreover, P3 achieved better economic results with lower costs. The authors argued the economic benefit is due to less frequent replacements of the battery in a hybrid ESS. Hence, it was evaluated whether it is convenient to replace battery with a more expensive lithium battery. A new project starting from P3 and replacing the AGM battery with a lithium-ion one was simulated and indicated as P4 in the study.

An economical comparison was carried out among these four alternatives: the three experimental projects with real data plus the simulated scenario. It was hypothesized that all prototypes must supply the same transport service with a daily range of 100 km. The comparison of the projects is based on life cycle cost with the net present value (NPV) indicator [12]. The incomes are the same for all of them and costs change from one to another. The project with lower costs has a better NPV.

The present study demonstrates that the best results are achieved in P4, which is characterized by hybrid ESS (as the one of P3) combined with lithium-ion batteries (as the one of P2). P4 combines two technologies with a more efficient usage that give longer life expectation to the electric and storage components.

This study is organized in four sections: the current introduction; Section 2, which presents the details of compared projects; Section 3, which presents the results; and Section 4, the conclusions.

### **2. Details of the Compared Projects**

This section details nominal specifications and analysis of real data for the three projects (from 1 to 3). P4 combines specifications of P2 and P3 and its input data are explained in Section 3.

Project 1 had the lead–acid batteries provided by the OEM. The batteries have no sensors, but the chopper (DC regulator) of the motor provides upon request voltage and current to and from the batteries.

Energy storage system prototypes were manufactured, specifically in P2 and P3. In P2, a battery managemen<sup>t</sup> system (BMS) measures voltage, current and temperature of each battery-cell individually for safety and advanced management. In P3, a buck-boost DC-DC converter was inserted between the ultracapacitors and the batteries. Therefore, P2 and P3 had very accurate measurement systems installed directly on the energy storage.

These three prototypes have different measurement systems. The first had a 1 Hz sampling rate; the traveled distance was about 100 km. P2 and P3 had an acquisition rate, respectively of 2 Hz and 10 Hz. These two projects had less data and the traveled distance was about 20 km each.

Energy consumption was compared by observing the average consumption per kilometer of many trips, each with different driving cycles, terrain orography, payload and driving styles.

Range and charging times are also different from among projects. In order to be compared and fulfil the same transport service, they required some adaptation:


The choice of LiFePO4 is due to the availability of experimental data [14], where it was estimated the maximum life cycle of a battery with conditions comparable to current bus usage.

A performance and economic evaluation were done. The first one is based on maximum speed, maximum acceleration and time to reach maximum speed with a standing start.

Indeed, the economic evaluation is a cost benefit analysis using the net present value as main indicator of economic value, was performed over a twenty-four-year time frame, to consider a least common multiple of the lifetime expectations of the different technologies (called also cycle life, CL) [15,16].

The periodic replacement of exhausted batteries during the lifetime of the bus has also been considered. An ESS lasts up to a few years depending on the usage. Bus lifetime ranges from 10 to 15 years depending on its size, for example a 12-m long bus has 15 years of depreciation in Italy. The bus used in this analysis is 6-meter long and a lifespan of 12 years was assumed. SC lifetime is longer than a million cycles (according to manufacturer specifications) [17,18].

The bus lifetime was assumed 12 years as reference value, so, the economic evaluation expects at least 24 years, considering at least a replacement for the bus and all ESS components. The SC lifetime is more than twenty years, considering the expected life cycle and their usage in a bus, while, battery life depends on several factors [19]. The main factors are: depth of discharge (DoD), discharge rate (measured in multiples of the nominal capacity C), charge rate, aging, working and environmental temperatures.

For the tests of the all projects, the working temperature of the battery was maintained within the limits prescribed by manufacturer. While the environmental temperatures of these projects were the same, all tests made in the ENEA campus were conducted with mild weather.

It can be assumed that for a bus application, number of cycles life of ESS components is lower than the calendar life (it is the elapsed time before a battery becomes unusable, whether it is in active or inactive use). The manufacturer of the AGM battery, used in P1 and P3, declares 20 months of life, while lithium lasts up to five years [20,21].

Meanwhile, DoD has the highest impact to a battery; for the P1 equipped with a lead–acid battery, the DoD was about 80% and it had 500 cycles to failure [22], as shown in Figure 2. Hence, if the transport service application requires a full charge every day, the battery must be replaced every 500 days.

**Figure 2.** Lead–acid battery behavior [22].

Figure 3 shows DoD effects applied to three lithium batteries with different chemistry [23]. A LiFePO4 battery (as those of P2) has about a thousand CL if used up to 80% of DoD or, if used only up to 40%, it will last three times longer.

**Figure 3.** Effects of DoD to a few types of lithium battery [23].

Such effects become very important from an economic point of view especially in the case of higher costs of lithium compared with lead–acid.

C rate, during charging and discharging, reduces the battery lifetime even more. As described in Figure 4, the capacity of a lead–acid battery drops when discharge rate raises from 0.5C to 10C, then battery capacity decreases from 100% of initial value (battery fully charged) to 70% [24]. Lithium-ion batteries suffer from the same issues but have different effects [14,25], accordingly to the discharge rate capability and battery life cycle given by battery manufacturers.

Experimental results of three cited project are described one by one as follows.

**Figure 4.** Rate capacity effect [24] with a lead–acid battery.

### *2.1. Project 1 (P1): Lead–acid Batteries*

The goal of the project was to develop an on-demand transport system between ENEA facilities. The bus was equipped with a lead–acid battery of 43 kWh at 72 V, with a capacity of 600 Ah. It was composed by 36 batteries of 100 Ah–12 V each. The configuration consisted of 2 strings in series, each one composed of eighteen batteries, meaning six groups in parallel of three batteries in series.

It had long running acquisition including different missions. Each one of them included a running distance of at least 500 m. It started and stopped at zero speed (minimum measuring time of 10 s).

This project ran for almost three months and data for about 100 kms were collected. In order to evaluate such amount of data, focusing only on the average consumption and with a wide variety of driving conditions (slopes, payload, etc.), data were divided into more than 110 stretches. Figure 5 shows the histogram of the occurrences for the average consumption.

**Figure 5.** Occurrences of average consumption for different stretches.

This bus consumes from 0.35 to 0.70 kWh/km, with a modal value of 0.45 kWh/km.

Figure 6 shows the current and voltage trend during a stretch. The ESS provided more than 320 A and each battery up to 55 A of maximum current. The chopper of P1 did not show negative values through the interface used, but it also computed energy consumption with both negative and positive currents.

**Figure 6.** Voltage and current trend for P1 during a bus ride.

Figure 6 highlights the voltage of battery drops due to high internal resistances.

The internal resistance further reduces the life expectation of lead–acid batteries. A new battery features this behavior only with low state of charge (SOC), but it gets worse with age and number of cycles.

Table 1 shows the results of a few rides during transport service of P1.


**Table 1.** Road testing results of P1.

### *2.2. Project 2 (P2): Lithium-Ion Batteries*

This project developed a prototype of fast charging battery pack for a small minibus [26,27].

The prototype battery pack was composed of 17 kWh of lithium batteries. It is composed of 96 cells of 3.7 V and 60 Ah each. The configuration was four strings of twenty-four cells in series each and the whole battery reached 76 V–240 Ah. It was capable of 3C charging rate.

The chemical composition was LiFePO4; it could be charged with 1.4 kWh in 110 s as shown in Figure 7, where current and energy during fast charging are plotted. Current values of Figure 7 are negative due to sign convention of the measurement system. Hence, the energy decrease means a charge.

**Figure 7.** Battery energy and current trends during charging.

Figure 8 shows voltage and current trend during a bus ride.

**Figure 8.** Voltage and current trends of P2.

There are two important differences between P1 and P2:


The lowest value of minimum voltage is probably due to high-power-request battery with degraded state of health or even low levels of SOC. These low-voltage situations cause malfunctioning in auxiliary devices (i.e., DC–DC converters, steering pump, brake pump, relays, etc.) and increase currents.

Such situation starts a chain reaction that alters the battery composition. A given power request with lower voltage means higher current (in comparison with another one at higher voltage). The higher current, in turn, means higher losses in heating and further lower voltage (due to rise of internal resistances) and again much higher current.

Table 2 shows the results of a few rides for P2; that needs 422 Wh of energy per kilometer.


**Table 2.** Road testing results of P2.

\*: based on all data measured.

### *2.3. Project 3 (P3): Hybrid Storage SC and AGM Batteries*

Figure 9 shows the prototype while it charged at bus stop. The project design was published [28–30].

**Figure 9.** Charging phase of the prototype of P3 with flash charging technology.

The bus was equipped with a hybrid storage system composed of AGM batteries and supercapacitors. The goal of this project was to develop a flash charging technology for public transport that can charge small quantities of energy very quickly at every stop.

The SC provides through the DC–DC converters some energy directly to the chopper, reducing the energy provided by the battery [31,32].

Figure 10 shows voltage and current trends during charging. This phase lasts 45 s and charges up to 302 Wh, current reaches 350 A. Supercapacitor voltage ranges from 200 V to 375 V. These features can be further improved by optimizing the charging phase; after some tests, an optimistic hypothesis is 20 s (to be validated).

This project requires a charging station at least every 600 m. Larger distances between two charging station can deplete the energy stored on board [33].

**Figure 10.** Supercapacitor voltage and current trends during the ultra-fast charging phase.

Table 3 shows main results of experimental campaign, it also shows more parameters than previous projects due to a larger number of installed sensors (i.e., DC–DC converters and SC).

Every hour the control strategy of DC–DC drains 35 Ah from the battery, so this prototype guarantees an autonomy of 3 h, or about 35 km, if the battery capacity is 120 Ah as in the P3 prototype. After that, it requires a slow full charge of AGM battery.


**Table 3.** Main test results of hybrid storage prototype (P3).

This project tested only a few strategies to manage the hybrid storage; a next step will be to reduce the current drained from the battery in order to increase its life.

Further developments come from using different strategies in order to keep the energy provided by battery close to zero, for example a depleting strategy allows to fulfil daily mileage required.

Figure 11 shows that the maximum current drained from the hybrid storage is about 320 A, 100 A of which are provided by the battery. The maximum current provided by the battery is the main difference between P3 and previous P1 and P2. Moreover, it brings a grea<sup>t</sup> benefit to lead–acid battery that has less voltage fluctuation (between 69 V and 78 V) than P1.

**Figure 11.** Battery voltage and current trends.
