Thermal analysis and simulation of a Li-ion battery pack for a lightweight commercial EV

Thermal analysis and simulation of a Li-ion battery pack for a lightweight commercial EV

Applied Energy 192 (2017) 159–177 Contents lists available at ScienceDirect Applied Energy journal homepage: www.elsevier.com/locate/apenergy Therm...

6MB Sizes 0 Downloads 31 Views

Applied Energy 192 (2017) 159–177

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Thermal analysis and simulation of a Li-ion battery pack for a lightweight commercial EV Paolo Cicconi ⇑, Daniele Landi, Michele Germani Università Politecnica delle Marche, via Brecce Bianche, Ancona 60131, Italy

h i g h l i g h t s  A knowledge-based approach to support the design of Li-ion batteries is proposed.  The proposed methodology is focused on the context of customized products.  A configuration tool implements design rules to define a battery layout.  An analytical model calculates the electrochemical heat generated by a Li-ion cell.  CFD simulations are used to investigate the thermal behavior of battery packs.

a r t i c l e

i n f o

Article history: Received 3 October 2016 Received in revised form 1 February 2017 Accepted 4 February 2017

Keywords: Electric vehicle Li-ion batteries Virtual prototyping Battery cooling Battery configuration

a b s t r a c t The scope of the present research is the reduction of cost and time related to the design, prototyping and testing of a Li-ion battery pack, which is used in commercial full electric vehicles using tools for rapid product configuration and simulation. This objective is particularly important for small companies that produce many different batteries in small lots. To develop the product design support system, a preliminary study was necessary. A 3D model was analyzed to simulate real thermal behavior, reproducing a real electric load using a standard ECE-15 cycle. Experimental tests have been conducted on the vehicle and battery to validate the model. An analytical thermal model was developed to evaluate the heat generated by electrochemical reactions inside a Li-ion cell. The outcome of this analytical model was used as the boundary condition in the CFD simulation of the battery model to evaluate the cooling behavior. The rules and results deduced from these studies have allowed the implementation of an easy-to-use knowledge-based configuration tool that supports the designer in the definition of the layout of the battery pack to save time and evaluate costs. As a test case, the battery for an urban freight vehicle was designed using the proposed approach. The achieved results show good performance and robustness of the simplified approach in terms of temperature distribution evaluation and design process efficiency. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Conventional freight vehicles have a significant impact in urban areas in terms of noise pollution, the emission of greenhouse gases (GHGs), road congestion, intimidation and safety for cyclists and pedestrians. Concerning the impact of urban freight transport on air quality, almost all freight vehicles are diesel-powered [1,2] and release particulate emissions, which can damage human health [3]. The particulate matter PM10 is a factor in conditions such as asthma, heart disease and respiratory diseases in urban areas [4,5]. Therefore, the improvement of air quality is generally a high priority for the city authorities [6,7]. The European Commis⇑ Corresponding author. E-mail address: [email protected] (P. Cicconi). http://dx.doi.org/10.1016/j.apenergy.2017.02.008 0306-2619/Ó 2017 Elsevier Ltd. All rights reserved.

sion provided air quality limits and a set of regulations and directives to promote the diffusion of more sustainable transport solutions [8]. Currently, a possible solution to reduce the dependence of road transport using oil-based fuel is to increase the use of low-emission vehicles (LEVs) such as electric vehicles (EVs) [9], plug-in hybrid electric vehicles (PHEVs), hydrogen-powered vehicles, natural gas vehicles, compressed natural gas (CNG)powered vehicles or liquefied natural gas (LPG)-powered vehicles [9]. Other solutions consider an integration of the LEV fleet management with logistical plans, regulations, incentives, ICT applications and intelligent traffic systems (ITS). The focus is on public and private transportation systems with special attention to freight delivery and intermodal solutions. Regarding LEVs, different public and private organizations consider EVs to be the best solution for a considerable reduction of the impact of road transport [10,11]. In

160

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Nomenclature BEV CFD DOD EV FEM FEV GHGs

battery electric vehicle computational fluid dynamics depth of charge electric vehicle finite element method full electric vehicle greenhouse gases

fact, EVs guarantee zero emissions in urban areas, high energy efficiency, low operation costs and low maintenance [12]. In this context, the development of the battery technology is essential to replacing the use of fossil fuel for urban freight transport. An electric powertrain requires a fast charging infrastructure, a battery with high storage capacity and long lasting high power, safety and low weight [13]. Battery vehicles and plug-in HEVs can also reduce noise pollution because their electric motors are extremely quiet compared to diesel engines. In addition, the size of EVs for freight delivery is limited to not preclude the battery autonomy; therefore, the use of smaller vehicles can improve safety and reduce disruption for pedestrians and cyclists. Quiet and small freight vehicles can also be used for night deliveries to avoid traffic congestion. Initially, the market for EVs and PHEVs was a niche market that focused on customized vehicles [14]. Electric light duty vehicles were produced as a conversion of traditional commercial vehicles such as lorries, vans and light trucks. Currently, various large companies sell electric vehicle floats for general purposes, and the market became more competitive [15]. These vehicles were able to implement modular kits for waste collection, the retail sector, express courier delivery or instant delivery (hotel, restaurants, catering, etc.). Today, the market take-up of the battery vehicles is still slow because of high vehicle costs, poor availability of charging stations, low autonomy and consumers’ uncertainty about the maturity of the technology. Despite this, small and medium enterprises (SMEs) have also been involved in the manufacturing and development of EVs, where the market is limited in size and a lack of manufacturing economies of scale continues to exist. For this situation, several capital investments are involved in the market of EVs and PHEVs [16]. The range of a battery vehicle is influenced by the weather conditions (neither cold nor hot climates are suitable), the driving cycle, and the technology employed. The nominal range of 150–200 km may be sufficient for freight delivery in urban areas; however, this range could decrease by 80 km if a vehicle operates outside of the standard conditions. The energy storage temperature is an important operative parameter in BEV with Liion batteries. In fact, while high temperatures decrease the efficiency of a Li-ion battery cell, enhancing the aging reactions and low temperature values can reduce the battery capacity. Additionally, temperatures over 70 °C or below 20 °C can damage the storage unit [17]. A uniform temperature distribution of battery cells is required to ensure good performance and battery preservation. In addition, the maximum temperature must not exceed 25 °C so as to not decrease the battery health and efficiency [18]. Today, SMEs are active in the field of tailored applications for the automotive industry as technological suppliers. Their effort is especially essential in the development of electric powertrain components such as batteries, motors, inverters and charging units. Energy storage units have an important role in EVs in terms of performance and economic impact over the lifecycle. In automotive terms, batteries are considered depleted when the original capacity is degraded by 80%; however, the remaining storage capacity is still useful in other Second Life applications. It is esti-

HEV KBE LFP NMC OEM SME SOC

hybrid electric vehicle knowledge based engineering lithium iron phosphate nickel manganese cobalt original equipment manufacturer small medium enterprise state of charge

mated that the re-use of powertrain batteries after their service life has ended may have the potential to offset the high initial cost of these batteries today [19]. Several researchers evaluated the environmental gain of possible Second-Life applications [20], while others focused on the economic return of this novel scenario [21]. Lithium ion (Li-ion) technology is widely used in rechargeable storage systems for automotive products. A battery pack is often composed of a set of Li-ion cells with nominal voltage from 3.2 to 3.8 V. Polymeric Li-ion cells, such as the LiFePO4 technology, offer high voltage, low self-discharge rates and high energy density, which is suitable for powertrain applications. The lithium titanate material (LiTiO2) is used in racing and high powered applications due to its ability to continuously provide a maximum peak of power 8 or 10 times higher than the nominal value. The main drawbacks of Li-ion batteries are their high cost [19] and long-term stability and safety issues related to the development of electrochemical reactions during the charge and discharge cycles. The exothermic reactions generate heat, which increases the temperature within each cell. The high range of temperatures depletes the Li-ion battery, causing aging reactions inside the battery and inducing risks such as damage and burning. Lithium inflammability is an important issue to consider during the design of a battery pack. One of the main issues of the Li-ion batteries is that the capacity of each cell, measured in Ampere-hours, degrades over a number of charge and discharge cycles. The capacity fade is the result of various factors, including irreversible electrochemical reactions that form a solid electrolyte interphase (SEI) in the negative electrode and oxidative reactions of the positive electrode [22] An accurate estimation and prediction of battery failure time provides information for the timely replacement of degraded batteries before the batteries reach the end of their useful life [23]. Operational factors such as the discharge rate, usage and storage temperature, and aging mechanisms are responsible for the capacity fade in Li-ion cells. Therefore, the operational temperature is a key factor in battery performance, life and safety. Thus, during the early embodiment design, it is important to consider the thermal battery behavior and evaluate the working temperature level to estimate the battery layout and cooling system. In this context, the issue considers the calculation and simulation of the heat source generated by electrochemical reactions. Generally, the configuration of a Li-on battery is a complex decision, which involves many factors such as energy density, cell properties, shape, size, and lifetime. Therefore, a configuration tool to support the design phase must consider all these issues. The paper aims to define an approach to reduce the cost and time related to the design, prototyping and testing of Li-ion battery packs that are used in full electric vehicles. The use of product configuration tools and simulation studies are highlighted in this paper. A configuration tool to support engineers in the definition of the layout for battery packs has been implemented. The context of the paper is the design of customized Li-ion battery packs produced in small batches. Concerning the simulation of Li-ion batteries, an analytical approach has been developed to evaluate the heat generated by electrochemical reactions inside Li-ion cells. CFD

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

(Computational Fluid Dynamics) simulations are proposed to investigate the effect of the battery layout on the cooling behavior. As a test case, a Li-ion battery pack for urban electric lightweight vehicles has been analyzed considering the ECE-15 condition as a driving cycle. The resulting prototype of the battery pack was tested in a real commercial vehicle. The remainder of the paper is organized as follows. The first part of Section 2 describes the technology involved in battery modeling. The proposed methodological approach is introduced in addition to the design tools involved. Section 2 also describes the configuration tool developed to define the layout of the battery packs. In Section 3, the test case, which focused on the electric retrofit of a freight vehicle, is proposed. The phases of the battery design are described according to the proposed approach. Finally, s 4 and 5 include discussions and the results.

2. Methodological approach 2.1. Traditional design tools In this sub-section, an overview about the traditional practices and tools for the design of Li-ion batteries is presented. Currently, manufacturers of automotive batteries use different approaches for the design of Li-ion battery packs. The choice depends on the number of batteries to be produced. Typically, experience is important in SMEs, where the traditional design approach is based on CAD modeling and trial-and-error procedures. In this case, the battery designer uses basic calculation tools to analyze and develop new solutions. He/she can calculate the electrical layout configuration and define the geometrical shapes and the size of the cooling system. In this traditional workflow, the model validation is based on the physical prototyping, which gives feedback about performance, quality and safety. Any design error requires additional time-expensive iterations with negative effects over the time-to-market. Documents such as drawings, tables and reports are considered basic tools to support the engineer during the design phase. The design of the tailored commercial vehicles and their parts requires rapid and flexible tools and methods to reduce the timeto-market (TTM) [24]. This target view is necessary in the context of modern SMEs, where OEM producers require an efficient and lean supply chain [25]. Virtual prototyping (VP) techniques are widespread in mechanical and electrical companies to reduce the time and cost related to many physical prototypes. Even if VP tools and methods are included in the traditional design procedures of large companies, the same approach is different in the context of SMEs. In fact, the complex simulations regarding multi-physic and multi-domain phenomena introduce high computational costs and long processing times if standard hardware is employed. Additionally, different VP tools still entail high setup costs and require trained staff. Concerning the design of Li-ion batteries, traditional commercial CAE (computer aided engineering) tools include features that calculate the electrical and thermal behavior of Li-ion polymeric cells, but a deep electrochemical characterization of materials and reactions is necessary. However, the main technical data of polymeric cells are difficult to obtain because they are related to the battery chemistry and technology employed during the manufacturing process. Therefore, the design and configuration of Li-ion batteries is often defined using electronic spreadsheets with macros and databases, and using a test bench is the only way to validate a prototype. An example of an analytical tool that supports the battery design is the BatPaC software, which was developed by Uchicagto Argonne [26]. This tool can predict the battery performance and

161

cost. BatPaC considers whether the battery production is focused in a single plant and whether the assembly line can feed a single product or various products. Regardless, several enterprises have implemented the use of computer-aided engineering (CAE) systems to investigate the performance of virtual products without using physical prototypes. CAE tools, in the context of the Li-ion battery design, can be used to evaluate heat generation and electrochemical reactions. Many commercial tools based on finite element methods (FEM) offer advanced features that predict cell performance by solving electrochemical formulas, while CFD tools can verify the cooling behavior [27]. Several researchers propose equivalent electrical circuits to correct the electrochemical behavior of a Li-ion cell [28]. Another approach proposes the online calculation of internal short circuit detection [29] to calculate the internal temperature of a Li-ion cell. Another approach proposes a 3D electrochemical-thermal coupled model to simulate various internal short circuit scenarios for large format lithium ion batteries [29]. This approach can predict the voltage and temperature responses of the battery. However, there is a lack of design tools that are able to support the engineer in all phases of battery design. One of the latest innovative methods proposed in the literature is based on the FOSTER network. Starting with the initial cell characterization and the FEM analysis of a single module of cells, a network comparison predicts flow cooling behavior and superficial cell temperatures at different current rates [30]. The single cell characterization considers the linear model of generated and transferred heat, and the final temperatures are the weighted sum of the contribution of each cell. In all design cases, experimental data are necessary to calibrate the analytical models and simulations. Another approach to evaluate the heat generation inside Li-ion cells is to consider the battery element as a body and calculate the electrochemical heat on the boundary of its domain [18] without having a finite difference method applied to each sub-layer of the Li-ion cell. The paper proposes this analytical method to reduce the complexity of the thermal analysis while considering a cell as a 0-D domain. This approach simplifies the calculation of the electrochemical heat using a rapid analytical method with experimental data. The only commercial CAE tools required in this approach is the CFD solver, which can support the designer in investigating the cooling behavior by changing geometric and operative parameters. 2.2. Design platform A methodological approach has been defined to support the designer during the early phase of a Li-ion battery project. The scope of the research is to provide a solution to reduce the cost and lead time in the context of SMEs and small production. Specifically, this research aims to apply virtual prototyping tools and methods to support the engineer in the early evaluation of the battery cooling. In fact, during the early design phase, correct estimation of the battery thermal behavior can reduce time and costs related to physical prototypes and tests. As cited in Section 1, the thermal behavior of a battery is an important issue to guarantee high energy efficiency and prevent failures and damage. The proposed design platform described in Fig. 1 reports the phases of the methodological approach analyzed in this research. This platform is divided into 4 modules: HD tools, KBE, DB and VP tools. The use of these tools focuses on the design of tailored batteries. The input is the specification of the battery with the electrical layout, the operative conditions, and the type and size of the Li-ion cells to be involved. Specifically, the electrical layout concerns the information related to the cell count and the type of electrical connections. The operative conditions must be elaborated upon by the designer, who calculates the current and voltage profiles related to an operation such as real driving cycles. Finally, data about a Li-ion

162

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 1. The battery design platform.

cell concerns the electrical capacity, energy and power density, heat capacity, thermal conductivity and safety limits. The output of the proposed design platform is the geometric layout of the battery pack, the simplified 3-D model, the definition of the cooling parameters and the distribution of the temperature in relation to the operative conditions. The first module, called HD tools, concerns electrical testing that is focused on a Li-ion cell. Typical electrical tests involve the acquisition of voltage curves and open circuit voltage (OCV) in different conditions for the temperature, SOC and current rate. The result of these experimental tests is used to characterize the model of a Liion cell to solve the electrochemical issue (Eq. (1)). The second module, known as KBE, includes 3 knowledge-based tools that support the designer during the technical analyses to estimate the thermal behavior of a battery pack with the cooling performance. These tools have been implemented using the development environment of Microsoft Visual Studio.NET. Specifically, this module is divided into 3 different levels: Testing Configuration, Cell Analytical Thermal Analysis, and Pack Layout Configuration. The first phase, which is related to the Testing Configuration, includes the definition of the electrical tests used to characterize the model of a Li-ion cell using HD tools. In this phase, the rules allow the user to recover historical data from a DB or choose from a list of electrical tests to be performed. Fig. 2 describes the tool implemented for the management of the electrical tests, which support the characterization of a Li-ion cell. The phase Cell Analytical Thermal Analysis (inside the KBE module) concerns the use of a knowledge-based tool that is connected to a parametric spreadsheet to estimate and quantify the heat source generated by a Li-ion cell. Specifically, this second tool solves the thermal model of a Li-ion cell through the solution from the electrochemical equations. In this calculation, the boundary conditions are related to the operating conditions. Cell parameters such as OCV are acquired by a database (DB) or from the result of experimental tests (HD tools). Fig. 3 describes the form of the tool implemented for the analytical calculation of the heat generated by a Li-ion cell. The KBE tool, Pack Layout Configuration, supports the designer in the definition of the layout of a battery pack arrangement (Fig. 4). This tool implements rules to support the engineer in the definition of a geometric layout and cell arrangement. A database provides a collection of template models that can be rapidly resized using parametric functions implemented inside a CAD system. The employment of a KBE approach aims to reduce

the time related to the phase of product configuration because the proposed tool supports the designer during the decisionmaking process. The layout configuration process is described in more detail in Section 2.3. The third module, introduced in the proposed design platform, is called VP tools and concerns the employment of 3-D simulations. More specifically, two levels of virtual prototyping are proposed: the FEM thermal analysis that is focused on a single Li-ion cell (Cell Thermal Analysis) and the CFD analysis extended to the complete battery pack (Pack CFD Thermal Analysis). The first level of the VP tools module considers the analysis of the heat distribution within the body of a Li-ion cell. In this study, the heat source, calculated by the Cell Analytical Thermal Analysis tool, is an input used to solve the thermal distribution of a Li-ion cell during the operation. The second level of the VP tools module concerns the virtual prototyping of a Li-ion battery pack. This analysis is focused on the thermal and cooling behavior during the battery operation when considering the arrangement of all cells. The output is the simplified 3D model of the battery pack with the cooling analysis. The report of the cooling analysis includes the temperature distribution across each cell and the field of the fluid flow. This type of analysis, which considers the interaction between solids and fluids, can also estimate the differences in terms of temperature between the hottest and coldest cells. A deep virtual study of the battery thermal management can improve the product safety and energy efficiency. The last module described in Fig. 1 is the database (DB) that collects data and documents from electric tests, CAD models and templates. The CAD system was considered a tool within our design platform, and it can be seen as a horizontal tool that interacts with all modules, such as KBE, DB and VP tools, during the design process. 2.3. Layout configuration As described before, a prototypical configuration tool was developed to support the designer during the configuration of a battery pack. The outcome concerns the geometric definition of a battery layout considering the dissipation of the battery heat to improve the air cooling. Using this tool, the designer configures the 2D layout of a battery using a user-friendly GUI described in Figs. 4 and 5. While the designer defines the configuration properties and the related geometric parameters, an algorithm elaborates the resulting 2D scheme from a library of CAD templates. More-

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

163

Fig. 2. The Testing Configuration tool.

Fig. 3. The Cell Analytical Thermal Analysis tool implemented to estimate the electrochemical heat power related to a Li-ion cell.

over, a generation routine, which was developed using API functions and VB scripts, can automate the modeling of a simplified 3D battery pack with every cell inside. The CAD library also includes pre-assembled 3D models to reduce the complexity during the generation of a battery model. The simplified 3D model will

be the input for the following virtual thermal analysis, where the heat produced by the electrochemical reactions is calculated using the proposed analytical calculation. The GUI highlighted in Fig. 4 was developed with the scope to enhance the reuse of past battery arrangements and to include

164

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 4. The main interface of the Pack Layout Configuration tool for a configuration of cylindrical cells.

templates and knowledge-based rules. Implicit rules and formulas have been implemented to pre-calculate the conditions of the cooling flow and the value of the geometrical parameters required for the definition of the battery layout. While implicit rules have been simplified using lookup tables, formulas have been described using the mathematical equations of the heat exchange. All formulas and any database can be updated by the design user. One of the first tasks in the proposed configuration is the definition of the type of Li-ion cell given its size. While some Li-ion batteries can be assembled in one pack, others require an arrangement in different modules. Therefore, this tool considers both solutions and allows the user to choose. Fig. 6 describes some possible solutions for the battery layout. Specifically, the shape of the cell arrangement and the fluid direction are considered parameters that can characterize the pack layout configuration. While the shape can be linear (Fig. 6a and b) or offset (Fig. 6c and d), the fluid direction can be parallel (Fig. 6a and c) or crossed (Fig. 6b and d). The parallel flow involves a symmetry between the inlet and outlet. Generally, the choice of the pack shape and the fluid direction is influenced by the type of Li-ion cell involved. More specifically, cells can be classified by geometric shape: cylindrical, prismatic and soft-pouch (Fig. 7). While the cylindrical cells are suitable in a crossed pattern layout, the others are often used for linear arrangements. Fig. 4 shows the parametric definition related to an offset arrangement of a cylindrical module of a battery pack on the left.

3. Test case The proposed methodology has been used to support the design of a Li-ion battery for a small urban freight vehicle (Fig. 8). This vehicle is a prototypical full electric pickup truck with chassiscab. As described in Fig. 8 and Table 1, the proposed freight vehicle is an N1-category truck [31] with 3 seats and a back motor axis. The target vehicle is a retrofit of an existing gasoline-fueled van, which had been designed for urban use with a maximum velocity of 80 km/h. This paper highlights the proposed design workflow focused on the Li-ion battery. The comparison between the different Li-ion cells is proposed in the following sections as well as the virtual thermal modeling related to one-cell components and the complete battery pack. The target of the test case is to define a suitable Li-ion cell to apply to the battery pack of the proposed freight vehicle considering the thermal behavior and cooling performance to provide high energy efficiency and safety. The vehicle has been tested in accordance with EU normative guidelines and take the NEDC (New European Driving Cycle) conditions into account [32]. The NEDC is introduced by UNEC (United Nations Economic Community for Europe) as the most representative typical usage of a car in Europe [32]. This driving cycle (Fig. 9) consists of four repeated ECE-15 Urban Driving Cycles (UDC) and an Extra-Urban driving cycle (EUDC) and falls under the European vehicle regulation UN ECE R 101.

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

165

Fig. 5. The main interface of the Pack Layout Configuration tool for a configuration of soft-pouch cells.

Fig. 6. Examples of some battery plant layouts: (a) linear structure and parallel air flow; (b) linear structure and crossed air flow; (c) offset structure and parallel air flow; (d) offset structure and crossed air flow.

Fig. 7. Different geometrical configurations of Li-ion cells (cylindrical, soft-pouch, and prismatic).

Since the proposed vehicle has a maximum velocity of 80 km/h, the NEDC driving cycle was limited to this velocity value. For this paper, the real and virtual vehicle model was generated through an analysis of the ECE-15 driving cycle, which is limited to 50 km/h. Fig. 10 describes the comparison between the ideal ECE-15 driving cycle and the real driving cycle analyzed with a traditional freight vehicle. The ECE-15 driving cycles were reproduced in a laboratory using a test bench for cars and light vehicles. While the maximum speed is 50 km/h, the average value is 19 km/h with a 1 km path. An ECE-15 cycle consists of low acceleration phases,

stops and constant speed paths. Even if some critics show how NEDC and ECE-15 do not reproduce real driving cycles, the real tests were useful for defining the electric traction motor to apply in the prototypical electric vehicle. After the analysis of these tests, a 90-kW electric motor with 260 V was chosen as the substitute of the traditional 1.2-liter ICE engine in the retrofit vehicle. As described in previous chapters, the use of the presented approach allows for a time reduction in the early design phase, a reduction in terms of cost, and an increase in market competitiveness for an SME. To validate the proposed methodology, different

166

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

tests on different types of cells, including cells with different geometries, chemistry and capacities, have been carried out. Therefore, different types of batteries that are suitable for electric mobility applications have been analyzed. As a first step, electrical tests were performed to support the determination of the physical parameters, which are necessary for the cell thermal characterization of the model. In the following sections, we present the different types of cells analyzed, the resulting electrical tests, simulations and battery pack configurations. Finally, the design of a cooling system is considered for a hybrid electric vehicle. 3.1. Cell electrical test

Fig. 8. The prototypical test case vehicle (an electric freight vehicle).

Table 1 Main technical details of the analyzed vehicle. Vehicle category:

Chassis N1

Side of guide: Cab seats Motor axis Maximum ground weight Empty chassis weight Length Width Height Rear overhang

Left or right 3 Back 3500 kg 1750 kg 4770 mm 1770 mm 1955 mm 1240 mm

As previously described, to determine the heat generated by a Li-ion cell, it is necessary to measure the trend of voltages and currents during different phases of charging and discharging. Therefore, a battery cycler was introduced to charge and discharge each Li-ion cell. During the charging, the alternative current (AC) from the grid is converted to direct current (DC) to increase the battery state of charge. The cycler delivers electricity at different levels of the current rate as scheduled and in accordance with the limit values provided by the manufacturer of the Li-ion cell. However, during the battery discharge at the test bench, the DC current is converted to AC. Fig. 11 shows a basic representation of the laboratory test equipment, which was used for the acquisition of the experimental data. The main components are the cycler, host PC, IR camera and Li-ion cells to be tested. This equipment allows the main electrical quantities to be acquired and monitored. These quantities are the voltage (V), current (A) power (W), capacity (Ah) power (W), SOC (%), and internal resistance (ohms). Before each cell test, the user programs every run of the cycler though the host PC. The acquisition is the runtime. An infrared camera provides the monitoring of temperatures during the electrical tests within a climatic chamber. Post-monitoring software allows the temperature values of the cells (minimum, maximum and average values) to be reported in tables and graphs. The temperature data are useful for supporting the definition and validation of the battery thermal model. Table 2 highlights the main characteristics from the datasheets of the three types of cells, which are analyzed in this paper. These Li-ion cells are softpouch models with different chemistries (LFP and NMC) and capac-

Fig. 9. The NEDC driving cycle.

Fig. 10. The comparison between the ideal and real ECE-15 driving cycle.

Fig. 11. The scheme of the battery test equipment.

P. Cicconi et al. / Applied Energy 192 (2017) 159–177 Table 2 Datasheet data for the analyzed battery Li-ion cells. Type

1

2

3

Chemical Geometry Dim Length Width Thickness Nom. Voltage Nom. Capacity Max discharge Max charge Weight

LiFePO4 Soft Pouch 222 mm 129 mm 7.2 3.25 V 14 Ah 140 A 14 A 380 g

LiNiCoMnO2 Soft Pouch 216 mm 129 mm 7.2 3.65 V 20 Ah 100 A 20 A 425 g

LiNiCoMnO2 Soft Pouch 327 mm 453 mm 10.7 3.65 V 150 Ah 300 A 150 A 3300 g

167

ities. While the first cell is an LFP element with a capacity of 14Ah, the second and third are NMC cells with 20 Ah and 150 Ah of capacity. To choose the more suitable Li-ion cell, the proposed cells have been tested at different current rates of charge and discharge according to the datasheets. The profiles of the voltage during the charge and discharge for each cell are reported below in Figs. 12, 13and 14. The voltage curves are plotted as a function of SOC (charging phases) and DOD (discharging phases). Each test was carried out at a constant temperature of 25 °C inside the climatic chamber in natural convection conditions. All experiments were performed under natural convection conditions to reduce the heat dissipation during the testing of each cell. The aim of this

Fig. 12. Voltage curves 1C in the charge and 1C, 5C, 10C in the discharge of a soft-pouch cell of 14 Ah.

Fig. 13. Different profiles of charge at 1C, and 2C and 1, 2, 3, 5C in the discharge of a soft-pouch cell 20 Ah.

Fig. 14. Voltage curves (0.5C and 1C) in the charge and 0, 3C and 2C in the discharge of a soft-pouch cell of 150 Ah.

168

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 15. Comparison between OCV curves.

approach was to validate the estimation of the heat generated inside a Li-ion cell and the thermal distribution. While the previous graphs describe the voltage curves, the following graphs show the OCV (open circuit voltage, also known as E0) analysis. Fig. 15 shows the comparison of the OCV curves for each cell tested (14 Ah, 20 Ah and 150 Ah). During the OCV tests, the cells are gradually discharged at a rate of 0.3C. Each 5% SOC, the test is stopped, and there is an 8 h waiting period before measuring the final OCV value. Although the waiting phase usually lasts 8 h, in this case, only 3 h of waiting was used to acquire the OCV values. After a complete discharge, the same cell was recharged again to 0.3C 5% in the SOC phase to obtain two OCV profiles for the charge and discharge. Fig. 15 shows the OCV trends for each cell analyzed at the test bench. 3.2. First analytical thermal analysis This section focuses on the evaluation of the electrochemical heat source, which is generated within Li-ion cells. As mentioned previously, this type of thermal analysis is important to calculate the temperature inside a Li-ion battery pack. Before showing the results regarding the thermal analysis of Li-ion cells, a short introduction to the thermal issue is provided to understand the technology involved in this topic. Afterwards, the calculation approach was applied to each battery cell described in the previous section. The calculation of the heat generated inside the Li-ion cells is a complex process that requires an understanding of how the reaction rates change as an effect of the temperature variation. Many researchers have studied the thermal issue related to cells and battery packs. While some proposed numerical solutions based on 1-D mesh with uniform heat generation, others studied coupled electrochemical–thermal models applied to a 3-D domain. Generally, the heat generated by a Li-ion cell depends on three phenomena: the activation of the interfacial kinetics, the concentration of the species transport, and the Joule heating (Ohm losses), which concerns the movement of particles during the charge and discharge phase. The first important study of an electrochemical and thermal analysis of lithium-ion battery cells was proposed by Bernardi [33]. In her analysis, the heat generated depends on the thermodynamic equilibrium inside a battery cell. She applied the first law of thermodynamics to the domain of the cell as described in Eq. (1) [33].

@E0  Q_ ¼ IðV  E0 Þ  IT @T

Z X X av g j  H  av g Þ @cj dt DH i r i  ðH j @t i j

ð1Þ

The letter Q indicates the quantity of heat generated per unit of time, V represents the cell voltage, E0 is the OCV value, I is the current (>0 in charging and <0 in discharging) and T is the tempera-

ture. The term DHavg describes the variation of enthalpy for a i chemical i-reaction, ri is the rate of the i-reaction, Havg represents j the partial molar enthalpy of species j, and cj is its concentration. The term t is the time, t is the volume, and the apex ‘‘avg” indicates a property evaluated at the averaged volume concentration [34]. The thermal approach, which is proposed in this paper, is based on a study developed by Thomas and Newman [35,36]. This study evaluates the heat produced by a single cell simplifying Eq. (1). In fact, Eq. (1) can be reduced to the first two terms. While the first term represents the exothermic and irreversible heat (Joule effect), which depends on the cell’s internal resistance (see Eq. (2)), the second term is the heat due to the entropy changes related to specific reactions. Additionally, this second term can be either endothermic or exothermic in the function of the current and the state of charge (SOC). This term indicates the reversible heat generated and can be replaced by Eq. (3), which calculates the ratio oE0/oT with the term DS/nF, where DS is the entropy variation, F is the Faraday’s constant and n indicates the number of exchanged electrons in the reaction.

Q_ irr ¼ IðV  E0 Þ ¼ I2 Ri

ð2Þ

@E0 I Q_ r ¼ IT ¼ T DS nF @T

ð3Þ

The last two terms in Eq. (1) can be neglected [35] because one depends on side reactions accounting for aging, which are assumed to be slow enough to be neglected, and the other term is the heat of mixing, which is generated from the formation and relaxation of concentration gradients within the cell. This term can be considered almost zero because the materials used have good electrochemical transportation properties; thus, the concentration gradients are limited, and the heat of the mixing can be ignored. Considering the dimensions and materials of a soft-pouch cell, it is correct to assume that these last two terms of Eq. (1) are negligible. Therefore, the resulting formula, which describes the quantity of the thermal power produced, is expressed by Eq. (4).

@E0 Q_ ¼ Q irrþ Q r ¼ IðV  E0 Þ  IT @T

ð4Þ

The proposed analytical approach aims to solve the thermal problem of a single Li-ion cell considering the effect of the electrochemical heat and the convective effect related to the cooling fluid and geometry. The output of this analytical calculation is the estimation of the heat dissipation and the average maximum temperature achieved by a Li-ion cell during operation. As the assumption, the irradiative condition was not considered in the analytical solution focused on one battery cell. Regardless, the irradiative equations were considered in the CFD analysis focused on battery

169

P. Cicconi et al. / Applied Energy 192 (2017) 159–177 Table 3 Report of thermal power calculated for each cell in a discharge state.

Fig. 16. Thermal balance for one cell.

packs. Fig. 16 describes the energy balance between the electrochemical phenomena and thermal dissipation from a convective flow. While the following sections describe the computational analysis for each soft-pouch Li-ion cell described, Fig. 17 reports a comparison between the calculated values of the heating power generated at different current rates (1C, 5C and 10C) from the 14-A soft-pouch cell. This graph gives feedback about the thermal magnitude related to each cell. Each discharge phase is described in the function of DOD (depth of discharge). Generally, even if the heat generated is almost constant during a discharge at a constant current rate, the battery temperature increases at the end of the discharge (or charge). In fact, the most severe conditions occur at the end of the charge or discharge. During operation, part of the thermal heat is dissipated by the cooling system, but another part increases the internal cell temperature according to its thermal capacity. Table 3 shows the maximum values of thermal power during the discharge, which are calculated for each proposed cell (14 Ah, 20 Ah and 150 Ah) by the analytical formulation. The 150-Ah cell reaches 180 W in a 2C condition, which means reaching 300 A. The most serious thermal condition for the 14-Ah cell (Table 3) is the rate of discharge at 10C, where the output power reaches 21 W at 140 A. While the heat generated mainly depends on the operating conditions of the battery and the current rate and depth of the charge or discharge, the calculation of the average temperature of the battery is influenced by the type of battery, chemical and geometrical dimensions, and the external environmental conditions, which determine the flow of heat by convection and radiation. To validate the thermal output values reported in Table 3, in the following sections, the calculation of the temperature through the analytical approach is shown as a comparison with the real developments. 3.3. Cell virtual thermal analysis The computational analysis concerns the thermal simulation of cell behavior using FEM software, which includes a CAD kernel for

Charge @ 1C Discharge @ 1C Charge @ 2C Discharge @ 2C Discharge @ 3C Discharge @ 5C Discharge @ 10C

14 Ah

20 Ah

150 Ah

3,0 W 3,5 W 4,0 W 5,0 W 6,5 W 8,0 W 21,0 W

10 W 12 W 23 W 27 W 42 W 170 W –

37 W 42 W – 180 W – – –

the geometry modeling. The use of this virtual prototyping tool allows the designer to verify the functionalities and performance of a virtual product, reducing the cost of physical tests. Generally, several geometrical parameters and cooling conditions can be evaluated using a virtual analysis. As cited before, this research investigates two different levels of virtual analysis using simplified models: the thermal behavior of a single cell and the cooling of a battery pack. The thermal solution of a cell element is based on the calculation of the heat source as described in the section for the Analytical Thermal Analysis. Using an FEM solver, the designer can directly model or import the battery-cell geometry and define the pre-processing parameters, such as the mesh and boundary conditions. The boundary conditions concern the convective heat transfer and the environmental data, such as temperature and relative humidity. While the value of the electrochemical heat source is set according to the analytical calculation, the solution of the thermal profile is calculated by the equation of the FEM solver. The following sections report the FEM simulations regarding each proposed Li-ion cell at constant discharge rates. Each simulation was compared to the experimental testing to validate the proposed approach. 3.3.1. Thermal behavior of the 14-Ah cell The 14-Ah cell has a smaller capacity than the other three battery elements. Despite the low capacity, this type of LFP cell provides high-power density. In fact, this cell can be discharged at a 10-C rate continuously. However, a discharge current of 140 A (10C) causes a strong increase in temperature as described in Fig. 18, which shows the real trend of temperature in different discharge conditions (1C, 5C and 10C current rate). Fig. 18 also shows the comparison between the real trend of temperature and the values calculated by the proposed analytical method, which simulates the same condition. The values of the heat source, which reproduces the electrochemical heat, were taken from Table 3 when solving Eq. (4). While the operative temperature condition was set to approximately 24–25 °C, the only thermal exchange condi-

Fig. 17. Comparison between the analytical calculation of heat generated at a 1C, 5C and 10C discharge for the 14-Ah cell.

170

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 18. The comparison between the real temperature profiles and the analytical ones for different discharge rates (1C, 5C and 10C) applied to a 14-Ah cell.

However, the curves of the voltage are not constant during the charge and discharge phases (Fig. 14). The presence of cobalt makes these battery cells expensive to produce but also more profitable to recycle. Fig. 21 shows the trend of temperature in different discharge conditions (1C, 2C, 3C and 5C current rates), where the temperature values were measured by an IR camera. While Fig. 22 reports the comparison between the real temperature profile, which was acquired using an IR camera, and the temperature values calculated by the virtual method for a 5C discharge, Fig. 23 shows the same comparison at a 2C current rate. In both cases, the gap between the real and virtual values is approximately 5%.

Fig. 19. A comparison between the real temperature range for a 1C charge and the analytical values calculated for the 14 Ah cell model.

tion analyzed was the natural convection. Afterwards, the virtual Li-ion cell model was validated by physical tests using an IR camera for the monitoring of the battery temperature (Fig. 20). Fig. 19 shows the difference between the real temperature profile and the values calculated by the proposed analytical method for a constant charge at 1C condition. In all proposed cases, the gap between the experimental and analytical values does not exceed 5%.

3.3.2. Thermal behavior of the 20-Ah cell The 20-Ah cell is an NMC type (LiNiCoMnO2). This type of Li-ion battery can achieve a higher peak of voltage than the LFP cells.

3.3.3. Thermal behavior of the 150-Ah cell This type of Li-ion cell (150 Ah) is used particularly in automotive and industrial applications due to its high capacity and power density. The NMC chemistry provides a high peak of voltage during the charge and discharge cycles; however, the voltage curve is not constant (Fig. 14), as discussed for the 20-Ah cell. Fig. 24 shows the temperature curves acquired during the discharge tests conducted at different current rates (1C and 2C). Figs. 25 and 26 show the comparison between the virtual temperatures and the real values monitored using an IR camera. Fig. 25 shows the case of a continuous discharge at a 3C current rate (which means 300 A). As described in previous figures (Fig. 25, Figs. 22 and 23), there is a small gap of approximately 5% between the calculated profile of the temperature and the real data monitored using an IR camera. While Fig. 27 shows differences between the minimum and maximum temperature values, which are mon-

Fig. 20. Thermal images related to the discharge for 1C (on the left) and the discharge for 10C (on the right).

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

171

Fig. 21. Temperature behavior for different discharge rates (1C, 2C, 3C and 5C) of the 20-Ah battery.

Fig. 22. A comparison between the real temperature range for a 5C discharge and the analytical values calculated for the 20-Ah cell model.

Fig. 23. A comparison between the real temperature range for a 2C charge (acquired by an IR camera, green solid line) and the analytical values calculated (blue solid line) for the 20-Ah cell model. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

itored after a 3-C discharge, Fig. 28 reports the map of temperatures simulated with an FEM tool. The comparison highlights a narrow gap between the numerical simulations and physical testing. 3.4. Pack configuration According to the proposed methodological approach, a configuration tool was used to support the configuration of the pack lay-

out. The configuration was based on the selection of soft-pouch Li-ion cells with 150 Ah, as discussed in the previous section. The battery pack consists of 6 modules with 16 cells per block. The external air temperature is 25 °C with 55% r.h. Additionally, the battery pack is air-cooled. Fig. 29 shows the basic layout configured for this test case. The Pack Layout Configuration tool suggested a gap distance between each module and an air velocity of approximately 20 m/s. Table 4 shows the dimensions related to the geometrical parameters highlighted in Fig. 29. The dimensions of the analyzed battery are 1570 mm  480 mm  380 mm. The value ZM represents the height of each module, and ZL is the battery height considered. The following figure highlights the 3D model related to the defined battery pack (Fig. 30). A layout with a linear cell arrangement has been chosen since the geometry of each battery element is a soft-pouch. Additionally, the implemented rules suggested a parallel flow against a crossed flow, which is more suitable for cylindrical elements. The battery nominal voltage is 350 V, and 200 A is the maximum discharge current. The nominal capacity is 52 kW h with 70 kW of maximum peak power. The battery housing is stainless, while the cooling system consists of two compact fans (100 mm in diameter) arranged on the top of the pack. Both fans pull air from the battery pack to the exterior. The air inlet sections are arranged in the frontal face to capture the cool air flow from the front area of the vehicle.

172

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 24. Temperature behavior for different discharge rates (1C and 2C) of the 150-Ah battery.

Fig. 25. A comparison between the real temperature range for a 2C discharge and the analytical values calculated for the 150-Ah cell.

Fig. 26. A comparison between the real temperature range for a 1C charge and the analytical values calculated for the 150-Ah cell model.

3.5. Virtual analysis of the battery pack The virtual analysis of a battery pack is an important issue for the designer in the definition of the cooling system. This approach allows for an analysis and evaluation of the operative temperatures of the Li-ion cells inside the battery. The achieved temperatures depend on the geometric layout, cooling size and working conditions. The CFD simulations can reproduce the thermal behavior of a battery pack under different conditions of load and air cooling. In this paper, the activity of the virtual analysis is focused on the battery simulation with an ECE-15 driving cycle. In these simulations, the profile of the electrochemical heat is an input calculated by the analytical approach discussed in the previous sections.

Fig. 27. Thermal images related to the discharge of 3C for a 150 Ah battery.

Fig. 28. Temperature map simulated in a natural convection for a 150 Ah battery at a discharge of 3C.

The estimation of the current profile, which is related to the ECE-15 cycle, has been evaluated to solve the simulation of the vehicle’s system. This analysis was developed using the Portunus software, which is a commercial tool used to solve mechatronic systems such as EVs and HEVs. The virtual system of the vehicle considers the main physical force applied during the motion, including the aero drag and the rolling resistance. Generally, when a vehicle is moving, the powertrain’s engine must overcome different resistance force, such as ordinary and accidental force. The ordinary force includes rolling resistance, friction resistance, aerodynamic resistance and resistance related to the road irregularities [37]. Accidental resistance involves the acceleration and deceleration phases and the presence of uphills, downhills and curves.

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

173

Fig. 29. The layout of the analyzed battery pack.

Rolling resistance (Eq. (7)), which is due to contact between the wheels and the road, has been calculated using an experimental formulation that embeds materials, velocity (v) weight, road inclination (a), the friction coefficient (b0) and rolling friction coefficient (b1).

Table 4 Chart dimensions related to the battery layout. XL XS XM YL YS YM ZL ZM

1570 mm 40 mm 470 mm 480 mm 40 mm 180 mm 340 mm 380 mm

F Rolling ¼ ðb0 þ b1  e6  v 2 Þ  m  g  cos a

ð7Þ

The drag force is caused by the resistance caused by the air during the motion of a vehicle (Eq. (8)). The fuel consumption decreases if the vehicle’s aerodynamics are not good. Specifically, the truck performance is strongly influenced by the aerodynamics of its shape, which is described by the drag coefficient (cx). The aerodynamic drag resistance also depends on the reference area (A), mass density (q) and flow velocity (v).

F Aero ¼

1 ðcx  q  A  v 2 Þ 2

ð8Þ

The force due to the vehicle’s weight is only considered in downhill and uphill paths (Eq. (9)), and it is related to mass (m) and acceleration (g). However, this paper does not consider this force because the simulated ECE-15 driving cycle is a flat path.

F Weight ¼ m  g  sena

Fig. 30. A 3D frontal isometric view of the analyzed battery pack showing the two frontal inlet sections and the three top outlet sections.

As a first step, the power levels were calculated by analyzing the real driving cycle of an ICE vehicle. This test vehicle was the same model as the proposed EV but with a 1.2-L gasoline engine. The virtual system analyzes data from the real driving cycle to estimate the required power from the sum of all force applied to the vehicle, such as the inertial force, friction losses, drag resistance, weight force uphill and downhill, and accessory loads such as cooling.

F Engine ¼ F Net þ F Rolling þ F Aero þ F Weight

ð5Þ

Eq. (5) describes the resulting force applied to a vehicle when it is moving. The net force (Eq. (6)) is calculated as a scalar multiplication between the vehicle’s mass (m) and its acceleration (a), which is the derivative of the velocity over time.

F Net ¼ m  a

ð6Þ

ð9Þ

Fig. 31 shows the power profile related to the proposed vehicle. This profile was calculated using the model implemented within the system simulation software. By considering the power profile, the characteristics of the electrical components (such as battery, motor and inverter) and their efficiencies, a current profile was generated related to the electric freight vehicle during the ECE15 cycle (Fig. 31). This current profile was used to predict the thermal loads produced by the electrochemical reactions in the Li-ion cells. In summary, the current profile described in Fig. 31 was used to compute the time-dependent heat source to apply a boundary condition in each virtual Li-ion cell (CFD model). As a result of the CFD analysis, Figs. 32 and 33 show the temperature level reached by the battery model at the end of the ECE-15 cycle regarding the vehicle in the full load condition (3500 kg). The simulation was calculated with the air temperature at 25 °C and a cooling air flow rate of 110 m3/h. The maximum battery temperature is almost 29 °C in this simulation, while the minimum value is almost 25 °C. The gap between the maximum and minimum temperatures is approximately 4 °C. A further simulation was conducted to evaluate the effect of the cooling system. Therefore, the same battery pack was simulated, as described above, under the same operating conditions but without

174

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 31. The calculated profile of power [W] for the ICE vehicle during the ECE-15 driving cycle (red line), and the estimated profile of the current for EV during the ECE-15 driving cycle. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 32. The battery temperature map and the velocity streamlines related to the ECE-15 simulation at a full load condition.

Fig. 33. The battery temperature map (in the middle section) related to the ECE-15 simulation at the full load condition.

the air cooling system. The battery housing was considered closed. The inlet and outlet sections were not permeable in this second simulation. Figs. 34 and 35 show the resulting temperature profile. The achieved temperature distribution shows a uniform distribution. The gap between the maximum and minimum values is close (less than 1 °C). However, the average difference between the configuration with and without the analyzed air cooling is approximately 5 °C for the analyzed driving cycle.

3.6. Electric vehicle: testing A prototype of the resulting battery pack was assembled and tested within the freight vehicle (Fig. 36) to validate the approach and the proposed model. As cited before, the prototypical vehicle is a retrofitted van. The electric powertrain has the characteristics described in Table 5. The analyzed battery pack consists of 6 modules with 16 elements (150-Ah cells). Each cell should provide at

175

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 34. The battery temperature map related to the ECE-15 simulation at full load conditions without cooling.

least 2000 cycles with 80% residual capacity or 3000 cycles with 70% residual capacity. The prototypical vehicle was tested with the ECE-15 driving cycle (Fig. 10) to analyze the system behavior and monitor the battery performance considering the achieved temperature levels. The test bench machine was calibrated to evaluate the real vehicle behavior at the full load condition (3500 kg) during the defined driving cycle in accordance with the simulation analysis. Fig. 37 describes the comparison between the velocity profile of an ideal ECE-15 cycle and the profile acquired at the test bench. Fig. 38 shows the real current profile acquired at the dyno. During the test, the real current profile was monitored in addition to the internal battery temperature. A thermocouple probe was placed inside the battery pack to monitor the average air temperature. The real temperature data were compared to the simulation results to validate the proposed approach (Fig. 39).

4. Discussion The discussion on this research can be divided in three levels: the product design, Li-ion cell analysis, and test case with the final battery pack simulation. Concerning the product design, this paper shows how a structured knowledge-based methodology can support the rapid and flexible definition of an automotive Li-ion battery pack. The approach is focused on SMEs or large companies that produce customized batteries in small batches. The architecture of the proposed design platform can be employed in the design of automotive batteries because it aims to reduce time and cost during the early design phase. The reduction of time is possible because a workflow has already been defined by assigning tools and tasks for each step. Additionally, design rules and formulas are formalized inside a KBE kernel within a configuration tool (Pack Layout Configuration). The reuse of information, knowledge and past configurations enhance the reduction of time. The use

Fig. 36. The prototype of the battery in the electric freight vehicle.

Table 5 Retrofit details of the proposed freight vehicle. Electric motor

Electric 260 V AC 60 kW

Battery pack Battery lifecycle

96 NMC cells; 350 V; 150 Ah 2000 cycles at 80% residual capacity 3000 cycles at 70% residual capacity Carbon steel chassis 7 kW at 380 V Standard 6 h 80 km/h 100 km

Battery housing Battery charger Recharge time Max velocity Battery range

of an analytical model for the first thermal analysis allows the operative temperature profile of each Li-ion cell to be quickly estimated without using expensive and slower numerical tools. The validation of the analytical thermal analysis requires experimental tests. This activity should be repeated only if models of the Li-ion cells that are not presented in the database are analyzed. The second level of the paper is focused on the analysis of the Li-ion cells. The test case section describes three different types of cells with different chemistries, capacities and performance levels. The comparison between analytical temperature and real data shows how the proposed approach can calculate the average temperature of a Li-ion cell during charge and discharge by simple thermal characterization. When the parameters of the thermal model of a cell are defined, the analytical simulation can be extended to different driving cycles or other operative conditions. The estimation of the electrochemical heat source and the prediction of the temperature gives important feedback to the engineer during the early design phase, where most of the project and product costs are defined. The temperature’s trend is an important

Fig. 35. The battery temperature map (in the middle section) related to the ECE-15 simulation at the full load condition without cooling.

176

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

Fig. 37. Comparison between the ideal ECE-15 cycle (red line) and the real cycle (blue line). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 38. Trend of the battery current during the ECE cycle.

Fig. 39. Battery pack temperature; real red line, virtual blue line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

issue in Li-ion batteries because of its effect on performance, efficiency and safety. This paper proposes reports and datasheets on three types of Li-ion cells: 14-Ah LFP, 20-Ah NMC, and 150-Ah NMC. As described, the temperature increases with the size of the battery and the C-rate applied during the charge and discharge. However, large Li-ion cells have a higher thermal capacity and can generate a lower increase in temperature compared to the small cells for the same discharge current. In the proposed test case, the 150-Ah cell was chosen due to its higher thermal capacity and for the specific power per kilogram of material. The third level of the discussion is focused on the test case and the simulation of the battery pack. The proposed electric vehicle for freight delivery is an important example of urban sustainable transportation. The electric retrofit of petrol vehicles requires tools and methods for the design of customized solutions. As a demonstration, the described approach is used to support the battery

design for a prototypical vehicle. The design methodology enhances the energy optimization of the battery system because thermal evaluations and cooling simulations support the designer in the decision-making process. CFD simulations allow the correct cooling configuration to be defined in the early design phase. The most important result of the research is the combination of the analytical heat source estimation with the configuration of the battery pack and the battery cooling simulation. The same approach can be proposed without using a CFD tool, but a fluid-dynamic numerical solver is necessary to find the correct evaluation of the thermal distribution and dissipation. The gap between the estimated temperature and real values inside the prototypical battery pack is close and confirms the validity of the approach and the analytical model developed. As a final discussion topic, it is important to highlight that the suitable use of analytical and numerical tools with a knowledge-based approach can improve the energy performance of a battery system and reduce the time and cost related to the design phase and testing of prototypes. This paper proposes the use of tools for rapid product configuration and simulation. These tools are defined as ‘‘rapid” because they aim to support the designer in the reduction of the lead time, particularly in the context of SMEs. As cited before, the formalization of knowledge and past configuration allows new product configurations to be defined using a KBE approach. This approach is a novelty for applications concerning batteries. Past studies focus on battery modeling and characterization without analyzing the development of design tools that can support the designer in the definition of a product configuration. Specifically, the ability to implement additional know-how is a key factor for the context of SMEs. In this context, the battery design often considers the study of custom applications. The use of advanced CAE software tools presents limits such as the complexity to integrate knowledge and rules. Additionally, they are expensive for such customized applications and therefore are not affordable for an SME. The proposed tools and methods aim to fill this design gap. The highlighted test case was performed in collaboration with an Italian SME, which produces Li-ion battery packs in small batches. The authors estimate that the design of a battery pack can take from 1 to 6 months depending on the size and application. In the case of batteries for customized hybrid and electric vehicles, at least 1 month, more or less, is necessary for the study of a noncommercial product. However, these data are difficult to quantify and find in the scientific literature. The use of a configuration tool as described will contribute to reduce the time of the design phase because it provides a set of possible configurations. As a first design step, the engineer-user must define the main battery specifications and the possible battery pack layout inside the configuration form (as described in Section 2). This form proposes data and geometric

P. Cicconi et al. / Applied Energy 192 (2017) 159–177

dimensions, which can also be modified by the user. Developed algorithms can read all input data and elaborate upon it to generate an automated simplified 3D CAD model. This model is suitable for further CFD simulation using any type of commercial or noncommercial CFD tools. Between the configuration and the 3D analysis, the engineer-user can use an analytical tool to estimate the heat generated by a Li-ion cell under different current profiles. The integration of tools such as the configurator and analytical and CFD solvers in a same design platform can be seen as a research novelty in the field of Li-ion batteries. The design of a Li-ion battery pack is mainly the configuration and arrangement of a set of cells. Therefore, the theory of knowledge-based configuration is suitable for application for the battery design, and this paper aims to follow this approach. 5. Conclusion This research shows a knowledge-based approach to support the design of Li-ion battery packs considering the analytical calculation, testing data and numerical simulations. Additionally, a configuration tool implements the design rules and knowledge to support the designer in the definition of a new layout of battery packs. The battery design is focused on the thermal analysis related to the electrochemical heat generated by each Li-ion cell. The CFD simulations are used as a tool to evaluate the battery performance in terms of cooling during operation. The application test case is focused on the battery design of an electric freight vehicle, which is a retrofit of a lightweight gasoline pickup truck. The proposed test case considers the design of a 52 kW h battery with NMC Li-ion cells. The resulting simulations show an approximate 5% gap between the real temperature values and the simulated data. The analytical calculation, which is focused on the thermal analysis of one cell, also shows a reduced gap with the physical testing. This is because one of the main results of the study is the correct definition of the cell thermal model for the calculation of the electrochemical heat provided by each Li-ion cell. The real temperature profiles were monitored by an IR camera for the study of one cell and by a thermocouple probe for the temperature inside the battery pack. The calculation of the electrochemical heat is based on the current profile related to the operative condition of the ECE-15 driving cycle. The related current profile was simulated by implementing all force applied to the freight vehicle and using data from a real driving cycle. Another important result of this research is the definition of a configuration tool to support the definition of the battery layout to maximize the cooling effect using design rules and formulas. This tool was used in the described test case to generate the simplified 3D model that was then used in the CFD simulations. References [1] Wandud Z. Diesel demand in the road freight sector in the UK: estimates for different vehicle types. Appl Energy 2016:849–57. [2] Zeng Y, Tan X, Gu B, Wang Y, Xu B. Greenhouse gas emissions of motor vehicles in Chinese cities and the implication for China’s mitigation targets. Appl Energy 2016 [in Press]. [3] Tente H, Gomes P, Ferreira F, Amorim JH, Cascão P, Miranda AI, et al. Evaluating the efficiency of diesel particulate filters in high-duty vehicles: field operational testing in Portugal. Atmos Environ 2011;45(16):2623–9. [4] Jung S, Lim J, Kwon S, Jeon S, Kim1 Jeongsoo, Lee J, et al. Characterization of particulate matter from diesel passenger cars tested on chassis dynamometers. J Environ Sci 2016 [in Press]. [5] Diesel vehicle emission and death rates in Tokyo. Japan: A natural experiment. Sci Total Environ 2011;409(19):3620–7. [6] Bishop JDK, Martin NPD, Boies AM. Quantifying the role of vehicle size, powertrain technology, activity and consumer behavior on new UK passenger vehicle fleet energy use and emissions under different policy objectives. Appl Energy 2016;180:196–212. [7] Riesz J, Sotiriadis C, Ambach D, Donovan S. Quantifying the costs of a rapid transition to electric vehicles. Appl Energy 2016;180:287–300.

177

[8] Miranda AI, Ferreira J, Silveira C, Relvas H, Duque L, Roebeling P, et al. A costefficiency and health benefit approach to improve urban air quality. Sci Total Environ 2016;569:342–51. [9] Casals LC, Martinez-Laserna E, García BA, Nieto N. Sustainability analysis of the electric vehicle use in Europe for CO2 emissions reduction. J Clean Prod 2016;127:425–37. [10] Batista T, Freire F, Silva CM. Vehicle environmental rating methodologies: Overview and application to light-duty vehicles. Renew Sustain Energy Rev 2015;45:192–206. [11] Noori M, Zhao Y, Onat NC, Gardner S, Tatari O. Light-duty electric vehicles to improve the integrity of the electricity grid through vehicle to-grid technology: analysis of regional net revenue and emissions savings. Appl Energy 2016;168:146–58. [12] Feng W, Figliozzi M. An economic and technological analysis of the key factors affecting the competitiveness of electric commercial vehicles: a case study from the USA market. Transp Res Part C 2013;26:135–45. [13] Hu J, Morais H, Sousa T, Lind M. Electric vehicle fleet management in smart grids: a review of services, optimization and control aspects. Renew Sustain Energy Rev 2016;56:1207–26. [14] Propfe B, Kreyenberg D, Wind J, Schmid S. Market penetration analysis of electric vehicles in the German passenger car market towards 2030. Int J Hydrogen Energy 2013;38(13):5201–8. [15] Bohnsack R, Pinkse J, Kolk A. Business models for sustainable technologies: exploring business model evolution in the case of electric vehicles. Res Policy 2014;43:284–300. [16] Dijk M, Wells P, Will RK. The momentum of the electric car last? Testing an hypothesis on disruptive innovation. Technol Forecast Soc Change 2016;105:77–88. [17] Tairan Y, Naixing Y, Xiongwen Z. Investigation of the thermal performance of axial-flow air cooling for the lithium-ion battery pack. Int J Therm Sci 2016;108:132–44. [18] Cicconi P, Germani M, Landi D, Mengarelli M. Virtual prototyping approach to evaluate the thermal management of Li-Ion batteries. In: 2014 IEEE Vehicle Power and Propulsion Conference. [19] Golembiewski B, Stein NV, Sick N, Wiemhöfer HD. Identifying trends in battery technologies with regard to electric mobility: evidence from patenting activities along and across the battery value chain. J Clean Prod 2015;87:800–10. [20] Cicconi P, Landi D, Morbidoni A, Germani M. Feasibility analysis of second life applications for Li-ion cells used in electric powertrain using environmental indicators. In: Proceeding ENERGYCON. [21] Jiao N, Evans S. Business models for sustainability: the case of second-life electric vehicle batteries. 13th global conference on sustainable manufacturing – Decoupling growth from resource use, procedia CIRP, vol. 40. p. 250–5. [22] Ji L, Zheng H, Ismach A, Tan Z, Xun S, Lin E, et al. Graphene/Si multilayer structure anodes for advanced half and full lithium-ion cells. Nano Energy 2012;1:164–71. [23] Guo J, Li Z, Pecht M. A Bayesian approach for Li-Ion battery capacity fade modeling and cycles to failure prognostics. J Power Sources 2015;281:173–84. [24] Landi D, Cicconi P, Germani M. A methodological approach for supporting the thermal design of Li-ion battery for customized electric vehicles. In: 2014 ASME international mechanical engineering congress and exposition, proceedings (IMECE). [25] Guptaa S, Palsule-Desaic OD. Sustainable supply chain management: review and research opportunities. IIMB Manage Rev 2011;23(4):234–45. [26] Nelson PA, Ahmed S, Gallagher KG, Dees DW. Cost savings for manufacturing lithium batteries in a flexible plant. J Power Sources 2015;283:506–16. [27] Rao Z et al. Simulate and experimental research on air-cooled thermal energy management of lithium-ion power battery. In: Proceedings of 25th international battery, hybrid and fuel cell electric vehicles symposium & exposition. [28] Mahamud R, Park C. Reciprocating air flow for Li-ion battery thermal management to improve temperature uniformity. J Power Sources 2011;196:5685–96. [29] Feng X, Weng C, Ouyang M, Sun J. Online internal short circuit detection for a large format lithium ion battery. Appl Energy 2016;161:168–80. [30] Rao Z, Wang S. A review of power battery thermal energy management. Renew Sustain Energy Rev 2011;15:4554–71. [31] Classification and Definition of Vehicles, ECE/TRANS/WP.29/78/Rev.4. [32] European Parliament and Council. ‘‘Council Directive 70/220/EEC of 20 March 1970 on the approximation of the laws of the Member States relating to measures to be taken against air pollution by gases from positive-ignition engines of motor vehicles”. 1970. [33] Bernardi D, Pawlikowski E. A general energy balance for battery system. J Electrochem Soc 1985;132:5–12. [34] Bandhauer TM, Garimella S. (n.d.). A critical review of thermal issues in lithium-ion batteries. J Electrochem Soc 158(3). pp. R1–R25. [35] Newman LR. Heat-generation rate and general energy balance for insertion battery systems. J Electrochem Soc 1997;144:2697–704. [36] Chen SC. Thermal analysis of lithium-ion batteries. J Power Sources 2005;140:111–24. [37] K. Reif (Ed.), Brakes, Brake control and drivers assistance system function, regulation and components, Basic principles of vehicle dynamics, 2014, Springer Fachmedien Wiesbaden, 275, pp. 12–27.