Method development for testing propulsion batteries at a workshop

IN

DEGREE PROJECT VEHICLE ENGINEERING, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020,

Method development for testing

propulsion batteries at a workshop

Parameter identification through experiments and

investigation of challenges with workshop

implementation

KIM STRINNHOLM

KTH ROYAL INSTITUTE OF TECHNOLOGY SCHOOL OF ENGINEERING SCIENCES

Method development for testing propulsion

batteries at a workshop

Parameter identification through experiments and investigation of challenges with workshop implementation

Kim Strinnholm

Master of Science in Engineering

Master programme in Vehicle Engineering KTH Royal Institute of Technology

Supervisor at (Scania CV AB): Karl-Johan Jeansson Supervisor at KTH: Prof. Mikael Nybacka

Examiner at KTH: Prof. Mikael Nybacka Date of presentation: 15/12-2020

TRITA-SCI-GRU 2020374

KTH Royal Institute of Technology School of Engineering Sciences

KTH SCI SE-100 44 Stockholm, Sweden URL: http://www.kth.se/sci

Abstract

The electrification within the automotive industry goes faster than ever, which drives an increased demand for more knowledge about batteries. Vehicle manufacturers should be able to tell how long the batteries will last and have a service program for electrified vehicles, just as there is for traditional, fuel-driven ones. Scania is in the process of developing new service methods for their hybrids and fully electrified vehicles where this thesis has been a part of this development by investigating the possibilities of having a workshop test to measure the capacity of the propulsion batteries.

During the thesis, essential parameters for cycling the batteries and measure the capacity with high accuracy have been identified and investigated by conducting lab tests. In parallel to defining the properties of a successful capacity measurement, the implementation of such a measurement at a workshop has been studied alongside a brief discussion about scheduling strategies. Conducting a capacity measurement in a workshop environment introduce new challenges, and the critical question arises, how long can the capacity measurement take?

It is identified that the state of charge window size, the temperature, and the relaxation time are essential parameters to control. From the experimental part of the thesis, it can be concluded that the start temperature should lay in the range of 15-25^◦C with a relaxation time of 5-10 minutes providing a satisfying accuracy. A SOC window size of 20-80 % seems to be the most optimal balance between time spent and accuracy in the measurement. Furthermore, it is identified that the workshop’s equipment is heavily influencing the time it takes to conduct a test. It is concluded that it is necessary to be able to charge and discharge the batteries.

Keywords

Master thesis, batteries, lithium batteries, test methods for batteries, test development for workshop, capacity measurement, state of health, tracking SOH

Sammanfattning

Elektrifieringen av fordons industrin går snabbare än någonsin, vilket driver en högre efterfrågan på mer kunskap om batterier. Fordons tillverkare ska kunna redogöra för hur länge batterierna kan användas och ha ett service program för elektrifierade fordon, likt det som redan finns för traditionella, bränsledrivna fordon. Scania håller på att utveckla nya service metoder för sina hybrider och elektriska fordon där detta examensarbete har varit en del av denna utveckling genom att undersöka möjligheterna kring en verkstadsmetod för att mäta kapacitetn hos framdrivnings batterier.

Under examensarbetet har väsentliga parametrar för cykling av batterier och mätning av kapacitet med hög noggrannhet identifierats och undersökts med laboratorietester.

Parallellt med arbetet för att definiera egenskaperna hos en precis kapacitets mätning har implementationen av en sådan mätning i en verkstad studerats tillsammans med en kort diskussion om strategier för schemaläggning av dessa tester. Det introducerar nya utmaningar att utföra kapacitets mätningen i en verkstad och den viktiga frågan uppstår, hur lång tid tar en sådan kapacitets mätning?

Det har identifieras att SOC fönster storleken, temperaturen och relaxeringstiden är essentiella parameterar att kontrollera. Slutsatserna är att temperaturen bör ligga i intervallet 15-25 ^◦C med en relaxeringstid på 5-10 minuter ger en tillfredställande noggrannhet. Ett SOC fönster motsvarande 20-80 % är förefaller vara den mest optimala i avvägningen mellan tidsåtgång och precision. Vidare kan den tillgängliga utrustningen på verkstaden pekas ut som en starkt påverkande faktor till tiden det tar att utföra ett sådant verkstadstest.

Nyckelord

Examensarbete, batterier, lithium batterier, test metoder för batterier, utveckling av tester för verstäder, kapacitets mätning, följa SOH

Acknowledgements

I would like to thank my examiner Mikael Nybacka, my supervisor at Scania Karl- Johan Jeansson and the battery testing department at Scania for help and support throughout the thesis. I would like to say a special thanks to Nicklas Holmström who is working in the battery lab at Scania for valuable support with tests and discussions throughout the thesis and for contributing with a lot of insight and knowledge.

List of tables

4.5.1 Power levels . . . 36

4.5.2 Test times for one battery pack . . . 36

4.5.3 Test times for three battery packs . . . 36

4.5.4 Test times for eight battery packs . . . 37

4.5.5 Total test time when comparing discharging speeds at different start SOC. 37 4.5.6 Total test time when comparing charging speed. . . 38

4.5.7 Test times when varying SOC window size . . . 38

5.0.1 The reference capacity for the tested battery packs . . . 41

5.2.1 Capacity comparison for different temperatures . . . 46

5.2.2 Capacity comparison from degraded battery, varied temperature . . . . 46

5.3.1 Capacity comparison for different SOC windows . . . 47

5.3.2 Capacity comparison summarised . . . 48

5.3.3 Capacity comparison between packs and SOC windows . . . 49

5.3.4 Capacity comparison from degraded battery, varied SOC window . . . 49

List of figures

2.2.1 Degradation of lithium-ion batteries . . . 11

2.3.1 Energy density versus OCV . . . 13

2.4.1 Dual polarisation model . . . 13

2.4.2Voltage response during pulse . . . 14

2.5.1 OCV curve example . . . 15

3.3.1 Spread in OCV curve . . . 22

3.3.2 SOC difference in OCV curve . . . 23

3.4.1 Voltage during long relaxation . . . 24

4.6.1 Temperature increase in B14 . . . 40

5.1.1 Relaxation error vs time for min and max voltge . . . 43

5.1.2 Relaxation error comparing SOC windows . . . 44

5.1.3 Relaxation error comparing temperatures . . . 45

5.1.4 Relaxation error for degraded battery pack . . . 46

5.4.1 Voltage drop during relaxation . . . 50

5.4.2 Voltage drop during relaxation with CV step . . . 50

Acronyms

BEV Battery Electric Vehicle BMU Battery Management Unit BOL Beginning Of Life

ECM Equivalent Circuit Model EOL End Of Life

OCV Open Circuit Voltage SEI Solid Electrolyte Interface SOC State Of Charge

SOH State Of Health SOP State Of Power

Table of content

1 Introduction

1.1 Background . . . 1

1.2 Purpose . . . 2

1.3 Goals . . . 4

1.4 Boundaries and limitations . . . 5

1.5 Outline . . . 6

2 Theoretical background

2.2 Degradation of the lithium cell . . . 8

2.2.1 Solid Electrolyte Interface formation . . . 9

2.2.2 Lithium plating . . . 9

2.2.3 Internal stress . . . 10

2.2.4 Linear and non-linear phase . . . 10

2.3 Chemical processes in brief . . . 11

2.3.1 Over voltage . . . 11

2.3.2 Temperature dependency . . . 12

2.3.3 OCV-SOC relationship . . . 12

2.4 Modelling of a lithium battery . . . 12

2.5 OCV curve . . . 15

3 Experimental setup

3.1.1 Temperature . . . 17

3.1.2 Relaxation time . . . 17

3.1.3 SOC window size . . . 18

3.1.4 Test matrix . . . 18

3.2 Capacity measurement . . . 19

TABLE OF CONTENT

3.3 Open circuit voltage curve. . . 20

3.3.1 Definition of OCV curve . . . 20

3.3.2 Reading the open circuit voltage curve . . . 21

3.4 Test procedure . . . 22

3.5 Monitored data for performance evaluation . . . 25

4 Workshop strategy

4.2 Equipment . . . 28

4.2.1 External factors . . . 28

4.2.2 Charger . . . 29

4.2.3 Discharging. . . 30

4.3 Scheduling strategy . . . 31

4.3.1 Trigger the test . . . 31

4.3.2 Scheduling . . . 32

4.4 Vehicle configurations . . . 32

4.4.1 One battery pack. . . 33

4.4.2 Multiple batteries . . . 33

4.5 Time duration . . . 34

4.5.1 Theoretical time calculation . . . 34

4.5.2 Comparing test times . . . 35

4.6 Parameters varying in workshop environment . . . 38

4.6.1 Start SOC . . . 38

4.6.2 Start temperature . . . 39

5 Result

5.2 Temperature . . . 45

5.3 SOC window size . . . 47

5.4 Constant voltage step . . . 49

6 Discussion and conclusions

6.2 Conclusions . . . 55

6.3 Future work . . . 56

TABLE OF CONTENT

References

Chapter 1

Introduction

The introduction describes the background of this thesis along with the purpose and the goals. Further, it describes the defined, and encounted limitations during the thesis and lastly the outline of the report is displayed.

1.1 Background

The automotive industry has grown from simple car-like vehicles to very complex hybrids, electrified powertrains and vehicles driving autonomously [1]. One of the main focuses today within the automotive industry is electrification with battery electric vehicles, which is seen as an essential step towards reducing the global environmental impact from the transportation sector [2].

The development of electrified solutions within the automotive industry has rapidly increased since Toyota introduced its first mass-produced hybrid in 2000, with a range of 2 km on electricity up to 50 km/h [3], to today’s Tesla Roadster with a top-speed of over 400 km/h and a range of 1000 km [4].

The demands from society are increasing rapidly and exceed the phase of the development. Some of the most important areas from a consumer perspective right now are range, charge times and lifetime of the battery [5].

There are many areas within the battery development where there are still uncertainties: the lifetime of the battery being one of them and how it is affected by factors like faster charging times, usage-strategy, environmental conditions etc.

CHAPTER 1. INTRODUCTION

One important topic related to batteries’ lifetime and degradation is battery diagnostics and how to evaluate their performance and health. State Of Health (SOH) is a way to describe the battery’s health, which is a subjective measurement and lacks a common standard. The parameters used to evaluate the SOH dictate the SOH estimation quality, and due to the lack of a common standard, this can vary between different companies. A complete SOH estimation takes charge acceptance, internal resistance, capacity, power delivery and more into consideration but usually it refers to the division between current capacity and rated capacity. The latter provides a simple definition, assigning a number to the degradation and how much it affects the vehicle’s range [6].

The SOH algorithm is usually running constantly during vehicle usage to collect data for estimating SOH onboard the vehicle. The SOH estimation fuses data from the different monitored parameters, which is measured under various conditions and loads, to estimate the capacity. Due to the varying conditions at which it operates and the fact that the used SOC window is narrow, the algorithm suffers from errors in the capacity estimation.

By defining the SOH as the battery’s remaining capacity, as described previously, the state of health can be calculated by doing a capacity measurement of the battery. This is the chosen method to move forward with for this thesis, in collaboration with the truck manufacturer, Scania.

1.2 Purpose

Scania is in the process of expanding the maintenance capabilities and workshop support for the electrical drivetrains, both considering plug-in hybrids and a future market of fully electric vehicles. There is currently a workshop method for testing hybrids, but it is shown not to be optimal and does not fulfill the future need of testing batteries in different configurations. This workshop method is time-consuming, with more than 3 hours of testing and the test’s accuracy has been varying. With this thesis’s help, Scania wants to take the first steps towards a new workshop method for testing plug-in hybrids and fully electric vehicles with a faster and more accurate test.

This thesis is expected to deliver the required material and background research to create a workshop method for how to test the capacity of the propulsion battery packs

CHAPTER 1. INTRODUCTION

on a truck. Today Scania already have a State Of Health estimator onboard that estimates the capacity of the pack. However, it is crucial to have a method for accurately determining the capacity when making critical decisions due to it being an estimation.

A workshop method for measuring the capacity is thereby essential to increase the confidence when making decisions, such as if the battery pack has reached End Of Life (EOL), and it is time to change the battery. It is a crucial decision to be confident in since the battery packs are costly to change for the customer and might even cause the customer to deem the vehicle obsolete.

Since the SOH estimation is not providing high enough confidence for being used in the final verdict, it is the workshop method’s main purpose to support such a decision.

The performance of the onboard estimator will vary depending on usage of the vehicle, loads, where in the lifetime the vehicle is and so forth. If a scenario is pictured where the estimation gives an error of 5% and the battery pack is changed too early, the distance of which it is equivalent to, it becomes evident how crucial the decision is.

Below the pictured scenario is given in bullet-points.

• At this particular moment the error is 5% in the estimation

• Lithium batteries for automotive purposes are usually considered EOL at 80 % SOH.

• Cyclable energy in one pack could be 100-250 MWh, counting with 250 MWh

• Consumption for heavy vehicles, between 1-3 kWh/km, counting with 1.5kWh/km

• Vehicle with 10 battery packs

If the estimation gives a SOH value of 80 percent which means it is time to change the battery, instead of the actual 85 percent in this case, it corresponds to 62.5 MWh if its counted with a total lifetime of 250 MWh. With the above mentioned example consumption and number of batteries, the drivable distance is almost 42’000 km. The number gives a perspective on the importance of having high accuracy when deciding if the energy storage have reached EOL or not. Increasing the drivable range from the energy storage is valuable for Scania’s brand as a premium truck manufacturer, valuable for the driver and have a significant impact on the environmental footprint of

CHAPTER 1. INTRODUCTION

the battery by being able to use it for a longer period of time.

It has also been identified that the workshop test for the battery capacity can be used as a supporting tool in the bigger picture of battery maintenance. From the Battery Management Unit (BMU) development team, it has been raised that the measurement can be used to calibrate the capacity value in the BMU, which is normally estimated.

This calibration could increase the accuracy in systems like State Of Charge (SOC) algoritm, State Of Power (SOP), and the balancing system.

From a workshop perspective, the test can also be used for concluding when the next test of the propulsion batteries should be scheduled and could be a tool for troubleshooting when problems arise with the electrical system and the battery.

1.3 Goals

The thesis consists of two distinct parts, where one is an experimental part with a parameter study, and the other is more engineering-related, aiming to discuss and formulate a strategy for testing the vehicles at a workshop. The later one should deal with the challenges of implementing the measurement at a workshop and consider different vehicle configuration and scheduling strategies.

Overall, the thesis’s goal can be summarized as investigating the possibilities of creating a workshop test for evaluating if the propulsion batteries onboard a vehicle have reached its EOL. It can further be broken down to more specific goals, as stated below.

1. Optimising the time spent on a capacity measurement while keeping the error as low as possible (However never higher than 5 %, considering the example given in section 1.2)

2. Evaluate parameters of interest to control 3. Discuss scheduling strategies for testing

As mentioned in the thesis’s purpose, the time spent on the test is of importance to reduce since the cost of a workshop visit is increasing with an increased testing time and hence the first bullet point is stated. The second goal has to do with the parameter study and investigate appropriate boundaries for the parameters of interest to control, keeping the accuracy high and time spent low. The third goal means that it is important

CHAPTER 1. INTRODUCTION

to consider the bigger picture of the challenges with implementation in a workshop environment and when to conduct the test. It is not part of the thesis’s scope to present a final strategy but rather present suggested strategies to continue to work with in the future.

1.4 Boundaries and limitations

The thesis topic is broad, where the focus is to consider all the aspects of implementing a workshop test rather than an in-dept optimisation of a single parameter. Hence, some of the topics to cover in the thesis will not be explored in dept to have time to cover all relevant areas to consider for getting the complete picture of implementing a capacity measurement at a workshop. The time for lab testing is limited due to the high occupancy of required equipment in the lab. Hence, the evaluation of parameters from lab tests should be seen as a proof of concept rather than a final conclusion with high statistical significance. The test time in the lab was limited to the occasions when there were stops in the regular activities or between them. The test objects were also limited to the ones available in the lab at the time for the tests.

It has been chosen to only consider vehicles with DC charging, meaning that the vehicle does not carry its own AC/DC converter. Since Scania already have a workshop method for hybrids, and the trend is going towards larger battery packs and DC charging, it has been chosen to exclude hybrids and only focus on plug-in vehicles and fully electric vehicles.

It was limited to what extent the degradation of batteries could be considered in the thesis, since it was not possible to get testing time on batteries close to end of life.

With regards to the degradation effect on the OCV curve, Scania already work on compensating for that and thus, it was assumed that there was always access to an accurate OCV curve.

For scheduling strategies, several parties need to be involved in the discussion and it needs to be evaluated from several perspectives. Efforts were made to set up such a team for discussions, but the individuals approached at Scania were highly occupied, which led to only brief discussions within the topic.

When comparing results it has been chosen to use made-up number for capacity of the cells, which is said to be 56 Ah for the new generation cells and 33 Ah for the old

CHAPTER 1. INTRODUCTION generation cells.

1.5 Outline

The thesis includes the following chapters with a description of its purpose:

• Theoretical background - Aims to give the reader an understanding of batteries as electrochemical devices and what the challenges are. It also covers degradation mechanisms to give the background for having a capacity measurement and why it is required to have a testing strategy.

• Experimental setup - Explains the parameters of interest to evaluate with lab tests, how the tests have been conducted and motivating some of the choices made when setting up the tests.

• Workshop strategy - This chapter focuses on the implementation of the test and the challenges from a workshop perspective. Here it is elaborated on the fact that the test is dependant on other factors than just the test parameters investigated in the lab. Further it is also discussed how a test strategy can look like.

• Results - The results from the lab tests are presented.

• Discussion and conclusions - The discussion, conclusion and suggestions for future work within the area are presented in this chapter.

Chapter 2

Theoretical background

In this chapter the theoretical background will be presented for the reader to get a brief insight into the theory behind the work and choices to be done during the thesis.

The theory will be necessary for understanding a lithium battery’s characteristics and conclude what parameters will be important when constructing a capacity test aiming to be used in a workshop environment.

It is worth noting that for the remaining part of this thesis, ”battery” will refer to a single cell or several cells connected to form a battery. A battery pack is explicitly referring to several cells forming a battery pack, including all electronics and enclosure, ready for mounting in the vehicle. Battery packs in plural are several packs connected in parallel to form a system containing more energy, typically found on plug-in hybrids and fully electric vehicles. It is also worth noticing that onboard estimators and online estimation will refer to the systems onboard monitoring the battery status and calculate battery parameters such as SOC. It also means that the system continuously collects data during operation and computes these values in real-time during operation.

2.1 Batteries as electrochemical devices

An electrochemical device is a common name for a device with the goal of creating a chemical reaction that generates electrical energy or the opposite- using electrical energy to cause a chemical reaction. For this thesis’s scope, the battery is the electrochemical device to look closer at.

The electrochemical devices typically consist of two electrodes, an anode and a cathode,

CHAPTER 2. THEORETICAL BACKGROUND

submerged in an electrolyte. The chemical reaction that generates electrical energy or is caused by electrical energy is happening at the electrode/electrolyte interface. [7], [8]

Batteries are used within many different areas and they come in a variety of shapes, sizes, and types of chemistry. The lithium-based batteries are prevalent today, and the lithium-ion is one of the leading technologies. The basic layout of a lithium-ion battery is a positive and negative electrode with a liquid electrolyte containing a separator made from a micro-porous polyethylene and/or polypropylene sheet with the purpose to let lithium ions pass it but still act as an isolator to protect the electrodes from short- circuiting [7].

A substantial amount of the gathered research and development within batteries for the automotive industry is focusing on increasing the range by carrying more energy while keeping the weight down and increasing the battery’s lifetime by reducing and altering the degradation mechanisms. The in-dept approach for achieving these goals is not in focus for this thesis, but if the reader wishes to read more about it, more information can be found in [8].

However, the degradation of batteries is important for the scope of this thesis to follow the discussions about strategies for scheduling workshop tests. Many identified degradation processes are happening inside a cell, derived from mainly two branches:

calendar aging and cycle-induced degradation. The calendar aging refers to all processes/reactions happening inside the cell during periods of not being used, which has been studied in-dept by [9]. The other important branch is degradation caused by using the battery, cycling it by taking out and putting energy back in. To cover the full degradation process in a battery, defining more categories and subcategories are necessary, but these are among the main contributors and the most prominent categories. Next, in section 2.2, a brief overview of the main degradation processes and what performance loss it contributes to is presented.

2.2 Degradation of the lithium cell

In this section, some of the most critical degradation factors of the lithium cell will be covered. It is important to establish why it is vital to monitor battery health and what parameters might be of interest when trying to capture this degradation with a

CHAPTER 2. THEORETICAL BACKGROUND workshop test.

2.2.1 Solid Electrolyte Interface formation

One important part of the lithium battery is the formation of a Solid Electrolyte Interface (SEI). The name comes from the fact that it is a solid electrolyte barrier formed between the electrolyte and electrode, which lets lithium ions pass through but blocks electrons and the liquid electrolyte [10]. The anode’s SEI layer is both an essential part of a lithium-ion battery and part of the degradation problem. The layer contributes to the cell’s chemical and electrochemical stability by protecting the electrolyte from reacting with the electrode, which is how the layer is created in the first place. [11], [10], [7]

During the cells first charge after production, the electrolyte reacts with the electrode, consuming lithium to create this passivisation layer, which is essential and part of the problem since the reaction consumes cyclable lithium [12]. Ideally, this SEI layer created in the Beginning Of Life (BOL) should protect from further reaction but if the layer becomes unstable, side reactions will slowly build up a thicker layer with time [7].

This consumption of lithium is one of the main contributors to the loss of capacity and thereby reduced range in a vehicle [12], [13].

2.2.2 Lithium plating

Lithium plating is another major degradation contributor and possible safety hazard of lithium cells [8]. It is related to critical charging conditions such as high charging rates, cold temperatures or overcharging. Furthermore, it is affected by cell design and manufacturing flaws [8]. Lithium plating severely reduces the performance of a cell and capacity by consuming lithium ions and electrolyte. Even more important than the range is that it introduces a safety hazard because the lithium plating can grow dendritically, which could cause a puncture of the separator, short-circuiting the cell [14], [8]. A short-circuit could potentially cause a thermal runaway in the battery pack, leading to a fire or explosion.

CHAPTER 2. THEORETICAL BACKGROUND

2.2.3 Internal stress

Another important degradation factor is internal stress in the cell, causing cracks in the electrode due to repeated intercalation and deintercalation of lithium ions. Cracks and separation of the electrode reduce the cyclable lithium ions and expose the electrode to the electrolyte, causing SEI growth, which further reduces the useful lithium and deteriorates the cell [15]. It becomes evident that this further reduces the capacity of the cell and cyclic performance.

2.2.4 Linear and non-linear phase

According to [13], degradation can be described as having a linear phase to begin with and shifting towards a non-linear phase closer to the EOL of the cell. Having a two stage degradation is contributed to by both SEI layer growth as well as lithium plating.

The non-linear phase is characterised by a distinct rise of the lithium plating. In the linear region, on the other hand, the degradation is dominated by SEI growth [13]. This is an important factor since it means the battery will gradually start to age faster, and when the cell is getting closer to EOL it is necessary to monitor the health continuously.

In Figure 2.2.1, it is shown how the characteristics of lithium batteries degradation could look like. The curve’s attributes will depend on many factors, such as chemistry, load spectrum, environmental condition, and more, but it is showing a general view of it.

In Figure 2.2.1, three distinct parts of the curve have been highlighted by drawing dotted lines. It is seen that there is a high derivative at the beginning correlated to a settling period of the cell. The next section represents the linear phase described above, and the third section shows that the degradation starts to accelerate again at a certain time. This knee point is usually referred to as the EOL point in the automotive industry since the range will drastically drop after this point. This part of non-linear degradation is also correlated to the rise of lithium plating, as described above, increasing the risk of an internal failure in the cell.

CHAPTER 2. THEORETICAL BACKGROUND

Figure 2.2.1: Typical characteristics for degradation of lithium-ion batteries.

2.3 Chemical processes in brief

2.3.1 Over voltage

During load, the measured voltage at the cell is never the same as the Open Circuit Voltage (OCV) voltage. During charging and discharging, there is an overvoltage and undervoltage respectively, applied to the OCV which represents the energy dissipated as heat. From now on, both scenarios will be referred to as overvoltage. The overvoltage occurs from electronic conduction, ionic mass transport, and the charge transfer phenomena in the cell. [16]

The contributions to the overvoltage can be derived from ohmic and non-ohmic effects.

The ohmic conduction, called ohmic drop, is a pure resistive effect. The ohmic drop comes from the ionic resistance in the electrolyte and the active mass’s electronic resistance. It also comes from the interface between the current collectors and the electrodes and the conductive tabs. [16]

The second part of the overvoltage, the non-ohmic part, is related to polarization losses, including the activation polarization and the concentration polarization. The activation polarization is the formation of non-reagent products at the interface between the electrode and electrolyte, related to the charge transfer process. The second loss is the concentration polarization, which is spatial variations of reagents

CHAPTER 2. THEORETICAL BACKGROUND

in the electrolyte or electrode, causing concentration gradients in the boundary layers.

[16], [17]

2.3.2 Temperature dependency

The cell’s temperature is highly connected to the chemical reactions and hence connected to the previous topic about overvoltage. In [18] it was experimentally shown that with lowered operational temperature, the polarization increased and the power density decreased. In [19], temperature effects on lithium batteries are reviewed, and it is explained that with reduced temperature the chemical reaction activity slows down as well as charge transfer velocity. It is further explained that this leads to a decreased ionic conductivity in the electrolyte.

2.3.3 OCV-SOC relationship

In [20] it is elaborated on the relationship between SOC and OCV, which is a crucial relationship. The author of the report also expresses the importance of acknowledging the changed characteristics of this relationship with degradation of the cell. It is essential to compensate for the change to keep the accuracy high for systems relying on the OCV curve. Furthermore, the author presents a generalised model to account for this.

In Figure 2.3.1, the SOC-OCV relationship can be seen on the left side and on the right side is the dQ/dV curve, which shows the energy content as a function of the OCV voltage. It can be seen that the energy content of the cell is far from evenly distributed on the OCV curve, which means that different parts of the SOC scale will contain different amounts of energy.

A more in-dept discussion about the SOC-OCV relationship and how it is changed with the SOH can be found in [20]. For the scope of this thesis though, the most important part to perceive is the distribution of energy along with the SOC window.

2.4 Modelling of a lithium battery

Modelling of lithium batteries is a key component for a successful implementation of a lithium-ion battery in any system today, to ensure safe operation and to be able to

CHAPTER 2. THEORETICAL BACKGROUND

Figure 2.3.1: OCV versus SOC, red line, and dQ/dV as a function of OCV, blue line [20].

display the status of the battery. Two essential factors for the user, especially in EV applications, are to display the SOC and how far it can be driven on the remaining energy in the battery. It is also a key factor for optimal utilisation of the battery.

There are three main categories for modeling lithium batteries, mathematical models, electrochemical models, and electrical equivalent circuit models described in more depth in [21]. It is chosen to discuss further the Equivalent Circuit Model (ECM) for the rest of this section since it is the most common automotive industry approach.

O.C.V.

R₀

C₁ R₁

C₂ R₂

V_out

Figure 2.4.1: Dual polarisation model with 2 RC branches.

As mentioned in [21], the equivalent circuit model has less complexity and it is possible to simplify it in order to find a balance between complexity and accuracy. A balance that is very important for automotive industry today since there are limitations on computational power. In Figure 2.4.1 an ECM is shown with two RC branches. The first resistor, R0, represents the pure ohmic resistance. R0 has a distinct phenomenon seen in Figure 2.4.2 where there is an instant voltage drop when applying the load and releasing it. It is related to the ohmic drop explained in subsection 2.3.1.

CHAPTER 2. THEORETICAL BACKGROUND

Figure 2.4.2: Discharge with a current pulse between time≈ 0.2 and 7 seconds.

The next, non-linear drop comes from the charge transfer resistance within a few hundred milliseconds, which is also described in subsection 2.3.1, modelled with the RC branches in the model. The gradual voltage drop after this is mainly due to the slow diffusion process, elaborated in more detail in [6], [22]. The reversed process can be seen when removing the load. First, the pure ohmic drop disappears, then the charge transfer resistance, the slow recovery after that is the slow diffusion process [22]. Adding several RC circuits will increase the model accuracy, but it will also highly increase the computational complexity.

Vout(t) = O.C.V − I × (R0+ R1× (1 − e^R1×C1−t ) + R2× (1 − e^R2×C2−t )) (2.1)

In Equation 2.1 the equation for the output voltage, seen in Figure 2.4.1, is presented.

here R₀is the pure ohmic internal resistance while R₁,R₂,C₁and C₂ are the equivalent polarisation resistance and capacitance. The current I is the load applied and t is the time since the load was applied.

The output voltage will be different from the OCV and it will be dependant on the internal resistances and the time constants τ = R× C of the cell. It becomes evident that these time constants, and many more, if it should be more accurately modelled, will affect the time it takes to reach equilibrium.

CHAPTER 2. THEORETICAL BACKGROUND

0 10 20 30 40 50 60 70 80 90 100

SOC [%]

3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4 4.1 4.2

Voltage [V]

OCV curve

Charge Discharge

Figure 2.5.1: Example of OCV curve from an older cell generation Scania have used.

2.5 OCV curve

An Open Circuit Voltage curve is often constructed by physical measurements to associate the charging and discharging to the SOC. The OCV curve is mainly used when modelling and used by systems onboard to monitor the battery pack status and calculate the remaining range.

There is no common standard for battery OCV measurements [23], and companies are defining their own test procedures with the different methods that exist, such as incremental cycling or low rate charge and discharge.

In figure Figure 2.5.1, a typical OCV curve can be seen and the hysteresis between the charging and discharging is directly noted. This phenomenon is a complex process that is dependant on the chemistry of the cell and related to the existence of several thermodynamic equilibrium potentials for the same SOC [23]. There are several explanations for the occurrence of hysteresis, and it can be studied in more detail in [23], [24], [8]. However, for this thesis’s scope, it is only necessary to acknowledge the difference since it will affect reading the SOC values.

Chapter 3

Experimental setup

This chapter aims to explain the parameters identified as important to control and monitor at a workshop test and how they have been evaluated using tests in Scania’s battery lab. Presented will also be the scaling of the capacity measurement and how to use the OCV curve for this.

3.1 Parameters

Many parameters can be controlled or monitored while doing a capacity measurement, and many factors affect the measurement to some extent. With the literature’s help presented in chapter 2, in collaboration with the battery lab at Scania, and considering the workshop environment explained in chapter 4, the most important parameters were selected. It was concluded that the temperature, SOC window, and relaxation time are important parameters to set up guidelines for and evaluate with lab tests.

The charging/discharging current is also important, but it has not been necessary to include as a parameter in the test matrix. The maximal allowed charging current, given from the cell responsible at Scania, is 1C (56A with the example cell) for now but might be possible to increase in the future. It is also limited by the available equipment at a workshop, described in section 4.2. From the internal performance tests done at Scania it can also be seen that going from C/3 to 3C, there is a decrease in measured capacity of roughly 2.3% and from C/3 to 1C it is 1.2% decrease. From these results, the difference is small compared to the influence of other parameters. The testing time with C/3 would give three times longer charge/discharge session.

CHAPTER 3. EXPERIMENTAL SETUP

In the same test report, the temperature dependency of the capacity measurement was evaluated in the same way, and it could be seen that from 50^◦C to 0^◦C, the capacity dropped with over 11%.

3.1.1 Temperature

In subsection 2.3.2, the temperature’s impact on the cell is explained briefly and it can be concluded that the temperature is of importance for a capacity test. The temperature is highly correlated to the electrochemical process inside the cell and will affect the diffusion rate for example. It gives a lower resistance and faster relaxation time at higher temperatures and the opposite for lower temperatures, higher resistance and longer relaxation times.

As described in subsection 4.6.2, it has to be considered that the start temperature of the battery pack can vary. Hence, it was important to evaluate if it is possible to recommend a temperature spectrum suitable for the test. If not, it would be very time consuming to adjust the battery pack’s temperature before conducting the test every time.

However, it is important to conduct the test within an optimal temperature range of around 25^◦C in the cells to ensure reliable and accurate results. It was concluded to evaluate the capacity measurement at 25^◦C and 15 ^◦C to see if it is possible to allow the test to be conducted already at 15 ^◦C. A test sequence at 0^◦C was also tested for comparison.

3.1.2 Relaxation time

The relaxation time is important since it is connected to the accuracy of the measurement and the time spent on the test. As discussed in section 2.3, the relaxation time is tied to the chemical processes in the cell. The relaxation time can be influenced by the temperature and the SOH of the cell.

Considering the proposed relaxation time and a basic scenario with one battery pack, the relaxation time would correspond to almost 25% of the total testing time which is highlighting the importance of reducing it while keeping the accuracy high.

CHAPTER 3. EXPERIMENTAL SETUP

3.1.3 SOC window size

The SOC window size will influence the capacity measurement since if not a 0- 100% SOC charge/discharge window is used, the energy measured will then be scaled with the window size used. Scaling the energy assumes the energy density is constant throughout the whole window, which is not the case, as described in subsection 2.3.3. Using part of the SOC window means the losses outside of the window is not being accouted for. In theory, the bigger window used the more accurate the capacity measurement would be, but it also increase the testing time since the charging/discharging procedure takes longer to conduct. Considering this, the SOC window needs to be evaluated as a parameter to find an optimal size, giving an acceptable accuracy while minimising the time spent.

3.1.4 Test matrix

The SOC windows chosen to test were 15-55%, 20-80%, 15-90%, and 40-80%. The SOC window of 40-80% was raised as compelling at a later stage in the project to compare with the 15-55 window since the size is the same, 40%, but they cover different parts of the full SOC window. The window of 20-80% was of high interest and the 15-90% was added later in the process since Scania mentioned that previous tests showed it would be a beneficial window size.

• SOC: 15-55%, 20-80%, 15-90% and 40-80%, tested at 25^◦C

• Temperature: 0^◦C, 15^◦C and 25^◦C, tested at 20-80%

• Relaxation: Evaluated from 0-30 minutes, measured for all varied parameters

• CCCV charging with CV step until C/6, tested at 40% and 25^◦C

The relaxation time was investigated from 0-30 minutes, where in section 3.4 it is elaborated on why it is enough. Limiting the total relaxation time while keeping the accuracy as high as possible, is a way of getting more tests done since the testing time was a limiting factor in the thesis.

A possible strategy would be to test only one battery pack in a multi-pack configuration, as elaborated on in subsection 4.4.2, but then the pack need to be matched with the rest of the battery packs again, as described in subsection 4.4.2. The limit of a 10V difference between the packs when connecting them again is critical to follow to ensure

CHAPTER 3. EXPERIMENTAL SETUP

a safe re-connection and not risking to break any components.

A constant voltage step after the constant current charging is necessary to reduce the voltage drop during relaxation and land closer to the target voltage. It would have been beneficial for the overall workshop test time to find at what current the CV step can be turned off and still be within the 10V limit. However, there was not enough time in the battery lab to evaluate this. Instead, one CCCV step was done where the charging were turned off at C/6A. The choice of doing it at 40% SOC was based on that it could be a reasonable SOC voltage at which the vehicle would come into the workshop and it fitted into the already defined test matrix, making the tests as time-efficient as possible.

3.2 Capacity measurement

The capacity is a measure of how much energy a battery can hold, which is measured by integrating the current over time. The test at a workshop can not do a 0-100%

SOC sweep since there are risks associated with using the full window and it is time- consuming, which means that the measurement needs to be scaled according to the size of the window used.

Q_tot =

∫t2 t1 Idt

SOC window = Q_measured

SOC(OCV 2)− SOC(OCV 1) (3.1) In Equation 3.1, it can be seen that the integrated current from t1 to t2 gives the measured energy and since the used window size will be less than 100% it needs to be scaled by dividing with the window size. It is important to see that this assumes the behavior to be linear outside of the window used, which is not the case, as explained in the theoretical background. It introduces an error to the measurement, and hence it can be concluded that the bigger the SOC window is, the higher accuracy the measurement will have in theory.

The SOC levels will be given from the OCV curve, covered in dept later in this chapter, by measuring the OCV before and after charging or discharging, called OCV1 and OCV2, respectively. The voltage measurement before charge/discharge and after the battery has relaxed will be referred to as OCV1 and OCV2. When measuring the OCV of the battery and divide with the number of cells, it will give an average value for the battery pack, but in reality it will be small variations from cell to cell

CHAPTER 3. EXPERIMENTAL SETUP

due to imbalance of the pack. Having an imbalance is unavoidable and comes from manufacturing tolerances, small differences in the internal structure of each cell, and local differences in the battery pack. These differences could be small resistance differences or manufacturing flaws on bus bars or the battery’s connections. It can also be due to temperature variations in the pack causing uneven operation and degradation.

There is no standard in place defining how to read the voltage to get the corresponding SOC. The final result will differ depending on if the average voltage, minimum, maximum, or something else is used. However, to use the same definition as the BMU onboard of the vehicle, the pack capacity was defined as the limiting cell’s capacity in the pack. This is done by calculating the capacity of all cells and then finding the cell with the lowest measured capacity.

3.3 Open circuit voltage curve

Since there is no common standard for defining the OCV curve, the used method for generating the curve is presented briefly in this section. Further, it is also stated how the OCV curve is used.

3.3.1 Definition of OCV curve

As was described in the theoretical background, the OCV curve is a way to couple the cell characteristics with the SOC and it can be generated in different ways, but it is a presentation of voltage versus SOC. Since the capacity measurement should be comparable with the estimation done in the BMU it was chosen to use the same definition of the OCV curve as the BMU is using. The used OCV curve is generated with tests done at Scania according to their specifications. The whole test is done at 25

◦C, and it is started by doing three standard cycles. The standard cycle is defined as a standard discharge followed by a standard charge. The standard discharge is defined as 1C discharging till lower voltage limit, and the standard charge is defined as 1C constant current (CC) charge to 100% SOC.

The OCV curves are then generated by doing a charge and discharge procedure, with the main requirements summarised as follow:

• C/3 current in both directions

CHAPTER 3. EXPERIMENTAL SETUP

• 100-10% SOC: 5% steps

• 10-0% SOC: 1% steps

• Requirement for starting each new step was 30 min relaxation and a voltage difference of < 2mV over 10 minutes

• Interpolation between the data points

If the reader wishes to go deeper into the definition of OCV curves and different methods for generating them, it can be found in [25].

3.3.2 Reading the open circuit voltage curve

As described in the theoretical background, it will be two different curves when charging and discharging to get the OCV curve since it is not an ideal cell. Different ways of reading the OCV curve have been explored in the thesis to identify how it is affecting the result of the capacity measurement. Sometimes, the charge and discharge curve’s average value is used out of simplicity, which is also one of the tested ways of using it.

To compare what method gives the best accuracy, explained in section 3.5, the following methods for reading was used.

• Average OCV curve when reading OCV1 and OCV2

• For charging, only use charge OCV curve when reading OCV1 and OCV2

• For charging, use discharge OCV curve when reading OCV1 and the charge OCV curve for OCV2

This is done in the same way for the discharging sequence with only the discharging curve for both, and in the third version it is first the charge OCV curve and then the discharge OCV curve. The third option can be considered the most proper way of reading the OCV curve since in most tests, the battery is being discharged to the start value for the test.

The reason for it being highly important how to read the OCV can be seen in Figure 3.3.1.

CHAPTER 3. EXPERIMENTAL SETUP

Figure 3.3.1: Zoomed OCV curve showing the difference between charge, discharge and average curve.

A given cell voltage reading on the y-axis can give a spread in SOC, as in this case at 3.545 volt a difference of over 3 percent can be observed.

In Figure 3.3.2, the difference in SOC for every voltage reading is presented. The orange curve shows the difference in percent by calculating the difference in the SOC reading between charge and discharge, for every given voltage value. Here it can be seen that the discrepancy can be as high as 4.3 % and it is high in the regions where the first measurement point is, which is at 15-20 %.

3.4 Test procedure

It was necessary to prepare a test setup that reduced the time needed at the test lab since the battery lab was highly occupied. The tests during the thesis were relying on getting test time in between the regular performance tests that were scheduled. The tests were also limited to the battery packs available in the lab at the test sessions.

All tests were performed in a climate chamber where the air in the chamber was controlled as well as the liquid cooling system of the battery pack, where they both were set to the target test temperature. The packs were complete battery packs ready for mounting into a vehicle, and the sensors onboard the pack were used for collecting data. One pack, named A1, was built in the lab and may have small differences in the

CHAPTER 3. EXPERIMENTAL SETUP

0 10 20 30 40 50 60 70 80 90 100

SOC [%]

3 3.2 3.4 3.6 3.8 4 4.2

Voltage [V]

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

SOC difference for a given voltage [%]

OCV curve

Charge Avarage Discharge SOC difference

Figure 3.3.2: OCV curve from a test made at Scania with the difference in SOC between charge and discharge, for a given voltage.

busbars and assembly process, compared to the ones from series production.

First, the test time for each test had to be reduced to conduct as many tests as possible.

It was identified that the time spent on relaxation periods substantially influenced the total time spent since there is a relaxation period initially, after charging and then after discharging. There is a compromise on how much time to spend on each relaxation period and how close to a stable value the voltage reading is. By having a long relaxation period after discharging and measure how the voltage dropped with time for a long period, it is possible to see approximately how long time it takes to reach close to equilibrium. It is worth noticing that the tests presented below were done on the older cell model, and it is assumed that the behavior is similar on the newer model.

In Figure 3.4.1a it can be seen that after 30 minutes, an average voltage of 3.582 volt is measured, and after around 63 minutes it is 3.584 volt. A difference of 2 mV is an increase of approximately 0.06%. This is a discharge down to 30% SOC at 15^◦C. In the other test, shown in Figure 3.4.1b it is discharging down to 10% SOC at 25^◦C.

It can be seen that after 30 minutes, the average voltage is 3.423 volt. The last registered value is 3.429 volt after two hours of relaxation. A difference of 6 mV representing an increase of approximately 0.18%, but it can be noted that the SOC value is lower which means it is in a region with slower chemical reactions.

In these tests, it can be seen that the change in voltage, after some time, is barely

CHAPTER 3. EXPERIMENTAL SETUP

2000 2500 3000 3500 4000 4500 5000 5500 Time [S]

3.48 3.5 3.52 3.54 3.56 3.58 3.6 3.62

Cell voltage [V]

Maximum, minimum and average cell voltage

Umax1 Umin1 Umax2 Umin2 Umax3 Umin3 avg X 1975

Y 3.488

X 3775 Y 3.582

X 5775 Y 3.584

(a)

4000 5000 6000 7000 8000 9000 10000 11000 Time [S]

3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8

Cell voltage [V]

Maximum, minimum and average cell voltage

Umax1 Umin1 Umax2 Umin2 Umax3 Umin3 avg

X 3458 Y 3.214

X 5258 Y 3.423

X 1.066e+04 Y 3.429

(b)

Figure 3.4.1: Relaxation step of cells with the three highest, lowest and average voltage during one (a) and two (b) hours after discharging, in different tests.

perceptible considering other sources of error introduced in a capacity measurement at a workshop environment. Hence, it is concluded that in order to reduce testing time, the relaxation period is set to 30 minutes for every relaxation. When evaluating the decreasing error in the measurement with relaxation time, it is compared to the 30 minutes value. When looking at the final capacity calculation, it is also from a 30 min relaxation time.

Since the battery pack was already used for tests in the lab when the thesis’s scheduled test sessions were, it almost always started on a higher SOC then the chosen start SOC values. Before the tests were carried out there was a discharge cycle to approach the right start SOC values.

Both the charging cycle and the discharging cycle were conducted with 1C = 56A current since it is the desired testing rate for workshops. There was a relaxation period of 30 minutes after correcting the state of charge to the start value before the charging process. The next relaxation period was after the pack was charged to the end SOC, and the final relaxation period was after the discharging process. The same strategy for relaxation time was applied for the constant voltage step and the current was monitored manually and shutdown at C/6 as explained in subsection 3.1.4.

The first strategy was to control the test by monitoring the SOC since it would be easy for the workshop to work with. However, due to a changed behaviour over time from the SOC algorithm it was chosen to control on voltage for the start value by having a CV step to C/3. It was then calculated how much energy should be charged to reach

CHAPTER 3. EXPERIMENTAL SETUP

a certain SOC level. To give an example, if start SOC is 20% and end SOC is 80% the energy to charge is 0.6 times the rated capacity. Monitoring energy was used when discharging too, to use the same amount of energy in both direction.

3.5 Monitored data for performance evaluation

As mentioned at the beginning of the chapter, the parameters of interest are SOC window size, relaxation time, temperature, and current. In this section, it is elaborated on how the parameters were evaluated.

The most important factor was naturally to calculate the capacity from the measured energy. The capacity was calculated as described in section 3.2, and in order for the workshop measurement to be comparable with the BMU estimation, the same definitions were used. The BMU is always identifying the smallest estimated cell discharge capacity to be the pack capacity. The minimum capacity was calculated by first calculating the capacity of all cells with the average OCV curve and then identifying the smallest cell to resemble a process where the BMU is pointing out the smallest cell. The pack’s capacity was then calculated by reading the cell voltages from the identified cell, and compared by reading the OCV curve in different ways described in subsection 3.3.2.

The calculated capacity was then compared with a reference capacity for the pack given by the Scania lab. The reference capacity obtained by conducting a Scania designed test sequence where it finalises the sequence by doing a 100-0% SOC sweep and assigning the measured energy as the pack’s capacity. The performance of the capacity measurement with respect to the reference capacity was then used to compare the results for different SOC window sizes, relaxation times, and temperatures. This was done for both a charging and discharging cycle. The charging cycle is most comfortable to implement at a workshop, as described in subsection 4.6.1, but the discharging cycle was also evaluated to see how large of a difference it is. From a customer perspective, the discharging capacity is of greater importance.

The relaxation time is a big part of the total testing time and the goal was to reduce the testing time while keeping the accuracy high. This was evaluated by plotting the voltage error, comparing the mountainous voltage reading with the final voltage value,

CHAPTER 3. EXPERIMENTAL SETUP

during the relaxation period. This is shown in Equation 3.2.

Relaxation error = voltage(1 : end)− voltage(end)

voltage(end) × 100 (3.2)

The Relaxation error becomes a vector here since voltage is, giving a good illustration of how fast the voltage approach a more stable voltage and thereby a more accurate reading.

Chapter 4

Workshop strategy

This part of the report will elaborate on how to approach the capacity measurement from a workshop perspective. Furthermore, reflect on how the test would be conducted at a workshop, strategy for scheduling the test or trigger it, and discuss the difference between a test at a workshop and in a battery lab. It is essential to see what limitations there is with a workshop test and why it balances precision in the measurement and time spent on it.

4.1 Introduction to challenges at a workshop

When conducting a capacity measurement of a battery, everyone wants as good accuracy as possible, and in a battery lab it is spent a lot of time and money on having the best equipment and test procedure. The challenge with developing a workshop test like this is that there is a delicate balance between precision in the test and time as well as money spent on it. The measurement accuracy should be good enough for answering the target questions with a certain degree of confidence, but outside of that the cost for increasing the accuracy is deemed too high.

The most crucial goal of the capacity measurement is to answer if the battery has reached EOL, since the confidence in the estimation is not high enough yet, as described in section 1.2. The balance between time and accuracy is not fully defined since there is no hard limit on the time, but it is given to have an error of well under 5

%, and the already existing test takes more than 3 hours. It is also preferred to consider the equipment existing in the workshop today and equipment planned to be acquired

CHAPTER 4. WORKSHOP STRATEGY

in the near future. It is further discussed under subsection 4.2.2.

The testing time is critical since it is costly for the customer to have the vehicle at a workshop. It is expensive both in terms of workshop cost and since the vehicle is standing still, and one of the most important factors for a transportation company is the up-time of the truck. Additionally, by having a test spanning over a shorter period, it is easier to plan it into the maintenance schedule.

Another challenge with designing a workshop method for testing the batteries, given that the time should be reduced, is not knowing at what SOC level or temperature the vehicle will arrive at the workshop with. In order to have a repeatability of the test and a consequent testing procedure, the method will require fixed start and end SOC value as well as a defined temperature span to be within. It is also important to have a big SOC window to increase the accuracy of the measurement while keeping in mind that an increased window size increases the test time.

The time will also depend on the available equipment and a large cost is connected to the charger’s ability, especially if it should have the capability to discharge. To reduce the dependency of initial status of the vehicle, it would be required to discharge the batteries and a bi-directional charger would be the optimal solution. However, it is an extra cost that might not be possible to justify. The equipment is elaborated on in section 4.2.

4.2 Equipment

This section will elaborate on how the workshop’s available equipment and set up on the vehicle will influence the test’s overall capabilities.

4.2.1 External factors

The equipment needed is heavily affecting the time spent on the test, but there is a relation between the charging equipment’s capability and its cost. The higher power capabilities of the charger, the faster the test can be done in theory. However, the charging speed is restricted by other factors, such as the allowed charging speed of the cell and the fuse installation at the workshop.

During the experimental tests, the maximum charging speed was set to 1C=”56A” from

CHAPTER 4. WORKSHOP STRATEGY

the cell responsible at Scania. There is room for increasing it in the future, but this was set as the upper limit for the scope of the thesis. The limitation from the electrical installation in the workshop building is also important to consider, where a 32A or 64A fused three-phase outlet makes a considerable difference in available power. The fusing at the workshop can be changed and will probably need to be changed to higher ones in the future, but for now it is seen as an unnecessary cost. From Scania, it was said that many workshops have a rating of 32A on the three-phase outlets.

4.2.2 Charger

As mentioned earlier, the ideal charger would handle 56A, preferably more to be future proof, and support bi-directional charging with the same rating. Having bi-directional properties, or what the concept is called in a broader perspective, Vehicle to Grid (V2G), is beneficial from a time perspective and an environmental perspective. Having bi- directional properties where the energy can be sent back to the electrical grid again would, on a bigger scale, mean saving a considerable amount of energy instead of being burnt.

In Equation 4.1, it is shown how much energy has to be used by the consumers onboard for every vehicle, considering a 8*56Ah battery setup, coming into the workshop with 31% SOC and counting with a nominal voltage of 666V. The start voltage of the test is 20% SOC.

E_burnt = 8× 56 × 0.11 × 666 ≈ 33kW h (4.1)

Around 33kWh per vehicle and test, given the conditions mentioned above, which is equivalent of using the dishwasher 16 times [26].

As mentioned previously in the chapter, this would be a solution that would not be economically justifiable considering today’s fleet of partially or fully electrified trucks.

The area of maintenance and workshop methods for electrified vehicles will however be more and more important. It will be necessary to equip the workshops with such systems when the rolling fleet of vehicles with this demand increases.

Scania is in the process of selecting a conservation charger as a recommendation for workshops to be able to maintain the battery SOC in the vehicle while conducting maintenance on it. The three most promising chargers, here called charger 1, charger

CHAPTER 4. WORKSHOP STRATEGY

2, and charger 3 have the following specifications.

• Charger 1: 25kW, 90kg, 63/32A, 11’705 Euro

• Charger 2: 40kW, 120kg, 63/32A, 16’900 Euro

• Charger 3: 24kW, 70kg, 32A, 14’000 Euro

Here, the specifications give maximum continuous power output, the weight, the outlet fusing size it is built for, and the cost. The higher power output version is more expensive but considering it would be possible to utilise the charger for the workshop capacity test and as a conservation charger, it can be argued that the increased price is justified. It is presented in subsection 4.5.2 how these chargers’ specifications are impacting the testing time.

4.2.3 Discharging

As mentioned in section 4.1, the SOC of the battery pack when the vehicle arrives at the workshop will vary. This means that it will be required to have the possibility to discharge the battery to lower the SOC if it is too high or if it needs to be discharged to match one pack with another.

Ideally, this functionality was inbuilt to the charger, which would also save energy if it is reinstated to the grid but since that functionality will not be available for now, other methods had to be explored for the discharging procedure. Another method explored was to discharge the battery with energy consumers onboard. After discussions with relevant departments at Scania, it was concluded that it is a feasible solution. The following energy consumers were discussed.

• DCC 6kW

• Heater1 6.8kW

• Heater2 9kW

The mentioned consumers and the summed up power of 21.8kW will be used in the calculations, but there are potentially more consumers that could be used, but due to lack of information about power consumption for the different systems, it is not counted in.

CHAPTER 4. WORKSHOP STRATEGY

4.3 Scheduling strategy

The capacity measurement’s primary goal is to verify if the battery has reached EOL in case the SOH estimator onboard says it. However, there is a potential for the workshop capacity measurement to be used in other ways as well that was briefly covered during the thesis. During the thesis, there have not been data available for analysing this and base a strategy upon. The section could rather be seen as presented options that may be of interest to investigate further when the development has come further, and there is more operational data available.

4.3.1 Trigger the test

The most straightforward strategy would be to wait for the online SOH estimator to say that the battery has reached EOL and needs to be replaced. The test would then be used to cross-check before either replacing or calibrating the BMU and continuing to use the battery.

There was however room for elaborating on the idea of basing the test sessions on the SOH estimation. First, to cover a scenario where it is over estimating the health with the potential error of 5%, as used in the example at the beginning of the thesis, it should be tested when the estimation reaches 85%. Doing this, a scenario is avoided where the estimation is showing 80% SOH but in reality it is 75%, which is passed the EOL (considering the example in section 1.2), and can potentially heavily affect the vehicle’s performance.

If it would be shown that the SOH estimator is drifting continuously throughout the whole lifetime of the battery, more frequent tests need to be conducted to calibrate the estimator with reasonable intervals. If this is the case, it would not be enough to do a measurement at, for example, 85% SOH and then expect the algorithm to capture when it is passing the EOL limit accurately. Instead, use the algorithm to trigger a test every 2% reduction during the last 5% of batteries’ life, for example.

Another parameter to build a test schedule from is the energy throughput, which is the amount of energy that has been passed through the battery pack in total. The BMU is already monitoring this parameter. One strategy could be to start doing a test after half the target energy throughput and then schedule more frequent tests when getting closer to 250MWh (from example in section 1.2). The reason for not starting to conduct