Techno-economic analysis of Battery Energy Storage Systems and Demand Side Management for peak load shaving in Swedish industries

(1)

IN

DEGREE PROJECT ENERGY AND ENVIRONMENT, SECOND CYCLE, 60 CREDITS

STOCKHOLM SWEDEN 2020,

Techno-economic analysis of Battery Energy Storage Systems and Demand Side Management for peak load shaving in Swedish

industries

BENJAMIN NESTOROVIC DOUGLAS LINDÉN

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT

(2)

Techno-economic analysis of Battery Energy Storage Systems and Demand Side Management

for peak load shaving in Swedish industries

Authors

Benjamin Nestorovic bne@kth.se and Douglas Lindén doulin@kth.se Department of Energy Technology

KTH Royal Institute of Technology

Examiner Björn Laumert

Supervisors

Monika Topel Capriles

Anna Nordling

WSP Systems – Energy

(3)

i

Abstract

The Swedish electrical grid has historically been robust and reliable, but with increased electrification in numerous sectors, out-phasing of nuclear power and a high market diffusion of wind power, the system is now facing challenges. The rotational energy in the system is expected to decrease as a result of higher shares of intermittent energy sources, which can affect the stability of the grid frequency negatively. To manage increased frequency drops, the new Fast Frequency Reserve (FFR) market will be implemented by June 2020 in the Nordic power system.

Simultaneously, it is expected that the demand of electricity will increase significantly in the transport and industry sectors in the coming years. Several DSOs already today indicate challenges with capacity and power security and have or will implement power tariffs as an economic incentive to prevent these problems. For energy intensive customers, such as industries, it will become important to reduce power peaks to avoid high grid fees.

Several peak load shaving strategies can be utilized by industries to reduce their power peaks and thus the power tariff. The aim of this study is to economically analyze peak load shaving for Swedish industries. This is done using Li-Ion BESS and DSM, and to maximize the utilization of the BESS by including energy arbitrage and FFR market participation into the analysis. Firstly, a literature review is conducted within the topics of peak load shaving strategies, energy arbitrage and ancillary services. Secondly, data is gathered in collaboration with WSP Systems – Energy, the initiators of the project, to conduct case studies on two different industries. These cases are simulated in the modeling software SAM, for technical analysis, and then economically evaluated with NPV. Also, nine scenarios are created for the emerging FFR market concerning the number of activations per year and the compensation price per activation.

The results from the case studies indicate that peak load shaving of 1 – 3 % with BESS provides a positive NPV for both case industries. However, higher percentages result in negative NPVs when no additional revenue streams are included. When considering energy arbitrage, it is concluded that the additional revenues are neglectable for both industries. Participating in the FFR market provides similar trends in the results as before. The exception is valid for scenarios with high numbers of FFR activations and compensation prices, where positive NPVs for all levels of peak load shaving can be concluded.

The peak load shaving strategy DSM is implemented for one of the industries, where efficiency measures are concluded to have the most impact on the economic evaluation. If all efficiency measures would be implemented, the electricity consumption would be reduced by 17 %.

Additionally, the power peaks would be reduced with 18 % and result in a significantly more positive NPV than peak load shaving using BESS. A sensitivity analysis concerning BESS capital cost and power tariff price concludes that the BESS price has a strong relation to the NPV, where a BESS price reduction of 60 % results in an NPV increase of at least 100 %. BESS prices have decreased the past years and are expected to keep decreasing in the future. Hence, investments in BESS can become more profitable and attractive in the coming years. Finally, for future research, it is recommended to combine the methodology from this study together with a load forecasting method. This combined methodology could then be practically applied to case specific industries with high peak loads.

Keywords: Peak load shaving, battery energy storage system, demand side management, Fast Frequency Reserve market, power tariff

(4)

ii

Sammanfattning

Det svenska elnätet har historiskt sett varit robust och pålitligt, men i takt med ökad elektrifiering i flera sektorer, utfasning av kärnkraft samt ökad mängd installerad vindkraft ställs nu systemet inför nya utmaningar. Bland annat förväntas rotationsenergin i systemet minska som ett resultat av högre andelar intermittenta energikällor i systemet. För att hantera detta kommer den nya Fast Frequency Reserve (FFR) marknaden finnas tillgänglig från och med juni 2020. Samtidigt förväntas även efterfrågan på el inom transport- och industrisektorn öka markant de kommande åren. Redan idag är effektbrist ett problem i vissa regioner, vilket kan komma att förvärras. Många nätägare ska eller har redan infört effekttariffer för utnyttjande av deras elnät, vilket är ett ekonomiskt incitament för att hantera effektproblematiken där kunder med en mer flexibel elkonsumtion kommer gynnas. För större elförbrukare, som exempelvis industrier, kan det bli ekonomiskt betydelsefullt att sänka sina effekttoppar och därmed undvika höga nätavgifter.

För att minska effekttoppar finns ett flertal så kallade peak load shaving-strategier, som kan utnyttjas av industrier för att minska kostnaderna för effekttariffen. Syftet med denna studie är att analysera peak load shaving för svenska industrier, med hjälp av ett Li-Ion batterilagringssystem och efterfrågeflexibilitet, samt maximera utnyttjandet av batteriet genom att inkludera energiarbitrage och deltagande i FFR-marknaden i analysen. Ett första steg i arbetet är att utföra en litteraturstudie för de berörda områdena. I ett andra steg insamlas data tillsammans med WSP, initiativtagaren av projektet, för att kunna göra en fallstudie på två industrier. För dessa fallstudier undersöks de tekniska förutsättningarna för att implementera peak load shaving-strategier genom modellering i simuleringsprogrammet SAM. Sedan utreds de ekonomiska förutsättningarna för fallstudierna, där NPV används som ekonomiskt nyckeltal. Dessutom skapas nio scenarion för den kommande FFR-marknaden för att uppskatta kostnader och inkomster.

Resultatet av fallstudien visar att 1 – 3 % kapade effekttoppar med batterilagring ger ett positivt NPV för båda industrierna. Över 3 % blir resultatet negativt utan ytterligare inkomstströmmar inkluderade. Energiarbitrage konstateras att bidra med marginella positiva fördelar. Vid inkludering av FFR-marknaden i analysen erhålls liknande trender i resultaten, bortsett från scenarion med relativt högt antal avrop och pris. I dessa fall blir även 4 – 10 % kapade effekttoppar ekonomiskt attraktiva.

För en av industrierna utvärderas efterfrågeflexibilitet, där effektivisering av elkrävande processer har störst inflytande på resultatet. Vid implementering av samtliga effektiviseringsåtgärder skulle elkonsumtionen minska med 17 %. Dessutom minskar effekttopparna med 18 %, vilket resulterar i ett signifikant mer positivt NPV, jämfört med användningen av batterilager. En känslighetsanalys gällande batteripris och effekttariffer, konstaterade att batteripriset har en stark påverkan på NPV.

Vid en batteriprisminskning på 60 % ökar NPV med minst 100 %. Därmed kan batteriinvesteringar bli mer gynnsamma och attraktiva om batteripriser fortsätter att falla, vilket flera prognoser indikerar. Slutligen rekommenderas framtida studier att kombinera metodiken från detta arbete med en prognostiseringsmetod för elanvändning i industrier. Denna kombinerade metod kan sedan praktiskt tillämpas på fallspecifika industrier med höga effekttoppar.

Nyckelord: Effektoppsreducering, batterilagringssystem, efterfrågeflexibilitet, Fast Frequency Reserve, effekttariffer

(5)

iii

Extent of work

This study is a Master’s thesis conducted at the Energy Department of the Royal Institute of Technology (KTH), in collaboration with WSP Systems – Energy in Stockholm. It was done within the Master’s programme Sustainable Energy Engineering, which is included in the five- year Energy and environment engineering programme. The thesis was conducted by Benjamin Nestorovic and Douglas Lindén. The two authors have been working fulltime with the thesis between the 13th of January and the 24th of May 2020.

Both authors carried out all stages of the work at the same rate. The literature review, resulting in a theoretical background, was divided equally between the authors, where Douglas had main responsibility for the literature review of peak load shaving and BESS. Benjamin had the main responsibility for reviewing literature of DSM and energy arbitrage, while both authors had an equal responsibility for ancillary services. Furthermore, the data collection was conducted jointly by the two authors. However, the case modeling for peak load shaving using BESS, including all revenue streams, for Stantek was performed by Benjamin and the modeling for Drying Industry was performed by Douglas. Also, the analysis in the report for the two industries was divided between the two authors accordingly. Extra time and effort were required for the investigation of the emerging FFR market as an ancillary service. Apart from the modeling and the economic evaluation, this included the creation of scenarios for the future occurrence and compensation price for the FFR market, where Douglas created the occurrence scenarios and Benjamin created the price scenarios. Regarding the DSM implementation, the main responsibility for the energy efficiency measures was held by Douglas, while the DR measures were mainly analyzed by Benjamin. All economic calculations, including the sensitivity analysis, were jointly performed by the two authors. All report writing was divided equally between the two authors, where the responsibility areas above generally reflect how the writing was divided.

The tasks performed within the scope and timeframe of this thesis correspond to a workload of 60 credits. Each part of the thesis was completed with equal contribution from the two authors.

(6)

iv

Acknowledgments

We want to express our warmest gratitude to everyone that has been supporting us throughout the study and a special thanks to everyone that have contributed to our work with data, feedback, and ideas.

Firstly, we are deeply grateful towards WSP Systems – Energy for providing the opportunity to conduct a study within this interesting topic. Thanks to all colleagues from the Energy department that have helped us throughout the project. We want to especially recognize our supervisor Anna Nordling for her feedback and support, and Anna Boss, who has helped us with the data collection.

Our sincere appreciation goes to Monika Topel Capriles, our supervisor at KTH, for her guidance and general support throughout the project.

Lastly, we want to thank Bill Edwall for his inputs in the initial stage of this thesis.

(7)

v

List of figures

Figure 1. Different phases in FFR activation (ENTSOE, 2019). ... 13

Figure 2. Summary of the peak load shaving strategies and the additional revenue streams. ... 14

Figure 3. Drying Industry’s yearly power demand on an hourly basis. ... 15

Figure 4. An example of a weekly electric load for Drying Industry. ... 16

Figure 5. Stantek’s yearly power demand on an hourly basis. ... 17

Figure 6. An example of a weekly electric load for Stantek. ... 17

Figure 7. Illustration of the cost structure in the power tariff. ... 19

Figure 8. A system overview and direction of electricity. ... 20

Figure 9. Graphical illustration of peak load shaving for Drying Industry using BESS. ... 27

Figure 10. Levels of peak load shaving and the corresponding NPVs for Drying Industry. ... 28

Figure 11. The NPV for the nine different scenarios utilizing BESS for Drying Industry... 29

Figure 12. A comparison of NPV between Ref 1 and no FFR participation. ... 30

Figure 13. A comparison of NPV between Med 2 and no FFR participation. ... 30

Figure 14. A comparison of NPV between High 3 and no FFR participation. ... 31

Figure 15. Discounted cash flow and NPV for the energy efficiency measures. ... 33

Figure 16. A comparison between the modified yearly load and the original load. ... 35

Figure 17. Comparison of NPV for peak load shaving with and without DSM measures. ... 35

Figure 18. NPV for FFR scenarios after implementing DSM measures. ... 36

Figure 19. Graphical illustration of 15 % peak load shaving for Stantek using BESS. ... 38

Figure 20. NPV for different levels of peak load shaving using BESS. ... 39

Figure 21. Graphical illustration of peak load shaving with the different FFR scenarios. ... 41

Figure 22. Sensitivity analysis conducted on Drying Industry for 10 % peak load shaving. ... 42

(10)

viii

List of tables

Table 1. Cyclic degradation (System Advisory Model, 2018) ... 7

Table 2. Description of different reserve markets available in Sweden. ... 11

Table 3. The alternatives for maximum full activation time. ... 13

Table 4. The number frequency drops in the electrical system. ... 21

Table 5. The number of FFR occurrences for each scenario by 2040. ... 21

Table 6. Limitation of Oskarshamn 3 during 2018. ... 21

Table 7. Price scenarios based on the limitation of Oskarshamn 3. ... 22

Table 8. The nine combined scenarios for compensation price and occurrence for FFR. ... 22

Table 9. General input data to the analysis. ... 24

Table 10. Specific data for the BESS used in the simulation. ... 25

Table 11. Drying Industry's monthly power peaks. ... 26

Table 12. Required BESS capacity for the different percentages of peak load shaving. ... 26

Table 13. Stantek’s monthly power peaks. ... 37

Table 14. Required BESS capacity for the different levels of peak load shaving. ... 38

Table 15. Monthly grid power targets for Drying Industry, for all levels of peak load shaving. . 53

Table 16. Monthly grid power targets for Stantek, for all levels of peak load shaving. ... 53

Table 17. NPV results for Drying Industry. ... 54

Table 18. NPV results for Drying Industry with the modified load curve. ... 54

Table 19. NPV results for Stantek. ... 54

(11)

ix

List of abbreviations

aFRR Automatic Frequency Restoration Reserve BESS Battery Energy Storage System

C-rate Charge/discharge rate ESS Energy Storage System EV Electric Vehicle

DOD Depth-of-Discharge

DR Demand Response

DSM Demand Side Management DSO Distribution System Operator

FCR-D Frequency Containment Reserve Disturbance FCR-N Frequency Containment Reserve Normal FFR Fast Frequency Reserve

KPI Key Performance Indicators Li-Ion Lithium-Ion

mFRR Manual Frequency Restoration Reserve NPV Net Present Value

O&M Operation & Maintenance Pb-A Lead-Acid

PV Photovoltaics

SAM System Advisor Model SEK Swedish krona

SOC State of Charge SOH State of Health

TSO Transmission System Operator

(12)

1

1 Introduction

1.1 Background

The electrical system in Sweden has historically been one of the most reliable electrical systems in the world. It is characterized by a large share of the electricity production in the northern parts of Sweden and most of the consumption in the south. Presently, hydro- and nuclear power each stand for approximately 40 % of the total electricity production (Energimyndigheten, 2020). Until recent years, the power balance between the geographical regions have been kept due to a robust and well sized system, but with an increasing electricity demand and an out phasing of nuclear power, grid stability has become problematic. Additionally, most of the projected wind power plants are to be installed in the northern parts of Sweden, while the consumption is forecasted to increase more in the southern parts. The capacity in the grid, transferring electricity to the south, will thus be overloaded and the transferring capacity needs to be strengthened for power security.

The Royal Swedish Academy of Engineering Sciences (IVA) roughly estimates that the transferring capacity needs to be doubled to handle these challenges. (IVA, 2019a)

The increased market penetration of renewable energy, such as wind power, causes the intermittency of the electric system to increase (IVA, 2017). It is Svenska kraftnät’s (Svk) responsibility as the Swedish Transmission System Operator (TSO) to maintain the electricity grid balance and ensure a high electricity quality. The inertia of the system depends on the type of electricity production in it, where production with high rotational energy, such as nuclear power, increases the inertia. It is forecasted by Svk that the rotational energy will decrease with approximately 30 % until 2040, which exposes the system to an increase of errors. Innovative ancillary services, such as power reserves, need to be implemented to avoid errors and maintain high stability (Svenska kraftnät, 2019d). One emerging power reserve is the Fast Frequency Reserve (FFR), which will be implemented in the Nordic power system by June 2020 with the purpose to provide frequency regulation.

Besides challenges with increased intermittent electricity production, the electricity demand in several sectors is forecasted to increase. The electricity demand in the Swedish transport sector is expected to increase with 5 – 10 TWh by 2030 and additionally 10 – 15 TWh by 2045 from today’s yearly demand of approximately 2.6 TWh (IVA, 2019c). In the Swedish industrial sector, the current yearly electricity demand is approximately 50 TWh. Due to a desired reduction of fossil fuels and increased electrification, the electricity demand is forecasted to increase with about 32 – 52 TWh by 2045 (IVA, 2019b). A reliable electricity supply is therefore essential for industries and is vital for a sustainable future progression within the sector. As of today, regional challenges concerning capacity and power security already exist. Distribution System Operators (DSOs) in regions such as Stockholm, Uppsala, Malmö and Mälardalen have problems meeting an increased electricity demand from customers, such as industry companies (Svenska kraftnät, 2019c). Industries in the south of Sweden have been forced to establish their business elsewhere and similar problems might occur in other areas. Additionally, the industry customers currently operating in these areas pay high electricity prices, where the power tariff constitutes a significant amount of the total cost. It is also expected that the power tariffs in the electricity market will increase in the coming years, due to the described existing and forthcoming challenges.

(13)

2 There is thus a need for industries to decrease their power peaks to thereby decrease their power tariffs, which simultaneously would increase the stability of the grid. The power peaks can be reduced in different ways, using peak load shaving strategies. There are several various strategies available for peak load shaving, where two common are Demand Side Management (DSM) and integration of Energy Storage Systems (ESS). DSM can be divided into energy efficiency measures and demand response (DR), where the latter requires the industry to have some flexibility in its electricity demand (Uddin, et al., 2018). According to (IVA, 2019a), there is a potential of 2 GW in electricity demand flexibility for Swedish industries, which makes DSM implementation an interesting research subject for achieving peak load shaving. Since not all industries can apply DSM measures, it is relevant to evaluate the potential of ESS and more specifically Battery Energy Storage Systems (BESS), which is the most common technology for peak load shaving applications (Uddin, et al., 2018). Due to the rapid price declination for Lithium-Ion (Li-Ion) BESS), there are presently examples of Li-Ion BESS being utilized for peak load shaving, but also for other applications such as energy arbitrage and ancillary services. (IVA, 2019a)

This study investigates peak load shaving using Li-Ion BESS for two case industries in Sweden.

DSM as a peak load shaving strategy is also examined for one of the industries. Furthermore, additional utilization of the BESS is investigated by including energy arbitrage and ancillary services with focus on the emerging FFR market for frequency regulation. This has not been fully done in previous literature, where DSM and BESS for peak load shaving combined with energy arbitrage and FFR market participation has not been researched.

1.2 Aim and objectives

The aim of this study is to perform an economic analysis of peak load shaving for Swedish industries, using Li-Ion BESS and DSM, and to maximize the utilization of the BESS by including energy arbitrage and FFR market participation into the analysis.

To successfully achieve the aim of the study, the following objectives are undertaken:

● Develop test cases for evaluating peak load shaving for Swedish industries.

● Utilize the Li-Ion BESS technology and identify the technically optimal sizing for case specific peak load shaving applications.

● Analyze the economic potential for reducing peak power tariffs for industries using Li-Ion BESS.

● Identify possibilities to achieve peak load shaving with DSM measures and analyze the economic impact.

● Explore the economic potential of energy arbitrage, when using BESS for peak load shaving.

● Utilize the Li-Ion BESS technology and identify its technically optimal sizing for FFR market participation.

● Analyze the economic potential for an industry participating in the FFR market with a BESS.

(14)

3 This study is conducted in collaboration with WSP Systems – Energy, with the purpose of evaluating future possible business opportunities within the field of peak load shaving in Swedish industries.

1.3 Scope and limitations

The study only analyses Swedish industries, participating on the electricity markets covered by Nordpool. The only reserve market of concern in this study is the FFR market. There is a limited amount of industries included in the study and the system model is based on their historic load curves from specific years. Load forecasting is outside the scope of this study. Due to data constraints, the DSM measures are only evaluated for one of the industries. A limitation of BESS technology is that only Li-Ion BESS is considered for the analysis, although other BESS technologies are mentioned in the review of literature. Finally, the energy arbitrage potential is calculated as a consequence of peak load shaving and is not jointly optimized with the peak load shaving controller.

1.4 Methodology

This study is divided into four phases.

1. Theoretical background 2. Data collection

3. Creating and modeling cases 4. Analyzing results

The first phase is to conduct a comprehensive literature review and to present a theoretical background within the topics of peak load shaving strategies, energy arbitrage and ancillary services. This is done to understand the boundary conditions and to gain greater knowledge within the mentioned topics. Although this phase is considered completed when most of the important theoretical parameters are included, more information is gathered throughout the study when needed. This is followed by a second phase of data collection, which is done in collaboration with WSP Systems – Energy. They possess a wide network of Swedish industries from different sectors, from where electricity demand data can be obtained. The third phase is to build cases based on the theoretical background from phase one and the collected data from phase two. These cases are modelled in the simulation software SAM for technical evaluation. The last phase is to analyze the economic potential of implementing peak load shaving strategies, along with the additional revenue streams. Phase one is presented in chapter 2, while phase two and three are presented in chapter 3. Finally, phase four is presented in chapter 4, where the results are described and analyzed, and chapter 5, where they are discussed and evaluated.

(15)

4

2 Theoretical background

In this chapter, a theoretical background based on a literature review is given for the different parts of the study. Firstly, the different peak load shaving strategies (BESS integration and DSM measures) are presented in subchapters 2.1 and 2.2, respectively. In addition, the BESS technology and more specifically the Li-Ion battery technology is described in more detail in subchapter 2.1.1.

Two further application areas (energy arbitrage and ancillary service) for the BESS are then presented in subchapters 2.3 and 2.4, respectively.

2.1 Peak load shaving

The concept of peak load shaving has been widely studied in different areas and with different strategies, where a common goal for the strategies is to reduce the power peak. A significant literature review on peak load shaving strategies was performed by (Uddin, et al., 2018), where the following three strategies were investigated: DSM, integration of ESS and integration of electric vehicles (EVs). The authors found that integrating an ESS is the most potential strategy of peak load shaving for residential buildings, industries and grids and that BESS is the most common technology. It is also mentioned that sizing a BESS properly is of high importance to have a well- functioning system. Oversizing the BESS can result in an unused capacity and it might affect the economic aspect negatively. Under-sizing forces the BESS to undergo more cycles of charge/discharge and the lifetime will ultimately be shortened (Martins, et al., 2018). (Chua, et al., 2016) developed a sizing method, which was tested on a building at the Universiti Tanku Abdul Rahman in Malaysia. Based on historical data, a rule-of-thumb was developed for customers to size their BESS.

Research within peak load shaving using BESS has been done to a great extent and different sizing methods and strategies have been studied. (Martins, et al., 2018) propose a linear optimization method for cost-optimal sizing of the battery and power electronics in an industrial setting. The study found that peak load shaving using BESS can shorten the payback period for large industrial loads. (Nayak & Nayak, 2017) used an improved harmony search algorithm, an optimization technique, with the objective to find a minimum annual operating cost. It resulted in a reduction in cost of purchased electricity for a grid connected solar photovoltaics-BESS (PV-BESS) system.

A BESS integrated in a distribution power system was studied by (Yang, et al., 2013). The used model intended to investigate the behavior of the battery when increasing voltage in the distribution system caused by high penetration of PV. The model considered state of health (SOH), which depends on both cyclic and calendric aging. Furthermore, degradation is a factor of importance and (Wankmuller, et al., 2017) concluded that degradation of the BESS can reduce revenues with 12 – 46 %.

Control strategies have been studied to increase the lifetime and to optimize the utilization of the BESS. (Oudalov, et al., 2017) presents a sizing methodology aiming to decrease costs and maximize economic benefits. Furthermore, optimal operation was studied using dynamic programming to charge and discharge the battery at an optimal time. It resulted in a cost reduction of the electricity bill. A similar study conducted by (Lu, et al., 2014) determined the optimal size of the battery using load forecasting. In addition, the control of the charging and discharging was optimized to minimize the peak-valley differences, as well as the variation in the daily usage. It is

(16)

5 concluded peak load shaving with limited battery capacity can be attained with an optimal control strategy. Furthermore, (Garcia-Plaza, et al., 2018) and (Reihanin, et al., 2016) respectively studied battery control for peak load shaving applications on a microgrid and an island power system.

(Garcia-Plaza, et al., 2018) highlighted the importance of internal factors (i.e. SOH and minimum voltage) of the battery for future development of these systems. (Reihanin, et al., 2016) investigated load forecasting with two different approaches, a non-linear and a real-time strategy, and found that BESS integration decreases the variability in the transmission system.

Several studies evaluate component sizing for different BESS technologies with economic parameters. (Telaretti & Dusonchet, 2016) examined the economic feasibility, of four different BESS technologies in peak load shaving applications. The four technologies were: Li-Ion, advanced Lead-Acid (Pb-A), sodium-sulfur and flow batteries. The results revealed that none of the BESS technologies were profitable for peak load shaving considering current BESS costs.

However, Li-Ion and flow batteries were the technologies approaching the break-even point and Li-Ion batteries are expected to be the most promising alternative in the near future, thanks to a forecasted cost decrease. A reversed approach was taken by (Rahmann, et al., 2017) that instead researched what the break-even costs are for different BESS technologies to make peak load shaving applications financially viable. The study showed that only Pb-A and zinc-bromide batteries have potential to reach the break-even, when considering current market prices, while Li-Ion costs are too high. Different costs are used in the study for all BESS technologies. However, the same round-trip efficiency, depth of discharge (DOD) and cycle life for all BESS technologies are used. Many other studies make sharp distinctions of technical performance parameters between different BESS technologies. In (Hesse, et al., 2017a), a literature survey on performance parameters for Pb-A BESS and two Li-ion BESS variations is done. It showed a major superiority in round-trip efficiency, DOD and cycle life for the Li-ion BESS technologies, but lower capital cost per kWh for the Pb-A BESS.

A case study aimed to investigate the Demand Side Flexibility of a shopping center in Väla, Sweden, with implementation of a BESS, addressed the need of further revenue streams to make the investment of the system more financial attractive. Participating in the Nordic reserve market is one alternative that is mentioned but not included in the study (Iggström & Svensson, 2019).

(Rahmann, et al., 2017) comes to the conclusion that more benefits from the BESS would probably be needed to make the different BESS technologies economically competitive. Similar conclusions are drawn in (Hesse, et al., 2017b) where combinating different applications increases utilization of the BESS and allows the sum of several application profits. It is shown that this profit increase in most cases outnumbers the increased cyclic degradation of the BESS.

2.1.1 BESS

A system is “a set of connected things or devices that operate together” (Cambridge Dictionary, 2020) . A BESS consists of many different components and even systems within the system itself.

Apart from a container holding the components in place, it also consists of power electronics, an inverter, a transformer, as well as thermal and energy management systems. Considering the capital cost, even engineering, permitting and project management can be included, all depending on where the system boundary is drawn and what to include in the analysis. The core component

(17)

6 in the BESS is the battery. It consists of one or several cells that can store energy in the form of electricity. Within the cell(s), there is a chemical reaction between two electrodes plunged into an electrolyte. The material of the electrodes and the electrolyte differ between various types of batteries, which gives them their unique characteristics. The battery discharge occurs by activating chemical reactions in the cells, which creates a flow between the two electrodes. The battery charge follows the same principle but reversed, caused by an external voltage across the electrodes.

Batteries can be compared in energy density (Wh/kg) or power density (W/kg), which can indicate the area of implementation. For applications where the battery is required to discharge for longer time periods, the energy density is a more relevant parameter, while the power density is more relevant when high power outputs are needed for shorter time periods. If the battery is intended for mobile applications, such as mobile phones or EVs, a generally higher density is preferable.

The lifetime of a battery is usually measured in how many cycles the battery can undergo before it needs to be replaced. However, it is not only the amount of cycles that affects the lifetime of the battery since the self-degradation also is a factor. The SOH is an indicator that presents the condition of the battery related to its initial capacity. For stationary systems, 80% SOH is generally used as a limit, indicating the need for replacement (Hesse, et al., 2017b). The SOH is calculated as presented in Equation 1, where Cnominal is the initial capacity, Cfade, calendric is the calendric degradation and Cfade, cyclic is the cyclic degradation. (Hesse, et al., 2017b)

𝑆𝑂𝐻 = (𝐶_{𝑛𝑜𝑚𝑖𝑛𝑎𝑙}− 𝐶_{𝑐𝑎𝑙𝑒𝑛𝑑𝑟𝑖𝑐}^{𝑓𝑎𝑑𝑒} − 𝐶_{𝑐𝑦𝑐𝑙𝑖𝑐}^{𝑓𝑎𝑑𝑒}) 𝐶_{𝑛𝑜𝑚𝑖𝑛𝑎𝑙}

Eq. 1

The calendric degradation of the battery is presented in Equation 2, where T is the room temperature, SOC is the state of charge and t is specific day. In this study, the room temperature is assumed to be 20℃.

𝐶_{𝑐𝑎𝑙𝑒𝑛𝑑𝑟𝑖𝑐}^{𝑓𝑎𝑑𝑒} = 1.02 − (0.00266 ∗ 𝑒⁻⁷²⁸⁰⁽¹^𝑇−²⁹⁶⁾¹ ∗ 𝑒⁹³⁰⁽^𝑆𝑂𝐶^{𝑇 −}²⁹⁶⁾¹ ) ∗ √𝑡 ^{Eq. 2}

The cyclic degradation, Cfade, cyclic, is calculated depending on the depth of each cycle, see Table 1.

Moreover, several studies conclude that the battery management impacts the lifetime, i.e. how the battery is charged/discharged. This can be measured in real time with SOC, which indicates the amount of energy stored in a battery at a specific time, related to the maximum capacity of the battery (Kousksou, et al., 2014; Vazquez, et al., 2010). Excessive discharging of the battery to low SOC values increases the cyclic degradation, which can be seen in Table 1. Hence, the SOCmin is set to 15 % for the analysis conducted in this study. Furthermore, when the battery is idle, a SOC below 100 % is preferable to decrease prominent calendric degradation (Hesse, et al., 2017b). The SOCmax is therefore set to 95 % for the analysis in this study.

(18)

7 Table 1. Cyclic degradation (System Advisory Model, 2018)

Depth-of-discharge (%) Cycles Elapsed Capacity (%)

20 0 100

20 5,000 80

20 10,000 60

80 0 100

80 1,000 80

80 2,000 60

BESS is predicted to have a major role in the future energy system, managing high shares of renewable energy sources. It is estimated by The International Renewable Energy Agency (IRENA) that BESS would need to have a capacity of 150 GW by 2030 worldwide to meet the goal of 45 % generated electricity from renewable energy sources (IRENA, 2015). Currently, 88 % of newly installed energy storage types are Li-Ion BESS (excluding pumped-hydro). The European commission estimates that stationary storage systems may reach up to 1,300 GWh by 2040, which indicates a significant market growth of Li-Ion BESS (Tsiropoulos, et al., 2018; IEA, 2020).

Stationary systems have a wide range of potential applications areas, where grid supporting services, behind-the-meter installations and off-grid solutions are most suitable for Li-Ion BESS (IRENA, 2017).

Different batteries have reached different stages of commercial maturity. Due to the rapid development of EVs, Li-Ion batteries have gained an increased performance and a fast price declination in the recent years. The average cost for stationary Li-Ion BESS applications was approximately 5,700 SEK/kWh (where 1 € = 10 SEK) in Europe 2017 (Tsiropoulos, et al., 2018).

The capital cost comprises the storage module (battery pack), the balancing of the system, the power conversion system and engineering, procurement and construction. Moreover, the BESS corresponding to this capital cost is assumed to have a Charge/discharge rate (C-rate) of 1 h-1. The C-rate is a measure of the rate a battery can be charged or discharged and is calculated by dividing the battery bank power with the battery capacity. When the C-rate, and hence the battery bank power, is increased, the capital cost increases. A common rule-of-thumb is that the capital cost increases with 15 % when the C-rate is doubled. Hence, a BESS with a C-rate of 4 h-1 has a capital cost of approximately 7,540 SEK/kWh, according to the average cost in Europe 2017 mentioned above.

Several studies indicate a price reduction for BESS in the coming years. According to McKinsey, a price between 1,700 – 2,700 SEK/kWh is plausible by 2025 (C-rate = 1 h-1) (Frankel, et al., 2018). In comparison, the scenario analysis conducted by (Tsiropoulos, et al., 2018), predicts that prices for Li-Ion BESS stationary systems can be lowered with 30 – 55 % by 2030 and

(19)

8 approximately 66 % by 2040. Compared to today’s average, the price would be 2565 – 3990 SEK/kWh by 2030 and approximately 1940 SEK/kWh by 2040. However, it is important to highlight price variations between different system applications, but also between different types of Li-Ion batteries. Because of the application diversity, different battery technologies are preferable depending on the requirements of the specific application.

2.2 DSM

Except from utilizing a BESS, there are more options for increasing economic competitiveness for peak load shaving applications. The study conducted by (Uddin, et al., 2018) on different peak load shaving strategies recommends future research to emphasize on application of DSM in combination with ESSs for peak load shaving purposes. DSM is the alteration of consumer energy demand through various approaches such as financial incentives or education to change behavior patterns. The goal of DSM is often to encourage less energy consumption during peak hours. DSM can be categorized into energy efficiency (reducing energy consumption) and DR (rescheduling and shifting energy consumption). Applicating DR requires additional investments due to need of data exchange infrastructure and increased variable costs for shifting loads. (Paulus & Borggrefe, 2011)

DSM can be beneficial for several stakeholders, such as utilities, customers, the public and the government. For energy customers, such as industries, DSM offers the possibility of power tariff reduction through energy efficiency measures. In the case of industrial companies, this leads to lower production costs and increased product competitiveness. Energy efficiency in an industrial context implies performing the same tasks, while using less energy. This measure includes a permanent reduction of energy consumption by using more efficient appliances and/or processes (Zhang, et al., 2017). Except from capital costs and increased operation & maintenance (O&M) costs, energy efficiency measures do not interfere with potentially crucial activities within the industry. Hence, energy efficiency might be a better option than DR for industries with production processes that cannot be reduced, flatten or shifted.

For industrial companies with flexible loads, the reactive and preventative method of DR can be utilized instead of, or as a complement to, energy efficiency. While energy efficiency reduces the energy consumption (and possibly the peak power demand), DR has historically focused on peak power demand reduction specifically. DR can be defined as the changes in energy consumption by demand-side stakeholders from their usual demand patterns. This change can be achieved through a wide range of actions taken at the customer side in response to different conditions within the electricity system. These can be changes in price of electricity, incentive payments to encourage lower electricity consumption at specific times or when the reliability of the system is of concern. (Uddin, et al., 2018)

A common instrument that can be used to implement DR on energy customers is dynamic pricing.

(Bergaentzlé, et al., 2015) mentions five pricing schemes:

1) Time-of-use pricing – Divides the day into different periods depending on a predetermined price. A simple and common pricing scheme with limitations, since the flexibility is restricted due to small price differences.

(20)

9 2) Critical-peak pricing – Based on the same strategy as Time-of-use pricing, but includes more time segments for price variations. The price difference between the critical-peak period and the baseline is often large enough to encourage a load reduction from the customer.

3) Peak-time-rebate pricing – Instead of increasing the price during peak hours, this pricing scheme rewards a reduced consumption during these periods.

4) Real-time-pricing – The consumer pays a variable price that follows the hourly variations of the wholesale price.

5) Inclining-block-rate – Reduces the average electricity consumption by increasing the price according to the consumed amount. The prices are divided into blocks, with certain thresholds for price increase.

The instruments mentioned above have different advantages depending on application area and on customer type. For instance, (Ashok, 2006) concluded that large industries can reduce their peak demand by rescheduling their production according to the Time-of-use tariff. In general, combining several instruments gives the most effective results. To secure the positive aspects of combining instruments, direct load control might be a prerequisite (Bergaentzlé, et al., 2015). This however increases the complexity of the overall system operation since many customers may not be willing to shift their activities. Furthermore, challenges with infrastructure need to be assessed before DR programs can be implemented, since the capital costs are high. (Uddin, et al., 2018)

2.3 Energy arbitrage

A possible application for BESS to combine with peak load shaving is energy arbitrage, which consists of charging and discharging the battery when economically favorable. For energy arbitrage to be applicable, the electricity consumer must pay a non-constant price for the electricity.

To reduce demand peaks, utilities can apply different instruments of dynamic pricing (see subchapter 2.2). The difference between peak and off-peak prices can thereby be exploited to increase the revenues for the BESS investment (Telaretti, et al., 2015). In a study conducted by (Telaretti & Dusonchet, 2016), the revenues from reduced peak demand charges and energy arbitrage are combined when peak load shaving is applicated using different BESS technologies.

The authors conclude that none of the BESS technologies provide positive economic results, but that a future decline in BESS capital costs will change the situation for primarily the Li-Ion technology. This is supported by (Telaretti, et al., 2015), which mentions Li-Ion BESS as the most promising in terms of both cost reduction and cycle performance.

Since energy arbitrage alone rarely provides a profitable investment, several studies emphasize the importance of combining applications in order to maximize profits. The study conducted by (Brivio, et al., 2016) explores multi-services BESS by simulation energy arbitrage together with ancillary services, with focus on the primary control reserve. (Gundogdu, et al., 2019) concludes that energy arbitrage profits can be made and that combining it with ancillary services (grid frequency regulation) throughout the day increases the revenues significantly. The significance of ancillary services is highlighted in the market analysis of energy arbitrage in Ontario, Canada, by (Bassett, et al., 2018). They conclude that the potential for energy storage facilities to earn revenue from ancillary services is significant compared to the marginal revenue generated from energy

(21)

10 arbitrage. Furthermore, (Hesse, et al., 2017b) proposes a combination of peak load shaving and ancillary services using BESS. The same recommendation for future research is made in (Schram, et al., 2018), where ancillary services is proposed to be suitable as a combination with peak load shaving.

2.4 Ancillary services

One application for ancillary services is frequency regulation, which is available through different reserve markets controlled by the Transmission System Operator (TSO). Using BESS for peak load shaving and frequency regulation is jointly optimized in (Shi, et al., 2018), where the results suggest BESS to achieve much larger economic benefits when it provides multiple services.

(Zakeri & Syri, 2016) showed great economic potential of ancillary servies for different ESSs on the Nordic power reserve market. Another study on the Nordic reserve markets saw a profitability increase of 16 – 28 % when a combined heat and power plant participated in the reseve market (Haakana, et al., 2017). In addition, (Hesse, et al., 2017b) mentiones the potential superiority of BESS compared to conventional power plants in the reserve markets thanks to the short response time of BESS. This is reflected in the recent development on the Nordic power market, where the different Nordic TSOs together have established the new reserve market: FFR. The FFR market is intended to be available for fast responding reserves, where especially BESS is suitable.

2.4.1 Power reserve markets

The nominated electricity market operator in Sweden is Nordpool, who delivers day-ahead and intraday trading, clearing and settlements to customers. Nordpool provides day-ahead and intraday trading in the Nordic, Baltic, Central Western European and UK markets. Hence, trading is done between different countries and regions through import and export. In the day-ahead market, Nordpool’s customers can sell and buy energy for the next day in a closed auction, where bids are matched to maximize social welfare (price is set where demand and supply curves meet). To keep the necessary balance between supply and demand, the intraday market provides trading closer to the actual delivery of energy. With the increasing amount of installed intermittent renewable energy, the demand for intraday trading and other solutions increase, since it becomes more difficult to balance the market during the day-ahead trading. (Nordpool, 2019)

To maintain balance in the electric grid, the Swedish TSO Svk utilizes power reserves. These can be production facilities, that can produce more electricity, or industries that can decrease their electricity consumption accordingly. There are several different reserve markets for stakeholders that want to participate in the system as power reserves. In Sweden, there are currently four available reserve markets; FCR-N, FCR-D, aFRR and mFRR (Svenska kraftnät, 2019a). The new FFR market will be activited for the first time during the summer of 2020. The different reserve markets available in Sweden are described in more detail in Table 2. General requirements for the reserve markets are (Svenska kraftnät, 2019b):

• prequalification approved

• real time measurement

• electronic communication

• endurance

(22)

11 Table 2. Description of different reserve markets available in Sweden.

FCR- Normal

FCR- Disturbance

aFRR (automatic)

mFRR (manual)

FFR

Smallest bid size

0.1 MW 0.1 MW 5 MW 10 MW 0.1 MW

Activated automatic during frequency deviation of

+-0.10 Hz from 50.00

Hz

automatic during frequency deviation of

more than - 0.10 Hz from

50.00 Hz

automatic via central signal

when frequency deviate from

50.00 Hz, during the hours the reserve is procured

manually when demanded by

Svk

automatic

Activation time

63% within 60 s and 100% within

3 min

50% within 5 s and 100

% within 30 s

100% within 120 s

within 15 min (longer time

allowed)

0.70 – 1.30 s

Volume requirement

~ 200 MW for Sweden

~ 400 MW for Sweden

~ 150 MW for Sweden

- 72 MW for

Sweden Other symmetrical

product that can regulate both up and

down

The four currently active reserve markets are based on the Frequency Containment Reserve (FCR) and the Frequency Restoration Reserve (FRR). FCRs are used by TSOs to maintain the frequency stable. FCR-N regulates frequency deviations of 0.10 Hz, both over and under 50.00 Hz, while FCR-D regulates frequency deviations below 49.90. If the frequency deviation lasts longer than 30 seconds, FRR replaces FCR. As shown in Table 1, FRR can be distinguished between automatic activation (aFRR) and manual activation (mFRR). (Svenska kraftnät, 2019b)

(23)

12 2.4.2 The FFR market

Due to the ongoing structural changes of the electric system, the Nordic TSOs have together studied inertia-related issues and potential solutions in several projects. Analysis of historical data shows that the largest frequency deviations occur during the summer season. This arises due to lowered electricity consumption during the summer, which causes power plants that provides inertia to the system to decrease their production. Daily differences in frequency deviation also occur due to decreased electricity consumption between 22.00 and 07.00. A Nordic study concluded that the FFR is the most promising market to adapt to these challenges. Statnett, the TSO in Norway, carried out a pilot for the new FFR market, based on the conclusion of the Nordic study. It used a mix of existing and new power reserves to examine the potential for the FFR market. The pilot was considered successful and valuable insight regarding the FFR market was achieved (Statnett, 2018). A common goal for the Nordic countries has been stated to implement the FFR market by the summer of 2020 and Svk is currently in the startup phase of a project to implement it. The aim is to have the FFR market in place by June 2020, operate it until September 2020 and fully implement it by 2021 (Svenska kraftnät, 2020a). The volume requirement of FFR is 72 MW in Sweden and 300 MW in the Nordics combined. (Svenska kraftnät, 2020c)

The implementation of FFR is divided in two steps. Firstly, a pre-qualification of potential FFR-providers is done by Svk to ensure that the providers fulfill the technical requirements. The pre-qualification can be done in two different ways of testing. There is a possibility to either simulate an external or an internal frequency signal to evaluate the suitability of the potential provider. The test method using an internal simulated frequency signal needs a complementation test with natural frequency variations, which examines the frequency measurement. During the pre-qualification, the data to be logged are (Svenska kraftnät, 2020c):

• Active power [MW] – resolution 0.01 MW

• Measured grid frequency [Hz] – resolution 10 mHz – accuracy 10 mHz

• Applied frequency signal [Hz] – resolution 10 mHz – accuracy 10 mHz

• Status ID for setting parameters (if these can automatically change during the test)

After the pre-qualification, the providers approved by Svk will continue to the second step, which is the procurement process. In this step, all approved providers will receive an invitation to leave offers regarding the FFR reserve. The offer should include the compensation (SEK/MW) the provider demands for being available for activation by Svk during the agreement period.

Compensation will be given even if the provider is not activated for a specific period, which can occur since Svk will have a safety margin on when FFR is needed. During the activation process, Svk will rank the different providers, starting with the lowest compensation demand first.

However, all providers will be compensated with the marginal price, which means that the last and highest activated offer sets the price for all offers for a specific activation hour. During 2020, Svk will conduct their FFR orders by telephone twice a week. The project goal is to implement an IT-solution for 2021 to make bids and orders easier. (Svenska kraftnät, 2020b)

FFR is intended to be a fast measure for responding to frequency deviation, which makes it suitable for BESS applications. To participate in the FFR market, there are several technical requirements

(24)

13 that need to be fulfilled. FFR can be divided into short (minimum 5.0 s) and long (minimum 30.0 s) support duration. Regardless of the support duration, the FFR provider needs to choose a frequency as their activation level, which has a corresponding maximum full activation time. There are three alternatives, which are presented in Table 3. In this study, alternative A is used for the analysis of the FFR market.

Table 3. The alternatives for maximum full activation time.

Alternative Activation level [Hz] Maximum full activation time [s]

A 49.7 1.3

B 49.6 1

C 49.5 0.7

After the support duration, the deactivation of the support starts, which has a certain timeframe.

The function of the deactivation varies whether the support has a short or long duration. When the support is deactivated, there is a buffer time before the recovery time. The FFR provider must have capacity for a new FFR activation cycle within 15 minutes after the previous activation cycle.

Some FFR providers might be able to supply several cycles without recovery, while others need to utilize the entire recovery time (ENTSOE, 2019). An illustration of the support activation, duration and recovery is presented in Figure 1.

Figure 1. Different phases in FFR activation (ENTSOE, 2019).

(25)

14

3 Data collection and case modeling

The two peak load shaving strategies analyzed in this study include BESS integration and DSM measures. The strategies are implemented for two case industries, which are further described in subchapter 3.1. The data collection for both industries are done in collaboration with WSP Systems – Energy. Peak load shaving with BESS, see subchapter 3.2, is considered for both case industries, while DSM, see subchapter 3.3, only is analyzed for one industry. Three revenue streams from the BESS utilization are considered and analyzed for both case industries, which consist of reducing the power tariff, exploiting energy arbitrage and participating in the Swedish FFR market. These revenue streams are described in subchapters 3.4, 3.5 and 3.6, respectively.

For the DSM measures, power tariff reduction and energy savings are the two considered revenue streams. Furthermore, the modified load curve after implementing DSM measures are used as input for analyzing a combination of peak load shaving with BESS and DSM, see subchapter 3.3. The economic calculations for both peak load shaving strategies and all revenue streams are conducted with net present value (NPV), see subchapter 3.7, where the sensitivity analysis also is presented.

The overall structure of the modeling approach is illustrated in Figure 2.

Figure 2. Summary of the peak load shaving strategies and the additional revenue streams.

3.1 Data collection

Two different case industries are included in this study. The first one is an industry with intensive drying processes, which are heavily dependent on fans for their operation. Due to company secrecy, the industry asked for anonymity in this study. Hence, this case industry is referred to as

“Drying Industry” in the proceeding parts of this thesis. Drying Industry, which is further described in subchapter 3.1.1, is investigated for peak load shaving utilizing both BESS and DSM. The second case, Stantek, is a smaller industry and operates in a different sector compared to Drying

BESS DSM

Energy arbitrage

Ancillary services

Peak load shaving strategies

Revenue streams Reduced power tariff

Energy savings

(26)

15 Industry. Stantek is only evaluated for peak load shaving with BESS and the industry is described in more detail in subchapter 3.1.2.

3.1.1 Drying Industry

Drying Industry is a company with operations in Northern Europe. The company is a global exporter and has a yearly turnover of about 2.3 billion SEK. This study focuses on the production facilities situated in Southern Sweden. Most of the consumed energy (83 %) is heat, which is delivered through a district heating system and is therefore outside the scope of this thesis. The annual electricity consumption is 17,000 MWh, where most of it is related to the production processes. There is a seasonal variation in the electricity consumption due to increased need for cooling during the summer months. The maximum power peak is 3.3 MW and occurs in July.

Generally, the power demand is between 1.5 MW and 2.5 MW, see Figure 3.

Figure 3. Drying Industry’s yearly power demand on an hourly basis.

The production processes do not follow the pattern of regular working hours and they generally have a limited flexibility. The consumption is relatively high during all hours of the week, where some fluctuations can be seen for certain energy demanding activities within the production processes. An illustration of a weekly electric load can be seen in Figure 4, which shows the daily and hourly variations of the electricity demand for a week in January.

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

0 973 1947 2920 3893 4867 5840 6813 7787 8760

Power (MW)

Hours

Drying Industry load

(27)

16 Figure 4. An example of a weekly electric load for Drying Industry.

The production is ongoing during unregular working hours for the entire week, including weekends, and since the electricity demand mainly comes from the production processes, Saturdays and Sundays have similar demands as weekdays. Furthermore, some peaks occur during nights as well.

3.1.2 Stantek

Stantek is an industry company situated in Kongsvinger, Norway. This study considers Stantek as a Swedish industry to be able to participate in the Swedish FFR market and to have the same price scheme for power tariffs as Drying Industry. The company produces thin plate steel products and is a subcontractor to other industry companies. The production processes are only active during weekdays since the industry is closed during the night and weekends. Stantek has an annual electricity consumption of approximately 395 MWh and a maximum power peak of 139 kW. On a yearly basis, there are significant differences in electricity demand. This can be seen in Figure 5, where the power demand exceeds 120 kW for only a few hours for one year. Since the productions processes constitute most of the electricity demand, there are no significant seasonal variations in electricity demand.

0.0 0.5 1.0 1.5 2.0 2.5

01/01 02/01 03/01 04/01 05/01 06/01 07/01 08/01

Power (MW)

Weekly Electric Load Drying Industry

(28)

17 Figure 5. Stantek’s yearly power demand on an hourly basis.

Even though the industry is closed during nights and weekends, some supporting processes are still active. Hence, the power demand rarely goes below 20 kW. This can be seen in Figure 6, where the hourly electric load for one week in January is illustrated. The peak loads are concentrated to working hours during weekdays. However, the hourly variations during these occasions are limited with a maximum difference of about 10 – 15 kW. This is a possible pointer of the potential for peak load shaving, since going beyond these values no longer increases the required battery capacity linearly.

Figure 6. An example of a weekly electric load for Stantek.

0 20 40 60 80 100 120 140 160

0 973 1947 2920 3893 4867 5840 6813 7787 8760

Power (kW)

Hours

Stantek load

0 20 40 60 80 100 120 140

20/01 21/01 22/01 23/01 24/01 25/01 26/01 27/01

Power (kW)

Weekly electric load Stantek

(29)

18

3.2 Peak load shaving with BESS

To analyze the technical aspects of peak load shaving with BESS and to determine the optimal battery capacity, the modeling software System Advisory Model (SAM) is used. SAM is developed by the US National Renewable Energy Laboratory (NREL) and supports simulations of several types of energy systems. One of many features in the software is battery storage, where it is possible to either store self-produced electricity or to store electricity from the electric grid.

Moreover, multiple peak load shaving tools are available in the software, where a grid power target is used for this study. The grid power target is calculated from the maximum peak for each month and varies with the level of peak load shaving, see Equation 3. The inputs in SAM are the electric load for the industry and the grid power target. Additionally, it is possible to specify detailed voltage, temperature and degradation models to determine both cyclic and calendric degradation of the BESS.

𝑃𝐺𝑟𝑖𝑑 𝑇𝑎𝑟𝑔𝑒𝑡,𝑀𝑜𝑛𝑡ℎ = 𝑃_{𝑀𝑎𝑥,𝑚𝑜𝑛𝑡ℎ} ∗ 𝑃𝑆_% ^{Eq. 3}

The power demand is modelled to not exceed the grid power target and when there is a demand above it, the BESS should manage the excess load. If the BESS fails to provide the needed power or is fully discharged before the demand goes below the grid power target, the BESS is considered too small and a larger capacity is simulated. It is thus an iterative process to find the optimal battery size for each level of peak load shaving, which are based on different grid power targets. The peak load shaving percentages included in the analysis are 1 – 10 %, 15 % and 20 %.

3.3 DSM

The different production and supporting processes are analyzed and the flows of energy are evaluated for one of the case industries. Both energy efficiency and DR measures are evaluated for the industry to reduce the peak load. The resulting reduction in peak load and total electricity consumption is used as input to the economic evaluation, where the capital cost and the O&M costs for implementing the measures are included. Moreover, possible options for DR measures and their implications are analyzed and discussed. Finally, the new load curve that has been altered from the energy efficiency measures is used as input for another peak load shaving analysis with BESS integration. This is done for the same variations of peak load shaving percentage as before.

3.4 Power tariff

A power tariff is a pricing method for DSOs to charge their customers for consumption and transmission of electricity. This is implemented to charge customers based on how much of the grid capacity they actually use and during what hours they do so. Typically, the charging scheme has a fixed cost (dependent on the size of the fuse), a transmission fee (multiplied with the bought electricity that month) and a power fee. The power fee is multiplied with the highest measured hourly peak for each month. A graphical illustration of the cost structure is presented in Figure 7.

Some DSOs have a high load fee, while others have plans to implement this soon. The high load fee is an extra fee customer must pay during months when the grid is particularly exposed to higher loads. It usually also depends on what hours in these months the electricity is bought. Generally,

(30)

19 the high load fee is applied from November to Mars between 06:00 – 19:00 but varies between the DSOs. This fee is calculated in the same way as the power fee.

Figure 7. Illustration of the cost structure in the power tariff.

In this study, Eon’s charging method is used for economic calculations. Eon currently charges 600 SEK as a fixed cost, 4.52 öre/kWh as a transmission fee, 88.40 SEK/kW as the power fee and has no high load fee. All prices above are presented without VAT. For Eon (and most other DSOs), prices differ depending on where in Sweden the customer is located. In this study, the prices from the southern pricing area of Sweden are used. As a disclaimer, the power tariff only applies for customers with needed fusehigher than 80 Ampere and 400 Volt. It is assumed that all case industries in this study have such a need. Eon also has a network subscription for customers who are in need of a connection with a 20 kV (or higher) grid (Eon, 2020). This is however not considered in this study.

3.5 Energy arbitrage

Energy arbitrage is analyzed as an additional revenue stream for the BESS utilization. To exploit energy arbitrage, the BESS needs to charge at a low electricity price and discharge when the electricity price is high. The new load after implementing the BESS for peak load shaving is extracted from SAM to Microsoft Excel. To calculate the energy arbitrage, a comparison of the yearly electricity cost is done between for the new load and the original load. Nordpool’s hourly spot price is used to calculate the yearly electricity cost in both cases. Energy arbitrage is only applied for the hours of peak load shaving. A joint economic optimization of peak load shaving and energy arbitrage is outside the scope of this study.

3.6 FFR market participation

As an additional revenue stream to the industry for the BESS utilization, ancillary services are considered. More specifically, the industry is modelled to participate as a reserve provider on the new FFR market. Since there is no historical data of the FFR market, three scenarios are created for both occurrence (number of activations per year) and price (compensation to the provider per

Fixed cost Transmission fee

Power fee

High load fee

Only during specific months and hours