Power mapping and aggregation as a service: A techno-economic view on Li-ion batteries for peak shaving and frequency regulation

(1)

UPTEC ES 20017

Examensarbete 30 hp

June 2020

Power mapping and aggregation

as a service

A techno-economic view on Li-ion batteries

for peak shaving and frequency regulation

Filip Angwald

(2)

Teknisk- naturvetenskaplig fakultet UTH-enheten Besöksadress: Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0 Postadress: Box 536 751 21 Uppsala Telefon: 018 – 471 30 03 Telefax: 018 – 471 30 00 Hemsida: http://www.teknat.uu.se/student

Abstract

Power mapping and aggregation of batteries as a

service

Filip Angwald

The world's energy supply today mainly consists of fossil fuels and nuclear power. Moving away from the use of these energy resources to renewable energy sources is considered a prerequisite for a sustainable future. In order to implement this change, it is necessary for renewable energy sources to be environmentally, technically and economically sustainable. A major challenge encountered in terms of technological sustainability is the intermittent nature of renewable energy sources. As the share of renewable electricity increases in the system, the electricity grid is facing new challenges such as increased instability of the frequency and capacity shortages. In order to meet these new challenges an increased flexibility from electricity users is proposed as a solution. Flexibility can be achieved either by controlling the use of electricity or utilizing energy storages. If different electric loads are to be controlled in a property, data regarding the power use of the loads must first be collected with a high time resolution in order to be able to properly analyze the data. Measures to shift or reduce the power peaks in a property can then be suggested and implemented. A battery storage can help reduce power peaks or shift loads in time and if done on a large scale that would reduce the strain on the entire Swedish grid. One of the ancillary services that the battery could offer is frequency regulation. Using energy storages for such an application could also provide a secondary revenue stream, aside from the revenue stream from peak shaving, and increase the profitability of the storage. Sweden has seen a dramatic increase in electric vehicles over the last decade and charging of the vehicles has become an issue for many property owners as it often creates power peaks.

The data collection regarding power use in properties performed in during this thesis showed that valuable data can be collected with the method and material used. With a battery price of 3000 SEK/kWh the payback time for a battery system can be reduced from 17,9 to 7,8 years if it is used for frequency regulation during the night. If power-intensive loads such as electric vehicle charging are

added to the model the payback period decreases to 6,1 years. With these results in mind, it can be concluded that the profitability of a battery storage can increase to the extent that the investment is of economic viability. In addition, the investment helps to improve the stability of the Swedish grid. The results are found to be relatively consistent with those of other similar studies.

Tryckt av: Uppsala Universitet ISSN: 1650-8300, UPTEC ES 20017 Examinator: Petra Jönsson

Ämnesgranskare: Cecilia Boström Handledare: Jonas Thyni

(3)

Populärvetenskaplig sammanfattning

Världens energiförsörjning idag kommer huvudsakligen från fossila bränslen och kärnkraft. En omställning från användningen av dessa energiresurser till förnybara energikällor anses vara en förutsättning för en hållbar framtid. För att genomföra denna förändring är det nödvändigt att de förnybara energikällor implementeras på ett hållbart sätt och en stor utmaning när det gäller teknisk hållbarhet är de förnybara energikällornas intermittenta karaktär. I takt med att andelen förnybar el ökar i systemet står elnätet inför nya utmaningar, såsom ökad instabilitet i frekvens- och kapacitetsbrist. För att möta dessa nya utmaningar föreslås en ökad flexibilitet från elanvändarna som en lösning. Flexibilitet kan uppnås antingen genom att kontrollera användningen av el eller utnyttja energilager. Om elanvändningen ska kunna styras i en fastighet måste data om effektanvändningen först samlas in med hög tidsupplösning för att kunna genomföra en analys på ett meningsfullt sätt. I denna studie föreslås effektkartläggning som en möjlig tjänst för att samla in data och kartlägga hur en aktörs effektanvändning ser ut och vilka kostnader som uppstår på grund av den. Åtgärder för att flytta eller minska effekttopparna i en fastighet kan sedan föreslås och lämnas över till aktören i fråga. Under perioder med högt effektuttag kan så kallade effekttoppar uppstå vilket kan ge upphov till förhöjda kostnader. För att minska kostnaden för effekttoppar kan batterier användas för att kapa topparna och denna studie undersöker ett sådant scenario. Även ett scenario där laddning av elfordon läggs till studeras för att undersöka hur det påverkar resultaten.

Batteriers tekniska och ekonomiska potential kan estimeras genom litteraturstudier och utifrån prisstatistik från den svenska regleringsmarknaden samt en teoretisk modell som utvecklats under detta examensarbete. Med ett batteripris på 3000 kr/kWh kan återbetalningstiden för ett batterisystem sänkas från 17,9 om det bara används för effekttoppskapning till 7,8 år om det dessutom används för frekvensreglering under natten. Om ett scenario där laddning av elfordon vid fastigheten också har lagts till modellen utvärderas blir återbetalningstiden 6,1 år. Med dessa resultat nås slutsatsen att lönsamheten för en batterilagring kan öka om batteriet används till flera olika applikationer under sin tekniska livstid och om det finns elintensiva laster som bidrar till höga effekttoppar. En del av den data om effektanvändning som samlades in under arbetet gick att validera mot data från elnätsägaren i området och det konstaterades att den insamlade data samt de beräknade kostnaderna stämde väl överens med de verkliga. Studien begränsas bland annat av att data endast insamlades under en månads tid vilket inte ger en komplett bild av hur effektanvändningen ser ut. Förutom att minska kostnader för el så bidrar tjänsten som beskrivs i detta arbete också till att förbättra stabiliteten i det svenska elnätet och omställningen till ett hållbart samhälle genom att möjliggöra en högre andel elfordon och förnyelsebar energi.

(4)

Executive summary

This master thesis was carried out in collaboration with the company Tvinn in order to investigate the possibilities and potential in two new business areas. In this report, the concept of power mapping is outlined and a process for how a power mapping service can be carried out is detailed and exemplified in practice. In addition, power peak shaving with batteries, the concept of aggregation of power resources and trading on frequency regulation markets is investigated. Lastly, an economic perspective is taken by estimating the payback period of batteries used for peak shaving and frequency regulation.

This study is to be considered as a foundation which Tvinn can use if they choose to move forward in these new business areas. The power mapping process which was outlined and tested can be realized quite quickly by following the steps suggested in this report. There is still more development needed regarding how to promote the service, pricing and what benefits the customer can expect from the service. For more reliant results regarding costs and potential savings, the time period in which data is collected should be longer than in this study. When considering peak shaving, batteries are found to be technically well suited for the task. However, the potential cost reductions are found to be very dependent on the power contract the consumer has. Going forward, the results in this report indicate that Tvinn should focus on consumers with power contracts which include fees for power use. Such fees are found to increase the potential cost reductions made by peak shaving substantially. The payback period in different scenarios which was estimated in this report indicate that if a consumer has charging stations for electric vehicles, the economic potential in peak shaving is higher than in the other scenarios. If Tvinn are to aggregate resources and trade on different markets, there is still a lot to be done before such a business can be realized. For example, Tvinn need to develop hardware and software solutions to control resources and algorithms for optimal trading for maximum profit.

In conclusion, there is a lot of potential in both these business areas and there is a clear connection between them. By providing power mapping and peak shaving as a service, a portfolio of resources can be collected over time with the end goal of having the technical capability to be an aggregator on the Swedish power market.

(5)

Foreword

This report is the result of a degree project of 30 credits, and the final part of the engineering energy systems program at Uppsala University. I would like to thank my supervisors Jonas and Jakob at Tvinn for support and guidance during the project. I would also like to thank my subject reviewer Cecilia Boström at Uppsala University for valuable comments and remarks. Many thanks to all the other people who have volunteered with time and commitment and made this study possible. In conclusion, I would like to thank my co-worker Nils who provided Python support, answered questions and contributed to a pleasant working environment.

Filip Angwald

(6)

Glossary and definitions

Aggregated resources: Several smaller units are merged into a larger volume that is

controlled according to a need.

Aggregator: A company that specializes in controlling and reselling aggregated consumption

or production flexibility.

Alternating current: Alternating current (AC) is an electric current which periodically

reverses direction.

Direct current: Direct current or (DC) is a flow of current which is static and does not

change direction.

Down regulation: Reduced production or increased consumption.

FCR-D: Frequency controlled reserve, used as primary control in case of major disturbances. FCR-N: Frequency controlled reserve, is used as primary control and is activated

automatically when the frequency deviates from 50 Hz. FCR-N aims to balance the grid in case of small and fast changes in frequency.

FRR-A: Automatic frequency control reserve that is centrally controlled and used as

secondary control.

FRR-M: Manual active reserve used as secondary control.

Grid support services: Services to the electrical grid such as frequency regulation and

voltage support.

IoT: Internet of things. Referring to components which have an internet connection.

Power peak: Power peaks in the electrical system occur when the power use is at its highest,

often when many electrical appliances are used at the same time.

Power profile: A series of data values regarding power use over some time period. Regulatory bid: Bid for up or down regulation where power and price are stated.

Resource owner: The owner of some resource which has an electric energy consumption. It

can be a building, company or electric vehicle.

Resource: Refers to production or consumption resource that can be adjusted as needed to

balance the power system.

Sub-ordered: A financial agreement between two parties where the availability of a resource

of some kind is purchased.

Transmission System Operator: Also called TSO. The entity responsible for the power

(9)

Up regulation: Increased production or consumption.

User flexibility: User flexibility means that customers can adjust their electricity consumption

at a certain point in time. This can be designed in different ways, for example through flexible pricing, agreed power reduction or forced load control.

(10)

1

1. Introduction

As a result of measures to address climate change, the share of renewable electricity generation in the Swedish electricity grid has increased dramatically over the past decade and there is a high probability that the share will continue to increase (Miljödepartementet, 2020). Of the new main renewable energy sources in Sweden, wind and solar, both are inherently intermittent and have led to the Swedish electricity grid facing new challenges. In addition, new types of loads have also been added that are challenging for the electricity grid, such as electric vehicle chargers, heat pumps, and an expansion of the manufacturing industry. This has led to what has been called a "lack of capacity" at some locations in Sweden, because while the amount of electricity produced in Sweden over a year is sufficient, there is a lack of capacity to transport the electricity to where it is needed at certain times (Uppsala Kommun, 2020). One way to deal with this issue is to increase the amount of decentralized and local electricity generation to reduce the amount of electricity that needs to be transported. The lack of capacity became an even more pressing issue in Uppsala, where the grid owner and operator Vattenfall made a request to the Swedish TSO Svenska Kraftnät (SVK) for an increased power supply but was denied the request. Vattenfall then had to deny new grid connections in the local grid as a result of the capacity shortage (Andersson & Bernström, 2019). The grid is also facing challenges with a decrease in inertia as the traditional power plants with large turbines and generators are being replaced with new technology. This has led to a more volatile frequency which more often deviates from the nominal value and thus the need for reserves for regulating the frequency has increased (Karlsson & Nordling, 2016). To meet the new demand for support services of that kind, new innovative solutions such as aggregation of flexible resources are emerging.

Tvinn1 is a relatively new company located in Uppsala working with software solutions in the energy sector. Tvinn wants to be part of the solution for the capacity shortage issue and at the same time promote a continued increase of renewable energy and electric vehicles in Sweden. This thesis aims to lay the foundation for Tvinn to develop and add new services in their portfolio that can potentially help to free up capacity in the electricity grid and promote efforts towards achieving climate goals. The services in question shall, on the one hand, be aimed at electricity users who wish to reduce their electricity costs or have more control over their energy system. On the other hand, Tvinn can offer aggregated resources on markets for frequency regulation and other emerging markets for capacity and flexibility. Therefore, this work aims to gather information to contribute additional knowledge to this area and to facilitate the creation of an aggregator business for Tvinn.

(11)

2

1.1 Purpose and goal of thesis

The aim of this work is to develop new services that Tvinn can offer customers in the future and to evaluate the potential economic benefits. The services consists of several parts which are explained in more detailed in other sections of this report but overall data collection, potential for energy storage and potential for aggregation are key questions in this work. In addition, the work investigates opportunities, obstacles and economic potential for Tvinn to act as a aggregator of flexible energy resources and trading on different markets in Sweden. There are many other studies which have discussed technical aspects of different energy storages such as suitability for frequency regulation (Lindgren, 2019) and the degradation of the storage related to such applications (Xu, et al., 2017). There are also numerus studies which discuss the potential for demand side response and flexibility by the consumer (Svenska kraftnät, 2017). This work will instead focus on the questions and issues Tvinn faces as a future aggregator and also estimate the payback period for energy storages in the form of batteries.

The questions discussed in this work are as follows:

1. How can power use data be collected and what equipment is required in order to perform a power mapping?

2. What will the payback period be for a battery storage system if it is used for; a) Shaving power peaks?

b) Shaving power peaks and used for trading on frequency regulation markets? c) Shaving power peaks with added load from electric vehicle charging and used for

trading on the frequency regulation markets?

1.2 Scope of thesis and limitations

This thesis will focus on the Swedish power system. The electricity grid in Sweden is part of the larger Nordic grid and have connections to neighboring countries such as Norway, Denmark, Finland and others. However, only the Swedish part of the grid will be considered in this study. When considering power in different systems, only active power will be discussed. The reason being that active power is what usually defines the fuse size of installations, the costs are based on active power and when trading on frequency regulation markets only active power is relevant. Reactive power and its possible effects on various power systems will not be discussed in this report.

In this study, data regarding power use is collected for the month of March 2020 for two different properties, which means that many aspects of the work will be based on a one-month period. Many costs and fees are usually based on a one-month period which makes it a reasonable time period for estimating monthly costs. However, when using the data over a month to evaluate possible incomes and revenue streams over several years the uncertainty increases drastically. It was not possible to collect data for a longer period during this thesis but regardless the data is used to estimate economical aspects such as payback period for batteries.

(12)

3

The reader may keep in mind that the calculations in this study should be considered as estimates.

Regarding energy storage systems, energy storage in this thesis will always refer to batteries and specifically batteries with Lithium-ion (Li-ion) chemistry. There are many different types of energy storage systems such as flywheels, pumped hydro and hydrogen gas but none are as commercially available and tested as the Li-ion battery (Zablocki, 2019). An important factor for Li-ion batteries is the battery lifetime and how the use of the battery affects it. There are numerous studies on this matter (Ovejas & Cuadras, 2019), and it will not be discussed at length in this report even though the applications of peak shaving and frequency regulation has a major impact on the battery lifetime. To know how much energy is used when selling a resource, such as a battery, on the frequency regulating markets one must analyze how the grid frequency varies over the time period when the battery is activated. Such an analysis is beyond the scope of this thesis but might make a good subject for further studies in this field.

In this study, calendar degradation of the battery will not be included in the calculations. How one uses a battery is crucial for the lifetime and for the purpose described in this report, i.e. for peak shaving and support services for the grid, the usage can be optimized in several ways. However, such optimization is not part of this study. Finally, the concept described in this report is primarily based on software solutions and IoT but those aspects are not inside the scope of this thesis and should be the subject for future studies. There are also many legislation and regulatory aspects which could have major effects on the concept, however these are not in the scope of this thesis.

1.3 Structure of report

This report will initially summarize the work that has been carried out followed by a glossary of the terminology used in the work. In the background, the problems that have given rise to this study and the work that Tvinn AB wants to carry out in the future are discussed. The theory chapter provides information about the equations and theoretical background which was used in following sections in the report. The Materials and methods chapter outlines how the work has been carried out and what equipment was used for data collection etc. The results are presented in the next chapter where power data, simulations and payback periods are presented. There is also a sensitivity analysis and a data validation for some of the results. Finally, the results are discussed and compared to those of other similar studies and conclusions are presented.

(13)

4

2 Background

This part of the report provides some background as to why the work was carried out and different preconditions. Information has been gathered mainly from literature, various reports in the energy field and the Swedish TSO website and its various regulatory documents.

2.1 Layout of the Swedish electricity grid

Today, most of the Swedish electricity production consist of hydro and nuclear power. Together they produced about 80% of the electricity in 2019. However, their share is declining as more and more new renewable energy sources such as wind power have seen a major expansion (Energimyndigheten, 2020). In the Swedish electricity grid, there are three levels: transmission grid, regional grid and local grid. The transmission grid consists largely of high-voltage power lines that can transmit the electricity with low losses. Since a large part of Sweden's power production takes the form of hydropower in northern Sweden, the transmission lines usually extend from north to south and branch out in regional grids that in turn branch out into local grids. The difference between the different grids are voltage levels, a high voltage leads to less losses in the lines since the losses are proportional to the current squared and therefore transmission lines have a voltage of 220-400 kV, regional grids have a voltage level of 40-130 kV and local grids below 40 kV (Svenska Kraftnät, 2020). Furthermore, Sweden is divided into four so-called electricity areas called E1, E2, E3 and E4, the price for electricity can vary between different electricity areas since there are some bottleneck effects which reduce the possibility of transmitting power between them (Energimarknadsinspektionen, 2012). In Figure 1 an overview of the electricity grid with its different voltage levels is shown.

Figure 1. Overview of the Swedish electricity grid division into grids with different voltage levels (IVA, 2017).

Historically, the Swedish state has had a monopoly on electricity trade, but in 1996 the rules were reformed and the market for electricity trade and electricity production was opened with the aim of creating a more efficient system. Electricity trading companies can buy electricity directly from an electricity producer or from the Nordpool electricity market and can then sell the electricity to an end user (SVK, 2014).

2.2 Electricity market and trading

Buyers and sellers of electricity can choose between trading the day before (day-ahead) or on the same day (intraday) as the delivery of power must take place. On the day ahead market electricity is traded for the next day. The price is then set based on an auction procedure for

(14)

5

each hour during the next day, where bids on the market are submitted until 12 o'clock. An hour later, Nord Pool will publish the prices for the following day. Thus, when trading on the day ahead market, the volume needed is purchased and the price is paid according to supply and demand for the next day's respective hours and prices. The intraday market is an adjustment market that can adjust any changes in the production and consumption of electricity among the market participants after the day-ahead market has closed. The adjustments can be made up to one hour before the delivery hour (Nordling & Burén, 2019).

2.3 Electricity contracts and costs

Any property or business that has a connection to the electricity grid needs an agreement with the company who owns the grid in their location. The local grid owner installs an electricity meter at the property in question that records the electricity usage and sends it to the grid owner. The user is then billed based on the measured energy and power used every month. There are a wide variety of contracts with different grid owners and operators and the structure of the contract can vary based on geographic location and what voltage level the user needs. Many contracts are divided into residential, low voltage and high voltage contracts and prices, tariffs as well as other fees can vary. High voltage refers to connections between 6-20 kV, low voltage and residential connections are usually 230 V (Vattenfall, 2018).

2.4 Frequency regulation in Sweden

The electricity market differs from other markets in that the commodity which is sold must be used the same moment its produced. If not, an imbalance will occur in the grid which in turn lead to a deviation from the nominal frequency value. If the production is greater than the demand, the frequency will increase and if the demand is greater than the production it will decrease. The reason why it is important to keep a constant frequency in the grid is because induction and synchronous generators as well as many other components only operates as intended if the frequency is kept at the nominal value (Watt-logic, 2017). However, both production and demand can be hard to accurately predict and there will always be fluctuations in the frequency. In the Swedish grid the goal is to keep the frequency ±0,1 Hz from the nominal value and to achieve this goal frequency regulation reserves are used.

In Sweden, SVK has the system responsibility for electricity and that the Swedish electricity grid is always in balance. To achieve this, SVK procures multiple resources to build reserves which can then be used when the frequency in the grid deviates from the nominal value of 50 Hz. At present, there are 4 different reserves that serve different purposes for frequency control, FCR-Normal, FCR-Disturbance, aFRR and mFRR (SVK, 2020). Figure 2 below illustrates how the different reserves are used.

(15)

6

Figure 2. The functions of the various reserves for frequency control in the Swedish grid (Weihs, 2019).

Figure 2 shows the reserves used in the event of a frequency disturbance. In the event of a frequency deviation, the automatic primary control is primarily used. In several power stations, today its mostly hydropower stations, production automatically increases as the frequency decreases and decreases as the frequency rises. Primary regulation is essential to maintain balance and stabilize the frequency as it changes. The reserves have been sub-ordered in advance by SVK and have equipment installed that sense the frequency at every moment of the day. This means that they are activated automatically and stabilize the frequency if the frequency changes within the frequency range they are meant to support. Primary regulation includes the frequency containment reserves FCR-N and FCR-D. Then the "automatic frequency restoration" (aFRR) reserves is used to restore the frequency to the normal range and "manual frequency restoration" (mFRR) to restore and stabilize the frequency to its nominal value.

2.5 The capacity shortages in Sweden

As a result of the climate goals, Region Uppsala has a project in progress to electrify public transport in Uppsala. Among other things, a new bus depot is planned for 2020. The bus depot is intended to house several electric buses and Region Uppsala therefore applied for a new grid connection to the depot with a capacity of 6 MW to allow fast charging of the buses. The local electricity grid owner Vattenfall denied the application because of the lack of capacity in transmission lines to Uppsala and instead offered a capacity for the bus depot of 1.5 MW during some hours of the day and 4 MW during other hours. Similar problems have arisen in other

(16)

7

Swedish cities such as Stockholm and Malmö (Andersson & Bernström, 2019). The reason for the situation of lack of capacity has been the fast growth in cities the recent years and the fact that local electricity production has in some cases decreased, for example local CHP plants have stopped generating electricity as the profitability has deteriorated (energigas, 2020). The power that the local grid owner must be able to guarantee to its customers can no longer be delivered to the local grid because the transmission lines supplying the electricity are not dimensioned for the increased demand. To solve the problem, SVK has plans to expand the transmission lines and the expansion is expected to be completed by 2030. As urban growth is expected to continue and the amount of new electricity-intensive components such as electric vehicle charging and heat pumps increases, the new capacity is expected to already be used when the construction of new transmission lines is complete and thus the capacity shortage will persist after 2030 (Andersson & Bernström, 2019).

In order to seek and provide solutions to the problem with capacity shortages a project called Coordinet2 has been initiated by the European Union where several countries are working together to develop solutions. In Sweden, Coordinet is developing a platform called Flexmarket with the purpose to reduce capacity shortages in the long term by building a marketplace for flexibility resources. It is seen as an important solution to address the capacity challenge. This can be done, for example, by means of electricity users such as households or industries that control their electricity use in a way that helps to even out the overall load in the electricity grid or by temporarily increasing/decreasing production or utilizing energy storages. The market is developed to be able to accommodate both large and smaller resource owners as well as companies who aggregate resources. Currently, Flexmarket is in a testing and demonstration stage where the market will be open for any aggregator who want to be part of the testing during the winter of 2020. Several other countries such as Germany and Great Britain has already developed functioning markets for flexible resources. For example, Piclo Flex3 in Great Britain provide a platform for resource trading which has proven to be an effective way of increasing the flexibility in the grid (Sahlén, et al., 2019).

2.6 Business model for power mapping, energy storage as a service and

aggregation

This section outlines the process for the new service Tvinn could provide to electricity users while at the same time building a portfolio of resources which can then be aggregated and utilized as a commodity on markets for flexibility or frequency control. First the process of power mapping and power peak shaving will be explained as this lays the foundation for the following section. The process of power mapping is defined in this thesis as the collection and analysis of data with regards to power use. The aim of the process is to map when power peaks occur and what the source of the peaks is. Then, the role of the aggregator in future energy systems is discussed. Illustrations that outline the processes are included to provide some

2_{Coordinet website:}_{https://coordinet-project.eu/}_{(visited 2020-06-05)} 3_{Piclo Flex website:}_{https://picloflex.com/}_{(visited 2020-06-01)}

(17)

8

visualization of the concept. Figure 3 below shows an example of what the procedure might look like in a power mapping and potential aggregation with Tvinn.

Figure 3. Flowchart for Tvinn’s business with power mapping and aggregation of resources.

A collaboration starts with a resource owner who is interested in reducing their electricity costs and streamlining their operations. Step two is to install measuring equipment in order to be able to record the power use during an agreed period of time that is likely to be at least one month long. Depending on the activity, different types of measuring equipment may be necessary to obtain the data required for the next step. By analyzing measurement data from the activities, a power mapping is carried out where Tvinn investigates which processes and loads contribute to power peaks during the time period examined. An economic analysis of how measures can reduce the operator's electricity costs is then carried out and a package with suggestions to reduce power peaks and costs is developed and delivered. If the resource owner has proven to be a good candidate for aggregation during the process, i.e. if there is a potential for load control or energy storage, then the resource owner may be offered to become an aggregated resource at Tvinn. The resource then needs to install more equipment for measurement and control before it can be connected to the Tvinn platform and participate in markets for frequency control and flexibility.

(18)

9

2.6.1 Power mapping and power peak shaving

For a long time, energy efficiency has been a topical issue both in terms of climate targets and for companies and organizations who wish to reduce energy costs. By reducing the amount of energy used in different processes, costs can be reduced, but this also leads to less fossil energy being used in the Swedish energy system. In the work on energy efficiency, a key part is energy mapping where processes that use a large amount of energy are identified and usually measures to reduce the amount of energy used are proposed (Duarte, et al., 2015). This study suggests a similar method and approach to investigate power usage. As more and more grid owners and operators include tariffs for power usage in their electricity contracts, the cost of power peaks is increasing and thus also the savings that can be made by shaving the peaks. Also, as previously stated, it is often not the energy but the available power which prohibits further integration of renewable energy. A more efficient use of power may have more benefits than the economical ones, such as the possibility for charging several electric vehicles. The power tariffs are often defined as some cost for the mean power over one hour, monthly. In this report those costs will have the form SEK/kW, month. This means that if some user normally has a low power use of 5 kW over a months’ time but a power peak of 100 kW only one hour the same month, the user will have to pay the power tariff for that 100 kW peak even though the normal use was much lower. Because of this, some electricity users might have a lot to gain financially from reducing power peaks which can be done by first identifying the cause of the peak and then finding ways to reduce it. Therefore, the first step in a power mapping process, similar to energy mapping, is to measure and collect data about the actual power use. Step two is to analyze the data that has been collected and determine whether it is possible to perform power-reducing measures and what those may be.

For Tvinn, the power mapping is also a method for examining whether the electricity user is a good candidate for aggregation, either by controlling flexible loads or by installing an energy storage at the site in question. Peak shaving has proven to be a suitable application for energy storages, especially battery storages since the battery can provide quick response time and variable power for relatively long durations (Staubo, n.d.). The energy required to shave the power peak is equal to the area below the curve of the power peak and it is evident from the power tariffs that short-duration peaks with a high amplitude has the highest cost saving potential. The cost of the battery depends heavily on the energy capacity of the battery, therefore an ideal electricity user has very few peaks over a months’ time with very high amplitudes but short durations. This would allow a relatively small capacity and would not require the battery to be used too often which would cause degradation of the battery and shorter lifetime. In order to quickly estimate what capacity a battery should have to provide a good balance between capacity and the amount of cycles necessary to shave power peaks in some time period a model has been developed during this thesis. The model can be used to build energy systems with different power sources and loads such as the grid, solar energy, EV charging etc. and from the measured power profile provide suggestions for a suitable battery capacity. This is explained in further detail in section 3 and 4.

(19)

10

2.6.2 Role of the aggregator in the energy system

As the share of intermittent energy sources increases in the Swedish energy system and more flexible energy and power sources are necessary, a new role as aggregator of resources has emerged. Aggregation is performed by combining several resources and through IoT solutions control the use of electricity from a central platform. As an aggregator, various services can be offered, such as grid support services to a TSO such as SVK or flexibility services in new emerging markets such as the Coordinet Flexmarket. Battery storages are technically well suited for aggregation and to deliver support services since the response times are relatively fast and both charge and discharge can be used to offer flexibility. There are several reports and studies which suggest using battery storages to support the electricity grid, among them being The royal Swedish academy of engineering sciences (IVA, 2017). The Swedish energy market inspectorate identifies some obstacles, mostly in the form of regulatory and legislative framework which needs to be updated as new technology emerge (Widegren, 2016). An overview illustration of the Tvinn concept with aggregated resources is shown in Figure 4 below.

Figure 4. Conceptual sketch of aggregated resources which can be offered on the Swedish power grid's frequency regulation market or the Coordinet market for flexibility.

The Tvinn platform can aggregate resources of several types such as industries, energy storages and electric vehicle chargers. Resources are divided into groups based on geographic location

(20)

11

as some markets, such as Coordinet's Flexmarket, will split resources depending on the local, or regional grid in which they are located. Thus, it is advisable to group resources by geographical location.

2.7 Brief overview on electric vehicles and charging

2.7.1 Electric vehicles in Sweden

The number of Electric vehicles (EV’s) in Sweden has increased dramatically in a relatively brief period due to electrification and new technology. Figure 5 shows the number of EV’s in Sweden year 2012 through to 2018. During the period presented, the number has increased drastically and is approaching 120 000 EV’s in 2020. Sweden has also seen an increase in the available infrastructure for charging EVs, starting with 505 charging stations in 2012 and increasing to 5,518 stations in 2018 (Statista, 2020). This development is expected to continue and according to the organization Power Circle, Sweden might reach 2,5 million EV’s in 2030 (Power Circle, 2019).

Figure 5. Amount of EV vehicles in Sweden from 2012 – 2020 in Sweden, divided by type of vehicle (Elbilsstatistik, 2020).

The abbreviations in Figure 5 are as follows: BEV - Battery Electric Vehicle, PHEV - Plug-in Hybrid Electric Vehicle and MC - Motorcycle. The greatest increase has been with PHEV’s followed by BEV’s (Elbilsstatistik, 2020).

2.7.2 EV charging

There are different types of charging depending on the type of power, AC or DC, and the amount of power ranging from 3.7 kW to 150 kW. As the technology develops higher charging power has become more common as fast charging of the vehicle is an important factor for many consumers. Today, the most common charge power is 22 kW according to statistics which show

(21)

12

that 44% of AC chargers in Sweden have 22 kW power output (Elbilsstatistik, 2020). The high power used when charging EV’s will have significant effects on power grids both locally and nationwide as power peaks and volatile load behavior become more frequent as the amount of EV’s continue to increase and such a development will require increased system balancing (Engel, et al., 2018).

(22)

13

3 Theory

3.1 Lithium-ion batteries

There are many ways to store energy using different technologies such as hydro storage, compressed air, flywheels and batteries to name a few. In this study the focus will be on electrochemical batteries with Lithium-ion chemistry. There are other commercially available batteries such as lead-acid and flow batteries, but the dominating battery technology today is the Li-ion battery due to its low production price, low maintenance and high energy density (Batteryuniversity, 2020). This section will provide an overview of the Li-ion battery technology.

A typical Li-ion battery cell consist of a negative electrode, often in the form of graphite, a metal positive electrode, an electrolyte and a separator. The flow of Li-ions occurs due chemical redox reactions where during charging the Li-ion flow from the positive metal electrode to the negative electrode and during discharge the flow is reversed. Lithium has an electro-potential of 3,05 V which is one of the highest potentials among all elements. This leads to the high energy and power density of the Li-ion battery and furthermore, Li-ion batteries have a high round trip efficiency and offers a high number of cycles compared to other chemistries such as Lead-acid (Clean energy institute, 2020). Multiple battery cells are often stacked in order to increase the energy content and then forms what is usually called a battery pack. A few common concepts when characterizing batteries are the C-rate, state of charge (SOC), depth of discharge (DOD) and available cycles. The C-rate refers to how quickly the battery is charged or discharged. If a battery with 1 kWh capacity is charged in one hour, it has been charged with 1 C. If it is then discharged in 30 minutes, it has been discharged with 2 C, i.e. it has been discharged with 2 kW of power constantly for 30 minutes. The C-rate is formulated mathematically in equation 1.

𝐶_{𝑟𝑎𝑡𝑒} = 𝑃𝑏𝑎𝑡𝑡𝑒𝑟𝑦

𝑄_{𝑏𝑎𝑡𝑡𝑒𝑟𝑦} (1)

𝐶𝑟𝑎𝑡𝑒– The C-rate of a battery storage system. 𝑃_{𝑏𝑎𝑡𝑡𝑒𝑟𝑦} – The power output of the storage system.

𝑄_{𝑏𝑎𝑡𝑡𝑒𝑟𝑦} – The energy capacity of the storage system.

The SOC provides information of the amount of energy in the battery at a certain time. If the battery has no energy content, the SOC is 0% and if the energy content is at its maximum the SOC is 100%. The DOD provides information of how deep a discharge cycle is, for example 100% if the battery discharges completely. The available cycles are the number of full charge and discharge cycles the battery can go through before the capacity of the battery decreases to some limit value. Often this limit value is 80% of the original capacity of the battery (Sandelic,

(23)

14

et al., 2018). There are several factors which can affect the cycle life, for example the operating temperature, the idle SOC and the DOD. High temperatures, high idle SOC and high DOD are known to reduce the cycle life of the batteries.

Li-ion batteries can have different properties depending on which chemistry is used for the positive electrode material. In Figure 6, four different types of Li-ion batteries are compared based on the electrode material.

Figure 6. Overview of Li-ion battery technologies comparing different parameters (Erdozia & Ferraris, 2017).

The abbreviations in figure 1 are as follows: • LFP- Lithium Iron Phosphate

• NMC- Lithium Nickel Manganese Cobalt Oxide • NCA- Lithium Nickel Cobalt Aluminum Oxide • LMS- Lithium Manganese Spinel

This section will not discuss these four chemistries in further detail, but it can be noted that there are differences, especially regarding safety and cycle life which might affect the suitability of certain applications such as frequency control and peak shaving (Erdozia & Ferraris, 2017).

3.1.1 Li-ion batteries for grid services

Battery technology may benefit the stability of the electricity grid in several ways. Cutting local power peaks, deferring expensive and time-consuming grid investments and frequency regulation are some possible tasks for battery storages. The grid regulations and legislation today are not yet adapted to the new ways of providing these kinds of services. Especially in

(24)

15

Sweden the system is built around the hydro power and structures for pricing are not adapted to battery technology (IVA, 2017).

3.2 Cost of a battery storage system

In recent years, different types of batteries have become increasingly common in different applications such as mobile phones, computers and electric vehicles. Still, the degradation of Lithium-ion batteries is currently not fully understood and is therefore difficult to model correctly as several factors affect the chemical reactions that can occur in the battery cells. This section discusses the costs of a Li-ion battery and some of the degradation processes that will limit the technical lifetime.

3.2.1 Capacity cost for battery storage

The cost of a battery is based on the capacity of the battery and is expressed in SEK/kWh in this report. According to an article from Bloomberg, the capacity cost of cells and the entire battery pack has decreased in recent years and Figure 7 shows that trend (Goldie-Scot, 2019).

Figure 7. Data from Bloomberg for cost per kWh for a Li-ion battery in dollars/kWh.

The trend clearly shows that the capacity cost of Li-ion battery packs fell from 650 dollars/kWh in 2013 to 176 dollars/kWh in 2018. However, the report also states that the price of a Li-ion battery varies greatly depending on who is the buyer and what volumes are purchased (Goldie-Scot, 2019).

3.2.2 Battery storage degradation

In this study, the potential profitability of battery storage for peak shaving and frequency control is investigated. The use of the battery also gives rise to most of the degradation, which is an argument against storing energy using stationary batteries for grid services and other similar applications. To give a deeper understanding of how different factors contribute to the degradation of lithium-ion chemistry batteries, the properties of such battery are described in this part. Battery life can be defined in several ways, but since it is a question of degrading the

(25)

16

battery capacity, it is difficult to set an absolute value for the lifetime. Figure 8 below shows how capacity decreases as a function of the number of cycles used.

Figure 8. Capacity decreases as a function of the number of cycles the battery goes through. The degradation is also affected by the DOD of the cycles (Batteryuniversity, 2020).

In order to maintain as much of the battery capacity as possible, deep cycles should be avoided to a large extent. Batteries are controlled with a built-in software, known as a battery management system (BMS), which regulates the in-and-output of the battery and automatically avoids excessively deep cycles (Relion, 2018). Battery manufacturers often indicate the life of a battery in the number of cycles that can be used before the battery has lost more than 80% of its capacity. As the graph above shows, in theory, a battery used only in the range of 75-25% charge rate can still have 80% of its capacity remaining after about 10,000 cycles while a battery used between 100-25% charge rate has 60% capacity left after the same number of cycles (batteryuniversity, 2020). The degradation that occurs is due to chemical reactions in the battery cells and it is permanent. A cycle is usually defined as charging from 0% to 100% and discharging the battery to 0% again. This study uses the term equivalent cycles, which is one way to calculate the total cycling from several cycles with different DOD. For example, two cycles with a DOD of 50% would count as a full cycle. Equivalent cycles are described in more detail in the below section on the Rainflow algorithm. The speed of the charging and discharge also affects the degradation. Several battery manufacturers believe that it is advantageous to charge the battery with 0.5C and avoid discharge above 1C.

In order to easily estimate the cost of using the battery over a given period of time, this study uses equation 2.

𝐾_{𝑐𝑦𝑐𝑙𝑒} = 𝑋𝑒𝑞𝑣

(26)

17

𝐾_{𝑐𝑦𝑐𝑙𝑒} – The estimated cost of using the battery for a certain number of equivalent cycles during a time period.

𝑋_𝑒𝑞𝑣 – Equivalent cycles during the time period.

𝑋_{𝑡𝑜𝑡𝑎𝑙} – The number of cycles that the battery can go through during its technical lifetime. 𝐾𝑐𝑎𝑝 – Refers to the cost of the battery per kWh.

When 𝑋_𝑒𝑞𝑣 has the same value as 𝑋_{𝑡𝑜𝑡𝑎𝑙}, the cost of the cycles is the same as the total capacity cost for the battery. The cost 𝐾𝑐𝑦𝑐𝑙𝑒 is used in this work to calculate which battery capacity provides the highest savings while using the battery as little as possible to reduce degradation. Batteries also undergo degradation regardless of whether it is used or not, which is called calendar degradation. Calendar degradation occurs as a result of chemical reactions in the cells and how quickly it occurs depends on, among other things, the ambient temperature and the SOC the battery has. Calendar degradation occurs at a faster rate if the battery is kept at a high SOC and high temperatures (Batteryuniversity, 2020).

3.2.3 The Rainflow algorithm

To determine how much the battery is used over some period of time, a method called the Rainflow algorithm is utilized (Alencar & Alencar, 2016). The algorithm was introduced in 1968 to measure fatigue of various materials as a result of repeated load cycles. The method has more recently also been used to count cycles and estimate capacity decrease for batteries. To perform the calculations, a specific package in Python called Fatpack4_{was used and with a}

data series of the battery SOC as the input parameter, the number of cycles and their respective discharge depths ranging from 0 to 1 was calculated automatically. The sum of the cycles then gives the number of equivalent cycles over the time period.

As the name indicates, the Rainflow algorithm can be likened to how water droplets would flow when they meet peaks and valleys. As the first step, the current data series is filtered so only local extreme points remain. Usually the data series is then rotated 90-degrees to better illustrate the intended approach. Figure 9 below shows an example of how the algorithm can presented graphically.

(27)

18

Figure 9. Illustration of how the Rainflow algorithm defines cycles based on valleys and peaks in a data series (Alencar & Alencar, 2016).

A drop of water begins its journey in point 1 and can then only flow to the right to point 2 corresponding to a local extreme point in the data series, the drop then falls vertically to the next level and continue its path. A drop journey ends if it encounters another drop that has fallen from a higher peak, for example from points 3 to 2' or if the drop comes to a point that has the same or higher absolute value as the starting point 1. Examples are point 3 and 6 where the drop journey ends at point 3 because point 4 has a greater absolute value than point 2. Similarly, point 7 has a higher absolute value than point 5 and therefore the route of the drop ends at point 6. In this way, all cycles and their respective depths are counted (Alencar & Alencar, 2016).

3.3 Calculation of payback period

Since the concept which is outlined in this report relies on battery storages to provide services for property and power grid owners, it is of interest to evaluate the payback period for the battery storage. The payback period refers to the amount of time it takes to regain the cost of an initial investment by accumulating the revenue created by the same investment. At some point the accumulation of the revenue created by the investment will be the same as the initial investment and that point is the payback period. The payback period is a good indicator to quickly estimate the profitability of an investment or project. The payback period can be formulated as in equation 3 below.

𝑇_{𝑝𝑎𝑦𝑏𝑎𝑐𝑘} = 𝐾𝑖𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡

(28)

19 𝑇_{𝑝𝑎𝑦𝑏𝑎𝑐𝑘} – The payback period in years.

𝐾_{𝑖𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡} – The initial investment.

𝑅𝑎𝑛𝑛𝑢𝑎𝑙 – The yearly revenue stream created by the investment.

The payback period does not consider that the value of money might change over time. It also disregards that some amount of money at present time is potentially worth more than the same amount at a later time (Kagan, 2020).

3.4 FCR compensation

3.4.1 Compensation on the FCR-N market

Equation 4 is used to estimate the revenue from the FCR-N market. The variables S and A denotes the number of times a bid on the FCR-N market has been sub-ordered (S) and activated (A) during some time period. If a bid is sub-ordered it receives capacity compensation (𝑅_{𝑐𝑎𝑝,𝑛}) based on the amount of power specified in the bid (𝑃_𝑜𝑢𝑡) and if it is also activated, it receives energy compensation (𝑅𝑒𝑛𝑒𝑟𝑔𝑦) for the amount of energy used during the activation period (𝑄_𝑛). However, after the resource is activated it will need to restore its initial state and therefore must buy back energy from the grid which comes with the cost 𝐾_{𝑒𝑛𝑒𝑟𝑔𝑦}.

𝑅_{𝐹𝐶𝑅𝑁} = 𝑆 ∙ 𝑅𝑐𝑎𝑝,𝑛 ∙ 𝑃𝑜𝑢𝑡

2 + 𝐴𝑛∙ 𝑄𝑛 ∙ (𝑅𝑒𝑛𝑒𝑟𝑔𝑦− 𝐾𝑒𝑛𝑒𝑟𝑔𝑦) (4)

𝑅𝐹𝐶𝑅𝑁 – The revenue from trading on the FCR-N market during some time period. 𝑆 – The number of times in a time period a bid on the market is sub-ordered.

𝑅_{𝑐𝑎𝑝,𝑛} – The compensation received for the sub-ordered capacity on the FCR-N market. 𝑅𝑒𝑛𝑒𝑟𝑔𝑦 – The compensation received for the activated bids on the FCR-N market. 𝐾_{𝑒𝑛𝑒𝑟𝑔𝑦} – The cost associated to the energy used for frequency regulating.

𝑃𝑜𝑢𝑡 – The power output of the resource.

𝐴𝑛 – The number of times in a time period a bid on the FCR-N market is activated. 𝑄_𝑛 – The energy supplied during the activation period.

The power output is divided by a factor of 2 since the product needs to be symmetric when offered on the FCR-N market and be able to provide both up and down regulation if activated.

(29)

20

3.4.2 Compensation on the FCR-D market

On the FCR-D market there is no energy compensation for activated bids. The revenue will consist of capacity compensation for sub-ordered bids only, but the energy lost when the resources are activated will still need to be replaced at some point. Equation 5 is used to estimate the revenue from the FCR-D market.

𝑅_{𝐹𝐶𝑅𝐷} = 𝑆 ∙ 𝑅_{𝑐𝑎𝑝,𝑑}∙ 𝑃_𝑜𝑢𝑡 − 𝐴_𝑑∙ 𝑄_𝑑∙ 𝐾_{𝑒𝑛𝑒𝑟𝑔𝑦} (5)

𝑅_{𝐹𝐶𝑅𝐷} – The revenue from bids on the FCR-D market during some time period.

𝑅_{𝑐𝑎𝑝,𝑑} – The compensation received for the sub-ordered capacity on the FCR-D market. 𝑄𝑑 – The energy supplied during the activation period.

(30)

21

4 Material and method

This section provides information on how this study was performed, which tools and instruments were used and how. A description of the two properties which this study focuses on are included as well.

4.1 Measuring power and data collection

4.1.1 Participating actors and data collection

At the beginning of this thesis, several companies and organizations in Uppsala were contacted to investigate whether there was an interest in participating in the study. Three respondents reported an interest in being involved and in the end a collaboration with two companies was initiated. The participants in the study will from here on to be referred to as "property A" and "property B", both properties are located in the city of Uppsala. The owners of the properties will be referred to as "property owner A" and "property owner B". Property owner A and B both agreed to having the necessary equipment installed in their respective property and to have data regarding their power use collected for this study.

4.1.2 Description of property A and B

The two properties differed both in terms of the type of activity carried out in them and the type of electricity contract they had. Property A was located in the industrial area Boländerna in Uppsala and 8 different companies rented space in the building. The property had a main fuse of 120 A for the entire building, but each company had a separate electricity meter from the company Vattenfall AB who owns the local grid. In this study, each company will be treated as a separate load and the total power use for the property is the sum of all loads. In conversation with property owner A and the business owners in the property, it was established that none of the companies had the possibility to control their power use on request since it might impact their respective businesses in a negative way. The property owner wanted to investigate whether it might be economically advantageous to switch to one electricity meter for the entire building instead of dividing the measurement and billing on each company in the building. The electricity contract would then need to be changed and a new electricity meter installed. This question is further investigated in the results section.

Property B was located in central Uppsala and there was only one company in the property. In property B there was no possibility to measure separate loads, but only one meter could be installed on the existing electricity meter from Vattenfall AB. Because of this and the fact that the company's power use was not considered flexible by the owner, property B was not relevant for load control. Property owner B informed that there were plans to install charging stations for electric vehicles on the property. Property owner B wanted to investigate how an energy storage might reduce the power peaks created by EV charging and what savings it might bring.

(31)

22

4.1.3 Measuring equipment and installation

In order to be able to carry out research regarding power use in the properties, data is required on how the electricity usage varies during a given period of time. To this end, a collaboration with the company Enegic5 was initiated. Enegic has developed a power metering device that were deemed suitable for this study as they were non-intrusive, relatively cheap and easy to use.

Figure 10 shows the power meter from Enegic used during the study. A total of 9 power meters were installed, 8 of which were in property A and one meter in property B. In the first case, a Wi-Fi router was also installed since there was no internet connection on the site. A internet connection was necessary to enable data transfer from the device.

Figure 10. The measuring device used in this thesis which was produced by the company Enegic.

The meter consists of a unit with printed circuit board (PCB) which controls the flow of data to and from the unit and a sensor which is installed on an existing electricity meter. Figure 11 shows the meter installed on an electricity meter from Vattenfall AB. The sensor of the meter is installed over the LED which is indicating the power used in the property by blinking with a certain frequency. The sensor is in turn connected to the PCB which sends data via the internet to Enegic's database.

(32)

23

Figure 11. The device installed on a electricity meter on one of the loads in property A.

To retrieve measurement data from the Energy database, an application programming interface (API) was used. An API is a specification for how different applications can communicate with each other and form an interface. For the purpose of this study, functions in Python were created that retrieved the desired measurement data from one or several meters over a certain period and with different time resolution. The shortest time resolution possible to retrieve from the meters was one minute.

4.2 Model for energy systems and simulations

4.2.1 Python programming language

Since the model developed in this study is developed for Tvinn as a possible business, it is of importance that the software and all code used should be open source and permitted for commercial purposes. In this study, the Python programming language was used to write all the code for calculations, graphs and the like. Introduced in 1991 by Guido van Rossum, Python has developed into one of the most widely used programming languages since it is both a powerful and simple language to use (Python.org, 2020). In addition, there is a large library of tools for many different purposes such as automation, machine learning and data management. This work uses several of those tools such as NumPy6, Pandas7 and Matplotlib8.

4.2.2 Open energy modeling framework

Models for energy systems are essential tools to better understand systems that over time have become increasingly complex. In both the planning and operation of different energy systems,

6_{NumPy website:}_{https://numpy.org/}_{(visited 2020-05-25)} 7_{Pandas webstie:}_{https://pandas.pydata.org/}_{(visited 2020-05-25)} 8_{Matplotlib website:}_{https://matplotlib.org/}_{(visited 2020-05-20)}

(33)

24

models and simulations are important to be able to maintain a high degree of efficiency. In this thesis the software developed by the project “Open energy modeling framework” (OEMOF)9 is used to build models of different energy systems that can have a variety of components and conditions. OEMOF is a framework written in Python for energy system modeling, simulation and optimization. OEMOF can be used for both district heating and electrical grids and depends on several packages and tools in Python such as NumPy, Pandas and many others. The framework has been developed in collaboration with several universities in Germany and private individuals. The aim of the project is to create an open and free framework that can be used to easily evaluate different aspects of energy systems. Many other similar tools are not free to use or may be developed for specific scenarios, which can make them difficult to use in other cases. OEMOF addresses these problems through an open, transparent and well-documented framework (OEMOF, 2020). OEMOF uses several packages and libraries to perform modelling and calculations. The SOLPH and PYOMO libraries are used to build and formulate linear or nonlinear optimization problems with several conditions. Each component that is added to the modelled system can have several properties, optimization variables, and by-conditions. For each model, an objective function is created that, depending on the components involved, can contain different variables and conditions. In OEMOF, generally the goal is to minimize the objective function for some period of time (Hilpert, et al., 2018). To solve the optimization problem created from the OEMOF model in this work, the open software

"GNU Linear Programming Kit" (GLPK) is used. GLPK is a python package designed to solve

large-scale linear optimization problems (GLPK , 2020).

Figure 12 shows a sketch of how the model can be built in OEMOF. Components are divided into sources and sinks where, for example, a grid connection can be added as a source and electricity usage for a property as a sink and the connection between them is added as an electricity bus. Sources can only have flow into the bus and sinks can only have a flow from the bus. In OEMOF models, energy storage is a unique component that can have flow both to and from the bus. In the model used in this work, the bus is an electrical bus but there are no restrictions regarding energy carriers and a gas bus can be added to OEMOF models if necessary. The bus is modelled as ideal in this work and thus there are no transmission losses. Alternative energy sources can also be added to the model, such as solar cells or wind power. Surpluses and deficits are handled in the model via a deficit source that must deliver the energy that may be missing in order to balance the system and a surplus sink that receives any excess energy.

(34)

25

Figure 12. Overview of how models are constructed in OEMOF with a few components as examples The line around the alternative power source is dashed since it was not included in the model used in this study, but the feature exists in the overall model which was developed.

OEMOF optimizes for the minimum cost in each time step and thus a cost can be added for the flow for each component. For the model used in this work, the hourly spot price for electricity at Nordpool is used for the grid connection. An average spot price for year 2017, 2018 and 2019 is used as prices since the price for only one year could misrepresent the cost. The cost of using the deficit source and the surplus sink should be significantly higher than all other sources and sinks in order to put them further down the order of priority. The models designed in OEMOF utilizes a so-called "Perfect foresight" approach in simulations, which means that the model is optimized at each time step based on the entire time interval with data. This means that all future events and possibilities are known, and decisions can be made on information which would not be available in a real-life scenario. This approach is quite common in economic models (Bohm & Wenzelburger, 1999) but the results should be interpreted with care since it will provide a perfect scenario where every optimization decision was the correct one at every time step.

Power mapping and aggregation as a service: A techno-economic view on Li-ion batteries for peak shaving and frequency regulation

Examensarbete 30 hp

June 2020

Power mapping and aggregation

as a service

A techno-economic view on Li-ion batteries

for peak shaving and frequency regulation

Filip Angwald

Abstract

Power mapping and aggregation of batteries as a

service

Populärvetenskaplig sammanfattning

Executive summary

Foreword

Table of contents

Glossary and definitions

1. Introduction

1.1 Purpose and goal of thesis

1.2 Scope of thesis and limitations

1.3 Structure of report

2 Background

2.1 Layout of the Swedish electricity grid

2.2 Electricity market and trading

2.3 Electricity contracts and costs

2.4 Frequency regulation in Sweden

2.5 The capacity shortages in Sweden

2.6 Business model for power mapping, energy storage as a service and

aggregation

2.7 Brief overview on electric vehicles and charging

3 Theory

3.1 Lithium-ion batteries

3.2 Cost of a battery storage system

3.3 Calculation of payback period

3.4 FCR compensation

4 Material and method

4.1 Measuring power and data collection

4.2 Model for energy systems and simulations