STOCKHOLM SVERIGE 2020 ,
Power and Energy Quality Optimization of a Battery Validation Facility
A case study of mapping the power profile of a battery cell testing facility
JOAKIM WAHLUND
KTH
SKOLAN FÖR ELEKTROTEKNIK OCH DATAVETENSKAP
Battery Validation Facility
Joakim Wahlund
Abstract—With Northvolt’s aggressive plan to establish Europe’s largest battery factory in northern Sweden, the need to build a pilot plant to demonstrate the possibilities of the company was needed. Quality, performance and cycle-life are key factors customers value particularly high when purchasing batteries, thus, a need for validating these parameters emerges.
Performance and life-cycle tests are performed in high power consuming cycling units divided into the major components - chamber and cycler. The cycler is programmed according to specific charge and discharge patterns (flows).
A validation facility has been established close to Northvolt Labs in V¨aster˚as, housing cycling units for different cell form factors. The major scope of this thesis was to find an optimal way to effectively and energy conscientious conduct operations in the facility. One part included testing the possibility to re-utilize energy from discharging cells in simultaneously charging cells as well as testing if the equipment can provide enough energy to sell back to the electricity distributor.
Additionally, an analysis to see if advanced planning can be used to balance the load profile of the validation facility and reduce energy costs was researched. A second part was to determine if the equipment imposed harmonics on the connected grid, and to decide if a filter is needed for PQ-compensation/harmonic filtering. Estimating the power demand of validation facility was needed to provide the grid owner an energy forecast. Firstly, to be able to analyze these questions, the power profile of the cycling units had to be established through measurements performed by installing power quality devices. It was determined that the chambers consumption was considerably higher than previously assumed and was in fact higher than the average charge-, and discharge-flow performed by the cyclers. A power profile was established for each type of cycling unit which was utilized in a Monte Carlo model to predict the power need of the facility based on new equipment arriving over the next couple of years.
A basic energy cost return-of-investment-calculation concluded that installing an high energy battery energy system was not economically feasible due to prevalent circumstances.
Sammanfattning– N¨ar Northvolt kom till beslutet att etablera Europas str¨osta batterifabrik i Sverige, var det uppenbart att projektet beh¨ovdes genomf¨oras i mindre skala f¨orst, f¨or att demonstrera vad f¨oretaget kan ˚astadkomma. Kvalitetvariabler s˚a som prestanda och cykellivsl¨angd ¨ar nyckelparametrar som st˚ar h¨ogt i anseende hos slutkunderna vid batterik¨op. Detta bidrog till beslutet att uppr¨atta en provanl¨aggning f¨or att m¨ojligg¨ora tester av just dessa parametrar. Kontrollen g¨ors i energikr¨avande utrustning kallade cycling-enheter som delas upp i huvudkomponenterna cycler, kammare och transformator.
Cyclern programmeras utefter upp- och urladdningsfl¨oden f¨or att kunna best¨amma cellernas prestanda och livsl¨angd.
Provanl¨aggningen som uppr¨attades i n¨ara ansluting till Northvolt Labs i V¨aster˚as, har ett stort antal av denna utrustning f¨or att kunna genomf¨ora tester p˚a battericeller av olika storlek och format. Det huvudsakliga m˚alet med denna
uppsats var att hitta det mest energisn˚ala s¨attet att utf¨ora den dagliga verksamheten i anl¨aggningen. En av delarna inkluderade att unders¨oka m¨ojligheten att ˚ateranv¨anda energi fr˚an celler som laddas ur, f¨or att i real-tid ˚aterinf¨ora den energin i celler som laddas upp. Vidare unders¨oktes det hur mycket energi som kan s¨aljas tillbaka till n¨atet under urladdningsoperationer.
Forts¨attningsvis s˚a genomf¨ordes en analys f¨or att se om avancerad testplanering kan utnyttjas f¨or att balancera byggnadens elektriska lastprofil och reducera elkostnaden. En andra del var att best¨amma om testutrustningen ˚aterinf¨orde harmoniska signaler p˚a eln¨atet och i s˚a fall om filtrering kr¨avs f¨or att f˚a elkvaliten till till˚atna niv˚aer. Estimering av det totala effektbehovet efter framtida expansion av testanl¨aggningen var ett ytterligare syfte med detta projekt. F¨or att g¨ora dessa analyser kr¨avdes en tydlig energiprofil av utrustningen som erh¨olls genom olika prestandatester efter installation av m¨atinstrument f¨or elkvalitet. Analyserna resulterade i att energikonsumptionen hos kammrarna var avsev¨art h¨ogre ¨an vad som antogs vid projektets b¨orjan. Faktum var att energikonsumptionen var s˚a mycket h¨ogre att det ¨overskred konsumptionen hos cycler-enheterna under de vanligaste testerna. En effektprofil skapades f¨or varje typ av testutrustning i anl¨agngingen. Med hj¨alp av Monte Carlo-metoder skapades en modell med syftet att f¨orutstp˚a framtida energibehov, baserat p˚a energiprofilen. En enkel kostnadsber¨akning gjordes ang˚aende installationen av ett batterisystem d¨ar slutsatsen kunde dras att det inte var ekonimiskt f¨orsvarbart under r˚adande omst¨andigheter.
C
ONTENTSI Glossary 3
II Introduction 3
II-A Background . . . . 4
II-B Problematization . . . . 4
II-C Purpose and research questions . . . . . 5
II-C1 Power Consumption & Demand . . . . 5
II-C2 Energy Quality . . . . 5
II-D Scope and Delimitations . . . . 5
II-E Disposition . . . . 6
II-F Related Work . . . . 6
III Power and Energy Quality 6 III-A Reactive power, Harmonics and Distortion 6 III-B Machine efficiency . . . . 7
III-C Monte Carlo method . . . . 8
IV Case & Methods 8
IV-A Grid Operator and Local Business
Landscape . . . . 8
2
IV-A1 Power costs . . . . 8
IV-B Battery Cycling Machines . . . . 9
IV-C ESS . . . . 10
IV-D Cells . . . . 11
IV-E Cyclers . . . . 11
IV-F Chambers . . . . 11
IV-G Life-cycle test . . . . 11
IV-H Measurement setup . . . . 12
IV-I Cycler internal utilization of discharge power . . . . 12
IV-J Consumption and quality testing . . . . 12
IV-K Operation of Chambers . . . . 14
IV-L Operation of Cycling units . . . . 14
IV-M Monte Carlo methods . . . . 14
V Results 15 V-A 200A Cycler test outcomes . . . . 15
V-B Chamber power characteristics . . . . . 16
V-C 400A Cycler test outcomes . . . . 16
V-D Power need estimation . . . . 16
VI Discussion 17 VI-A Cycler Unit Consumption . . . . 17
VI-B Cycler characteristics . . . . 17
VI-C Monte Carlo outcomes . . . . 18
VI-D Cost reduction from BESS implementation 19 VII Conclusions 20 VII-A Power need . . . . 20
VII-B Power quality . . . . 22
VIII Future work 22 IX Ethics 22 IX-A Supply-chain . . . . 22
IX-B Energy consumption . . . . 23
References 23
I. G
LOSSARYAbbrevation/Term Explanation
BESS Battery Energy Storage System
HE-BESS High energy BESS
HP-BESS High power BESS
PV Photovoltaic
C-rate Current-rate
SOC State of charge
MCM Monte Carlo method
DSO Distribution system operator
SWF Swelling force
SWFMC Swelling force module cycler
OPEX Operating expenses
CAPEX Capital Expenditure
II. I
NTRODUCTIONE NERGY has been, and will continue to be a topic of discussion for as long as mankind strive to improve the way she leads her life. With the ever increasing levels of CO2 and other greenhouse gases in the atmosphere, which are causing record breaking heatwaves across the globe [1]
nearly every single year for past decades. Companies and governments alike see the need to improve their environmental impact by putting effort into curtailing emissions and improving technology and processes that otherwise end up having adverse repercussions on the global climate. The Paris Agreement [2] is only one of the treaties around the globe where countries has come together in a joint effort to combat the global temperature increase. The European Union for instance, has several plans for its member nations’ to reach sustainability goals by 2020, 2030 and more [3].
In recent years, there has been an increasing interest in clean energy and especially in Electrical Vehicles (EV) and Battery Energy Storage Systems (BESS) that can be implemented to de-carbonize the major energy sectors across the globe. The opportunities have consequently been reported by several main stream media outlets [5] and [6] and many observations points towards its continued rise across several industries, due to forecasts pointing towards lowered cost of storing energy while at the same time being able to reach chemistry’s with a higher energy density, in the coming decade [7]. Results clearly shows that companies that are supporting the energy transition such as Northvolt AB, with its immense, record-breaking capital surge from large and established corporations such as ABB AB, Volkswagen AG, BMW Group, Siemens and more [4], will play a huge role creating green batteries in a sustainable fashion to enable the use of unpredictable energy sources and push the fossil–heavy vehicle market to a more EV-centric fleet in the coming decade. Due to Northvolt’s rapid growth and planned scaling in production size, new challenges arise, including power demand for high energy consuming battery validation facilities in terms of available power in the area, as the region is already close to the maximum amount that the power infrastructure can currently deliver.
Fig. 1. Northvolt’s prismatic and cylindrical cells. [4]
Fig. 2. Northvolt’s planned 32 GWh factory in Skellefte˚a - Sweden. [4]
A. Background
Northvolt AB (NV) has had one of the most impressive journeys among start-up companies in the history of Europe.
With former Tesla managers spearheading NV’s advance, the company has had the vision of reshaping the battery market in Europe in their sights since before founding the company in 2017. To be able to reach their goal of producing 32 GWh worth of batteries every year, NV started planning their magnum opus - the World’s 3rd largest building, an enormous battery factory located in Skellefte˚a, northern Sweden, in close proximity of 100% green energy producing hydro plants to be able to produce the most sustainable batteries to date [8]. Not stopping at 12M euros, NV has steamed ahead to break the record of highest investment into a new company that Europe has seen - closing an equity capital race at 1B euros in funding from partners and investors such as Volkswagen AG, Goldmann Sachs Merchant Banking Division, BMW group, AMF, Folksam group and IMAS foundation.
Starting with a pilot factory in V¨aster˚as, a one hour drive from Stockholm Sweden, relatively small (if comparing to the gigantic factory in Skellefte˚a), but nevertheless a production scale building is currently taking form, for the purpose of demonstrating the manufacturing process. At this site, 350 MWh of cells will be produced annually in the
process of the development of new chemistry’s emerging from the R&D department located nearby. Early on the need for a validation department was seen. Here cells of all produced form factors would be tested for performance and lifetime. An adjacent facility from a proprietor called Kungsleden was rented in a medium large industrial area called Finnsl¨atten in V¨aster˚as. In this building a validation team has been put into operation. The validation team’s main function will be that of testing the cells and making sure that they are up to par with both customer specifications and NV’s quality assurance guidelines.
The equipment used in the validation facility is used for, as the name suggests, to validate the performance of the cells being created in the R&D laboratory as well as batch-testing of cells created in the larger factory. Customer’s will provide their custom-made driving patterns henceforth called flows, of how the batteries will be used in their applications to the validation team in order to establish the cell’s performance are aligned with the customers intended cell use. This includes, charging, discharging, resting and aging of the cells.
B. Problematization
Large industrial companies in the manufacturing business
consumes a large part of the national consumption of energy
within the country it operates e.g. 25% of the total energy use in Sweden 2018 is used for industrial purposes [9].
Planning for a large facility which a huge power consumer takes on new challenges when the building’s area coincides with an area that is already near its power capacity [10].
Analysis is being conducted to see how the machinery affects the power grid, and a possibility has been seen that some of the equipment are inserting energy with poor quality when discharging battery cells, considering the sheer size of the planned equipment installation and one of the major incorporated components in the equipment being inverters.
The Property Owner (PO) controls and operates the 6 kV power grid which the validation facility (VF) is connected to. Additionally, they supply all businesses in Finnsl¨atten with power. With NV establishing in the area, the total power the PO has to provide will increase severely. The way large industrial energy importers choose to conduct their energy purchasing business is to, as an example buy 85% of the power need for three year periods in advance using energy contracts to be able to cut costs. In order to do this optimally power consumption data is crucial. Companies residing within the local grid studied in this thesis are both well established businesses who has conducted their operations there for in some cases, decades, where definitive and thorough historical data of their energy consumption and quality exists, as well as companies who has a power need that has a low impact on the quantity of purchased power and energy by PO. In contrast, NV is constructing buildings with needs of several megawatts, which as a consequence has substantial repercussions in the PO’s decision-making.
Northvolt is expanding a part of their validation facility to commission-complete status following an incremental installation plan, where day-to-day operations are still far from where it is expected to be in late 2020 in terms of cell batch sizes for testing. Estimations of how much power is required on average (as well as peak) in the coming year is thus needed. Predicting an accurate estimate with very limited information or historical data is troublesome.
Including a very slim allowed margin of error in the prediction further increases the challenges with providing accurate outcomes. In addition that providing an outcome that is not in the range of the allowed margin will increase OPEX and this is an additional purpose of why a thorough case study is needed.
The need to keep the momentum is part of the company modus operandi. During this start-up phase, when executing important work at a high speed in order to be able to reach milestones is very important, certain tasks will be neglected.
Both due to less time performing due diligence and researching vendors for equipment purchasing. This has caused unaccounted for critical parameters in the equipment technical specifications, most relevantly power quality and energy consumption limits. The need for power testing is thus apparent in order to plan for electrical installations, budgeting and providing PO a energy forecast to reduce the
risk of NV having to pay more than needed.
C. Purpose and research questions
The primary focus of this thesis is to characterize the cycling unit energy consumption and power quality profile, in order to provide aid in project decisions regarding the expansion of the testing facility, in terms of energy purchasing, dimensioning of cables and other electronic equipment. In addition to this a basic economical analysis of the possible cost reduction will be conducted, facilitated by implementing alternative energy solutions such as energy storage solutions to the facility.
1) Power Consumption & Demand:
1) How much power will the facility consume in the planned expansion?
2) How much of the peak power can be shaved by either implementing an BESS-system to the facility or effectively planning when cells are charging and discharging?
3) Can an battery energy storage system (BESS) be implemented to reduce energy costs in the long term?
2) Energy Quality: Quantifying the quality of the energy being produced during discharge is critical to understanding how the facility will operate as a complete unit. The following questions has been formulated to answer this.
1) Is there a need to improve the energy quality of the cycler unit in terms of harmonics imposed on the grid?
2) Is there a need to compensate for reactive power consumption in the machines due to the non-linear characteristics of the inverters inside the cycler?
3) How efficient is the cycler at converting its AC input power to DC power used for charging the cells?
4) How efficient is the cycler at utilizing the energy discharged by the cells?
D. Scope and Delimitations
The path is currently well paved in terms of Northvolt’s expansion, by fall next year the load of the facility is expected to increase ten fold, thus exceeding the current power capacity by megawatts. Cases could be made for three solutions
1) Increasing the power capacity by installing new cables or otherwise altering the power infrastructure connected to the facility.
2) With advanced planning of scheduling charge/discharge- tests to reduce peak load during the day
3) Introducing ESS systems which help in shaving power peaks as well as reducing overall energy costs in the long term.
Thus, a main purpose for this thesis is to propose solutions to Northvolt on how to first and foremost establish the power need of the equipment and secondly to tackle these issues, in addition to investigating the need for power filtering.
The research questions can be attacked from several angles.
With various levels of detail and specificity, hence the need to establish an approachable scope considering this thesis time constraint. Distinct delimitation’s and restrictions has to be formed. The power characteristics of the cycler unit could be investigated down to component level. Instead the cycler units’ will henceforth be considered a black box. With inputs of AC power and outputs such as DC power to charge and discharge cells, losses generated by the power electronics and auxiliary devices within the unit i.e, heat and base-load consumption during unit stand-by mode. The cycling unit will be divided into two separate equipment units to be able to easier identify characteristics between the the cycler and the chamber. In order to determine the total effects the equipment has on the grid, a measurement setup would be needed at the grid connection, this is not possible during the thesis, hence creating the inability to study the effect the VF’s operations has on the grid. Instead the power quality will be investigated on equipment level.
Even though the VF’s room’s facilitates more than one type of form factor, the majority of the energy will be consumed by the prismatic form factor cycling units. Hence the model being built based on test being run on prismatic cell cycling units instead of cylindrical. However, assumptions will be made in regards to the consumption of both cylindrical cell equipment as well as utilities such as lighting, air conditioning, liquid-cooling etc. To be able to provide an overall estimate for the entire facility.
Conducting a profitability analysis of installing an ESS could be the entire scope of a thesis project. Since this is not the major aim of this thesis, a basic investigation will be made in terms of evaluating the average cost savings from charging a fixed capacity battery energy system during the night, to be discharged during a peak load hour in the middle of the day.
E. Disposition
•
Introduction - The thesis begins with establishing the background of Northvolt and the relevant areas included in the scope of this project. It delves into why the work is needed and lists the problematization in a quantifiable manner.
•
Power and Energy Quality - The stakeholders in the power chain are presented and a basis of electric theory and characteristics of the cycling units, are presented in addition to information regarding implementations and limitations of possible solutions.
•
Methods & Case - Test procedures and measurement installation setup are given to connect the problematization with how the research question’s outcomes can be evaluated. Furthermore methods of forecasting the energy consumption is discussed.
•
Results - Data from measurements combined with graphical interpretation of the test results are presented in this section.
•
Discussion - Here findings and why results lead to certain outcomes are being discussed.
•
Conclusions - This chapter explains the conclusions drawn from the results.
•
Ethics - Ethical challenges within the thesis project are discussed.
•
Appendix - Abbreviations and terms are explained and additional data is presented
F. Related Work
To date, there are few studies that have investigated the power need of a facility, which could be described as being internal investigations within companies. Previous research exists of cost-and energy-effective methods of production planning, both on factory scale level [11], [12] and [13] as well as individual machine level [14]. Naturally there has been extensive research put into energy quality during the past decades as can be seen in [15] and more recently from a renewable power generation point of view, e.g. [16].
Resources has previously been allocated internally within NV to account for the earlier mentioned challenges with a focus on how to deal with the upcoming power demands.
Since infrastructure projects span several months at a time and, thus, hinder Northvolt’s advance in the expansion of their validation facility, estimates and predictions had to be made earlier. Internal investigations within NV have been conducted which has acted as a baseline for the work in this thesis. The work stands on testing equipment nominal values, ventilation and auxiliary power needs and assumptions in regards to needed DC power of the battery validation tests. Basic assumptions has been used which will be evaluated in the scope of this thesis.
One key assumption that was made during the previous research mentioned above, was that DC power from discharging cells could lower the overall power level of the facility by 25%.
III. P
OWER ANDE
NERGYQ
UALITYPower quality is a term that in this context is used for describing the overall quality of the electricity in regards to voltage levels, frequency disturbances and imposed harmonics.
A. Reactive power, Harmonics and Distortion
Reactive power is an important component in the power system, and can both serve as a solution in some cases but as a problem in others. Reactive power stems from the phase difference between the voltage ¯ V and current ¯ I. Not to be confused with the reactance X which is the imaginary part of the impedance ¯ Z = R + jX which gives the phase difference.
Imaginary part of the complex power in a three phase power system and develops due to several reasons. One of these reasons are due to the inductance. If there exists inductors such as coils the inductance will increase and affect other phenomena such as losses and voltage levels.
Reactive power occurs when there is a difference in phase
angle between the voltage and the current. Inductive loads
cause the current to lag the voltage and there is a discrepancy between them thus consuming reactive energy.
In contrast, capacitive loads such as conductors or even inverters cause the current to lead the voltage which from a electrical load’s perspective produces reactive energy
Furthermore reactive power is also evident when the current is irregular in amplitude over a time-period. Grid owners often enforce a maximum percentage of reactive power to active power being either imported to the grid, or exported to a load.
This is done to aid in balancing across the entire system.
The complex power is expressed as
S = P + jQ ¯ V A (1)
Whereas the apparent power
| ¯ S| = p
P
2+ Q
2(2)
And active power
P = | ¯ S| cos(φ) W (3)
Where | ¯ S| is the apparent power, P is the active power and Q the reactive power.
Q
50Hz= S sin(φ) V Ar (4)
There are two types of reactive power that can be distinguishable. reactive power fundamental (Eq. 4 and reactive power distortion (Eq. 5) henceforth simply called distortion power. Together they will make up for the difference between the cos(φ) and the power factor which is explained in [17]. Electricity consumers are often contractually obligated to maximally import or export reactive power at a certain percentage of the total active power in order for the DSO’s to be able to keep the system balanced [18].
Measurements will not be conducted on a building level, which is needed to establish the power factor and to see if the facility is sending harmonic currents back to the DSO.
The reactive power caused by current harmonics is also called the distortion power. Distortion in a power system is closely related to current harmonics within the system. When discussing the reactive power caused by the phase shift between voltage and current, it is reasonable to distinguish between the reactive power at the fundamental frequency i.e.
50 Hz and the reactive power created due to deviation in phase angle originating from harmonics at other frequencies than 50 Hz - called distortion power. [17]
D = q
S
2− P
2− Q
250HzV A (5) Irregularity in current is rather common, especially when current is passing non-linear loads such as inverters.
Harmonics in a power system are induced from currents from non-linear loads that are oscillating at frequencies that are integer-multiples of the system fundamental frequency (50 Hz in this case). Harmonics in the system can cause over-heating, distortion, losses and other phenomena that has a negative impact on the system [19]. Effectively filtering harmonics to reduce the risk of polluting the power grid is
crucial (depending on the amplitude of the harmonics in question) to improve power efficiency, mitigating additional power costs of consumed reactive power and to improve the lifetime of power electronics.
Three phase transformers can to some degree filter harmonics depending on the connection type between the primary and secondary side, but the fact remains that the harmonics will induce losses in the windings as well as the core of the transformer [20] which ages the transformers, which for instance is shown in [21]. Furthermore [21]
explicitly states that harmonics can cause aging of the components. The wave-form of the current gives the total harmonic distortion (THD) by utilizing Eq. 6 where the subscript of the currents used in the equation is the currents at that order of harmonic. I
1being the current at the first order, which in the case of Sweden is the current at the system frequency - 50 Hz. I
2being the current at the frequency of 100 Hz, I
3at 150 Hz and so on. Harmonics increase the currents in the system and the 3rd order harmonic is especially damaging to power components since it increases the current in the neutral conductor which leads to more rapid aging in insulation of grid components due to higher temperatures in the conductor, as well as reducing efficiency of generation equipment, and energy transport.
I
T HD= pI
32+ I
52+ I
72+ I
92+ I
n2I
1(6)
B. Machine efficiency
A cycler’s efficiency of utilizing inserted AC power to charge cells under testing with DC power is analyzed.
The cyclers are separated with their respective chambers to be able to distinguish the power going in to the cell versus the power consumed by the cycler. Since the chambers are essentially a combined oven and freezer it will always consume a certain amount of energy, in the best of worlds it would keep the temperature within the machine with a very low input of AC power to the cycler itself.
The power parameters of the cycler is assumed to consist of parts seen in Tab. I.
% of total power PAC
DC power to cells Pcells 85%
Base-load Pbl 10 %
Losses Plc 5%
TABLE I
ASSUMED POWER PROFILE OF THE CYCLER,WHEREPACIS THE MEASURED POWER OF THE CYCLER.
The base-load will be the power the cycler requires to run
auxiliary components such as PC, OCM, control system,
fans etc. This is a factor that has a high likelihood to heavily
affect the energy sent back to the grid when discharging
cells if it is too high. This is of course depending on how
the equipment is being run. If few cells are being charged,
the base-load power will be a larger percent of the total
consumed power of the cycler.
The AC to DC efficiency during a charging operation of the cycler will be defined as
Pbl Baseload power consumption
Pcells Total DC power the cells are being charged/discharged with
PAC Measured AC power
TABLE II
NOTATIONS USED IN EFFICIENCY EQUATIONS
η
cycler−charge= P
cellsP
AC(7) The P
ACis defined as positive when the machine is consuming energy and negative when the machine is producing energy.
The discharge efficiency is thus
η
cycler−discharge= P
ACP
cells(8) Excluding the base-load consumption of the cycler an efficiency (Eq. 9) of the inverters and other power electronics not included in the base-load consumption will be found.
η
cycler−inverter= P
AC− P
blP
DC(9) Where P
blis the base-load consumption of the cycler.
The cycler’s efficiency of utilizing the discharged energy from the cells and insert it into the grid is defined as
η
c−dCh= P
ACP
DC(10) And analogously to Eq. 9, removing the base-load consumption of the cycler during discharge will yield an internal efficiency of the cycler.
η
c−dCh= P
AC− P
blP
DC(11)
The losses of the cycler will be defined as
P
lc= P
AC,cy− P
DC,cy− P
base−load,cy[kW ] (12)
C. Monte Carlo method
IV. C
ASE& M
ETHODSThe validation facility partly exist to validate new types of experimental battery designs developed at the R&D facility and see how they perform under operation and in certain environments. The result of what is happening at the validation facility will heavily affect cell development procedures from the NV R&D and Northvolt Labs in the future.
A. Grid Operator and Local Business Landscape
Northvolt Labs breaking ground where it did - in the industrial R&D-prominent part of V¨aster˚as called Finnsl¨atten, was among other reasons the already existing infrastructure in terms of power supply and near vicinity to partner companies, such as the electrification giants ABB.
Moreover V¨aster˚as is relatively close to the capital, Stockholm, making travel connections excellent.
Finnsl¨atten’s commercial inhabitants, where ABB Machines is a heavy consumer, are connected to an industry power grid operated by PO. Apart from owning and being responsible for the power grid in the area, PO also owns a large portion of the buildings which the established companies are renting. This entails a rather complex business-political landscape in terms of how power is being bought and distributed when a new player joins the game, especially since the establishment of NV Labs, which will according to the current prognosis contend with the front-runners in terms of highest power need in the area.
The stakeholders in the current ladder of power for the validation facility, panning across all levels, from production, transmission and distribution down to consumers provokes a bureaucratic process that could be especially lengthy. Of particular concern is that NV does not buy their own power directly from a distribution system operators such as E.ON, Vattenfall or M¨alarenergi, but has to go through the PO to get their needed energy. The concern is however more business related in terms of processes that each stake holder has to go through in order to come to a decision based on the need of every entity in the chain, including environmental permits and other needed documentation to perform construction work. Navigating through this business landscape is not a part of the scope of this thesis, but is considered as a foundation for the problematization regarding the power need.
1) Power costs: The major costs of energy consumers operating at a steady power level are the active power import and the power subscription. Customers always pay for the energy that they have consumed, at a price that is determined by several factors such as if power has been bought beforehand using power contracts or the day-ahead markets, or if it is bought at an commonly more expensive rate during the spot-market auctions.
In general the power subscription cost is related to how much power a consumer would not excede at any point in time. If they surpass the power subscription limit an added cost would be added to compensate for the producers having to supply more energy in real time than what they had planned to according to the agreed upon terms with the consumer.
Importing or exporting large quantities of reactive power to
the power grid will cause volatile fluctuations in voltage
level in the grid which increases resistive losses and
generates a need to compensate in other parts of the grid.
Generally speaking, consumers are allowed to import or export 5% of their subscribed power in reactive power from, or to the grid. Exceeding this limit will induce an extra cost per imported/exported kVAr above the limit which is relatively high. The same consequences are applied for active power, exceeding the power subscription more than two times per subscription length will yield fines of 1.5 times the cost of the difference between your originally agreed upon subscription and the average between the power peaks of the two times the subscription was exceeded. The more power a customer subscribes to, the more expensive it will thus be to exceed it. Contrary to exceeding the limit, there is also a risk of overestimating the subscription amount. Obviously it costs more to subscribe to a higher power level, meaning if a distributor reserves an amount that a customer does not reach, the customer still has to pay for that power subscription cost. In the case of this thesis, the business landscape is not as simple as just explained, since Northvolt purchases everything in the facility through the property owner.
B. Battery Cycling Machines
Building a new company that heavily relies on external investors puts a huge demand on the company image. When it comes to battery manufacturers it is especially critical to not only deliver cells according to agreed upon specification but to at all costs produce batteries that the highest level of safety characteristics. Moreover, customers demands assurance that the batteries are functioning to their specific applications, for instance as energy storage in modern EV’s, where currents are high both during discharge and when the car is stationary during charging - safety is of key importance [22]. The customers sends their estimation of how a user would use their product and how the battery parameters are typically fluctuating. This profile is then inserted into the cyclers which simulates the real-world circumstances of the batteries.
The battery cycling machines (See Fig. 4 operates at different currents. Each cycler has the ability to perform its operation on many cells at once by placing them in so called channels. Each cyclers has a set amount of channels and thus it is possible to examine how much power is needed per channel. NV has performed a preliminary analysis into the power need during each ramp-up phase until reaching the final goal of reaching the middle of 2020 expansion goals.
Since there is a plan of action of when these rooms will be set to use, and what types of tests will be run in the machines over the course of the next couple of years - it is possible to forecast the power consumption in the foreseeable future with a rough certainty. This model was based on the DC power required to charge cells, and had a fairly reasonable assumption that the chambers would require 15% of the power, to keep the ambient temperature at 20 °C, of what a cycler will use during a test on average. Battery cyclers are high-consuming electronic equipment which
requires vast amounts of energy to charge the numerous cells needed according to the project development goals seen in Fig. 3. Where the difference between samples are the maturity of the product. The earliest prototype is the A-sample, where goals are set lower than the final product.
In order to reach a ready C-sample that is ready for large scale industrialization large quantities of tests are needed.
All operations that the cells are subjected to occurs within a so called cycler chamber. The chamber can be set to different specific temperature to test the cells under several common conditions, from the cold weather in Sweden to the hot and tempered environment in more southern places of the world. The cycler chamber is being controlled separately from the cycler. The cycler contains the electronics and software that performs the actual charge and discharge in the channels within the chamber that are filled with cells, while the chamber only controls the heating and cooling of the chamber, by the use of compressors and resistive heating elements. Throughout this report, the term cycler unit will be used to refer to the battery cycling machines that performs performance and lifetime tests on battery cells for validation purposes. The cycler unit can be divided in to its main components, the cycler, the chamber and the transformer.
The chambers exist in different sizes in terms of how many channels they fit. The channels are slots where a cell can be placed. One of the analyzed chambers can most house 16 individual prismatic cells as can be seen in Fig. 4. There exists several different form factors of cells, in this thesis the primary focus will be put on prismatic cells of different capacities. They all have their own form factor but the same shape, namely prismatic cells. The size scale with the capacity. The differing capacities of the cells means different amounts of current to be able to test the same operational levels on the cells. Central to the entire discipline of battery design and manufacturing is the so called C-rate. The C-rate is the current rate and is different for batteries with different capacities. For instance the 1C of a X Ah battery means that the current it is discharging or charging at is X A. 0.5C would then be X
2 A. Then it is easy to compare that in order to run a battery at for instance 5C would require a lot more current if a cell has the capacity 5X Ah, than if it would have 1X Ah . This induces a need for cyclers that can handle these high currents for high capacity cells, and is why there are several types of cycling units, distinguished by their highest current they can input to a channel.
The channels in the cyclers all have their separate inverters,
and as is known, inverters are not linear loads, which
inherently causes harmonics above the fundamental
frequency in the power system. The chamber is typically
powered by a three phase 400V AC 50 Hz supply, but it
depends on which part of the world the machines are
operated. A compressor operates the cooling and a heating
element the influx of hot air to the chamber, where the
operational range is -40 °C to 80 °C.
Time A-sample
B-sample C-sample
Product development
Product development
Product validation
Product validation
Process
tuning Process
validation Start of Production
Fig. 3. Simplified process of a typical cell development
Chamber Cycler
400 V
UMG104
UMG605
Transformer
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Power module
Temperature control
Fig. 4. Cycling unit setup with measurement points. See Sec. IV-H for further explanation of the power measurement devices UMG605 and UMG104.
C. ESS
In recent year battery energy storage systems (BESS) has stepped forward from the shadows of the custom application business to emerge as a true competitor in the market of many commonplace applications, an extreme example is for instance the one installed by Tesla in Australia [23]. In places where blackouts or brownouts are common, having an integrated BESS in the grid system is an environmentally friendly way of providing fast-response power and energy to the grid in question instead of using e.g. diesel or natural gas generators.
It is not only an efficient way of storing energy for areas with numerous blackouts that such as third world countries where the power supply is unreliable, it is getting more common in developed countries where the grid capacity has reached a limit where power peaks are imposing high amounts of strain on the grid components and thus the TSO’s and DSO’s has to raise electricity prices to levels far above average to compensate for the damage to the grid.
Hence more and more industries, municipalities and other entities are investing money into BESS solutions where they
can utilize the cheap electricity prices during low-demand hours and this during the high-demand peak hours. this increase is largely due to the ever decreasing costs of buying end-user battery solution, of course an outcome of decreased prices in producing battery cells for example. Studies of BESS-solutions for industry applications has shown the benefits of peak-shaving for cost-minimization purposes, e.g.
Oudalov et. al. studied the theoretical optimal sizing for bulk battery systems already in 2007 [24] and even earlier research has been conducted in the 90’s by the likes of Miller et. al. and more. [25] who furthermore has been a part of commissioning a 5 MVA - 2.5MWh at a battery recycling plant in Vernon, California [26].
Different types of BESS currently exists, two different
characteristical types are high energy and high power, where
the high energy are used in energy management scenarios
where the BESS has to output energy for a time range (of
around 1 hour). In contrast, the typical usage for high energy
BESS (HE-BESS) are for peak-shaving, facilitating
integration of unpredictable energy sources such as
photo-voltaic (PV) systems or wind farms in addition to
being a solution for island operations without a proper
connection to a regional grid whereas high power BESS (HP-BESS) are commonly used to compensate voltage dips, reactive power or flickers on the demand-side [27] [28].
The system design of industrial and commercial BESS-solutions alike are very similar. The bulk battery pack is connected to a PCS, a power conversion system converts the variable DC voltage of the batteries to a three phase AC voltage used by the majority of machinery and utilities in industries.
Finding optimal size and usage of a BESS for a facility can be a thesis project in and of itself, therefore a more basic approach is employed in this project. For economical analysis a HE-BESS with a capacity of 25% of the facility’s average hourly power need is used. The investment cost is estimated to be 1.4 million euros. As an example the HE-BESS would have a 250 kWh capacity if the building has a 1 MW power level during a typical hour. This would entail that one quarter of the operational power need of the facility could be sustained for one hour during an probable blackout event. To put this into an example; the power need of the facility is assumed to be 2 MW on average continuously throughout the day and night and over the whole year, which is not far off from the reality since tests will run for several weeks or more. The main idea is to reduce energy costs but value could be arguably added in other aspects, such as presenting proof of concepts.
Energy cost reduction is made possible by charging the ESS during night when the prices are somewhat lower. and utilizing this during the maximum peak hour of the day, since there is only enough to sustain 25% of the operations as mentioned in the previous paragraph, there would only be a reduction of 25% during normally priced peak hour of the day.
D. Cells
Manufacturing batteries requires the ability to provide fully functional and beyond reproach performance of cells according to customers demands. In this case there are three different types of cells, each with a different capacity and physical size, each cell belong to a separate customer project. The cells are undergoing several validation tests to confirm its operational status, performance degradation over time and capacity checks.
Certain projects are going into specific types of cycling units with different internal current ratings.
Project Capacity [Ah] % of total
A x 25
B 1.5x 50
C 3x 25
TABLE III CELL CAPACITY PER PROJECT
E. Cyclers
The cyclers are aptly named after their purpose, that is to cycle cells to be able to measure the decrease in capacity and performance over the course of its life time, and furthermore how fast the cells can charge or discharge. Currently there are several types of cyclers in the completed fire cells. The cyclers are fed through an 400/220V 50Hz three phase transformer in the current setup but this could be changed in the future. Due to operating several types of cyclers with limited measurement possibilities it was necessary to establish a possible linearity between the types of cyclers in order to extrapolate data from one measurement to another. i.e, if a certain current in a 200A machine will correspond in a linear fashion to a test running at another current in for instance a 400A cycler.
F. Chambers
The chambers are the cabinets that house the cells during testing. They are able to regulate the temperature within themselves in a range of -40 to 80 °C. The chambers employ compressors and a propylene glycol solution for removing heat from within the chamber, and a resistive heating element for heating. To keep the temperature even throughout, fans are used to distribute the air inside the unit.
The normal operation of the chambers are predicted to keep the internal temperature at room temperature i.e., 25 °C. In this investigation, a fully utilized chamber can fit 16 cells inside.
G. Life-cycle test
One of the key tests and major components of the testing schedule are life-cycle test. It exists to measure how and if the cell degrades internally when subjected to a long period of many charges-, and discharge-cycles. A life-cycle test can typically consist of performing charge and discharge cycles in the order of hundreds, at a fairly low C-rate of 0.5. After a specific time a capacity check as well as a internal-resistance check is performed in order to measure the capacity of the cell after having gone through the extensive cycling. This process is then again repeated numerous times to gain insight into the cell’s quality after many uses. Since cycle-life is a key part when discussing batteries it is truly of the essence to have a high cycle-life in modern Monte Carlo batteries. cycle-life will of course depend on the C-rate - i.e. at which current cells are charged and discharged. Where C simply corresponds to the Ampere hour rating of the cell. Common C-rates are 0.33C, 0.5C, 1C and so on, where the number is a multiplier of the cell’s rating.
Customers will of course put a lot of weight on the cycle
range of the cells. Since they will of course want to be able
to use them for as long as possible, not only to to keep the
end customer from needing to change batteries too often, but
also to lessen the environmental impact.
Life-cycle flows are the most common test that will be ran in the facility on regular operation. Thus, for determining the facility power consumption, a typical usage case of a machine during its most common process was needed.
The chosen method for life-cycle testing was to run a 100%
utilized, 400A cycling unit and measure the chamber and the cycler independently. The test ran in a total of 13 hours, cycling the cells 5 times. Active, reactive and apparent power was the primary recorded parameters. cos(φ) and power factor was recorded as well why they are differentiated is due to [17] where the phase angle between V and ¯ ¯ I is not equal to the cos(φ) in Eq. 3
To model the less prominent 5%-cases, where tests are consuming high currents, a high power peak was artificially added to the real data from active power measurements which was later used in a Monte Carlo model for facility power prediction, see section IV-M.
H. Measurement setup
An quick and easy setup was decided upon to use in collecting power and energy data from equipment in the facility. The Janitza UMG605 and UMG104. The main advantage of this type of device is that it records every needed electrical quality parameter needed for compliance with European energy quality standard EN-50160 which is applicable in European electrical equipment. The measurement setup was installed according to Fig. 4. No other device was evaluated. The Janitza devices was deemed a good fit due to the fact that it has the advantage of machine-to-machine protocol OPC UA for enabling cloud services to collect and analyze data with aligns with NV’s vision of data collection.
The validation facility is divided into several rooms (fire cells) containing mostly cyclers but also auxiliary equipment.
The long term plan in the validation facility is to expand one fire cell at a time hence starting in the first room, collecting data from one or a few machines in addition to utility and auxiliary equipment consumption, and then using that to make an applicable and adaptable forecast model from a which a simulation of how much power is going to be needed during parts of the project. Thus a case study is the chosen method for enabling a rapid expanse in addition to providing meaningful data.
Data of power consumption is collected by the measurement devices utilizing current transformers. There are different types of devices used in for the data collection. They all have in common that the are able to record active power, reactive power, harmonics, voltage and current from which most electrical parameters calculated by the devices is retrieved.
Applied methodology in this case study is to start measurements on one a single unit of a certain type. The perks of having a sufficiently enough mobile measurement
system is that the devices can be moved to another type of cycling unit in order to compare results. Two types of cycling units were tested. the 200A and the 400A types. The devices was installed on the cycler and on the chamber separately. If the cycler had two chambers, which is the case for the 200A machines, only one of the chambers’ was measured. An assumption that they would behave equally was made with the basis that they had undergone both factory and site acceptance tests and had been approved accordingly.
As can be seen in Fig. 4 the UMG605 will include influence from the step down transformer, which should have an marginal impact on losses since according to EU regulations there are certain constraints on the minimum efficiency of transformers [29].
I. Cycler internal utilization of discharge power
One key aspect of determining the cycler characteristics is to confirm or reject the assumption that the cyclers can internally utilize discharged energy during the test procedures to use while charging other cells connected to the same cycler within the same chamber. If this hypothesis is not rejected, advanced planning using optimization algorithms to establish the best possible schedule of cell-testing to minimize power need and thus considerably reducing energy costs and not contribute to unnecessary energy usage.
A test procedure was built to establish the degree of which the cyclers can regenerate energy internally .
1) Test 1: Establishing a baseline. All channels are performing the same flow which can be seen in Tab.
IV.
2) Test 2: Upper shelf re-energizing the lower shelf by mismatching the flows so that the top shelf is discharging will the bottom shelf is charging at the same constant power per channel and cell.
3) Test 3: Upper shelf divided into 2 parts A and B.
Analogous steps are taken for lower shelf - C and D.
A and C running the same flow while B and D are running the same but mismatched to A and C.
This process will yield, in the case of Test 1, two peaks per parameter level (or two peaks per cycle) as can be seen in Tab.
IV. Adding these two peaks together will give the baseline energy need for one cycle. For Test 2 and Test 3 it will be somewhat different. There should be two very similar peaks for a charge+discharge per cycle. The sum of these two will surmise to a number that will be compared to the baseline case, which will be the indication that shows if there is a possibility to perform scheduling optimization.
J. Consumption and quality testing
Determining the efficiency and the consumption of the
cyclers at different levels of operation is crucial to be able to
provide an overall picture of the upcoming consumption’s
Discharging
Charging
Upper shelf
Lower shelf
Channels:
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 Channels:
Step 1 Step 2
Test 1: Baseline
Fig. 5. Step 1 and 2 of baseline test
Discharging
Charging
Upper shelf
Lower shelf
Channels:
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 Channels:
Step 1 Step 2
Test 2
Fig. 6. Step 1 and 2 of Test 2, where the the chambers’ shelves are performing the opposite charge-process of the other shelf. To see if any energy discharged in the upper shelf can be used in the other.
Discharging
Charging
Upper shelf
Lower shelf
Channels:
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 Channels:
Step 1 Step 2
Test 3
Fig. 7. Step 1 and 2 of Test 3, where the the shelves are divided into two sides with one side performing the opposite charge-process of the other.
from the soon-to-be installed new cycling equipment in wave 3. By performing tests at different ”loads” on the cycler it is possible to see a typical consumption in addition to see if the energy consumption is compliant with the specification of the machines. Ideally it is in Northvolt’s interest for the
cyclers and chambers to operate below nominal power or as low as possible. Since the demand of power in the facility will be a large portion of the operational expenditure will stem from energy costs.
The main focus was to study the consumption at different
levels. Hence conducting a test at constant current (CC) and constant power (CP) was conducted for both discharge and charge. During these tests a whole array of parameters are also able to be recorded for analysis. CP-tests allow for easily comparing input AC-power with used DC-power within the cells during charging, and the same is also true for the opposite process when discharging.
The base-load of the cycler will be measured as the power consumption when no flows (which are what the process of cycling cells at different C-rates, SOC’s and so on) are running in the cycler. I.e. the base-load is the stand-by consumption of the cycler.
The reactive power needed during the discharge part of the flows are also of particular concern hence recording it is of great importance. The total harmonic discharge of the voltage was observed in real time.
Power is recorded as an average over the two minutes the operations in the consumption flow are running.
Step Operation Constant parameter
1 Rest -
2 Discharge 25A CC
3 Rest -
4 Charge 25A CC
5 Rest -
6 Discharge 50A CC
7 Rest -
8 Charge 50A CC
9 Rest -
10 Discharge 75A CC
11 Rest -
12 Charge 75A CC
13 Rest -
14 Discharge 100A CC
15 Rest -
16 Charge 100A CC
17 Rest -
18 Discharge 75W CP
19 Rest -
20 Charge 75W CP
21 Rest -
22 Discharge 150W CP
23 Rest -
24 Charge 150W CP
25 Rest -
26 Discharge 225W CP
27 Rest -
28 Charge 225W CP
29 Rest -
30 Discharge 300W CP
31 Rest -
32 Charge 300W CP
TABLE IV
TEST PROCESS FOR CONSUMPTION AND EFFICIENCY EVALUATION
Special testing cells are used in the tests due to having enough cells to fully utilize the cyclers.
K. Operation of Chambers
To be able to properly forecast the energy consumption a rigid knowledge of how the chambers operates at different conditions is needed. One key aspect which is detailed in the specification is that the chamber must be able to keep the set
temperature within a certain limit of ± 1 °C. The Chamber has a liquid cooling system and a heating system connected via a control system that regulates the ambient temperature within the chamber. Due to the importance of keeping the temperature as stable as possible within the cells, to be able to ensure a proper testing environment, the control system is working hard to keep the temperature as stable as it possible can, especially at 25 °C.
The chamber has a liquid cooling system consisting of pumps and heatsinks that leads the heat away from the chamber when it needs to decrease the temperature within the system.
Power removed from the chamber by the cooling system can be calculated using thermodynamic parameters.
P
cooling= (T
out− T
in) · Q · µ
c· ρ
c(13) Where Q is the flow, µ
cthe specific heat capacity and ρ
cthe density of the solution.
The chambers can house different amounts of cells in so called channels. The amount of channels in a chamber depends on the cycler it is connected to, and the cells which needs to be tested. Higher capacity cells require higher current to be able to test them at extreme conditions hence the need for such high currents as 800A and 1000A. Roughly speaking the chambers make up the majority of units within the firecells.
Some cyclers have one chamber whilst others has two. More details on this can be seen in Tab. V
L. Operation of Cycling units
The operation of the cycling is of course heavily dependent on the types of tests being run. As previously stated the baseline consumption of the cycler is one important factor in the mapping of the overall efficiency of the unit. The baseline power need is measured when the cycler is powered on but not running any specific operations on the cells.
Linearity between cycler types are investigated.
Measurements of minimum two types of cyclers are thus, needed.
M. Monte Carlo methods
Probability methods and simulations can be a very useful tool when attempting to tell the future using forecasts and estimations. Monte Carlo methods are a common method to use in economics and finance [30] as well as determining power need [31]. It uses a randomization process that simulates uncertainties of parameters used in models where the distribution of the outcome later on can be analyzed. In essence is randomizes parameters and takes the outcome as a sample to provide approximate solutions to problems that are quantitatave.
The Monte Carlo model (MCM) can look very different
depending on the field it is used in or even on a case to case
basis, they only share the use of randomness to solve
deterministic problems. In the context of this thesis it will be
used to help in rationalizing conclusions of outcomes from probability distributions to serve as an aid in power related decision-making. For instance, knowing the probability of reaching peak consumption is decidedly useful both in risk-management when ordering power from the DSO, and when determining other construction aspects that need to take the power need into consideration.
In this project the stochastic parameter is the consumption of the individual cycling units within the facility. Since making decisions with the rated power or current rating of the equipment will yield power requirements that does not align with the way they are intended to be used, a fundamental model based on testing of currently installed equipment is used as a base for the Monte Carlo method.
The equipment used in the model (see Tab. V) are already in commission or will be under the coming months. For each
Unit type Cycler type Qty Chambers (avg.) Sn(MVA)
Pris. Cycl. 200A 4 2 1.1
Pris. Cycl. 250A 13 2 1.959
Pris. Cycl. 400A 4 1 X
Pris. Cycl. 450A 7 1 0.84
Pris. Cycl. 800A 28 2 2.45
Pris. Cycl. 1000A 2 1 0.22
Cylin. Cycl - 9 1
Pris. Form - 5 1
SWFPC - 9 1.78 0.2
SWFMC - 5 1 1.14
TABLE V
EQUIPMENT USED IN THEMCMWITH THEIR RESPECTIVE TOTAL NOMINAL POWER. FOR THESWFPCTHERE ARE IN SOME CASES2AND IN
OTHER1,CHAMBERS PER CYCLER.
type of prismatic cycling unit there is an individual consumption data set, as can be seen in Tab. V.
Of course different types of machines have a different consumption profile. Based on the life-cycle measurements of the 400A cycling unit, the resulting power was then manipulated to fit other types of cyclers based an assumption that was made regarding which project (see Tab. III) was going to be used in which type of cycling unit. For instance one possible scenario could be to use project A in the 200A and 400A cycler, project B in 400A and 800A, and finally project C in 800 only. This would yield an average power per cycler, since the cells from each project has a different capacity, running all the life-cycle test at 0.5C requires a different amount of power for each type of cycler. These then acted as the data sets for which a random value was taken for each cycler to provide a total power need each second. The method provides outcomes of possible power needs at an instant in time. This is then repeated numerous times to generate a histogram of which a probability density function can be fitted over.
A MCM can be built in most environments, but for this task Python was used, due to it being a robust programming language that is simple to use as well as being open source.
Apart from providing a foundation on which reasoning can
be built, the outcome also yields a presentation of the power one can expect to be needed.
The model in this report performs the MCM-iteration 10 000 times. The difference in between tests has been recorded at less of 5 kW which in this case is < 1%.
V. R
ESULTSA. 200A Cycler test outcomes
Power consumption of a 200A cycling unit with 100%
utilization i.e. 32 cells is seen in Fig. 8
Fig. 8. Cycler consumption of a 100% utilized 200A cycler at different levels of DC parameters in the cells.
Constant parameter PAC (kW)
25A discharge 0.95
25A charge 6.77
50A discharge -0.9
50A charge 10.95
75A discharge -2.24
74A charge 15.98
100A discharge -3.25
100A charge 21.44
TABLE VI
MEASURED POWER CONSUMPTION AT DIFFERENT CONSTANTDC - PARAMETERS ACTING UPON UTILIZED CHANNELS WITHIN THE CHAMBER
The test described in Tab. IV for a 400A cycler has its outcome
presented in Fig. 9. The number adjoining the peaks denote
the operation for all the 16 channels within the chamber during
the test. Fig. 10 shows the results for Test two. The third and
final internal energy utilization test is presented in Fig. 11
described in Fig. 6. Details regarding the plots in Fig. 9,10 and 11 can be seen in Tab. X.
Fig. 9. Cycler consumption of a 100% utilized 400A cycler at different levels of DC parameters in the cells.
B. Chamber power characteristics
The chambers currently operates with no regards to conserving energy consumption. A hypothesis was constructed that the control system is poorly calibrated when set to a temperature that is far off from the current temperature in the chamber.
The results at different set temperatures is visualized in Fig. 12 and listed in Tab. VII. During the life-cycle tests the chamber consumption was observed - Fig. 13.
°C ∆T (average) Cooling effect Power [kW]
45 0.75 3.69 7.71
35 0.57 2.81 7.79
30 0.55 2.71 4.77
28 0.47 2.31 4.62
25 2.09 10.29 16.45
10 2.01 9.90 14.96
5 1.92 9.47 14.35
-5 1.34 6.60 10.88
-10 1.32 6.50 10.35
-20 1.19 5.86 9.8
-40 0.98 4.82 8.21
TABLE VII
POWER PROFILE OF A CHAMBER CONNECTED TO A200ACYCLER DURING DIFFERENT SET TEMPERATURES
C. 400A Cycler test outcomes
For cos(φ) during life-cycle test see Fig. 14 as well as power factor (Fig. 15 ) was recorded during the life-cycle test.
Fig. 10. Cycler consumption at different levels of DC parameters in the cells for Test 2 described in Fig. 6.
PDC,ch[W] Chan. PRest PDC[kW] PACpeak 1 PACpeak 2
75.00 16 3.23 1.20 2.06 4.53
150.00 16 3.23 2.40 1.15 5.93
225.00 16 3.23 3.60 0.22 7.47
300.00 16 3.23 4.80 -0.50 9.09
TABLE VIII
TEST1 -TEST RESULTS OF RUNNING CELLS WITH CONSTANT POWER
D. Power need estimation
Simulating the future with the Monte Carlo model yields a probable outcome for the power and the risk in reaching power peaks that might surpass the grid subscribed power according to the terms in the facility agreement with the building proprietor. Different cases were chosen to be studied and gave an idea of how much certain inputs affected the final outcome.
First and foremost a baseline case had to be established.
This baseline case is aligned with the current plan of incoming soon-to-be commissioned equipment in the facility.
This provided a starting ground from which other cases evolved and gave insight of where the largest gains in terms of money saved, as well as its correlated parameter energy saved. Additionally, which types of equipment where the most power/energy heavy was also an outcome of the simulation. The baseline case outcomes is shown in Fig. 17.
Fig. 18 shows case two, i.e. how much the histogram would
shift on the x-axis when improving the chamber’s consumption
by 50%. When excluding the SWF module cyclers, the result is
Fig. 11. Cycler consumption at different levels of DC parameters in the cells for Test 3 described in Fig. 7.
PDC,ch PRest PDC PAC-Disch. ηdCh PAC- Charg. ηCh
75.00 3.23 1.20 2.06 0.98 4.53 0.92
150.00 3.23 2.40 1.15 0.87 5.93 0.89
225.00 3.23 3.60 0.22 0.84 7.47 0.85
300.00 3.23 4.80 -0.50 0.78 9.09 0.82
TABLE IX
EFFICIENCY RESULTS FROM A400ACYCLER TEST AT DIFFERENT RATES OF CONSTANT POWER PER CHANNEL. ALL16/16CHANNELS RUNNING
THE SAME FLOW USING TESTING CELLS.
changed to what can be seen in Fig. 19. Finally the distribution plots within the same frame as in Fig. 20 is shown to compare the three cases.
VI. D
ISCUSSIONA. Cycler Unit Consumption
The cyclers had a significantly larger power consumption than was what originally thought. Poor efficiency and a high base-load consumption rose the power per cycler from an assumed low power need to a significantly higher level. The initial assumption made by NV; that the cyclers would consume roughly the same amount of energy that it takes to charge cells, can thus, be deemed unlikely.
Performing life-cycle testing 95% of the time also brings forth the unwelcoming fact that the cyclers will not act as energy producers under discharge operations but instead still be considered as consumers of active energy with the added realization of having a poor power factor when all cells are discharging.
Fig. 12. Power consumption during levels of set temperature for an individual chamber. Losses can be seen as the part of the power that is not removed by the cooling system
Test PDC,ch PACPeak 1. PACPeak 2 Cycle power [kW]
Baseline 75 2.06 4.53 6.59
- 150 1.15 5.93 7.08
- 225 0.22 7.47 7.69
- 300 -0.50 9.09 8.59
Test 2 75 3.25 3.24 6.49
- 150 3.51 3.51 7.02
- 225 3.85 3.87 7.72
- 300 4.24 4.27 8.51
Test 3 75 3.26 3.27 6.53
- 150 3.53 3.53 7.06
- 225 3.83 3.86 7.69
- 300 4.27 4.28 8.55
TABLE X
ENERGY CONSUMPTION RESULTS DURING A TEST-CYCLE STEMMING FROM INTERNAL ENERGY UTILIZATION TESTS