Degradation-Aware Derating of Lithium-Ion Battery Energy Storage Systems in the UK Power Market

Abstract

As more renewable energy sources are integrated into the United Kingdom’s power grid, flexibility services are becoming integral to ensuring energy security. This has encouraged the proliferation of Lithium-ion battery storage systems, with 85 GW in development. However, battery degradation impacts both system lifespan and the economic viability of large-scale projects. With rising commodity costs and supply chain issues, maximising the value of energy storage is critical. Traditional methods of mitigating battery ageing rely on static limits based on inflexible warranties, which do not fully account for the complexity of battery degradation. This study examined an alternative, degradation-aware current derating strategy to improve system performance. Using an optimisation model simulating UK energy trading, combined with an electro-thermal and semi-empirical battery model, we assessed the impact of this approach. Interviews with industry leaders validated the modelled parameters and the relevance of the alternative strategy. Results show the degradation-aware strategy can extend battery lifetime by 5–8 years and improve net present value and internal rate of return over a 15-year period compared with traditional methods. These findings highlight the economic benefits of flexible, degradation-aware operational strategies and suggest that more adaptive warranties could accelerate renewable energy integration and lower costs for storage operators.

1. Introduction

The threat of climate change has created an urgent imperative for countries around the globe to transition away from relying on fossil fuels by developing low-carbon energy systems [1,2]. The UK government expressed support for a low-carbon future by outlining a plan to decarbonise the nation’s power system by 2035 [3]. An essential part of achieving this goal is integrating renewable energy sources into the UK power grid [4].

However, the predicted increased use of renewables in the UK power grid presents a new set of challenges. The increased use of intermittent renewable energy sources, coupled with a decrease in reliably generated, dispatchable energy from traditional fossil fuel generators, creates problems with energy security and balancing instantaneous supply and demand [5,6]. The non-synchronous nature of intermittent renewable technologies decreases the power grid’s inertia, with knock-on impacts on the safety and reliability of such systems.

Flexibility in energy systems involves manipulating the way energy is produced, generated, or stored to improve a power system’s ability to balance generation and demand [6]. There are various options for achieving low-carbon flexibility, with energy storage presenting itself as the most diversely applicable. Lithium-ion batteries (LIBs) are the most used energy storage technology globally for stationary storage applications today [5,7] and have significant projected growth within the UK power market [8,9]. Large-scale, stationary battery systems require substantial investment. Therefore, battery lifetime plays a crucial role in developing feasible business cases and maximising profit [10].

However, battery degradation is complex and difficult to manage. There are numerous interconnected mechanisms that contribute to LIB ageing—including mechanical, electrochemical, and thermal factors. Degradation models can be used to inform operational strategies that minimise degradation. A common, low-cost method for limiting degradation is derating, where the charging and discharging rates of batteries are limited. Derating strategies are typically based on parameters that are monitored by most battery management systems, such as temperature or state of charge (SOC). Conventional methods of managing battery degradation often take a simplistic view of degradation, involving linear degradation curves and static, conservative operational thresholds outlined in conventional battery warranties.

This study evaluated a degradation-aware derating strategy for assessing the potential for this alternative strategy in comparison to common operational methods.

1.1. Literature Review

There are various academic studies that explore the underlying mechanisms behind LIB degradation. The literature consistently classifies the origins of LIB ageing into two categories: calendar and cycle ageing [11,12]. Calendar ageing occurs as a result of storing energy in a LIB. These ageing phenomena are driven by temperature, duration of storage and SOC [5,6]. Cycle ageing encompasses the losses associated with charging and discharging the battery. These losses are attributed to a complex combination of factors, including temperature, charging/discharging current, charge cut-off voltage, charge cut-off current and C-rate. Peterson et al. present results indicating that Depth-of-Discharge (DoD) contributes significantly to LIB ageing [13].

Battery degradation models use various techniques to capture the complexity of LIB ageing. In [14], an empirical degradation model is provided as a function of simulation time, charge throughput and charge capacity. Accelerated ageing test data from 60 nickel-manganese-cobalt cells were used to parameterise the model. The authors split calendar and cycling ageing into two different analyses. For cycle ageing tests, the DoD and SOC varied at a constant temp and C-rate, whereas, for calendar ageing tests, the temperature and voltage varied. From the datasets, two mathematical descriptions for cycle and calendar ageing were found and combined. Similarly, Peterson et al. used aggregated EV driving data to parameterise an equation relating capacity fade to cycling time, integrated battery capacity, and the energy throughput of the cells [13]. In [15], an empirical degradation model was combined with physicochemical sub-models that linked degradation rates of specific degradation mechanisms to influencing factors such as SOC and temperature. Combining these two methods made this semi-empirical model generalisable enough to be applicable to different contexts, dissimilar to the testing conditions used to gather the parameterisation data [15]. The authors validated the model against batteries with different states of health (SOHs) and different battery chemistries and achieved a maximum error of 5% across all scenarios.

Degradation models can be used in conjunction with operational schemes to predict and extend battery lifetime to maximise system performance. All operational strategies aim to control the key parameters influential to LIB degradation, such as temperature, C-rate, and SOC [16]. Active thermal management, or temperature control, is traditionally used in large-scale LIB systems to manage degradation and prevent thermal runaway [17,18]. The literature suggests that the optimal operating temperatures for Lithium-ion battery systems are between 15 and 35 °C [19,20,21]. Battery warranties for large-scale systems generally stipulate a static operating temperature within this range despite the more complex relationship between temperature and degradation illustrated in various degradation models [16]. Active thermal management also incurs issues of parasitic power draw and impacts the specific energy density of the LIB pack, creating a trade-off between the efficiency and lifetime of a system [22].

As such, alternative methods for managing LIB degradation and lifetime have been explored in the literature. Hybridisation has been explored in the context of combining smaller LIBs with larger modular battery packs to improve economic performance or combining LIBs with supercapacitors to exploit complementary technological capabilities [23,24,25,26]. Active system balancing methods such as decentralised system balancing [27] and offline energy balancing based on convex optimisation [27] have also been explored. However, these active strategies both suffer from barriers of complexity and increased infrastructure costs.

Barreras et al. provided an overview of derating strategies for LIB, indicating that industry derating standard involves the use of look-up tables that map specific stress factor values to the necessary derating factors [28]. Derating factors are the ratios by which some operational conditions need to be reduced from their maximum capability in order to achieve the desired degradation rate. Based on [28,29], it is generally the charging or discharging current that is varied in order to keep the stress factor being considered within the specified limits and achieve the desired degradation rate. In [30], current derating was investigated as a way of extending the battery life of isolated mini-grids in developing countries. SOC derating was shown to be the most effective strategy, increasing battery lifetime up to 7 years while using a static SOC limit of 50%. Similar SOC derating based on static limits was investigated in the context of second-life batteries used in a PV and battery energy storage system (ESS) in [31]. The analysis showed that reducing the permissible SOC limits from 95–15% to 85–20% reduced battery degradation and increased project lifetime such that second-life batteries became more economical than the use of new batteries.

Although these studies found success using static SOC derating strategies, practical issues arise with SOC derating, such as how to perform online SOC estimation [28]. More generally, the industry standard of using static derating based on look-up tables has the flaw of being empirical and, therefore, based on a case-by-case formulation [16]. In [29], the variations in the responses to derating between battery types, and even within different batteries of the same type, illustrated that derating can only be effective using such a case-by-case method due to the varying degradation rates of different battery types. As such, these methods are not easily transferable. In addition, the static limits imposed on stress factors do not capture the full complexity of degradation. This is illustrated in [16], wherein a derating based on static manufacturer’s limits does not achieve the same lifetime increase as a more dynamic derating strategy.

As a result of these limitations, more complex operational strategies have been explored, considering battery degradation in conjunction with the economic performance of BESS. These can be divided into objective-based approaches and constrained-based approaches. The first category involves including degradation in an objective function to optimise performance, whereas the latter involves constraining the operation of the system according to degradation principles. Falling into the first category, Kumtepeli et al. proposed an Energy Management System controlled by a mixed-integer quadratic programming problem aimed at cost minimisation. The degradation cost in the objective function was calculated using a degradation model that includes both cycle and calendar ageing effects [32]. Similarly, in [33], the objective was to optimally schedule how a BESS should charge/discharge based on system requirements as well as degradation rates. Two different empirical degradation models were used to introduce a degradation penalty into the objective function of a profit-maximising optimisation model, cycling through different degradation penalty amounts [34].

Falling into the category of constrained-based operational strategies, Li et al. introduced a method for forecasting the performance and decline of second-life batteries in electric grid systems, integrating an electrochemical model, health monitoring, and a cost-reduction algorithm tailored for grid use [35]. In [36], a battery life model that incorporates historical ageing, state of charge (SOC), and charge-discharge rate (C_rate) for online applications was introduced. A sliding-window dynamic programming strategy was investigated to optimise short-term grid service revenues and long-term battery life losses. Similarly, a multi-objective approach to maximise revenue while taking cycling losses in lithium-ion batteries into account was proposed in [37]. Klein et al. presented an optimal charging strategy that used non-linear model predictive control to maximise charging rate while maintaining safe operating conditions and long-term battery ageing [38]. Yin et al. developed a fast-charging strategy that included charging time and SEI formation as variables and cut-off voltage and Lithium-ion concentrations as constraints [39].

Operational strategies based on stipulations in battery warranties can be considered simplistic, constraint-based approaches. Warranties are manufacturer guarantees of Lithium-ion battery (LIB) performance, offering compensation if performance is not met. Warranties define operating limits for parameters critical to battery degradation, such as temperature, state of charge (SOC), and C-rate, along with terms like duration and acceptable capacity degradation [40,41]. While electric vehicles (EVs) and plug-in hybrids (PHEVs) have well-established warranties, typically 8 years or 100,000 miles, with an end-of-life (EoL) threshold of 80% state of health (SOH), stationary battery ESS lacks a standard warranty framework due to limited data on long-term degradation [42]. However, most warranties for stationary systems are static, with fixed cycle limits per year, regardless of actual usage or degradation patterns [43]. Flexible warranties, which adjust degradation curves and operational limits based on real-time system performance, offer a more favourable approach than static warranties by giving operators better insight into the trade-offs between short-term intensive cycling and long-term battery life [40]. However, there is limited research on the impact of warranty terms on LIB performance, likely due to non-disclosure agreements between manufacturers and system operators. Given the importance of warranties in shaping the economic and operational strategies of LIB projects, further research on developing flexible warranties in conjunction with degradation-aware operational strategies is crucial.

1.2. Contribution

This study is the first to compare the economic and system lifetime performance of large-scale, stationary Lithium-ion battery systems in the UK power market, evaluating traditional warranty-based operational strategies against a constraint-based, degradation-aware derating strategy.

The work presented in this study used an advanced semi-empirical battery degradation model that is experimentally validated, as outlined in [44]. The modelling also considers battery temperature variation and investigates the basic impact of different thermal management strategies by comparing passive and active thermal management techniques. In addition, this investigation accounted for the long-term impact of battery degradation in control regimes by implementing scenarios with battery degradation-aware derating strategies, which impose dynamic BESS operating constraints dependable on allowable degradation rates [16]. Interviews with leading industry members were performed to validate the assumptions behind the counterfactual warranty-based scenarios and to gauge the major obstacles to further optimising battery lifetime.

Through the outlined approach, this paper sought to address the following questions:

What are the implications of conventionally used static operational limits for battery degradation, as defined in traditional battery warranties?

How far can battery lifetime be improved by implementing degradation-aware derating strategies?

How can the maximum degradation rates used in a degradation-aware derating strategy be tailored to balance system lifetime and economic performance for a large-scale battery storage system within the UK power market?

What are the economic implications of considering the full complexity of battery degradation for large-scale, stationary systems operating in the UK, and how can this apply to future warranty development and operational strategies for Lithium-ion battery systems?

A summary of the contribution of this paper in the scope of existing BESS market studies is outlined in Figure 1.

2. Materials and Methods

In order to tackle the questions outlined above, this paper draws together four sections of a methodology: (1) case study development, including defining an optimisation problem to simulate energy arbitrage cycling behaviour and establishing static operational limits through interviews with industry members; (2) a semi-empirical battery degradation model for a lithium iron phosphate (LFP) battery, which is the most dominant battery chemistry used in stationary energy storage systems [46]; (3) the degradation-aware operational strategy developed by Schimpe et al. [16], which combines individual calendar and cycle ageing reduction strategies to minimise degradation and, finally, (4) an economic analysis. The semi-empirical battery degradation model forms part of the Full Battery System model (FBSM). This model also includes calculations to scale up cell level assumptions to pack level and an equivalent circuit electro-thermal model, used in the context of this paper to simulate various thermal management scenarios. Sowe et al. implemented the same FBSM [30], and a full description is provided in [44,47].

2.1. Case Study Development

The two most influential parameters that define battery ageing are the battery cycling behaviour and the battery ambient temperature. Both aspects were defined for various case studies so that the impact of degradation-aware derating on both parameters could be investigated and the economic consequences established.

One of the most profitable applications for LIB ESS in the UK energy market is energy arbitrage-charging the battery when wholesale electricity prices are low and selling that stored energy when the wholesale prices are high for a profit. The UK’s liberalised electricity market and fluctuating wholesale market prices make energy arbitrage an attractive business opportunity for LIB investors and system operators. Additionally, energy arbitrage as a revenue stream does not share the same risk of saturation as other profitable LIB ESS revenue streams, such as frequency response services, which are limited in the volumes procured. Energy arbitrage is also a suitable context for a derating strategy, as the cycling levels are up to the discretion of the system operator and not dictated by long-term contracts with the National Grid Energy System Operator (NGESO). Therefore, this case study explores energy arbitrage operating within the UK’s day-ahead wholesale power market. The base case investigated in this paper involves a 50 MW/50 MWh, 1 h duration Lithium-ion battery system located in the Birmingham area.

The battery cycling behaviour for the base case was obtained using an optimisation problem based on the methodology developed by Wankmüller et al. [34]. While the optimisation problem is simple, this was deemed an appropriate solution as the focus of this investigation is battery degradation as opposed to arbitrage cycling behaviour. Perfect foresight of wholesale market prices was assumed for the development of this optimisation problem. Previous analyses of energy arbitrage profit predictions illustrated that this assumption leads to between 10 and 15% profit overestimation [48,49]. This is a modest overestimation considering the simplicity of the arbitrage model and was deemed acceptable for this investigation, considering the focus of this work centres around exploring the impact of derating rather than implementing complex price forecasting techniques. The objective of the energy arbitrage model is to earn the highest revenue possible, leading to an objective function that reads as follows:

$$max\sum _{t=1}^T{P}_{DA}(t)\left(E_{sell}(t)-E_{buy}(t)\right),$$

(1)

where $P_{DA}\left(t\right)$ refers to the day-ahead wholesale price at hour $t$, $E_{sell}(t)$ is the optimal amount of energy the battery should sell or discharge at hour $t$, and $E_{buy}(t)$ is the optimal amount of energy the battery should buy or charge at hour $t$. The day-ahead wholesale market operators are hourly; therefore, each time step from $t$ to $T$ is one hour. Constraints were introduced to maintain a state of charge between 0 and 100% and the C-rate from exceeding 1C, considering this is a 1 h duration system. It is noted that the useable energy capacity might be lower than the installed capacity in many instances, leading to a potential maximum DoD of less than 100%. However, this was neglected in this analysis as specific system design operations vary depending on asset owners and suppliers. Binary variables were also introduced to prevent charging and discharging from occurring simultaneously, as explained in [36]. The day-ahead wholesale price data for 2020 were sourced from [50] and used to populate $P_{DA}(t).$ The same dataset was used by Núñez et al. [51] in a similar optimisation problem. The cycling behaviour defined by this version of the optimisation problem is referred to as the Base Case.

There is a lack of academic literature critically assessing the implications of battery warranties on battery lifetime, battery degradation and the economic potential of LIB ESS. Therefore, a synthetic warranty was developed based on LIB warranties pertaining to systems similar to the system developed in the case study. Interviews with various stakeholders in the LIB ESS supply chain were conducted in order to gauge the usual parameters of a battery warranty and approaches to thermal management. In addition, the interviews aimed to establish how battery warranties and thermal management influenced the different stakeholders’ decisions regarding system sizing and operation. Stakeholders who took part in the interviews included energy storage investors, LIB ESS integrators, stakeholders involved with LIB Engineering, Procurement and Construction, and, finally, system operators.

Based on interview results, two synthetic warranties were developed. Warranty 1 has more relaxed cycling limits and Warranty 2 involves more stringent cycling limits. Both warranties included a DoD limit, a C-rate limit, an annual average daily cycle limit, and a maximum daily cycle limit. The warranties were used to adjust the constraints of the arbitrage model such that the most revenue was earned while adhering to the stipulations of the warranty. Operation under those warranty rules is simulated along with the base case and the degradation-aware operation cases.

The FBSM consists of an electro-thermal model used to simulate the change in cell temperature as a result of joule losses heating up the battery and heat transfers between the cell’s surface and the ambient. The full description of this electro-thermal model is defined in [47], and the same assumptions were used in [30]. Ambient cell temperature is a required input parameter. For the Base Case and Warranty scenarios, it was assumed that the battery was kept in a storage container and a heat pump system was used to maintain a constant ambient cell temperature within the container. For this study, the constant ambient temperature was chosen to be 18 °C, based on the optimal operating temperatures quoted in the literature and expert industry interviews conducted as part of this research. The simulation also accounted for the fact that heat transfer would occur between the surface of the cell and the cell’s ambient as the cell, e.g., heated up due to the dissipative losses. Therefore, the active thermal management system consumes power to maintain a constant ambient cell temperature. The power consumption at each time step was defined as follows:

$$P_{TM}(i)=h_{cell}(T_{i+1,ambient}-T_{i,cell})\left({\displaystyle \frac{1}{COP}}\right),$$

(2)

where $COP$ is the coefficient of performance of the heat pump, taken at a standard value of 3 for cooling and 4 for heating, and $h_{cell}$ is the heat transfer coefficient [52]. To investigate the effectiveness of the degradation-aware derating strategy to assist with thermal management and control cycling behaviour to extend lifetime, passive air cooling was the only thermal management technique simulated when derating was used. As a result, there was no power draw associated with the thermal management strategy for the derated scenarios investigated. The ambient cell temperature for the degradation-aware derating scenarios was taken as the outdoor temperature for Birmingham at each time step sourced from the renewables.ninja platform [52].

2.2. Battery Degradation Model

The degradation model included in the FBSM was parameterised using experimental data from a Sony LFP-graphite cell. The model has been validated over a wide range of temperatures and operational conditions with a relatively low-capacity prediction error of less than 20%.

The model calculates calendar ageing as a function of both temperature and SOC. Cycle ageing is represented by three individual ageing mechanisms, which depend upon charge/discharge current in addition to temperature and SOC.

$$\begin{array}{cc}& Q_{Loss}(T,SOC,I_{ch},Q_{Tot},Q_{Ch})\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} =Q_{L,Cal}(T,SOC,t)+Q_{L,Cyc,HighT}(T,Q_{Tot})\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +Q_{L,Cyc,LowT}(T,I_{Ch},Q_{Ch})\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +Q_{L,Cyc,LowT,HighSOC}(T,I_{Ch},SOC,Q_{Ch})\hfill \end{array}$$

(3)

$$\begin{array}{cc}& Q_{Loss}(T,SOC,I_{ch},Q_{Tot},Q_{Ch})\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} =k_{Cal}(T,SOC)\cdot \sqrt{t}\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +k_{Cyc,HighT}(T)\cdot \sqrt{Q_{Tot}}\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +k_{Cyc,LowT}(T,I_{Ch})\cdot \sqrt{Q_{Ch}}\hfill \\ \hfill & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +k_{Cyc,LowT,HighSOC}(T,I_{Ch},SOC)\cdot Q_{Ch}\hfill \end{array}$$

(4)

$Q_{L,Cal}$ represents calendar ageing and $Q_{L,Cyc,HighT}$,$Q_{L,Cyc,LowT}$ and $Q_{L,Cyc,LowT,HighSOC}$ represent cycle ageing under high temperature, low temperature, and low-temperature high SOC conditions, respectively. The input parameters $T$, $SOC$, and $I_{ch}$ represent cell temperature, state of charge, and charging current at time $t$.

Capacity fade is commonly used as a metric to quantify battery ageing by defining a battery’s state of health (SOH) [1], as shown in Equation (3):

$$SOH={\displaystyle \frac{Actual Battery Capacity}{Nominal Battery Capacity}}$$

(5)

The SOH of the battery at a given time $t$ is calculated by summing the previous degradation:

$$SOH(t)=1-\sum _0^t{\delta }_{Loss,t}$$

(6)

The model was developed by performing multiple calendar and cycling ageing tests through a temperature range of 0 °C to 55 °C. Using Arrhenius and Tafel equations, stress factors, and the various k terms shown in Equation (4), impacting degradation was established under different conditions. The interested reader is referred to the original paper by Schimpe et al. [44] for more information on the model’s development, including experimental procedures, parameterisation, and model validation.

2.3. Degradation-Aware Derating Strategy

Separate degradation rates for calendar and cycle ageing were determined using the degradation model described above. Look-up tables were created, mapping the degradation rates to reference conditions. The calendar ageing rate was found to be dependent on SOC and cell temperature. The cycle ageing rate was divided into two separate rates for charging and discharging, with the former being dependent on cell temperature, SOC, and charging current and the latter only on cell temperature. The relationship between a fixed maximum degradation rate, cell temperature, SOC, and charging current in shown in Figure 2 and Figure 3 for calendar and cycle ageing rates, respectively.

The maximum battery current was controlled to ensure that reference maximum degradation rates were not exceeded at each time step. For calendar ageing, this was achieved by using the look-up table and SOC at the current time step to determine the maximum permissible current. The electro-thermal model component of the FBSM was used to determine the maximum cell current such that the joule losses did not cause the cell temperature to violate the maximum cell temperature limit through self-heating effects.

For cycle ageing, the charging current was determined directly from the look-up tables by using SOC and cell temperature at the current time step to determine the maximum permissible current. The minimum between the calendar and cycle current thresholds was taken as the maximum permissible current value for each time step. The maximum permissible current value was compared to the cycling behaviour defined by the optimisation problem defined in Section 2.1, and the current was saturated where necessary. For the full explanation of the development of the degradation rates, the look-up tables and full details on the degradation-aware derating algorithm, the interested reader is referred to Schimpe et al.’s paper [16]. An overview of the degradation-aware derating strategy can be seen in Figure 4.

The Base Case operates without any derating limits and with active thermal management implemented, as described in Section 2.1. Note that the Base Case is not a realistic operational strategy and is included purely as a reference scenario to compare the results of subsequent scenarios.

Warranty 1 and Warranty 2 operate according to the interview-derived warranty guidelines, including active thermal management, as described in Section 2.1. Additional parameters stipulated in these synthetic warranties include limitations on the maximum daily full equivalent cycles (FECs), DoD, and C-rates.

Derated uses degradation-aware derating strategy, with maximum degradation rates set to reference conditions. No active thermal management is included in this simulation to assess the impact of more variable cell temperature control on degradation.

Optimal Derated is identical to Derated, but the values for the maximum degradation rates are optimised for maximum economic performance. Further details follow in Section 3.2.

Finally, the Optimal Derated + Active TM is identical to Optimal Derated except that active thermal management, as described in Section 2.1, is simulated.

Table 1. Summary of scenarios investigated.

Scenario	Operational Strategy	Thermal Management
Base Case	Profit optimisation; No cycling, DoD or C-rate limitations.	Active
Warranty 1	Profit optimisation; Maximum 3 FECs per day, yearly average of 2 FECs per day, 80% DoD limit, 0.95C C-rate limit.	Active
Warranty 2	Profit optimisation; 1.5 FECs per day, early average of 1 FEC per day, 80% DoD limit, 0.95C C-rate.	Active
Derated	Maximum degradation rate used for derating is calculated at reference conditions.	No active thermal management
Optimal Derated	Maximum degradation rate used for derated is tuned for maximum economic performance.	No active thermal management
Optimal Derated + Active TM	Maximum degradation rate used for derated is tuned for maximum economic performance.	Active

provides an overview of all scenarios.

2.4. Economic Analysis

For all scenarios, the revenue from each year was calculated using the day-ahead wholesale prices, extended to 15 min intervals, and the cycling behaviour for each scenario. The money earned from selling energy contributed positively toward the total revenue, and the money used to purchase the energy to charge the battery contributed negatively to the total revenue. The total revenue at each year, ${rev}_n$, was discounted using a standard discount factor of 10% and summed to calculate the total discounted revenue (TDR) over the time period using Equation (7), where $N$ is the project lifespan in years, and r is the discount rate. The cost of the power draw from the active thermal management scenarios was calculated for each year using the power draw at each time step defined in Equation (2) and using the day-ahead wholesale market prices. This was under the assumption that the battery would need to purchase additional power from the wholesale market to power the active thermal management system. To investigate the economic impact of this additional power draw, the cost was subtracted from the revenue for each year, and the new TDR was calculated.

$$NPV=\sum _{n=0}^N{\displaystyle \frac{{rev}_n}{(1+r{)}^n}}$$

(7)

The net present value (NPV) was also calculated using the capital costs (CAPEX) and operational costs (OPEX) for the 50 MW/50 MWh system size with values provided in

Table 2. Parameters used for economic analysis.

Parameter	Value	Source
CAPEX (£/kWh)	350	[54]
OPEX (£/kW per annum)	3.65	[55]

. The CAPEX was included as a negative investment cost in the first year of the analysis. The OPEX was calculated for each year thereafter and subtracted from the revenue for that year. All economic analyses were conducted over two time frames: a fixed time frame of 15 years and over the system lifetime using an 80% SOH as an EoL condition. An 80% EoL condition was chosen because, although in practice, stationary battery systems are run until 60–65% SOH was reached, calendar and cycle ageing tests performed on LFP SONY cells illustrated no rapid capacity loss after 80% SOH [53]. For the 15-year analysis, if the SOH dropped below 80%, it was assumed that the system required augmentation to remain in operation. As a result, augmentation costs of 20% of the total CAPEX costs were subtracted from the cash flow of the year before the SOH dropped below 80%. After the augmentation, it was assumed that the system earned revenue commensurate with how it earned at the beginning of its lifetime, i.e., if the system was augmented in year 10, the revenue earned in year 11 was the same as that earned in year 1. A sensitivity analysis was performed for the NPV to investigate the influence of OPEX and CAPEX prices on economic performance.

An overview of the full methodology is provided in Figure 5.

3. Results

3.1. Battery Degradation

As shown in

Table 3. Table showing simulation results over 15 years.

	Base Case	Warranty 1	Warranty 2	Derated	Optimal Derated	Optimal Derated + Active TM
Total FEC	17,061	9878	3920	9547	11,401	15,526
Max Charging C-rate (C)	1.0	0.80	0.80	0.96	0.98	1.0
Average Charging C-rate (C)	0.81	0.78	0.70	0.46	0.57	0.81
Min Discharging C-rate (C)	−1.0	−0.80	−0.80	−1.0	−1.0	−1.0
Average Discharging C-rate (C)	−0.81	−0.78	−0.64	−0.73	−0.74	−0.78
Average SOC (%)	49.5	59.2	62.0	20.4	25.5	36.3
Final SOH (%)	73.1	75.8	79.1	85.2	81.6	76.3
Average Cell Temperature (°C)	21.2	19.2	18.4	11.0	11.3	19.9
Max Cell Temperature (°C)	30.8	26.9	27.0	36.5	39.9	30.0
Min Cell Temperature (°C)	18	18	18	−4.1	−4.1	18

by the FEC metric, the Base Case cycled the most from all scenarios, followed by the Optimal Derated + Active TM Scenario, Optimal Derated Scenario, Warranty 1 Scenario, and the Derated Scenario. The Warranty 2 Scenario cycled the least.

The average SOC for the Derated Scenario was significantly lower than the Base Case’s, 20.4% compared with 49.5%. This was linked to the framing of the Base Case’s optimisation problem—to maximise profit, it made sense to fully charge the battery at any opportunity such that large price differentials of the wholesale market could be exploited fully. This led to a higher SOC than when SOC was actively managed in the Derated Scenario. The Optimal Derated Scenario and the Optimal Derated + Active TM Scenario both had slightly higher average SOC than the Derated Scenario due to the more relaxed cycle degradation rates, which allowed for a higher average charging C-rate. The average discharging C-rate for both derated scenarios, however, was only marginally different. This meant that the cell reached a higher SOC faster, and the SOC decreased at the same rate in the Optimal Derated scenarios compared with the Derated Scenario, resulting in a higher average SOC in the former scenario.

Both warranties stipulated a limit on the number of permissible cycles per day and per year. This stipulation, in addition to the DoD limit, meant that the batteries would have a higher average SOC, corresponding to how much cycling is limited. This limitation on cycling meant that the cell spent more time storing energy instead of cycling. Therefore, stricter cycling limits resulted in more time spent storing large amounts of energy with a high SOC. Consequently, Warranty 2 had a higher average SOC of 62.0% than Warranty 1’s average SOC of 59.2%

A summary of the battery degradation results for the 15-year simulation is shown in Figure 6. The Base Case incurred the highest total capacity fade of 26.9%. Warranty 1 incurred only 24.2% of the total capacity fade. Further, SOH improvements were achieved by Warranty 2, which incurred a total capacity fade of 20.9%. The Derating scenario performed the best with regards to capacity fade, incurring only 14.8% at the end of the 15 years, whereas the Optimal Derated scenario incurred 3.6% more capacity fade than the Derated scenario and the Optimal Derated + Active TM scenario incurred 8.9% more than the Derated scenario.

As shown in Figure 6, calendar losses dominated the scenarios where active thermal management was simulated—the Base Case, Warranty 1, Warranty 2, and Optimal Derated + Active TM scenarios. As explained in Section 2.2, calendar ageing is linked to high SOC and high cell temperature. Active thermal management meant that relatively constant temperatures were maintained in these scenarios. The difference between the maximum and minimum temperatures ranged from 12.8 °C for the Base Case to 9.0 °C for Warranty 2. This small variation in temperature meant that temperature played a less significant role than SOC regarding calendar losses.

The same trend can be seen for the remaining scenarios. The average cell temperature for the derated scenarios was low, 11.0 °C for the Derated scenario and 11.4 °C for the Optimal Derated scenario. These low cell temperatures meant that SOC played a more significant role in determining the calendar losses for the derated scenarios. Therefore, the calendar losses incurred across all scenarios correspond with the scenario’s average SOC.

The total cycle losses corresponded to the FEC for each scenario, as shown in Table 3. Most cycling losses from all four scenarios fell under the category of cycling at low temperatures (referred to as Cyc Loss (Cold) in Figure 6). This was also the only loss category where the Optimal Derated scenario achieved the highest capacity fade from all scenarios despite cycling less than the Base Case and Optimal Derated + Active TM scenarios. This is due to the Optimal Derated scenario’s relative FEC compared with other scenarios and the lack of active thermal management leading to higher cell temperatures than the Base Case and Optimal Derated + Active TM scenarios.

Reduced capacity fade had a significant impact on system lifetime, as shown in Figure 7. Warranty 1 and Warranty 2 increased the system lifetime relative to the Base Case by 2.45 years and 5.95 years, respectively.

The Optimal Derated scenario achieved a 10.75-year lifetime increase and the Derated scenario impacted the lifetime of the system by the largest margin, achieving an increase of 21.9 years relative to the Base Case. The Optimal Derated + Active TM scenario achieved a lifetime increase of 1.9 years.

As illustrated in Figure 4, there are four current limits that form the degradation-aware derating strategy. A breakdown of the influence of each current limit is provided in

Table 4. Table showing the breakdown of which ageing reduction value resulted in I_Req being derated for the Derated scenario during the 15-year simulation.

	Cal Charging	Cal Discharging	Cyc Charging	Cyc Discharging
% Derating Origin	0.52	3.98	95.5	0

and

Table 5. Table showing the breakdown of which ageing reduction value resulted in I_Req being derated for the Optimal Derated scenario during the 15-year simulation.

	Cal Charging	Cal Discharging	Cyc Charging	Cyc Discharging
% Derating Origin	1.12	6.15	92.7	0

. The main limitation of the current was the cycle charging limit. This is linked to the average cell temperature in the Derated scenario being 11.0 °C because the cell’s temperature was controlled by passive air cooling from the mild ambient temperature in Birmingham. The influence of charging C-rate on cycle losses at lower temperatures, as shown in Equation (3), meant that the charging current would need to be reduced the most to adhere to the maximum cycle degradation rate.

The Derating strategy successfully reduced calendar losses compared with the other three scenarios. This is attributed to the low average cell temperature and average SOC of 11 °C and 20.4%, respectively. While this low temperature was contributed partially by the calendar ageing reduction methods described in Section 2.3, the breakdown of the derating strategy illustrates that the calendar charging and discharging limits only contributed 4.50% to the current derating. The low average cell temperature was largely achieved by the passive air cooling and the heat exchange between the cell and the relatively low ambient temperature of Birmingham rather than active current derating due to temperature-dependent calendar ageing.

The low average SOC, however, was actively caused by the derating strategy. This was caused by the charging cycle current being limited most severely, contributing 95.5% to the current derating. This means that increasing the SOC through charging was limited. On the other hand, cycle discharging was not limited at all. This is because, as shown in Figure 3, the cycle discharging limit operates as a step-function, with the discharging current being reduced to zero when the maximum cell temperature limit is exceeded. The maximum cell temperature in the derated scenario was 36.5 °C, which is well below this maximum temperature limit. Therefore, discharging from high SOC to low SOC occurred unencumbered, allowing lower SOCs to be maintained.

3.2. Sensitivity Analysis: Degradation Rates

As shown in Table 4 and Table 5, the maximum charging limit originating from the cycle reduction section of the derating strategy has the most impact on the battery current. Therefore, to assess if the degradation-aware derating algorithm can be tuned to increase profitability, the maximum cycle ageing degradation rate was varied in a sensitivity analysis.

As shown in Figure 8c, decreasing the maximum cycle degradation rate increased system lifetime by limiting charge throughput more stringently. As a result, TDR was also reduced. Increasing the maximum cycle ageing degradation rate reduced the lifetime of the cell and, therefore, reduced the total charge throughput through the cell, as the cell was cycled for less time. However, the higher degradation rate allowed for higher average C-rates when the cell was cycled, allowing for maximum price differentials in the arbitrage markets to be better exploited. Therefore, the faster degradation rate corresponded to a higher annual average revenue.

There is a clear turning point where the optimal balance between short-term and long-term revenue was reached illustrated in Figure 8b. This turning point corresponds to a 50% increase in the maximum cycle degradation rate from reference conditions. Increasing the degradation rate above this value reduced the lifetime such that the increase in short-term revenue no longer outweighed the reduced long-term revenue. Consequently, the overall discounted revenue decreased. A 50% increase in the maximum cycle degradation rate from reference conditions was determined to result in optimal economic performance for the derating strategy. The Optimal Derated scenario and Optimal Derated + Active TM scenario used this maximum degradation rate. The initial maximum cycle degradation rate taken at reference conditions was used for the Derated scenario and was qref, loss, Cyc. HEC = 2.5 × 10⁻⁴% of 3 Ah. Therefore, the maximum cycle degradation rate for the Optimal Derated scenario was calculated to be qref, loss, Cyc. HEC = 3.75 × 10⁻⁴% of 3 Ah.

It was noted that the significant drop in IRR and NPV shown at a 100% increase to the degradation rate from reference conditions shown in Figure 8a,b, respectively, is a consequence of the system lifetime at this degradation rate dropping to approximately 14 years, as shown in Figure 8c. As 80% SOH is used as an EoL condition in this analysis, augmentation costs are required at the end of year 14 to continue system operation. This significant cost, equating to 20% of the total CAPEX, as explained in Section 2.4, causes a significant reduction in the most recent cash flow, causing a significant drop in both IRR and NPV. Considering that the NPV and IRR are calculated over a fixed 15-year lifetime, the increased earning potential of the repowered system is not captured in subsequent cashflows when using this degradation rate.

3.3. Economic Results

The NPV for each scenario was given relative to the Warranty 1 scenario. The relative NPV was included as a metric to describe the value provided by each warranty and derating strategy as opposed to commenting on the profitability of the case study. The numeric NPV value for all scenarios was found to be negative, as expected because it is industry standard to employ revenue stacking to achieve profitable LIB ESS business cases. However, the objective of this investigation was to assess the value of derating rather than to devise a profitable business case; therefore, it was deemed appropriate to include the relative NPV as opposed to the numeric NPV as an economic metric.

Table 6. Economic results when the simulation is run for 15 years.

	Warranty 1	Warranty 2	Derated Scenario	Optimal Derated Scenario	Optimal Derated Scenario + Active TM
Average annual revenue (£ million)	0.70	0.50	0.43	0.50	0.72
NPV relative to Warranty 1 (%)	-	−7.5	−3.9	0.5	2.7
IRR relative to Warranty 1 (%)	-	−312	−0.1	8.9	16.6

provides economic metrics for all scenarios over a 15-year simulation period.

The Optimal Derated + Active TM scenario achieved the highest NPV relative to the Warranty 1 scenario, followed by the Optimal Derated scenario. Both the Derated and Warranty 2 scenarios achieved a lower relative NPV. The Optimal Derated scenario earned the same average annual revenue as the Warranty 1 scenario; however, the increase achieved through the derating strategy lifetime meant that no augmentation costs were required, resulting in a higher NPV and IRR.

A sensitivity analysis was performed by varying the CAPEX and OPEX values. The NPV relative to the Base Case calculated over 15 years increased with increasing CAPEX costs for all scenarios due to increased augmentation costs decreasing Warranty 1’s NPV. The NPV relative to the Base Case remained unchanged by increasing OPEX as the OPEX is accounted for as constant yearly payments.

4. Discussion

Warranties are an essential aspect of successful, large-scale LIB ESS projects, dictating permissible operating strategies. Merchant trading business cases are becoming increasingly popular in the UK LIB market, leading to growing interest in economic performance and lifetime considerations.

The degradation-aware derating strategy, as implemented in the Derating scenario, increases battery lifetime by a more significant margin than either Warranty scenario. While both warranties reduced cycling losses and increased lifetime relative to the Base Case, they increased calendar losses due to increased SOC maintained by static cycle limits. This illustrates that static warranties do not capture the complexity of degradation. In addition to not maximising lifetime, both Warranty scenarios were shown to be overly restrictive in terms of cycling, which has significant economic implications.

The optimal degradation-aware derating strategy involved tailoring the permissible degradation rate, which controlled the system’s operation to maximise economic performance. This strategy achieved a higher lifetime and improved economic performance compared with both warranties. This illustrates that overly restrictive, static warranties are not optimal in a profit-maximising setting, and flexible warranties or derating strategies can increase system lifetime while being more economically viable than traditional warranties.

In practice, this could mean having a real-time adjustable warranty wherein degradation rates at certain conditions are the guarantee of the manufacturer, allowing the system operator to have more control over how this information is used. This creates more operational flexibility from a system operator’s point of view. Conversely, this also invites more risk from an investor’s perspective. However, the higher economic potential might outweigh the risk from the investor’s point of view, particularly when considering the long-term revenue potential provided by the extended lifetime this method provides. This potential is contingent on changing CAPEX, as this influences augmentation costs. From the manufacturer’s perspective, this approach means more complicated warranty development. However, as the market moves toward more flexible warranties, this approach offers a complete understanding of their asset’s degradation to aid in flexible warranty development.

While derating was also used to assist with thermal management, the scenario largely relied on passive ambient air cooling to maintain low cell temperatures. Therefore, the extent to which degradation-aware derating can perform solely as a thermal management technique, particularly for large stationary systems, is not definitive. Such systems would still require active thermal management equipment for safety reasons, such as to mitigate thermal runaway concerns. This is particularly important from an investor and insurance outlook. However, the constant ambient temperature maintained by the active thermal management scenario was not optimal in terms of degradation. This was confirmed by the Optimal Derated scenario, involving varying ambient and cell temperatures, achieving a lower degradation than the Optimal Derated + Active TM scenario, where the ambient temperature was kept constant.

The Optimal Derated + TM simulates a hybrid scenario where degradation-aware derating is implemented to manage the system’s lifetime while maximising profitability, but active thermal management is also simulated to manage safety concerns. This scenario is shown to be the most profitable, achieving the highest NPV and IRR. Active control of the ambient temperature increases the average cell temperature compared with the Derated and Optimal Derated scenarios. As shown in Figure 2, increased cell temperature increases the permissible charging C-rate, which allows the battery to cycle more and earn the most average annual revenue than the other derated scenarios. As a result, the system cycles the most under this scenario and incurs the highest cycle losses after the Base Case. However, by not imposing strict cycle limits but rather actively controlling the C-rate, the degradation-aware derating enables lower calendar losses than both Warranty scenarios. This means that despite more intense cycling, the Optimal Derated + Active TM scenario can maintain a comparable lifetime to Warranty 1 and earn more revenue. Once again, this illustrates the economic implications of more restrictive warranties. Warranty 1 and Warranty 2 do not fully capture the complexity of degradation, limiting economic performance while not providing significant improvements in the system’s lifetime.

5. Conclusions

The degradation-aware derating strategy explored in this work was successful in reducing calendar and cycle ageing losses, improving over traditional methods. Additionally, this form of derating was successful as an operational strategy aimed at maximising the LIB system lifetime while retaining economic performance. This was achieved by implementing a more complex view of degradation while harnessing the economic benefit of an extended lifetime. The derating strategy requires simple metrics for operation, which are monitored through most battery management systems, allowing for easy integration into existing systems. The adjustable degradation rates used in this strategy allow for economic optimisation and flexibility. This illustrates the benefits of flexibility that come from a better understanding of degradation. The integration of degradation rates has the potential to assist manufacturers with flexible warranty development and aid battery operators in enhancing the performance of large-scale Lithium-ion battery projects.

While the results of this project illustrate the significant impact that tuning the maximum degradation rates has on the economic and technical performance of the derating strategy, the potential of the flexibility this provides has not been fully explored. For example, an optimal degradation rate could be used in different scenarios throughout the battery’s lifetime depending on wholesale market conditions or other factors, such as the system operator’s shifting objectives. Integrating a degradation cost linked with a variable degradation rate into the arbitrage optimisation function is a potential area for future work to explore this idea further.