Electric vehicles, laptops, phones, cordless hoovers, robot dogs (what?). They all have one thing in common. Lithium-ion batteries power them, and their capacity for storing energy depletes over time. Grid-scale storage faces the same laws of physics as your robotic canine friend - batteries degrade depending on a range of factors, including environment (e.g. temperature), time and usage (e.g. duty cycle, depth-of-discharge and more). For asset owners, managing battery degradation is asset management. In this thread, we explain the key considerations.

Note - this is a hotly researched topic, and as with any immature technology, the world is still figuring it all out...

What is battery degradation?

Lithium-ion cells operate by ion movement between positive and negative electrodes. This concept is as old as the first batteries (over a century), and in theory, this mechanism should work forever. But, of course, theory ≠ reality. Instead, the components of a battery cell (electrodes, electrolyte, current collectors and additives) go through physical and chemical changes during operation, which reduce a battery’s storable energy (capacity) and reduces the maximum deliverable power of the cell.

These physical changes reduce performance and can occur faster or slower, depending on how the cells are operated.

Put simply, operating batteries ‘vigorously’ reduces their performance faster than operating them ‘lightly’.

Metrics, metrics, metrics

Cycles - ‘cycles’ is a widely used term to count the number of full charges and discharges of a battery. This figure tends to be net and additive. For example, you don’t need to do a full 100% charge in one go. Instead, this could be through lots of smaller charges that all add up to 100% (i.e. one cycle). When talking about cycles, keep in mind where you measure it - some folks count cycles at the battery cells (i.e. DC side of the power converters), and others measure it elsewhere (AC side, metered grid connection etc.). These nuances can include or exclude the rest of the system's losses, which can completely change the metric.

State of Charge (SoC) - the level of charge in a cell relative to its capacity, usually shown in % (0% = empty, 100% =full). For some lithium-ion cells, operating within a restricted range can extend the lifetime of the cell. For example, only operating the cell between 20% and 80% SoC.

Depth of Discharge (DoD) - a figure of merit sometimes used instead of SoC, which shows measures how much charge has been removed from the battery relative to the total amount of charge (i.e. DoD = 100% - SoC)

State of Health (SoH) - a figure of merit on a cell's condition compared to ideal conditions, usually shown in %. This metric is often used alongside energy capacity as an indicator of degradation. On day 1 of commissioning, you would expect SoH to be pretty darn close to 100%. An asset that has been in operation for some time will have a SoC of <100%.

Temperature (T) - temperature has a huge impact on cell performance and degradation. Cells can get very hot very fast, so cooling systems are firmly under the microscope here. It’s also important to keep cells above a minimum temperature (typically 10-20 deg C) for optimum performance.

What data do we have on battery degradation?

Let’s start with academic research:

(Applicable) battery degradation data is sparse. Sure, there’s plenty of test data from academic research into cell degradation. But, this research tends to be for single or small groups of cells and is mostly for other use cases (such as electric vehicles, aviation, consumer electronics), which have different characteristics to grid-scale storage (e.g. cell arrangements, temperatures, load profiles). There’s also a time, cost and space element to lab testing that means that researchers often have to simulate and make assumptions.

Testing degradation over 10 years takes, well, 10 years. Anyone fancy stretching a PhD over 10 years? Thought not.

It’s common practice to simulate, extrapolate or speed up cycling (10 years cycling in 1 year, perhaps?). The cost side has challenges, too- procuring a 40-foot shipping container full of battery cells (only to degrade them) is an expensive endeavour, so research tends to scale things down to more manageable test rigs. This is sensible but does still miss the additional dynamics of real-world operations - thermal loading and cooling systems inside containers, seasonal temperature variations, outages and the like. Certainly worth a nod here to the Centre for Research into Electrical Energy Storage and Applications (CREESA), led by the University of Sheffield, who have procured, built and operate a 5MW energy storage system purely for academic research purposes - kudos. However, we can’t apply data from the CREESA system to other assets in the UK because it uses lithium-titanate cells, an undoubtedly superior cell chemistry to most, but also very expensive and consequently not deployed elsewhere in the UK. So, in summary of academic research into cell degradation - loads of fantastic work going on, but there remains separation between research and real-world usage for grid-scale energy storage. Applicable data for utility-scale energy storage asset degradation remains elusive.

What can we learn about battery degradation from operational assets?

The GB energy storage market is maturing; circa 1GW of grid-scale energy storage installed across ~60 assets in the last 5 years provides plenty of data points for ‘real-world applications. However, asset owners and manufacturers tend to keep this data to themselves. Most operational assets in the GB are aged between 1-4 years and have operated mostly in frequency response services (like FFR).

Based on anecdotal accounts of cell warranty agreements, we believe most operational assets in GB to have a retained usable capacity of 90-95% (i.e. battery degradation of 5-10%). But, this is warranted, not actual.

What do manufacturers say about battery degradation?

On the whole, manufacturers tend to take a conservative approach and specify lithium-ion cells' life as a number of full charge/discharge cycles, but this is a simplification of real-world operation. A rule of thumb in the industry - expect to retain between 75-85% of energy capacity in 10 years for your energy storage asset operating in 1.5-2 cycles per day (for a 1-hour system). This is (of course) subject to tons of assumptions regarding specification, load profile and usage.

Measuring battery degradation

Over the course of your asset’s lifetime, measuring battery degradation will be crucial to you being an ‘informed operator’ and especially important if battery degradation forms part of your cell warranty contract. With extended warranties now being offered of up to 15 years duration, regular measurements are vital. In some cases, battery degradation tests can determine whether an OEM must replace equipment or pay compensation (£££), so the testing regime must be agreed to upfront and written into the contract.

The most comprehensive way to measure battery degradation is to complete an energy capacity test. The system is charged to 100% SoC and then discharged continuously at a set power until 0% SoC is reached (i.e. 1-hour discharge at rated power from 100%-0% SoC). This is a fantastic test of performance because it’s incredibly ‘vigorous, which will also test the ancillary services (such as cooling). However, these tests themselves will cause some battery degradation (imagine picking your car up from an MOT to find it has an extra 1000 miles on the clock). To reduce these tests' stress, some OEMs advise administering this test as under 50% rated power or less or restricting the SoC levels (i.e. discharge from 80%-20% SoC and extrapolate). The jury’s out.