Battery procurement decisions are frequently made on a single figure: landed price per kWh of nameplate capacity. It's an understandable starting point — it's easy to obtain, easy to compare across suppliers, and easy to put in a tender scoring matrix. It also tells you almost nothing about what the battery will actually cost you over the life of the asset.
A total-cost-of-ownership view changes the comparison, and in ESC's experience it changes the ranking of suppliers more often than it doesn't.
What purchase price leaves out
Usable capacity, not nameplate capacity
Two batteries quoted at the same price per kWh of nameplate capacity are not necessarily equivalent if their recommended depth of discharge differs. A cell that should only be cycled to 80% depth of discharge to hit its rated cycle life delivers less usable energy per rand than a chemistry rated for deeper cycling — the comparison needs to be normalised to usable capacity, not the number on the spec sheet.
Replacement interval, informed by the real operating environment
As covered in ESC's note on operating temperature and LFP battery life, the same battery can have a materially different effective lifespan depending on where and how it's deployed. A TCO model that assumes the datasheet cycle life, rather than a site-adjusted estimate, will consistently understate the long-run cost of the cheaper option if that option is more temperature-sensitive.
Logistics — the cost that scales with remoteness
On an accessible urban site, the cost difference between a battery replacement in year 5 versus year 8 is mostly the capital cost of the battery itself. On a remote site, it also includes freight, import duties where applicable, a technician visit, and often a security or access cost. In ESC's experience this logistics cost is frequently underweighted in TCO models, and on hard-to-reach sites it can be large enough on its own to reverse a purchase-price-based ranking.
Warranty — real protection, or a document?
A longer warranty period looks like risk transfer to the supplier, but its value depends on enforceability: whether the supplier is likely to still be trading and able to honour claims in year six or seven, whether there's local service presence to action a claim rather than a distant head office, and what evidence the warranty actually requires. Claims are commonly disputed where BMS logs show the site operated outside the warranty's stated conditions — which loops back directly to the thermal and cycling assumptions built into the original specification.
The cheapest battery at the point of purchase is not reliably the cheapest system over its life. On remote, thermally demanding sites, ESC has seen the ranking between two suppliers reverse entirely once replacement interval and logistics are modelled explicitly — without either supplier's headline price changing.
Building a usable TCO model
- Normalise every quote to cost per usable kWh, not nameplate kWh, using the manufacturer's recommended depth of discharge.
- Replace the datasheet cycle-life assumption with a site-adjusted estimate that accounts for the actual thermal and duty-cycle profile.
- Include real freight, duties and site-access logistics cost for each expected replacement event, not just the first purchase.
- Discount future replacement and O&M costs appropriately (NPV terms) so the comparison reflects present value, not undiscounted totals.
- Treat warranty terms as a probability-weighted input, not a guarantee — informed by supplier track record and local service presence.
- Run the model with a sensitivity range on replacement interval — it is consistently the single biggest swing factor.
This doesn't need to be an elaborate financial model. A simple spreadsheet that makes these assumptions explicit, rather than implicit, is usually enough to change a procurement decision — and more importantly, to make the trade-off visible before it's locked in rather than after the first premature replacement.