Repair and RMA
This page answers one question: when something fails, is it repaired or replaced — and how long does the claim take?
The commercial answer and the technical answer often differ. A board that can be repaired in four hours may still be replaced, because the site cannot wait four hours. What determines the right choice is not capability alone but the cost of downtime, the availability of the part, and — the factor most often overlooked — whether the intervention preserves or destroys your warranty position.
1. Repair or replace
| Repair | Replace | |
|---|---|---|
| Typical candidates | Power semiconductors, driver boards, control boards, BMS slave boards, communication modules, fans and cooling components | Battery modules, complete power modules, items where the failure cause is not conclusively identified |
| Downtime | Longer — the asset waits for the repair, unless a swap unit is available | Shorter — swap and go |
| Cost | Lower unit cost | Higher unit cost |
| Warranty implication | Depends entirely on authorisation — see section 3 | Normally clean, if the replacement is a genuine part fitted by an authorised party |
| Documentation burden | Higher — the repair itself must be evidenced | Lower |
| Best pattern | Repair off-line, hold the repaired unit as a swap spare | Swap first, decide the repaired-or-scrapped question afterwards |
The practical pattern on a live site is almost always: replace now, decide later. The failed unit leaves site, the asset returns to service, and the repair-or-scrap decision is taken off the critical path.
Cell-level repair — opening a module, testing individual cells, and replacing matched cells — requires clean working conditions, cell matching and characterisation capability, and specific safety infrastructure. BessRe’s cell-level repair capability is planned for 2027. Until then, module-level replacement is the honest answer and we say so.
2. How an RMA actually runs
An RMA is a state machine, not a form. Each state has an owner, an expected duration and a set of documents without which the next state cannot be entered.
| # | State | Owner | Typical duration | Documents required to exit |
|---|---|---|---|---|
| 1 | Fault recorded | Service provider | Same day | Fault description, serial number, event log extract, site and asset identification |
| 2 | Warranty validation | Manufacturer | 1–3 working days | Proof of purchase or commissioning date, evidence the installation and any prior work were performed by authorised personnel |
| 3 | Approved or rejected | Manufacturer | — | Written decision. A rejection should state the ground |
| 4 | Reverse logistics | Service provider | 2–10 working days | ADR classification under SP 376, correct packing instruction, dangerous goods documentation. See Replacement & spare parts |
| 5 | Assessment and disposition | Manufacturer or repair centre | Variable | Failure analysis where the manufacturer performs one |
| 6 | Outcome | Manufacturer | — | Repair, scrap, credit, or replacement despatched |
| 7 | Closure and archive | Service provider | Same day | Full record retained by asset, not by ticket |
Published field service benchmarks for standard RMA processing put the cycle at roughly 5–7 working days. They also record a finding worth pausing on: approximately 40% of customer chase-ups are caused by lack of visibility into RMA status rather than by the elapsed time itself. The claim is not late; the customer simply cannot see where it is.
That is a solvable problem, and it is why we report RMA status by asset on a standing cycle rather than waiting to be asked.
3. Every step of the RMA is warranty evidence
This is the section that costs owners money when it is skipped.
Manufacturer warranty terms are explicit about who may intervene. Sungrow’s terms provide that installation, commissioning and removal shall only be carried out by certified personnel authorised by the manufacturer or its approved service partners, and that warranty claims are submitted through the company’s Global Service Platform by the installer or authorised service partner. Independent guidance on the same terms notes that unauthorised repair, adjustment or modification voids the warranty, and that failure to cooperate with the manufacturer’s inspection and testing requirements can have the same effect.
The practical consequences are three:
- An undocumented intervention is indistinguishable from an unauthorised one. If you cannot show who did the work and under what authorisation, you are arguing from a position you cannot evidence.
- The claim channel is usually the authorised party’s, not the owner’s. If your O&M provider is not authorised, someone else has to file, and you have introduced a dependency into every claim.
- Cooperation obligations are real. Terms that require the owner to permit inspection or testing are enforceable, and a refusal — even an inadvertent one, such as scrapping a failed unit before the manufacturer has seen it — can end the claim.
Practical rule: do not scrap anything until the disposition is written down. A failed module in a container is an asset with an open claim. The same module in a skip is a loss.
4. What happens to what comes out
The removed unit has four possible destinations, and the choice is partly technical and partly regulatory.
- Repair and return to stock — where the unit is repairable and the economics work
- Return to manufacturer — under RMA, for repair, credit or failure analysis
- Harvest for parts — where the unit is beyond economic repair but contains serviceable components
- Disposal — where the unit is waste
For battery modules the disposal route is constrained by producer responsibility obligations and by the practical availability of licensed reception capacity, which in the Benelux is narrower than most owners assume. Modules classified as damaged or defective under ADR SP 376 face additional restrictions on storage and transport.
Throughout, the removed unit remains the owner’s property. We arrange the movement and the documentation; we do not take title and we do not take custody of the risk. That is a deliberate structure, and we recommend the same wording in any service contract you sign with anyone.
Illustrative scenario
A power module fails on a Flemish C&I site. It is swapped from local stock the following morning and the asset returns to service within one working day.
The failed unit is photographed in situ, its serial number recorded, and the relevant event log window exported before anything is disconnected. The RMA is raised the same day with the fault description, log extract and evidence that the original commissioning was performed under manufacturer authorisation. Warranty validation returns approved in two working days.
The unit is classified under SP 376 as stable defective, packed under P908, and shipped to the manufacturer’s European reception point. Failure analysis identifies a driver board fault; the manufacturer repairs and returns the unit, which enters local stock as the next swap spare.
Elapsed time from fault to claim closure: eleven working days. Site downtime: one working day. The two numbers are different, and only one of them affects revenue.
Illustrative scenario based on typical Benelux configurations — not a client reference.
How we work
We file claims through the manufacturer’s own channel where we hold partner status, because a claim filed through the correct channel by an authorised party is a claim that gets processed rather than argued. We document every intervention to the standard a warranty dispute would require, whether or not one is expected. And we report RMA status by asset on a standing cycle, because the benchmark data is clear that most chase-ups are about visibility, not delay.
Frequently asked questions
How long does an RMA take? Published benchmarks for standard processing are around 5–7 working days for the claim cycle. Reverse logistics adds time, and dangerous goods classification adds more. Site downtime and claim duration are different numbers — a well-run process closes the downtime long before it closes the claim.
Who pays for the return shipping? Set by the warranty terms and variable. Establish it before your first claim, not during it.
Will opening the unit for inspection affect my warranty? Potentially yes. Manufacturer terms commonly treat unauthorised repair, adjustment or modification as voiding cover. Inspection by an authorised party under a documented process is a different matter from an undocumented intervention. If in doubt, document first and open second.
Can you repair rather than replace? At board and component level, yes. At cell level, our capability is planned for 2027 and we will not claim it before it exists.
What if the manufacturer rejects the claim? A rejection should state its ground. Where the ground is a documentation gap, it is often recoverable. Where the ground is unauthorised intervention, it usually is not — which is the argument for getting the authorisation question settled before work starts. See Commissioning.
Talk to our team about RMA handling for your fleet — info@bess.re.
Repair or Replace? The 2026 EU Compliance & Local Delivery Guide
Repair economics, the compliance map, and local delivery capability — for Chinese BESS manufacturers entering Europe.
Download the whitepaper →