A battery plant with hundreds of small inverters may require more frequent attention than one with a handful of large units, though it can still lose less capacity overall. For owners comparing string and central power conversion systems (PCS), however, the availability benefit is not always clear.

The distinction between central and string inverters matters when an availability claim becomes part of a project’s business case. Smaller, more frequent outages can be less disruptive, but the benefit depends on how often equipment fails and how quickly it returns to service.

What smaller converters change

The PCS converts electricity between the batteries and the grid. A central arrangement groups more battery capacity behind each large converter. A string arrangement divides that duty among more, smaller converters, allowing each affected section to be isolated while others continue operating.

In a simple comparison at equal installed power, if all converters have the same failure rate and repair time, splitting the plant into smaller units leaves the expected total loss of converter capacity unchanged. There are more units that can fail, but each failure removes a smaller share of the plant.

Smaller, independent outages can make available capacity more predictable and reduce the impact of any single incident. This can matter when the plant has committed to deliver a particular level of power.

This is not necessarily redundancy. If no spare capacity can take over a failed unit’s duty, the plant continues at reduced capacity.

Why unit MTBF can differ

Mean time between failures (MTBF) depends on what is inside each converter. A larger converter can contain more power modules and supporting circuits whose failure trips the entire unit.

In a parts-count reliability model, the failure rates of those independent, essential components add up. With identical components, equal stresses and constant failure rates, doubling their number halves the predicted unit MTBF. A smaller assembly can therefore have a higher unit MTBF, even while a plant containing many such assemblies experiences more faults in total.

There is a potential benefit to the smaller assembly. According to vendor claims, the average MTBF of string inverters can be around twice that of central inverters. There is no magic: fewer components can bring reliability benefits.

What a 50 MW example shows

To explore the difference, I compared two illustrative layouts for a nominal 50 MW project: 13 central converters rated at 4 MW each, and 218 string converters rated at 230 kW each. Their installed converter capacities are 52 MW and 50.14 MW, respectively.

I assumed a mean time between failures of 150,000 hours per central unit and 300,000 hours per string unit. These are modelling inputs based on statistics from current market players (most OEMs have this data available only under NDA. Exact numbers are therefore not publicly available, but this range covers the majority of vendors though individual performance will vary).

The example deliberately gives each string unit half the failure rate of a central unit; this is not an inherent property of either architecture.

For the string units, I assumed replacement within 24 hours of a known fault. I compared that with central units restored in either 24 hours, where fast replacement is possible, or 72 hours. Every failure in this example takes the affected converter fully offline.

Illustrative converter faults and capacity losses

These figures sum partial outages and express them as equivalent hours of the full 50 MW being unavailable. They do not mean the whole plant shuts down for that long and they also do not measure lost energy or revenue.

Under these assumptions, the string fleet has about 8.4 times as many faults, but each fault removes roughly one-seventeenth as much converter capacity. The resulting annual capacity loss is about 52% lower than for the central layout with 24-hour repairs, and 84% lower than with 72-hour repairs. The assumed unit reliability and restoration time explain that advantage.

The clearest operational lesson is the change within the central design. Reducing restoration time from three days to one cuts the modelled loss by two-thirds without changing the converter size.

The result changes when the assumptions change. If string-unit MTBF were also 150,000 hours, its annual loss would rise to 1.41 equivalent hours, close to the central result of 1.46. The remaining difference reflects installed capacity. Keeping string MTBF at 300,000 hours but extending restoration from 24 to 72 hours raises its loss to 2.11 hours, worse than the central layout repaired within 24 hours.

In real-world scenarios, a trained on-site OEM team may be able to restore string inverters in about 5 to 6 hours. That possibility is not included in the model above.

The service plan matters

A 24-hour string replacement assumption requires a compatible spare on site, trained people, safe access and the ability to return the unit to operation. If the operator waits for a scheduled maintenance visit, the small fault can remain unresolved for much longer. A layout with frequent small incidents needs an operation and maintenance contract that can respond to them.

Central designs deserve the same scrutiny. Replaceable modules and local support can shorten repair time. If a particular design also keeps part of a converter running during repair, that capability should be included in its assessment. The full-unit outages assumed here would not represent it fairly.

The example assumes independent converter faults at a constant rate during normal operating life. It does not forecast early failures or ageing, and it excludes detection delays and faults in shared equipment. A transformer fault, common control problem or shared protection trip can affect several converters together. More string units do not remove those dependencies.

Overall

When reviewing an availability claim, I would ask how much capacity each credible fault removes, who restores it and what evidence supports the promised restoration time. The contract should reflect the string or central design, with appropriate spare-part delivery and response times.

String PCS can offer a useful combination of smaller outages and quick replacement. Owners will capture that benefit when spare parts, staffing and response obligations support it. I would give those commitments as much attention as the architecture itself when comparing bids.

Sergey Syrvachev is Director of Product Engineering at HyperStrong.