String Inverter vs Central Inverter for UK Utility-Scale Solar

The string inverter vs central inverter decision is arguably the most consequential technical choice an EPC contractor will make on any utility-scale solar UK project. According to data from Solar Power Europe, over 14 GW of new solar capacity was installed across Europe in a single recent quarter, and the UK remains one of the most active ground-mount markets on the continent. Yet despite this scale of deployment, inverter architecture selection is still frequently driven by procurement familiarity rather than rigorous engineering analysis. That is a costly habit. The wrong inverter configuration on a 50 MW UK ground-mount site can translate into yield losses exceeding 2.5% annually, O&M cost overruns of tens of thousands of pounds per year, and G99 compliance headaches that delay grid energisation by months. This article cuts through the generic product-sheet comparisons and gives EPC engineering teams the technical depth needed to make the right call on every project.

String Inverter vs Central Inverter: Understanding the Core Architecture Difference

The fundamental distinction between string and central inverter architectures lies in where MPPT (Maximum Power Point Tracking) happens relative to the array. In a string inverter solar farm layout, each inverter unit handles a small group of series-connected modules, typically between 10 and 30 panels per string, and each inverter runs its own independent MPPT algorithm. A 50 MW site might deploy 400 to 600 individual string inverter units, each rated between 80 kW and 125 kW, tracking the optimal power point for its dedicated strings in real time.

By contrast, a central inverter solar farm consolidates the DC collection from an entire block, often 2 MW to 5 MW per unit, into a single high-power conversion stage. Strings are combined in DC combiner boxes, and the resulting parallel DC bus feeds one MPPT channel (or a small number of MPPT inputs in modern central units) before conversion to AC. The architectural consequence of this topology is that a single poorly performing string, affected by soiling, shading, or module degradation, drags down the entire block’s performance rather than being isolated to one small inverter unit.

For engineers working on inverter configuration UK solar projects, understanding this distinction at the electrical level, rather than just the commercial level, is the starting point for every subsequent technical decision on the project.

Distributed MPPT vs Single MPPT: What It Actually Means for UK Sites

Here is a real-world scenario that illustrates the practical stakes of this distinction. Consider a 20 MW ground-mount site in Lincolnshire with a gently undulating terrain profile. On the eastern edge of the site, morning fog from a nearby drainage channel causes intermittent partial shading on roughly 8% of modules between 07:00 and 09:30 during autumn and winter months. With a central inverter architecture, those shaded strings pull down the MPPT operating point for the entire 2 MW block they belong to, causing the unshaded 92% of that block to operate below its true maximum power point. The actual energy loss is not 8% of the block; it compounds to between 12% and 18% of block output during those shading hours, depending on the IV curve mismatch severity.

With a distributed string inverter architecture, each unit tracks its own strings independently. The shaded strings operate at their own reduced power point, while the unshaded strings on adjacent inverters run at full MPP. The energy loss is contained to the directly affected strings. This is not a theoretical advantage. It is measurable in PVsyst simulation outputs and in operational SCADA data from deployed UK sites, and it is one of the primary reasons that sites with irregular terrain, nearby tree lines, or phased panel cleaning schedules are increasingly specified with string inverters by experienced EPC contractors.

G99 Compliance Implications: How Inverter Topology Affects Grid Code Obligations

For any EPC contractor solar UK working on projects above 1 MW, compliance with Engineering Recommendation G99 is not optional and the inverter architecture choice has direct implications for how that compliance is demonstrated to the DNO. This is an area where many procurement teams underestimate the complexity until they are already in the grid connection approval process.

For detailed guidance on navigating the grid connection process, see how to pass grid connection approval for UK solar projects.

Harmonic Distortion Limits and Topology-Specific Challenges

G99 imposes strict Total Harmonic Distortion (THD) limits at the Point of Common Coupling (PCC). Central inverters, by virtue of their large transformer-coupled output stages, tend to produce lower harmonic content on a per-unit basis and often have well-established type-test documentation that satisfies DNO technical review teams quickly. This is partly because central inverter manufacturers have been engaging with grid code requirements for longer and have mature compliance portfolios.

String inverters, however, present a more complex picture at utility scale. A 50 MW site running 500 individual string inverter units is essentially a distributed harmonic source array. Each unit may be individually compliant with IEC 61727 and EN 50549-1 harmonic limits, but the aggregated harmonic injection at the 33 kV transformer LV busbar needs to be assessed as a combined system. DNO engineers reviewing G99 applications for large string inverter arrays will typically request harmonic summation studies, and the contractor must demonstrate that the aggregate THD remains within the limits specified in ER G99 Section 12 even under partial generation conditions.

This is not a reason to avoid string inverters at utility scale. It is a reason to commission those harmonic summation studies early and to work with inverter manufacturers who can provide measured field data from comparable UK installations rather than just simulation results.

Reactive Power Capability and Power Factor Control

G99 also requires generators above certain thresholds to provide reactive power capability, typically a power factor range of 0.95 leading to 0.95 lagging at the PCC. Both central and string inverter architectures can meet this requirement, but the implementation differs significantly. Central inverters provide reactive power from a single large controlled source, making reactive power dispatch straightforward to model and demonstrate in the G99 application. String inverter deployments must coordinate reactive power across hundreds of units, which requires either a centralised plant controller communicating with each unit via Modbus or SunSpec protocols, or a hierarchical control architecture where zone controllers aggregate reactive power commands.

The plant controller specification and commissioning costs for a large string inverter array are non-trivial and must be included in the BOS cost model from the outset. A plant controller system for a 50 MW string inverter site can add between £80,000 and £150,000 to project costs, depending on complexity and the number of communication nodes required.

PVsyst Modelling Differences: Why Your Simulation Methodology Matters for Inverter Selection

One of the most technically important and most frequently mishandled aspects of the solar PV inverter comparison process is how each architecture is modelled in PVsyst. EPC engineers who use a single generic modelling approach for both central and string inverter configurations are building systematic error into their yield assessments, and that error flows directly into IRR calculations and lender technical due diligence reports.

For a deeper understanding of how PVsyst simulations are structured for UK solar projects, see the PVsyst shading analysis standard applied to UK solar projects and explore specialist PVsyst services for photovoltaic projects.

Inverter Clipping Loss Modelling for Each Architecture

PVsyst calculates inverter clipping losses (referred to as overload losses in the software) based on the ratio of peak DC power to inverter AC rated power, the DC/AC ratio. For a central inverter configuration, this ratio is applied to a large block simultaneously. The clipping loss model assumes that when DC power exceeds the inverter’s AC rating, the operating point is shifted to reduce output, a single-point event affecting the whole block.

For a string inverter configuration, each inverter unit clips independently based on the DC power from its specific strings. This means that on a partially cloudy day, some string inverters may clip while others are running below their rated capacity. The aggregate clipping loss is therefore lower than a naively modelled central inverter equivalent, because the distributed architecture naturally smooths the DC power envelope across hundreds of independent units. PVsyst handles this through the “inverter array” modelling approach, where multiple string inverters are modelled as sub-arrays, but this requires the engineer to correctly define the string groupings and ensure the software is not aggregating them into a single virtual central inverter. Errors in this setup are surprisingly common and lead to either overstated or understated P50 yield estimates.

MPPT Range Modelling and UK Irradiance Profiles

The UK’s irradiance profile is fundamentally different from southern European sites. The UK has a high proportion of diffuse irradiance, lower peak GHI, and frequent low-irradiance operating conditions, particularly between October and February. String inverters typically have wider MPPT voltage ranges (often 200V to 1000V) compared to older central inverter designs (500V to 850V operating range on some legacy units), and this matters enormously for UK operating conditions.

During low irradiance periods, the PV array’s open-circuit voltage (Voc) remains relatively stable but the MPPT operating voltage can drift toward the lower end of the range. Inverters that fail to track effectively below certain DC input voltage thresholds will shut down or de-rate during these periods. In a UK climate, where diffuse irradiance conditions may represent 30% to 40% of annual energy production hours, MPPT range performance is directly correlated with annual energy yield. PVsyst allows engineers to model this through the inverter’s efficiency curve and threshold parameters, but the accuracy of the simulation is only as good as the inverter datasheet inputs. EPC engineers should always request extended efficiency curve data from manufacturers covering operation at 10%, 20%, and 30% of rated power, not just the peak efficiency figure that appears on the product sheet.

BOS and Cabling Cost Differential: The Numbers That Change at Different Project Scales

The BOS cost comparison between string and central inverter architectures is not static. It shifts materially depending on project scale, site geometry, and AC voltage levels chosen for the collection network. Understanding where the crossover point lies is essential for any inverter selection solar process on UK utility projects.

For a comprehensive overview of how system design parameters interact with BOS cost decisions, see how a PV solar system is designed for utility-scale applications.

DC Cabling: The Central Inverter Penalty on Large Sites

A central inverter architecture demands long DC home runs from the array to the inverter building. On a 50 MW site where the central inverter station is located in the centre of the site, DC cable runs can reach 300 to 500 metres from the most distant rows. At DC currents of 600A to 1200A common in central inverter DC combiner circuits, voltage drop across these runs becomes a significant loss mechanism. A 1% voltage drop on a DC home run represents a direct energy loss, and on a 50 MW site with a 25-year asset life, that loss compounds to a material NPV impact.

String inverters eliminate long DC runs almost entirely. Each unit is mounted on or near the racking structure, typically within 30 to 50 metres of its connected strings. DC cable cross-sections can be smaller, and voltage drop losses on the DC side are dramatically reduced. However, this architecture shifts complexity and cost to the AC collection network. With 400 to 600 string inverter outputs each producing 400V or 800V AC, the AC combiner network, including MV transformers, AC switchboards, and protection relays, becomes more elaborate than the equivalent central inverter AC connection scheme.

The AC Combiner Network Cost Reality for String Inverter Arrays

A frequently underestimated cost in string inverter project budgets is the AC combiner infrastructure. Each cluster of string inverters, typically 8 to 16 units, feeds into a cluster box or AC combiner panel, which in turn connects to a pad-mounted MV transformer. The procurement cost of these combiner panels, the additional MV transformer units required to serve distributed AC collection points, and the LV AC cable required to connect string inverter outputs to those combiners can add between £40,000 and £70,000 per MW to BOS costs on a large UK ground-mount project compared with a central inverter layout using bulk DC collection and fewer, larger MV transformers.

The economic crossover between the two architectures on a pure BOS cost basis typically occurs around the 20 MW to 30 MW project scale, where the DC cable penalty of central inverter layouts begins to be outweighed by the AC combiner complexity costs of large string inverter arrays. Below that scale, string inverters generally have a lower total BOS cost. Above it, the comparison is more nuanced and requires project-specific cable quantity takeoffs rather than generic assumptions.

Maintenance, Redundancy, and O&M Cost Modelling Over a 25-Year Project Lifetime

The O&M implications of inverter architecture selection are where the long-term financial case for each technology becomes most visible, yet they are routinely underdeveloped in project financial models submitted to lenders and investors.

Central Inverter Single-Point-of-Failure Risk Profiling

A 2 MW central inverter failure on a 50 MW site takes 4% of the plant offline for the duration of the repair or replacement event. The mean time to repair (MTTR) for a central inverter failure ranges from 3 days for a minor component replacement with parts in stock to 6 to 8 weeks for a major transformer or IGBT module failure requiring factory-level intervention or a custom replacement unit. During that downtime window, lost generation revenue accumulates rapidly. At a wholesale power price of £60/MWh and assuming full capacity factor conditions, a 2 MW central inverter offline for 4 weeks represents approximately £115,000 in lost revenue, before accounting for any balancing or imbalance costs for contracted sites.

Manufacturers typically publish Mean Time Between Failures (MTBF) for central inverter units of between 80,000 and 120,000 hours for the electronic components, but the mechanical and thermal management systems, cooling fans, heat exchangers, and capacitor banks, have shorter replacement cycles of 5 to 10 years. Proactive O&M contracts for central inverter sites must budget for these scheduled component replacement events, which for a site with 10 central inverter units can run to £15,000 to £30,000 per maintenance cycle.

String Inverter Modular Replacement and Fleet Redundancy

The O&M profile of a string inverter array is structurally different. When a string inverter fails, the maximum generation loss is the output of that single unit, typically 80 kW to 125 kW on a large-scale deployment. On a 50 MW site, that represents 0.16% to 0.25% of total capacity. Replacement units can typically be sourced within 24 to 72 hours from the manufacturer’s local depot or the contractor’s own spares holding, and installation is a routine field task requiring no specialist lifting equipment or factory support.

Well-run O&M programmes on string inverter sites maintain a rolling spares holding of 1% to 2% of total inverter count on site, enabling same-day or next-day replacement of any failed unit. The annual string inverter failure rate on modern units (post-2018 manufacturing) typically runs between 0.5% and 1.5% of fleet units per year over the first 10 years, rising toward 2% to 3% in years 15 to 25 as capacitor and fan components age. A realistic 25-year O&M cost model for a 500-unit string inverter fleet on a 50 MW site should budget for approximately 50 to 100 unit replacements over the project life, at a unit replacement cost of £1,500 to £2,500 per inverter including labour. This is a predictable, linear cost curve, which is far easier to model and finance than the unpredictable large-scale failure events that characterise central inverter O&M risk.

Inverter Warranty Duration and Technology Obsolescence Risk

A factor that rarely receives adequate attention in the string vs central inverter debate is technology obsolescence over a 25 to 30-year project life. Central inverters from major manufacturers have product support commitments of 10 to 15 years from manufacturing date, which means that a central inverter installed at project financial close may have its parts supply discontinued before the project reaches its halfway point. The cost of sourcing replacement parts for an obsolete central inverter unit can be 3 to 5 times the original component cost, and in some cases, a full unit replacement becomes the only economically rational option.

String inverters, being a higher-volume product category with more competitive manufacturers, have historically maintained better parts availability across product generations. Additionally, the modular nature of string inverter deployment means that individual unit upgrades, replacing older units with more efficient next-generation models, can be incorporated into scheduled maintenance cycles without requiring the kind of large-scale plant outage and recommissioning event that a central inverter retrofit demands.

Making the Final Inverter Selection Decision: A Framework for UK EPC Contractors

Given the technical depth of the considerations above, it is worth consolidating a decision framework that EPC engineering teams can apply at the pre-FEED stage of a UK utility-scale project. There is no universal correct answer, but the following factors consistently tilt the decision in one direction or the other.

Projects that typically favour string inverter architectures include:

• Sites with irregular terrain, multiple aspect orientations, or nearby shading sources such as trees, hedgerows, or agricultural structures where distributed MPPT delivers measurable yield advantage
• Projects below 30 MW where BOS cost advantages of string topologies are strongest
• Sites where O&M access is constrained or where the client requires minimised single-point-of-failure exposure in the asset management contract
• Projects with complex G99 reactive power dispatch requirements where a sophisticated plant controller is already part of the scope
• Phased development sites where generation can begin on completed phases without waiting for the full central inverter infrastructure to be commissioned

Projects that typically favour central inverter architectures include:

• Very large flat sites above 50 MW with uniform module orientation and minimal shading where the MPPT advantage of distributed architecture is marginal
• Projects where the developer has an existing central inverter O&M service agreement and in-house technical capability for large unit maintenance
• Grid connection points where the DNO has specific harmonic injection requirements that are more straightforwardly demonstrated with fewer, larger inverter units
• Sites in remote locations where logistics costs for transporting multiple small string inverter units significantly exceeds the cost of delivering fewer large central inverter units

Conclusion: Getting the Inverter Architecture Decision Right from Day One

The string inverter vs central inverter debate in utility-scale solar UK projects is not a question that can be answered with a single universal recommendation. It requires project-specific analysis of terrain and shading conditions, accurate BOS cost modelling at the specific project scale, a realistic assessment of G99 compliance implications for each topology, properly differentiated PVsyst simulation methodology, and a long-term O&M risk model that accounts for the fundamentally different failure profiles of each architecture over a 25-year asset life.

What is clear is that the era of defaulting to central inverters for anything above 10 MW out of engineering conservatism is over. Modern string inverter technology at utility scale is mature, G99-compliant, bankable, and in many UK site conditions, demonstrably superior in energy yield performance and O&M risk profile. Equally, central inverters retain genuine advantages on very large, geometrically regular sites where distributed MPPT provides marginal benefit and where operational simplicity is a priority for the long-term asset owner.

The EPC contractors who will build the most competitive and best-performing UK solar portfolios over the next decade are those who apply rigorous, project-specific engineering analysis to this decision on every project, rather than relying on procurement habit or vendor preference. The tools, simulation methodologies, and grid code frameworks to make that analysis rigorously are all available. The competitive advantage lies in using them consistently and with the technical depth that project finance lenders, DNOs, and long-term asset owners increasingly expect.

Heval Arcan