Overbuilding the grid connection: what more module capacity than connection capacity means for yield and investors

A solar park with 1.4 megawatts peak of modules that may feed in at most 1 megawatt: to a layperson that sounds like money left on the table. In fact it has been standard practice for years. Modules deliver their rated output only on a few clear middays; for most of the year, a connection sized for the full module capacity sits half unused. Building more modules onto the same connection loses some electricity on those middays and gains it in all other hours.

For investors the topic has become more tangible in recent months. Since February 2025, operators can fix the feed-in limit contractually with the grid operator; electricity above it then no longer has to be taken off. The draft of Germany's so-called grid package (Netzpaket) would make overbuilding mandatory for new ground-mounted parks. Anyone investing in a solar park should therefore know how far the park is overbuilt, what that costs and what the connection contract says.

What does it mean when a solar park overbuilds its grid connection?

It means the module capacity is larger than the capacity the park may feed in at the grid connection point (Netzverknüpfungspunkt). Two figures are set against each other. Under the EEG, installed capacity is the rated DC output of the modules (§ 3 no. 31 EEG); the inverters do not count, as the Clearingstelle EEG|KWKG confirms. The agreed feed-in capacity is set out in the connection contract (Netzanschlussvertrag) and limits how much AC power may flow into the grid.

The ratio of the two is usually called the DC/AC ratio in yield reports. A value of 1.0 means no overbuilding, 1.43 means a feed-in limit at 70 percent of module capacity, and 2.0 means twice as much module as connection capacity. Overbuilding has long been the norm: according to Fraunhofer ISE, module capacity in Germany is on average just under 10 percent above inverter capacity; at large south-facing power plants the inverters typically feed in at around 85 percent of module capacity, a ratio of roughly 1.18.

The limit can arise in three ways: through smaller inverters, through power control at the connection point or through a contractual agreement with the grid operator. For the kilowatt hours it makes no difference. Legally it does, as the section on § 8a EEG explains.

Why do projects build more modules than the connection can take?

Because the grid connection is the scarcest part of a solar park. Grid operators work through connection requests in order and cannot keep up with the volume. For battery storage at medium voltage and above alone, the Federal Network Agency reports around 9,700 connection requests totalling about 400 gigawatts in 2024; around 25 gigawatts were approved in the same year. The federal government describes the same order of magnitude in its explanatory memorandum to the grid package and calls it a flood of applications that blocks other applicants.

Whoever has a connection point therefore wants to use it for as many hours a year as possible. Price matters too: the hours in which the modules could deliver more than the connection takes are sunny middays in spring and summer. That is when solar power is cheapest on the wholesale market. More and more often the price is even negative; a new park is curtailed then anyway, as Negative electricity prices: what they mean for solar and storage investors explains. What is clipped on those middays is worth little.

How much electricity and revenue does clipping cost?

Less than most people expect, as long as overbuilding stays moderate. We held a grid connection of 1 megawatt fixed and raised the module capacity step by step from 1.0 to 2.0 megawatts peak, once for a south-facing array tilted at 25 degrees and once for an east-west array at 15 degrees. The basis is hourly generation profiles for a site in central Germany and actual day-ahead prices for 2024 and 2025. In hours with a negative price the park does not feed in, as is usual for new plants under § 51 EEG.

Two patterns stand out. First, the park always loses less revenue than yield, because the clipped hours are the cheapest of the year; in 2025 this effect was even stronger than in 2024 because of the many midday hours with very low prices. Second, orientation matters a great deal. In our profile a south-facing array exceeds 70 percent of its module capacity in 182 hours, an east-west array in only 8 hours. Its output is spread more evenly across the day, so it can be overbuilt up to 1.43 practically without loss. The reasons for these profiles are explained in South, east-west or vertical: which solar park layout pays off for investors.

One caveat belongs here: hourly values smooth out short output peaks, for example when the sun breaks through clouds and the modules deliver more for a few minutes than the hourly mean. In reality clipping is therefore somewhat higher than in the table. A yield report should calculate it at quarter-hour or minute resolution.

Does the additional module pay off for the investor?

Usually yes, well beyond 1.4, but each additional module earns a little less than the one before. The right question is therefore not how many percent are lost, but what the additional kilowatt peak still earns and what it costs. The following table shows this for the south-facing array, per kilowatt of grid connection.

Up to 1.43, each additional kilowatt peak earns almost as much as the first. Revenue per kilowatt peak across the whole park barely falls: from 57.3 to 57.1 euros, and to 54.3 euros at 2.0. Between 1.6 and 2.0 it is still 78 percent in 2025, but only 72 percent at 2024 prices. The connection, by contrast, is used much better: at 2.0 it feeds in almost twice as many kilowatt hours a year as without overbuilding.

For returns this means: the additional module needs no additional grid connection, no larger substation and usually no further transformer. If an additional kilowatt peak costs clearly less than an average kilowatt peak of the park, it pays off even if it earns only three quarters as much. Two counterarguments belong in an honest picture. More modules tie up more capital and more land. And yield per kilowatt peak falls. Anyone comparing parks by purchase price per kilowatt peak therefore needs to know how far each is overbuilt; two parks with the same price per kilowatt peak can feed in very different amounts.

This leads to the most important check: the yield report must include clipping at the feed-in limit. A report that simply multiplies module capacity by the site's specific yield overstates a heavily overbuilt park. How to read a yield report is described in Reading a yield assessment: what P50 and P90 mean for a solar park direct investment.

What do § 8a EEG and § 17(2b) EnWG regulate?

They allow the feed-in limit of a grid connection to be fixed contractually with the grid operator. Both provisions came with the so-called solar peak act (Solarspitzengesetz, BGBl. 2025 I No. 51) and have applied since 25 February 2025. § 8a EEG covers renewable energy plants, § 17(2b) EnWG all connectees, including storage and consumers.

Under § 8a(1) EEG, grid operator and plant operator may agree a connection-side limit on maximum active power feed-in. The limit may apply permanently or only in certain time windows; its level may differ per window. The operator must keep to it at all times by technical means. § 17(2b) EnWG also distinguishes static and dynamic limits; according to the explanatory memorandum, models in which the grid operator lowers output depending on grid conditions down to an agreed floor are also possible.

What the agreement must contain is listed in § 8a(2) EEG: the level of the limit, any time windows, the duration and what applies afterwards, technical implementation, liability for exceeding the limit and, where several operators share one connection point, their consent. For that case the law also requires rules on joint responsibility and joint and several liability.

Three points are decisive for investors. First, the agreement is voluntary: the grid operator may offer it but does not have to. It only has to check whether one would be possible at the nearest connection point if it assigns a more distant one (§ 8a(3) EEG). Second, § 8(2) EEG expressly allows a connection point already used by another plant to be shared, provided that plant's operator consents; the industry calls this cable pooling.

The third point matters most to investors: where a flexible connection agreement exists, the grid operator only has to take off electricity generated within the agreed limit (§ 11(1) sentence 3 EEG). Anything above falls outside the take-off duty. It follows that clipping at the agreed limit is not curtailment by the grid operator that would be compensated under § 13a EnWG; it is a consequence of the park's own design. If the park is curtailed below the limit because of a grid bottleneck, the usual rules in Curtailment of a solar park: when the grid operator pays and when it does not apply. For dynamic models the contract decides; § 8a EEG and § 17(2b) EnWG contain no express compensation rule for them.

What changes if the grid package is passed?

For new ground-mounted parks, overbuilding would become mandatory. Under the government draft, ground-mounted solar plants connected from 1 January 2027 would only have a right to grid connection with a feed-in limit of 70 percent of installed capacity (§ 8(1) sentence 4 EEG-E, Drucksache 21/7866, p. 14). Where several plants share a connection point, the limit applies to their combined feed-in. According to the explanatory memorandum, it is meant to apply permanently and regardless of bottlenecks and to encourage east-west arrays with storage (p. 58).

Translated into our calculation: a 70 percent limit corresponds to a DC/AC ratio of 1.43. A south-facing array loses about 1 percent of yield and 0.3 to 0.8 percent of revenue as a result, an east-west array practically nothing. According to Fraunhofer ISE, large south-facing parks today typically sit around 1.18; for them the limit means a small yield loss or, read the other way, the chance to build more modules on the same connection. Existing parks would not be affected because the rule applies to connections from 2027.

The transition is the main open question. The Bundesrat has asked for the limit to apply only from 2030, to be confined to capacity-limited grid areas and to exempt projects permitted or fully applied for by the end of 2026. The federal government rejects confining it to bottleneck areas but is examining a transition rule (Drucksache 21/8299, pp. 21 f. and 33 f.). For a park planned today that connects in 2027, this can decide whether the limit applies.

The draft also enshrines a right to overbuild: several plants may share a connection point even if their total exceeds the feed-in limit. This also applies to a plant that has yet to be commissioned (§ 8(2) EEG-E). Storage at an existing connection could no longer be refused for lack of capacity as long as the previous connection capacity stays unchanged. For grid areas with heavy curtailment the draft has separate rules, described in Curtailment of a solar park: when the grid operator pays and when it does not. And under the Federal Network Agency's draft for the new grid fee system, a feed-in charge from 2029 would be calculated on the agreed feed-in capacity; an overbuilt park pays correspondingly less per kilowatt peak, as Grid fees for generators from 2029: what Germany's planned feed-in charge costs a solar park and who is grandfathered shows.

What happens when storage uses the same connection?

A battery behind the same feed-in limit can absorb clipped solar power and feed it in during the evening when the connection is free. How much that earns depends on the degree of overbuilding. At 1.43, a south-facing array produces so little surplus that a battery barely earns anything from it: in our calculation 0.8 to 1.4 euros per kilowatt of connection per year. At 2.0 the picture changes. A battery with half the connection capacity and four hours of storage then shifts 76 to 80 percent of the clipped volume into later hours and earns 15 to 17 euros per kilowatt of connection a year from it.

The rescued midday energy alone does not carry a battery; at usual overbuilding ratios it is too small. A battery at a solar park earns its money mainly from price spreads on the power market and from system services. The surplus from overbuilding is a side income. How park and battery work out on one connection is covered in Co-located vs. stand-alone: which gives the better risk structure and Retrofitting battery storage at a solar park: what does the MiSpeL ruling change?.

Legally, a shared connection brings an obligation investors should know about. If park and battery belong to different operators, both must consent to the agreement, jointly ensure the limit is kept and are jointly and severally liable for breaches (§ 8a(2) sentence 2 EEG). Anyone investing at a shared connection should therefore know who controls output, who pays in case of a fault and how connection capacity is split in hours when both want to feed in.

Which questions should investors ask about the grid connection?

• What is the module capacity? Which feed-in capacity is stated in the connection contract? Which DC/AC ratio results?
• Does the limit come from the inverters, from power control or from a flexible connection agreement under § 8a EEG?
• If there is an agreement: is the limit static or dynamic? Does it apply permanently or until a grid expansion? What applies afterwards?
• Who may reduce feed-in and down to what level? Is compensation regulated in the contract?
• What liability does the operator face if the limit is exceeded? How is compliance ensured technically?
• Does the yield report include clipping at the feed-in limit? At what time resolution is it calculated?
• Do other plants or a battery share the connection point? Who controls the combined output? How is joint and several liability settled between the operators?
• Is the park expected to connect on or after 1 January 2027? Does the model account for the planned 70 percent limit and possible transition rules?

General checkpoints for any provider are listed in How to tell a trustworthy provider of energy direct investments.

How we check grid connection and overbuilding

For every solar park we present to investors, we compare module capacity with the feed-in capacity in the connection contract and record the DC/AC ratio. If it is above 1.2, we check whether the yield report includes clipping. We also recalculate it with our own model. If there is a flexible connection agreement, we read it for the limit, term, control rights and liability; at a shared connection point also for the rules between the operators. For parks connecting in 2027 or later, we include the case with a 70 percent limit until the legislator has decided.

We do not promise returns. We show which figure rests on which assumption. What makes a solar park attractive as a direct investment overall is set out in Investing in solar parks: revenues, costs and tax leverage at a glance. That is exactly what we check with you in a no-obligation initial consultation, gladly using a specific offering document.