Croatia’s wind fleet demonstrates how little an annual capacity factor can reveal about the operational and commercial behaviour of a large wind portfolio.
In January 2026, Croatia had 30 wind farms in regular operation, with total installed capacity reaching 1,237.15 MW. A further two projects, with combined capacity of 77 MW, were under construction or trial operation.
The operating fleet generated 250,003 MWh during the month, corresponding to a capacity factor of 25.57%. On the surface, those figures could describe a relatively conventional renewable-energy portfolio.
Hourly generation, however, reveals a much more demanding operating profile.
Croatian wind output reached 958.83 MWh in its strongest hour of January, while falling to just 3.79 MWh at its weakest. Daily production ranged from 1.79 GWh to 19.12 GWh, illustrating the extent to which aggregate wind generation can fluctuate even across a large installed fleet.
February reinforced the same pattern. Wind generation reached 1,016.42 MWh in the strongest hour but fell to zero during the weakest hour, while the monthly capacity factor stood at 26.2%.
For the electricity system, this represents a significant balancing challenge. For investors, it is a reminder that a portfolio capable of producing around 1 GW in one hour can behave like a fundamentally different asset only a few hours later.
The financial implications begin with forecasting. Wind farms nominate their expected production ahead of delivery, while traders can adjust their positions through the intraday market as forecasts become more accurate closer to real time.
Even with increasingly sophisticated forecasting tools, however, some difference between expected and actual production remains. The balancing system must absorb that deviation, creating a cost that ultimately has to be allocated somewhere within the market structure.
Who bears that cost depends on the regulatory and contractual framework. A wind farm may carry imbalance exposure directly, while a trader can absorb it in exchange for a fee or a less favourable PPA price. An aggregator can also combine several assets in an attempt to reduce the impact of individual deviations.
The underlying economic principle remains unchanged: variability has a cost, even when the wind resource itself is free.
Croatia’s generation data also show why aggregation does not automatically eliminate that risk. If several projects are exposed to the same regional weather pattern, their production can rise and fall simultaneously, meaning that the diversification benefit may be considerably smaller than the number of individual assets suggests.
The critical metric is therefore correlation.
Two wind farms located hundreds of kilometres apart can still have strongly correlated production if they are influenced by the same large-scale weather system. Conversely, projects operating in different wind regimes may partially offset one another even when they are geographically closer.
That distinction is likely to become increasingly important for wind portfolios across Southeast Europe.
A developer assembling multiple projects should therefore assess not only the P50 production forecast for each asset, but also the covariance between their hourly generation profiles. A lender financing a portfolio should similarly examine whether the combination of projects genuinely reduces volatility or simply aggregates generation that responds to the same weather conditions.
The issue is equally relevant for power purchase agreements. A corporate buyer purchasing electricity from several wind farms may assume that portfolio diversification will create a relatively stable delivery profile.
That assumption can prove wrong if all contracted projects respond similarly to regional weather conditions. The portfolio could then overproduce collectively during strong-wind periods and underproduce at the same time during weak-wind conditions.
This can increase exposure to replacement power precisely when market prices are high.
Transmission geography introduces a second source of correlation.
Many Croatian wind farms are concentrated in coastal and southern areas where wind resources are particularly attractive. HOPS reported that the operating fleet in early 2026 was concentrated especially in Zadar, Šibenik-Knin, Split-Dalmatia, Dubrovnik-Neretva and Lika-Senj counties.
Such geographic clustering can create efficiencies in project development and infrastructure planning, but it can also concentrate generation injections into the same parts of the transmission network.
The combination of correlated weather and shared grid infrastructure therefore deserves greater attention in the valuation and financing of wind assets.
Storage could eventually mitigate part of this variability, but it should not be treated as a universal solution. A battery designed to manage short-term forecast deviations requires a different operating strategy from one designed for two-hour wholesale-market arbitrage.
Similarly, a battery used to firm a PPA may require different power and energy characteristics from a system primarily designed to provide balancing services.
Croatia’s balancing framework is also becoming more open to renewable flexibility. HOPS allows technically prequalified individual network users, aggregators, independent aggregators and renewable-energy communities to provide balancing services, subject to the relevant qualification and contractual requirements.
This creates an opportunity to turn part of wind generation’s variability into marketable flexibility.
But the first step is understanding the physical behaviour of the underlying assets.
Annual production figures smooth out the extremes. Electricity systems, PPAs and project finances experience those extremes hour by hour.
Croatia’s wind fleet therefore offers a useful lesson for investors across Southeast Europe. The key question is not simply how much electricity a portfolio can produce over the course of a year.
It is how often those wind farms rise and fall together — and what that correlation ultimately costs when the power system has to respond.
