What’s the business model of a Virtual Power Plant (VPP)?

As the share of renewables rises in many energy markets around the world, the question of how to manage the infeed of those very resources arises more sharply. While the percentage of renewables in the grid and subsequently in the portfolio of power traders may be low and the volatility of renewables may not pose a big challenge in the beginning, there seems to be a certain threshold after which it becomes a financial burden for the portfolio manager not to forecast and dispatch renewables as accurately as possible. Let’s take your generic utility as an example. In the 2000s and the early 2010s, it may only have had a couple of megawatts of solar and wind in its larger portfolio, so the intermittency of PV and wind infeed didn’t have a big effect on its overall generation – their fluctuations were basically muted by the general baseline of conventional power production. But as soon as that utility ramped up its renewable generation to, let’s say, 20 % of its portfolio, solar and wind have a larger effect on the everyday management of the utility’s overall portfolio. Forecasts suddenly turn out to be wrong. Wind and solar are not curtailed if prices on spot markets turn negative and so instead of making money, they cost money. Nominations of expected infeed to the Balancing Authority (e.g. Transmission System Operators) have huge deltas. Trading limps behind and is punished by expensive balancing costs for ex post portfolio corrections. Why does this happen? Because by nature, it’s more difficult to project the trajectory of PV or wind power infeed than that of a baseline conventional power plant. There may not be the same historical data to rely on yet. In many cases, the utility doesn’t even know what its renewable portfolio is producing right now.

So the utility turns to the technology of a Virtual Power Plant to get a grasp on things: All renewable energy sources (RES) in the portfolio of that utility are networked through remote control units, so that the aggregated volume of power generation from all distributed plants is displayed live, and all or individual PV and wind farms can be curtailed from the trading floor. Furthermore, historic data is either added to the system by importing it from DSO-metered data or starts building itself once all plants are sending their infeed. Now you add some meteorological forecast data and put the whole dough through statistical analysis and machine learning algorithms to gain the best possible forecast of the entire renewable portfolio. Where is the money in that, you may ask? Four sources: First, the portfolio manager saves balancing costs that incur in many energy markets around the world when the forecasts of a power trader’s portfolio don’t meet the actual feed-in. Secondly, the portfolio manager or the trader can now curtail the RES portfolio as soon as prices fall below zero. Thirdly, in case the RES portfolio contains some sort of bioenergy or hydropower – or other dispatchable renewable energies like geothermal etc. – the trader may use the VPP to optimize and steer the generation schedule of those assets, e.g. by ramping them up when power prices are projected to go up. This way, he is now able to beat average prices on short-term markets and share the additional revenues with the plant operator. Lastly, by understanding the own RES portfolio better, the trader also learns more about the whole market situation and where spot market prices will be heading, since PV and wind infeed tend to become the price drivers in energy markets with higher shares of renewables. The trader then uses that information to take up not only a safer but also potentially more lucrative trading position.

Exhibit A shows how a VPP helps with the management of a RES portfolio from a trader’s perspective. But what about the actual physical fluctuations that the trader now knows more and earlier about than before but which nevertheless of course occur and which need to be physically balanced within the grid? First of all, more lead time due to improved forecasts already enables the entire market – other traders and dispatchers – to react more quickly to fluctuations and thus to counter them before they cause pain in the system. But more importantly, a VPP not only networks renewables to improve forecasting but also to provide ancillary services to the grid operator. After receiving a signal to ramp up or down power generation from the grid operator, the central control system of the VPP splits up that signal to hundreds or thousands of individual signals for the individual dispatchable renewable power plants, taking into account their restrictions on response time, filling levels, heat generation, you name it. It then automatically sends the ramping signal to the involved networked units and ramps them up or down to support the grid frequency and to counter the very same fluctuations other units in the VPP – mostly PV and wind – have caused in the first place. Again, where’s the money? Since short-term reserves are extremely important to run the electricity system and to provide the security of supply we all enjoy every day, they have a significant price tag. Tapping into that potential by providing short-term reserves to the grid operator is an excellent business case for aggregators, especially because they don’t need to invest in the physical buildup of flexible power generation (e.g. gas-fired power plants or pumped hydro storage) but merely in the networking of existing smaller-scale flexible units. Traditionally, the VPP operator splits the revenues from successful capacity tenders and actual deliveries of ancillary services with the operators of the dispatchable RES unit that is networked in the VPP. In non-liberalized markets where tenders for these reserves don’t exist yet, the Transmission System Operator (TSO) can run the VPP by himself to harvest the flexibility from distributed energy resources.

Basically a subset of Exhibit B, an aggregator might opt to not only aggregate power generating units but also (or exclusively) demand side units to provide ancillary services to the grid. The power grid doesn’t care if you ramp up the power generation of a hydropower plant or if you ramp down the power consumption of, let’s say, an air conditioner or the cooling system of a warehouse. The effect on the grid frequency is the same, so there also is a business case for aggregators (the term “Virtual Power Plant” doesn’t seem to apply here) on the demand side. In some energy markets, there are even – on top of the ancillary services markets – special tender schemes for reserves that are supplied by the demand side in order to promote their development. While the implementation of demand response aggregation is a very regional phenomenon today (US, Australia) and often limited to commercial and industrial power consumers, it holds gigantic potential once tiny flexibilities on the household level (e.g. EV batteries, heating pumps, AC units) are aggregated and become more active on the energy market.

Moreover, a fleet of aggregated demand response units may also be used to not only deliver ancillary services to TSOs but also to dispatch consumption of the networked units to times of low prices on spot markets – this way lowering the costs for power procurement for each asset with a share for the VPP operator. Thanks to the automated consideration of each unit’s individual restrictions, these time-of-use tariffs may differ from unit to unit.

Like every typology, this one also has its flaws. The boundaries between the different business cases are fluid, of course, depending on the structure and regulation of the energy market where the aggregator is active, but also depending on current and ever changing price signals in these markets. You can also easily imagine – and find – aggregators that combine all three business cases in one Virtual Power Plant. But one thing is for sure: The large-scale integration of renewables into grids and markets needs a systematic and well-coordinated approach through Virtual Power Plants – and there is money to be made with it.