Virtual Power Plants: Aggregating distributed energy resources

There's a quiet revolution happening on the grid. Millions of small energy assets - rooftop solar panels, home batteries, EV chargers, heat pumps - are popping up across Europe and the world. Individually, none of them matter to a grid operator. But aggregate them behind smart software, and you've got something that looks and behaves like a power plant. No turbine, no smokestack, no single location. Just coordination at scale.

That's a Virtual Power Plant. And if you're building anything in the energy space right now, you need to understand how they work - because VPPs are becoming the backbone of a flexible, decarbonized grid.

What is a VPP, really?

A Virtual Power Plant is a software platform that aggregates distributed energy resources (DERs) and orchestrates them as a single, dispatchable unit. From the grid operator's or market's perspective, a VPP can bid into wholesale markets, provide balancing services, or respond to dispatch signals - just like a conventional power plant.

The "why now" is straightforward. DER penetration is accelerating. The grid needs flexibility. And the economics of centralized peaker plants are getting worse while the economics of aggregated distributed flexibility are getting better. Software eats everything, including the grid.

What gets aggregated?

The asset mix in a VPP typically spans several categories:

Behind-the-meter batteries. Residential and commercial energy storage systems. Think Tesla Powerwalls, BYD units, or any of the dozens of systems now hitting the European market. These can absorb or inject power on command.

EV chargers and vehicles. A fleet of EVs plugged in at home or at a depot is a massive battery sitting idle most of the day. Smart charging (and eventually V2G) lets you shift load or even push energy back to the grid.

Heat pumps and thermal loads. A heat pump doesn't care if it runs at 14:00 or 14:30 - the house stays warm either way. That flexibility is shiftable load, and it's enormous in aggregate. Thermal mass is basically free storage.

Rooftop solar + inverters. Smart inverters can curtail output or adjust reactive power. Not the most popular option with asset owners, but technically valuable for grid management.

Demand response. Industrial and commercial loads that can be reduced or shifted on signal - HVAC systems, refrigeration, industrial processes. Old-school DR, but now with real-time telemetry and faster response times.

The key insight: none of these assets were built to be grid resources. The VPP software layer is what turns them into one.

How aggregation works in practice

This is where it gets interesting - and where most of the engineering complexity lives. A VPP platform has to handle several things simultaneously:

Forecasting. You need to know what your portfolio can deliver before you commit to it. That means forecasting solar generation, EV availability, battery state of charge, building thermal state, and consumption patterns - all continuously, all probabilistically. Bad forecasts mean penalties in the balancing market.

Optimization and dispatch. Given the forecast, market prices, grid signals, and asset constraints, the optimizer decides what each asset should do and when. This is a constrained optimization problem that runs on short cycles - typically 15-minute intervals aligned with market settlement, but sometimes sub-minute for frequency response.

Control signals. The dispatch plan needs to reach the devices. This means communicating with potentially tens of thousands of heterogeneous endpoints - via OCPP for chargers, Modbus or proprietary APIs for batteries, SG-Ready or EEBus for heat pumps, cloud APIs for smart inverters. The protocol zoo is real and painful.

Telemetry. You need to know what actually happened. Real-time monitoring of device state, energy flows, and grid conditions feeds back into the forecasting and optimization loop. Latency matters here - if you're doing frequency containment reserve, you're talking about seconds, not minutes.

Settlement and reporting. After the fact, you need to prove to the market operator or DSO what you delivered. Metering data, baseline calculations, and settlement reports close the loop. In many markets, this is still surprisingly manual.

If you're building this stack, the unglamorous truth is that 70% of the work is integration, data quality, and reliability engineering. The optimization algorithm is the easy part.

Grid value

Why do grid operators and markets care? Because VPPs can deliver several things that the grid desperately needs:

Peak shaving. Reduce demand during peak hours by discharging batteries, curtailing flexible loads, or shifting EV charging. This directly reduces the need for expensive peaker plants.

Balancing and frequency response. Aggregated DERs can respond to frequency deviations, helping keep the grid stable. The response times of batteries are actually superior to most conventional generators.

Ancillary services. Voltage support, reactive power compensation, congestion management - VPPs can participate in a growing menu of grid services, depending on market design.

Deferred grid investment. This is the big one that doesn't get enough attention. If a VPP can reliably reduce peak load in a constrained area, the DSO can defer or avoid a multi-million-euro transformer or cable upgrade. The economics here are compelling.

Business model

The VPP aggregator sits in the middle, and the economics work through revenue stacking - capturing value from multiple streams simultaneously:

The aggregator bids the portfolio into day-ahead and intraday energy markets, captures spread from balancing services, and contracts with DSOs for local flexibility. Each stream alone might not justify the platform cost, but stacked together they create a viable margin.

Participants - the asset owners - get paid for the flexibility they provide. This can be a fixed monthly payment, a share of market revenues, or reduced energy costs (dynamic tariffs make this particularly interesting). The incentive has to be tangible and transparent, or people won't opt in.

The aggregator's margin comes from the spread between market revenues and participant payouts, minus platform and operations costs. Scale matters enormously here. Managing 500 assets is almost as expensive as managing 50,000, but the revenue is 100x. This is fundamentally a software-scale business.

At Pstryk, we see dynamic tariffs as a natural entry point into this world. When your customers are already responding to price signals, you've built the behavioral and technical foundation for aggregation.

Key challenges

Let's be honest about what's hard:

Interoperability. Every device manufacturer has their own API, protocol, or cloud platform. Standards like OCPP, EEBus, and IEEE 2030.5 help, but real-world integration is still a slog. You'll spend more time reverse-engineering proprietary firmware than writing optimization code.

Latency and reliability. Consumer internet connections, cloud-to-device round trips, device firmware bugs - the control chain has many points of failure. For energy-only markets, this is manageable. For fast frequency response, it's a serious engineering challenge. Edge computing helps, but adds complexity.

Regulation. Market rules in most European countries are still catching up to aggregation. Prequalification processes for VPPs vary wildly. Some markets don't allow sub-meter aggregation. Others have minimum bid sizes that effectively exclude small aggregators. The regulatory landscape is improving, but it's fragmented.

Customer trust. You're asking people to let software control their battery, car charger, or heat pump. That requires trust in the technology, the company, and the value proposition. One bad experience - a cold house, an uncharged car - and you lose that customer. Comfort and mobility guarantees aren't optional.

What's next

Three trends will accelerate VPP adoption:

AI-driven optimization. Reinforcement learning and more sophisticated ML models will improve forecasting accuracy and dispatch decisions, especially as portfolios grow and asset interactions become more complex. The gap between a good optimizer and a great one translates directly to margin.

Dynamic tariffs at scale. As more markets mandate or incentivize time-of-use and dynamic pricing (the EU's electricity market reform is pushing this), consumer behavior becomes more flexible by default. Dynamic tariffs are the demand-side on-ramp to VPP participation.

Interoperability standards maturing. Matter for energy, EEBus adoption in HVAC, OCPP 2.0.1 for chargers - the standards are getting better. It'll take years, but the direction is clear. The aggregators who invest in protocol abstraction layers now will have an advantage.

The grid of 2030 won't be powered by a handful of massive plants. It'll be coordinated across millions of small, distributed assets - managed by software, optimized in real-time, and monetized through markets.

If you're building in this space, the opportunity is massive. The stack is complex, the integration is messy, and the regulatory environment is uneven. But the physics and the economics are on the right side. Software turns distributed chaos into grid-scale value. That's the VPP thesis, and it's playing out right now.