The term "virtual power plant" has been used to mean so many different things over the past decade that it has nearly lost its utility as a technical descriptor. Marketing teams use it to describe everything from a single behind-the-meter battery to a nationwide demand response program. Utility strategy documents use it as a catch-all for any aggregation of distributed resources. The actual operational definition is more specific, and the gap between the concept and the operational reality is where most VPP projects stumble.
A virtual power plant is a portfolio of distributed energy resources coordinated to behave as a single dispatchable generation or load resource from the perspective of a grid operator or wholesale market. The "virtual" qualifier is what distinguishes it from a physical power plant: the assets are geographically distributed, heterogeneous in type and capacity, and owned by different parties. The "power plant" part means the aggregate must be controllable, predictable, and reliable enough to commit capacity into operational and market processes that depend on that capacity showing up when needed.
That second requirement is where dispatch software is not optional. It is the thing that makes a VPP a VPP rather than a collection of DERs that happen to share a spreadsheet.
The Coordination Challenge
Consider what coordinating a VPP actually involves at an operational level. You have a portfolio that might include 40 behind-the-meter batteries at commercial facilities, 15 MW of curtailable industrial load from a handful of manufacturers, and 8 MW of residential flexible load from a demand response program. The individual assets range from 20 kW to 2 MW. They are connected to different distribution feeders, served by two different distribution utilities, and their owners have different contractual obligations to the aggregator.
A dispatch instruction to deliver 10 MW from this portfolio in the next 15-minute interval requires the dispatch software to: determine which assets are available (not in maintenance, not blocked by a distribution constraint, not already committed to another product), calculate how much each asset can contribute within its physical and contractual limits, allocate the 10 MW target across available assets, send dispatch signals to each asset's control interface, monitor actual response, and adjust allocation in near-real time if some assets under-deliver.
All of this needs to happen in seconds. The dispatch signal from the ISO or the utility arrives at the aggregator, and the aggregator's system has to disaggregate it into per-asset instructions and confirm execution before the interval opens. There is no time for a human in the loop on each event.
Asset Diversity as Both Strength and Complexity
One of the genuine advantages of a heterogeneous VPP over a single large battery is portfolio diversity. A portfolio that includes batteries, flexible loads, and solar curtailment has different response characteristics in different conditions. Batteries are available regardless of weather or time of day but are constrained by SoC. Flexible loads are constrained by process requirements and occupancy patterns but are not energy-limited in the same way. Solar curtailment is only available when the sun is shining.
This diversity means the aggregate portfolio can provide a wider range of services across more operating conditions than any single asset type could. A battery-only VPP runs out of energy during an extended high-demand period. A flexible-load-only VPP cannot respond during nights and weekends when buildings are unoccupied. The mixed portfolio has a much higher probability of being able to respond to any given dispatch event.
But diversity also means dispatch complexity. Each asset type has different control interfaces, different response times, different minimum and maximum dispatch levels, and different constraints on repeated activation. The dispatch optimizer needs a model of each asset class that accurately represents these characteristics, not a simplified aggregate. Getting those models right is a significant data and engineering effort when a new portfolio is onboarded.
Forecasting as the VPP Foundation
A VPP that cannot predict its own available capacity is unreliable as a committed resource. Before submitting a day-ahead bid or a capacity commitment to a distribution utility, the aggregator needs a forecast of what each asset in the portfolio will be capable of delivering during each interval of the next day.
For battery assets, available capacity depends on projected SoC at the start of each interval, which depends on prior dispatch decisions and charging behavior. For flexible loads, available curtailment depends on occupancy, process state, and weather-driven load profiles. For demand response participants in residential programs, availability depends on how many participants are actively enrolled and responsive, which is itself uncertain.
We are not saying you need perfect forecasts to operate a VPP. You need forecasts that are accurate enough to make defensible capacity commitments. If you commit 8 MW to a capacity product and your portfolio can actually deliver only 5 MW under average conditions, you will face performance penalties that undermine the economics of the program. The forecast accuracy threshold is determined by the penalty structure of the product you are committing to, not by a universal standard.
Dispatch Signal Architecture
How dispatch signals reach individual assets is a practical architecture decision that shapes the entire VPP operation. The options range from direct inverter control via SCADA to cloud-based API signals to aggregator-operated local gateways at each site.
Direct SCADA control offers the fastest response time and highest reliability for battery assets that have grid-compliant control interfaces. But it requires the aggregator to maintain communication infrastructure to each site, which scales in cost with portfolio size. Cloud-based API dispatch works for assets with internet-connected control systems and is lower infrastructure cost, but introduces latency and single-point failure risk if the cloud connection or the asset's internet link is unreliable.
Local edge controllers at each site are a middle path: the edge device receives dispatch schedules from the cloud ahead of time and executes them locally without depending on real-time connectivity for each dispatch event. This architecture is more resilient to communication outages and has near-zero dispatch latency for pre-scheduled events. It is less flexible for real-time re-dispatch in response to unexpected system conditions, because the edge device cannot receive an updated schedule if communication is disrupted.
For a VPP committing to ancillary services with real-time AGC signal tracking, the edge controller architecture is essential. For a VPP providing day-ahead capacity commitments that rarely require real-time re-dispatch, cloud-based dispatch with good connectivity monitoring is often sufficient and considerably cheaper to build and maintain.
Settlement Accuracy at the Portfolio Level
VPP settlement presents a specific accounting challenge: the ISO or utility settles with the aggregator as a single entity, but the aggregator must then attribute revenue and costs to individual asset owners. This disaggregation is not straightforward when the portfolio performed as a unit but individual assets delivered different fractions of the committed quantity.
If the VPP committed 10 MW and delivered 9.8 MW, which assets get credit for the shortfall? If two batteries were at low SoC during the interval and under-delivered, but the flexible loads over-delivered to compensate, how is the revenue split? The settlement methodology needs to be defined in the aggregator's contracts with asset owners before operations begin, and the metering infrastructure needs to support the per-asset measurement granularity required to apply that methodology accurately.
This is operational detail that does not appear in VPP marketing materials but determines whether asset owners continue participating after the first settlement cycle. Operators who have built VPPs from scratch know that settlement transparency and accuracy is one of the top reasons participants leave programs early. The dispatch software needs to produce per-asset performance records that are audit-ready, not just aggregate delivery numbers.