Integrating solar PV into the grid

Georgios Tzounas, Assistant Professor at University College Dublin (UCD), outlines the modelling, simulation, and control challenges that arise when solar PV is installed in a power system that was not originally designed for it.

“We are moving into a phase where we need to understand power systems not just as static infrastructure, but as dynamic, evolving systems, and solar PV is a major contributor to that change,” Tzounas says.

While solar PV brings the benefit of clean, decentralised generation, it also introduces significant variability and uncertainty. “The primary energy source, solar radiation, is highly volatile, and that has consequences for system stability, especially in the short-term dynamic range we focus on,” he explains.

This critical time window, ranging from seconds to milliseconds after a disturbance, is where foundational stability mechanisms such as frequency and voltage control take place.

Challenges

One of the key challenges Tzounas points to is the impact of cloud cover on solar irradiance. “A cloud event can lead to a 60 per cent drop in irradiance within a minute. That is not a theoretical risk; it is been observed in real data. These are fast dynamics, and they matter a lot if you are trying to keep the system stable.”

Such rapid changes in solar generation create abrupt shifts in power flow, which can trigger control responses or even cascade into larger system disturbances if not properly managed. The challenge, Tzounas asserts, is that these dynamics are not well-represented in many standard power system models.

“Most of our models were developed for systems where generation was centralised and predictable. They do not account for the kind of stochasticity we see in solar PV.” He argues that modelling tools must be adapted to reflect the real-world behaviour of solar assets, especially at high levels of penetration.

Another complexity introduced by solar PV is the sheer number of small-scale generation units such as rooftop panels, small commercial installations, and community-scale arrays all feeding into the distribution network. “You now have hundreds of thousands of devices, many of them operating with limited visibility to the grid operator, and often with proprietary or unknown control systems,” he says.

This proliferation of devices introduces what Tzounas calls “granularity” to the system. “It is not just more generation, it is more potential points of interaction, more data, more control complexity.”

“If we do not rethink some of our engineering foundations, we may find that the system behaves in ways we do not anticipate or cannot control.”

In response, some system operators are moving toward distributed aggregation models, where clusters of solar PV and other distributed energy resources (DERs) are treated as a single controllable entity. “That helps from a management perspective, but it introduces new questions around coordination, timing, and how those aggregated resources respond dynamically to changes in the grid,” Tzounas says.

Simulations

Even with improved models, effective simulation is far from guaranteed. “Defining a model and solving it numerically are two different things,” Tzounas explains. He points out that many commercial simulation tools rely on numerical methods that may not be stable or accurate when applied to solar-dominated systems.

“Systems with delays, fast switching events, or high levels of stochastic input like solar PV are challenging to simulate correctly. The numerical methods themselves can introduce errors or instabilities. So the simulation may run, but it may not be trustworthy.”

This is especially concerning for scenarios where solar PV is expected to provide critical grid services such as frequency regulation or voltage support. “If your simulation does not reflect the real behaviour of the system, you cannot trust your control design,” he adds.

Control

Unlike traditional generators, solar PV systems are interfaced through power electronics, and their contribution to the grid is entirely defined by their control algorithms. “They do not have physical inertia. Their behaviour is software-defined. That gives us flexibility, but also responsibility,” Tzounas says.

He notes that some current control strategies attempt to make solar inverters behave like synchronous machines. “That is understandable; it is a known benchmark. But we should not assume it is the best solution. The control possibilities are broader, and probably more effective if we rethink them from first principles.”

For example, conventional schemes tie active power to frequency control and reactive power to voltage control. That works in large transmission systems, but breaks down at the distribution level where solar PV is typically installed. “There, the relationship between voltage and active power is stronger, and you cannot rely on the same assumptions. We may need to control both frequency and voltage through a combination of signals and power flows.”

A 2024 study by Tzounas and researchers from EirGrid and UCD suggests that hybrid control schemes, using both active and reactive power to regulate both variables, can deliver better performance. “It is not just a matter of better performance, it may be necessary to avoid oscillations and instability, especially as more solar PV comes online,” he says.

Concluding, the UCD assistant professor states that the integration of solar PV is “not plug-and-play”. “It requires a revaluation of how we model, simulate, and control power systems especially in the face of growing renewable targets.

“The complexity is increasing. The traditional tools are being pushed beyond their limits. If we do not rethink some of our engineering foundations, we may find that the system behaves in ways we do not anticipate or cannot control.”