Forecasting load growth and planning the utility investments necessary to handle changing electricity demand used to be a straightforward exercise.

Utility planners knew that customers in their service territory more or less shared the same load pattern. The only real wrinkles planners had to navigate as they considered grid upgrades were entirely new loads — like from a new subdivision — or widespread embrace of technologies like LED lighting that reduce demand.

The rapid influx of distributed energy resources (DERs) such as rooftop solar photovoltaics, energy storage and electric vehicles has introduced significant complexity to the formulation of accurate forecasts planners need to cost-effectively accommodate DERs while maintaining grid reliability. Even forecasts that accurately project the total number of new EV chargers or solar arrays that will be added to an area served by a substation can be problematic.

“A planner can assume that every x number of households has an EV charger and then space it out on the network so that every third person gets one,” said Gerhard Walker, director of product management at Opus One Solutions, a Toronto-based software engineering and solutions company that works with utilities across North America to effectively integrate DERs. “But when it comes to DERs, location is really important."

"If a Level 2 EV charger is put in at the end of a very long feeder, that additional load might cause a problem," Walker said. "Even if the forecast is correct that one new charger is added, its impacts on the grid can play out a hundred different ways depending on where it is installed.”

From worst-case to probabilistic planning

The traditional utility planning approach has been to make investments based on worst-case scenarios, usually to accommodate the highest system peak load. “We have to divorce ourselves from the old approach where we are looking at low and high load or low and high generation scenarios because those end up giving you extremes that would most likely never happen,” said Walker. “It can cause you to massively overbuild the infrastructure or under-build, both of which have problems.”  

The increasing prevalence of DERs across distribution grids demands a more flexible and accurate approach to forecasting and planning. This can be achieved by embracing forecasting based on probability studies, Walker said.

DERs come in numerous flavors and combinations. Those assets can have very different grid impacts depending on a host of factors, including where they’re located, how many other DERs are nearby, and even the time of day. Uncertainty about future DER uptake is also challenging the efforts of planners to forecast how DERs may affect the distribution grid.

Put another way, the sheer number of DER scenarios and how those many iterations could impact the grid are impossible to evaluate using traditional planning tools. “It’s very computationally intensive problem-solving because instead of running a snapshot for a single-hour worst-case scenario, you’re running an entire year of power flows, and you’re doing it thousands and thousands of times,” said Walker.

“It’s orders of magnitude more resource-intensive than traditional deterministic planning.”

Computing power enables probabilistic planning

In recent years, the emergence of cloud computing and parallel processing has made possible probabilistic planning that considers a vast range of DER scenarios and their likelihood. For distribution planners and utilities, this level of granular detail can guide investment decisions — either in infrastructure upgrades or the potential of non-wires alternatives.

Using a probabilistic approach, planners evaluate thousands of randomly generated scenarios around a given confidence interval, allowing them to understand the probability that deviations from their forecast will cause issues to the system. This type of scenario analysis also creates a clearer picture of the worst-case scenario and its probability of occurring.

“For non-wires alternatives, the probabilistic approach is good because it allows you to look at such a broad spectrum of possibilities. For example, you can run all probable scenarios and see how a battery behaves in every one of those situations, and you can start evaluating how many it actually solves,” said Walker.

“A probabilistic planning approach will tell you a solution solves 95 percent of your problems. So as a planner, you have to ask whether the 5 percent it doesn’t solve are likely or unlikely scenarios. The utility is basically stepping back from worst-case planning to make an optimized decision on what to invest in based on the probability of an event occurring.”

These decisions can be improved and refined when utilities and their regulators define what probabilities should actually trigger an investment. With traditional deterministic planning, there is no risk tolerance; any potential risk to the grid represented by DERs results in potentially expensive steps to mitigate it. While this regulatory approach hasn’t happened yet, it could lead to more efficient utility investments.

“If you have only a 4 percent probability of a problem down the road, is it worth an expensive upgrade?” said Walker. “At some point, we are probably going to have to have guidance from the regulatory side about what is an acceptable risk tolerance.”

Already, though, some utilities are embracing a probabilistic approach to help forecast and plan for DER installations. One California utility, for example, is using it for solar forecasting as a way to gauge hosting capacity. “It becomes very specific to the location of the solar installation,” said Walker. “If everything is installed at the end of the feeder, my hosting capacity drops significantly.”

As more and more DERs are deployed, it will become increasingly important for utilities to accurately forecast and plan for the impact they will have. “The traditional methods are going to involve planners making certain assumptions that are less and less true,” said Walker. “I think a probabilistic approach is the future for planning.” 

To learn more about using probabilistic methods to optimize DER planning, join our upcoming webinar with Opus One Solutions on June 23.