Planning network investment as DER adoption grows

Distributed energy changes the shape of the load-flow problem, not just the width of its error bars. A system-average forecast can’t tell you where it defers investment and where it forces it.

For most of the last century, the distribution network had one job in one direction. It moved power from the substation to the customer. Distributed energy resources quietly ended that. Rooftop solar, behind-the-meter storage, EV chargers and the management systems coordinating them (DERMS) have changed the edge of the grid. That edge now both consumes and produces, on a schedule no historical load curve predicted.

The reflex in many planning rooms is to treat this as a forecasting problem with a wider error bar. From an asset-management standpoint, it isn’t. It’s a change in the shape of the problem, and in the resolution at which you have to model it. A wider error bar still assumes the same underlying load duration curve. DERs break the curve itself. They decouple a feeder’s net demand from the weather, the time of day, and each other. An aggregate forecast cannot represent that behaviour.

The same megawatt can defer investment or force it

Here’s the uncomfortable part: a distributed resource is not inherently good or bad for your capital plan. Sited and integrated well, a cluster of DERs can defer or eliminate a reinforcement. The load it offsets is load you no longer have to build for. The deferral itself has quantifiable value. Money has a cost. Pushing a reinforcement out by several years therefore carries a present-value benefit. You can put that benefit on the books before a single new asset is installed.

Sited poorly, the same resources do the opposite. Behind-the-meter generation can drive reverse power flow through assets and protection schemes designed for one-directional flow. Concentrated PV can push feeder voltage above statutory limits in the middle of a low-load afternoon. And clustered EV charging can create localised thermal loading, forcing a reinforcement you would not otherwise have needed. The identical megawatt of solar is an asset on one feeder and a liability on the next.

The identical megawatt of solar is an asset on one feeder and a liability on the next.

The deciding factor is rarely the quantity of DER on the system. It’s where it lands relative to existing headroom, and how the network beneath it behaves hour by hour. That is precisely the resolution most planning models don’t operate at. They were built to answer “how much demand, and when does it peak”. Not “what is the voltage at the end of this feeder at 2pm on a sunny Sunday when half the connections are exporting.”

Why aggregate forecasts quietly mislead

A top-down forecast can tell you that DER adoption across a service territory will reach some percentage by some year. What it cannot tell you is which substations gain headroom and which feeders breach their hosting-capacity limit first. Nor can it tell you where a connection request will clear and where it will trigger a reinforcement. Those answers live in the physical constraints: thermal ratings, statutory voltage bands, fault levels, protection coordination, and reverse-flow limits. These constraints only resolve when you model the network from individual assets and circuits upward, with real topology and impedances. A system average, worked downward, discards them.

Hosting capacity is the clearest example of why the average lies. It is not a single territory-wide number. It is a per-feeder, sometimes per-node quantity, governed by whichever constraint binds first. That could be voltage rise on a long rural radial or a thermal limit on a loaded urban feeder. It could also be the fault level near the substation. Two feeders carrying identical nameplate DER can sit on opposite sides of their limit because their length, loading and topology differ. A connectivity-aware, power-flow-based model captures that. An average discards exactly the spatial information that determines the answer.

Plan from the average and you make one of two expensive mistakes. You overinvest in reinforcement the DERs would have rendered unnecessary, sinking capital into capacity distributed generation was already providing. Or you underinvest and discover the binding constraints only after they surface as voltage complaints and outages. At that point the remedy is reactive, urgent and far dearer than the planned alternative.

The leadership question underneath the technical one

What executives are really asking isn’t “how much DER is coming.” It’s “how does this change what I need to build, and when can I avoid building it?” That reframing matters. The upside of getting DER integration right isn’t only reliability, it’s the capital you defer or never spend.

The discipline that unlocks it is treating avoided investment as something you quantify and defend, not merely hope for. If storage on a constrained feeder lets you postpone a reinforcement, that deferral has a net present value that belongs on the same risk-and-return footing as every conventional build in the plan, rankable, comparable, and justifiable to a regulator alongside the steel it displaces. Distributed resources stop being a nuisance to absorb and become a planning asset to position.

So how do you tell the asset from the liability?

Getting there requires asset models that are genuinely DER-ready: built to evaluate hosting-capacity headroom, reverse-flow and voltage-rise scenarios, and the time-varying behaviour of the edge, not just annual peak growth. The networks treating DER as an analytical opportunity rather than an operational headache are the ones finding capital they didn’t know they had, by modelling from the asset up, where the value actually hides.

How Direxyon does this. Where most tools produce a single deterministic projection, Direxyon models the uncertainty explicitly and does so from the asset up, not from the system average down. Feeder-level hosting-capacity, reverse-flow and voltage-rise scenarios are encoded into a decision twin that reflects each feeder’s actual topology, thermal rating, and DER adoption profile. Those scenarios are then run through Monte Carlo simulation across the full portfolio, so each integration strategy yields a distribution of probable outcomes rather than one forecast that will inevitably be wrong. The output includes the present value of a DER-enabled reinforcement deferral with confidence intervals you can put on the books and defend before a regulator and a clear ranking of which feeders represent genuine capital-deferral opportunities versus which ones will require reinforcement regardless of how optimistically the DER is positioned.

Frequently Asked Questions