The Forgotten Price of Keeping the Lights On: A Fair Market for Voltage Stability
Every watt of solar and wind power we install in far-flung fields and deserts comes with a quiet, unglamorous problem buried in the mathematics of the grid: some of those buses — the points where power plants plug into the network — are too "weak" to handle the renewable energy flowing into them. When the grid at a connection point has high impedance, the local short-circuit ratio (SCR, a snapshot of how strong the local grid is) drops, and voltage becomes unstable. Too much voltage instability and you get flicker, tripped protections, or in the worst cases, blackouts. As grids fill with solar panels and wind turbines — the majority of which operate in "grid-following" mode, meekly tracking the grid rather than anchoring it — this weak-bus problem stops being an engineering footnote and becomes a near-term operational crisis.
A new paper by Peng Wang and Luis Badesa of the Technical University of Madrid (Wang & Badesa, 2026) takes aim at the invisible economic machinery underneath this problem. The question is not just how to keep the grid voltage-stable, but how to pay the generators who make it so. Utilities routinely pay power plants for energy and for "ancillary services" — things like frequency regulation and voltage support that keep the grid reliable. But voltage stability, and specifically the strengthening of weak SCR buses, has been stubbornly hard to price. The researchers propose a new mathematical method — a "primal-dual formulation" — that computes shadow prices for voltage stability services and, crucially, guarantees that the plants providing them actually make money doing it. Across their case studies, the method produced revenue-adequate prices every time, allowing synchronous generators, virtual synchronous generators (VSGs), and grid-following inverters to recover their costs without the awkward, opaque "uplift payments" that current markets lean on as a crutch.
It sounds like an accounting problem. But the stakes are physical. If we cannot price voltage stability fairly, generators that keep the grid strong can lose money by doing their job — and the clean-energy transition can stall at the point of physics rather than politics.
The Science
The story starts with physics. A grid-following inverter (GFL-IBR — inverter-based resource) is a current source. Unlike a spinning synchronous machine that has inertia and naturally regulates its terminal voltage, a solar or wind inverter simply injects whatever current the control system commands and lets the grid tell it what voltage results. That is fine on a strong bus with low impedance and a high SCR. It becomes dangerous on a weak bus with high impedance, where the same current injection swings the voltage dramatically.
Chu and Teng (2022) derived an explicit operating-point constraint that captures this. In steady state, the active and reactive power injected by a grid-following unit must satisfy
where is the SCR of the GFL-IBR's bus and the equivalent impedance the unit sees. Reading the formula, two levers emerge for keeping voltage stable. A GFL-IBR can provide reactive power support, pushing the right-hand side of the inequality up so it can safely inject more active power. Or, someone else can strengthen the grid by reducing the equivalent impedance — raising directly. That second job is what synchronous generators and virtual synchronous generators do by staying online: their spinning mass (or virtualized equivalent) lowers the impedance, acts as a voltage anchor, and lifts the whole stability envelope.
So voltage stability is a service with real value and real cost. A synchronous generator that stays on just to firm up a weak bus burns fuel, pays no-load and startup costs, and gets nothing for it in an energy-only market. A grid-following inverter that curtails its active power to make room for reactive support forgoes revenue. Unless markets pay for this, the units have no reason to provide it.
The catch is that computing the price is mathematically treacherous. The unit-commitment problem — deciding which generators turn on and off each hour to meet demand at minimum cost — is a mixed-integer problem. Its commitment variables are binary: a generator is either on or off, no fractions. And shadow prices, the dual variables that reveal the marginal value of a constraint like voltage stability, are only cleanly defined for convex problems. Chop the integer variables away and the prices get distorted; keep them and the "shadow prices" stop being well-defined.
Two prior methods have tried to work around this. The "dispatchable" approach relaxes the binary commitment variables into a continuum, allowing generators to be partially committed — a fiction that under-prices their contribution to voltage stability because a plant's SCR support only materializes when it is actually fully online. The "restricted" approach preserves the integer commitments by solving the problem twice and deriving a separate "commitment price" for synchronous generators, but that price only covers part of a unit's fixed costs and shoves the rest into uplift payments — unit-specific, opaque, allocated to no one knows whom. And weather-driven renewables, having no commitment variables at all, are left out of the compensation entirely.
What They Found
The innovation in this paper is to sidestep the convexity trap by constructing the market and the prices together, in a single primal-dual framework. Instead of solving the unit-commitment problem first and then hoping to extract meaningful prices from a badly behaved optimum, the researchers solve the primal and dual problems simultaneously, minimizing the duality gap — the distance between the cost-minimizing primal solution and the revenue-maximizing dual one. They keep the binary commitment variables intact, but add explicit "non-negative profit" constraints that tie the primal variables (power output) to the dual variables (prices). This guarantees, by construction, that every participating unit earns at least as much as it spends.
The profit constraints are built to be honest about the full cost of staying in the market. For synchronous generators, that means no-load costs, marginal generation costs, and startup costs. For inverter-based resources, the researchers go a step further and fold in daily-amortized investment costs — the capital expense of building the converter, spread across its operating life. This is a meaningful departure, because it means the prices signal not just "cover your fuel" but "this resource is worth building in the first place." A grid-following inverter that only gets paid its operating cost will never be financed; one that recovers its capital has a business case.
Revenue adequacy of participating units by pricing method
Net profit (arbitrary units scaled from case-study results) of each participating unit type under the three pricing methods. Under dispatchable and restricted pricing, synchronous generators end up with negative net profits (requiring uplifts); under the proposed primal-dual method all three unit types achieve non-negative profits.
| Label | Value |
|---|---|
| SG (dispatchable) | -12 |
| SG (restricted) | -8 |
| SG (primal-dual) | 15 |
| VSG (dispatchable) | -3 |
| VSG (restricted) | 2 |
| VSG (primal-dual) | 9 |
| GFL-IBR (dispatchable) | 0 |
| GFL-IBR (restricted) | 0 |
shows the revenue adequacy results across the case studies. The proposed primal-dual method produced non-negative profits for every participating unit type in every scenario — synchronous generators, VSGs, and GFL-IBR alike. In contrast, the dispatchable and restricted pricing methods left synchronous generators losing money, with negative net profits in many hours, requiring uplifts to keep them in the market. The dispatchable method also failed to remunerate renewable units because relaxing the commitment variables erased their voltage-stability contribution. In the case studies detail, the primal-dual prices found the same optimal unit commitment as the original problem — the duality gap stayed extremely tight, confirming that the relaxation around the optimal solution preserved the economically meaningful result while restoring the feasibility that integer variables had destroyed.
Revenue composition of a voltage-supporting unit
Illustrative composition of a voltage-stability-supporting unit's revenue under the proposed primal-dual pricing. Energy payments form the backbone, while the SCR-enhancement price (dual variable mu) and reactive-power-support price are surfaced as transparent, real-time ancillary service revenues that keep the unit profitable.
| Label | Value |
|---|---|
| Energy payments | 62 |
| SCR-enhancement price | 22 |
| Reactive-power support price | 16 |
breaks down how the revenue is composed. Energy payments form the backbone, but the voltage-stability service prices — the SCR-enhancement price and the reactive-power-support price — contribute a meaningful share for the units located at weak buses. This is exactly the price the restricted and dispatchable methods mangled: they either under-valued it (dispatchable) or pushed its cost into unpublicized uplifts (restricted). The primal-dual approach surfaces it as a transparent, real-time ancillary-service price.
Remaining uplift payments by pricing method
Illustrative level of remaining uplift / make-whole payments required for synchronous generators across the three pricing methods. The primal-dual formulation eliminates unit-specific uplift payments entirely because its profit constraints guarantee revenue adequacy by construction.
| Label | Value |
|---|---|
| Dispatchable | 8 |
| Restricted | 5 |
| Primal-dual | 0 |
shows the key difference in transparency: with the primal-dual method, no unit-specific uplift payments are needed at all. The "make-whole payments" that today's markets dispense after the fact to plug revenue gaps disappear entirely, because the prices were designed from the start to prevent the gaps.
Why This Changes Things
The most radical thing about this result is what it removes. Uplift payments are not just an accounting nuisance; they are a market-integrity problem. They are allocated per unit, decided after the fact, and not publicized in real time (Hohl et al., 2023). No one can see the true cost of keeping the grid voltage-stable, and no one can be sure who is bearing it. By pricing voltage stability directly and transparently, the primal-dual method turns a hidden subsidy into a visible market signal — the difference between a tip jar passed around after dinner and a price written on the menu.
There is a deeper significance too, one that reaches beyond this specific service. The unit-commitment problem's integer variables are the arch-enemy of electricity-market design everywhere. Every time a market operator has to price a non-convex resource — a minimum-run coal plant, a must-start nuclear unit, a synchronous condenser — it confronts this same duality-gap problem. The primal-dual trick, with its explicit profit constraints and duality-gap minimization, is a template for pricing any service whose cost cannot be separated from an on/off commitment decision. Voltage stability is the first application; it will not be the last.
The investment-cost dimension also matters for the energy transition specifically. Weak buses tend to be where renewables actually live — offshore wind farms, desert solar plants, remote onshore wind. These are low-SCR regions by nature, and the plants there are the ones whose voltage stability most needs supporting. A market that prices that support fairly and that lets inverter-based resources recover their capital costs turns voltage stability from a constraint into a revenue stream. That is the difference between a grid that tolerates renewables and a grid that is built to welcome them.
What's Next
The authors are careful about what this method does and does not claim. The offline training that approximates the impedance ratios in linear form, borrowed from Chu and Teng (2022), is an approximation — the true inverter of the admittance matrix is highly nonlinear, and the linear surrogate carries some error. The duality gap, while small, is not guaranteed to be exactly zero once the profit constraints and binary variables are added. Strong duality is compromised; the solution stays near-optimal rather than provably optimal. That is an honest caveat, and one that invites future work on tightening the relaxation and bounding the gap more rigorously.
There is also the question of scale and dynamics. The framework handles static voltage stability — steady-state constraints at each operating point. The dynamic transients that follow a disturbance — the milliseconds-to-seconds swings that trip real-world protections — are a different beast, and pricing those would require folding in rotor-angle and frequency dynamics that this model does not carry. A natural next step is extending the primal-dual logic to dynamic stability services.
And then there is the transition question: as grids shift from many synchronous machines to mostly-inverter resources, the mix of who can provide SCR enhancement changes. The paper already includes virtual synchronous generators, which emulate inertia and voltage anchoring digitally — but the authors note that GFM (grid-forming) control, which lets inverters behave more like synchronous machines, remains an open frontier. The pricing framework, designed around physical voltage-stability constraints, is agnostic about which technology meets them. It just needs to be paid.
The quiet message of this paper is that the clean-energy grid is not short on physics or on hardware — it is short on markets. The technology to keep weak buses stable exists. What has been missing is a way to make that stability worth a generator's while. By computing prices that guarantee revenue adequacy, eliminate uplifts, and reward both spinning machines and inverters for the voltage they hold steady, Wang and Badesa have handed the grid a price tag for its own resilience. Sometimes the most important advance in a revolution is not a new machine, but a fair way to pay for the one you already have.