On 22 September the System Coordinator published a new run of its Long-Term Programming model, the PLP. It is the model that projects system operation week by week under, at this time of year, 31 different hydrological conditions. It is where the water value used to schedule day-to-day operation comes from. Read hydrology by hydrology, the run paints a picture the sector had not seen: entire weeks in which the marginal cost is zero at every hour, not just at midday.
That has a direct consequence for a fleet that will exceed 6,000 MW of batteries by year-end (and for companies that are long in the energy balance). A battery earns the difference between the price at which it charges and the price at which it discharges. If both ends are zero, every cycle wears the equipment and recovers nothing.
Three seasons in seven months
The run splits the horizon into three very different stretches, and does so almost identically at Crucero, Polpaico and Charrúa.
Between September and October the profile is the one the system already knows: a zero price during daylight hours and around 60 USD/MWh at night. That 60-dollar spread is, quite literally, a battery's business.
Between November and January the spread disappears along with the price. In the weeks of 23 and 30 November all 31 hydrologies leave Crucero at zero for all 24 hours.
From February the price returns, at between 48 and 54 USD/MWh.
None of this has a precedent in the north. In 107 weeks of actual prices, since September 2024, Crucero never had a week with the median at zero in all three blocks of the day; its lowest night block was 13.6 USD/MWh. In the south it is already happening: Charrúa recorded its night block at zero in six of the last nine weeks, and Puerto Montt had an entire week at zero in late August. What the PLP projects is that southern condition extended to the whole system.
What a battery sees in a flat price
For a generator, what matters is the level of the price. For a battery, what matters is its shape: the difference between the hour it charges and the hour it discharges. With a whole day at zero that difference vanishes, and cycling carries two costs that nobody pays for. One is round-trip losses, which cost nothing at a zero price. The other is wear: every cycle consumes a fraction of the equipment's useful life and of the manufacturer's warranty, which is expressed in cycles or in energy discharged, and that cost does not depend on the price.
So why would they cycle?
If the price is zero at every hour, the efficient answer —from the owner's point of view— is for the battery to do nothing, except what the system needs from it for security or reserves. Charging with a surplus that was going to be curtailed anyway and discharging into a night that does not need it either displaces no generation at all: it turns curtailment into losses and wear.
The current rules do not guarantee that outcome. Stored energy is valued at what it cost to charge, at the scheduled marginal cost of the charging hours, so a battery that charged at zero enters the merit order with a zero variable cost. In real time, moreover, batteries follow the charge and discharge profile of the day-ahead or intraday schedule in force.
The system needs them to minimise operating cost
In this condition, whether one battery or another charges or discharges makes no difference to the operator. The optimisation problem becomes degenerate: there are thousands of solutions with the same total cost and the schedule delivers one of them. Which batteries cycle, which do not, and what state of charge they end the day at is decided by the algorithm, using criteria that are not public. As we warned in our July proposal, the apparent neutrality of a mathematical solution can hide significant distributional decisions. The system has already been through this with the allocation of curtailment.
Discrepancy 31 is about the other problem
This discussion is taking place while the Panel of Experts hears discrepancy 31-2026. Aela Generación is challenging the June Economic Transfers Valuation Report because its El Salvador battery was instructed to inject in intervals in which the actual marginal cost at its busbar fell below its variable cost, which since January has been the opportunity cost the Coordinator itself calculates for it. Aela is asking for side payments to cover that difference. The Coordinator replies that there was no operation out of economic order, because the dispatch resulted from the scheduling process.
The parties that have joined the case show how far it reaches. AES Andes points out that Supreme Decree 32, published in June, extends the opportunity-cost calculation to all storage facilities, so whatever the Panel decides goes beyond a single battery. ACENOR, for its part, points out who pays: uncovered production costs are charged to those who withdraw energy to supply end customers.
The discrepancy asks how to pay for the value of stored energy when the system uses it before the price recognises it. A wet year raises the opposite question. With a zero price at every hour, a battery's opportunity cost —read from the model as the marginal value of one additional MWh at its state of charge— is also zero. Zero variable cost and zero marginal cost: even if the Panel sided with Aela, in a flat week there would be no difference to compensate. Opportunity cost solves the intertemporal problem of energy, but it does not put a price on the cycle.
A design built for thermal plants and reservoirs
The Chilean market is scheduled on audited costs, without bids. A thermal plant needs only its fuel cost and a non-fuel variable cost per MWh. A reservoir has a water value that the model calculates. A battery fits neither case. Its relevant cost is not the energy it bought, which in these weeks cost nothing, nor only the future value of that energy, which in these weeks is worth nothing either. It is cell wear, which depends on the number of cycles, their depth and the state of charge at which it operates, and which today is not part of the cost table used to schedule the system.
Finally, investment. If for three months of a wet year energy arbitrage does not exist, a battery's income over that period rests on capacity payments and ancillary services. Hydrology ends up weighing on the storage business far more than the models that assess these projects with an average year assume.
Questions for discussion
There are no settled answers to all of the above. Some questions the sector should perhaps discuss before November arrives:
- Should wear enter the cost table? As an audited non-fuel variable cost, for instance. And if it does, with what metric —USD per MWh discharged, per equivalent cycle, by depth of discharge— and who audits it?
- What tie-breaking rule does the schedule apply when the solution is degenerate, and should it be public? Pro rata allocation, a minimal per-cycle penalty or preserving a given state of charge are alternatives with different distributional effects, and every participant should be able to know which one applies.
- Who stores energy for the next tight spell when reservoirs and batteries lose their value at the same time, and a wet year can end in a tight autumn?