How to optimise a hydropower plant: from reservoir to market

Optimising a hydropower plant means deciding how much and when to generate to get the most value from the available water, while respecting physical and concession constraints. In practice you work on several horizons at once: long-term reservoir planning, the hourly schedule for the next few days, bids on the markets and the handling of the unexpected, from maintenance to faults.

Water as stored energy

A run-of-river plant generates according to the flow at the time and has little freedom of choice. A plant with a reservoir, on the other hand, is an energy store: it can hold back water and use it in the hours when the market pays most. This flexibility is what creates value, but it also makes the decision difficult, because every cubic metre turbined today will not be available tomorrow.

The starting point for optimisation is therefore the value of water over time. Generating makes sense when the market price exceeds the value that same water would have if it were kept. When plants are arranged in cascade along a river system, the decision becomes more complex: what an upstream reservoir releases becomes inflow for those downstream, and the system has to be optimised as a whole, not plant by plant.

Planning reservoirs over the long term

Long-term planning decides how to spread the water resource over weeks and months. It has to deal with two uncertainties: how much water will arrive, from rainfall and snowmelt, and at what price the energy can be sold. That is why the work is based on inflow and price scenarios, and stochastic optimisation techniques help choose robust strategies rather than betting on a single forecast.

The result is a trajectory of reservoir levels, with target levels to be reached in given periods. A good plan keeps spills, that is, water released without generating energy, to a minimum and leaves room to capture high-price hours.

The constraints to respect

No optimisation is useful if the schedule it produces cannot be carried out. Typical constraints concern:

• reservoirs: minimum and maximum levels, minimum environmental flow, filling and drawdown rates;
• plants: available capacity, number of daily start-ups, ramps, minimum output or flow, binding schedules;
• canals and penstocks: maximum diversion flows and minimum flows by period.

On top of these come concession obligations and the needs of the local area. The optimisation model must represent all of them, and must adapt to every river system without bespoke development, because no two hydraulic systems are alike.

From optimisation to bidding: entry price, MGP and MI

Day by day, the long-term plan becomes an hourly production schedule for each plant, updated when inflows, prices or availability change. This is where the bidding strategy comes from. The entry price is the minimum price at which it is worth offering the energy: below that threshold it is better to keep the water. Calculating it hour by hour makes it possible to set the quantities and prices to offer on the Day-Ahead Market and the intraday markets, and to adjust the position when conditions change during the day.

With the power market moving to quarter-hours, schedules and bids have to be managed at a finer granularity, and the link between optimisation and bidding must be direct so as not to multiply manual work.

Unavailability, maintenance and Terna

An optimal plan becomes wrong the moment a unit stops. Unavailability, whether planned for maintenance or forced by a fault, must be recorded immediately, notified to Terna and reflected in the production schedule, otherwise bids are based on capacity that is not there. The more the process relies on emails and phone calls, the longer the delay between the field and the market.

Work on the hydraulic network also has downstream effects: closing a penstock or changing a flow can alter the output of several plants. Simulating the work before carrying it out makes it possible to choose the timing and configuration with the least impact on safety and production.

Measure and improve

The cycle closes with metering. Comparing actual output with the plan and the bids shows where deviations arise and improves subsequent decisions. The same analyses, extended over the long term, are used to assess asset investments and scenarios ahead of concession renewals.

Indicators of availability and of energy lost, through unavailability or spills, show management how much value is being lost and where to act. For those who also take energy from third-party producers, the same cycle applies to plants they do not operate directly: schedules and unavailability must arrive in a standard format, and Terna metering becomes the basis for settlement and invitations to invoice.

How MAS does it: IDRO Suite

IDRO Suite, the MAS software for hydropower producers, brings these steps together in a single platform. IDRO Optimizer, for optimising plants and reservoirs, handles long-term planning and intraday optimisation, calculates the entry price and prepares bids on MGP and MI, which pass to ERGObid for submission.

MAIA, for unavailability and maintenance, acquires faults from SCADA systems, communicates with Terna via AGILE and updates the production plan. THESEUS, for simulating hydraulic interventions, assesses the effect of an operation before going on site. Those taking part in the capacity market use CAMPO, for Capacity Market nominations and obligations; those who also manage third-party producers' plants have SOLIS UPNR for production forecasting and THOR as the producer portal.

IDRO Suite manages more than 3,000 MW of installed capacity. We explored the link between water management and the market in our article on how to get the most value from hydropower with IDRO. To see the suite on your plants, request a demo.