Quick Dips

Solving Hydro Headaches

Imagine trying to sort out a system which will allow optimal automatic scheduling of water flow through 166 kilometres of hydro generation scheme, involving eight power stations, each with different capacities and requirements.

That's the headache recently solved by Dr Kevin Broughan, Alfred Sneyd, Yan Wang and Carolyn McKenzie at Waikato University. They were contracted by ECNZ to develop a mathematical model of the Waikato hydro system, allowing precise control over a 48-hour scheduling period.

"The program involved several thousand variables," says Broughan. Apart from the need to check the system every 30 minutes over the 48 hours, the programmers had to also bear in mind the individual constraints throughout the system.

"The Waikato hydro lake levels may not vary by more than about one metre, and the generation capacities of the eight stations vary from a minimum of 51 MW at Waipapa to a maximum of 360 MW at Maraetai," says Broughan. "Since Waipapa is immediately downstream from Maraetai and Lake Waipapa has a relatively small surface area, it will overflow quite quickly when Maraetai is generating near full capacity."

This meant that the code needed to be able to anticipate peak demand to avoid wasteful spillage at this end. Further downstream, the hydro lakes are further apart, with water releases upstream taking several hours to reach the next lake.

"Another special feature is that one geothermal and one thermal station use the Waikato River water for cooling, and minimum flow rates must be maintained past both stations. Also the first station Aratiatia must spill water twice daily for one and a half hours, to maintain flow through scenic rapids."

The general aim of the management system is to schedule the hydro generation of each station so that the overall system demand is met with the minimum usage of water, Broughan explains. This has to done while ensuring that all lake levels are maintained within the operating ranges, and that economic factors are taken into account.

Seven separate complex mathematical phases are involved in the program which cover everything from lake level constraints to getting the data into a usable format in the first place. The model was checked against the real world, using historical data from the actual operation of the hydro system.

"Our code represented a significant improvement on the present methods," says Broughan. Savings of just a few percent can mean tens of thousands of dollars saved, as well as large quantities of water.