Transpower Ensures Reliability of New Zealand National Grid with Reserve Management Tool
Challenge
Solution
Results
- Critical updates rapidly implemented
- Simulations verified using real data
- Updates made in-house
“We record frequencies on the grid, inject them into our Simulink model, and compare the simulation results to the actual system response. With Simulink we can continually calibrate and improve our model, and ultimately improve the accuracy of our reserve estimates.”
Transpower owns and operates the New Zealand national grid, a high-voltage transmission network that delivers power to industrial, business, farming, and domestic electricity consumers. Transpower must balance load and generation to keep the grid’s frequency at 50 Hz. If one generator fails, power from other generators must be made available within a few seconds to offset the loss and ensure the system frequency doesn’t fall to a level from which recovery is not possible.
Because it is costly to hold generated power in reserve, Transpower must precisely calculate the minimum reserve needed to ensure the reliability of the grid within established risk tolerances.
To meet this need, Transpower has developed the Reserve Management Tool (RMT) in MATLAB® and Simulink®. RMT includes a sophisticated model of the entire grid, including generators, loads, and high-voltage direct current (HVDC) links between New Zealand’s two main islands. Simulations based on up-to-the-minute grid information are run every half hour to determine the currently needed reserve.
“Transpower has two directives. One is to make sure the lights stay on by operating the grid reliably, and the other is to do this economically,” says Heidi Heath, investigations engineer at Transpower. “By driving Simulink simulations with real-world data, we not only make better tradeoffs between cost and reliability, we actually reduce costs while improving service.”
Challenge
In the past, Transpower used spreadsheets to calculate the required minimum reserve. This approach required a technician to perform numerous manual steps and run several supporting scripts. The system was inflexible, making it difficult to incorporate new generators as they came on line. Transpower needed a reserve management tool that could be updated rapidly as the grid changed.
“Over time, we needed to comply with more and more sophisticated rules to handle contingent events, and the cost pressures increased the need to estimate required reserve more accurately,” says Heath. “We anticipated adding more generators, as well as making changes to the HVDC links. We needed a tool that we could adapt in-house so that we would not need to wait for a third-party company to make these changes.”
Solution
Using MATLAB and Simulink, Transpower built a new RMT to model and simulate New Zealand’s entire national grid.
Transpower engineers created MATLAB and Simulink models of wind, hydroelectric, and geothermal plants, which make up most of the generators on the grid. The individual plant models use the input frequency and proportional-integral-derivative (PID) controllers to calculate power output. The PID controller gains were tuned using Control System Toolbox™.
To develop Simulink models for the more advanced combined-cycle thermal plants, Transpower used block diagrams and test data provided by the companies operating the plants.
The engineers also developed models for more than 30 interruptible and non-interruptible loads distributed across the country’s two main islands, as well as for the HVDC links that connect the islands.
Working in Simulink, engineers assembled a complete model of the grid with models of the loads, HVDC links, and approximately 80 generators.
Using MATLAB Compiler™ and Simulink Coder™, the team created a standalone executable of the complete model. The Transpower IT team integrated the model with a Scheduling Price Dispatch (SPD) tool, which is used to procure the required reserve at the lowest price.
The new RMT tool is currently run every 30 minutes to calculate the required reserve.
Results
Critical updates rapidly implemented. “When a new generator is added or some other change is made to the grid, we need to update RMT quickly,” says Heath. “Over the past year, we’ve made 18 such updates in Simulink, which would have been more difficult and less accurate with our spreadsheet-based system.”
Simulations verified using real data. “When a system event occurs, we can drive our Simulink model with measured data, see if RMT predicted results accurately, and verify that the generators responded as expected,” says Heath. “This approach enables us to improve the accuracy of our grid model and individual generator models, and it helps us identify possible grid improvements.”
Updates made in-house. “With MATLAB and Simulink, we can make changes to the system using our own expertise instead of having to rely on an outside vendor,” says Heath. “As a result, we have full control over when the updates will be completed, as well as control over sensitive data.”
