Faculty Members
Assoc. Prof. Andrea Raith Email Assoc. Prof. Andrea Raith
Dr. Geoff Pritchard
Postdoctoral Fellows

Visiting Researchers in 2008
Visiting Researchers in 2009
Visiting Researchers in 2010
Visiting Researchers in 2011
Visiting Researchers in 2012
Visiting Researchers in 2013
Visiting Researchers in 2014
Visiting Researchers in 2015
Visiting Researchers in 2016
Visiting Researchers in 2017
Visiting Researchers in 2018
Visiting Researchers in 2019
Visiting Researchers in 2020
Visiting Researchers in 2021
Visiting Researchers in 2022
Visiting Researchers in 2023
Visiting Researchers in 2024
Dr Jannik Haas
Dr Hans Christian Gils
Manuel Wetzel
Prof. Michael Ferris
Current Graduate Students
Dominic Keehan
Rishi Adiga
Electric Power Optimization Centre
GEMSTONE is the name we give to the EPOC project that aims to develop a zero carbon energy system for New Zealand. The core problem underlying GEMSTONE is to determine the scale and timing of investments and shutdowns of electricity generation over a long time horizon.
GEMSTONE entails a suite of different models at different time scales:
POUNAMU (html)
Daily operation to deal with renewable intermittency.
JADE (html)
Medium-term operation to deal with dry-year risk.
EMERALD (html)
Long-term investment in generation capacity.

Recent GEMSTONE Projects
Security of Supply in the New Zealand Electricity Market (report) (literature survey)

The New Zealand Government recently announced that the nation's electricity should be entirely produced by renewable sources 'in a normal hydrological year' by 2035. As a result, power stations that use coal and gas are being phased out, the largest of which is Huntly Power Station. This report examines the effect of Huntly's units being shut down, and how the potential deficit in supply could be resolved.

To do this, policies of when and where to release water from New Zealand's seven largest reservoirs and how much electricity to produce by thermal means are constructed using Stochastic Dual Dynamic Programming. These policies are then simulated using historical data which captures the weekly inflows of water into each of the reservoirs. The various policies and simulations represent various combinations of electricity-producing units at Huntly being available, as well as renewable energy sources being added into the New Zealand Electricity Market.

Huntly Power Station (Huntly) has both coal and gas units and is owned by Genesis Energy, which is one of the five big electricity-generating companies in New Zealand. We investigate the role that Huntly plays in the electricity market, which companies would benefit if Huntly's coal units are shut down, and which generation sources would be best suited to making up the deficit of electricity supply.

From our analysis, it is apparent that the removal of units at Huntly increases the strain on New Zealand's hydro network. Water becomes more valuable, and this impacts the estimated distribution of operating surpluses for the five largest market participants. If Huntly were to shut down entirely, electricity shortages would become more likely. Hence, it is clear that new generation assets will be required in order to account for the deficit in electricity supply if some or all of Huntly's units are shut down. We postulate that geothermal and wind energy are best suited to replace Huntly's coal units, and that the addition of all currently-consented geothermal and wind energy sources into the New Zealand Electricity Market would significantly decrease the carbon emissions associated with the generation of electricity.
Infinite-Horizon in Stochastic Dual Dynamic Programming (report) (literature survey)

The New Zealand government aims for 100% of New Zealand’s electricity generation to come from renewable sources by 2035. This objective is causing additional uncertainty around the future of Huntly Power Stations' coal-fired units which already have intermittent use because of their function as a 'peaker' during periods of extended low reservoir levels.

Determining the future of Huntly requires a model of the New Zealand Electricity Market (NZEM). This project builds on previous research. Hydro-thermal scheduling models of the NZEM such as JADE and DOASA have been used to research the value of Huntly in the NZEM as well as future renewable generation mixes.

JADE and DOASA models are solved using the stochastic dual dynamic programming algorithm (SDDP). An explicit assumption of SDDP is an exogenous, predefined terminal marginal cost function. This assumption reduces the accuracy of these models and their results.

We extended the JADE model to an 'infinite-horizon' SDDP with an endogenous terminal marginal cost function. Computational improvements reduced the run-time of the 'infinite-horizon' SDDP from greater than 40 hours down to 30 minutes thus enabling accurate solutions to determine the value of Huntly's coal-fired units and the future renewable generation mixes.
On reducing New Zealand's CO2 emissions (presentation)