ITP Renewables’ open-source capacity expansion model of the National Electricity Market (NEM) – openCEM – is now available free of charge at www.openCEM.org.au.
openCEM has been developed by ITP’s Analytics division, together with the Centre for Energy and Environmental Markets at University of New South Wales, the Energy Transition Hub at the University of Melbourne, and the software development specialists ThoughtWorks.
We expect users of the modelling tool to include policy makers, project developers, investors and the interested public. Users can run unlimited scenarios to explore the implications of a wide range of input assumptions including technology characteristics and costs, and electricity demand profiles, out to 2050.
openCEM optimises for a least-cost solution that maintains energy security. It reveals when, where and what kind of energy generation and storage capacity, and transmission capacity should be added, and when and where carbon intensive generation should be retired, in order to achieve user specified outcomes.
One problem that openCEM solves is that similar models are privately held, expensive to commission, and offer the public limited detail on how they actually work. By contrast, openCEM is free to use and open to scrutiny and expert review. Users can inspect every line of code and can control every input assumption. We expect a community of expert users to collaboratively improve the model over time.
The development of openCEM received funding from the Australian Renewable Energy Agency as part of its Advancing Renewables Program and the Governments of New South Wales, Victoria and South Australia.
There were also significant in-kind contributions from the development partners, and the US Strategic Energy Analysis Center of the National Renewable Energy Laboratory. The development process was also generously assisted by many firms and organisations participating in an Industry Reference Group.
The operation of the openCEM involves two steps. For every simulated year, a capacity expansion model finds the least cost additions to generation, transmission and storage from an existing NEM configuration required to satisfy forecasted electricity demand, subject to renewable energy resource availability, and economic retirement of existing generation capacity.
Step two involves running a chronological dispatch model of the NEM under the capacity conditions obtained in the first stage. This stage verifies how the electricity system operates, on an hourly basis over a given year, ensuring reliability standards are met and that the load is met under all conditions, including peak demand events. The resulting generation, storage and transmission decisions obtained after this process carry forward to the next year, and so forth.
Starting assumptions about technology and fuel costs and performance characteristics are taken from the Australian Energy Market Operator’s (AEMO) Integrated System Plan (ISP).
For example, the “base case” scenario takes most of its cost, performance and demand data from the ISP Neutral Scenario, and shows results that are qualitatively similar to the ISP modelling.
Each time AEMO publishes updated data, or other sources of data become available, it will be possible to update openCEM’s starting assumptions. Other scenarios to be added will be based on the CSIRO’s GenCost data.
openCEM subdivides the NEM into the same 16 planning zones that AEMO uses for the ISP and their National Transmission Network Development Plan.
For each zone the capacity expansion model considers a configurable list of generators, storage devices and hybrids (generator + storage). Intra-regional transmission power flow is modelled in the form of “zonal interconnectors” between zones and the cost and benefit of building new interconnectors between planning regions at selected locations is evaluated.
It is also important to note that openCEM optimises capacity and dispatch to minimise the cost of electricity. In essence it has ‘perfect knowledge’ of demand, weather and available capacity and its decisions to build new assets (generation, storage, transmission) and how to operate them is based exclusively on the marginal costs and cost trade-offs between alternatives.
Capacity expansion models like openCEM do no attempt to model the bidding behaviour (strategic or otherwise) of market participants or to take into account complex electricity supply contracts that distort the idealised operation of the electricity market.
There are generally three levels of energy sector models to evaluate a diverse set of metrics depending on the scale (time scale, geographical scale) and level of detail required for a particular evaluation.
The first are tools that assess the functioning of discrete parts of an electricity system down to sub-second intervals, with a focus on managing frequency and voltage. The second are grid integration models (such as openCEM) that can assess the evolution of entire networks over times frames of many years. The third are tools that focus more on how economic activity in other sectors will impact electricity demand and supply costs.
Due to computational limitations, lower level models are confined to parts of the grid and short timeframes; grid integration models simplify the behaviour and resolution of generation and transmission; and high level multi-sector models span entire economies at multi-year scales, but with coarse assumptions on system behaviour. The intermediate approach of openCEM combines multi-year system evolution with hourly dispatch calculations.
There are two ways to use the modelling tool. The simpler option is to explore a range of “pre-run” scenarios (based on a set of identified assumptions) on the openCEM website.
Results such as generation capacity, dispatch and wholesale electricity cost are displayed visually using a range of graphics. At this stage ten scenarios have been selected to demonstrate the capability of the model, but more will be added in response to user feedback and requests.
The more sophisticated option is for users to download and install openCEM on their computer. Users can then run their own scenarios, with tailored assumptions about inputs such as technologies, policies and demand profiles.
Note that a single run of the model typically takes 2-3 days on a good desktop computer, and setting up the scenario consists of filling some information in scenario input files that requires some basic understanding of how openCEM works.
After the simulation is finished, a complete dataset of the simulation is produced that contains all input assumptions (e.g. costs, policies, traces, demand, etc.) as well as all dispatch and building decisions.
The openCEM code repository features installation and usage instructions, and documentation about how all the output data is organised, as well as a more detailed background on the model methodology.
Also, at the time of release, a separate repository called openCEM_examples will host the input files for 10 different scenarios, which gives a complete starting point for novice users to run scenarios in their own computers and start experimenting with the tool.
ITP will keep developing openCEM, and our hope is that, consistent with the philosophy of open-source projects, a collaborative community of users will emerge and support future improvements.
Please explore openCEM at www.openCEM.org.au.
Oliver Woldring is Strategy Group Manager, ITP Renewables