Renewable generation is expected to play an increasingly important role in the electricity grid in Australia and other countries. A significant challenge associated with many forms of renewable generation such as wind and solar PV is that the level of power generation is highly variable and can only be predicted in a limited way. Energy storage provides a way of smoothing the supply, managing the demand, and matching the two. However the cheapest form of storage for large amounts of energy (hydro) is very limited as an option for Australia given its geography. Other forms of storage (various types of batteries, compressed air, etc.) are possible but more expensive. Hence a challenge is to determine the minimum amount of storage that is required and how this storage should be used, taking into account multiple uses of storage:
• Short term smoothing: responding to rapid changes in power output (sudden drops or spikes) while giving other power generators a chance to adjust.
• Demand shifting: storing energy over a period of hours or a day or two in order to match the availability of power with periods of demand.
• Transmission upgrade deferral: storing energy close to sources of generation and moving it at a steady rate to avoid the need for larger capacity or additional transmission lines.
Clearly there is also a trade-off between using storage and installing additional capacity of generation to help cover peak demand or additional transmission capacity.
Hence, to determine the ideal amount and location of storage, it is necessary to consider the overall network design in the presence of significant uncertainty in both supply and demand.
Answering questions regarding the optimal energy and power capacities of storage, where it should be installed and how to use it requires the solution of large stochastic programming problems with a strong network design element and potential multiple objectives. Each of these areas is challenging in its own right. It is expected that significant research is required into efficient formulations and effective algorithms in order to be able to solve instances of a reasonable size. This PhD will focus on the formulation and algorithm development with emphasis on the use of parallel heuristics based on mathematical programming ideas. Exact methods will also be tried but are expected to provide only heuristic solutions for realistic instance sizes.
The problem of how much storage to install and how to use this storage can be formulated in terms of a large scale stochastic programming problem with a time-layered network design structure. This project will include the following activities:
• Formulate the question of how much energy storage to install and how best to use this storage as a mathematical optimisation (stochastic programming) problem. It is expected that there will be a number of sub-problems that will be identified as potentially of interest depending on the assumptions made, e.g. optimisation of existing storage, sizing of a single storage facility at a fixed location, green-field network design and so on. Given CSIRO has already done some work on control of storage facilities for smoothing of renewable generation, this project is intended to focus on electricity network planning and support, potentially with significant levels of renewable energy generation.
• Create test data sets based on the structure of the Australian electricity grid and typical usage. The aim is not to carry out exhaustive numerical studies or to create comprehensive data sets, which is a non-trivial research exercise in its own right. Rather, relevant data will be obtained from parallel projects carried out by CSIRO, some renewable energy output data will be estimated from climate models, and where necessary data may need to be randomly generated based on patterns observed in Australian networks. However, the models and algorithms to be developed in the PhD project will be guided by the problem structure encountered in Australia.
• Develop custom algorithms that can deal with these large optimisation problems. This is expected to build on:
|•||Experience by the supervisors in dealing with network design problems and multi-criteria optimisation algorithms|
|•||Literature on stochastic programming/network design mostly in the context of stochastic supply chain optimisation.|
|•||Lagrangian heuristics and approximate dynamic programming methods.|
|•||The availability of multi-core and cluster computers that to speed up computation through the use of parallelisation. While this is relatively straight forward from a software implementation point of view, algorithms will need to be designed with the aim of parallelisation in mind.|
• Analysis of algorithms in terms of computational performance. Some comments will also be made on the structure of solutions such as conditions under which storage becomes a significant part of the optimal configuration. However it is not expected that there will be enough time as part of this project to assemble comprehensive and clean data sets so that the conclusions will not include specific recommendations for the future development of the Australian grid.
For further information, including proposed academic and industrial supervisors, please visit the RMIT Industry Doctoral Training Centre (IDTC) website.
RMIT University, as a member of the Australian Technology Network of Universities (‘ATN’) and collaborator on the Industry Doctoral Training Centre (Mathematics & Statistics) (‘IDTC’) initiative, is currently offering this scholarship to Australian or New Zealand citizens or permanent residents of Australia who are interested in undertaking doctoral degrees in industrial mathematics, applied mathematics or applied statistics and working in collaboration with an industry partner on a project of immediate relevance to Australian industry.
Study Subject: Mathematical Sciences - Operations Research
Web address: http://www.rmit.edu.au/browse;ID=6gsbcgtpswyi
Provided by: RMIT University
To be undertaken at: RMIT University, in collaboration with CSIRO.
One scholarship is available for a high achieving student interested in taking up higher degree by research studies, working on a project of relevance to CSIRO, leading to the award of the degree of Doctor of Philosophy.
The value of this scholarship is AUD30000 (per annum) for up to 4 years, subject to satisfactory performance evaluated annually. This award constitutes a living allowance for the discretion of the student. Students of the IDTC program will be required to undertake scheduled travel within Australia to engage in coursework and other professional development activities in partial fulfilment of the conditions of the IDTC program. The associated travel costs will be met by the industry partner and/or RMIT University.
To be eligible for this scholarship you must:
• be an Australian citizen or an Australian permanent resident
• be commencing your PhD full time in semester 1, 2012; and
• have either of the following:
- at least a first class Honours degree in any field of Science or Engineering or
- other approved qualifications or expertise
Students from non-mathematical degrees are encouraged to apply.
Applications for this scholarship must be made directly to RMIT University. For more information please visit the RMIT Industry Doctoral Training Centre (IDTC) website.
The closing date for applications is Friday 31 October 2011.
Terms and conditions are subject to change. Always confirm details with the scholarship provider before applying.
Dr John Gear (RMIT IDTC Program Coordinator)
RMIT University, School of Mathematical & Geospatial Sciences
Phone: +61 3 9925 2589
Dr Melih Ozlen
RMIT University, School of Mathematical & Geospatial Sciences
Phone: +61 3 9925 3007