Intelligent Energy
We are developing new ways of controlling and managing energy usage and generation through the application of techniques in machine learning, optimisation, and automatic adaptive control. Our research is being applied to the problem of controlling local renewable energy generators, residential energy consumption, and the energy consumption of large electrical loads such as heating, ventilation and air-conditioning (HVAC) systems. Ultimately our research outcomes will allow a significant reduction in CO2 emissions associated with small-scale electricity generation, as well as more effective use of electricity in large loads such as air-conditioning systems. Our technology will provide information to consumers that will afford them greater choice and enable them to responsibly modify their consumption patterns.
These projects form part of CSIRO's Energy Transformed Flagship program, and involve researchers from both the ICT Centre and CSIRO's Division of Energy Technology.
Smart Grids
Traditional national and international electricity grids have been designed to optimise the distribution of energy from a limited number of large power stations capable of sustained output. The capabilities of these grids are being challenged by a growing number of distributed renewable energy resources, and pressure from consumers seeking access to cheaper "greener" energy. Electricity networks will need to become "smarter" in order to cope with intermittent generation, increasing demand, and the need for more information about the availability and consumption of energy.
This project is addressing the design of the "Smart Grid" of the future, including methods for:
- Planning network topology and optimal placement of sensors (smart meters).
- Anticipating energy demand.
- Detecting faults and anomalies.
- Recovering from cascading failures and blackouts.
- Developing likelihood v. severity profiles for different types of faults.
- Reconciling multiple objectives (e.g., demand vs. supply, customer comfort vs. utility profits vs. CO2 reduction, etc.) to achieve real-time control of the distributed grid.
Intelligent HVAC Management
Heating, ventilation and air-conditioning (HVAC) systems in commercial buildings play an important role in regulating the indoor environment in order to provide people with a comfortable and safe work environment. In most buildings the performance of the HVAC system can significantly influence energy consumption as well as indoor air quality, so HVAC system management and control is an active and important research area.
As HVAC systems in buildings become increasingly larger and more complex, they require close monitoring to evaluate their performance with respect to energy efficiency and equipment operation. The increasing complexity has made it very difficult to detect problems caused by improper installation, inadequate maintenance, or equipment failure. Typical issues include mechanical failures such as stuck, broken, or leaking valves, dampers, or actuators; control problems related to failed or drifting sensors, poor feedback-loop tuning or incorrect sequencing logic; fouled heat exchangers; design errors; or inappropriate operator intervention. Such faults often go unnoticed for extended periods of time until the deterioration in performance becomes great enough to trigger comfort complaints, cascading equipment failure, or excessive power consumption.
A fault detection technology that can quickly and accurately detect performance deterioration offers the potential for faster remedial action to improve the indoor environment and ensure the reliability and safety of an entire HVAC entire system. The energy-saving potential of early fault detection is estimated at 10-40% of total HVAC system energy consumption depending on the age and condition of the equipment, maintenance practices, climate, and building use.
In this project we are developing new ways to automatically detect faults in HVAC systems using dynamic machine-learning techniques. Such techniques offer faster set-up, easier retrofit to existing HVAC systems, and ongoing performance improvements over traditional rule-based fault detection systems which require expert knowledge in order to model the operating behaviour and predict error conditions specific to each particular HVAC installation. Our system is capable of automatically learning the characteristics of an HVAC system operating normally, and then using the statistical relationships between groups of measurements to identify anomalous deviations from the norm and identify faults in all subsystems for which sensor information is available, regardless of the specifics of the installation.
We are currently trialling this technology in collaboration with the Hornsby Shire Council.
Local Energy Storage
As
we seek to reduce green-house gas emissions by increasing the
proportion of renewable energy generation, energy storage technologies
will play an increasingly important role in the overall energy system.
This project is assessing the requirements and market for storage
technologies, with a particular focus on storage technologies located
at customer premises or close to load centres. We aim to guide the
application and development of storage technologies and management
algorithms.
The project will consider:
- National analysis of storage requirements as a function of renewable energy penetration and generation technology mix.
- Placement, sizing, and impact of storage on the low-voltage and medium-voltage networks that service urban and rural customers.
- Getting
value from local storage technologies at residential and commercial
customer premises and at strategic network locations.
The Adaptive Systems team is contributing methods for managing local storage, balancing supply and demand, and forecasting renewable generation output.
Minigrids
A minigrid is a set of generators and distribution infrastructure that supplies a localised group of customers. A minigrid may be electrically isolated, or it may be integrated with the central grid but still capable of serving the local load independently. By locating the generators close to the point of consumption, minigrids offer opportunities to improve the economics of meeting energy needs by avoiding the costs of transmitting energy over large distances or transporting fuel from distant supply sources.
This project aims to demonstrate the reliability and efficiency of renewable-energy powered "smart" minigrid systems in order to improve community, government and industry acceptance of the technology. CSIRO's Energy Transformed Flagship will develop and deploy a pilot "Smart Mini Grid" system in Australia that will demonstrate, in practice, how smarter control of renewable energy sources combined with intelligent management of energy-consuming loads (such as heating and cooling) can improve the affordability, efficiency, and reliability of energy supply.
Our researchers have expertise in the areas of:
- Planning of minigrids: The use of mathematical techniques to optimise the design of a minigrid, including considerations such as types of generation, minimisation of electrical losses, and maximising the penetration of renewable generation sources.
- Fault detection: Achieving reliable operation of minigrids is particularly challenging when the grids are operated with high levels of renewable generation and minimal energy storage. We are developing methods for accurately detecting faults with a minimum number of sensors.
- Dynamic load and generation management: Matching energy load with supply can be a challenge in a minigrid. We focus on learning and predictive methods to better anticipate load and available supply, thereby improving the load/supply balance.
- Optimal storage sizing: We aim to minimise the amount of storage needed in the minigrid, by examining the trade-off between amount of storage and system reliability.
- Utility grid interaction: We are investigating the issue of reliability in the utility (traditional) electricity grid, and how minigrid techniques may improve the reliability of the utility grid.
Contacts
Publications
- Joseph Lizier, Mikhail Prokopenko, and David Cornforth. "The information dynamics of cascading failures in energy networks". European Conference on Complex Systems 2009.
- Ying Guo, Astrid Zeman, and Rongxin Li. "A reinforcement learning approach to setting multi-objective goals for energy demand management". International Journal of Agent Technologies and Systems 2009; 1(2):55-70.
- Ying Guo, Rongxin Li, Geoff Poulton, and Astrid Zeman. "A simulator for self-adaptive energy demand management". 2nd IEEE International Conference on Self-Adaptive and Self-Organizing Systems (SASO 2008); Venice, Italy. IEEE; 2008: 64-73.
- Ying Guo, Astid Zeman, and Rongxin Li. "A reinforcement learning approach to setting multi-objective global goals for energy demand management". 7th International Conference on Autonomous Agents and Multiagent Systems (AAMAS '08); Estoril, Portugal. ACM; 2008: 65-72.
- Jiaming Li, Glenn Platt, Geoff Poulton, Josh Wall, and Geoff James. "Dynamic zone modelling for HVAC system control". International Conference on Modelling, Identification and Control (ICMIC 2008); Shanghai, China. 2008.
- Ronxing Li, Jiaming Li, Geoff Poulton, and Geoff James. "Agent based optimisation systems for electrical load management". 1st International Workshop on Optimisation in Multi-Agent Systems (OPTMAS 2008); Estoril, Portugal. 2008: 60-69.
- John Ward, Glenn Platt, and Jiaming Li. "The Virtual Power Station: reliably meeting electricity system demands with photovoltaics". 3rd International Solar Energy Society Conference, Asia Pacific Region (ISES-AP 2008); Sydney, NSW. 2008.
- Jiaming Li, Geoff Poulton, Geoff James, Astrid Zeman, Peter Wang, Matthew Chadwick, and Mahendra Piraveenan. "Performance of multi-agent coordination of distributed energy resources". WSEAS Trans. System and Control. 2007; 2(1):52-58.
- Elth Ogston, Astrid Zeman, Mikhail Prokopenko, and Geoff James. "Clustering distributed energy resources for large-scale demand management". 1st IEEE International Conference on Self-Adaptive and Self-Organizing Systems (SASO '07); Boston, Mass. IEEE; 2007: 97-106.
