2018 projects

Planning electricity transmission networks

ElectraNet is responsible for the long-distance, high-voltage transmission of electricity throughout South Australia, connecting traditional and renewable power generators and large industrial customers to the transmission network and connecting South Australia to the eastern states.

Stochastic generation technologies, including distributed rooftop solar along with emerging grid scale, distributed and even mobile (electric vehicles) battery technologies are changing the way that the transmission network is used. Emission reduction goals will ultimately see the fleet of conventional coal and gas fleet retired. To plan future transmission infrastructure, ElectraNet must predict, up to 20 years in advance, when and where power will be generated and used.

To make this problem tractable, the current planning approach has two stages. First, a long-term plan is developed. The long-term plan takes into account existing generation, renewable resources that might be used for future generation, and load predictions. Load is modelled as a sequence of monthly load distributions for the next 20 years. Each monthly load distribution indicates the proportion of time that the load will exceed a given level during the month, but does not specify when. The long-term planning process uses a large linear program to optimise the fleet of generators that will be used to meet the monthly load distributions.

Next, the long-term plan is tested by simulating the operation of the proposed transmission network and generator fleet for the next 20 years, using hourly time steps. At each time step, the optimal mix of generators is selected from the proposed fleet.

This second stage often shows that the first stage—the long-term plan—undervalues interconnectors, because it does not model the hourly variability of generation and the resultant power flows in the transmission network. Furthermore, the impact of storage, which is likely to cycle daily, may not be effectively taken into account in the long-term model.

The aim for MISG is to suggest modelling and optimisation approaches that will more accurately reflect the increasingly stochastic nature of future generation in the long-term modelling of the system.

Inertia in electricity networks

In electricity generation systems, large thermal or hydro power plants use rotating synchronous machines to generate three sinusoidal voltage phases with a frequency near 50 Hz. The electrical flux and mechanical dynamics of a synchronous generator can be modelled by well-known differential equations that depend on terms including the mechanical torque being applied to the generator, the electrical load on the generator, and the moment of inertia of the rotor.

When synchronous generators are connected together, they synchronise to form a stable equilibrium. If the load on the grid increases then the increased load will be shared automatically amongst the synchronous generators in a way that maintains synchronisation. Generators with higher inertia will take a higher proportion of the additional load, and all generators will slow at about the same rate until the mechanical power applied to the generators can be increased or load can be reduced. This slowing of the generators causes a drop in the frequency of the grid power. Similarly, if the load on the grid decreases then the frequency will rise. The rate of change of frequency (RoCoF) is a key indicator of system stability; if frequency changes too quickly then the system will become unstable and may have to be shut down.

New renewable generation systems, such as wind turbines and photovoltaic power systems, use power electronics to control their output to ensure that they are synchronised to the traditional generators. They have zero or low effective inertia. The growth of these loosely-coupled generators is reducing the total inertia in the system, which can impact on the stability of the system. Furthermore, the new generators are scattered across the power system, and can be far from big load centres and from synchronous generators.

The aim for MISG is to help ElectraNet understand the impact of low-inertia generation, far from loads and synchronous generation, on the stability of the South Australian power system. How should the equivalent inertia of a wind farm be calculated, and how does it impact the rate of change of frequency in the system? Ultimately, the aim is to understand how energy storage and devices such as synchronous condensors (large rotating synchronous machines that provide mechanical inertia without generating) can be used to stabilise power systems.

Optimising carcase cuts in the red meat industry

The red meat industry is moving into a new era of objective carcase measurement. Until now, meat processors have sorted carcases based on weight. New measurement technologies give detailed, accurate information on the amount of lean meat, fat and bone tissue in each carcase before it is processed. The lean meat and fat tissue present on a carcase influence the profitability; generally, carcases containing more fat have a lower profitability. Profitability of a carcase is also influenced by the market prices being paid and the types of meat cuts being used. If a meat processor is able to leave more fat on a given set of meat cuts then they are able to reduce wastage and increase profitability, due to the vastly higher wholesale price of lean meat versus fat trim.

The red meat industry has started to develop an optimisation tool to determine the most profitable options for each carcase, given the cut orders received from each market. However, this optimisation tool is very specific, allocating individual cuts from individual carcases for a fixed set of carcases and orders. The tool needs to be more flexible, because new carcases arrive and orders change during processing. Our problem is to determine what types of approaches may be taken to increase the flexibility of the optimisation, and how we can re-work the tool to make it more useful in the ever-changing environment of a meat processor.

Combining ABS publicly available data with other publicly available data

There is a wealth of publicly available data produced by Commonwealth and State and Territory Government agencies. The Australian Bureau of Statistics (ABS) produces—a official economic, social and population statistics and data. Other agencies publish data that if combined with ABS data, can help provide more information about policy or issue of interest. For example, the ATO publishes personal income tax data by postcode. State and Territory government agencies publish data sets covering education, health, social conditions, employment, income and wealth. All of these data sets are available at different geographic levels and so are therefore difficult to combine. For example, the ATO publishes aggregated incomes by suburb, whereas the ABS uses the Australian Geographic Statistical Standard.

The ABS is seeking to develop techniques to analyse these geographically different data sets to identify patterns and correlations that will support enhanced decision making. The key questions include:

  • What can we learn about sub-populations or sub-state areas by combining publicly available ABS data with other publicly available data?
  • Can unstructured searches of data trained on high-quality census and survey data be used to discover characteristics of sub-populations, provide estimates of future characteristics, and identify leading indicators?