Decentralised Thermal Storage to Stabilise Electrical Distribution Networks

The saturation of greenhouse gasses in our shared atmosphere must reduce to sustain life as we know it. We must rapidly decarbonise our energy sector, thereby transitioning from fossil fuel to renewable sources. Such a transition may be difficult at the required scale, as traditional electrical distribution networks were built on centralised power generation with minimal storage and not intended to operate with fragmented and erratic power generation. Renewable power generation has left coal fired power plants in fiscal uncertainty, such that many planned new coal fired power plants are being abandoned in the capital planning stage. However, given renewable power generation depends on local climatic conditions (wind/solar radiation etc) there is not reliable renewable power to meet electricity on demand. One logical solution to assist uninterrupted electricity supply is to install substantial energy storage. The challenge is that electrical energy storage is expensive at up to USD600/kWh, and at present, not feasible at the scale that will allow 100% renewables by the global target of 2050. Electrical grids that have transitioned to high renewable penetration are demonstrating the challenges and subsequent opportunities of advanced grid energy management. The South Australian electricity system is world leading with the highest installed capacity of rooftop photovoltaics, now exceeding 1 kWp per capita. All coal-fired power plants have been retired, significantly reducing the rotational inertia that is critical to maintain network frequency control. To assist in supply security, the world’s largest battery (at the time of installation), gas peaking plants and increased interconnects to the National Electricity Markets (NEM) have been installed. However, additional controls are needed to alleviate limitations in renewable generation and further transition to 100% renewables. Residential storage electric water heaters consume on average more than 2 MWh per year each (~25% of domestic energy consumption), representing a potential national controllable load of 10.8 GW, and total annual consumption of approximately 5.94 TWh. Typical domestic thermal loads peak in the early morning and early evening, which unfortunately align with peak electrical demands. However, by exploiting thermal storage, electrical consumption to heat water can be shifted to more favourable grid times, whilst providing un-interrupted thermal supply to the home. Consumers are financially incentivised to self-consume rooftop PV, however as shown in this thesis, classic storage electric water heaters can only access 12.7 % of electricity from rooftop PV. Whilst timers to align energy consumption to periods conducive to PV generation can substantially increase PV contribution to 38.1 %, development of multi-zone water heaters with modulated energy control can result in a PV contribution of 84.2%, reducing the annual grid consumption of each water heater to 0.4 MWh. Such improvements are only available by reducing emergency heating volumes in the water heater and maximising the thermal storage potential of the water heater for PV self-consumption. In assessment of air source heat pumps (HP) under the same conditions with the incremental control of a simple timer, the annual grid energy consumption can be reduced to 0.28 MWh. Of significance is that both advanced immersion element and HP water heaters use comparable purchased energy when PV-self-consumption is optimised. Modern electrical grids operate in a liberal market, where the wholesale real-time energy price and the provision for frequency control ancillary services (FCAS) is settled based on supply and demand. The South Australian grid with more than 50% renewable supply is demonstrating economic volatility, challenging market operators to maintain power quality. To accurately assess the potential financial and grid benefits of residential water heating demand response that is sympathetic to grid conditions, numerous quasi-dynamic time-series simulations were developed. Time-series inputs include annual recorded climatic conditions, influencing thermal storage heat loss and electrical energy produced by rooftop PV. The PV electricity that is available at any point in time is determined by subtracting the simulated PV generation from the domestic electrical consumption (excluding water heating) as obtained from field data in the South Australian electrical network. The water heating thermal demand was also determined from recorded field data from the same electrical network. The last significant time-series input was that of the grid conditions (real-time and frequency reserve price) as obtained from the Australian Energy Market Operator (AEMO). Using the time-series input data and detailed thermodynamic modelling of a fleet of water heaters, , it is shown that wholesale energy expenses can be reversed, reducing water heating energy costs by 113% to 217% (meaning consumers can get paid) depending upon household occupancy, incorporating FCAS contingency reserves that produce approximately 40% of the identified savings. Peer-to-peer (P2P) energy trading is emerging to increase renewable energy hosting capacity. Whilst individuals can optimise PV self-consumption with up to 90% of water heating energy originating from their own rooftop PV, only 39% of the grid rooftop PV is self-consumed without P2P. To simulate a fleet of water heaters a novel architecture is developed to interconnect multiple time-series simulations with Python scripts, critically retaining thermodynamic modelling accuracy of individual consumers, whilst unlocking centralised communication and aggregated control. Using this new tool, P2P energy transactions more than doubled the network PV self-consumption to 83%. P2P was shown to reduce the average water heating energy costs by 31%, whilst reversing the deleterious grid impact of PV generation. Whilst it is tempting to extrapolate the demand response potential of storage electric water heaters to HP water heaters, HP’s operate with inherent demand response limitations. Examples of such limitations are less income generated from negative real-time price conditions and increased response times that render HP’s unable to participate in the lucrative FCAS market. However, literature indicates that advances in PV self-consumption and optimising coefficient of performance (COP) can be obtained by controlling HP compressor speed. This thesis contributed to literature by developing new methods to test and characterise variable speed HP water heaters, allowing time-series simulations to optimise PV self-consumption and reduce electrical consumption and associated costs. By intelligently selecting compressor speed, HP electrical consumption could be reduced by 28% without rooftop PV or battery energy storage. When combining variable speed HP and rooftop PV, the HP grid energy is reduced by 88%, with this extended to 99% when utilizing rooftop PV and a 13.5 kWh battery. By exploiting the interdependence of thermal and battery energy storage coupled with rooftop PV, the annual household electrical costs can be reduced by ~$1600, where HP and battery controls contribute ~$100 and ~$500 respectively. To stretch the effectiveness of battery storage, machine learnt daily electrical deficits are predicted using boosted regression modelling, proving very effective in delivering 99% of the potential savings, increasing battery annual savings to $574. However, given the large capital expense of batteries, only ~1/3 of the capital expense is recovered in a lifetime of 10 years. Whilst the capital expense of the variable speed compressor control is recoverable in a similar lifetime, the lowest net lifetime cost option was again apparent using a fixed speed compressor and simple timer control to exploit probable rooftop PV, reducing household lifetime costs by 27%. This thesis concludes that consumers can realise substantial reductions in purchased grid energy with advanced HP and storage electric water heating design and controls. A symbiotic relationship is shown between grid operators and consumers, such that profitable business models can be developed, with the net effect of aiding grid operation and subsequently advancing the transition to 100% renewable energy.