Civil Engineering - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemA scenario analysis approach to distributed energy system optimisation( 2016)Distributed Energy Systems (DESs) can provide less carbon-intensive, more resilient and highly efficient alternatives to centralised electricity generation for growing urban populations. Their successful planning depends on the selection of technologies and capacities, which are heavily reliant on an unknown future energy landscape. Furthermore, precinct development, electricity and natural gas price changes, future advances in DESs efficiencies and costs and government interventions can all influence the financial performance of installations. Existing DES optimisation methods typically assess a single year of representative data and yield a single immediate investment solution without capturing or adapting for future shifts in influencing factors. Accordingly, to address the need to include these future changes, this thesis developed a framework that facilitates the selection of optimal DES investment strategies over the lifespan of a project for a range of scenarios at the precinct scale. This approach enables a more representative assessment of DES performance by considering future forecasts as well as providing the additional freedom to defer investment to later during the project life. Building energy performance simulation software (DOE 2.2) with the addition of available measured data was used in conjunction with a bottom-up archetype approach to determine precinct scale electricity and heat load profiles for an analysis period of up to 20 years. Existing hourly intermittent supply models and long-term representative meteorological years were used to estimate solar photovoltaic (PV) and small-scale wind turbine outputs. When adjoined with a combined heat and power dispatch logic formed the electricity and heat supply model. The objective of the optimisation is to minimise the net present value of costs (NPVC) associated with the supply of heat and electricity to the precinct. Optimisation decision variables pertained to investment capacity and technology for each year of assessment. Solar PV installation capacities were approximated as continuous variables and wind turbine and gas generators consisted of integer variables for type and number of units leading to a total of 80 decision variables for a 20-year assessment period. Review and testing of a variety of optimisation algorithms showed the inability of genetic algorithms (GAs) to converge within a reasonable computer run-time. An iterative hybrid approach was therefore developed where GAs were initially employed due to their ability to handle integers and global search strengths after which particle swarm optimisation (PSO) was implemented to optimise continuous variables and the process repeated. The scenario analysis approach developed was tested for the Parkville Campus, the primary precinct of The University of Melbourne between 2016 and 2036. Four scenarios were developed, capturing a range of gross floor area (GFA) growth from 11% to 65% based on the Universities strategic plan as well as future electricity and gas prices, DES investment costs and government interventions. The buildings within the Parkville Campus were categorised based on their use and construction era, resulting in a total of 16 archetypes, where each were assigned an hourly representative electricity and heat demand profile. Electricity demand was estimated based on existing measured data where average annual demand was found to vary from 95 kWh m-2 a-1 for modern office space archetypes up to 266 kWh m-2 a-1 for traditional laboratory archetypes. Heating demand was estimated using building energy simulations for hot water and space heating which were scaled to match monthly measured natural gas demand data. This resulted in modern teaching space archetypes requiring an average of 25 and 99 MJ m-2 a-1 for hot water and space heating respectively whereas traditional cafés and restaurants were found to require 201 MJ m-2 a-1 for hot water and 119 MJ m-2 a-1 for space heating. Different trial runs were conducted for each Parkville Campus scenario, (1) no DES investment, (2) the traditional optimisation approach where DES investment was ‘now or never’ or only considering a single year solution and (3) the long-term optimisation approach with DES investment was allowed at the beginning of each analysis year. Constrained optimal DES investment strategies for all scenarios selected the upper bound of 6.7 MWp of solar PV installations whilst the selection of combined heat and power (CHP) systems ranged from none to 9.3 MWp depending upon the trial case and the scenario in assessment. Scenarios with high GFA growth and the inclusion of nearby residential apartment development provided the best business case for CHP, however the long-term investment strategy recommended deferring this investment until 2022. Both the long-term approach and traditional DES investment strategies provided benefits to the university over the do-nothing approach, with different scenarios yielding: (1) a reduction of the NPVC between 3% and 18%; and (2) greenhouse gas (GHG) emission reductions between 7% and 40%. Compared to traditional optimisation approaches, it was found that the long-term scenario analysis approach resulted in greater reductions in NPVC of between 0.34% for low growth scenarios and 3.76% for high growth scenarios. This translated into a saving of between A$0.9m for low growth and A$14.6m for high growth scenarios for a twenty year NPVC of A$241.7m and A$332.7m respectively. The scenario analysis approach did not consistently yield reductions in GHG emissions compared to traditional optimisation approaches, due to the deferral of DES investment. By allowing deferral of DES investment, the long-term scenario optimisation framework was shown to have several key advantages over traditional ‘now or never’ optimisation approaches, including: reduction in upfront capital expenditure by allowing step wise investment in DES in line with projected precinct electricity and heating demand growth; provision of installation lead time for planning, approval acquisition and allocation of plant area; reduced overall NPVC particularly for high growth scenarios; lower risk of oversized DES as investments are made when required. The long-term scenario optimisation framework requires significantly more data collection and simulation than traditional approaches and is therefore only warranted where precincts are expected to undergo meaningful change over the assessment period. A more detailed heat network loss model and optimal dispatch logic algorithm would be required if this framework is to be used at the functional design stage. Finally, the inclusion of GHG emissions as a second objective and the use of multi-objective optimisation could provide decision makers with an assessment of the trade-offs between NPVC and GHG emission reduction.