A Mixed Integer Linear Unit Commitment and Economic Dispatch Model for Thermo-Electric and Variable Renewable Energy Generators With Compressed Air Energy Storage
The objective of this PhD thesis is to create a Unit Commitment and Economic Dispatch (UCED) modelling tool that can used to simulate the deterministic performance of a power system with thermal and renewable generators and energy storage technologies. The model was formulated using mixed integer programing (MIP) on GAMS interface. A robust commercial solver by IBM (CPLEX) is used as solver. Emphasis on the development of the tool has been given on the following aspects. a) Technical impacts of Variable Renewable Energy (VRE) integration. The UCED model developed in this thesis is a high resolution short-term dispatch model. It captures the variability of VRE power on the intra-hour level. In addition the model considers a large number of important real world, system, unit and policy constraints. Detailed representation of a power system allows for a realistic estimation of maximum penetration levels of VRE and the related technical impacts like cycling of generators (part-loading and number of start-ups). b) CO2 emissions. High levels of VRE penetration can potentially increase consumption of fuel in thermal units per unit of electricity produced due to increased thermal cycling. The dispatch of units in the UCED model is based on minimizing system wide operational costs the most important of those being fuel, start-up costs and the cost of carbon. Fuel consumption is calculated using technical data from Input/Output curves of individual generators. The start-up cost is calculated based on times the generator units have been off and the energy requirement to bring the unit back to hot state. Thus dynamic changes on fuel consumption can be captured and reported. c) Technical solutions to facilitate VRE integration. VRE penetration can be facilitated if appropriate solutions are implemented. Energy storage is an effective way to reduce the impact of RE variability. The UCED model includes an integrated Mixed Integer Linear (MILP) compressed air energy storage (CAES) simulation sub-model. Unlike existing CAES models, the new “Thermo-Economic” (TE) CAES model developed in this thesis uses technical data from major CAES manufacturers to model the dynamic effect of cavern pressure on both the compression and expansion sides during CAES operation. More specifically the TE model takes into account that a) a compressor discharges at a pressure equal to the back-pressure developed in the cavern at each moment, b) the speed of charging can be regulated through inlet guide vanes; higher charging speed can take place at the expense of additional power consumption, c) the maximum power output during expansion can be limited by the levels of cavern pressure; there is a threshold pressure level below which the maximum output decreases linearly with pressure. Since it uses actual power curves to simulate CAES operation, the TE model can be assumed to be more accurate than conventional Fixed Parameter (FP) models that don’t model dynamic effects of cavern pressure on CAES operation. The TE model in this thesis is compared with conventional FP models using historical market prices from the Irish electricity market. The comparison was based on the ability of a CAES unit to arbitrage energy for making profit in the Irish electricity market. More specifically a “Base” scenario was created that included the operation of a 270MW CAES unit with technical characteristics obtained from a major CAES manufacturer and assumed discharge time of 13hr. Various sensitivities on discharge time, natural gas prices and system marginal prices (SMPs) were modeled. An additional scenario was created to show the benefit on CAES profitability if the unit participated in both the energy and ancillary services markets. All scenarios were modeled using both the TE and FP CAES models. The results showed that the most realistic TE model returns around 15% less profitability across more scenarios. The reduction in profitability grows to around 30% when the cavern volume (discharge time) is reduced to half (6 hours). The latter is related to the sensitivity of the TE model on cavern pressure that is being built faster when the volume is reduced. A CAES unit won’t get a positive net present value (NPV) in Ireland under any scenario unless SMPs are greatly increased. Thus, it was shown that that existing FP CAES models overestimate CAES profitability. More accurate models need to be used to estimate CAES profitability in deregulated markets. Additionally, it might deem necessary to create additional markets for energy storage units and increase the possible revenue sources and magnitude to facilitate an increase of storage capacity worldwide. The second step of analysis involved the integration of the CAES and UCED models. The UCED model developed in this thesis was validated and applied using data from the Irish grid, a power system with more than 50 thermal generators. A vast of existent data was used to create a mathematical model of the Irish syste