Abstract:
The organic Rankine cycle (ORC) is a heat recovery technology with applications in renewable energy generation such as geothermal power and waste heat recovery. In this thesis the ability of model based control and optimisation techniques to increase the value generated from geothermal ORCs is examined. Existing geothermal ORCs use decentralised proportional-integral (PI) control loops to regulate plant operation. This thesis analyses the benefit to be gained by applying advanced process control to geothermal ORCs. The design and operation of geothermal ORCs relies on analysis that does not consider the full range of disturbances that are likely to impact the plant. This thesis also investigates the additional value of considering the disturbances during design and operation of geothermal ORCs. In the literature ORCs are modelled mechanistically and this approach is also used in this thesis. Models consist of unit operations connected by process streams. Typically these are lumped parameter models but some distributed parameter modelling is observed in heat exchanger models. The model equations consist of thermodynamic state, mass and energy balance, heat transfer, and adiabatic compression and expansion calculations which describe the physical processes in the plant. Steady-state models have been constructed for geothermal and waste heat recovery ORCs and dynamic models have been constructed for waste heat recovery ORCs but there is a gap in the literature in dynamic models of large scale geothermal ORCs like the one examined in this thesis. To address this gap a dynamic model of a commercial geothermal ORC plant was built using the process simulator VMGSim, and validated using twenty-four hours of plant data. This validation showed that the plant data agreed reasonably well with the model output. A novel outcome of this model was that the results indicated that the dynamics of the working fluid cycle are fast compared to plant disturbances. The existing PI controllers are able to provide control that is adequate and in general advanced control techniques cannot provide additional benefits commensurate with their cost and complexity. The control of small scale ORCs including the application of advanced control techniques such as model predictive control was examined in the literature. The literature on ORC control focusses mainly on highly variable heat resources such as those seen in waste heat recovery applications. There is a gap in the literature in the control of large scale ORCs in geothermal applications, although some research has been done on the control of hybrid systems that combine geothermal with other heat resources such as solar. From the results of the dynamic model it is known that the dynamics of the system are fast enough that sophisticated control techniques are unlikely to have an impact on plant performance above simpler PI controllers. Instead, a specific area where advanced control could provide a benefit was examined. The impact of feed-forward control on using the geothermal wellhead valves to maintain pipeline pressure was examined using the dynamic model. A novel result of this study was a demonstration that feed-forward control can reduce the amount of geothermal fluid released to the atmosphere. This has sustainability benefits for the geothermal reservoir and also prevents emission of CO₂ and other pollutants present in the geothermal fluid to the atmosphere. This study also found the impact of the feed-forward controller on net power was minimal. There is substantial literature on optimisation of a wide variety of ORCs and heat source types including geothermal ORCs. Optimisation research in this area has examined ORCs using multiple objective functions including net power, efficiency, exergy, and thermoeconomic functions that measure both thermodynamic and economic value. One area that the literature does not consider explicitly is the consideration of disturbances when performing ORC plant optimisation. This thesis seeks to address this gap and does so in three ways. The first is in investigation of sizing of the air-cooled condenser in the modelled geothermal ORC plant. The size of this heat exchanger was examined with respect to the range of air temperatures that were recorded over a period of one year at the site. The original sizing of the condenser was done by assuming the average air temperature as the design point. A new condenser sizing was found by applying the heuristic that the sizing of the condenser should be based on the 95th percentile of the recorded air temperatures. An economic analysis was then performed that considered how the net power of the plant would be changed across the entire air temperature range. This concluded that an increase in the air condenser to the new size would have a payback period of only a couple of years, which indicates it may be profitable. The second way the consideration of disturbances when optimising geothermal ORCs is addressed by this thesis is through building and validating a steady state model of a commercial geothermal ORC in VMGSim and MATLAB. This model includes the geothermal gathering system which is essential to properly understanding the behaviour of geothermal ORCs. MATLAB was used to converge the model more quickly and coordinate the solution of large datasets. This was used for model validation and to optimise geothermal flow rate and turbine choked area for the plant, which identified an improvement in net power by adjusting the geothermal flow rate and turbine choked area. The behaviour of the plant for a range of turbine choked areas and geothermal flow rates and for different fouling conditions in the heat exchangers was also analysed which allowed the nature of heat transfer between the geothermal gathering system and ORC to be determined. This revealed a link between the pressure-flow dynamic of the geothermal gathering system and the pressure-flow dynamic of the ORC which will be useful to plant designers in the future. The third and final way disturbances were accounted for in the optimisation of ORCs was by applying self-optimising control to the plant steady state model. This demonstrated that this method can show an improvement in net power when the plant is subject to disturbances. Using a MATLAB program, controlled variables were selected that optimise the plant when they are held at a constant set point over a range of disturbance scenarios. This method is a slightly modified version of the existing self-optimising control method that greatly increases the speed of the analysis without impacting the improvement in net power output of the plant. An approximate self-optimising control method was developed and applied to the plant steady state model to show that it can provide an improvement in net power when the plant is subject to disturbances. Using a MATLAB program, a controlled variable—which is a linear combination of plant measurements—is selected that optimises the plant when it is held at a constant set point over a range of disturbance scenarios. This method is a modified version of the existing self-optimising control method that greatly increases the speed of the analysis without significantly impacting its accuracy. It can also be used with a model created in process simulation software, which allows it to be implemented more easily. It was found that the controlled variable that was selected caused the plant to operate at its optimum point across the range of expected disturbances. From the work presented in this thesis it is demonstrated that advanced process control will not bring significant benefits to large scale geothermal ORCs, but in certain niche applications it can provide a benefit. It is also shown that by considering plant disturbances improvements can be made to plant design and operation that generate greater value from geothermal ORCs.