A Novel Mathematical Layout Optimisation Method and Design Framework for Modularisation in Industrial Process Plants and SMRs

Nuclear power has been proposed as a low carbon solution to electricity generation when intermittent wind and solar renewable energy are not generating. Nuclear can provide co-generation through district heating, desalination, hydrogen production or aid in the process of producing synfuels. However, current new large nuclear power plants are expensive, time consuming to build and plagued by delays and cost increases. An emerging trend in the construction industry is to manufacture parts off the critical path, off site in factories, through modular design to reduce schedules and direct costs. A study from shipbuilding estimates work done in a factory may be 8 times more efficient than performing the same work on site. This productivity increase could be a solution to the problems in nuclear power plant construction. It is an emerging area and the International Atomic Energy Agency records over 50 Small Modular Reactor designs in commercial development worldwide. Most Small Modular Reactor designs focus on integrating the Nuclear Steam Supply System into one module. The aim of this Applied Research Programme was to develop an efficient and effective analysis tool for modularisation in industrial plant systems. The objectives were to understand the state of the art in modular construction and automating design through a literature review. The literature review in this thesis highlighted that automating earlier parts of the plant design process (equipment databases, selection tools and modular Process and Instrumentation Diagrams) have been developed in modular industrial process plant research but 3D layout has not been studied. It was also found that layout optimisation for industrial process plants has not considered modularisation. It was therefore proposed to develop a novel mathematical layout optimisation method for modularisation of industrial plants. Furthermore, the integration within the plant design process would be improved by developing a method to integrate the output of the optimisation with the plant design software. A case study was developed to analyse how this new method would compare against the current design process at Rolls-Royce. A systems engineering approach was taken to develop the capabilities of the optimisation by decomposing the three required constituents of modularisation: development of a model to optimise layout of modules utilising the module designs from previous research (Lapp, 1989), development of a model to optimise the layout equipment within modules and development of a combined and integrated model to optimise assignment and layout of equipment to modules. The objective function was to reduce pipe length as it can constitute up to 20% of process plant costs (Peters, Timmerhaus, & West, 2003) and to reduce the number of modules utilised. The results from the mathematical model were compared against previous layout designs (Lapp, 1989), highlighting a 46-88.7% reduction in pipework and considering pipework costs can be up to 20% of a process plant cost, this could be a significant saving. This does not consider the significant schedule and productivity savings by moving this work offsite. The second model (Bi) analysed the layout of the Chemical Volume and Control System and Boron Thermal Regeneration System into one and two modules, reducing pipe cost and installation by 67.6% and 85% respectively compared to the previously designed systems from (Lapp, 1989). The third model (Bii) considered the allocation of equipment to multiple modules, reducing pipe cost and installation by 80.5% compared to the previously designed systems from (Lapp, 1989), creating new data and knowledge.
Mixed Integer Linear Programming formulations and soft constraints within the genetic algorithm function were utilised within MATLAB and Gurobi. Furthermore, by integrating the optimisation output with the plant design software to update the new locations of equipment and concept pipe routing, efficiency is vastly improved when the plant design engineer interprets the optimisation results. Not only can the mathematical layout optimisation analyse millions more possible layouts than an engineering designer, it can perform the function in a fraction of the time, saving time and costs. It at least gives the design engineer a suitable starting point which can be analysed and the optimisation model updated in an iterative process. This novel method was compared against the current design process at Rolls-Royce, it was found that an update to a module would take minutes with the novel optimisation and integration with the plant design software method, rather than days or weeks for the manual process. However, the disadvantage is that more upfront work is required to convert engineering knowledge into mathematical terms and relationships. The research is limited by the publicly available nuclear power plant data. Future work could include applying this novel method to wider industrial plant design to understand the broader impact. The mathematical optimisation model can be developed in the future to include constraints in other research such as assembly, operation and maintenance costs.