TOWARDS ACCELERATED SIMULATIONS FOR FLUID FLOW AND HEAT TRANSFER OF LARGE INDUSTRIAL PROCESSES

Abstract: The manufacturing sector is one of the biggest energy consumers. The iron and steel markets in China are growing very fast. Several studies have been performed to evaluate the Chinese steel sector in terms of energy savings and CO2 emissions. The results of the studies showed that the major energy savings expected within 2020 and 2030 timeframe will be from industrial furnaces in steel mills. For the Swedish steel industry, it is important to be very efficient in order to remain competitive. The hot rolling process in the steel industry is a long process, where big slabs are heated in a furnace above the recrystallization temperature to roll the metal into a thin sheet and then the sheet is cooled at the Runout table using water. The amount of energy used during the process directly influences the price of the products. Moreover, the government policy on energy usage and CO2 emissions, the competitive market and the water scarcity, demand an optimal process operation to reduce energy consumption and greenhouse gas emission. Computer simulation is the best and most convenient way to approximate real-world processes; therefore, there is a need to have a real-time online simulation tool for process optimisation, decision support and diagnostics in different industries.Computational fluid dynamics (CFD) is a robust tool for simulating almost any kind of real-world process related to fluid flow, heat transfer and combustion. However, simulating real-world processes in real-time using CFD is very challenging due to the complexity involved in the physical phenomena studied. In this thesis, CFD simulations have been performed in small scale to understand the physics and perceive the complexity involved in the heating process of steel slabs and the cooling process of the steel sheets at hot rolling steel industries. The results from the simulations are successfully validated using experimental and theoretical results published in open literature. Past experience suggests using mesh-based commercial CFD solvers for simulating industrial processes, only if accurate and detail results are desired. However, the computational performance of these solvers shows limitations from a real-time perspective and indicates the need for alternative CFD methods and solvers. In the literature review performed as part of the first stage of this work, we have identified different alternative methods which can be used to perform CFD simulations in real-time or near real-time for large industrial processes. The thesis discusses the limitations of different types of CFD methods and points out the difficulties and challenges in utilising these methods for simulating large industrial processes. Our preliminary simulation work brings light towards the goal of multi-phase multi-physics real-time simulations.