Data-driven control of discrete-phase concentration in a vortex involved two-phase flow with optimized energy consumption

• Main Author: Zhang, Xingyu
• Other Authors: Li Hua
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/144249
• Institution: Nanyang Technological University

Control of the discrete-phase (e.g. particles or bioaerosols) concentration in a two-phase flow has a wide range of attractive applications in engineering. However, it is difficult to achieve the objective concentration with optimized energy consumption, due to the complex flow of the fluidic phase and its effect on the discrete phase, especially in a single-inlet fluid field with fixed geometric structure. Moreover, the limits of computational ability constrain the design of control algorithm to achieve an optimized discrete-phase concertation in a two-phase flow. Therefore, this thesis aims to develop novel methods systematically to control the discrete-phase concentration in a two-phase flow, by modulating the flow pattern via the existing vortex in the fluid field, with optimized energy consumption for control actuation. The proposed methods are conducted to control the bioaerosol concentration at indoor corner, defined as the area around ceiling edges, in order to improve the indoor air quality that is very important to resident health. Three major academic achievements from the present research are summarized respectively as follows. The first achievement of the present research provides a feasibility analysis for control of bioaerosol concentration at an indoor corner, through airflow from ventilation outlet with optimized energy consumption. In the proposed control strategy, a gas-particle flow model with a chaotic moving vortex is developed to simulate the bioaerosol movement at an indoor corner via the Eulerian-Lagrangian scheme. In this model, the flowrate of the inlet fresh or air-conditioned air is tuned by a flowrate controller, which is installed at ventilation outlet or inside ventilation duct, in order to control the vortex trajectory. A residual concentration is defined as the ratio of the residual number of densely released particles at the corner to the area of the corner, in order to indicate the probability of bioaerosol movement, which is released sparsely and randomly by a microorganism in a real-world situation. The control strategies are then developed to optimize residual bioaerosol concentration and energy consumption by the orthogonal scheme. Based on the simulation results, a small variation of the airflow rate from the ventilation outlet contributes to reduction of the residual concentration of bioaerosols that are originally released from the corner, or to increase of the residual concentration of detergents that are released from the ventilation outlet. It is thus concluded that a Pulse Width Modulation controller, with low amplitude, proper duty factor, period and time delay, is more energy efficient to improve the local indoor air quality. The second achievement is the development of a machine-learning based method for control of local bioaerosol concentration, via the forced laminar corner flow with optimal energy efficiency at an indoor environment. A recirculation zone determined by the inlet flow rate traps particles partially with one or more vertexes around the corner. The profile of the recirculation zone is then determined mathematically, by the minimum net mass flux principle with a grid search technique. After that, the variation of the recirculation zone profile is learned via Neural Network (NN), in which dataset is collected from the simulation by the Eulerian-Lagrangian scheme. Moreover, a model predictive control (MPC) algorithm is implemented to achieve an optimal profile of the recirculation zone with optimal energy consumption, based on the linearized NN model. Finally, the proposed NN-MPC is employed to improve the local bioaerosol environment at indoor corner through a flow-rate-controllable air inlet located at ceiling. The third achievement is an extension to turbulent from laminar flow mentioned above, for modelling of the continuous phase dynamics in two-phase corner flow, in order to control the discrete-phase concentration with optimal energy efficiency. The extended turbulent two-phase corner flow model is carried out by computational fluid dynamic (CFD) simulations, in which the turbulent effect is simulated by Menter’s Shear Stress Transport (SST) - model, and the particle phase simulated by Eulerian-Lagrangian scheme. A great agreement is achieved against experimental data of a backward facing step, with an error of 12.2% for the maximum negative velocity in the recirculation zone. The simulation results are then used as input/output data for system identification by linear transfer function models, describing the relationship between inlet flowrate and particle concentration. A proportional-integral-derivative (PID) controller is tuned to minimize the residual particle concentration with optimized energy consumption for control. Simulation results demonstrate highly effective controllability of PID controller, reducing maximumly 24.16% of the particle concentration compared with natural ventilation (zero control). In addition, the efficient energy consumption is discussed among different sets of controller parameters, based on the requirements of the energy and residual particle concentration. However, there is still a space for improvement of the controller based on a better system identification algorithm, since it is difficult for the linear transfer function model to capture the nonlinear dynamics in the turbulent two-phase flow. Moreover, it is easy and fast for PID control to tune with effective improvement of particle removal, although other control methods probably achieve better control performance.