| CFD simulations have been widely used in aviation, aerospace, meteorology, shipping, water conservancy, chemical industry, construction, machinery, marine, environment, etc.. However, the CFD softwares are mostly validated based on the deterministic geometrical models and boundary conditions. However, there are many uncertainties in real engineering problems, for instance, the uncertainty of boundary conditions, geometrical tolerances, the uncertainty of gas physical property and so on. These uncertainties would bring potential risks to the design which is based on the deterministic simulations and lead to the failure in industry design. Therefore, it is necessary to take those uncertainties into account during the simulations.Non-deterministic analysis for CFD mainly focuses on the influenceof uncertain model input or model parameter onoutput parameters, namely propagation of uncertainty inflow field. And the distribution function of the randomvariable is supposed to be known. The Intrusive Polynomial Chaos (PC) method and Non-intrusive Probabilistic Collocation (NIPRC) method are mainly used in the paper. Firstly, the two methods are validated by case the stochastic cavity flow with several uncertainties. The results show the accuracy and efficiency of multi-dimensional Polynomial Chaos method on the simulation of propagation of uncertainties in the flow field when the number of uncertainties is not large, which compared with the results of Monte Carlo (MC) method. However, with the number of uncertainties increasing, computational time cost is increasing rapidly. NIPRC method need not change the deterministic solver, which applies to complicated model. The NIPRC, PC and MC are compared in the paper. The results show that the NIPRC method has higer convergence speed and convergence precision.Non-deterministic CFD simulations were performed on the airfoil NREL S809with uncertain angles of attack (AOA) using the NIPRC method in chapter4, which is supposed to be in Guassian distribution. The simulation results show that the influence of the AOA uncertainty mainly locates at the leading edge of the airfoil for a low AOA. That influence for a high AOA expands to the middle and the trailing edge of the airfoil.In chapter5the paper investigates the uncertainty of incoming wind speed of an NREL Phase VI wind turbine. The wind speed imposed on the inlet boundary is supposed to be an uncertain parameter. The flow simulations are performed at four mean wind speeds. The results show that the input uncertainty of inlet velocity results in a high uncertainty zone in pressure distribution near the root, and decreases from root to tip. At a low wind speed, the maximum pressure values are along the surface and appear near the leading edge in suction side. With the increases of the wind speed, the maximum values of the pressure are away from the blade surface, which leads to insensitivity of the load and power. The separate vortex regions are more sensitive to the input uncertainty of inlet velocity. Then the mistuning problem caused by the error of setting angle has been investigated based on the2.5MW wind turbine DF90. The setting angle of one wind turbine blade has been set as a random variable. The NIPRC method is employed to analyze the flow variables resulting from the mistuning problem. The simulation results show that the mistuned blade has influence on the upstream and downstream blades, and the influence on downstream blade is more obvious. Moreover, the blade load variation caused by the positive and negative deviation of pitch angle caused is asymmetric. The load change of the mistuned blade varies oppositely to the upstream and downstream blade load variations.In chapter6, NIPRC method is used to simulate the effects of tip clearance uncertainty on the aerodynamic performance of NASA Rotor37. The tip clearance height is supposed to be a stochastic variable in Beta distribution. Four different typical uncertain height of the plain tip clearance are investigated, respectively, such as, the height at the leading edge, at the trailing edge and at both the leading and trailing edges. The statistical properties of the total performance of the rotor and flow structure within the blade passage under different working conditions are analyzed. The results show that: with an uncertain plain tip clearance, the variation of total performances is decreased with the reducing of the outlet pressure. The variation extremisms of the Mach number appear in the regions where the leakage flow, the shock wave and the separated flow locate. When the tip clearance is linearly distributed from the leading edge to the trailing edge, the uncertain tip clearance at leading edge has strongest influence on the aerodynamic performance under high mass flow rate conditions. With the mass flow approaching to the choking point, the variation caused by the tip clearance at trailing edge drops down sharply. When the mass flow is lower than that at the near-peak efficiency point, the Std. of the total aerodynamic performances caused by the uncertainties at both the leading and trailing edge heights are larger than that at single... |
PDF Full Text Request