This paper presents the optimization of 3D valveless diaphragm micropump for medical applications.The pump comprises an inlet and outlet diffuser connected to the main chamber equipped with a periodically moving diaph...This paper presents the optimization of 3D valveless diaphragm micropump for medical applications.The pump comprises an inlet and outlet diffuser connected to the main chamber equipped with a periodically moving diaphragm that generates the unsteady flow within the device.The optimization,which is related exclusively to the diaphragm motion,aims at maximizing the net flowrate and minimizing the backflow at the outlet diffuser.All CFD analyses are performed using an in-house cut-cell method,based on the finite volume approach,on a many-processor system.To reduce the optimization turn-around time,two optimization methods,a gradient-free evolutionary algorithm enhanced by surrogate evaluation models and a gradient-based(GB)method are synergistically used.To support the GB optimization,the continuous adjoint method that computes the gradient of the objectives with respect to the design variables has been developed and programmed.Using the hybrid optimization method,the Pareto front of non-dominated solutions,in the two-objective space,is computed.Finally,a couple of optimal solutions selected from the computed Pareto front are re-evaluated by considering uncertainties in the operating conditions;these are quantified using the polynomial chaos expansion method.展开更多
文摘This paper presents the optimization of 3D valveless diaphragm micropump for medical applications.The pump comprises an inlet and outlet diffuser connected to the main chamber equipped with a periodically moving diaphragm that generates the unsteady flow within the device.The optimization,which is related exclusively to the diaphragm motion,aims at maximizing the net flowrate and minimizing the backflow at the outlet diffuser.All CFD analyses are performed using an in-house cut-cell method,based on the finite volume approach,on a many-processor system.To reduce the optimization turn-around time,two optimization methods,a gradient-free evolutionary algorithm enhanced by surrogate evaluation models and a gradient-based(GB)method are synergistically used.To support the GB optimization,the continuous adjoint method that computes the gradient of the objectives with respect to the design variables has been developed and programmed.Using the hybrid optimization method,the Pareto front of non-dominated solutions,in the two-objective space,is computed.Finally,a couple of optimal solutions selected from the computed Pareto front are re-evaluated by considering uncertainties in the operating conditions;these are quantified using the polynomial chaos expansion method.
