T HE effective control of flow separation promises substantial performance improvements for a wide variety of air vehicles. Although the methods are well known, there is very little by way of theory or numerical models that can adequately predict lift enhancements, drag reduction, etc. An attemptwasmade to address this problem by conducting a computational fluid dynamics (CFD) validation workshop for synthetic jets and turbulent separation control [1], where one casewas dedicated to predicting the nominally two-dimensional flow over a hump. The baseline (uncontrolled) case was considered in addition to control by means of steady suction [2] and zero-net-mass flux (oscillatory) blowing [3]. The workshop determined that CFD with steady or unsteady Reynolds-averaged Navier–Stokes (RANS or URANS) consistently overpredicted the reattachment location, regardless of turbulence model or method. Within the separation bubble, most computations predicted velocity profiles well but considerably underpredicted the magnitude of turbulent shear stresses. Large-eddy simulations and other costly methods appear capable of overcoming this deficiency, but the focus of the current study is on the more affordable RANS and URANS methodologies. See, for example, Krishnan et al. [4], Morgan et al. [5], and Saric et al. [6]. Although these individual test cases were challenging to CFD codes, only a single test case was considered for both steady suction and zero-net-mass flux blowing. During the course of the experimental investigation, however, steady and unsteady surface pressures were acquired for a wide variety of control parameters, including Reynolds number, suction flow rate, and frequency and blowing amplitude in the zero-net-mass flux case. Significant variations, or changes, in control authority were observed depending upon the control input or prevailing Reynolds numbers. For example, varying the suction flow rate at different Reynolds numbers indicated an increase in the rate of drag reduction; in contrast, no Reynolds number effect could be detected with regard to zero-net-mass flux blowing. In addition, for the oscillatory case, the flow was seen to be highly dependent on control frequency and peak blowing amplitude. Different, sometimes counteracting, mechanisms dominated the separated flowfield during different parts of the control cycle. Comparisons of these changes, or trends, with numerical results will clearly facilitate more precise evaluation of CFD’s value for predictive purposes. This is because the uncontrolled, or baseline, state is usually known and hence the correct prediction of changes to the baseline state as a result of a particular control strategy would be of great value for assessing different concepts or downselection. To address this need, Rumsey and Greenblatt [7] undertook a detailed parametric study and this Note presents a summary of the main findings.
Read full abstract
Jul 20, 2013
J A Esfahani + 2
Open Access