National Aeronautics and Space Administration Numerical Simulation of a fully turbulent flow in a pipe at low Reynolds numbers Gennaro Serino [email protected] Dr. Nagi Mansour [email protected] Background & Motivation Numerical simulations of a fully developed turbulent flow in a pipe at low Reynolds number have been carried out by using the open-source CFD solver SU2 developed at Stanford University. The aim is to validate the SST turbulence model implemented in the solver. DNS statistics data of fully-developed turbulent pipe flow by X. Wu and P. Moin, (JFM, 2008) have been considered for the validation. Data are available at the following link (http://www.stanford.edu/group/ctr/research_data/pipe/) and the reference publication is the following: "A direct numerical simulation study on the mean velocity characteristics in turbulent pipe flow" by Xiaohua Wu and Parviz Moin, Journal of Fluid Mechanics, Vol. 608 pp. 81-112, 2008. Results have been obtained at the Center for Turbulence Research at Stanford University (http://ctr.stanford.edu/). Further simulations have been carried out by including the Joule heating due to an imposed current along the streamwise direction of the pipe. A joule-heated fully turbulent flow has been thus simulated in order to understand the physics in presence of an heating source. This problem is intended to model the conditions experienced in the constricted Arc heater of the HAF facility at NASA Ames Research Center. • to edit turbulent Master textflow styled Fully Click developed in alevel pipe with the SST turbulence model in SU2 • Third level – Fourth level Geometrical Setup » The full-3D geometry has been considered for the cases and the mesh has been obtained with “Pointwise V17.1R2” which is capable of exporting into the native format of SU2 thanks to the proper plug-in. The number of points in the boundary layer and the wall spacing have been set in order to have a y+<1 on all the computational domain. Fifth level Geometry and mesh details (Pointwise V17.1R2) • Click to edit Master text styles Effect of Joule heating on turbulent flows in a pipe – Second level Joule heating • Third level – Fourth In an arc jet facility, a gas is heated and expanded to level » very high temperature and speeds. The heating is provided by a continuous electrical arc produced by two sets of electrodes . The aim is to simulate the surface temperature and pressure which are experienced by atmospheric reentry vehicles. The local heating W is a function of the imposed current I and of the conductivity σ which is a function of pressure and temperature. Fifth level Schematics of a constricted Arc heater arrangement Reference conditions for the simulations Comparison SU2-DNS data : non dimensional mean velocity profiles (Re=5300) Simulations have been carried out by considering the Joule heating in a fully turbulent pipe flow with SU2. The aim is to recreate the condition in the ARC heater facility thus a current of 1000 A has been imposed at the inlet. The Reynolds number has been set to 5300 respect to the pipe diameter in order to compare results with the previous simulations. As expected, by increasing the temperature, the flow becomes less turbulent and the velocity profiles look less smoother at the same streamwise coordinate until they superimpose the end of the pipe. Comparison SU2-DNS data : non dimensional mean friction velocity (Re=5300) x/L = 0.25 x/L = 1.00 x/L = 0.75 x/L = 0.50 SU2 simulations of turbulent flow in a pipe with and without Joule heating : comparison of velocity profiles at different streamwise location (Re=5300) Comparison SU2-DNS data : non dimensional mean velocity profiles (Re=24000) Comparison SU2-DNS data : non dimensional mean friction velocity (Re=24000) Conclusions Comparison SU2-DNS data : non dimensional mean velocity profiles (Re=44000) www.nasa.gov Comparison SU2-DNS data : non dimensional mean friction velocity (Re=44000) – Fourth level » Fifth level The SU2-SST turbulence model has been validated focusing on low-Reynolds cases. DNS data have been used to compare the results and sufficient agreement has been achieved. Only for the lower Reynolds number case (Re=5300), the solution does not match the DNS data in proximity of the wall. This could be addressed to a deficiency of the model or to the necessity of a finer grid. In the future, finer meshes will be used and the results will be compared to those hereby presented. Finally, the Joule heating effect has been included in order to simulate the conditions experienced by the flow in the ARC heater facility. Results have demonstrated the good coupling of the turbulence model (SST) and the Joule heating due to an imposed current with the SU2 code.