Speed-up algorithm based on parallel projection correlation technique for planar PIV: Accuracy and limitation
Autor: | N. S. Dushin, Nikolay I. Mikheev, Dinar Zaripov, Renfu Li |
---|---|
Rok vydání: | 2018 |
Předmět: |
Turbulence
Parallel projection Computer science Computation 01 natural sciences 010305 fluids & plasmas Computer Science Applications Physics::Fluid Dynamics 010309 optics Boundary layer Multigrid method Particle image velocimetry Modeling and Simulation 0103 physical sciences Potential flow Electrical and Electronic Engineering Shear flow Instrumentation Algorithm |
Zdroj: | Flow Measurement and Instrumentation. 60:88-94 |
ISSN: | 0955-5986 |
DOI: | 10.1016/j.flowmeasinst.2018.02.019 |
Popis: | Implementation of advanced experimental techniques such as High-speed Time-resolved planar Particle Image Velocimetry is associated with big data processing and implies time-consuming operations. The Parallel Projection Correlation (PPC) technique is proposed in this paper for computation speed-up. Despite the fact that this technique reduces the calculation time, it has been revealed that its accuracy strongly depends on the interrogation window size. To show the limitations of PPC, three synthetic test cases are considered in this investigation: one-dimensional uniform flow, linear shear flow and turbulent boundary layer flow. Accuracy analysis is carried out for different interrogation window sizes. PPC is applied together with the iterative multigrid window-deformation technique (WIDIM). It is demonstrated that the maximum size of interrogation window that can be used for the shear flow analysis is 16 × 16 pix. The temporal filter applied at the initial and intermediate iterations of WIDIM is proposed to overcome this limitation enabling to use 64 × 64 pix interrogation windows for high-gradient shear flow. Finally, the results of turbulent boundary layer problem obtained by PPC are discussed in terms of accuracy of the streamwise velocity and turbulence intensities. |
Databáze: | OpenAIRE |
Externí odkaz: |