Gaurab Samanta1, Antony N. Beris1, Kostas D. Housiadas2, and Robert A. Handler3. (1) Chemical Engineering, University of Delaware, 150 Academy street, Newark, DE 19716, (2) Mathematics, University of the Aegean, Karlovasi, Samos, 83200, Greece, (3) Naval Research Laboratory, 4555 Overlook Ave. SW, Washington, DC 20375
Previously, Direct Numerical Simulations (DNS) of viscoelastic turbulent channel flows have been successfully used to demonstrate drag reduction [1]. This led us to explore ways to extract dynamic information of large scale structures (coherent structures) hidden into petabytes of data generated by typical DNS runs as these structures help to reveal turbulence intermittency. In this regard, we made use of the Karhunen-Loeve (K-L) decomposition technique, applied only to the velocity field, which allowed us a systematic way to truncate velocity information, and carried out a comparative study of Newtonian and viscoelastic turbulent channel flows, investigating time series data of high energy K-L modes via correlation (auto and cross) calculations [2]. However, the polymer conformation has a major role to play in the drag reduction mechanism [1] and it is essential to investigate the effects of velocity data reduction on conformation statistics. So, we calculate here the viscoelastic conformation field developed in response to the velocity field reconstructed from selected KL representations and compare the results against the full DNS predictions. We also independently truncate the velocity information represented in a Fourier-Chebyshev series in a DNS and study its response on the polymer conformation. The relevance of these results in building low dimensional models for viscoelastic turbulence will be addressed.
1 Housiadas and Beris, J. Non-Newtonian Fluid Mech., Vol. 122, 243-262 (2004).
2 Samanta, Oxberry, Beris, Housiadas, and Handler, J. Turbulence, (Under review).