We propose an integrated experimental and numerical investigation of the dynamics of inertial particles in isotropic turbulence. The experimental work will be performed in a new soccer ball turbulence facility that will be built as part of the project. The new facility will be capable of producing isotropic turbulence with a Reynolds number (based on the Taylor microscale) of 500. Into this flow, we will introduce metal coated hollow glass spheres and image those spheres using an advanced version of digital holographic particle image velocimetry (DHPIV). Using a unique optical setup, we will image the particles in single and double exposure modes to obtain position and velocity statistics. Additionally, we will perform direct numerical simulations (DNS) that will be used to: (i) advance the DHPIV technique; and (ii) complement the experimental measurements.
Intellectual Merit:
The measurements we propose to make in the lab and in silico will allow us to quantify two important aerosol processes: (i) two-particle dispersion of inertial particles; and (ii) inertial particle collision rates, both for the first time. DHPIV will capture position and relative velocity statistics, and through kinematic relationships these data will be used to quantify the dispersion rate and the collision kernel as a function of the particle parameters (Stokes numbers) and Reynolds number. The new facility, combined with a judicious choice of particles, will allow us to isolate the effects of each parameter. A crucial aspect of the velocity measurement is the accurate pairing of particles in the two images. Current algorithms do not work well for inertial particles that don't necessarily follow the flow or remain highly correlated. With the aid of DNS, we will develop a new matching algorithm based on sweeping the time lapse between images. The DNS too will be advanced under this study. Our current algorithm is capable of performing 10243 simulations on our 32-node cluster. However, in order to match the conditions of the proposed experiments, we must increase the resolution. We will modify the data structure of our code so as to take advantage of recent developments in the 3D fast Fourier transform. The new code will be able to run on 100's and even 1000's of processors on the Texas Advanced Computing Center, enabling 20483 simulations and Reynolds numbers of 500. In this way, we will continue our tradition of making quantitative comparisons with the experiments. Additionally, DNS yields more information than the experiments about the flow field, as well as allows us to study Lagrangian statistics. We will perform these studies to test assumptions we have made in the analysis of the experiments, as well as to advance our theoretical understanding of particle dispersion and collision.
Broader Impacts:
The motion of discrete particles in a turbulent fluid is of great significance to a broad range of engineering flows as well as natural flows. From understanding the competition between growth and oxidation of soot particles in a diesel engine, to quantifying the impact these particles have on the global climate, we are challenged to describe the dispersive and collisional properties of particles in order to get these predictions right. Historically our understanding of turbulence has gone hand-in-hand with our ability to measure the key variables, either experimentally or computationally. The goal of this proposal is to measure the statistical quantities that will allow us to quantify these two important aerosol processes. These results will stimulate an exciting new theoretical understanding, both within our group and elsewhere. Our approach is unconventional in that we intermingle DNS and experiment completely. Indeed, a strength of this work has been our ability to make quantitative comparisons between DNS and experiments. We meet weekly via videoconference to thoroughly discuss all aspects of the work. This provides a rich environment for students, who are exposed, at a high level, to all of the activities. The PIs have been heavily involved with outreach throughout their careers. Recent activities include recruitment and mentoring of women and underrepresented minorities at their respective institutions, outreach within the community, and organization of a series of high profile workshops at the NSF directed towards encouraging underrepresented minorities into the academy.
This project involves a collaboration between Cornell University and SUNY-Buffalo to investigate the dynamics of inertial particles in homogeneous turbulence. The project entails both experimental and numerical studies of particle-laden turbulence, where the numerical work is being performed at Cornell and the experimental work is being carried out at SUNY- Buffalo. The groups meet periodically by videoconference to coordinate activities and share results. The experiments are being performed in a new turbulence flow chamber that was designed and built as a part of this project. Particle measurements are being made using digital holographic particle image velocimetry (DHPIV) and more traditional particle image velocimetry (PIV). The simulations are being carried out on in-house computers and on national facilities. Intellectual Merit. The goal of this project is to bring together, for the first time, experimental measurements and numerical simulation of relative velocity statistics for inertial particles suspended in a turbulent flow environment. Relative velocity statistics, in combination with earlier measurements of the particle radial distribution function, a statistical measure of particle pair separations, would make it possible to compute the average particle collision rate. To measure the particle velocity, the experiments will use two exposures of DHPIV taken in rapid succession; however, an earlier study (de Jong et al. 2010) demonstrated that the errors in this measurement are unacceptably large. We are using the direct numerical simulations (DNS) to assist in the development of improved image processing to reduce those errors. Because of the much higher Reynolds number in the new chamber, we originally proposed to upgrade the DNS code to allow for simulations with up to 2048^3 grid points. While this work was cut from the original budget, we were still able to achieve this goal through the assistance of a graduate student supported by the NSF Graduate Research Fellowship Program. The DNS was used to develop a new strategy for measuring the particle relative velocities. Due to unavoidable delays in the experimental program, we have not yet completed the comparisons between the experiments and the DNS. In response, we modified the goals of the Cornell project to include: (i) advancing theory; and (ii) studying inertial particles in homogeneous turbulent shear flow. The funding provided by this NSF grant was augmented by other internal funds to allow this expansion in scope. Figure 1 shows an image from the 2048^3 simulation made with the upgraded DNS code on NCAR's Yellowstone computer. The yellow lines show the spontaneously formed vortex tubes and the white dots show the locations of the inertial particles. Clear evidence of particle clustering is apparent, and this is known to increase the particle collision rate. These DNS are the most highly resoved simulations ever performed in the US. This research has led to 5 publications in leading peer-reviewed journals and 3 papers currently in review (with several others planned). Additionally, 10 conference presentations were made. The grant partially supported two PhD students, one of whom has completed his degree and one who will finish in a year. The grant partially supported a postdoctoral research fellow who has advanced a theory for inertial particles; the remainder of his support came from internal Cornell funds. Broader Impacts. The motion of discrete particles in a turbulent fluid is of great significance to a broad range of engineering flows as well as natural flows. From understanding the competition between growth and oxidation of soot particles in a diesel engine, to quantifying the impact these particles have on the global climate, we are challenged to describe the dispersive and collisional properties of particles in order to get these predictions right. The measurements in this study provide key data for understanding the dispersive and collisional properties of inertial aerosol particles over a wide range of conditions. This has an immediate impact on our ability to model multiphase turbulence in these applications. Moreover, the results are stimulating new theories, within our group and elsewhere. These will lead to an improved understanding of immense societal problems such as global climate change. deJong, J., Salazar, J.P.L.C., Woodward, S.H., Collins, L.R. and H. Meng, Measurement of inertial particle clustering and relative velocity statistics in isotropic turbulence using holographic imaging, Int. J. Multiphase Flow 36:324-332, 2010.
