The calibration of hot-wires and determination of turbulence quantities using time averaged response equations are considered and studied both numerically and experimentally. The calibration study involves the development of a non-linear optimization technique, a comparison of the various heat transfer models, and a determination of the sample size required for accurate calibration. The conventional method of determining turbulence quantities based on a series expansion of the hot-wire response equation is compared with the alternate method based on squaring the response equation before time averaging. The non-linear technique of treating the raw calibration data is based on the iterative procedure of Gauss-Newton. The numerical experiments were carried out using the Monte Carlo simulation technique. This technique was used to simulate the hot-wire response and hence to determine the sample size required for accurate calibration. The effects of measurement error, the range of velocity, the optimization techniques used and the spacing of data points on the sample size required were studied. Using the same technique, pseudo turbulence data and the corresponding hot-wire responses were generated to serve as a standard, against which the two signal interpretation schemes were compared. This study included the estimation of truncation errors in the conventional method, the effect of varying the turbulence intensity and the effect of errors in measurement. The experimental programme involved performing velocity and yaw calibrations in the potential core of the jet and the measurement of turbulence in the fully developed turbulent pipe and jet flows. The results indicate that the non-linear calibration technique is systematic, accurate and easy to use. The yaw calibration obtained using the non-linear technique resulted in a negative value for the yaw factor (k(u272)). The extended power-law model yields the best compromise between low errors in velocity and low uncertainty in the estimated calibration constants. The calibration studies, using Monte Carlo technique, indicate that the sample size required is in the range of 20-30. Both the numerical and experimental results indicate that the conventional method of determining turbulence quantities with corrections for high turbulence intensity yields reliable results. By accurately determining the sensitivity of the hot-wire, the accuracy of turbulence measurements can be improved.Dept. of Mechanical, Automotive, and Materials Engineering. Paper copy at Leddy Library: Theses u26 Major Papers - Basement, West Bldg. / Call Number: Thesis1985 .S883. Source: Dissertation Abstracts International, Volume: 46-09, Section: B, page: 3189. Thesis (Ph.D.)--University of Windsor (Canada), 1985.
展开▼
