Novel approach for tomographic reconstruction of gas concentration distributions in air: Use of smooth basis functions and simulated annealing.
Drescher-AC; Gadgil-AJ; Price-PN; Nazaroff-WW
Atmos Environ 1996 Mar; 30(6):929-940
Techniques for measuring the gas concentration distribution imaged via computed tomography (CT), algebraic reconstruction technique (ART) and smooth basis function minimization (SBFM), were examined. A mixture of 10% sulfur-hexafluoride (SF6) as a tracer gas in helium was released within a wooden chamber. SF6 concentration distributions were measured over a 1 hour period at numerous chamber sites. Both point sample measurements and path integral measurements were obtained. Minimal ray variance was determined when the ART was applied to the ray integral data generated from the point sample measurements. However, the ART did not accurately reconstruct peak magnitudes. Secondary artifacts not evident in the original kriged maps were observed in the ART reconstructions. An accurate, preliminary guess was required in order for the ART to reconstruct the true measured values. This was attributed to the fact that the ART was a highly underdetermined system, with 195 unknowns and 56 equations. The SBFM approach, which eliminated indeterminacy, was then developed. In this method, the data were modeled with two Gaussian basis functions and a functional form was imposed on the concentration profile, constraining the solutions to a more realistic domain. Results indicated that the SBFM method was more accurate in the reconstruction of concentration profiles than the ART. Although the ray variance was higher with SBFM than with the ART, the variance was still considered acceptable. Maps derived from the ray integral data were more consistent with the measured point sample data when the SBFM method was employed. The authors conclude that the use of computed tomography coupled with SBFM is a promising method for the efficient and reliable measurement of gas concentration distribution.
AENVEQ; NIOSH-Publication; NIOSH-Grant; Grants-other; Air-samples; Measurement-equipment; Mathematical-models; Simulation-methods
Environmental & Indust Health University of Michigan 1420 Washington Heights Ann Arbor, MI 48109-2029
Other Occupational Concerns; Grants-other
University of Michigan at Ann Arbor, Ann Arbor, Michigan