CALCULATION OF KINETIC-PARAMETERS FROM CHAMBER TESTS USING NONLINEAR-REGRESSION

Different functions of exponential type, which are relevant for the ma thematical modeling of test chamber kinetics are examined. The degrees of freedom, the time range and the noise level are systematically var ied. The possibilities and limitations for the recovery of kinetic par ameters are tested with nonlinear regression analysis by the use of ex perimental and synthetic data sets. An equation is presented, which al lows superimposing the generated data with accurate noise characterist ics. As expected, the parameter uncertainties being obtained from the inverted covariance matrix are found to be unreasonably large for a li mited number of data pairs and it is shown that error estimation using the confidence region at the 68.3% level gives more reliable results. It is demonstrated how the analysis of simulated data can yield valua ble information about the conditions being necessary to obtain reliabl e results from test chamber experiments.