Analysis of noise and bias in fermentation oxygen uptake rate data

The calculation of many derived fermentation variables such as the respiratory quotient (RQ) and mass transfer coefficient (K(L)a) requires the differences between the molar percentages of oxygen and carbon dioxide in the fermentor inlet and exit gas, called the %OUR and %CER. Noise and bias in %CER data is of order that in the exit gas carbon dioxide analysis. However, the relative amount of noise in the %OUR is one to two orders of magnitude greater than the noise in the raw oxygen analyses because the %OUR is calculated as a small difference between two large quantities. The noise in the %OUR is white with a Gaussian amplitude probability distribution of absolute standard deviation 0.0145. A chi-square filter of the %OUR data is shown to considerably improve the quality of the calculated RQ and K(L)a for a fermentation of Streptomyces clavuligerus. (c) 1992 John Wiley & Sons, Inc.