One impetus for fed-batch operation came from developments in making penicillin. Early investigations found that lactose gave better penicillin yields than when using other sugars. Lactose obtained from companies that made evaporated milk was about 3 times the cost of glucose that is the usual sugar for microbial fermentations. When it was realized that the organisms were hydrolyzing lactose to its monomers, galactose and glucose, at a rate that about matched their use by the organisms, a logical experiment was to feed glucose instead. The result was a 3 to 4-fold increase in the yield of penicillin. When the feed rate was too slow, product formation was restricted. When the feed rate was too high, the organisms prefered to grow instead of making penicillin. The remarkable improvement in yield made it highly worthwhile to monitor glucose concentration and to establish the feed rate to hold it in the range for optimum yields. The premium price for lactose disappeared as companies switched to feeding of glucose.
Any multivariable system such as a fed-batch bioprocess poses problems for research and development. Many runs with permutations of variables are needed to unravel the interactions between timing of feed, concentration of nutrients, and the physiology of the culture. This is a situation that calls for computer simulation to aid in planning runs with the real process.