Fluctuations in dilution rate may result from drift in pumping rates and changes in the bioreactor working volume (D = flow rate/volume). In bench scale bioprocessing, peristaltic pumps are susceptible to short and long term changes in flow rates. The working volume is normally assumed to be constant and is rarely measured on-line, but it can fluctuate significantly with changes in aeration, surface tension, and biomass concentration. Therefore, the desired steady state in continuous culture is difficult to achieve with a chemostat as the dilution rate approaches the maximum specific growth rate, and this is often the region of highest productivity.
Auxostats eliminate the operational problems found at high growth rates. Stability arises in the bioreactor through manipulation of the pumping rate of fresh medium using feedback control strategies. The dilution rate becomes a dependent variable that adjusts to the rates of supply and demand for the controlled variable. Random short-term disturbances in the pumping rate and bioreactor volume are handled by the control algorithm and long-term changes can be accounted for by volume measurements and intermittent pump calibration. This permits stable operation in the `high gain' areas near the maximum growth rate, but only when robust and accurate measurement and control for the state variables can be devised. Fig. 2 is a sketch of an auxostat.
Auxostats have an inherent advantage because they can reach steady state more rapidly than can chemostats. Chemostats have a very poor response time as the dilution rate approaches the maximum growth rate, where it increases exponentially. Faster response times are a basic benefit of feedback control (Seborg et al., 1989). Disturbances and changes in setpoints are very common in bench scale bioprocessing (planned and unplanned), and feedback control permits quicker attainment of a new steady state. This allows auxostats to move quickly through different operational regimes. The choice of the feedback variable for an auxostat is quite broad, however it must relate to microbial growth (Fredrickson et al., 1970). Some of the earliest work employed turbidostats where the biomass concentration, as reflected by turbidity, was controlled through the manipulation of dilution rate (Bryson, 1952). This technique is normally based on measuring optical density, however it is unreliable because of fouling of the sensing element (Watson, 1969). It is still used as research tool for non-fouling types of cultures (Aarinio et al., 1991).