Modelling and control of gas flow in anesthesia

This paper describes the development of a model of an anesthetic circle system. The different components of the model are validated by a series of experiments. The results of these experiments show that the simulated system closely follows the actual system. The mathematical model is used for controlling the gas concentrations during inspiration, both within the time frame of an individual inhaled breath, and between breaths. A PI controller is used to get an indication of the problems to be expected in predictive control. The dominant problems are the discontinuities in the control signal and the stiction in the mass-flow controller valves. The discontinuities in the control signal arise because control is only applied during inhalation, which results in an unsatisfactory transient response, but the rise time can be improved by including a feedforward component in the controller. Stiction occurs because the valves are closed during expiration which results in delayed opening on subsequent inspiration. A range of methods were considered for overcoming stiction, and the most effective approach was found to be adding a strong pulse to the control signal at the start of inhalation