Dynamic Modelling of the Patient Circuit for High Frequency Ventilation

Sammanfattning

Artificial breathing is vital when it comes to treatment of critically ill patients where the natural breathing mechanism is insufficient. With the help of mechanical ventilators, the natural breathing mechanism of the patient can be assisted or even exchanged with the artificial breathing from the machine. Small errors and unexpected events in these systems may lead to serious damages on the patients, causing even more harm than good. Therefore, these systems require a lot of testing and monitoring to ensure functionality. With the use of accurate simulation models, testing time can be reduced by running test in the simulation environment instead of on the actual machine. The simulation models can also be used for monitoring functions in real time, making sure the ventilation of the patient is working as expected. When it comes to simulating a ventilator system controlled with high frequency ventilation techniques, the existing simulation models fail to reproduce the high frequency dynamics that appear during high frequency ventilation. This paper proposes a modelling approach for mechanical ventilator systems exposed to high frequency dynamics. Focus is placed on modeling the patient circuit including the inspiratory and expiratory tubes, the humidifier with the dry line tube, the Y-piece, the tracheal tube and the patient lungs. The model is based on mathematical models representing the thermodynamic and pneumatic behaviour of the system. It is built using Simulink with regular and customized building blocks from Simscape. Compared to pre-existing simulation models, this model includes the inertia effects of the gas which is crucial when it comes to accurately modeling the system while being exposed to fast changes in flow and pressure. To evaluate the model performance, the simulated pressure and flow at the patient port are compared to measured data from an experimental setup. From the results of this thesis it was seen that the model is very sensitive to the patient model used in the simulation environment, which means that the patient model has to be remodeled in order to archive a better model performance. Compared to the pre-existing model used for comparison in this thesis, it is seen that an increased parameter model produces more accurate results.