A Rule-Based Fault Detection Algorithm for a Purge System of a Ventricular Assist Device

A simple fault detection algorithm was designed to monitor the purge pump operation of a rotary type ventricular assist device (AB-180 Circulatory Support System [AB-180 CSS], CardiacAssist, Inc., Pittsburgh, PA, USA). This algorithm receives the fluid pressure of the purge system, calculates the pressure derivative, and then compares with the preset thresholds to detect potential abnormal conditions, such as leaking and kinking along the purge tubing, abnormal purge rate, and stoppage of the purge pump. The thresholds were identified from bench top experiments simulating the possible failure conditions in the purge system. The algorithm was then implemented in a data acquisition system and verified with different pumps and controllers. Because of its simplicity, the algorithm only takes an insignificant computation time to process the measured signal. Therefore, it can be incorporated into the existing controller without any major design change in the controller hardware. The same concept is also applicable as an alternative way to monitor an extremely low flow rate in other devices in which directly monitoring flow rate from the devices is important, but either too expensive or too difficult to implement.