A controller for HVAC systems with fault detection capabilities based on simulation models

This paper describes a control scheme with fault detection capabilities suitable for application to HVAC systems. The scheme uses static simulation models of the system under control to generate feedforward control action, which acts to supplement a conventional PI(D) feedback loop. The feedforward action reduces the effect of plant non-linearity on control performance and provides more consistent disturbance rejection as operating conditions change. In addition to generating feedforward control action, the models act as a reference of correct operation. Faults that occur in the HVAC system under control cause the PI(D) controller to provide a greater than normal control action to compensate for fault induced inaccuracies in the feedforward models. The controller monitors the level of feedback compensation and generates alarms when thresholds are exceeded. The paper presents results from testing the controller with a dual-duct air-handling unit.