Fault Tolerance Assessment of HVAC-control in Office Buildings

In this PhD research, a methodology is developed and demonstrated, to evaluate the sensitivity of model based controllers against degradation faults.

A model based controller is a controller that implements knowledge about the building and installation structure and usage, into a model structure, model order and parameter values, based on as-built data, measurements or simulations).

Robustness is used with a double meaning: on one hand, the impact of degenerative faults on global KPIs (such as energy usage, discomfort, investment costs, maintenance costs, replacement  costs) and on the other hand, the control stability (number of switchovers, osccilations). Degradation faults are sensor faults and actuator faults in the BMS.

This evaluation utilizes cumulative KPI’s over a reference year, obtained through latin-hypercube selected computations with an emulator and control model. These KPIs, determined in fault-free and fault-present situations, are translated to economical scale and weigthed according to its chance of occurrence (FMEA-principle) and the preferences of the operator (estimated costs and benefits).

This methodology is demonstrated on three extreme office building cases, with either high or low thermal inertia, and slow or fast HVAC emission systems (CCC vs FCUs). This research aims for the comparison of the weighthed performance of these emulator models, when controlled with several (4) types of model based controllers, in contrast to the default control strategy. For every controller type and building case, the analysis and  comparison was made using a fault free, univariate fault present and a large subset of combinations of 7 degenerating pieces equipment, 3 fault types and 5 degradation types.

This study highlights the minimal requirements and the practical applicability of FDDe systems ánd model based controllers in office buildings. This methodology allows to identify and economically analyse the realistic potential of model based HVAC controllers in office buildings, taking into account realistic faults.