Effects of multiple simultaneous soft faults in variable-speed rooftop units: Experimental analysis and criticisms in detection and diagnosis

Soft faults in cooling systems, such as heat exchanger fouling or improper refrigerant charge, commonly arise from inadequate maintenance practices. To mitigate energy performance degradation, effective fault detection and diagnosis techniques must be implemented. However, diagnosing multiple simultaneous faults remains particularly challenging, especially in variable-speed systems, where control-induced interactions can significantly distort conventional fault signatures originally established for constant-speed operation. Based on the outcomes of previous experimental investigations on single faults, this study experimentally investigates the combined effects of evaporator fouling, condenser fouling, and refrigerant undercharge in a 17.5 kWc variable-speed rooftop unit. The analysis examines how individual fault signatures interact, dominate, compensate for one another, or deviate from linear superposition. The results indicate that sensible cooling capacity degradation is primarily driven by evaporator fouling, with reductions of 10.6% and 19.1% observed under evaporator fouling alone and under combined evaporator and condenser fouling conditions, respectively. In terms of thermodynamic variables capable of detecting the anomalies, the discharge pressure and the subcooling remain condenser fouling-dominated in the fouled unit with correct refrigerant charge; conversely, the undercharge induces nonlinear and competing effects, such as a mitigation of the discharge pressure increase (with residuals up to 2.4 bar). A symptom matrix is proposed to classify linear dominance, nonlinear contributions, and competing interactions under multiple-fault conditions, revealing that diagnostic rules derived under single-fault or fixed-speed assumptions cannot be directly extended to variable-speed systems operating under concurrent disturbances, where control-induced compensation reshapes fault observability.