Detecting Cooling Coil Fouling Automatically—Part 1: A Novel Concept

Abstract

Fouling of surfaces within the heat exchangers of heating, ventilating, and air-conditioning (HVAC) systems of buildings is an equipment fault that wastes appreciable amounts of energy; however, it escapes detection under current building automation technology. A novel concept is introduced to automatically detect this fouling on water-side and air-side surfaces of water-cooled HVAC air coils. The concept incorporates a model-driven component contributing a table of human expert information and an embedded data-driven component assimilating real-time data sampled from HVAC plant instrumentation. Supervisory programming (the “agent”) conducts real-time surveillance for coil fouling using the data-driven component, which is a model replicating the current dynamic thermal behavior of the coil. The surveillance is a specific characterizing transient (a “query”) exercised periodically on the data-driven dynamic coil model as a surrogate for exciting the real coil. When a query returns a suspect result, the agent determines if coil fouling or some other change caused that result by using the tabulated expert information. The concept makes use of all data sampled from the plant, reflecting transient and steady behavior. Fouling can be discerned from other developments, such as instrument drift, and the agent can distinguish air-side fouling from water-side fouling, estimate the severity of fouling, and estimate an uncertainty for its classification. The values tabulated describe three-dimensional surfaces characterizing the varied impact fouling generically has on coil thermal effectiveness when considered over the state space of coil operation. Categorical use of that information by the concept is justified by analysis in the effectiveness-NTU state plane. A companion paper (Veronica 2011) gives results of exercising a crucial task within the concept on simulated data by using one form of a dynamic data-driven model; a multilayer perceptron.