28.10.2014 Views

Fault Detection and Diagnostics for Rooftop Air Conditioners

Fault Detection and Diagnostics for Rooftop Air Conditioners

Fault Detection and Diagnostics for Rooftop Air Conditioners

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

32<br />

has two possible causes: refrigerant is undercharged when service was done or there is a<br />

refrigerant leakage. There<strong>for</strong>e, low charge can be a system-level operational or service<br />

fault.<br />

Since the rooftop unit is under positive gage pressure system when charged, noncondensable<br />

gases can only be introduced during service. Non-condensable gases tend to<br />

accumulate in the condenser. Its primary impact is to increase heat transfer resistance <strong>and</strong><br />

results in high condensing pressures <strong>and</strong> temperatures. So, non-condensable gas is<br />

considered to be a component-level service fault.<br />

In summary, the characteristic of a component-level fault is that its source impact<br />

is confined to a specific location or component <strong>and</strong> all the other impacts on the system<br />

originate from this source impact. On the contrary, the source impact of a system-level<br />

fault cannot be confined to a specific location or component. Operational faults usually<br />

develop through running <strong>and</strong> occur r<strong>and</strong>omly or gradually, while service faults are<br />

introduced with service.<br />

1.2.2 Interactions<br />

As depicted in Figure 1-6, a rooftop unit can be represented as a black-box, which<br />

is driven by faults, disturbances <strong>and</strong> overall system driving conditions, including T<br />

aoc<br />

,<br />

T<br />

aie, <strong>and</strong> φ<br />

aie<br />

, <strong>and</strong> outputs overall system state variables. It is difficult to tell which<br />

factors contribute to the current operation state directly from overall state variables. The<br />

SRB method uses normal state models to predict the normal operation states according to<br />

the overall driving conditions <strong>and</strong> generates residuals to decouple the interactions<br />

between driving conditions <strong>and</strong> faults, <strong>and</strong> further uses statistical analysis to further<br />

decouple the actions from disturbances, but leaves the couplings among the different<br />

faults untouched. This is the reason why the SRB FDD methods cannot h<strong>and</strong>le multiplesimultaneous<br />

faults.<br />

32

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!