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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

3. Comparison of black-box modeling approaches using<br />

laboratory data<br />

3.1 Laboratory experimental data<br />

Breuker (1997) gathered data under controlled conditions in a laboratory on a test unit<br />

at a number of controlled operating conditions <strong>and</strong> used the data to test the ability of<br />

several model types to produce accurate estimates of unit per<strong>for</strong>mance. This test unit<br />

is a three ton packaged rooftop unit (Carrier Model 48DJE004610) which has a<br />

constant speed <strong>and</strong> hermetically-sealed reciprocating compressor (Copel<strong>and</strong> Model<br />

CRH3-0275-TFD) <strong>and</strong> uses fixed-orifice type expansion devices <strong>for</strong> refrigerant flow<br />

control. The grid of indoor conditions used in the testing is shown in figure 3.1. The<br />

larger set of data, labeled "Training Data" in the figure, was gathered at indoor dry<br />

bulb temperatures of 70, 73, 76, 79, <strong>and</strong> 82 F, indoor wet bulb temperatures of 55, 58,<br />

61, 64, <strong>and</strong> 67 F, <strong>and</strong> ambient temperatures (not shown in figure) of 60, 70, 80, 90,<br />

<strong>and</strong> 100 F. The smaller set of data, or "Test Data", was gathered at indoor dry bulb<br />

temperatures of 71.5, 74.5, 77.5, <strong>and</strong> 80.5 F, indoor wet bulb temperatures of 56.5,<br />

59.5, 62.5, <strong>and</strong> 65.5 F, <strong>and</strong> ambient temperatures of 65, 75, 85, <strong>and</strong> 95 F. As the<br />

figure shows, however, not all dry bulb conditions were simulated at all wet bulb<br />

conditions. A total of 40 distinct combinations were simulated in the "Test Data" <strong>and</strong><br />

94 combinations were simulated in the "Training Data" set. The conditions were<br />

selected because they completely cover the normal com<strong>for</strong>t region defined by<br />

ASHRAE (1993). The reason <strong>for</strong> gathering the data in two distinct sets was to allow<br />

<strong>for</strong> training <strong>and</strong> testing on separate sets of data. Training on the large set <strong>and</strong> testing<br />

on the small set of data tests the ability of the model to interpolate accurately.<br />

Training on the small set will test the ability of the model to extrapolate beyond the<br />

training data.<br />

18

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

Saved successfully!

Ooh no, something went wrong!