M1 Milford Plant

IPMVP Option B Readiness — Plant Summary
Coating Readiness Need More Data
• Baseline at 68% — need more collection time before coating.
OAT Bin Performance
Bin (°F) Readings Avg kW Avg EER
60–65° 438
0.2 2543.6
65–70° 195
0.5 1576.2
70–75° 52
0.4 1217.0
75–80° 140
0.3 1516.2
80–85° 118
0.3 2324.8
85–90° 0
90–95° 0
95–100° 0
M1 Milford Plant 68% Partial
OAT Bin Coverage iNeed ≥5 of the 8 five-degree bins (60–100°F) with compressor-on readings. Filled: 60–65°F, 65–70°F, 70–75°F, 75–80°F, 80–85°F. Empty: 85–90°F, 90–95°F, 95–100°F. 5/5 bins Passed. Still empty: 85–90°F, 90–95°F, 95–100°F.
Min Readings/Bin iStatistical power per temperature bin. The thinnest filled bin must have at least 10 readings for a defensible average. 52/10 readings
Total Readings iTotal compressor-on readings with valid OAT. Larger datasets produce tighter regression confidence intervals. Target: 500+. 7241/500 readings
Monitoring Window iCalendar days from first to last compressor-on reading. At least 14 days captures weekday/weekend cycles and weather variation. 6.1/14d Need 8 more day(s) of collection.
kW-vs-OAT R² iHow strongly outdoor temperature predicts electrical power draw (linear regression). R² ≥ 0.75 means OAT explains at least 75% of kW variance. 0.00/0.75 Weak OAT-kW correlation. Unit may be oversized or have variable internal loads.
Diagnostics
Data Completeness iOf all the readings the sensor should have sent (based on its reporting interval), how many actually arrived? Small gaps from WiFi hiccups are normal. Only a concern if large chunks of time are missing. 9.1% Significant data gaps (90.9% missing). Verify sensor is online and gateway has stable connection.
CV-RMSE iCoefficient of Variation of RMSE — measures how well the kW-vs-OAT regression fits the data. Lower is better: <15% excellent, <25% good, 25%+ may need more data or a non-linear model. 382.0% fair
Sensor Channels iAll mapped sensor channels should have at least one reading. Missing channels may indicate dead sensors or configuration errors. 5/5
Regression Model iBaseline regression equation: kW = slope × OAT + intercept. Required by IPMVP for 3rd-party verification of the baseline model. kW = 0.0019 × OAT + 0.21
EER Model iEER vs OAT regression. A positive slope is expected physics for high-OA units: higher outdoor temp → higher cooling load → more efficient compressor operation (closer to design conditions). Negative slope is typical for recirculation-dominant units. EER = 0.0302 × OAT + 15.81 (R²=0.006)
NMBE iNormalized Mean Bias Error — measures systematic over/under-prediction. IPMVP threshold: <±5% excellent, <±10% acceptable. 0.0% excellent
Durbin-Watson iTests whether regression residuals are independent over time. DW ≈ 2.0 = good. DW < 1.5 = positive autocorrelation (model misses a trend). DW > 2.5 = negative autocorrelation. 2.02 good
Load % Bin Performance
Load % Readings Avg kW/ton Avg EER
0–20% 97
4.449 213.4
20–40% 201
0.754 588.8
40–60% 267
0.418 1131.2
60–80% 508
0.319 1489.8
80–100% 1,047
0.24 1838.7
100%+ 5,121
0.196 2950.0
Device Status
Device Readings Last Seen Avg kW Avg Temp (C) Avg Flow (GPM) Quality
flow_meter 64,219 2026-04-21 17:37:30 -- -- 624.3 --
pqm_251510818 36,031 2026-05-06 06:05:44 18.3 -- -- --
pqm_251510832 35,861 2026-05-06 06:07:02 58.6 -- -- --
return_temp 2,926 2026-05-06 06:10:23 -- 6.4 -- --
supply_temp 2,632 2026-05-06 06:09:42 -- 0.9 -- --
Plant Efficiency (kW/ton) Over Time
|
Chiller Power (kW)
Flow Rate (GPM) & Delta-T (F)
Supply / Return Temp (C)
Cooling Output (Tons)