CASE02 · Manufacturing · Heat treatment
Normalizing high-consumption ACF batches with equipment-specific baselines
Compare E5 RETAINING PLATE batches at the same weight and use separate weighted least squares baselines for ACF#3 and two tempering furnaces. The initial scenario targets excessive consumption without treating the estimate as verified savings.
01 · CASE ANALYSIS
Operating challenge
At a constant production weight of 134 kg, ACF#3 electricity consumption ranged from approximately 83 to 360 kWh. Processing duration ranged from 118 to 340 minutes. Waiting, quality rework and heating conditions therefore needed to be examined alongside production weight.
Tempering furnaces #1 and #2 processed multiple products. Their baselines needed processing duration and production quantity as well as weight.
02 · CASE ANALYSIS
Data and baseline models
Daily MES production records were aligned with equipment-level EMS electricity data. Noise and influential observations were excluded before constructing weighted least squares (WLS) baselines.
In the model notation, Wday represents daily production weight, Tday processing duration and Qday production quantity. The low adjusted R² for furnace #2 calls for closer review of exceptional days and operating conditions.
| Equipment | Baseline model | Reported performance |
|---|---|---|
| ACF#3 | Ebase_day = 1.5287 × Wday − 21.5208 | Adjusted R² 0.9481; RMSE 19.680 kWh; CV(RMSE) 9.12% |
| Tempering #1 | Ebase_day = 7.1244 + 0.0686Tday + 0.0281Wday | Adjusted R² 0.8482; CV(RMSE) 18.11% |
| Tempering #2 | Ebase_day = 24.5317 + 0.1194Tday + 0.0009Qday + 0.0046Wday | Adjusted R² 0.3219; CV(RMSE) 22.80% |
03 · CASE ANALYSIS
Proposed operating change
For E5 batches at 134 kg, the lower-quartile normal consumption level was selected as a practical first target. The scenario compares 183.3 kWh with 175 kWh and reports 4.5% initial potential.
04 · CASE ANALYSIS
Measurement and verification
Keep the E5 product, batch weight, target temperature and heating profile comparable. Record waiting and quality rework so that their effects are not confused with a change in furnace efficiency.
The scenario remains an estimate until field application and M&V. Tempering furnace #2 requires stronger exception management because its baseline errors are relatively large.
| Comparable condition | Baseline | Target | Potential |
|---|---|---|---|
| E5; 134 kg | 183.3 kWh | 175 kWh | 4.5% (reported) |
05 · CASE ANALYSIS
Engineering lesson
A single production-weight model is not necessarily sufficient for every furnace. Choose variables for each equipment boundary and evaluate residuals before turning a modeled opportunity into an operating target.
APPLY TO YOUR SITE
Connect your operating data to an actionable energy strategy.
Discuss the measurement boundary, data quality, operating constraints and verification approach for your facility.
Discuss your project All English cases