Measurement points and breakdown
The main meter gives only a total. Meaningful improvement needs lines, machines and utilities measured separately, with measurement points chosen by energy intensity and by how much can actually be changed.
- Point selection weighs consumption share and actionability
- Utilities are measured separately, not allocated to production
- Sampling frequency follows process dynamics
- Meter accuracy is verified on a regular basis
Data collection and edge processing
Moving raw field data to a central system inflates link cost and latency. Collection, filtering and summarization happen at the edge; only what a decision requires travels onward.
- Edge summarization lowers link cost and latency
- Data buffers at the edge when connectivity drops
- Timestamps are applied at source from one clock
- Raw data remains retrievable when needed
Anomaly and loss detection
A consumption profile is judged against production volume. Rising consumption for the same output signals equipment out of adjustment, leakage or overdue maintenance. Anomaly is defined against expectation, not against an absolute value.
- Consumption is assessed relative to production volume
- Idle consumption is tracked as its own indicator
- Shift and weekend profiles are compared
- Anomalies are routed into maintenance work orders
Verifiable savings
The effect of an improvement is measured by bringing before and after onto the same footing. A comparison made without normalizing production volume, outdoor temperature and product mix will mislead.
- Comparisons use normalized data
- Before and after windows are of equal length
- Effect is reported as measurement, not estimate
- Reporting obligations draw on the same data