Solutions

Energy performance and EnRC

Expected consumption instead of a flat baseline: EnRC models, actual-versus-predicted tables, and the indices operators and reviewers share.

Request a quote Open the glossary

ISO 50001:2018
TS EN ISO 50001:2018

A flat last-year baseline cannot explain a hotter summer or a different production mix. ISO 50001 energy performance needs expected consumption given the relevant variables — then a fair comparison with what the meters actually recorded.

mf-python models turn those variables into energy reference consumption (EnRC). Actual-versus-predicted tables, WMAPE bands, and SEU narratives feed the company energy report. CUSUM, efficiency and intensity indices, and specific energy consumption sit next to the model so operators and reviewers share one picture of performance.

What you get

EnRC, not a frozen baseline: Relevant variables produce expected consumption for the period you are reviewing, so a change in throughput is not mistaken for a saving.
Actual versus predicted in the open: Tables and WMAPE bands show whether the model is fit to use in the report, not only in a data-science notebook.
CUSUM and intensity indices: Cumulative sum, efficiency and intensity indices, and specific energy consumption sit beside the model for the same SEU.
Narrative that matches the SEU list: Each modelled SEU can carry the expected-versus-actual story into the company energy report without a second export.

Frequently asked

Which models does AEYS use?: Training runs through mf-python (OLS and related estimators). The EnMS cares that expected consumption is documented and comparable, not that every plant uses the same algorithm.
Can we start without a full year of tagged data?: Yes. File uploads can train the first models. Canonical mappings take over when the same columns arrive every month.

What you get

EnRC, not a frozen baseline

Relevant variables produce expected consumption for the period you are reviewing, so a change in throughput is not mistaken for a saving.

Actual versus predicted in the open

Tables and WMAPE bands show whether the model is fit to use in the report, not only in a data-science notebook.

CUSUM and intensity indices

Cumulative sum, efficiency and intensity indices, and specific energy consumption sit beside the model for the same SEU.

Narrative that matches the SEU list

Each modelled SEU can carry the expected-versus-actual story into the company energy report without a second export.

Frequently asked

Which models does AEYS use?

Training runs through mf-python (OLS and related estimators). The EnMS cares that expected consumption is documented and comparable, not that every plant uses the same algorithm.

Can we start without a full year of tagged data?

Yes. File uploads can train the first models. Canonical mappings take over when the same columns arrive every month.