The same three steps for thirty years. The difference is that the assessment now starts from your data rather than from a walkdown and an opinion.
What are you trying to settle? Some questions the historian can answer and some it cannot, and finding that out costs one conversation rather than one month.
You export the data. We run it and produce a report per step, so every number traces back to the file you sent. The assessment ends with a ranked list of what is worth acting on, what was set aside and why, and the questions your operators need to answer.
Your engineers challenge the findings. This is the point of showing the working. A finding that dies here saves the cost of acting on it.
What to change, in what order, with the reasoning attached. Where a fix cannot be made at the loop or the unit, the proposal says so rather than quoting for work that will not help.
Specification, code and documentation, aligned with the standards your site already follows.
| For a continuous plant | A historian export of the tags in scope. Two to four weeks at one minute is comfortable; a tag list with descriptions and engineering units makes every report readable. |
|---|---|
| For a batch process | Twenty or more batches of the same product, run the same way, with a note of which ones were good. |
| For control loops | PV, setpoint and controller output for each loop, plus the loop list. Days rather than weeks is enough. |
The honest caveat about data. An assessment can only be as good as the window it learns normal from. If the export is too short for the number of tags, we will say so and tell you what to ask for, rather than produce a confident-looking report that cannot be trusted.
Once process data reaches the MES or the data lake, the question becomes what to do with it. These tools are one concrete answer, and they work on exported data without touching the control system.
Knowing which loops are genuinely poor, and which are only passing on someone else's problem, changes the scope of a project before it is quoted.
Comparing ways of running the same product, with the statistics to say whether the difference is real.
The reports are written to be read by engineers who were not in the room, which makes them usable as a work process rather than a one-off study.
Bring a question and a rough idea of what data exists. We will tell you whether it can be answered.