How 4 phases in 12 months took a plant from frequent stops to consecutive weeks of continuous operation.
A chlor-alkali plant operating at the edge of reactivity.
Unplanned stops were part of the routine. The team woke up to fight fires. Energy was disappearing into the compressors without anyone knowing exactly how much — or why. The legacy historian was accumulating data that nobody could turn into decisions.
Chlorum knew it needed to change. But it also knew that the wrong answer would be worse than the problem: buying an "AI-powered" system and installing it on top of a fragile foundation. Bad data in, bad decisions out — at higher speed.
The right question was not "what technology should we buy?". It was "how do we build a proactive operation, respecting the plant's pace?"
More than 70% of digital transformation projects don't finish.
Not for lack of technology. For lack of method. Many digital transformation projects are not conceived with the aim of impacting business results — they are conceived to show that the company "is transforming".
The pattern we observe repeatedly: the company tries to buy maturity quickly. It skips steps. It has no reliable data, but installs an analytics platform. It has no adequate instrumentation, but hires an AI project. The project stalls. The organisation loses confidence. The next initiative starts with scepticism.
Maturity is not gained by intensifying effort alone. Maturity requires time for reflection and learning — and a method that respects that.
Four phases. A 12-month journey. Each phase delivers value and sustains the next.
We started in December 2023. We finished in December 2024. It was not a straight line — no industrial journey is. But it was a deliberate sequence, respecting the pace of the operation and the team.
We replaced the legacy historian with a modern architecture. We connected PLCs, process instruments and energy meters via OPC UA and Meka Edge. For the first time, the right data reached the right place — in real time.
We built reliable OT→IT pipelines. We contextualised the data: every variable gained industrial meaning, not just a numeric tag. We separated the transactional from the analytical environment — without this, any analysis is suspect.
Real-time control charts. Cp and Cpk per variable — the operation could now see where the process was out of control before failure. Digital Forms & Checklists replaced paper on the shop floor. Web dashboards for the entire operations team.
Anomaly Detection via AutoML complementing the statistics — with the Mekatronik team validating every output before notifying maintenance. Process Sentinel + Notifier alerting the right people at the right moment. No black box: every alert has a traceable origin.
An unprecedented result in the operation — measured, witnessed, replicable.
Chlorum did not buy an off-the-shelf "AI-powered" technology. It bought a four-phase method that respected the plant's pace. AI only appeared in phase 4 — on top of a solid foundation of reliable data and statistical control. That sequence is not a detail: it is what makes the difference between a project that works and a project that stalls.
"More than 70% of the digital transformation projects don't finish. Many digital transformation projects are not conceived with the aim of impacting business results."
Dênis Leite · CEO, Mekatronik · Hannover Messe 2026
Technology chosen for each phase — not technology pushed.
Each tool was selected to solve a real problem at that moment in the journey. Nothing was installed before having an adequate foundation to run on.
Stops that keep repeating. Energy escaping without explanation. The plant running below its potential and nobody knowing exactly by how much. If this resonates, it starts with a diagnostic conversation. No immediate proposal — just to understand the problem up close.
