Industrial data analytics: from capture to proactive decision
We read what your machines measure and turn it into decisions, in five phases: capture, analysis, preventive, predictive and proactive. Each phase builds on the previous one, and you can stop at whichever makes sense for your plant.
A machine gives no warning before it stops. Its data does
Fitting sensors is not enough to read that data. First you have to measure properly, then understand what is normal for each machine, and only then alert, anticipate and change the process. That is why we work in phases: each one leaves something useful in your plant and prepares the next.
From measuring to anticipating, in five steps
Same machine, same signal. Each step gets a little more out of it.
Measure what today is estimated
We fit sensors where they are missing and read those already in your PLCs, SCADA and ERP. Everything arrives in one place, time-stamped, including from older machines.
What you getYour machines’ data in one place, continuously.
Know what is normal for each machine
With a few weeks of data we learn how each machine works across its shifts, start-ups and format changes, and separate what is normal from what signals a problem.
What you getA dashboard with the status and history of every machine.
Alert before it fails
We set meaningful thresholds for each signal. When one goes outside normal, we alert you with what is happening and what should be checked.
What you getAlerts by machine and by signal, each with its explanation.
Know when it is going to fail
A model learns from the history and estimates when the threshold will be crossed, with its margin of error. So the intervention is planned, not improvised.
Predictive maintenance with IoT →
What you getProbable failure date and its margin.
Change the process so the failure never happens
With the digital twin of the machine or line, we test changes in speed, load or sequence virtually before touching the plant.
What you getChanges tested in the digital twin before touching the line.
Machine variables and sources you already have
There is almost always more data than it seems. We start by reading what already exists and only measure what is missing.
What we measure on the machine
The most common ones. In the assessment we choose those that make sense in each case.
- Vibration
- mm/s
- Temperature
- °C
- Pressure
- bar
- Power consumption
- kWh
- Stoppages and production cycles
- h
Where each piece of data comes from
We connect to what already works in your plant, without changing it.
- New sensors
- Where nothing is measured today
- Existing sensors
- Those the machine or installation already has
- PLC
- Machine run, stop, cycles and alarms
- SCADA
- What can already be seen in the control room
- ERP
- Orders, shifts and recorded stoppages
Each phase has its solution
Industrial sensorisation
We connect sensors, PLCs and ERP, including older machines with no communications.
2 · AnalysisIndustrial monitoring
Consumption, temperatures and stoppages across the whole plant on one dashboard, with alerts when something goes out of range.
3 and 4 · Preventive and predictivePredictive maintenance
We read the signal of the machine that worries you and tell you what is wrong and when to step in.
5 · ProactiveDigital twin
A virtual replica of your machine or line to test changes before touching the plant.
Industrial data analytics, in a nutshell
If your question isn’t here, ask us during the assessment.
What is industrial data analytics?
It is turning what a plant’s machines measure (vibration, temperature, pressure, consumption or stoppages) into information for decisions: what is happening, what is going to happen and what should change.
Do you have to go through all five phases?
No. Each phase delivers something useful on its own. Many plants start by measuring and analysing one machine, and move on when it makes sense.
Do I have to change my machines or systems?
No. We read what your PLCs, SCADA and ERP already generate, and only add sensors where they are missing, including on older machines.
When does the first useful alert arrive?
It depends on the machine. First we read its signal for a few weeks to learn what is normal; the alerts follow from there.
What is the difference between predictive and proactive?
Predictive tells you when something is going to fail. Proactive changes the process so it never fails, testing the change first in a digital twin.
Where do we start?
With an assessment: we look at which machine or process worries you, what data already exists and what would need to be measured.