A rising temperature can mean several things. The machine may be making a different product, running under a different load, or operating after a recent adjustment. A maintenance investigation becomes useful when those facts are considered together.
Predictive Maintenance in Orenda brings the available measurements, production records, service history, and machine documents into one investigation. The team can inspect the evidence behind a finding and decide what to do next.
Begin with the machine and the question
Choose a named machine and a specific question. “Why did the temperature rise during this run?” gives the investigation a clearer job than “Analyze the factory.”
Record the period and the available signals. Confirm units, reading times, and operating states. A stale value, a changed sensor, or a new scaling rule can otherwise look like a change in machine behavior.
Compare similar operating conditions
A comparison is most useful when the periods are comparable. Look for the same product or recipe, a similar load, and a known machine state. Note a changeover or maintenance intervention instead of mixing it into an unexplained average.
For example, an investigation might compare the latest run with earlier runs of the same product, then check whether a recorded service intervention falls between them. This is an illustrative method, not a claim that a particular machine fault has been diagnosed.
Bring the manual into the investigation
A manual may explain the alarm, component, inspection step, or expected operating condition. Keep it linked to the correct machine and version so the team can open the passage behind the answer.
Orenda processes and indexes machine documents locally. The investigation can use permitted passages together with measurements and records. A missing page or an unindexed document should lead to a clear next step: add the relevant source or complete its preparation.
Keep the result connected to its evidence
A useful finding should make four things easy to locate:
- what changed and over which period;
- which measurements or records support the finding;
- which document passage is relevant;
- what the team should check next.
Open the underlying source before deciding on work. Save the investigation so the next engineer can continue from the same evidence rather than reconstructing the conversation.
Follow through after the report
The report is part of the service record. Review the proposed action, carry out the agreed check, and record the outcome. Later investigations can use that retained history.
Orenda supports saved investigations, repeat schedules, and HTML or PDF exports. The workflow can continue through reviewed maintenance actions where the relevant modules and permissions are installed.
Understand the local and hosted parts
Machine records and original documents stay on the Box. Document preparation and indexing are local. Hosted AI analysis sends selected authorized measurements and document excerpts through the configured inference service. A customer-managed endpoint can also be configured.
The deployed applications, machine grants, available sources, and inference connection determine what an investigation can use.
What should the first pilot prove?
Choose a recurring maintenance question on one machine. Prepare the relevant signals, a known production period, previous service notes, and the manual. Run an investigation with the maintenance team and review whether the cited evidence helps decide the next check.
A maintenance investigation is not automatically a validated remaining-life model. If the requirement is a calibrated failure probability or a precise failure date, define that separate validation task with suitable machine data.
Start with a result your engineers can examine: a clear change, an understandable source, and a practical next check.
