Technology and method

Connect the evidence. Keep control on site.

Start with the systems you already operate. Robometic brings BMS and DCIM records, maintenance history and targeted condition observations into a structured engineering review. Every assumption, gap and model limitation remains visible.

Five layers, in order.

The order matters. Each layer is agreed before the next one is added.

  1. Existing data

    BMS trends, DCIM records, maintenance history, operating logs and the site's own procedures.

  2. Defined gap

    A specific question the existing evidence cannot answer, agreed with the site before any measurement is added.

  3. Targeted observation

    Selective temporary or permanent measurement where it adds decision value, using the site's access and safety rules.

  4. Dependency model

    Which equipment supports which load, where paths and controls are shared, and which constraints apply.

  5. Engineering review

    Assumptions, limits and next checks set out in writing for the authorised site team.

Existing systems first

The site's BMS and DCIM already hold much of the operating evidence. The assessment starts there, with read-only access agreed for the engagement.

Where a record is missing or inconsistent, that is an evidence gap. It is recorded, not estimated away.

Selective sensing

Additional sensing is introduced only for a defined question. The choice, placement and method are agreed with the site and are not published as a product specification.

Sensing does not replace protective devices, leak detection or the site's required instrumentation.

Asset and dependency model

The assessment maps the cooling path around the equipment in scope: shared headers, controls, valves, air-side distribution and the loads that depend on them.

The map shows where a single intervention affects more than one part of the system.

Engineering rules before models

Where evidence permits a bounded scenario comparison, it is built on stated engineering rules and assumptions. A model is not presented as the source of truth.

If the evidence does not support the comparison, the pack says so.

Model limits

Validation is established per application.

Site-specific modelling is a separate engineering engagement with its own validation basis. Its assumptions, capability and accuracy are established before any result is relied on.

A scenario outside the validated operating envelope is referred for engineering review. It is not given a reassuring answer.

What is agreed first

The question, the system boundary, the data sources, the validation basis and the review route.

What is reported

Assumptions, limits and uncertainty alongside any result. The site team sees what the model can and cannot support.

What a result can carry

Assumptions, limits and uncertainty travel with every result. A quantitative failure probability, an uptime assurance, a capacity rating or a redundancy classification sit outside the method, and no live facility data appears on this website.

Evidence in view

See condition changes in context.

Bring condition observations, operating context and review points into one view. The example below shows how an observed change becomes a focused engineering question.

Illustrative normalized condition trend A normalized trend across six observation windows. A condition index rises from window four and enters a review window. An operating context line is shown for comparison. Markers at windows four and five are marked unknown. No engineering units, threshold or capacity result are shown. ILLUSTRATIVE NORMALIZED TREND NOT MEASURED DATA · NOT A THRESHOLD REVIEW WINDOW, ILLUSTRATIVE 0.00 0.25 0.50 0.75 1.00 NORMALIZED, ILLUSTRATIVE illustrative baseline band unknown at review unknown at review W1 W2 W3 W4 W5 W6 OBSERVATION WINDOWS · NORMALIZED FOR ILLUSTRATION condition index operating context The chart shows direction and review points only. It does not assign a probability, a capacity rating or an approval.
Illustrative normalized readings from an example system.

Maintenance review

From observation to a reviewable request.

The storyboard shows the sequence the site can inspect. It is deliberately plain: observe, corroborate, bound the question, then let the authorised team decide.

Illustrative maintenance review storyboard Four review panels: observe condition and context, corroborate with site records, bound the dependency and unknown, and refer to the authorised site team. Each panel states evidence in and the question out. ILLUSTRATIVE REVIEW STORYBOARD 01 Observe Condition and operating context, recorded together. EVIDENCE IN Signal, load, state, history. QUESTION OUT What changed, and when? 02 Corroborate Compare with records, observations and checks. EVIDENCE IN Site records and checks. QUESTION OUT What explains the change? 03 Bound the question Shared path, controls, power and the evidence gap. EVIDENCE IN Dependencies and unknowns. QUESTION OUT What remains uncertain? 04 Review Authorised site team reviews the pack. EVIDENCE IN Question, limits, next checks. DECISION Site authorises or refers.
Evidence enters onceThe same record feeds the assessment and the review pack, so the site can trace each statement.
Unknowns are not failuresA recorded gap is a result. It changes the next check, not the truth of the observation.
The site decidesRobometic advises. Isolation, permits and protection remain with the authorised team.

The human boundary.

Advisory, not control

No connection that commands equipment. The site keeps control of its systems and its procedures.

Read-only by default

Access is agreed for the engagement, limited to what the question needs, and removed or reviewed at the end.

Authorised people decide

The site's authorised team reviews the evidence, checks the constraints and decides whether the work proceeds.

Technology connects the evidence. Engineering review establishes what it means. Your team retains control of the decision.

Data handling

Site-controlled evidence.

Access, storage and handling are written into the engagement agreement. The site's machine data is used for the agreed assessment, not pooled into a shared dataset or used for an unrelated purpose.