Conflicting equipment states in a digital twin: which source should the display trust?
Do not compress production, connectivity and maintenance into one truth value. Establish authority and validity by field, then choose a task-specific primary display while retaining conflicts.
On this page5 sections
Production software may report that equipment is working, an IoT platform may report disconnection, and a work-order system may show maintenance in progress. These are not necessarily contradictory values of the same field. Separate production, connectivity and maintenance first, then examine each source and validity period. A primary model color may emphasize the current task, but it should not erase the other facts.
Start with a multidimensional state table
| Dimension | Authoritative field to confirm | When its evidence expires |
|---|---|---|
| Production | The operating or job state confirmed by the production system | Retain the last state and source time, marked awaiting update |
| Connectivity | The collection chain's connection assessment | Show uncertainty about data, not an inferred physical shutdown |
| Maintenance | The work-order system's actual maintenance stage | Retain its evidence without rewriting production feedback |
A work order created before maintenance begins should not automatically become “equipment stopped for repair.” Define each field's meaning, object identity, source, time and validity condition. Naming the source system alone does not define what its field proves.
Resolve conflicts within the same field
Suppose two sources both claim to supply the current operating state, one reporting running and the other stopped. First confirm the same physical asset and relevant time. Then apply the approved source rule. If the primary source expires, specify whether a secondary source may take over, for how long and with what confirmation. These decisions should not be hidden in whichever connector was implemented last.
Do not use “last message received wins” as a universal business rule. Backfilled records can arrive late. A fresh, valid primary value can take precedence under the agreed contract. If it expires and no fallback is authorized, show pending confirmation. When equally authoritative valid sources disagree, retain the conflict for the responsible owner rather than convert uncertainty into normal operation.
Distinguish processing time from observation time
The Azure Digital Twins documentation distinguishes property processing time, lastUpdateTime, from optional source observation time, sourceTime, and provides ETag-based conditional updates. This supports managing stored updates; it does not decide which business system has authority over a field.
Even a concurrency-protected update can successfully store a business state that is too old. Conflict evaluation still needs original state, source time, receipt time, quality and the applied rule version. If clocks are unreliable, retain that limitation instead of forcing an ordering based only on timestamps. A successful write and a trustworthy current interpretation are separate checks.
Choose the primary color for a particular task
For operational inspection, lost communication may need to interrupt any implication of “currently normal,” while details retain the last operating state and maintenance stage. A maintenance-planning view may emphasize maintenance progress instead, provided that the color is not presented as physical operating feedback. This is an example of interface policy, not a universal alarm-priority standard.
Record a decision table connecting input combinations to the primary display, supporting labels and accessible evidence. A new unhandled combination should enter an unknown or review state rather than default to green. Pair color with text so that users know which state dimension they are reading and can inspect the remaining dimensions.
Accept deliberately contradictory examples
Test running production with disconnected telemetry; a created but unstarted maintenance order; an expired primary source with a valid secondary source; opposing valid sources; and an old state arriving late. Check the primary display, dimensional details, conflict list and recovery behavior. Human confirmation should retain the person, time and evidence rather than silently editing original source records.
Connecting Custom Software to ERP, CRM and OA : Who Owns the Data? covers system responsibilities, and Which Systems Should a Data Center Digital Twin Integrate: DCIM and Environmental Monitoring Checklist describes data-center source scope. This article turns those boundaries into field-level conflict and display decisions. For digital twin development services (Chinese), deliver the decision table and counterexamples. When another source is added later, check rule coverage before attaching its states to the model.