A Digital Twin Needs a Maintenance Plan

The handover is not the finish line

A digital twin can be impressive at handover: equipment is located, systems are linked and operating data is visible. Its long-term usefulness depends on what happens after a valve is replaced, a tenancy changes or a control sequence is revised. If those changes never reach the model, the organisation acquires a polished representation of a building that no longer exists.

Hong and Li’s documentation guidance concerns building-AI research rather than a digital-twin standard [HONG]. Its emphasis on traceable data and model context nevertheless exposes a relevant management problem. NIST’s lifecycle approach to cyber incident preparation adds a separate reason to know which systems and responsibilities are current [NIST].

Specify what the twin must support

Begin with a decision: troubleshooting, asset planning, space management or operational simulation. A model built for visual navigation may not contain the data necessary for equipment diagnostics. A simulation model may depend on assumptions that are unsuitable for real-time control. Calling both a digital twin does not make their capabilities interchangeable.

For the chosen use, define authoritative sources and update frequency. A maintenance record may own component status, a drawing system may own location and a control system may own current operating values. The twin should expose these sources instead of quietly becoming a competing record that staff must reconcile manually.

Figure 1. Four checks before accepting a performance claim Original Osmos Global framework informed by Hong and Li (2026), §§3.1–3.5, CC BY 4.0. No source diagram reproduced. Prepared 1 September 2026.

What this means: A strong algorithm cannot compensate for an invalid comparison.

Make change a managed transaction

Every relevant change needs a trigger, responsible person and confirmation step. The workflow should capture what changed, when it became effective, which dependent relationships were affected and whether the twin was updated. Where an update cannot be completed immediately, mark the affected information as provisional.

Include ordinary events, not only major projects. Renaming a point, replacing a sensor with a different range or reassigning a space can break analytics without changing the appearance of the building. A change register should identify these small events before they accumulate into unexplained model errors.

Fund stewardship and portability

The operating budget must cover model stewardship, data interfaces, quality checks and correction of discrepancies. If the procurement case includes only initial modelling and software licensing, it understates the lifecycle cost. The owner should know who maintains the twin when the implementation partner leaves.

Portability is equally practical. Test whether asset identities, relationships, history and documentation can be exported in a usable form. A file that can be downloaded but cannot be interpreted by another team does not provide a credible exit. Require an export rehearsal while the supplier is still accountable for acceptance.

Illustrative decision rehearsal

Imagine a tenant refurbishment changes room boundaries and replaces several terminal units. The digital model still presents the old layout, while the maintenance platform contains the new assets. An engineer looking for a fault may follow a valid-looking link to the wrong equipment. This is an illustrative coordination failure, not an assertion about a particular project.

The handover process should reconcile the physical change, asset register, control points and model relationships. The change owner should identify which downstream analytics and documents require revalidation. Closing the construction task before those updates are confirmed leaves an information defect inside an apparently completed project.

For the first operating cycle, select a recent real change and trace how it propagated through the organisation’s records. Look for manual re-entry, missing approvals and conflicting identifiers. Use the result to define a repeatable update workflow with a visible completion criterion.

At the next governance review, ask whether users can distinguish current, provisional and historical model states. Historical information can be useful for understanding a past fault, but it should not be confused with the current configuration. A reliable twin makes time and version explicit. This is the practical difference between preserving a useful operating record and maintaining an attractive but progressively stale representation of the building.

Measure trust, not visual complexity

Useful measures include the proportion of decision-critical assets with verified records, unresolved discrepancies, update delay after physical change and whether users can trace a conclusion to its source.

These are proposed management measures, not a published industry benchmark. Their purpose is to reveal whether the model remains fit for the decisions it supports.

No cited source establishes a universal return on investment for digital twins. Benefits depend on use, maintenance and integration. Osmos’s recommendation is therefore modest: procure a sustainable information service, not simply a three-dimensional deliverable. A less elaborate model with reliable stewardship can be more useful than a richer one nobody maintains.

Source notes

[HONG] Tianzhen Hong and Han Li. Good practices for documenting AI-based studies on energy and buildings. Energy & Buildings / Elsevier; author copy hosted by Lawrence Berkeley National Laboratory, 2026-01-20. Sections 2, 3.1–3.6 and 4; pp. 1–4. DOI: 10.1016/j.enbuild.2026.117043. Accessed 1 September 2026. https://eta-publications.lbl.gov/sites/default/files/2026-06/1-s2.0-s0378778826001039-main.pdf [NIST] Alexander Nelson, Sanjay Rekhi, Murugiah Souppaya and Karen Scarfone. Incident Response Recommendations and Considerations for Cybersecurity Risk Management: A CSF 2.0 Community Profile. National Institute of Standards and Technology, 2025-04-03. Section 2; Table 2 GV.SC-05/08; Table 3 RC.RP. DOI: 10.6028/NIST.SP.800-61r3. Accessed 1 September 2026.

Editorial and visual note

This is original Osmos Global analysis informed by the cited publications. Reported findings are distinguished from Osmos recommendations and illustrative scenarios. Source findings and trademarks remain attributable to their owners. Original visual designs do not imply endorsement by source organisations. The content is general research and does not replace site-specific professional advice.