BIM Validation for Refurbishment: A Practical Checklist

A refurbishment model is useful only when the project team knows what it can be trusted to answer. BIM validation is the disciplined check that connects the model to the available site evidence, the intended decision and the limits that still need investigation. It is not a claim that every part of an existing building has been proven.

That distinction matters in occupied buildings, phased upgrades and MEP-heavy work. Existing drawings may be incomplete, later fit-outs may have changed the space, and services can be concealed or obstructed. A practical validation process gives designers, contractors and owners a shared basis for deciding what to coordinate now, what to verify on site and what to leave clearly unresolved.

What BIM validation means in refurbishment

For an existing building, validation asks two related questions. First, does the model represent the conditions that matter for the next project decision? Second, does it contain the agreed information in a form the receiving team can use? The answer is never simply “yes” because a model looks detailed. It should be supported by specific checks against the evidence and the brief.

Visible geometry can be compared against a registered point cloud, drawings, photographs and survey notes. Information can be checked against agreed names, classifications, locations, attributes and deliverables. Crucially, the process also marks what is unknown. A ceiling void that could not be accessed, a wall build-up inferred from limited evidence or an asset without a confirmed serial number should remain an open condition, not become a silent certainty in the model.

The ISO 19650 information-management process treats information production as a sequence that includes checking, review, approval, authorization and acceptance. For a project team, that is a useful practical principle: agree the acceptance test before modelling begins, then make each review answerable to it.

Start with the decision, scope and tolerance

The first validation step happens before a model is reviewed. Write down what the model needs to support. A ceiling-services replacement, a tenant fit-out, a structural opening or an energy retrofit each creates a different set of risks. The model may need to establish routes and clearances, confirm floor levels, locate major structure, document existing plant, or provide a base for proposed design. It does not need to do all of those things by default.

Then define the scope in a way a reviewer can test. Identify the building areas, disciplines, elements, source data, coordinate reference, output formats and responsible parties. State the tolerance or acceptance basis that is meaningful for the decision. A feasibility model can be checked differently from one supporting fabrication or congested MEP installation. Generic detail labels do not replace this conversation.

Make the acceptance checks observable. For example, specify the views or sections that will be reviewed, the rooms that need a site comparison, the data fields that must be populated and the format of the limitations record. This keeps the review from becoming a general request to “check the model” and makes it easier for a project lead to see when the agreed work is complete.

A strong brief separates three layers: what is expected to be present in the model, what must be checked against the available evidence, and what the client accepts as a known limitation. This stops a reviewer from rejecting a model for an item outside scope, while also stopping a delivery team from hiding a genuine evidence gap behind broad wording.

Build a validation plan before the model is complete

Waiting for a full model before checking it is one of the most expensive habits in existing-building work. By then, an incorrect coordinate base, a misunderstood level convention or an omitted service zone can be repeated across hundreds of elements. Instead, agree short review gates: one after the source data is prepared, one after representative areas are modelled, one before other disciplines coordinate against it, and one before issue.

At each gate, choose a small number of high-risk locations. Plant rooms, shafts, congested corridors, façade interfaces, level changes and areas where proposed work must connect to retained fabric normally reveal more than a random sample of open-plan space. Check plans, sections, elevations and three-dimensional views, because a model can appear aligned from one view while drifting in another.

Reality-capture data remains useful after the first model review. A point cloud preserves measured spatial relationships and can be returned to when a design option changes. As a practical workflow reference, SHARE3DCAM’s existing-building guide separates capture, point-cloud preparation, existing-condition review, modelling, model-to-cloud verification and deliverable preparation. The relevant lesson is not a particular software path. It is that validation is a planned stage between the model and the next decision.

A practical BIM validation checklist

Use the following checklist as a starting point, then add project-specific tests. It is designed for the questions a refurbishment team needs answered before it relies on an existing-condition model.

• Confirm the reference system: check units, levels, grids, origin and any survey control against the agreed project reference. The proposed design cannot be coordinated safely if its starting position is uncertain.
• Test the source-data coverage: confirm that required rooms, routes and interfaces are represented in the survey material. Mark inaccessible, reflective, obscured or unscanned areas clearly.
• Compare critical visible geometry: use plans, sections and elevations to compare walls, slabs, columns, openings, major plant and visible services with the point cloud or survey evidence.
• Check the modelled extent: make sure each discipline has modelled only what the brief requires, and that omitted items are documented rather than accidentally left out.
• Review clearances and interfaces: where the refurbishment changes ceilings, equipment or services, compare the proposed envelope with retained beams, ducts, trays, pipes and access zones.
• Check information fields: test a sample of names, types, classifications, locations and required properties against the agreed schedule or information requirement.
• Review deliverables as received: open the planned model views, exports, drawings and schedules as the receiving team will. A correct authoring file is not enough if the issue package is unclear.
• Issue the limitations record: identify known deviations, assumptions, incomplete evidence and follow-up surveys. This is part of the deliverable, not a private production note.

The checklist should be proportionate. A small office fit-out might use a concise review sheet and a handful of measured checks. A hospital, laboratory or industrial upgrade can need formal issue tracking, discipline-specific rules and repeated coordination reviews. In both cases, the aim is the same: make the next decision from evidence that the team understands.

Use the model-to-site check for MEP decisions

MEP refurbishment is where untested existing conditions become costly particularly quickly. A new route can look clear in a design model but collide with a retained beam, cable tray, pipe branch or access zone on site. Validation should therefore compare the proposed envelope against the actual conditions captured or observed, rather than only federating design models that share the same outdated background.

Check the specific decision that is at risk. That can mean a riser route, a plant replacement path, a valve-access zone, a ceiling coordination strip or the clearance around new equipment. Solibri’s refurbishment validation example describes combining point-cloud data with planned BIM to check clashes, clearances and alignment against site conditions. The same principle applies outside healthcare: the practical test must relate to the work about to be installed.

Where the site evidence is incomplete, do not force the model to give a false answer. Raise a site-verification action, give it an owner and keep it visible through coordination. That is a better outcome than resolving a clash in software only to discover that the retained condition was different.

Make responsibilities visible at every review

Validation fails when everyone assumes someone else has checked the important condition. The survey team can confirm what was captured and how it was registered. The modelling team can explain how the evidence was interpreted. The design or coordination lead can decide whether the model is sufficient for the next stage. The client or appointing party can accept the agreed information for its stated use. These roles can sit within one organization or across several, but the handoffs should be written down.

Keep the review language practical. Each finding needs an owner, a due point and one of a small number of clear outcomes: accepted, revise the model, verify on site, outside scope or decision required. That gives a coordination meeting a useful purpose. Instead of debating whether the model is “accurate enough”, the team can decide whether a specific retained beam, ceiling zone or plant connection has enough evidence for the proposed work.

The appointing party is especially important because it defines what useful information looks like. ISO’s current process description notes that the party receiving information is responsible for reviewing and accepting submitted models against the required information need and acceptance criteria. In day-to-day project terms, a model review should be based on the questions the receiving team must answer, not a generic visual inspection of everything in the file.

Record exceptions so the handover stays useful

A validation record does more than list problems. It tells the receiving team what was checked, against which evidence, by whom, when, and what remains outside the model’s reliable use. A simple table can capture the area, check, evidence, outcome, action owner and status. Link the record to model views or issue references when that makes the location easier to find.

Keep observations specific. “Services to verify” is not an actionable limitation. “Ceiling void above meeting rooms 2.14–2.18 was inaccessible during survey, verify existing branch connections before final route installation” gives the next team a decision and a location. The same is true for assumptions about wall thickness, structural build-up and hidden connections.

This is also how a model remains useful as the project evolves. The point cloud, notes and validation decisions form a traceable basis for later design changes instead of forcing a new team to rediscover why an element was modelled a particular way. For a fuller view of planning the evidence from capture through handover, read our Scan to BIM workflow guide.

When an independent delivery partner helps

Project teams often have enough site data to begin, but not enough time to turn it into a controlled, reviewable existing-condition model. The useful delivery partner is the one that begins with the intended decision, checks the evidence before modelling at scale and issues the model with its limitations visible.

ASBUILDEXTECH supports Scan to BIM delivery for existing buildings and can help structure a model-validation approach around refurbishment, renovation and coordination work across Germany, Luxembourg and Europe. Where proposed services must be tested against retained conditions, the team can also plan an MEP coordination package with the relevant reference models, review zones and responsibilities agreed up front.

Related post

Plan the workflow before you validate it.

Scan to BIM Workflow: A Practical Guide for Existing Buildings

A clear route from point cloud to an existing-building model that is ready for design, refurbishment and coordination decisions.

Read the guide →