Skip to content

BIM SERVICES

BIM modeling, coordination and model-based drawing production.

Discipline models built to an agreed level of detail, coordinated until the clashes are closed rather than counted, audited automatically against your information requirements, and reviewed and signed by senior engineers.

What we model, and to what level of detail

Most disappointing BIM engagements go wrong in the first week, when nobody wrote down what the model is for. A model built for planning consent, a model built for coordination and a model built to produce fabrication drawings are three different objects that happen to look alike on screen.

So the level of detail and the level of information are fixed per element class before modelling starts, using the LOD framework the project already runs on, and then they are enforced by the audit rather than by memory. Within that agreement we model:

  • Structural models — concrete, steel and composite framing, foundations, cores and connections, at the detail the downstream use actually needs.
  • Architectural models — envelope, internal fabric, openings, finishes and room data.
  • MEP models — services routes, plant, containment and clearances, modelled to be coordinated rather than to look complete.
  • Façade models — curtain walling, rainscreen, cladding and glazing set out with the brackets, movement joints and interfaces back to the frame that the other packages have to build around.
  • Infrastructure and civil models — alignment-based work for roads, rail, tunnels and drainage, exchanged as IFC 4.3 where the receiving party can read it.
  • Existing conditions from survey and point cloud data.
  • The federated model that holds them together, with a documented origin, shared coordinates and levels that agree across disciplines.

Level of detail is agreed in the vocabulary your project already uses rather than ours, and the numbers mean different things to different people, which is why they get written down. Roughly: LOD 200 is generic geometry good enough to plan around, LOD 300 is designed geometry at real size and position, LOD 350 adds the interfaces between systems that coordination actually turns on, LOD 400 is detailed enough to fabricate from, and LOD 500 is what was verified as built. A fuller explanation of the LOD levels is on the blog. What matters here is that the agreement is per element class, not per model — the frame can be at 400 while the ceiling void is at 200, and that combination is a perfectly sensible model as long as somebody decided it.

Coordination that closes out, not clash counts

A clash report with four thousand items is a way of not doing coordination. The number goes down when the rules are right, and the rules are the work: which pairs of disciplines are tested, at what clearance, with which tolerance, and which results are noise that should never have reached a person.

We set a clash matrix per project, run it on the federated model on a fixed cycle, and manage the results as issues with an owner and a state — raised, proposed, agreed, resolved, verified — exchanged in BCF so your design team works in their own tools. Zones are signed off as coordinated, in a sequence that matches the construction programme, so the areas being built first are the areas resolved first.

What comes back to the design team is a short list of real conflicts with a proposed resolution against each. What comes back to the project manager is the state of every zone, not a screenshot of a clash.

Scan-to-BIM and existing conditions

Refurbishment, plant modification and extension work all start with the same question: what is actually there? Point cloud data from laser scanning or photogrammetry is registered, cleaned and turned into a model at a level of detail that matches the decision it supports.

The important discipline is honesty about deviation. As-built elements are modelled with the survey deviation recorded rather than idealised to the design position, and a deviation report identifies where the built structure has moved far enough from the design model to matter — before a new steel frame is fabricated to fit a wall that is forty millimetres out.

Deliverables are the model, the registered point cloud, the deviation analysis, and 2D drawings cut from the model where the project still runs on drawings.

The other half of this work has no scanner in it at all. A great deal of existing building stock is documented only as a 2D drawing set, and converting that set into a model is a different discipline: the drawings disagree with each other, the sections were never updated after the amendment, and levels are stated three ways. CAD-to-BIM is therefore done as reconciliation rather than tracing — the conflicts between sheets are listed and resolved with you before geometry is committed, and the assumptions behind each resolution are handed over with the model. A model built by quietly picking whichever drawing was on top is worse than no model, because it looks authoritative.

Working inside ISO 19650

ISO 19650 is not a certificate on a page; it is a way of running information so that everyone knows which version is real. We deliver inside it as a matter of routine.

That means information requirements read and answered rather than filed, a delivery plan with dates that match the design programme, file and container naming to the agreed convention, status and suitability codes applied honestly so that shared is not confused with published, and models issued to the common data environment the project actually uses rather than to an email thread. Where a project does not run a common data environment, we run the exchange and hand over the register at the end.

Where a project has not set that machinery up yet, we build it rather than complain about it. That means reading the exchange information requirements and answering them clause by clause, writing the execution plan before appointment and maintaining it after, producing the master and task information delivery plans with dates that reconcile against the design programme, and setting up the template, family library, view templates, naming convention and classification — Uniclass, OmniClass or the client's own — so the standard is something the authoring environment enforces rather than something a document describes. Where the team is doing this for the first time, the handover includes the configuration and a walk through it, so the next project does not start from zero. A common data environment only works if everyone understands what its status codes mean.

ISO 19650-5 requirements are applied where a project is security-minded — defence, energy, data centres and critical infrastructure — including what is modelled, what is federated and who receives which container.

Machine-audited models

Model quality is measurable, so it is measured. Every model is audited automatically before issue:

  • naming and classification of every element against the project convention;
  • required parameters populated for the agreed level of information — not just present, but populated;
  • geometry sanity: duplicated elements, orphaned or unhosted objects, elements off-grid or off-level, warnings that have been left to accumulate;
  • IFC export validity, so what leaves the authoring tool is what the project receives;
  • quantities reconciled between the model and the schedules cut from it;
  • drawings reconciled against the model revision they were produced from;
  • federation integrity: shared coordinates, level names and units consistent across disciplines.

The audit report is issued with the model. A senior engineer reviews the model and the audit together and signs the issue.

From model to drawings, quantities and asset data

A model earns its cost downstream. We cut the deliverables from it rather than drawing them again: sheet sets produced from views with annotation and schedules that update with the model, quantity take-off structured to the measurement rules the commercial team uses, and asset and handover data mapped into the client's format rather than exported raw.

Exchange is deliberate. Native files where the recipient uses the same tool, IFC where they do not, and both checked against each other before issue. Where the model feeds detailing — steel, precast or reinforcement — it is built so that the detailing package can be produced from it rather than modelled a second time.

Time, cost and the data that outlives the project

The extra dimensions are only worth the money when somebody is going to make a decision with them, so they are scoped against that decision and not sold as a package.

4D links the model to the construction programme so a sequence can be looked at rather than imagined. The useful part is not the animation; it is that the model has to be broken down the way the work is actually done — by zone, by pour, by lift, by possession — and doing that breakdown exposes sequencing conflicts, crane reach problems and temporary works that a bar chart hides. Where the programme changes, the link is re-run rather than re-built.

5D ties quantities to the cost plan using the measurement rules your commercial team already applies, so the model produces a number they recognise instead of one they have to translate. The quantities are the same ones the automated audit reconciles against the schedules, which means the cost model and the drawing set cannot silently disagree.

Asset and handover data is the deliverable that survives everybody. Parameters required at handover are agreed at the start and audited throughout, rather than harvested from a finished model in the last fortnight — which is the usual way it is done and the reason it is usually wrong. Output is COBie where the client asks for COBie, and the client's own asset register schema where they have one, mapped rather than dumped.

One partner, one signature

BuildTwin builds the model, audits it and signs it. One delivery lead, one information delivery plan, one company answerable for what is in the container — not a pool of people you have to manage.

The work runs across hospitals, data centres, transport structures, industrial plants and commercial frames, in your authoring environment, on your templates, to your naming convention. Capacity does not change the arrangement: when the programme needs more zones coordinated per week, the system produces more; the audit and the signature stay with the same senior engineers.

FAQ

Common questions

Which level of detail do you model to?

Whichever one the downstream use requires, fixed per element class before modelling starts and then enforced by the automated audit. A model for planning, a model for clash coordination and a model that produces fabrication drawings need different levels of detail, and agreeing that in week one is the single cheapest decision on a BIM engagement.

Do you work in our Revit environment and templates?

Yes. Your template, families, view templates, naming convention and worksharing arrangement are the starting point, not something we translate to afterwards. Models are delivered as native files and as IFC, checked against each other before issue, so parties outside your authoring tool receive the same geometry and data.

What does working to ISO 19650 mean in practice?

Information requirements answered rather than filed, a delivery plan whose dates match the design programme, container naming and status codes applied consistently, models issued into the project common data environment with an honest suitability code, and a register at handover that shows what was issued when. Where the project runs no common data environment, we run the exchange and hand over the register.

Can you produce a model of an existing building from a point cloud?

Yes. Point cloud data is registered and cleaned, then modelled at a level of detail matched to the decision it supports. Deviation between the built structure and the design intent is reported rather than smoothed away, which is what stops a new frame being fabricated to fit a wall that has moved.

Do you deliver 4D sequencing and 5D quantities?

Yes, scoped against the decision they support rather than sold as a bundle. 4D requires the model to be broken down the way the work is built — by zone, pour, lift or possession — and that breakdown is where the sequencing conflicts and temporary works problems surface. 5D uses your commercial team's measurement rules so the model produces a number they already recognise, reconciled by the same audit that checks the drawing schedules.

Can you turn an existing 2D drawing set into a model without a scan?

Yes, and it is treated as reconciliation rather than tracing. Old sets contradict each other — sections that were never updated, levels stated three ways, an amendment that reached the plan but not the detail. Those conflicts are listed and resolved with you before geometry is committed, and every assumption is handed over with the model. A model built by quietly picking whichever drawing was on top is worse than no model, because it looks authoritative.

Do you write the BIM execution plan and set up the standards?

Yes, where the project needs it. That covers answering the exchange information requirements clause by clause, the pre- and post-appointment execution plan, the master and task information delivery plans with dates that reconcile against the design programme, and the template, family library, naming convention and classification set up so the authoring environment enforces the standard rather than a document describing it. The configuration is handed over, so the next project does not start from zero.

How do you keep drawings and quantities consistent with the model?

They are cut from the model rather than drawn separately, and the automated audit reconciles them before issue: schedules against model quantities, and each sheet against the model revision it was produced from. A drawing that no longer matches its model is caught by the check, not by the reviewer who happened to notice.

Put one zone through the audit.

The AI Factory is in private preview. Request access and we will model and coordinate a single zone on your template, and hand back the audit with it.