A Complex Roof Modelling Example, Built Properly

A complex roof modelling example is rarely difficult because of one unusual pitch or dormer. The risk sits in the relationship between every element: uneven ridge lines, offset valleys, parapets, historic chimney stacks, rooflights and junctions that do not follow the original drawings. If that geometry is estimated rather than measured, errors quickly travel into elevations, sections, planning submissions, quantities and design coordination.

For refurbishment, heritage and conversion projects, the roof is often where an existing building stops behaving like a set of clean, repeatable components. A dependable model needs to show what is there, distinguish it from reasonable interpretation, and provide a practical basis for the design team’s next decisions.

A complex roof modelling example: from capture to model

Consider a listed Victorian property being converted for a new commercial use. The building has a main pitched roof behind a front parapet, two later rear extensions, a lead-lined valley, three dormers of differing sizes, chimney stacks that lean slightly from vertical and a small plant enclosure added during a previous alteration. Existing drawings show the footprint reasonably well, but the roof plan is incomplete and the elevation heights conflict with one another.

The architect needs a coordinated Revit model to test a proposed rooflight strategy, assess visibility from surrounding streets and develop alterations without assuming that the upper level is regular. The brief is not simply to produce an attractive 3D shape. It is to document the roof geometry reliably enough that the architect can design from it.

That changes the workflow. A technician should not begin by tracing a roof from photographs or forcing every surface into a familiar Revit roof tool. The first task is to establish dependable site evidence.

Capture the geometry before interpreting it

A 3D laser scan records dense spatial information across the roof structure, parapets, external walls and accessible internal areas. For a roof with restricted access, scanners positioned from upper-floor windows, terraces, adjacent spaces and ground level can capture much of the external form. Supplementary measured observations and photographs help verify materials, drainage directions, flashing lines and details that may be partially obscured in the point cloud.

The quality of registration matters particularly at roof level. A slight misalignment between scan positions can make a ridge appear broken or a chimney look out of plumb when it is not. Control, overlap and a careful review of the assembled point cloud are therefore part of the documentation process, not an administrative step before modelling begins.

There are limits. A survey cannot reliably model what cannot be seen or accessed. Concealed roof build-ups, hidden timber structure and areas screened by neighbouring buildings should be identified as assumptions or excluded from the agreed scope. Clear limitations are more useful to a design team than false precision.

Establish a modelling datum

Before creating roof elements, the model needs stable reference information. In this example, the building’s principal floor levels, external wall faces and parapet tops provide the framework. The modeller checks these against the point cloud and sets named levels that make sense for project coordination, rather than creating a separate level for every minor change in height.

This is a balancing exercise. Too few levels can make the model difficult to interrogate, especially where extensions sit at different heights. Too many can make documentation cluttered and increase the chance of elements becoming incorrectly constrained. The right structure reflects the building’s real organisation while keeping the model usable for architects, consultants and contractors.

At this stage, wall geometry is also checked. Roofs do not exist independently of the surfaces supporting or concealing them. An irregular parapet can hide a substantial difference in eaves height, while a historic wall that bows outward may affect where a valley actually lands. Modelling the surrounding envelope first prevents the roof from being developed against an idealised footprint.

Model each roof plane by evidence, not convenience

The main pitched roof in this example is best understood as a series of measured planes. Ridge, eaves and valley lines are extracted from the point cloud, with sections taken at key locations to confirm pitch and changes in direction. Rather than relying on a single nominal angle, the modeller verifies whether the two slopes genuinely meet at a straight ridge and whether the eaves line remains consistent behind the parapet.

For a simple roof, Revit’s footprint roof tools may be sufficient. For an irregular roof, a combination of roof elements, shape editing, in-place geometry or mass-based surfaces can be more appropriate. The choice should follow the intended output and required level of development, not a preference for one technique.

A native roof element is useful where the design team needs to edit pitch, build-up or drainage logic later. A carefully controlled in-place component may be more suitable for a small, non-standard lead-lined junction that must be represented accurately but is unlikely to be redesigned as a parametric object. The key is to avoid creating an unnecessarily complicated model that no one can maintain.

In this case, the main roof planes are modelled as editable roof elements. The skewed valley at the rear extension is formed from measured reference lines and checked in section. The plant enclosure, which has no meaningful architectural roof build-up and is represented primarily for coordination, is modelled as simpler geometry. Each choice is deliberate and proportionate to the brief.

Resolve dormers, chimneys and drainage junctions

Dormers often expose weaknesses in existing documentation. Their cheeks may not be parallel, their roofs may drain in unexpected directions, and their openings may sit off-centre relative to rooms below. A plan-only approach can make them look acceptable while producing incorrect elevations and sections.

For each dormer, the modeller reviews plan, elevation and section views against the point cloud. Cheek walls, sill heights, roof slopes and flashing locations are checked as a connected assembly. Where leadwork is not required at fabrication level, it can be represented with a clear, appropriate level of detail rather than an elaborate approximation.

The same principle applies to chimneys. If a stack is visibly tapered, stepped or slightly out of plumb, the model should reflect this where it affects design clearances, heritage assessment or the appearance of proposed works. It does not need to reproduce every weathered brick unless that is part of the agreed documentation purpose.

Drainage deserves attention because it explains the roof’s geometry. Valleys, gutters, flat roof falls and outlet positions can influence proposed rooflight locations and the feasibility of alterations. In a survey model, these items may be shown as visible conditions rather than assumed hidden construction. If the fall of a concealed surface cannot be verified, it should not be presented as fact.

Validate the model in the views designers will use

A roof model can look convincing in a perspective view and still fail when used for design. Validation should happen in orthographic roof plans, elevations, building sections and local detail sections. These views reveal whether edges align, whether roof elements cut correctly against parapets, and whether the model produces credible drawings without excessive manual correction.

For the Victorian property, the crucial checks include the relationship between the rear valley and extension walls, the height of dormer heads beneath the main ridge, and the visibility of rooflights above the front parapet. A point-cloud overlay in selected views provides an efficient way to compare modelled geometry against captured conditions.

This process also identifies where tolerance matters. A model intended for concept design may represent uneven slate courses as a single plane while accurately locating ridges, valleys, openings and parapet heights. A model supporting fabrication, complex structural coordination or heritage repair may need a more detailed scope. LOD is not a promise of universal accuracy; it must be paired with a clear statement of what has been measured, modelled and assumed.

Deliver files that reduce work for the project team

The value of complex roof modelling is realised when the output is structured for use. A well-prepared delivery may include the Revit model, the registered point cloud where agreed, roof plans, elevations, sections and a concise record of scope, assumptions and areas requiring further investigation. Naming, worksets, view organisation and element categories should support the receiving team’s workflow rather than reflect an internal production process.

For architecture practices working under programme pressure, this structure matters as much as visual accuracy. A model that is technically correct but difficult to navigate can still create delay. Equally, a fast model based on unreliable geometry moves uncertainty downstream, where changes cost more and coordination becomes harder.

Space Captures approaches complex existing-condition work with that practical balance in mind: precision-first capture, proportionate modelling and clean, design-ready outputs. The aim is not to make an irregular roof appear simpler than it is. It is to give the design team dependable geometry, clear boundaries of certainty and a sound basis for the decisions that follow.