Capitalizing on Building Information Modeling (BIM) for new buildings is easy — you get to start with a fresh, clean file, and you can lay out everything you need precisely where it is supposed to be. But what if you’re doing an addition, a renovation, or a historic preservation? Most of what is important to a structural engineer is probably hidden behind an existing facade or finish. And yet, the model must go on. Fortunately, you’re not alone! This article summarizes the suggestions, ideas, and questions covered in a roundtable discussion that took place at AU Las Vegas 2016, led by Kate Morrical.
Kate Morrical talks about the importance of using BIM for existing buildings.
Disclaimer: All statements below reflect the experience of the attendees. They are not intended to be generalized conclusions as to the state of the industry.
Who Was There?
Among the 28 or so session attendees, we had mostly architectural folks, but structural, contractors, and facilities management were represented as well.
What Does “Existing BIM” Mean to You?
Work on existing buildings can take many forms. For some projects, you’re adapting a historic structure to serve modern functions. Sometimes, it’s refitting a space for a new tenant. Or it could kick in, in the form of facilities management, as soon as a new building is complete. Our participants have experience in all three types of work.
All agreed that meaningful discussions are necessary at the beginning of a project to ensure that all participants understand the goals and limitations of any models.
How Much of the Building Should You Model?
For structure, the original mindset was to model only what was needed for that particular project. But this can cause problems if other disciplines have more extensive scope (e.g., MEP systems throughout an entire building) and the structure isn’t there to use for coordination. Or if the scope of the project expands halfway through, and you have to go back and start chasing through the original document again. It’s great if you can get additional fee for this effort, but if you can’t it often pays off in its own right. This is especially true if you have future work on the same building (e.g., malls, office buildings, or healthcare facilities). Then you’ve front-loaded your research effort and don’t need to “mobilize” it twice.
Caveat to the previous statement: We didn’t have many MEP representatives, and the hypothesis is that MEP would prefer to not model existing elements until it’s known whether they will be needed for the new work.
A Side Note on Scope
How do you handle secondary structural elements like kickers and gusset plates? The immediate reaction was that they do not get modeled. But what if they’re critical for MEP coordination? In that case, bring it up early! If the request is made soon enough to include these elements, team members can plan for it.
Who Wants BIM?
Owners are not necessarily asking for BIM. Often it’s driven by the design team’s requirements. If a team member isn’t using BIM yet, do what you can to convince them. Coordination efforts are easier if everyone is in 3D.
How small of a project is too small for BIM? Pretty small! Projects of many scales can benefit from BIM, both for their actual work and in the cases of scope creep or future work as noted earlier.
A Comment on Model Element Authorship
Often the architectural model is the first to be started. Your MEP and structural team members request that you bring them in as early as possible so they can model their elements according to their own best practices. Models tend to turn out better if experts in each discipline do their own modeling. (We will come back to this point in the laser scanning section.)
Dealing with Uncertainty
There is always some level of uncertainty inherent in an existing building. This is especially true for structure and MEP, whose elements are often concealed inside architectural elements. How do you handle this?
Managing uncertainty is the biggest challenge with BIM for existing structures.
LOD, LOA, LOC
Level of Development (LOD) is not useful for existing elements. They exist, so by definition they’re fully developed. But that doesn’t mean we know everything about them.
Level of Accuracy (LOA) is a scale developed by USIBD for setting the accuracy of data gathering and for elements modeled based off of gathered data. For example, a laser scan could be done to meet LOA 20, and the elements modeled off of it could meet LOA 30.
Silman is working with a third scheme we call Level of Confidence (LOC). This is an adaptable scale that describes how much information you actually have about a given element. It enables you to model a wide flange section (for example) while indicating that you’re not absolutely certain that it represents the true field conditions.
In its simplest form, LOC is a single digit:
- 0 — Assumed based on existing documentation or symmetry
- 1 — Identified based on non-destructive evaluation (radar, thermography, etc.)
- 2 — Field-verified (laser scanned or probe)
Someone pointed out that “assumed” and “based on existing documentation” could qualify as two different levels. In that case, the scale would go from 0 to 3. You could also expand the scheme to multiple digits — for size, location, and material — if it made sense for your project.
Related to your confidence level, how much detail should you included in your modeled elements? Some of it might be determined by how much information you actually have, but consider historic windows. Is an eighth of an inch variance in width enough to trigger a different window type?
The general agreement was that precision levels have to be set on a project-by-project basis. Some buildings might be important enough to model eighth-inch differences. For others, a tolerance of one inch might be sufficient. The age of the building is also relevant in determining this. Buildings from the 1980s need to be treated differently from buildings from the 1880s.
We agreed that yes, you should rationalize your field measurements! (Our facilities managers would love to have everything modeled much more precisely, but acknowledge that this would mean we’d never get any work done.) An example given was a long wall clearly constructed in one shot, but modeled with 30 different types to represent minute variations in thickness. It wasn’t helpful for the goals of the model. Historic preservation projects can make an argument for using this kind of hyper-precision, but for most work it’s not necessary.
Two questions may help with framing these discussions:
- • Are you re-creating the design intent, or documenting actual existing conditions?
- • How important are minute variations? For example, consider an existing wall. Are you repainting it, or adding custom casework? The precision requirements for these two tasks can vary widely.
Circling back to the first topic, it’s something to discuss at the beginning of each project. To quote one of our attendees, “begin with the end in mind.”
We’d all love to have existing documents, but even when we get them, they’re not always complete, legible, or accurate. For some old buildings, technologies evolved faster than construction, so systems may vary across a building’s footprint or from floor to floor. (There go any assumptions of symmetry.)
Much of the room is still a fan of beginning the model from the existing documentation, just to have something to start with that has some level of authority. But to borrow a famous quote, “trust but verify.”
A few people in the room have their own laser scanners, but most still outsource it. Almost everybody has had a least one problem with models created by outsourcing. It seems to be another area of “let the professionals do what they do best.” Surveyors are great scanners. Architects/engineers are great modelers. Don’t be afraid of a division of labor that plays to everyone’s strengths.
We had a few different approaches to the use of point clouds as part of the design process:
- 1. Scans are overlaid on the model — which was created based on existing documentation — just for verification.
- 2. Scans are the starting point of the model.
- a. Sometimes made as an “initial deliverable” to hand over to the design team and be used as the basis for new documents.
- 3. Scans are part of the final documentation.
- a. This one was for plants and factories, especially to show connections to existing elements.
- b. The resolution can be adjusted to give the appearance of “lines”, or supplemented with photographs.
As with so many things, there’s no one right answer.
Future of Scanning
Request: The ability to select a group of points in a cloud and say “these represent a wall” — and then be able to assign material and geometry properties to that set of points. It would essentially bypass the middle step of modeling a probably-imprecise solid representation.
We heard about a research project underway in the UK that will be able to use a laser scan to identify the material of a scanned element, in addition to its location and color. This technology is probably several years away from commercial availability, but it’s potentially very interesting!
Useful Contract Language
Here are some things to consider if you are writing an RFP or contract for laser scanning and/or modeling based on a laser scan.
- • What do you want from the project? 2D construction documents, or an intelligent 3D model?
- • How important is it that elements are flat/straight? Is this a design-intent model or a fieldconditions model?
- • For design-intent models, Silman has found this language useful: “Elements with orthogonal design intent shall be modeled orthogonally.”
- • This means: Walls are straight, floors are flat, columns are vertical, and corner are right angles.
- • Yes, this triggers discussions of tolerances. But if you don’t say it, you might end up with 89.5 degree corners. And if there are elements where you need to know their deviation from orthogonal (for example, a bowed wall), you can make allowances for them.
- • Are you okay with in-place families? Or do you need system and component families to be used whenever possible? If the latter, consider including that in the contract.
Registration is the process of linking multiple point clouds together. It should be included in any competent scanner’s base fee. Tolerances may be an item of discussion, but execution is not. Surveyors often have expertise in setting control points to ensure the linked models match field conditions. (If they don’t, you might get a point cloud with a 36-inch deviation in floor elevation, instead of the actual 6-inch change.) And if your scan shows a wall as bowed when the actual field condition is straight…you need a better scanning partner.
In the context of laser scanning, “post-processing” refers to cleaning up the cloud data. Scan companies may be asked to remove mobile objects (people, cars) from the scan, or correct artifacts that are the result of the laser bouncing off reflective surfaces. This can be an expensive effort, but it can also be ameliorated if the requirements are made clear early enough that they can be planned for in the field (i.e., by getting people out of the scanned area ahead of time).
Photogrammetry is being used in some instances to supplement laser scanning data, for example high flat surfaces that aren’t visible from the floor. In one instance, scanning was used to document a railway site without shutting down the trains.
As some pointed out, there is often a disconnect between point clouds and useful data. Data must be interpreted and analyzed before it becomes useful information. Human interaction is still required to translate raw data into models.
Revit’s built-in tool for controlling phase graphics is effective, but it’s a pretty blunt instrument. Unless you combine it with filters or other overrides, all existing elements use the same line weight and fill pattern. And the biggest problem with filters is that you can’t use “phase status” — or even “phase created” and “phase demolished” — as filter criteria. Although you could set up custom parameters to contain this information — you could even use Dynamo to populate them automatically. Working within Revit’s capabilities may require adjusting your company’s graphic standards if you want to achieve efficiency.
Remember that there is a difference between a Revit phase and a construction phase. Not all construction phases require Revit phases. For existing buildings, this often means lumping all existing conditions into a single “as found” phase. If you are going to tweak any phase names or phase filter names and settings, be sure to coordinate that with the rest of the design team. Differences in phase names can be accommodated with phase mapping for linked files, but phase filter names must match if you want to use them to control linked models.
Do You Need a Demolition Phase?
Most of the time, you do not need a separate phase in order to show demolition. One participant pointed out an exception: Asbestos abatement might require a separate phase in order to keep it out of the regular demolition plans. Speaking of which…
If an element is going to be demolished, do you need to model it? It depends. (eBIM can turn us all into lawyers.)
Are you doing an assessment or evaluation of the existing structure?
- • Yes: You might want to model the demo.
- • No: A mass/solid/filled region might work out just as well
Are you giving the model to the contractor?
- • Yes: They might appreciate a more complete model
- • No: Just get the scope on there.
As with many other points, this circles back to the discussions on scope of work and scope of model.
What if you’re demolishing part of an element? Do your best to represent things accurately. You have several options for cutting holes or splitting elements. They each have pros and cons.
1. Shaft elements only support phases in one direction. “Existing” shafts may be demolished and filled in with new slabs. But “new” shafts will cut existing slabs even if the shaft has its “phase created” property set to a later phase.
DO use shafts if you have:
- 1. A new building
- 2. An existing opening
DO NOT use shafts if you have:
- 1. A new opening in an existing building.
- 2. A void can either be hosted on an element or on a plane. If it is hosted to the element, the edges will be normal to the cut surface. If it is hosted to a plane, the cut edges will be normal to the plane.
- 3. Voids obey “phase created/phase cemolished,” but you may have to do some work to get the edges to show up in plan.
- 4. I recommend creating standard families for rectangular and circular voids. If you need another shape, you can use multiple voids or create an in-place element.
- 5. You can build these into your void family. Parts obey phases, but they do not behave the same way as their parent elements. Use these with extreme care. In our discussion, they weren’t a popular option.
We’ll leave with a final recommendation: If you’re using Navisworks for coordination, running a clash detection between new elements and demolished ones can give you some useful information.
Kate Morrical is a digital design manager at the structural engineering firm Silman. She is responsible for coordinating procedures and standards for software used in the design process, whether CAD, Building Information Modeling (BIM), or design/analysis programs.