The Reality Capture Journey - Part 3

3, 2, 1, go! The scan data is ready, and the modeling team is fired up to tackle this monumental project. We are standing before a vast, blank canvas, but there is not enough BIM input to kick things off. When you need to set a project of this magnitude in motion without a BIM execution plan (BEP) or established standards, what should your initial steps look like? How can you get the ball rolling in the right direction to test the waters? I’ve got you covered!

In Part 3 of the ‘Reality Capture Journey’ series, I’ll walk you through our Scan to BIM strategy for a Central Utility Plant project, from a BIM and project management perspective. Get ready to learn about Revit model management, modeling strategies, and automation.

If you missed the first two articles on how we estimated and scanned this job, grab a coffee and settle in—they are worth the read. Now, let’s talk BIM stuff!

Square One: The KO meeting

As I emphasized in Part 1, when project guidelines are undefined, creativity and initiative become essential. A solid kickoff meeting (KO) will help you set the stage for clear expectations among stakeholders. By asking open-ended questions you can uncover everyone's goals: What are the key outcomes? How will the end product ultimately be used?

This is the perfect time to propose ideas, evaluate them collectively, and establish consensus. Wrap up the meeting with yes/no questions to solidify agreements. These decisions will become your baseline, and documenting them will be invaluable for managing risks in the long run.

When BIM projects is about, one risk worth addressing immediately is Revit coordinates. And having learned this the hard way in the past, we dedicated a substantial amount of time to getting them figured out. Let’s dive into the nitty-gritty!

Project Coordinates: Keep it simple

As you might have read several times in this series, information wasn’t abundant on this project. We didn’t count with a base coordinate system, level elevation input, or 3D geometry to align our scans to. With a grid system as our only guide, we embraced simplicity and set up our project as follows:

• Setting the Project Origin: We established grid intersection A1 on the Ground Floor as the model and project origin (0,0,0).
• Transferring the Grid System: We aligned the grid system (A1) to Revit’s Internal Origin, leaving the Survey and Project Base Points untouched.
• Defining the First Level: We set the Ground Floor Level at elevation 0.
• First Point Cloud Rotation: We rotated the first point cloud to align concrete column centerlines with grid intersections while keeping A1 as the origin. This allowed us to link the first point cloud file using the ‘Origin to Origin’ insertion method in Revit. Simple! ??

• Verification and Export: After rotating and exporting the first point cloud, we validated the alignment across four additional grids. The concrete columns throughout the building aligned with our grid intersections. We moved forward applying the rotation to the other areas and levels, and exported 25 RCS (point cloud) files.

• Additional Levels Setup: With all point clouds stacked in Revit, we set the remaining levels based on the average scanned floor elevations around A1.

• Cross-check: We created a Navisworks model to check that all files landed correctly. To do this, we modeled a few elements at different corners of the building, exported them in NWC format, and loaded the geometry and point clouds into Navisworks to verify alignment across all files.

A Few Things I Can’t Highlight Enough:

We ended up creating 21 Revit models. When it comes to setting up models for projects of this magnitude, there is no such thing as too much QC. As we start creating models downstream, every omission or mistake leads to exponential rework. Catching issues early is crucial.

Regardless of your project coordinate system, keeping your model geometry near the Revit Internal Origin should be non-negotiable. Revit does not play nice if your stuff is over 30,000 feet away, leading to export issues. Keep it close.

As you might imagine, this process required very close collaboration between the scanning and modeling teams, including countless calls to make this process smooth. Teamwork and communication were of essence.

Last but not least, thanks to incorporating survey control, we could rotate, align, and stack 25 point clouds with no major issues. We couldn’t have done this process without it!

Divide and Conquer: Zoning and Models Breakdown

As I often like to say, you can’t manage what you don’t know. For large, complex, and uncertain projects like this one, breaking things down into more manageable pieces gives you peace of mind. It provides a clearer understanding of the time, effort, and risks involved early on. Plus, it enables you to work more agilely, delivering progress, receiving feedback, and iterating as you go. Equally important, it helps you to prevent painful deviations of scope and schedule.

After defining high-level areas, and anticipating we’d have more than 20 Revit models, we broke the project down based on the following principles:

• Phase Delivery: To deliver the first package and receive feedback early.
• Limited Users: To avoid having more than 6 modelers per file.
• File Size Control: To keep small file sizes to prevent corrupt models.

Worth Discussing Before Breaking Down Projects:

In the spirit of avoiding rework, we discussed the following items with the end users:

• Architectural vs. Structural: With laser scanning data, sometimes it’s hard to distinguish architectural finishes from structural components like slabs and floors. Same happens with components behind furring, soffits, and ceilings. As much as we didn’t want to separate these elements into different models, we were required to.
• Mechanical Pipe vs. Plumbing: Separating mechanical pipes from plumbing pipes can be tricky, especially when they aren't labeled or easily distinguishable. We kept all of them together in each Mechanical model.
• Zone Boundaries: As you divide the project into zones, pipes and ducts might span across adjacent areas. To avoid splitting systems, we grouped all chiller zones together.

The Not-So-Fun Part: Standards

When standards is about, there is no such thing as too many questions. Reconciling model setup omissions and mistakes is really time-consuming (been there), especially with multiple files in play. Here are some items and details for you to think about:

• Naming Convention: From general to specific, the model name should describe its content. Here’s the structure we used: Model Author_Client_Project Name_Area_Discipline_Revit Version
• Content Naming: Establish a consistent naming convention for Revit families for everyone to follow.
• Units and Rounding: Confirm the preferred units and rounding early on.
• Revit Template Choice: Should you use the client’s template or your own?
• Worksets: Determine the appropriate level of granularity for worksets. Don’t break them down too much unnecessarily.
• Discipline-specific Content: Discuss and define what Revit categories should be modeled into each discipline file. Some could be ambiguous or dependent on client standards, for example:

a) Floors: We placed them in the structural model, instead of the architectural one.

b) Platforms, Catwalks, and Respective Railings: We located all of them in the structural model.

c) Diffusers, Plumbing, and Lighting Fixtures: We hosted them in the respective MEP model.

d) Equipment Pads: We placed them in the structural model.

e) Air Grills and Louvers: We located them in the architectural model.

The Fun Part: Modeling Automation

Well, I can talk about model automation forever, but this would be a mile-long article. While we applied several tools throughout the modeling process, this time I’ll focus on how we leveraged EdgeWise software to kick start the MEP modeling work.

If you’re unfamiliar with it, EdgeWise is a powerful tool designed to streamline 3D modeling from point clouds. Its advanced algorithms automatically identify and model elements like pipes, walls, ducts, and beams, significantly reducing manual input and speeding up timelines (thanks, ChatGPT!).

We’ve been using EdgeWise since 2017, and based on our experience, it truly shines when you have two things: a high-quality point cloud and exposed MEP systems. This project was an excellent case for its application.

Essentially, this is what our workflow looked like:

• Automated Extraction: We ran the extraction process, setting a minimum diameter of 2 inches for pipes and conduits to prevent false positives (such as railings).
• Manual Refinement: We complemented the automated workflow with a round of manual extraction to ensure we captured as much content as possible.
• Easy Connect Tool: We ran this feature to add fittings and connections.
• Geometry Clean Up: We cleaned up the results by converting the pipes and conduits to standard sizes to ensure consistency.
• Revit Export: We exported the resulting geometry into a new empty Revit model and ran our proprietary Dynamo scripts to replace EdgeWise fittings with our standard ones.
• Category Check: We used Dynamo scripts to convert pipes to conduits (and vice versa) as needed.
• Worksets Assignment: We divided the geometry into worksets by discipline, such as electrical, fire protection, or general piping, to facilitate the allocation in different models.
• Central Model Integration: Finally, we imported the geometry of each workset into each discipline-specific central model by using the Copy or Bind tools.

Once the clean geometry was imported into Revit, we worked by zones, conducting several rounds of manual refinement and quality check:

• First Pass: We added any fittings or connections the software missed, ensuring all lines were properly connected.
• Second Pass: We assigned pipes to the correct systems and added valves and accessories as needed.
• Final QC: Using CloudCompare software (more about this in the next section), we detected any missing lines not picked up by EdgeWise and added them manually.

Model Quality Check - Automated and Manual QC

Before outlining our QC process, let me emphasize one thing first: You should perform periodical checks (quality assurance) during the entire modeling phase to catch issues early. Even simple actions, like opening your models every other day to look for ‘the elephant in the room’, can save time and prevent miscommunication. Frequent client touchpoints and team regroup meetings also help ensure you’re modeling what is required, nothing more, nothing less.

Automated QC:

Regardless of how experienced your team is, human error is inevitable. At the end of the modeling process for each phase, we performed a thorough quality check, starting with automated QC, to catch 80% of the issues. The automated QC helped us identify:

• Omissions: Missing geometry.
• Misalignments: Elements modeled outside acceptable tolerances.

To do so, we utilized Cloud Compare. This is an open-source and very robust tool widely used in the geospatial industry because of its proficiency at handling large 3D point clouds. Here is how we applied it:

1) We exported the Revit geometry as STL file.

2) Loaded the point cloud files into CloudCompare in E57 format.

3) Set up our scalar field with the required tolerance.

4) Ran the analysis to obtain results. The output was colored according to the scalar field based on deviations and omissions.

5) Exported the output (omissions and misalignment results) as RCS files.

6) Imported the results in Revit for resolution.

For omissions QC, only points corresponding to missing geometry were exported, making it easier for modelers to locate and add missing elements in Revit. For tolerance QC, only points representing misaligned elements were exported for adjustment.

Manual QC:

There is no substitute for a thorough visual inspection to detect model inconsistencies. We conducted visual walkthroughs in Navisworks software, leveraging Navi’s appearance profiler to identify the misuse of categories, geometry overlap, and other issues that CloudCompare couldn’t detect.

Finally, and before cleaning up the model, we checked:

• Content Naming
• Geometry Reference Levels
• Worksets usage.
• Phases.

Final Thoughts

The lack of information and specifications is the norm in Reality Capture projects, and that’s exactly what we are hired for, isn’t it? While precision is often the primary goal for scan to BIM QC processes, quality isn’t solely about accuracy. True quality lies in meeting the project requirements and expectations set with your stakeholders. Always measure your work against your contract to ensure you deliver what you have committed to. And speaking about contracts, every learned lesson should shape your future documents and standards, paving the way for more predictable and efficient work.

I hope this article provided you with some actionable insights for your projects. Stay tuned for the final article in this series, where I’ll explore BIM for FM and digital twinning.