Generating BIM Under Pressure

Laser scanning and automated feature extraction dramatically speed the process of 3D building information modeling for a $125 million hospital renovation in Ohio.

By Kevin P. Corbley

Bethesda Hospital in Zanesville, Ohio, will undergo a major renovation and expansion in 2013. Architectural and engineering design work is already underway. The three-year construction plans call for upgrading medical offices, an outpatient cancer center, and 200 patient rooms. The new addition will contain an enlarged intensive care unit, operating rooms, and extra in-patient suites.

Built in 1963, the 350,000-square-foot facility has already seen its share of additions and upgrades with multiple wings built onto the original structure. With each expansion, new or enhanced mechanical systems have been installed to meet the growing requirements for air conditioning, heat, water, and electricity. As a result, the mechanical rooms—five in the basement and two upstairs—are jammed with pumps, valves, conduits, and pipe runs. “In the past 50 years, there have been so many changes to those rooms that the original design drawings are useless to us,” said Mark Hanna, founder and president of PrecisionPoint Inc., a provider of 3D laser scanning solutions based in Carmel, Indiana.

The architecture and engineering firms hired to design the expansion contracted PrecisionPoint to create highly accurate digital building information models (BIMs) for the entire facility using terrestrial laser scanners. Of the hundreds of corridors and rooms in the hospital that had to be scanned and modeled in three dimensions, the mechanical rooms presented the most formidable challenge to the six-week schedule due to their size, complexity, and importance.

At 5,000 square feet each, the mechanical rooms were among the largest in the facility. But unlike most of the patient rooms and medical offices whose floors, ceilings, and walls were the only features that had to be captured in the 3D BIM, those seven rooms required more detailed modeling because they are packed with critical mechanical, electrical, and plumbing (MEP) equipment.

In a large building or plant renovation like this one, Hanna explained, individual MEP features must be modeled accurately in three dimensions so new pipes and conduits can be designed for clean installation in the void space. This requires extraction of the MEP elements, which means pipe runs must be located precisely in space with their size, sag, and turn angles carefully measured and recorded in tables. This extracted information enables the designer working in the CAD environment to see and query MEP objects and their specifications to ensure there will be no interference between old and new equipment during installation.  

Scanning the Interior

In preparation for scanning the interior of the hospital, PrecisionPoint established a survey control network that would be used to register the inside point clouds with the same coordinate system as the exterior site survey. Another firm that had surveyed the hospital grounds provided control point coordinates to PrecisionPoint, which then used GPS and total stations to run a survey network starting outside the facility and extending down the inside hallways on each floor.

A series of targets was set up outside and inside the hospital as tie points for the surveying and scanning. The outside targets were six-inch spheres on bipods, while the majority of inside targets were paper, taped to the walls of halls and rooms. These were adorned with checkerboard patterns and unique numerical identifiers. During the survey, coordinates were captured for the center of each hallway target, which remained in place for the scanning.

Additional scan targets were set up in all the inside rooms. Those rooms with odd configurations or lots of equipment that occluded line of sight—like the mechanical rooms—were plastered with multiple targets. The objective was to ensure that three or more targets would be captured in every scan so all of them could be registered and later integrated into a single 3D model tied to the common coordinate system.

Next, PrecisionPoint brought in two Focus3D laser scanners from FARO Technologies Inc. to scan the interiors. Mounted on tripods, the devices were placed at a stationary point in the room to perform each scan. For simple rectangular or square hallways and rooms, the high-speed lasers captured a single 360-degree scan, including the ceilings, in three to four minutes. As expected, however, multiple scans were required for the complex mechanical rooms.

“We had very short lines of sight because there was so much equipment in the way … and we had to do very dense scanning to avoid having shadows or occlusions in the point clouds,” said Hanna. “Our technicians had to move the scanners to many different positions to capture all of the piping nested in the ceilings and hidden behind other equipment.”

Over a period of three days, they took at least 15 scans in each of the mechanical rooms. In addition to the pipes and ducts, each room contained large MEP equipment such as water pumps, compressed air tanks, air conditioning units, and fire suppression apparatuses.

Extracting the Pipes

Once the interior scanning was completed, the next phases of the modeling project were carried out on the desktop computers at PrecisionPoint’s Indiana headquarters. There, the technicians used FARO Scene software, which came with the scanners, to register the point cloud scans. This meant that individual scans for each room were overlaid on the survey control network using the easily visible targets as tie points so the scans would be registered to the coordinate system being used for the overall hospital design project.

From the Scene software, the point clouds were exported to the EdgeWise Plant software from ClearEdge3D. Based on extraction algorithm technology originally developed for use in mapping from airborne lidar data sets, this software had been modified specifically for extraction of pipes and related equipment in industrial plants, refineries, and MEP rooms. Its primary advantage is speed, allowing BIM modelers to extract pipe runs up to four times faster than manual methods.

Devon Kelley, PrecisionPoint’s BIM specialist, set a few parameters in the software, most importantly asking it to extract all pipes of two inches in diameter or greater. Anything smaller would be ignored. Proceeding one room at a time, Kelley selected the scans with the highest density of pipe runs in them. Since the software analyzed one scan at a time, he wanted it to capture as many pipes as possible from the first several scans.

The EdgeWise Plant software quickly identified pipes in the scans and measured their size, sag, and bends, recording the data in an attribute table. An enormous timesaver was the software’s ability to accurately estimate the radius of a pipe elbow, a common source of error in manual extraction. In addition, the software had built-in intuition that enabled it to trace a continuous pipe even if it temporarily disappeared from view in the scan as it ran behind a piece of equipment.

With about 90% of the pipes identified and measured automatically, Kelley took advantage of a manual editing function to clean up the pipe models. This involved aligning pipe runs that were occluded in one scan and then re-appeared in another. He also used the clean-up mode to double check the bend radii and compensate for pipe diameter measurements that were thrown off by insulation casings.

“The automated extraction was accurate and helped us get through the pipe modeling more quickly,” said Kelley, adding that the automated measurements proved accurate to the specifications of the designers. Without the software, he would have had to delineate, measure, and trace each pipe run manually in the point-cloud processing package. This would have added weeks to the modeling phase of the project.

Building the BIM

Once the extraction was completed, Kelley exported the point clouds and pipe models with their associated tables directly to AutoCAD for additional clean-up. From there, the data was imported into the Autodesk Revit software for final BIM modeling. The 3D locations of the pipes were represented in this application as centerlines.

Using the native Revit MEP pipe modeling tool, Kelley re-built each individual pipe run in its precise size and configuration, creating a solid 3D BIM of the mechanical room.  “You just choose the diameter and snap onto the centerline” to create a solid 3D model of the pipe, explained Kelley. (This summer ClearEdge3D will release a plug-in that will allow users to bring the EdgeWise solids and centerlines directly into Revit.)

He was then able to add attribute information to the pipe layers by reading identification information on the pipes that was visible in the point cloud. This allowed him to differentiate water pipes from compressed air lines, for example. He also modeled equipment such as air conditioners and pumps so that pipe runs could be properly aligned and attached to their sources and terminations. PrecisionPoint delivered BIMs for the entire hospital structure to the architecture and engineering firms by the specified delivery date just six weeks after scanning the facility.

Kevin P. Corbley is a business consultant based in Denver, Colorado, and can be reached at

» Back to our August 2012 Issue