A day using 4th Gen AR AI-ML in Construction - Use Case A (first steps): BIM is more than 3D's

Today, the BIM methodology approach and BIM Management in general have a natural tendency to optimize 3D development and interaction. In examples of Augmented Reality (AR) application in BIM, three evolutionary stages can be observed to date:

First Generation: Overlay through mobile devices.

Second Generation: Overlay and geometric interaction (e.g., zoom-in/zoom-out) with 3D in AR.

Third Generation: Overlay, geometric interaction, and input/output of dynamic data in AR.

Let us go further and discuss the future Fourth Generation, where BIM Analytics comes into play, which can be integrated into an augmented reality (AR) environment. Here we must clarify that the term "analytics" is often misused and confused with reporting or updating. Thus, correctly understanding what analytics is, it is necessary to add the following capabilities to BIM models to achieve this quantum leap:

Statistical Modeling

Big Data

We present here a future use case - representing the first steps in this technological saga - with the introduction of what we term "deep assessment," a methodology oriented towards generating actionable intelligence for enhanced decision-making in construction projects. We envision this proposal serving as a catalyst for innovation, inspiring practitioners and researchers alike to explore new frontiers in construction analytics and data-driven project management.

Leah receives a call from Mark the project manager. “Apologies for the late call, but I need your help. It shouldn’t take long if the system we are testing is fully operational.”

Leah has just arrived at the office on a hot and sunny day at The Line—an ambitious construction project in the United Arab Emirates, where the first of many skyscrapers exceeding the height of the Burj Khalifa is under development. Leah is responsible for testing the completion deep-assessment software, a critical component of the project’s digital infrastructure.

The software integrates with BIM models to generate state sets for each BIM element. These state sets are continuously synchronized with a dedicated analytics cloud, where the data is processed through a statistical model designed to provide highly accurate completion assessments. This tool represents a shift from traditional progress tracking toward precision-driven assessments, offering unparalleled insights into project completion.

“Hi, Mark. No worries, I had planned to stay on site for one more hour. What can I do for you?” Leah replied.

Mark responded, “Look, today I was walking through the site, the third tower zone, with one of the investor’s technical supervisors. He was updating progress information using the 360o view device attached to his hard hat.”

“Did he find anything wrong?” Leah asked.

“I’m not sure,” Mark said. “As we walked, he synchronized his 360o view device with his iPad, displaying a double view—one taken three weeks ago and the other showing the current site conditions.”

“I see what you mean,” Leah replied thoughtfully. “It’s similar to our 360o view device, except for that cool site synchronicity feature you mentioned. But it seems like something’s on your mind. Tell me more.” She opened her laptop as she spoke.

“To be honest, he flagged several elements as ‘hot,’ indicating zones that are behind schedule—specifically the northern foundation area, which was recently modified.” Mark said, his tone reflecting genuine concern.

“I understand,” Leah replied. “Yes, we are a bit behind schedule… Let’s check. Can you open the cloud-synced BIM model? Navigate to this area.”

Mark opened the model and ran the deep-assessment application. "It shows elements in yellow, and the view from three weeks ago shows the same elements in yellow. What does that mean?" he asked.

Leah zoomed in on the area and explained, “These elements remain yellow because they weren’t completed.”

Mark frowned and replied, “That’s what concerns me. According to the last project control report, this zone was at 60% progress, while three weeks ago it was at 45%. I specifically ordered a progress curve for this area. I thought there would be significant differences compared to three weeks ago, but visually, the foundation’s status hasn’t changed. I’m not sure if the information in the progress curve is accurate.”

Leah clicked on the dashboard and sent a link to Mark. The dashboard displayed both the lower and upper limits of the completion curves, surrounding the progress curve. The lower curve indicated a deceleration pattern, and the complementary curves, plotted using the Markov Chain Analysis (MCA) model, revealed an intriguing trend.

“Look at the link I sent you,” Leah said.

“I see it,” Mark confirmed.

The MCA-projected curves showed no change in the estimated completion date, likely because this area is under a float scheme. However, the projected curves for early completion had shifted by five days, indicating a delay. “See the complementary MCA curves. This means the contractual completion date is still achievable, but finishing earlier has become more difficult,” Leah noted.

The MCA model analyzes the variations in three distinct completion states: not started, Work-In-Process (WIP), and completed. In the BIM model, these states are represented using white, yellow, and green, respectively.

As each BIM element (e.g., column, foundation, wall) either changes state or remains in its current state, it is only a matter of time before all elements leap to the completed state (green). The MCA assesses these transitions and estimates the week when all elements will turn green, based on a probabilistic matrix. The probability data based on state leaps used in this matrix is derived from the two previous towers, which are above 50% progress. The third tower has several key aggressive milestones, as the learning curve from the earlier towers is expected to be applied to its construction, making these milestones supposedly achievable.

Additionally, the MCA model incorporates a special matrix in which each leap has a 0.5 probability of occurring, representing a highly improbable scenario. The primary probabilistic matrix generates a realistic completion date by assuming deterministic and mutually exclusive states, while the secondary matrix provides an optimistic—but unlikely—early completion date.

“Mark, do you remember what I explained about the completion curves and the two projections generated by the MCA model?”

“Yes, I do. It was a kind of new progress curve,” he replied.

“Not exactly. Remember, progress is not the same as completion—progress includes WIP elements, but completion accounts only for finished elements. There’s no room for WIP in the completion curve,” Leah clarified.

Mark nodded as he recalled, “Ah, yes! I remember it was mentioned by our PMO Risk Mgr Omar Samaniego during his training on how to avoid the 90% Syndrome by using completion curves instead of relying solely on the progress curve.”

“Exactly, Mark. Now, if you click the status button on the model, you’ll see the two completion dates estimated by the MCA model.”

“I see them, Leah. The overall completion date hasn’t changed, but the early completion date has shifted. By the way, I know you were updating the site status today. Was everything reflected in this estimate?”

“Yes, of course,” Leah replied. “Every state change I made in the BIM model, based on what I observed at the site, was synced to the cloud database and reflected in the dashboard. I ran the MCA model just 15 minutes before you called me, and these are the results.”

“Well, it seems the final completion date is not at risk—we still have nearly two years until the contractual milestone. Having this diagnostic at such a granular level, yet conveniently dosed, really helps. But what should I tell the investor’s supervisor about those elements that remain unchanged—still in yellow?” Mark asked.

Leah responded thoughtfully, “First, you need to remember that while the progress curve is useful for tracking financial investment in the project, it tells us very little about granular, daily issues. For instance, even though additional excavation was done and some pre-foundation concrete was poured, the BIM model keeps those elements in yellow because they represent WIP, not completed deliverables. There is not foundation completed yet, and this is fact. In traditional progress control, the quantities achieved are tracked, but they don’t reflect completed elements. Take the excavation and soil stabilization for the large foundation, for example—it added to the progress curve but still left the foundation in a WIP state. Now, according to the concrete casting plan, it was decided to wait until at least three foundations were ready before concrete casting. In this case, it makes sense and is something you can explain to the investor’s supervisor. But remember, Mark—'eating the elephant one bite at a time' may cost a bit more, but it reduces the risks during execution. And, by the way, there are other areas where the delays aren’t justified. I’ll brief you on those tomorrow.”

Mark smiled and replied with relief: "Well, so don't worry about tomorrow, because tomorrow will bring its own concerns. Each day has enough trouble of its own, which doesn't mean we shouldn't be prepared. Meanwhile, we'll help ourselves with the software!

?“Yes! See you tomorrow, Mark. By the way, Tomorrow, Ishmael from ARBIM will visit the site to showcase a new version of the deep-assessment app—a leaps-delay detector. He explained that it uses blinking yellow color for WIP elements that show no variation within a predefined time period. This feature will enable the team to visually and quantitatively identify and assess the trends and risks associated with these delays. Also, I heard that a new AR version of the deep-assessment system is in development.”

“Good to know! See you tomorrow! Leah”