Automation, Controls

How virtual design, artificial intelligence impact engineers

Architecture, engineering and construction professionals will be impacted by virtual design and artificial intelligence tools

By Karen Pierce October 23, 2020
Courtesy: UNIFI Labs

 

Learning Objectives

  • Learn how artificial intelligence and virtual design overcome limitations in BIM.
  • Understand how these innovations impact existing workflows.
  • Identify the skill sets required to effectively use these emerging technologies.

For decades, the architecture, engineering and construction industry has developed software and hardware tools that enhance efficiencies in designing, documenting and planning throughout the building phase. In the mid-1980s, computer-aided design programs began to take over the traditional hand-drafting efforts in the workplace. Nevertheless, there was a problem with these approaches; it lacked nonphysical data and collaborative tools. Various stakeholders working on a project could not readily identify building conflicts or troubleshoot concerns in the specifications and certainly not in real time.

Then building information modeling emerged in the early 2000s. This application was not just a glorified CAD product. Instead, it layered nonphysical data alongside the physical, providing a unified design schematic that brought together design, 3D visualization, building specification and project documentation.

BIM also allowed everyone involved in the design, construction and implementation phase of a project to work together. It put engineers, contractors, clients and architects on the same page, allowing better collaboration.

The industry is now amid an upheaval in workflows and skillsets as virtual design and artificial intelligence gain momentum. These latest trends have been gaining popularity over the past few years. Independently, both technologies have the potential to dramatically change AEC firms’ operations by enhancing their design capabilities during the design phase.

On the other hand, combining these two innovations can overcome the remaining limitations of both BIM and CAD and usher in a new age for the industry. Another point of discussion is how the short- and long-term effects of COVID-19 could increase the use and development of these tools.

Figure 1: This kitchen rendering was created in ArchiCAD with Cinerender. Courtesy: UNIFI Labs

Figure 1: This kitchen rendering was created in ArchiCAD with Cinerender. Courtesy: UNIFI Labs

The challenge with BIM

Although the 3D efforts of the BIM process put forth an extraordinary effort of bringing together the broad spectrum of interest groups involved in a typical construction project, it is not perfect. As buildings become more complex, stakeholders become more involved and with a substantial increase in client requirements, it has become increasingly difficult for AEC firms to fulfill these new deliverables using standard toolsets.

Additionally, while a better specification and process might be out there, accessing it is not always practical. It has become increasingly important for firms to understand how virtual design and artificial intelligence might influence their current operations. This process may also include understanding what products or toolsets are missing and how they can take incorporate them.

BIM also requires professional expertise at all points in the design phase. Before the past few years, there had been no tool to set the design specifications and then let the software discern the optimal solution. However, this is starting to change with generative design programs such as Dynamo.

Although BIM initially was a one-stop-shop for whole building modeling, this conflicts with realities of delivery of building design and communication. BIM application workflows for architects diverges from BIM workflows for engineers and contractors. Getting the two to communicate effectively can continue to be challenging due to working in their different paradigms.

Some firms try to get around this problem by increasing their software features, which often leads to diminishing returns. Somebody still has to consider all the information manually, which could lengthen the entire project schedule to an unacceptable manner

Finally, for all its bells and whistles, there is one consideration that BIM cannot accommodate: the rationale behind the project itself. On physical and technical levels, BIM is sufficient at telling users whether a specification is internally consistent. However, it does nothing to ensure that designers created a specification that meets the needs of clients.

This criterion is outside of its scope. The current software has no idea why it is doing what it is doing. It does not ask tough questions of designers and it does not prompt them to refocus their efforts on solutions that would benefit the end-user. All that rationalization must occur concurrently in the build phase, external to the BIM environment. That is changing.

Figure 2: This image is using Dynamo in Revit to review options for desk layout in an office to maximize desk area per person and minimize unused space. Courtesy: UNIFI Labs

Figure 2: This image is using Dynamo in Revit to review options for desk layout in an office to maximize desk area per person and minimize unused space. Courtesy: UNIFI Labs

The promise of virtual design

The AEC industry is continuously evolving and it has recognized the limitations of the traditional BIM process and, collectively, is seeking needed improvements — some organizational and some technological.

Between 1970 and 1990, virtual reality software and hardware were in numerous industries. It mitigated risk for training the military; it advanced medical practice and even affected automobile designers. After the 1990s, virtual reality software became even more widespread as the gaming industry took off.

On the surface, virtual design does not add a considerable effort to the design process of BIM. The cynic could argue that there is only a change in the user interface. Rather than reviewing designs on screen or paper, designers and stakeholders now step inside virtual versions and inspect them from the inside out. Therefore, at best, it might improve productivity by a few percent.

Enthusiasts, however, view the technology differently. They contend the real promise of virtual design, they point out, is how it reshapes creativity. Even the most astute architects can struggle to appreciate the aesthetic merit of their creations looking at aerial plans alone. Architects and engineers become immersed in the action and they can do more than guess at the subjective nature of the experience.

As architects and engineers navigate a post-COVID workplace, the earlier Dynamo example of maximizing desk space and minimizing unused space changes to maximizing desk space and maintaining social distancing guidelines.

Virtual reality takes some of the ambiguity out of the project details. With augmented reality and virtual design, designers and stakeholders alike can get a feel for what it will be like to look out of the building from the inside — something that was not possible before. Software platforms like Revizto create explorable, true to scale virtual reality experiences within seconds. Designers can physically experience the model. They can visualize the construction project seamlessly as well as do remote walks of the site, improving troubleshooting. Designers can don their virtual reality goggles and look at a room to determine whether it makes sense. It is easy to miss noticeable flaws when looking at an image on the screen, but these flaws can become extraordinarily apparent when standing in a virtual 3D environment.

Figure 3: A Hypar workflow shows several generative design options for the building design. The current service core option is displayed in context with alternative options available for selection. Courtesy: Hypar.io

Figure 3: A Hypar workflow shows several generative design options for the building design. The current service core option is displayed in context with alternative options available for selection. Courtesy: Hypar.io

The promise of artificial intelligence

Leading AEC firms are contemplating how to implement machine learning technology to improve BIM standard practice and solve many longstanding problems with the approach.

Hundreds of papers exist on the premise of artificial intelligence for a wide range of industries. In April 2019, Google released TensorFlow, a powerful and open-source machine learning platform. Users with a background in deep learning algorithms, a computer and an internet connection, now have access to a platform that could compete directly with Google itself.

Before the COVID-19 pandemic, firms like EvolveLab and Smartvid.io have specialized in software platforms that harness the algorithms behind artificial intelligence. As the United States begins to reopen from the quarantine lockdown, it is clear on every construction site that safety protocols are not the same as they were. Smartvid.io uses artificial intelligence to measure compliance with proper personal protective equipment and social distancing automatically, and Smartvid.io can also analyze photos from just about any project management platform to create a virtual walk-through to detect additional safety concerns.

On the design side, many firms are not applying their project data or even lessons learned from past projects to current projects. They are not applying their project data because making effective use of the mountains of data generated by construction projects is challenging. There is value in this, but extracting it is either difficult or impossible for most firms. Firms begin from nothing every time, continuing to explore only the most successful experimental learnings from earlier projects.

With artificial intelligence, however that could all change. Artificial intelligence combined with BIM data could allow AEC firms to go far beyond rational analysis and at a much lower cost. Engineers could connect artificial intelligence tools to their databases, feed them data and then extract new insights.

Artificial intelligence may also enter the BIM market in another way: by allowing specialists to explore the space of construction more rapidly. Working out whether a design is possible requires the manually intensive endeavor of physically going in and creating it. With machine learning, it is possible to effectively “evolve” a design that meets specifications. Architects and building contractors could set the parameters and then tell the computer to continually try new combinations until it produces something that fulfills the client’s criteria. Design specialists then use their judgment to review the machines’ design outputs and decide which designs take the next step.

Artificial intelligence may also be able to help at the data classification level. Inspecting each element of a BIM model and then assigning it to a category is a tedious task. AEC professionals spend an enormous amount of their time, inputting specific details to ensure that BIM projects remain robust and collaborative. However, machine learning could easily take over this process by looking for commonalities among elements, scanning each data point in BIM design and assigning a category. If the machine cannot figure it out, the machine can ask a human to categorize and store the added information in its databanks.

Figure 4: This infographic highlights this generative design process. Courtesy: UNIFI Labs

Figure 4: This infographic highlights this generative design process. Courtesy: UNIFI Labs

Evolving AEC workflows

Throughout the design and engineering process, AEC firms collect data from a broad array of sources such as design criteria, simulation data, structural analyses and similar. Historically, this information was stored in a repository and remained unused and unloved. With the advent of machine learning, it is now possible to use it for many project-related purposes. These include:

  • Generative design.
  • Managing risk.
  • Denoising rendering.
  • Budgeting.

For example, rendering a true-to-life model of a project on the computer once took hours to achieve. However, when engineers added denoising algorithms to the graphics processors, it significantly reduced the raw data necessary to produce a viable visual output. Artificial intelligence could allow designers to make changes much faster than in the past, allowing them to modify building aesthetics in near real-time.

We will see something similar happening in the realm of generative design as well. Workflows will change from the current BIM paradigm. Teams of designers and engineers will no longer need to manipulate the design structures manually and continually check their viability. Instead, a machine learning algorithm will use mountains of data to simulate a series of design experiments subject to user-provided constraints. It will then develop a set of solutions, trialing different approaches until it finds one that satisfies the user’s specifications.

Because of the fluidity of artificial intelligence, workflows may wind up looking quite different. Designers and architects will no longer be down in the trenches, dealing with the details, such as whether a wall can support a load. The artificial intelligence will do that. The engineers’ role will be setting the constraints, specifying the parameters and evaluating whether the arrived-upon designs fit the client brief.

The way companies go about budgeting will change dramatically too. Cost flow information already forms part of the accessible datasets. Artificial intelligence systems could potentially calculate the costs of real-time projects by incorporating learnings from prior builds. As projects progress, engineers and designers could add new elements to the BIM models and then determine how it affects their artificial intelligence-generated budget.

In a sense, therefore, both virtual design and artificial intelligence force engineers, architects and builders to take a step back and ask the more important questions. These innovations will protect them from becoming bogged down in the details and focus on the clients they serve.

Figure 5: A Hypar workflow showing a generated building design in a referenced urban context. Additional generated façade options are available to replace the current façade design. Courtesy: Hypar.io

Figure 5: A Hypar workflow showing a generated building design in a referenced urban context. Additional generated façade options are available to replace the current façade design. Courtesy: Hypar.io

Skill sets required

Both virtual reality and artificial intelligence are valuable innovations for the AEC industry. However, unless firms learn the necessary skills, they will not be able to leverage them and rollout will be as slow as it was for BIM.

Data and artificial intelligence skills AEC firms will always need personnel who understand how to use and implement their data and artificial intelligence resources effectively. Still, collecting, managing and interpreting data requires human interaction. artificial intelligence will also not tell companies how or when to deploy the software, share best practices or lessons learned. They will still need people for executive decision-making.

Data and artificial intelligence will usher in a new paradigm for the sector. Rather than a single design possibility (the result of a traditional collaborative BIM endeavor), generative artificial intelligence will produce a menu of options. Firms will need people with the skills to evaluate each and select the best to support the desired outcome.

Companies will also need professionals who understand how to label data and link it to new artificial intelligence tools. Practically all data firms collect throughout their operations is valuable but putting it to effective use is a challenge.

Cost and scheduling skills — Under traditional BIM workflows, there is a simple analysis regarding the timeline and cost at which projects achieve completion. However, when using artificial intelligence, this will no longer be the case, as superintendents will have the option of weighing different considerations against each other.

Do they want to spend more money and complete the project faster? Do they want to achieve a higher-quality design, but dedicate more time to the project? What about a project that delivers the smoothest crew cadence at the lowest cost? Thus, when AEC firms introduce artificial intelligence to the design process, they wind up requiring professionals who can make executive decisions according to entirely new schemas that will be need to be addressed.

Management and planning skills Management and planning have always been critical in the AEC sector, but these innovative technologies will change the game. artificial intelligence-optimized workflows will make the task of site management more data-driven. Those in charge will see a shift from the day-to-day practicality of keeping the wheels of the project, moving to plan how things will unfold ahead of time. Managers need to become comfortable with the idea that they must create plans that satisfy multiple criteria relating to time, cost and crew availability.

Artificial intelligence, virtual design and AEC firms

Whether artificial intelligence and virtual reality will solve all the problems inherent in BIM remains to be seen. Human error will remain an obstacle for some time, but the stumbling block’s nature is going to change. No longer will it be incongruities in BIM that hold up projects. Instead, it will revolve more around the misapprehension of client needs. artificial intelligence will take care of a lot of the grunt work that goes into the design, but there will still be a need for human decision-making at an elevated level. That is never going to go away.

AEC companies that can position themselves to take advantage of these technologies will gain a competitive advantage. artificial intelligence will enable them to assess clients’ needs better and ensure that they meet them, regardless of what software tools they use. That same technology will revolutionize the design process itself. Professionals will find themselves engaged in radically different workflows that require entirely new skill sets.

How AEC firms operate, therefore, is going to change significantly in the future. Development time will decrease and the quality of service will improve. As virtual design and artificial intelligence workflows evolve, it will lead to less worrying about the details and more focusing on the big picture.


Karen Pierce
Author Bio: Karen Pierce is a senior BIM specialist at UNIFI Labs. She has 10 years of experience in the AEC industry and has led the charge on expanding BIM technology in various roles.