Mining the Cloud for a Better Building Industry
Figure 1: Project Fractal automates the generation of thousands of design options by driving input changes through uploaded Autodesk Dynamo Studio graphs
By attending several conferences this year I’ve had the opportunity to hear some of the most advanced thinking around the intersection of computation and architecture, engineering, construction, and operations. Around the world I’ve consistently heard declarations that it isn’t the model that’s important, it’s the project data, as though the building and infrastructure industries have undergone a profound change. In response, I always ask, “When wasn’t the data important?”
The Great Pyramid of Egypt deviates from an exactly square base by 0.07 percent. I would argue that this level of precision, with what we would consider primitive tools, arises from a concentration on the importance of data when making decisions about buildings and infrastructure. Other examples from the ancient world include the accuracy of celestial observation implicit in the layout of Stonehenge and Roman Empire aqueducts that serve Rome effectively to this day. None of the outcomes served in these examples would be possible without a concentration on critical data at the center of the project, be it celestial mechanics or gravity hydraulics.
So, if data has always been at the center of the built environment, why the current rhetorical focus on the idea of data centricity in multiple presentations, keynote addresses, and classes at events in every major region of the world? I believe the thought being conveyed in every one of these venues is the realization that supporting technology has progressed past the point where we need to wholly concentrate on producing conventional instruments of service, as consumptive of time as that activity remains. As a class of professions concentrating on the built environment, we have an opportunity to leverage and repurpose the data those instruments have explicitly and implicitly recorded in the design and project delivery process.
Again, this was always the case. Those responsible for managing and maintaining facilities have employed design and construction drawings of various degrees of accuracy for centuries, and yet there is an intuitive sense observable in the AECO industries that something has changed. That change is not merely the increased accessibility of data implicitly recorded for creating instruments of service, but also the increasing capability in the industry to automate the production and delivery of information formerly consigned to human labor.
CAD programs provided some automation of drafting tasks, but no explicit support for producing better outcomes for a project. BIM has helped the industry progress farther, at first supporting a degree of understanding of a proposed building’s behavior through accurate visual renderings. Step by step, as the data implicitly recorded in models has been leveraged into additional analyses and simulations, we have begun to derive meaning from what were formerly highly abstracted representations of the built environment encoded in ways accessible only to those with specialized training. What we are beginning to see is not only the democratization of information, but of intent.
However defined and prioritized, effective outcomes have always been the intent of every building project, but the industry has lacked the readily available means to consistently and accurately deliver on the intent of its projects, as opposed to the mere physical instantiation that may or may not serve the project’s goals. Due to the increasing use of scalable cloud computing, wide availability of relevant data, and worldwide connectivity between professionals, the AECO industries now have a unique opportunity to move beyond instruments of service, and even physical artifacts, to universally serve the intent of projects through these means.
Up to the present day, the impediments to consistently and accurately serving the constituent needs of building projects have been myriad and nearly insurmountable. The shortage of relevant expertise, delivered only through the direct consultation of too few highly experienced professionals, has too often resulted in suboptimal outcomes in which the goals of a project have been only partially served or entirely obscured in the completed project. As a group of building professions, we have attempted to address such problems through specialization in specific building types and building systems, and we hope that the most experienced people we place in critical project positions will be in the right place at the right time to ensure proper service to the project’s intent.
Hope, unfortunately, is not a plan.
Figure 2: Project Discover used scalable computation and genetic algorithm optimization techniques to generate and evaluate thousands of options for Autodesk’s new Toronto office
While such an approach is anecdotally effective, as evidenced by many successful projects, the outcomes for the building industries remain dismal. It need hardly be enumerated that the AEC industry’s cost overruns, missed schedules, and suboptimal results are endemic and nearly universally challenging on almost every project. It is commonly understood and still widely tolerated as inescapable that 25 to 50 percent of all labor and materials on every project will be wasted, the type of process issue that has long ago been optimized out of existence in the manufacturing industry. Simply put, manufacturers who could not equal or surpass the efficiencies and quality controls of their competitors went out of business long ago.
Digital cloud technologies now afford the AECO industries an opportunity to become both more consistently efficient, and more importantly, more consistently effective on every project, regardless of the assigned professionals. Building expertise has traditionally been difficult to acquire, maintain, and extend. Professional licensing requirements in most jurisdictions include continuing education, but the intersection of such fulfillments with timely project needs is chancy at best. Likewise, expertise conveyed through conferences may or may not arrive at the right moment to increase project quality or deliverable productivity. Expert consultation is likewise subject to uncertain availability and outright shortages on projects.
The only consistently reliable way to deliver relevant data and expertise is through software environments capable of supporting AECO needs, either by connecting professionals to relevant information or directly encoding the expertise in the form of analytical, simulation, computation, data management, and machine learning services.
Delivering such capabilities in the past faced limitations of computation and memory access inherent in desktop environments. By incremental steps leveraging scalable cloud computation and data orchestration—from realistic cloud rendering through energy analysis to BIM collaboration technologies—Autodesk has sought to increase the timely application of relevant expertise to building problems by leveraging the cloud.
Regardless of the services on offer, cloud technologies provide two fundamental advantages: connectivity and scalable computation. These basic capabilities can be shaped and applied in thousands of ways to support individual workflows. Where Autodesk’s portfolio has encompassed a vast array of needs in the AECO industry, the availability of API access to product features was always an implicit recognition that no commercial software developer can anticipate every need and use of their products, especially in industries responsible for more than $7 trillion of economic activity annually worldwide.
As Autodesk transitions and creates new capabilities in the cloud, we’re looking to leverage existing technologies in more granular ways, exposing valuable capabilities as cloud services that can be orchestrated by partners and customers for unique needs unsupported by initial commercial offerings.
Several current services on Autodesk Forge, our cloud development platform, started life as technologies necessary to support BIM 360 capabilities. As we expand our commitment to the opportunities afforded by the cloud’s connectivity and scalable computation, we’ll be looking to open new services to wide use, not just by partner developers, but directly by building professionals.
Several current development investments at Autodesk illustrate the strategy.
Project Fractal, available for experimentation today, uses scalable cloud computation to produce thousands of options by driving an uploaded Autodesk Dynamo Studio graph, extending a desktop visual programming environment with a service aiding the exploration of captured project intent.
Project Discover, employed in the design of Autodesk’s new Toronto office, uses scalable algorithmic optimization to generate thousands of options balancing multiple project goals. Project Quantum is designed as an ecosystem of desktop and web software environments, cloud services, and data repositories that capture and inform project expertise by supporting arbitrary connectivity between multiple capabilities in orchestrated workflows.
BIM 360 Insight, currently in development in conjunction with six major construction companies, applies machine learning to thousands of their collective recorded BIM 360 issues to alert their project managers of elevated risk to current projects, with the system becoming more accurate and nuanced the more it’s used.
None of these capabilities could be fully realized without employing the connectivity and scalable computation of the cloud, and all of them begin to address the application of specific recorded and automated expertise to building problems to help raise the quality of the resulting built environment. As automation gains ground in industry after industry, raising productivity, reducing risk, and increasing quality, a pervasive concern in many quarters is the future value of professional contribution.
Figure 3: In this prototype, project Quantum keeps Autodesk Revit and Autodesk Civil 3D topography synchronized between platforms
Figure 4: BIM 360 Insight uses machine learning in the cloud to find patterns in thousands of issues recorded by participating construction companies, using past data to predict current risks
To my mind, the concerns are misplaced. Merely framing a building problem for an automated approach is a human endeavor requiring considerable thoughtful investment and client discussion. Selecting the proper data and services to apply to a defined problem is itself a creative endeavor. Prioritizing conflicting project goals can only be performed by professionals advising their clients. No machine is going to be able tell us what’s important, what critical values we must serve, but by the careful application of the new capabilities the cloud affords, machines may be able to help us explore and understand the full consequences of our building and infrastructure design and construction choices, allowing us the luxury of informed and confident decisions that pursue prioritized intent on every project.
Supporting the scalable cloud automation of building, infrastructure, and manufacturing design, fabrication, and construction isn’t merely a good idea, it’s a necessity. By 2050 our planet will be inhabited by 10 billion people who will need 14 million new buildings and attendant infrastructure just to maintain the world’s current wildly unequal standard of living. As building professionals, we should aspire to far more than maintaining the inadequate status quo. Digital technologies have finally reached a threshold where they can offer more than the flawless repeated execution of precisely crafted commands.
As partners in our professional endeavors, software environments supported by cloud technologies that can deliver the right building expertise to the right people at the right time will transform the AECO industries, dependably raising the level of practice worldwide and allowing us to be more effective in delivering on project intent because we’ll have at our disposal the collective expertise of our industries. Just as Google helps us find anything, and Amazon helps us buy anything, we’re hoping that Autodesk can help make anything. Better.
Anthony Hauck has been involved in architecture, engineering, construction, and technology for more than 30 years. As an architect, millwork project manager, software developer, and IT Director, he has always looked to technology to help solve issues facing the building industry. He joined the Autodesk Revit team in 2007, holding a succession of product management positions in the group until joining Autodesk AEC Generative Design in 2015 as its Director of Product Strategy, where he is responsible for helping define the next generation of building software products and services for the AEC industry.
