How Practitioners are Applying AI and CFD Research to Advance Wind Engineering 

By Neetha Vasan 

The best professional advice is equally informed by scientific research and the practical experience of past application. The value of applied research, derived from collaboration between researchers and practitioners, cannot be underestimated. One example of this is the advancements in Computational Fluid Dynamics (CFD), Artificial Intelligence (AI), and Machine Learning (ML) in wind engineering. Wind tunnel testing (WTT) has been and continues to be the primary choice for accurate quantitative evaluation of wind flows around buildings. Widespread use of CFD in wind consulting and the subsequent lessons learned have furthered the application of AI and ML in wind and microclimate studies. With these advancements, the wind engineering industry now uses CFD with more confidence and for more applications in bluff-body aerodynamics and microclimate assessments than available research would have allowed 20 years ago. 

The co-dependent relationship between research and consulting

A key difference between scientific and applied research is the environment in which each occurs. Scientific research often happens in a sterile environment with controlled inputs. On the other hand, applied research— especially in professional consulting— is faster-paced and has greater volume and diversity of application. Additionally, the work is driven by demand for time-sensitive solutions and outcomes with real-world impact. The communities a consultant serves depend on the consultant’s work to advance their own objectives. 

These factors allow consultants greater visibility into real-life problems that are often challenging, unique, and lacking sufficient scientific research to aid in developing a solution. This scenario, in turn, brings the practitioners back to researchers to aid in collecting more scientific data. In this way, the researchers stay abreast of challenges they may not have considered otherwise. Thus, the co-dependent — and often cyclical — relationship between strong research and its practical application by professionals to move an industry forward is made clear. 

The importance of CFD research and development in consulting

In the consulting and service industry, the saying goes that of good quality, low cost, and fast turnaround, you can only achieve two at any given time. Since it is often impossible to succeed in all three simultaneously, the consultant’s job is to find a balance between the three.

Consultants need the right tools to balance each aspect of a project effectively and efficiently. These tools or processes must be proven, meet industry standards, and follow best practices to reduce risk. Processes must also be repeatable, accurate, and precise. Meeting these requirements is no small feat, and CFD is no exception to the decades of research often required for engineering tools to attain industry standards.

Computational approaches to wind engineering have been a focus of industry research for decades. However, it was not until recently that wind consultants’ application of CFD, AI, and ML gave researchers the necessary data and industry validation to advance developments in this space. Although researchers have successfully used CFD in the research of bluff-body aerodynamics—including structural loading calculations, which are typically resource-intensive—engineering consultants typically only use it for less demanding applications. CFD needs further development to be applied at the fast pace of the consulting industry, while being able to deliver results that are of good quality, at a low cost, with the desired fast turnaround. 

Computational wind engineering (CWE) has come a long way in the last several decades, driven by a research and development curve that has been notably steep in the last 20 years. Just 10 years ago, consultants would use CWE only for qualitative, early-design, informational assessments with low impact on critical design decisions. Today, as illustrated in Figures 1 and 2, technological advances enable broader applications and high-fidelity simulations at a reasonable speed, to the point that CFD has become a valid alternative to wind tunnel testing for many applications. 

Figure 1: Choice of Wind Assessment Approach for Different Development Stages

Figure 2: Choice of Wind Assessment Approaches for Different Areas of Assessment

How wind engineering consultants use CFD and AI to deliver more effective solutions 

Advances in CFD and AI applications in wind engineering have changed the consulting landscape, enabling wind engineering specialists to deliver better and more efficient solutions to the challenges faced by their clients and stakeholders. Consultants can now use CFD to support communities that are striving to lower the environmental footprint of building projects, and proactively plan for livability in cities where these projects  are built. 

Case Study: San Francisco, California

A consulting engagement in San Franciso illustrates how innovative tools impact a community. San Francisco has established environmental impact bylaws that mandate WTT to determine the effect of building-induced wind on the pedestrian experience. Because the city is moderately windy in summer, tall buildings could result in pedestrian-level wind conditions that exceed the prescribed safety criteria.

I worked with a developer who had proposed a building that would occupy a large lot in San Francisco and rise approximately 400 feet (over 120 stories). This project would tower above its surrounding buildings, which were all  less than a third of its height, and would therefore be exposed to strong winds at high elevations. The proposed development would substantially displace natural wind flows in the area, causing problematic wind conditions for pedestrians. An expert in building-related wind microclimate could make this educated prediction in a matter of minutes. 

Throughout the building permit process, the city-mandated wind assessment required 35 iterations of wind tunnel testing to satisfy the pedestrian safety regulations. While this was a robust and accurate assessment, this methodology is expensive and time-consuming. In aggregate, the WTT , consultation meetings, and charrettes totaled over $300,000, and took more than two years to work through the city’s regulatory process before obtaining final approval. And then, after all that time and expense, the design was ultimately scratched for feasibility reasons. 

When a different architectural team was brought on to create an entirely new design, the developer did not repeat the same mistake. This time, they used CFD and ML to mold the design in the initial design stage, when there was more freedom to make architectural changes. This enabled the developer to identify high-risk areas and explore several massing and architectural wind mitigation elements before entering the city’s regulatory process. This approach significantly improved the project’s time and cost (Figure 3).

In the original plan (Design 1), all the design iterations happened in the WTT process, which drove up the cost tremendously. In the second iteration (Design 2), early planning allowed a wind-responsive design from the start and reduced time in the regulatory WTT process. As consultants, we leveraged each tool at the right project stage and delivered an efficient—and much more cost-effective—process. 

Figure 3: Cost-effective and efficient design process using the right tools at the right stages.

Providing Holistic Assessment and Solution Development

Most available research addresses one area or problem in wind engineering at a time. As a result of this research structure, wind consultants have traditionally had a conservative “symptom-based” approach to solutions. This approach often included connections between different wind problems as an afterthought, rather than as a proactive consideration. 

For example, blocking winds for pedestrian comfort and safety may affect natural ventilation, and in cold climates, redirecting wind to mitigate a problem may affect snow drift accumulations in the winter. Failing to account for these connected impacts increases the risk of a redesign, or the need to resolve costly oversights. Using the right tools allows for the simultaneous assessment of multiple connected disciplines. This interdependent analysis is more critical than ever, as cities increasingly prioritize livability for their citizens—including their residents’ physical and psychological needs. The ability to conduct multidisciplinary assessments that consider many areas of the human experience and address them holistically revolutionizes how we do consulting today.

In another example, a builder/developer client wanted to explore different tower shapes for their new project. Their critical goals were to lower pedestrian-level wind impact and be informed about the wind pressures that would affect façade cladding. We conducted a preliminary assessment using ML to identify hot spots and critical problematic areas. This assessment was quick and inexpensive, because we could use a single model for multiple assessments in parallel, without additional burden to project timelines and costs. Additionally, we could view all the results on a single board in a comprehensive format, which made discussions simple and intuitive. This visibility gave the client confidence to pick the preferred design. We then performed more detailed studies to inform the final design decision. 

Detailed studies are more costly than the quicker ML runs, so having a low-fidelity tool for the preliminary decisions that require multiple iterations and comparisons creates a huge advantage for consultants and clients. This relatively new advancement has turned the impossible to possible in just the last few years, thanks to the work of the wind engineering research community.

Figure 4: Lo-fi and hi-fi tools used at different project stages for efficiency gains in the design process.

Engineering Better Outcomes

Wind engineering researchers and practitioners form a synergistic relationship strengthened by a collective effort to advance the field. Recent advances in CFD, AI, and ML have significantly transformed the technologies available to consultants, and increased access to wind and environmental studies for developers. A decade ago, using CFD was considered unacceptable in many cases, but today, there is more confidence in its application. Many cities have now incorporated CFD into their bylaws, offering a variety of possibilities for wind consultants to help shape different stages of the planning lifecycle. 

The more versatile the toolkit, the more critical it is to select the right tool for the job. Experienced practitioners should use the appropriate assessment tools, understanding their nuances and limitations while being aware of how results are interpreted. As responsible practitioners, we must recognize the limitations of CFD and other computational tools, and remain conscious of the community’s dependence on our expertise.

Neetha Vasan, M.A.Sc., LEED AP, is Senior Scientist/Associate at RWDI, an engineering consulting company that has served diverse sectors for 50 years. Ms. Vasan is an accomplished senior microclimate specialist who has led and executed demanding wind engineering projects on three continents. With 15 years of global project management experience and significant expertise in aerodynamics and wind tunnel, she is responsible for supervising building and wind engineering project deliverables, managing quality assurance, directing technical and quality development research initiatives, training and mentoring colleagues, leading technical workshops for municipal clients and industry conferences, and applying her unique understanding of North American municipal regulatory processes to add value to clients in the U.S. and Canada. Ms. Vasan is a member of the American Society for Civil Engineers (ASCE) and a Technical Subcommittee Chair with the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE).