Barriers to CFD utilization in the building industry
I saw a thread today regarding how CFD Experts are not tapped for knowledge in the building industry. The poster went on to question why that might be. It’s certainly accurate that we don’t utilize CFD very often, and even more true we don’t adopt CFD modeling usage at the correct point in design. I began to think to myself about why it’s such an unused talent and what the barriers to that might be.
Though there are exceptions, there are two ways CFD modeling is primarily leveraged. First, we see CFD utilized to simulate HVAC or exhaust conditions under wind load or in unusual, or sensitive spaces such as large atria, or high load server farms, when we as designers are looking to examine a particular quality. Secondly, CFD is utilized to develop complex fire and smoke delivery models in conjunction with egress modeling to demonstrate smoke control system effectiveness.
As a fire protection and life safety professional, I primarily utilize CFD in conjunction with combustion, radiation transport, and fire dynamics modeling to produce an accurate simulation of a fire and smoke condition that could threaten the occupants of a building. As is standard practice in my specific industry, I utilize Fire Dynamics Simulator(FDS), developed by a group of professionals way smarter than I at National Institute of Standards and Technology(NIST). Though passably competent at HVAC modeling, FDS is primarily useful for low-speed thermally driven flow and analyzing heat, species and particle flow within the model through qualitative and quantitative means.
There are also two ways to design a building into code compliance. We can design based on prescriptive code requirements (i.e. the local building code may require 10 air changes an hour for a certain occupancy) which are generally, though not always, based on practical experiences or scientific information. The designer can also under certain conditions utilize performance based design. What this means is you take the actual code requirements and basis for the code requirements and demonstrate ‘alternative compliance.’ The IBC does occasionally make specific callouts for performance based design, smoke control and egress analysis is certainly on that list. The primary mode of executing performance based design in this case is through CFD, fire, and egress modeling.
I should think everyone by now has heard the platitude: Done Well, Done Fast, Done Cheap: Pick two. Budget and time by necessity factor into work planning. By default, it is much simpler and faster to comply through means of prescriptive requirements. The engineer can demonstrate minimum compliance simply by indicating a code reference. This seems to this professional to be dissimilar to how design happens in other industries tied to CFD, where the engineer can simply operate within budget, material, and clear design constraints. By contrast, the A/E requires approval of the building at the culmination of building design from an Authority Having Jurisdiction (AHJ), and a particular AHJ may have different priorities or experiences with compliance reports. Thus, when the designer utilizes performance based design, the burden of responsibility on the design and litigation risk to the company increases. Especially so in design affecting life safety, the engineer is required to justify assumptions, metrics, and conclusions. This work does not entirely consist of CFD analysis, much time is spent analyzing materials, comparing to available resources, and referencing white papers.
The result is additional man hours and money pursuing this goal of performance based design. By its nature, CFD cannot be utilized at every building design opportunity — computational run time on complex models presents a real opportunity cost to competent users, and every implementation is unique, else clear prescriptive requirements would be in place replacing the need for CFD. To be done properly, CFD development should occur in conjunction with building design progression from our architectural, mechanical, and fire protection partners to create a solution that works for all stakeholders. FDS for instance utilizes Navier-Stokes through a second order accurate predictor-corrector scheme. If what I just wrote seems gibberish, this means FDS’s algorithm is very accurate but computation time can be quite large (dependent on many factors, though often on the order of days and weeks). To lead a team through an atrium design, careful work planning, design control management, and educational exercises within a team become absolutely critical, as design progress must be integrated into your model in time for permit submission.
Engineers have always been practitioners of applied science. As HVAC is not a particularly new practice many of the scientific principles utilized have since been put into practical application resources. This often means that it is much simpler and time effective to utilize prescriptive design in particular for HVAC design for typical project applications. As a result, it’s been my professional experience that most CFD work performed falls under the purview of smoke control design.
Fire safety design with respect to smoke control however, lives in a slightly other sphere. The science is for the most part multi-disciplinary and thus more complex, including but not limited to aspects of fluid dynamics, thermodynamics, psychology, chemistry, statics, mechanics and dynamics. By proxy, the science lives less close to the building code requirements, and can often be determined though observations both historical and scientific of failure criteria such as bodily or structural harm. Widely accepted CFD best practices do not exist in the United States, though there is some movement on this inside of NFPA and SFPE right now to attempt to rectify this. Other countries such as Sweden have made some best practice recommendations for CFD smoke control design, but it is this professional’s opinion that fire science is a new enough discipline that it hasn’t well propagated yet. Since best practices are difficult to discover, this presents a substantial barrier to CFD experts who wish to examine fire and smoke analysis, and fire protection experts who wish to delve into CFD even if they understand the scientific principles behind CFD (more on that later.) Furthermore, and perhaps more importantly, this same barrier also exists on the Authority Having Jurisdiction side.
Some AHJs who do not fully understand what the model is computing internally can be wary of results, because CFD models can show wildly different results depending on different assumptions of fire package, including wattage, smoke production, growth rate, and selected mesh cell size and have no readily available metric to measure and give them a warm fuzzy feeling about what we produce as engineers for their evaluation and approval. Additionally, it’s easy to hide, on purpose or by accident, design assumptions in a CFD model, so an AHJ must also understand the fire and fluid science as well as the programming of the model to truly feel comfortable.
AHJs are becoming much savvier about CFD implementation as time goes on, and it is certainly possible and encouraged that we as experts educate the AHJ so they can judge our model fairly and accurately, but this is not always possible within the budgetary and time constraints of a fast tracked or budget conscious project.
Primarily in building design, we want to produce the best possible product for our client, within these time and budget constraints. Ideally, this means that the client, stakeholders, architects and engineers will gather and determine design criteria for a desired atrium for example. Criteria would typically include but not be limited to the square footage, furniture package, program requirements, areas of inclusion or omission from an atrium, and qualities of control and detection systems. Thereafter, the A/E ballparks an internal project budget and construction budget. This process doesn’t always happen so cleanly, and sometimes the division of labor is distributed, especially with the increase of design-build projects. However, any lack of clarity from any party can make CFD difficult to execute. Due to the lack of definitive metrics in the building code for analyzing CFD design, an AHJ could in theory deny occupation for a building if they deem the model inappropriate or incomplete for any reason and could demand the prescriptive requirements are met instead. In practice this rarely happens, but occasionally there is some trepidation and additional measures must be taken from the engineer to assuage any concerns.
Any further doubt from the AHJ on the quality of the mode of implementation, quality of the model or the engineer, or real-world applicability of the CFD model can give the client or the AE hesitation to incorporate CFD into their design. Thus, unfortunately, CFD modeling is sometimes relegated as a last resort to address a design flaw, instead of proactively to enhance the design.
Generally, there are two categorical ways to leverage CFD as a design tool. The designer can:
A) use CFD to improve the end product by deviating from normal design requirements to save the client money on materials or to facilitate an innovative design that wouldn’t normally fit into prescriptive requirements, or
B) use it to verify design criteria already in place. Ideally, you want to use CFD as a design tool as part of a team through the design process so you can free your team up for some design flexibility.
Worst case, someone is brought in very late in the design process to rationalize a non-compliant design after the fact. I see it frequently. Occasionally I have observed an unscrupulous designer prey on the unwillingness of the AHJ to seem inexperienced or unknowledgeable about this honestly quite advanced material. This work does a disservice to our industry as a whole, especially so to the occupants of the buildings we design for. It stunts the growth of our industry as we only will see more CFD work if the AHJ and the designer can have intelligent conversations about the results and assumptions included in the design report. Comfort level of the AHJ in having these discussions is absolutely critical to the success of a project, and can often be remedied by early and often conversations about CFD and design goal with the AHJ which, once again, eats project hours.
The same effect often happens when the designer is misaligned with general understandings of fluid dynamics, or the impact of certain assumptions. I’ve seen many times that default variable settings are left in place for inappropriate conditions. Given a tool, it can seem quite simple to produce results, but as with anything quite so technical and programming driven, it is, as always, Garbage In, Garbage Out.
Another barrier to CFD usage I see is a differentiation on CFD technique between software packages. For instance, what might make one an expert at CFD for fire dynamics near certainly doesn’t make them an expert in energy modeling. The tools are completely different, for example FDS is not able to utilize dynamic meshing. Not only does this lengthen computational time, it also further increases the knowledge barrier to competent modeling creating the need for an understanding of the inner workings of the computer, including how Message Passing Interface functions and what computational resources you can expect to see a bottle neck from when you are building the model.
To conclude, I believe the barriers to CFD utilization in the building industry can be summarized thusly:
1) Time or cost constraint on the project which easily can add over 50k USD to the budget of the design work versus prescriptive design.
2) Performance based design work increases the risk of litigation to the firm executing the analysis.
3) Computation time introduces an internal time constraint on the AE team, as it lags design work, exacerbating the demand on a team.
4) Lack of available building code best practices and metrics can discourage new practitioners from adopting CFD as a design tool.
5) AHJs require education to comfortably and correctly asses the quality and applicable validity of a CFD model. Lack of available building code metrics can cause AHJs to face CFD performance based design with trepidation exacerbates this condition. AHJs can thus reject even a properly performed analysis for any basis which causes a chilling effect to the AE and client on CFD utilization. The primary reducer of this risk is open and honest communication with the AHJs at an early stage of design which isn’t always afforded by the project time and budget constraints.
6) As a near consequence, CFD is often brought in after the problem has already arisen instead of proactively, which can lead to additional ill will and distrust on the part of the client and AHJ.
7) Poor CFD models serve to muddy opinion and erode the knowledge base of the AHJs.
8) Huge knowledge barrier of entry for CFD experts from other areas of expertise to shift to fire dynamics, the most common usage of CFD in building applications.
All of this said, CFD is still seriously underutilized as it remains a powerful and flexible tool for building design. I can see advancements in acceptance of CFD as a tool for informing design from the AE field as well as from the AHJ and owner. I am appreciative that more and more often, AHJs are open and curious about the possibilities, and colleagues get excited about the possibilities that CFD presents outside of normal design constraints. There are plenty of growing pains to push past for this section of the industry, certainly. I am encouraged with what the years will bring.
If this posts interests you, and want to talk about how CFD is going to change in the upcoming years, please send me mail at seanmccreadyfpe@gmail.com or drop me a message on LinkedIn.
