A Worst Case Scenario
On November 7, 1940, one of the most spectacular engineering failures of the 20th century occured in Tacoma, Washington. During the early morning hours, the Tacoma Narrows Bridge began to undulate violently in 35+ mph winds. After several hours (and luckily after the bridge had been closed by police), the oscillations took the form of a twisting motion that would cause the main section of the brigde to tilt at up to a 45o angle from its original position. After approximately one-half hour of this twisting motion, the main span of the bridge collapsed. The bridge, which had cost $6.4 million, had been open for only a little over four months.
The fundamental design flaw in the Tacoma bridge was a failure to account for the resonance vibrations generated in the structure due to the aerodynamic forces. Subsequently, the replacement bridge design was tested a wind tunnel to measure potential wind effects before being built, and such type of testing became standard for all future brigde designs.
The Tacoma Narrows Bridge disaster illustrates two main points: 1) that fluid dynamics can play a role in many critical phenomena outside the traditional spheres of aerospace and mechanical engineering and 2) that analysis of these phenomena during the design phase of a project can potentially save a lot of time and money.
The advent of Computational Fluid Dynamics (CFD)
Of course, even after the Tacoma Narrows Bridge disaster, fluid dynamics analysis was still not widely employed. Why? Well, because fluid dynamics is difficult - the equations governing fluid flows are non-linear, and can only be solved analytically in a few cases. While these cases proved to be useful approximations of the real solution in many situations, the fact was that only through scale model testing in wind tunnels, towing tanks, or other experiments could more precise and detailed information be obtained. Given that the number of venues available for such testing were limited and that the test were not inexpensive, many projects that might have benefited were not subjected to fluid dynamic analysis.
All this has changed in the past few decades with the exponential growth of computer power and the application of Computational Fluid Dynamics (CFD). The object of CFD is to use computersto solve the previously intractable conservation equations for fluids in order to accurately simulate. Thus, computational fluid dynamics allows the analysis of fluid flow problems in detail, faster and earlier in the design cycle than possible with experiments, costing less money and lowering the risks involved in the design process. This trend is only likely to grow more pronounced in the future as computers become increasingly cheaper and more powerful while traditional forms of testing become increasingly expensive. This concept was expressed graphically in what is known as Ruppert's 2nd Law. As shown in the graph to the left, the expectations are that over time, the ability for computations to simulate complex problems will increase while cost decreases; on the other hand, the cost for experimentation is likely to continue steadily increasing. Therefore, the region in which CFD is more praticable and less expensive (represented by the region in which the red curve falls below the corresponding green curve) is steadily growing.
The Flexibility of CFD
The ready applicability of CFD today has caused its usefulness to be recognized in many areas. While its most common usage remains in engineering fields such as aerospace, mechanical, and chemical, CFD has made in-roads in many other areas. Examples of this range from biomedicial research on blood flow to analyzing the cooling effects of air blowing over computer chips. CFD is even used on problems not usually thought of as involving fluids, such as plate tectonics (which act as fluids over million-year time scales), magnetic fields (a discipline known as magneto-hydrodynamics, or MHD), and galaxy formation.
Even with the advancements in CFD, it is still not a foolproof science: some fluid phenomena are still not well understood and cannot be simulated well numerically, translating the results from computational to physical space can be tricky, and without some careful validation work the full "Quality and Trust" in the simulations cannot be established. Therefore, CFD analysis still demands relying on some expertise in the field, but there are several different means of obtaining that expertise. For an organization that will require large and regular amounts of CFD work, such as aerospace or automobile companies, the best option is often to maintain their own staff of in-house CFD experts. This allows the development of proprietary codes that might provide a competitve advantage, but at the cost of internal code developers and maintainers.
Another option more palatable for smaller organizations is to obtain ready-made codes from government or commercial sources and use them, minimizing the need for code development. While the government CFD programs are often free, they tend to be less user-friendly, more research oriented, and lack good technical support. Alternately, commercial codes are relatively user-friendly and are supported and updated, but they are essentially opaque black-boxes that are only as good as the original code developer. Since no one code can account optimally for all possible flow phenomena, commercial codes can produce poor or occasionally wildy inaccurate results for more complex problems.
Increasingly, the option chosen by many industries and laboratories is to outsource their CFD work to academia, research labs, or consulting companies that specialize in fluid dynamics. This allows the local engineers and scientists to focus on design and development rather than numerical simulation, requesting CFD simulations only when necessary. Further, competition between potential outsource targets helps keep the costs down, and drives the CFD organizations to keep up-to-date on (or in some cases develop) the latest CFD technologies. Through outsourcing, CFD analysis has become more cost-effective for a wide variety of small-to-midsize companies. Even larger companies and labs when faced with a highly challenging problem will outsource: as can be seen in the figure, the most complex problems involving time-dependency and three spatial dimensions are pretty much the domain of a few, select industries, certain government labs, and academia.
At the University of Kentucky, we are committed to performing the state-of-art research necessary to be a leader in computational fluid dynamics. We are investigating many of the most vexing challenges in fluid dynamics, including turbulent and transitional flows. On the computational side, we are developing advanced numerical techniques and solvers to achieve more accurate results more rapidly. We are also focused on how best to transfer these technologies to application-oriented industries and government projects, both the processes needed to establish "Quality and Trust" in CFD simulations and the best means of supplying the technology. Finally, the UK CFD group is dedicated to training the next generation of CFD experts, perhaps the most direct form of technology transfer available.
To examine how our mission is applied to actual research projects, please examine our Projects section.