From Wind Tunnels to High Performance Computing

Back in the 1930s, the Tennessee Valley Authority (TVA) was created by congressional charter in May 1933 as a federally owned corporation in the United States and ultimately declared constitutional by the U.S. Supreme Court in 1936 [1, 2].The TVA was, in part, established as a way to combat the seeming unfair business practices of for-profit power generation that that favored owner over consumers. It also ended up being a post-Depression era works program for a rural/agrarian South (covering most of Tennessee, parts of Alabama, Mississippi, and Kentucky, as well as smaller parts of Georgia, North Carolina, and Virginia). TVA's power mix as of 2012 was 10 coal powered plants, 29 hydroelectric dams, three nuclear power plants (with six operating reactors), nine simple cycle natural gas combustion turbine plants, and five combined cycle gas plants. TVA is the largest public power utility in the U. S. and one of the largest producers of electricity [1]. 

During World War II, the TVA was instrumental in America’s war effort as the massive power generation was used to manufacture many aspects of the U.S. war machine, most notably enriching uranium and making aluminum for airplanes. Vice President Harry Truman is credited with saying “I want aluminum. I don't care if I get it from Alcoa or Al Capone." [1]. Then, during WWII, it was clear that the U.S. was behind Germany in aircraft development and the late 1940s and 1950s saw an increase and, in many ways, the rebirth of aerodynamics research and airplane design in the U.S. This was due, in large part, to a number of wind tunnels that were built close to the power facilities of the TVA. 

These wind tunnels, in an almost literal sense, fueled the Cold War and advanced the state of the art in aerospace technology. In my opinion, however, these advances, while substantial (and gave us the understanding to build everything from the Bell X-1 to the SR-71 to the Space Shuttle), started to be farther and fewer between in the 1970s and 1980s. It was no coincidence that CFD was looked at as a tool in the 1980s when testing facilities began to yield fewer and fewer insights into aerodynamics.

This build up of power sources and subsequent aerodynamic capability echoes the more recent development of CAE, which is also reliant on power hungry devices: computers. As we all know, it was the computing resources that were the limitation of early CAE application, not necessarily the codes themselves (certainly numerical techniques have developed over the decades, but they were, again, in my opinion, developed and applied as computers got faster). However, while the 20th century consumed power directly for wind tunnels, I think the 21st century will consume power to cool computers that are running code. This is what we now know as the discipline of High performance Computing (HPC). The wonderfully poetic “close the circle” thing to HPC is that as the massive wind tunnels powered by the TVA are being shut down, there are huge computer clusters taking their place. Oak Ridge National Labs, the facility that was originally used to show that plutonium can be extracted from enriched uranium back in the 1940s [4] and powered by the TVA, is home to one of the fastest computer on the planet: the Titan XK7, which has a theoretical peak performance exceeding 27,000 trillion calculations per second (27 petaflops), featuring 299,008 CPUs, and a total system memory of 710 terabytes [5]. And while the test guy in me waxes poetic for those majestic wind tunnels from the golden age of aerodynamics, my current self thinks about the wonderful possibilities of leveraging the ever increasing and faster HPC resources and applying CAE to future challenges.

I hope you join me in that quest.

What are your thoughts on this (music or otherwise)? Or anything else for that matter? Send me an e-mail at: thecaeguy@nafems.org.

-The CAE Guy

[1] Tennessee Valley Authority, https://en.wikipedia.org/wiki/Tennessee_Valley_Authority. Due to lightness of this column, I thought it OK to reference Wikipedia,although it is generally deemed a less than reliable source, I will confess – like many people - to commonly using it as a starting point, but in this instance, it is also the ending point [3].

[2] Ashwander v. Tennessee Valley Authority, 297 U.S. 288 (1936).

[3] https://en.wikipedia.org/wiki/Wikipedia:Academic_use. The irony of referencing Wikipedia for why Wikipedia should not be a reference is,frankly, delightful.

[4] Oak Ridge National Laboratory, https://en.wikipedia.org/wiki/Oak_Ridge_National_Laboratory[5] Titan Cray XK7, https://www.olcf.ornl.gov/computing-resources/titancray-xk7/