Advanced Computing in the Age of AI | Thursday, March 28, 2024

Did Analytics Answer Malaysia Flight MH370 Water-Entry Question? 

(Source: Wikimedia)

The tragic disappearance of Malaysia Airlines Flight MH370 in March 2014 generated few facts and many questions. But a team of computational math, engineering, modeling, and advanced scale computing experts believe they solved at least one of the many puzzles: How the Boeing 777-200ER entered the ocean.

Just as the team of academics behind this research hope their investigation sheds light on the mystery that happened over the Indian Ocean more than a year ago, they believe the tools and techniques they used also can help enterprises answer big-picture questions and analyze vast quantities of seemingly unrelated data to better position companies to take advantage of new opportunities, markets, or products.

Having watched news coverage about the vanished plane and heartbreaking presumed deaths of 239 people, Goong Chen, professor of mathematics at Texas A&M University (TAMU) in Austin and TAMU Qatar, realized nobody had yet studied Flight MH370's water entry from a mathematical perspective. Rather, most analytics focused on the trajectory, he told EnterpriseTech. Applied mathematics plus crash analysis crunched on high performance computers, coupled with input from experts in other fields, determine how the plane entered the ocean, said Chen. That, he said, would provide a richer picture of the global quandary – and a further demonstration of how mathematics and advanced scale computing can answer questions about the unknown for organizations across a spectrum of disciplines.

Investigating the water entry took insight and data from multiple experts.

"I really needed to rely on collaboration with aerospace engineers, and ocean engineers, and impact mechanics – when an aircraft crashes into something there is a huge impact so we need to do impact analyst, crash analysis – so I gathered together a team," he said.

Like an enterprise initiative that can include a mix of experts from IT, engineering, and lines of business, Chen partnered with professionals in various fields both within and outside TAMU. They included Cong Gu, a PhD student in TAMU's math department; Philip Morris, Boeing/AD Welliver professor of aerospace engineering at Pennsylvania State University; Eric Paterson, Rolls Royce Commonwealth professor of marine propulsion and department head of aerospace and ocean engineering at Virginia Tech; Alexey Sergeev, postdoctoral fellow at the Qatar Environment and Energy Research Institute in Qatar; Yi-Ching Wang, PhD student in the math department at TAMU, and Tomasz Wierzbicki, professor of applied mechanics and Director of Impact and Crashworthiness Laboratory at the Massachusetts Institute of Technology (MIT). They wrote an in-depth paper – "Malaysia Airlines Flight MH370: Water Entry of an Airliner" – about their research for the American Mathematical Society, describing the equations used.

Virginia Tech's Paterson, an expert in computational fluid dynamics (CFD), was an early adopter of OpenFOAM – Field Operations and Manipulation – a free, open source, object oriented C++ software that allows developers to write their own solver applications, he said in an interview. OpenFOAM is scalable to thousands of processors, Paterson said, and was well suited to this investigation.

"I brought the OpenFOAM software to him and his group and then they ran with the ball. My role was mainly advisory. His students and post docs did the heavy lifting," he said. "They did the simulations and analysis. I kept them on the straight and narrow."

Analyzing the Entry

Studying water entry is not new for mathematicians, Paterson said. But this particular case was different – and not only because Chen's interest was triggered by a tragedy in the news.

"What was new here was we did simulations for something really complex – an airplane – a fuselage with wings and engine cells. Obviously that simulation of an airplane is indeed much closer to MH370 than a bullet," he said. "There are not a lot of people who have done water entry for very complex vehicles. What facilitated that was HPC; flex meshing algorithms, and open source OpenFOAM CFD software. You have to model the atmosphere, the ocean, and all the fluid dynamics of that event and that's what CFD is. In my mind that's the real accomplishment."

Chen's team at TAMU took data – such as atmospheric pressure, water surface tension, gravitational acceleration, and airplane dimensions – and used the university's EOS supercomputer in Austin and RAAD supercomputer in Qatar to conduct analysis, he said.

"Both supercomputing facilities were extremely helpful in granting us our supercomputer time allocation and providing technical support as well. At first we would run a job where we used 20 processes and it would take about a week to run one case," said Chen. "Later on, we made the algorithms somewhat more efficient and then we just concentrated on the most important parts, so for each job we input into the supercomputer, it usually took something on the order of one to three days. That is just the main core computing part. After that, we needed to do more core processing."

Based on the research, Chen's team came up with five possible scenarios; only one fit all parameters such as lack of debris and observable oil spill in the days following MH370's disappearance. As a result, "the nose-dive water-entry or a water-entry with a steep pitch angle, is the most likely scenario," they wrote in the AMA article. "This particular assertion is speculative but forensic, based mainly on the observations of the computed data in the [paper], combined with the understanding of aviation precedents, atmospheric and ocean surface conditions…"

While the research itself is interesting, it should not be extrapolated to MH370 since the aircraft and passengers have not yet been located and the results cannot be applied specifically to this aircraft, Paterson said.

"They looked at a different entry angles and make a hypothesis. You need hard calculations," he said. "If you had to pay for the ships that went out to do searching would you use the data to guide your searching? I wouldn't. It's more an example at this point of what could be done. It wouldn't help anybody necessarily find MH370."

Enterprise Lessons
They may disagree about the meaning of the results, but both Chen and Paterson concur enterprises can learn from this investigation.

"Top executives need to look at things in both the long and the broad range. There are always some not well understood gut feelings, insights, or reports. We're all hoping to predict what we do not know," said Chen. "High performance computing will play an extraordinary role. Large corporations such as Exxon Mobile, Shell, etc., they have very many in-house experts. But in-house experts could also have their limited view of some important trends or events because they have been settled in the corporate atmosphere and culture for too long.

"For a top executive, I would recommend they form some closer bonds with the academic community," he added. "Academics tend to think more broadly, more in long range about what they think are important trends of future predictions. Those things, of course, are usually very complex and they need large models and computational scientists that generate a lot of correct predictions. For such complex phenomenon, we cannot rely on the gut feeling of a prominent analyst in a certain market. Even prominent analysts make contradictory recommendations."

Enterprises should make more use of scalable open source applications like OpenFOAM, recommended Paterson. Today, 78 percent of companies run on open source, according to "The Future of Open Source," a 2014 report co-sponsored by Black Duck and North Bridge. One reason: 58 percent said open source offers the greatest ability to scale and 55 percent it provides superior security, the study said. Businesses should consider using these tools on more powerful systems, he said.

"The take away would be there are open source solutions that are free that allow you to do large-scale computational mechanics simulations," Paterson said.

About the author: Alison Diana

Managing editor of Enterprise Technology. I've been covering tech and business for many years, for publications such as InformationWeek, Baseline Magazine, and Florida Today. A native Brit and longtime Yankees fan, I live with my husband, daughter, and two cats on the Space Coast in Florida.

EnterpriseAI