Second Challenge Results
Scientists and engineers from 3M, BP, Dow Chemical, DuPont, ExxonMobil, Mitsubishi Chemical, and NIST challenged the molecular modeling community to predict physical properties of industrially relevant fluid systems. Contest entrants presented their work, the champions were announced, and prizes awarded, during a special session at the AIChE Annual Meeting in Austin, TX. November 7, 2004
Academic groups, research laboratories, and scientific software companies from around the world were given just over a year to develop methods for predicting vapor pressures and heats of vaporization, gas solubility, and enthalpies of mixing for materials specified by the contest committee. Working in secret, scientists at NIST and Dow Chemical obtained accurate experimental measurements of these properties, which were used to judge the predictions made by the contest entrants.
“The organizing committee managed to raise interest from around the globe,” commented Philippe Ungerer, Institut Français du Pétrole (IFP), France, whose team won the overall “best in show” prize. “There were entries from almost every continent, including China, Japan, and from three countries in Europe,” he observed.
This second Industrial Fluid Properties Simulation Challenge, which was organized in conjunction with both the American Institute of Chemical Engineers (AIChE) Computational Molecular Science and Engineering Forum, and the Theoretical Subdivision of the American Chemical Society (ACS) Physical Chemistry Division, was held to obtain an in-depth and objective assessment of current capabilities for the prediction of fluid properties, and to promote the use of molecular modeling in this area. Molecular simulation has been identified as a promising technology for predicting materials properties in the Vision 2020 Roadmap for the Chemical Industry.
The problems were closely focused on properties that are industrially relevant. There were three sections. The first section challenged entrants to predict vapor pressures and heats of vaporization for two different materials. “The heat of vaporization is an important property in the design of heat exchangers and other chemical process units,” says Jim Olson of Dow Chemical, Midland, MI, USA “There are a growing number of chemicals whose vapor pressures and heats of vaporization need to be determined outside the range of ordinary apparatus. Molecular simulation could offer an attractive alternative to these difficult laboratory measurements.”
The second section challenged molecular modelers to predict the solubility of gases in liquids – dissolved gases are a key component of many industrial chemical processes. The last section involves the prediction of heats of mixing for an amine in both hydrocarbon oil, and in water, over a range of concentrations at different temperatures. “One of the premier challenges for modeling thermal separation processes (e.g. absorption, distillation, extraction) is the ability to describe activity coefficients over the whole concentration and temperature range with sufficient accuracy” says Martin Schiller of DuPont, Germany.
“We were pleased with the level of entries this year,” says Raymond Mountain, NIST, chairman of the organizing committee. “The contest provides a useful comparison between a number of different molecular modeling approaches, and some were clearly more successful than others,” he noted.
In the first two problem sets (the prediction of vapor pressures and heats of vaporization and the prediction of the solubility of gases in liquids) the various atomistic simulation methods did a fairly good job, and the best of the predictions were in reasonable agreement with the experimental data. The first problem set was won by Professor Richard Elliott and his students from the Department of Chemical Engineering at the University of Akron. The second problem set was won by Professor Jeffrey Errington and his students from the Department of Chemical and Biological Engineering, University of Buffalo, NY.
The last section (prediction of enthalpies of mixing) was won by Professor Huai Sun and his students from the School of Chemistry and Chemical Technology at Shanghai Jiao Tong University, Shanghai, China. “This was certainly the most challenging of the problems, and we particularly appreciated the two groups that were brave enough to enter their results,” commented Fiona Case, Case Scientific, a member of the IFPSC organizing committee. “The results showed that we have some way to go before atomistic simulation methods can be routinely used for predictions of miscibility, particularly for aqueous solutions.” Sun’s group was one of two who entered all three sections of the contest, using the same method and forcefields to predict all the different materials properties. “This is important, since it reflects the way that modeling methods are actually used in industry,” commented Case. “These two groups really entered into the spirit of the contest, and it was encouraging to see them obtain reasonably good results across the different materials properties.”
The other group which entered all three sections was a European collaboration involving researchers from Institut Français du Pétrole (IFP), France; Université Paris-Sud, France; Universitat Rovira i Virgili, Tarragona, Spain; and the Commissariat a l'Energie Atomique in France. This group, headed by Philippe Ungerer, obtained the best overall score and was awarded the “best in show.”
“Normally you wouldn’t see several different groups trying to predict the same physical properties,” commented Anne Chaka, NIST. “But, this is important if we are to obtain an accurate assessment of the current capabilities of atomistic scale simulation.”
“One of the great things about this contest,” commented Joe Golab, BP Chemicals “is that people were showing results that weren’t particularly good. We wouldn’t usually see those results, but it is important for industry to know what works and what doesn’t.”
