The Industrial Fluids Simulation Challenge: Highlights of the First Event and a Preview of the Secon

1
The Industrial Fluids Simulation
ChallengeHighlights of the First Event and a
Preview of the SecondRay Mountain1, Fiona
Case2, Anne Chaka1, Daniel Friend3, Dave Frurip4,
Russell Johnson1, Joseph Golab5, Petr Kolar6,
Jonathan Moore4, James Olson7, Rick Ross8, Martin
Schiller9, Joey Storer4

• 1NIST, Gaithersburg, MD 2Colgate-Palmolive
  Company, Piscataway, NJ 3NIST, Boulder, CO 4The
  Dow Chemical Co., Midland, MI 5BP, Naperville,
  IL 6Mitsubishi Chemical Corporation, Okayama,
  Japan 7The Dow Chemical Co, South Charleston,
  WV 83M Company, St. Paul MN 9E.I. du Pont de
  Nemours Co., Inc., Wilmington, DE
• FOMMS 2003 participants in red

2
Background

• Idea arose out of Dow-NIST discussions
• NIST Workshop in 2001
• Reasons for contest
• benchmark current capabilities against real
  industrial problems (see next slide)
• Prize money donated by generous grants from Dow
  and BP

3
The Need for a Competition

• Accurate physical property data are critical in
  process and materials design, but it can be
  difficult to obtain reliable information,
  especially for unusual materials, mixtures, or
  state points far from ambient conditions where
  experimental determination becomes costly,
  unsafe, or too time consuming.
• Some data are available in the literature or can
  be estimated using empirical correlations based
  on literature data. Resources exist to aid the
  experimental evaluation of data at NIST, in the
  AIChE Design Institute for Physical Properties
  (DIPPR) consortium, and at contract measurement
  laboratories.
• Computer simulation holds out great promise in
  this area. In the future we hope to be able to
  build models of sufficient accuracy to
  confidently predict physical properties, even for
  materials that have never been studied
  experimentally.

4
The Need for a Competition, contd

• lack of validation of different methods and of
  reliable comparison studies is a major limitation
  to the industrial application of atomistic scale
  simulation
• Contest enhances alignment of academic efforts
  with Industrial needs

5
Background, contd

• First Contest ended Sept. 3, 2002
• One day symposium at AIChE meeting on Nov. 3,
  2002
• Champions announced
• Entrants presented results

6
The Problems

• VLE (Px and azeotrope)
• dimethylether/propylene
• propyleneclycoldimethylether/nitroethane
• Density (at two state points)
• water, cyclohexane, isopropanol, dimethylamine,
  1,2,3-trichloropropane, triethyleneglycol,
  pyridine, water/choline chloride, water/methanol
• Viscosity
• nonane, isopropanol, mixtures

7
The Answers

• Benchmark data supplied by NIST (Friend/Magee)
  and Dow (Olson)
• combination of literature data (published and
  unpublished) and new experiments at NIST and Dow
• unpublished DIPPR data used for VLE
• Results are unimpeachable!

8
The Contestants

• Ten entries representing seven different
  individuals or groups
• 3 entries for VLE, 4 for density, and 3 for
  viscosity
• Three commercial software vendors Accelrys
  (Rigby) and COSMOtherm (Klamt) Aeon (Sun)
• Academic Colorado School of Mines (Ely) BYU
  (Rowley) The Great Lakes Regressors U. of
  Minn. (Siepmann), Notre Dame (Maginn), SUNY
  Buffalo (Kofke), Wayne State (Potoff)
• National Labs Sandia (Martin and Thompson)

9
The Judging Process

• Double-blind
• Benchmark team did not see entries before
  benchmark data was decided
• Entries judged using a combination of
• quantitative criteria (closest to answer)
• qualitative criteria (FF fit to data?, etc.)
• typically 60/40 quant./qual.

10
Results Viscosity, ChampionThompson Martin
Nonane/isopropanol mixtures .1 Mpa and 300 K
Champion ?
Champion ?
11
Results Density Champion Huai Sun
12
Results VLE Problem Champion Klamt

13
Results VLE -- DP from Raoults Law

at 253.15 K for Dimethyl Ether(1) Propylene(2)
14
What We Learned from Challenge I

• Calculations gave reasonable answers but we are
  not there yet!
• A bit too ambitious (one person year for all
  problems)
• Need better definition of and communication of
  judging criteria!

15
Contest II, The Saga Continues!

• Expanded Contest Team to include Mitsubishi (P.
  Kolar) and 3M (R. Ross)
• 2 years (Fall 2004, AIChE Mtg.)
• 10 mos to plan carefully
• 14 mos for entrants to grind the crank!
• We expect entries to increase

16
Problem 1Vapor Pressure and Heat of
Vaporization

• Vapor pressure is an extremely valuable property
  in the chemical industry used mainly in the
  design of separation processes. It is also an
  important parameter in evaluating environmental
  and safety hazards of chemicals.
• Vapor pressure and heat of vaporization of two
  organic materials, each at four temperatures.
  For scoring, a numerical score will be computed
  as the sum of the absolute values of the
  percentage difference for each of four state
  points from the experimental benchmark data.
  Appropriate weighting will be applied to each
  part of the problem to reflect their relative
  difficulty.

17
Problem 2 Henrys Law Constant

• Solubilities of gases in liquids are properties
  of great industrial importance
• The inherent experimental difficulties associated
  with certain gas/solvent mixtures (e.g. oxygen
  solubility) make them promising candidates for
  the application of molecular simulation as a tool
  for physical property prediction.
• Calculate the Henrys law constants of 4 gases in
  1 liquid at 2 temperatures. For scoring
• 20 of the score will be based on predicting the
  correct trend with temperature for each gas.
• 40 of the score will be based on a quantitative
  comparison of the predicted Henrys constants
  with the experimental values, but only in a
  relative sense.
• 40 of the score will be based on an absolute
  quantitative comparison of the predicted Henrys
  constants with the experimental values.

18
Problem 3 Heat of Mixing

• Calculate the heat of mixing for 2 binary systems
  at 4 equally spaced compositions and 2
  temperatures (a total of 16 state points). The
  temperatures will be about 50 K apart and low
  reduced temperatures will be selected to avoid
  the influence of the pressure dependence (P lt 10
  bar) on the heat of mixing. For scoring
• 20 of the score will be based on predicting the
  correct trend with temperature (or the slope
  DHE/DT) for each system.
• 80 of the score will be based on a quantitative
  comparison of the predicted HE values with the
  experimental values.

19
Your Name Here!
We Encourage You to to Participate!!!