Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Ahren Jasper Applicant's institution: ANL Applicant's division: CSE Project Name: KTP Project title: Temperature- and Pressure-Dependent Kinetics Associated funding: DOE BES Other Systems: Chemistry's Blues cores b431-b460 Chemistry's linux 1300-core cluster "Linus" Science: In recent work, we have shown that quantitative predictions of gas phase chemistry can be made using a combination of classical trajectories (molecular dynamics), transition state theory, and master equation calculations. The present project focuses on using classical trajectories to predict energy transfer parameters for use in master equation calculations for real fuels of relevance to combustion. Whereas in past work, we have considered small test case fuels, such as methane, real fuels often contain much larger species such as dodecane and the large esters that appear in biodiesels. The main goal of this project is to use high performance computing to adapt our methods to the treatment of larger (real) fuels. Project description: We will (1) fit full-dimensional potential surfaces, and (2) run trajectories to calculate energy transfer. Step (1) involves parallel sampling of the intermolecular potential (which has a dimensionality of 3, 5, or 6 for atomic, diatomic, and polyatomic baths, respectively) using commercial quantum chemistry (QC) programs (principally Molpro). "Good" QC methods, such as counterpoise corrected CCSD(T)/CBS, can be prohibitively expensive, even in large parallel computing environments. We will therefore continue our work optimizing (a) the choice of QC method (considering MP2, dispersion corrected DFT, etc) and (b) sampling point placement (considering biased distributions and quasirandom (Sobol) sampling). In preliminary work, we have found (a) that MP2 is often accurate for Ar as a bath gas, and (b) a biased quasirandom sampling can increase efficiency by a factor of 15. Still, for large molecules, sampling will be expensive, requiring ~10000 1-hour QC calculations. Fitting the resulting data will be done using a genetic algorithm or linear least squares fitting. The cost of the fi tting is negligible relative to the cost of sampling. Step (2) involves using our parallel trajectory code DiNT to run energy transfer trajectories. 10,000 independent trajectories are required to characterize energy transfer for each system of interest, with each trajectory taking 1-60 minutes, depending on the size of the system. In this first step, we will consider medium sized alcohols, peroxides, and esters. For each of these three class, we will consider 9 cases (e.g., the three alcohols ethanol, butanol, and hexanol for each of the three baths He, Ar, and N2) for a total of 27 cases. The cost of both steps 1 and 2 will scale dramatically (in different ways) based on system size, with the largest systems considered requiring most of the computer time. For example, we estimate 10000 samples of the hexanol+Ar potential to require 5000 hrs, though we likely have to "waste" processors due to I/O limitations, thus increasing the core-hour cost by up to a factor of 3. Altogether, the quantum chemistry costs should require: 3 classes * 9 cases/class * 10,000 core-hours/case = 270,000 core-hours. In the second step, for each of the 27 cases, we will run 10,000 trajectories. These take 30 minutes, on average, for a total of 135,000 core-hours. The trajectory calculations run at near 100% parallel efficiency and scaling. Steps 1 and 2 thus require a total of 405,000 core-hours, to which we add 20% for testing and development, leading to a total request of 486,000 core-hours. Industry partnership: None Project URL: Requested allocation: 486000 Q1: 0 Q2: 86000 Q3: 200000 Q4: 200000 Justification: Storage requirements: Less than 1 TB The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System