Journal
MACROMOLECULES
Volume 48, Issue 5, Pages 1606-1613Publisher
AMER CHEMICAL SOC
DOI: 10.1021/ma502525x
Keywords
-
Categories
Funding
- Army Research Office [W911NF-08-2-0058, W911NF-09-2-0071]
Ask authors/readers for more resources
Despite strong industrial interest, only recently, have quantitative predictions for entangled polymer rheology become possible. Major advances are achieved with an alternative approach to entanglement modeling slip-links. For example,- the discrete slip-link model (DSM) allows for equilibrium and highflciw,rate preclietions with a strong connection to a molecular basis. Slip-link -models are applicable to arbitrary composition and chain architectures, but usually incur high numerical cost. Recent advances in coarse-graining allowed for a speedup of order one hundred, but further improvements are required for the ultimate goal simulations of nenhomogeneots,flows., On the othefhand, it is possible to rely on advanced computational devices to obtain additional sp- eed-up. 'GPUs are one of the most rapidly developing kind of computational devices, and provide immense computational power, but also have some critical restrictions on numerical operations. Since slip-link models have a uniquelevel of description, Pin numerical implementations are designed from scratch, rather than using available modeling frame I ivorks. n this work, we preSent a numerical algorithm for.a variant- of DSM the elustered flied slip-link model fully implemented on CUDA. Because the model is a -Well-defined mathematical object, the completely new GPU-optimized, algorithm produces results indistinguishable from the earlier CPU implementation We believe the performance of the algorithm will allow simulations of complex, nonhomogeneous flows of entangled poly:mer melts and solutions on -a relatively small GPU computational eluster. The single GPIJ- version of code for -equilibrium and homogeneous flow calculations is available for free download online,: which allows homogeneous flow calculations on an inexpensive desktop. We discuss implementation details and test performance and show -a comparison of the results with,experiments.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available