MULTISCALE MODELING AND SIMULATION IN INDUSTRY
New product and process development in an industrial setting requires attention to a variety of sometimes contradictory considerations. These considerations include formulation design, product performance targets, materials and processing costs, market trends, and governmental regulations. The relevant materials science also encompasses many phenomena. Hence a multidisciplinary and integrated modeling approach is desirable. The multidisciplinary nature of industrial modeling is illustrated schematically in Figure 1, which is reproduced from J. Bicerano et al., "Polymer Modeling at The Dow Chemical Company", J. Macromol. Sci. - Polymer Reviews, 44, 53-85, 2004.
Figure 1. Schematic illustration of multidisciplinary nature of industrial modeling.
Such work is most effective if it is performed in a strong and seamless partnership with experimentalists and if all factors relevant for commercial success (such as formulation design, product performance targets, materials and processing costs, market trends, and governmental regulations, as appropriate) are taken into account in a holistic manner.
Most work of this type requires the combined consideration of chemical and physical phenomena that occur at length scales and time scales that may differ by many orders of magnitude. In some instances, the length scales may range from ~10-10 m to >>1 m while the time scales may range from ~10-15 s to many years. Figure 2, which is reproduced from J. Bicerano, Prediction of Polymer Properties, third edition, Marcel Dekker, New York, 2002, illustrates the multiscale modeling paradigm that is the “holy grail” for such work. The system components may differ greatly among different types of systems modeled within the conceptual framework of this paradigm. For example, they may consist of any combination of polymers, molecular fluids, or solid particles. Different quantum mechanical methods, some ab initio and others semi-empirical, taking the electronic structure into account explicitly, may be used in calculations of chemical structures. Different force fields, optimized for different types of components, might be used in atomistic simulations on different systems. The interaction parameters may be defined somewhat differently in various types of mesoscale simulations. Self-consistent mean field theories and mesoscale simulation tools come in many flavors, some of which work better than others in helping understand specific types of multiphase materials. Finally, more than one simulation method (representing the system morphology at different levels of coarseness) exists for continuum simulations. However, this multiscale modeling paradigm holds regardless of the specific tools placed in its “method” boxes.
Figure 2. Schematic illustration of the multiscale modeling paradigm for predicting the morphologies and properties of many types of mixtures, solutions, dispersions, blends, block copolymers and composites; as well as for characterizing the interfaces between the different phases in such systems. The items enclosed in the boxes with thick solid borders are important portions of the input into and/or the output of the simulations. The simulation methods are enclosed in the boxes with thin dashed borders.
Another perspective into the nature of multiscale modeling is provided in Figure 3, which is adapted from J. Bicerano, Prediction of Polymer Properties, third edition, Marcel Dekker, New York, 2002. Heterophasic materials contain morphological features on length scales differing by many orders of magnitude, formed by the interplay between the thermodynamic driving force towards equilibration and the rate-dependent (dynamic or kinetic) phenomena on time scales differing by many orders of magnitude. These phenomena of vastly different length scales and time scales often have highly nonlinear interdependencies. In the past, the prediction of the properties of multiphase materials was limited severely by the inability to predict the morphology with sufficient relevant detail. This limitation forced researchers into making assumptions about the morphology as input for property calculations. There has been tremendous progress over the last two decades in developing the ability to predict multiphase material morphologies via multiscale modeling. This area is at the frontiers of theoretical and computational materials science. Continuing rapid advances in simulation methods, computer speed and software algorithm efficiency suggest a bright future ahead.
Figure 3. Schematic illustration of the length scales and time scales that are probed by simulation methods for modeling interfacial and phasic behavior. While the concept of the use of different types of simulation methods to probe different length scales and time scales will always remain valid, numbers are not shown along the length scale and time scale axes since the scales that can be probed with finer-scale methods will continue to expand with increasing speed of computers and the development of more efficient software algorithms.
For further discussions, see (1) J. Bicerano, Prediction of Polymer Properties, third edition, Marcel Dekker, New York, 2002; and (2) J. Bicerano et al., "Polymer Modeling at The Dow Chemical Company", J. Macromol. Sci. - Polymer Reviews, 44, 53-85, 2004.
