Number prohibitive computational times

In a VOC-NO mixture containing many different organics, the number of reactions becomes unmanageable for application in models used to describe an air basin or region. Thus the amount of computer time required for numerical integration of the rate equations associated with the thousands of individual species found in ambient air is prohibitive. Furthermore, even as computing power increases, in practice, the kinetics and mechanisms required as input are not all known. 

The results just discussed indicate that for bubble-point systems, the light-component K values are relatively insensitive to the number of subfractions (between 7 and 34) used for the heptanes plus, especially when the heptanes plus is characterized properly. This behavior has been noted for several oils and is important for two reasons (1) in a compositional reservoir model study, the number of components must be limited to 14-18 to keep computer time and costs from being prohibitively high and (2) when vaporization of an oil is being calculated, extremely high... 

In general, the development of approximation methods for the solution of the many-electron Schroedinger equation is a challenge for physicists because no exact numerical solutions can be found apart from very few cases of a small number of electrons, such as the helium atom. The main difficulty arises because of the electron-electron interaction, which is a two-particle operator. Thus, increasing the accuracy of solutions implies increasing the computer time needed for the numerical calculations, and the cost becomes prohibitive even for molecules with a few atoms. 

We saw in the previous section that the perturbation theoretical expressions governing two molecules (or linear chains) at medium distances (where a multipole expansion for the electrostatic term alone is insufficient) are rather complicated even in second order. On the other hand, perturbation theory in this form cannot describe the simultaneous interactions of more than two molecules (only with the aid of the still more complicated double perturbation theory), and it is also not very accurate. Therefore one must develop a new method, which is nearly as accurate as the supermolecule approach (which, for larger interacting molecules, is not feasible because of the prohibitively large amount of computer time), can treat an arbitrary number of interacting molecules (or linear chains) at medium intermolecular (interchain) distances, and is much faster than perturbation theory (PT). This problem was solved at Erlangen in a series of papers for both molecules and linear chains. 

See Pope (2000) for an alternative estimate of the time-step scaling based on the Courant number. The overall conclusion, however, remains the same DNS is computationally prohibitive for high Reynolds numbers. 

Table I (taken from Martin, Sykes, and Hioe16) contains the most recent exact enumerations of C for the triangular and fee lattices. Similar enumerations for other lattices have been given elsewhere ° 11 numerical analysis indicates that the close packed lattices lead to most rapid convergence, and these were therefore selected for an extensive enumeration project. It should be noted that C12 for the fee lattice is of order 1.8 x 1012. Using a direct enumeration procedure on a digital computer, the machine time required would be quite prohibitive. It is only by the way of sophisticated counting theorems17 and skilled programming that these numbers could be obtained.

Liquid-liquid interfacial tensions can in principle also be obtained by simulations, but for the time being, the technical problems are prohibitive. Benjamin studied the dynamics of the water-1,2-dichloroethane interface in connection with a study of transfer rates across the interface, but gave no interfacial tensions. In a subsequent study the interface between nonane and water was simulated by MD, with some emphasis on the dynamics. Nonane appears to orient relatively flat towards water. The same trend, but weaker, was found with respect to vapour. Water dipoles adjacent to nonane adsorb about flat, with a broad distribution the ordering is a few molecular layers deep. Fukunishi et al. studied the octane-water Interface, but with a very low number of molecules. Their approach differed somewhat from that taken in the simulations described previously they computed the potential of mean force for transferring a solute molecule to the interface. The interfacial tension was 57 11 mN m", which is in the proper range (experimental value 50.8) but of course not yet discriminative (for all hydrocarbons the interfacial tension with water is very similar). In an earlier study Linse investigated the benzene-water interface by MC Simulation S He found that the water-benzene orientation in the interface was similar to that in dilute solution of benzene in water. At the interface the water dipoles tend to assume a parallel orientation. The author did not compute a x -potential. Obviously, there is much room for further developments. 

---