Large-scale processing, simulating

Figure 6.1 Computer applications in catalysis research range all the way from understanding the role of molecular active intermediates to large-scale process simulations.

The presentation in this paper concentrates on the use of large-scale numerical simulation in unraveling these questions for models of two-dimensional directional solidification in an imposed temperature gradient. The simplest models for transport and interfacial physics in these processes are presented in Section 2 along with a summary of the analytical results for the onset of the cellular instability. The finite-element analyses used in the numerical calculations are described in Section 3. Steady-state and time-dependent results for shallow cell near the onset of the instability are presented in Section 4. The issue of the presence of a fundamental mechanism for wavelength selection for deep cells is discussed in Section 5 in the context of calculations with varying spatial wavelength. 

Often it is possible to consider the process or plant, as a system of independent sub-sets or modules, which are then modelled individually and combined to form a description of the complete system. This technique is also used in the large scale commercial simulation software, in which various library sub-routines or modules for the differing plant elements, are combined into a composite simulation program. 

Models may be used for analyzing data, estimating performance, reactor scale-up, simulating start-up and shutdown behavior, and control. The level of detail in a model depends on the need, and this is often a balance between value and cost. Very elaborate models are justifiable and have been developed for certain widely practiced and large-scale processes, or for processes where operating conditions are especially critical. 

The application of large scale computer simulations in modeling fluidized bed coal gasifiers is discussed. In particular, we examine a model wherein multidimensional predictions of the internal gas dynamics, solid particle motion and chemical rate processes are possible. 

The successes enjoyed by nanosciences in many fields [2-10] have resulted in a need for adequate theory and large-scale numerical simulations in order to understand what the various roles are played by surface effects, edge effects, or bulk effects in nanomaterials. The dynamics of colloidal particle transport calls not only for passive transport, but also for additional processes such as agglomeration/dispersion, driven interfaces, adsorption to pore wall grains, and biofihn interactions [4,11-14]. In many cases, there is a dire need to investigate these multi-scale structures, ranging from nanometers to micrometers in complex geometries, such as in vascular and porous systems [4,15-17]. 

For case 1, a screening unit of high flexibility and availability is needed. The extraction volume can be small. For case 2 the amount of extract should be apt to determine the course of the extraction with time. It has proved that about 100 to 500 g of solid substrate is appropriate, meaning that the extraction vessel volume should be about 1000 cm If a gas cycle is added, a parameter study can be carried out on extraction and precipitation. For case 3 the equipment should be able to carry out the total process or to simulate all the process steps in sequence in the same manner as intended in a production process. This means that for purposes of screening and a parameter study the precipitation of caffeine in a decaffeination process can be carried out by adsorption on active charcoal or by absorption in water. But for demonstration of process principles, the individual steps have to be carried out by the same unit operation and with the same type of mass transfer equipment as intended for use in a large scale process. Otherwise, no indication of the joint operation can be obtained. For information on scale up, the extractor volume must be at least one order of magnitude larger than in the laboratory type of equipment. 

ReaxFF [50] provides a generally valid and accurate way to capture the barriers for various chemical reaction processes (allowed and forbidden reactions) into the force fields needed for large-scale MD simulation. ReaxFF is parameterized exclusively from QM calculations, and has been shown to reproduce the energy surfaces, structures, and reaction barriers for reactive systems at nearly the accuracy of QM but at costs nearly as low as conventional FFs. 

It should be noted that the hybrid quantum/classical schemes apply not only for determination of geometries, energies, and reaction mechanisms. The Monte Carlo [67, 68] and molecular dynamics (MD) [69-72] simulations are quite popular as frameworks for which various QM/MM procedures serve as subroutines . Before employing hybrid schemes the large-scale MD simulations were performed only with low-level approximations for force fields. The use of hybrid schemes extends significantly the scope of their application, improve precision of the results that allows to improve the understanding of statistical properties and dynamical processes in liquids and biopolymers. 

In actual applications, a simpler kinetic model is more favored for large-scale reactor simulation, suppose that it can adequately describe the reaction and transport process in the reactor. The simplification is normally based on the experimental findings. For example, Gayubo et al. (2000) reduced the number of reactions from eight to four by eliminating some reaction steps with slow reaction rates. Therefore, the appHcation scope of the macroscale reaction kinetics is highly dependent on the experimental conditions studied. 

Large-scale atomistic simulations [35, 36] of the plastic deformation of nanocrystalline materials suggest that both the inter- and intra-granular deformation processes under uniaxial tensile and compressive stress in nanoindentation are leading conditions. In the scoping studies [37, 38], various parametric effects on the stress state and kinematics have been quantified. The considered parameters... 

Simulating the crystallization process is a computational challenge, precisely because crystal nucleation is an activated process. This implies that the formation of small crystal nuclei in a supersaturated liquid is infrequent but, when it happens, the process is quite fast, i.e. it proceeds on a time scale that can be followed in a molecular simulation. For instance, experimentally measured nucleation rates are typically on the order of (9(10 ) to (9(10 ) nuclei per cm per sec. We can estimate the number of time steps needed in a molecular dynamics (MD) simulation to observe one nucleation event. In a large-scale computer simulation, it is feasible to study the dynamics of (9(10 ) particles, but the number of particles in a typical simulation is some two to three order of magnitude less. For an atomic liquid, the volume of a simulation box containing one million particles is of order (9(10 ) cm. If a million nuclei form per second in one cubic centimeter, then it will take, on average, 10 seconds for a nucleus to form in a system of a million particles. As the typical time step in a molecular simulation (MD) is on the order of femto seconds, this implies that it would take some 10 " MD time-steps to observe a single nucleation event under experimental conditions. 

---