Desarrollo de Requerimientos (RD)

In microreaetors, laminar flow is usually encountered due to small length scales. Therefore, heat and mass transfer can be fully characterised, facilitating reliable

description o f processes and improving the fidelity o f the mathematical models. Modelling o f microchemical systems can provide analysis o f specific systems, assist in the design o f new ones and evaluate potential performance advantages in relation to macroscale systems. However, the scale necessitates the modelling o f heat transfer within the solid boundaries as well and can present problems o f multiple length scales and complex geometries.

Wenka, et al (2000) modelled fluid flow and heat transfer in a cross flow microheat exchanger using 3-D numerical simulations. The effect o f wall material conductivity on heat transfer efficiency was investigated and it was found that low conductivity materials such as glass and ceramics result in higher heat transfer efficiency because they reduce axial heat transfer (Stief, et al, 2000). Hardt, et al (2000) suggested three scenarios for improving heat exchanger performance, namely suppression o f axial heat transfer by including low thermal conductivity structures in high conductivity walls, meandering flow channels which increase the heat transfer area, and mixing at the inlet with fins which result in vortices and higher heat transfer. CFD simulations showed the last design to be the most effective, while further simulations o f combined reactors and heat exchangers proved the superior performance o f these devices compared to conventional fixed bed technology. Luo, et al (1999) also found that introducing mixing within the flow channels can improve the performance o f compact heat exchangers with channels in the mm range. Alépée, et al (1999), using Finite Element Modelling (FEM), found improved temperature uniformity in a double-side heated reaction channel for dehydrogenations compared to when only one reactor side was heated.

The velocity distribution from a single inlet to a number o f parallel microchannels on a plate was modelled by Commenge, et al (2000), using a simplified system geometry and mass and pressure drop balances. This model allows optimisation o f plate geometry so that equal velocities are obtained in each channel. The effect o f channel dimensions and diffusion coefficients on axial diffusion and residence time distribution (RTD) was studied by Walter and Liauw (2000) using CFD, who commented that when many reactants are present, the optimum conditions for a narrow RTD for one reactant may be different from that o f the others.

Bibby, et al (1998) used a finite volume CFD code to model the extraction between two immiscible liquids that flowed in separate channels and came in contact through a small slit. This configuration allows easy separation o f the two phases at the end o f the process. For the interface, mass transfer equilibrium partition coefficients were used, and the results o f Fe"^ fractional transfer compared well with experimental data (Shaw et al., 1998). Extraction between liquids separated by a perforated plate was modelled by TeGrotenhuis, et al (1998) by solving a convection-diffusion equation and using an equilibrium partition coefficient. For channel depths under 300 pm it was found that the resistance o f the separating plate became significant. Rector and Palmer (1999), using the Lattice-Boltzmann method, modelled diffusion during co-current flow o f two immiscible liquids in a microchannel; the liquids were either in direct contact or separated by a porous contactor plate. Mass transfer was found to be slower in the latter case due to decreased contact area between the two phases.

Angeli, et al (1999) studied gas-liquid hydrogenation reactions under isothermal conditions in a 2-D microchannel using CFD. Hydrogen was allowed to diffuse into

the liquid phase where the reaction took place, through an appropriate wall boundary condition. Stepanek and Marek (1999) used a systems approach to model a combined reactor-separator where an immobilised ion-exchange resin separated the product and increased yield. Biological reactions in microchannels have been simulated by CFD (Makhijani, et al, 1999), and by a diffusion-based mass transfer model (Dickey, et al,

2000).

Fedorov and Viskanta (1999) modelled wall adsorption during flow in 2-D parallel plates including both heat and mass transfer. For the wall they used both a no-slip and a slip boundary condition. The inclusion o f slip reduced the rates o f momentum and heat transfer between the gas and the wall; the former resulted in reduced pressure drop while the latter caused the gas temperature in the simulation to increase faster compared to no-slip flow. The inclusion o f adsorption caused the velocity, temperature and adsorbed species concentration within the channel to oscillate, a behaviour not seen in large scale systems. The authors attributed this to the much smaller characteristic time scales for mass transport. A finite difference model that included material and energy balances was used for the study o f dehydrogenation o f cyclohexane to benzene in microreaetors and demonstrated that Knudsen flow, which existed near the reaction channel exit, resulted in higher conversions than slip flow (Jones, et al, 2000).

Stief and Langer (2000) solved the balance equations in a non-homogeneous catalytic gas phase reactor for ethylene oxidation in order to investigate the effect o f periodic reactant feed and heating on conversion and selectivity. Large amounts o f wall material resulted in high thermal inertia and slower response to temperature changes

compared to concentration changes. For the combination o f reaction and reactor used, periodic operation actually decreased conversion. Zheng, et al (2000) modelled a membrane reactor for cyclohexane dehydrogenation where a Pd membrane selectively removed H2 from the product mixture, by dividing the reaction channels in slices and solving materials and energy balances. Hsing, et al (2000) used finite element simulations to model a T-reactor combined with heaters. As a model reaction they used Pt-catalysed ammonia oxidation. 3-D flow and heat transfer simulations were used to acquire effective heat transfer coefficients, which were then used in a reduced 2-D model that included reaction kinetics. The inclusion o f the catalyst layer, the reactor wall and the heaters presented problems o f large variations in characteristic lengths within the computational domain. To overcome this, the three different wall layers were reduced to one wall boundary condition, using Biot number analysis. For the particular reactor configuration modelled, it was found that the reaction was mass transfer limited, while ignition occurred downstream and the reaction front subsequently travelled upstream. Quiram, et al (2000) used the same model to design a micro-flow sensor based on hot-film anemometry. In order to account for heat transfer in the transverse direction, which is not modelled in the 2-D simulation, they added an extra heat loss term in the wall boundary condition. Snita, et al (1999) modelled electrophoretic separation and electroosmosis in microchannels by solving the momentum, heat, mass and electrical charge balances. They also studied the propagation o f concentration profiles in channels under the influence o f an electrical field.