Jeffrey S. Scroggs

A forward-trajectory global semi-Lagrangian transport scheme

A forward-trajectory semi-Lagrangian scheme for advection on the surface of the sphere is proposed. The advection scheme utilizes the forward (downstream) trajectory originating at Eulerian grid points and cascade interpolation, a sequence of 1D interpolations, to transfer data from the downstream Lagrangian points to the Eulerian points. A new and more accurate algorithm determines pole values. The resulting forward-trajectory semi-Lagrangian scheme can easily incorporate high-order trajectory integration methods. This avoids the standard iterative process in a typical backward-trajectory scheme. Two third-order accurate schemes and a second-order accurate scheme are presented. A mass-conservative version of the forward-trajectory semi-Lagrangian scheme is also derived within the cascade interpolation framework. Mass from a Lagrangian cell is transferred to the corresponding Eulerian cell with two 1D remappings through an intermediate cell system. Mass in the polar region is redistributed by way of an efficient local approximation. The resulting scheme is globally conservative, but restricted to meridional Courant number, Cθ ≤ 1.

A Global Nonhydrostatic Semi-Lagrangian Atmospheric Model without Orography

Abstract A global nonhydrostatic semi-implicit semi-Lagrangian (SISL) atmospheric model with orography has been developed. The height-based terrain-following σz coordinate of Gal-Chen and Somerville is used to incorporate the orography. A 3D vector form of the SISL formulation is proposed. It is based on the complete Navier–Stokes equations. The model is stable for large time steps of up to 1 h at horizontal/vertical resolution of 2.8125°/1200 m. Isolated bell-shaped mountain profiles and real orography are employed to evaluate the model performance. The sensitivity of the model with orography to the order of accuracy of the uncentering scheme, the reference temperature (T), and size of the time step are similar to that of the model without orography described in Semazzi et al. The authors find that for successful execution of the model, it is important that the orographic height Zs, the reference state mass variable (qs), and T satisfy the hydrostatic balance relationship in the terrain-following σz coor...

Optimal design of a high pressure organometallic chemical vapor deposition reactor

A team composed of material scientists, physicists, and applied mathematicians have used computer simulations as a fundamental design tool in developing a new prototype High Pressure Organometallic Chemical Vapor Deposition (HPOMCVD) reactor for use in thin film crystal growth. Early design of the HPOMCVD reactor dramatically evolved long before any physical reactor was built. This effort offers a strong endorsement of such multidisciplinary, computationally based modeling teams in the design of new products in areas of emerging technologies where heretofore extensive and costly experimental design was the central paradigm.

Simulation of a vertical reactor for high pressure organometallic chemical vapor deposition

The suitability of a vertical cylindrical reactor with highly constrained radial flow from a central gas injection port past a set of heated substrate wafers that are embedded in the top channel wall has been evaluated in the context of organometallic chemical vapor deposition (OMCVD) at elevated pressure. Numerical simulations showed that, in addition to the limitation on the channel height necessary for preventing buoyancy driven recirculation, negotiating the ninety-degree bend at the inlet is problematic and also constrains the channel height below a critical value, at which the radial flow area after the inlet bend is equal to the cross-sectional area of the central gas injection port. Restricting the channel height poses the danger of heating of the channel wall opposite to the substrate wafers causing potential problems with deposition of decomposition products and competitive polycrystalline film growth at this location. These problems can be avoided by actively cooling the channel wall opposite to the substrate and by keeping the retention time of the source vapor molecules and fragments thereof in the wafer location below a critical value.