Projects

Here at MSAL, we have created and continue to grow two codes: MDAG and L-HVAC. The purpose of these codes is twofold: 1) to create a tool that allows for state-of-the-art analysis of multiscale systems, and 2) to use as a teaching tool for advanced code development, debugging, and maintenance. Details on these codes are provided below. Also listed is a theoretical analysis performed through the laboratory - an extension of the van der Waals theory of capillarity. Investigators are listed in parentheses where applicable.

• Thermal Transport in Nano-enhanced Phase Change Materials website

Molecular Dynamics for Arbitrary Geometries (MDAG)

Molecular Dynamics for Arbitrary Geometries (MDAG) is a parallel classical molecular dynamics (MD) tool for investigation of a wide variety of nanoscale systems. The code is developed using a simple keyword format with zone-based molecular initialization that allows for implementation of a MD simulation of thousands of molecules with a few lines of input deck. Some features of the code are as follows:

1. Trivial parallel implementation using the Message Passing Interface (MPI) and GPU (CUDA) platforms.
2. Traditional or linked-cell force calculation
3. Electrostatic force calculation via Ewald summation
4. Trivial simulation restart capability
5. System visualization via opendx software
6. Hexahedral zone-based initialization (input deck creation via a hexahedral mesh generation code like TrueGrid)

MDAG is developed in structured C. Parallel implementation is through MPI. Current analysis projects featuring MDAG are the following:

1. Heat transfer characteristics of nanodroplet impingement (G. Haas)
2. Thermal transport characteristics of graphite nanofibers (M. Khadem)

Current MDAG development projects are the following:

1. Zone-based temperature and bulk velocity initialization (G. Haas)
2. Eulerian and Lagrangian tracer particle initialization and output (G. Haas)
3. Andersen thermostat implementation for NVT ensembles (M. Khadem)
4. Transport coefficient calculation using Green-Kubo relations (A. Wemhoff)

NOTE: Some MDAG research efforts are sponsored by the National Science Foundation

Lumped HVAC (L-HVAC)

Lumped HVAC (L-HVAC) is a lumped parameter code used to predict moist airflow thermodynamic properties in a heating, ventilating, and air conditioning system (HVAC). The code performs nonlinear implicit coupled calculations of flow resistance, absolute humidity, coil calculations, psychrometrics, and energy transfer to obtain predictions of system air properties for both transient and steady-state systems. Some features of this code include:

1. Virtual controls: thermostats and flow regulators connected to output devices (control dampers, chiller work input, fan speed/power)
2. Energy use calculations by the entire HVAC system: chiller work input and fan/pump work input
3. Trivial simulation restart capability
4. Adaptability to nearly any HVAC system

The long-term project goal is to apply the predictive capability of L-HVAC to an optimization code for purposes of developing optimal control sequences to minimize the HVAC energy use of a system. L-HVAC is originally developed in MATLAB, but a migration to structured C is in progress (A. Wemhoff and E. Wroble)

NOTE: Some L-HVAC research efforts are sponsored by the Office of Naval Research

Extension of the Neoclassical Theory of Capillarity

Extension of the Neoclassical Theory of Capillarity: J. D. van der Waals originally developed a theoretical approach to interfacial tension prediction using excess free energy generation due to the presence of a finite-sized transition region between bulk liquid and vapor phases of a fluid. In his analysis, he applied the van der Waals equation of state to obtain his results. The predictions were recently improved by V. P. Carey at UC Berkeley by applying the Redlich-Kwong fluid model to this analysis. At MSAL, we applied the two most advanced cubic equations of state known - Soave-Redlich-Kwong and Peng-Robinson - to the theory of capillarity but found out that the increased complexity in the latter two models provided worse predictions. Further investigation into this issue found that the reason is that all models overpredict vapor density, and these overpredictions act to adjust the calcualted surface tension to a reasonable value. The better vapor density predictions by the advanced models reduce this adjustment and therefore provide less accurate predictions of surface tension. In addition, relations were created that allow for reasonable estimates of surface tension using the advanced models. (A. Wemhoff)