Thermal conductivity of liquids and gases through experiment and molecular dynamic simulations

Thermal conductivity is most commonly used in the design of heat exchangers, distillation columns and other energy calculation. Accurate thermal conductivities of heat transfer fluids are needed to obtain precise heat-transfer coefficients. These heat-transfer coefficients are then used to estimate the size and cost of the heat exchanger. It is essential that high-quality prediction methods are available for properties when there is an inadequate amount of experimental data. Liquid thermal conductivity (LTC) and vapor thermal conductivity (VTC) are two properties for which there typically is a shortage of experimental data. The purpose of this work is to develop a method that can accurately predict thermal conductivity values for a wide variety of pure compounds.; A two-fold approach has been used to achieve the purpose of this research. First, very few experimental data have been compiled for nitrogen- and sulfur-containing compounds. Therefore, the thermal conductivity of selected compounds that contain nitrogen or sulfur has been measured. With the addition of the experimental data, the DIPPR database contains accurate experimental data for a wide range of families of compounds. This broad data set has then been used to develop a new group contribution method (GCM) that predicts the thermal conductivity of a pure fluid. The LTC and VTC prediction methods were developed from a training set that included 3297 and 757 experimental data points, respectively. The resulting produced average absolute deviations (AAD) of 3.8% and 3.7% for LTC and VTC prediction methods, respectively when applied to the training sets.; Second, molecular dynamic (MD) simulations are used to predict the thermal conductivity of a fluid. A new procedure has been developed to predict thermal conductivity from MD simulations that is quick and applies to platonic compounds. The accuracy of this method is strongly dependent upon the potential model used to model the fluid. The predicted thermal conductivity of a fluid using MD simulations should be applicable to all families of molecules, and is only limited by the potential model used.; The MD method developed was used to calculated the thermal conductivity of argon. The simulation data compared favorably to the experimental data with the added benefit of less CPU expense over other MD methods that are used to predict thermal conductivity. The thermal conductivity of nitrogen and butane were also calculated using the MD method. The results of the nitrogen and butane simulations also compared favorably to experimental data.