No abstract

This thesis presents design and operation of a newly developed virtual thermostat to control any desired local temperature in an HVAC controlled space. Conventional HVAC thermostats use a single point temperature sensor at one fixed location in a controlled space to control the temperature of the whole space. The single point temperature sensor does not represent the entire controlled domain. In addition, in most applications only a small zone in the controlled space needs to be controlled (for example where the people are located). Heating and cooling of the unnecessary zones lead to extra energy consumption that can be saved if only the required zones are controlled. Such a control system requires knowledge of the temperature distribution in the whole zone at all times. In this work, we have used computational fluid dynamics (CFD) to determine the distributive temperature inside the controlled space at all times and turn the HVAC system ON and OFF based on the temperature of the desired zone in the space. In this study, we have determined the energy savings by comparing the energy consumption of a conventional thermostat controlled system with a locally controlled one. Several cases with different heat losses, and different inlet air velocities and directions are studied. Energy savings of up to 22% is realized for the cases studied here. In addition, the study showed the effect of person's location on the energy saving using the local control. One of the main problems with such a control system is that the CFD calculations may take a long time (10-24 hours depending on the complexity of the iii problem), which is not practical for real-time temperature control. In order to reduce the computational times and simulate various operating conditions quickly, an artificial neural network (ANN) model is used to train the temperature control system. This resulted in reduction of the computational times to few seconds, making this system practical. This thesis, presents various elements of the newly developed virtual thermostat and its operating principals.iv

Carbon nanotubes (CNTs) have been reported to have excellent thermal and mechanical properties over the past two decades. However, the practical application of CNT-based technologies has been limited, due to the inability to transform the excellent properties of single CNTs into macroscopic applications. CNT network structure connects CNTs and can be possibly scaled up to macro-scale CNT-based application. In this paper, nonequilibrium molecular dynamics is applied to investigate thermal transport across two CNTs connected longitudinally by molecular linkers. We show the effect of different types and lengths of molecular linkers on interfacial thermal conductance. We also analyze the density of vibrational normal modes to further understand the interfacial thermal conductance between different molecular linkers and CNTs. These results provide guidance for choosing molecular linkers to build up large-scale CNT-based network structures.

No abstract