Modeling and Simulation of a Solar Energy System

Abstract

A Building Analysis Program is a modeling and simulation tool that determines the thermodynamic behavior of a building structure. It helps to detect the influence of materials, architectural features and air-conditioning systems during a building's design as well as in an existing building's improvement in terms of climate control.

A structure model is assembled from provided submodels, which are representations of the basic physical components comprising the entire structure model. For an accurate model description, the user must supply sufficient data describing the thermodynamic characteristics of the materials modeled, and must define the submodel interconnections.

This thesis deals with a new methodology for modeling a system's basic components and their interconnections so that an accurate model of the entire system can be feasibly constructed and simulated. The bond graph technique is used to model the physical laws of heat transfer between the basic components, and the modeling language Dymola is used to provide a convenient platform for creating hierarchical and modular component descriptions in very readable code. The bond graphs can be coded directly into Dymola, which can generate code for a simulation language such as ACSL. ACSL is then used to obtain the numerical solution of the equations which describe the system.

The goal of this work is to prove that continuous system simulation using bond graphs coded in Dymola are feasible for building analysis in terms of modeling capabilities, accuracy of the results, and computation time.

This new approach is compared with two state-of-the-art building analysis programs, namely CALPAS 3 and DOE-2. For the comparison, the thermal performance of a test house is determined by all three programs and the results for temperature profiles and heat flow rates have been found to be very similar. Dymola proves to be an excellent modeling tool and very suitable for modeling systems with modular components. However, the high accuracy achieved with continuous system simulation requires considerably longer computation times.


Interested in reading the entire thesis? (125 pages, 607,882 bytes, pdf)


Homepage


Last modified: February 16, 2006 -- © François Cellier