Research On Applicability Of Calculation Methods For Conduction Transfer Function And Heat Transfer Characteristics Of Underground Structures

Conduction transfer function (CTF) is widely used to calculate conduction heal transfer in building cooling/heating loads and energy calculations. The limitation of a methodology possibly results in imprecise or false CTF coefficients. There are three popular methods, state-space (SS) method, direct root-finding (DRF) method and frequency-domain regression (FDR) method to calculate CTF coefficients. The second chapter of this dissertation investigated the applicability of three methods as Fourier number and thermal structure factor were varied and in detail explained the sources that introduce error in the CTF solutions. The results show that the calculation error of SS and DRF methods becomes increasing as the reciprocal of the product of Fourier number and thermal structure factor becomes increasing. The maximal error reaches17.34%and16.83%(single-layered slab),38.7%and-17.49%(three-layered slab),81.61%and-72.50%(six-layered slab). However, the absolution value of the calculation error of FDR method always remains within0.6%no matter how Fourier number and thermal structure factor are varied. Thus, FDR method is more robust and reliable than SS and DRF methods. And it is more practical to calculate CTF coefficients and may be a better choice to calculate the cooling/heating loads for building structures for the architect/designer.As above-ground components of the building thermal fabric become more energy efficient, the heat transfer between the building and the ground becomes relatively more important and can no longer be neglected. Thus, it is necessary to design and analyze adequately the ground-coupled heat transfer problems. Most of the former calculation methods are based on conventional numerical solutions such as finite difference, finite element or finite volume technique. These methods are based on the mesh or element. Therefore, it is very complex and difficult to apply and program. To overcome the problems, in this dissertation, the element-free Galerkin method (EFGM) was introduced in chapter3. In the EFGM, the construction of the approximation requires only nodal data—no element structure is needed. The numerical integration of the Galerkin weak form needs only simple integration cells—no element connectivity is needed. The EFGM requires no postprocessing for the output of field variables which are derivatives of the primary-dependent variables since these quantities are already very smooth. In this dissertation, the EFGM was utilized to cope with the complex two-dimensional ground-coupled heat transfer problems. The energy status and heat transfer characteristics of underground structures were analyzed and investigated in detail. They can be used as a reference for the engineer and designer. The work and main conclusions of this dissertation are as follows:1. In chpter3. firstly, the fundamental principles of the moving least square approximation, which was employed for the construction of the shape function in the EFGM. were introduced and derived in detail mathematically. Secondly, the selection principles of both weight function and domain of influence were discussed. The penalty function and Lagrange multiplier techniques were used to enforce the essential boundary conditions due to their simplicity. A fast algorithmic of the shape function and the derivative was introduced. Thirdly, the discretization of the governing equations by EFGM using both penalty function and Lagrange multiplier techniques was derived from the variational principle. Finally, Gauss-Legendre integration method was used to compute the integration of the Galerkin weak form. The process of the EFGM program was interpreted. Based on the MATLAB work desktop, the EFGM program for the ground-coupled heat transfer problems was developed.2. The effects of weight functions and the scaling parameter on the accuracy and convergence rate of the EFGM were discussed in detail by comparison with the analytical solutions of several thermal examples. The results show that the EFGM has very high accuracy and convergence rate. The quartic spline weight function is the most appropriate for heat conduction problems. When an appropriate penalty parameter values, the penalty function method has almost given the same accuracy as the Lagrange multiplier method. In the unsteady analysis, the CPU time of the penalty function method is less than the Lagrange multiplier method. When the scaling parameter increases, no matter what kind of weight function, the CPU time has increased substantially.3. The heat transfer model of underground structures involves the wall, floor, roof, foundation, footing, gravel sand and the soil layer. The influence of the water table line and thermal insulations on energy consumption of underground structures is taken into consideration.4. Based on the definition of basic distance of nodes, zone uniform distributing node method is put forward for generating discrete nodes in the EFGM. 5. In the steady-state analysis, the sensitivity analyses of the water table, far field boundary, soil thermal conductivity, outdoor surface wind speed and distance of the roof from the ground surface were carried out. The relationship between energy consumption of underground structures and correlative parameters was investigated in detail. Furthermore, the relationship between energy consumption of underground structures and insulation thickness, length and layout of the location was investigated in detail.6. In order to obtain initial conditions in the unsteady analysis, three cycles pre-calculation method was proposed. The third cycle quasi-steady distribution of temperature field was taken as initial conditions. The results show that this method is correct and reasonable.7. In the unsteady analysis, parameter sensitivity investigations show that indoor energy consumption of underground structures is very sensitive to the soil thermal conductivity. Therefore, the soil thermal conductivity must be determined cautiously.8. In the unsteady analysis, due to insulation, the absolute value of the amplitude and mean of the floor, wall and roof heat flux significantly decreased, and the time delay substantially longer.9. The amplitude of1year period harmonic is highest and the second is1day period in Beijing. Sinusoidal temperatures are all attenuated by the soil layer, especially for the shorter period and smaller amplitude waves. The deeper underground structure locates, the bigger the level of the attenuation is.