|
|
Advanced Inverse Heat Conduction Methods
Komínek, Jan ; Čarnogurská, Mária (referee) ; Hajduk,, Daniel (referee) ; Raudenský, Miroslav (advisor)
Numerical simulations of thermal processes are based on known geometry, material properties, initial and boundaries conditions. The massive use of these simulations in the metallurgical industry (for example for simulation of heat treatment of steel) is limited by the knowledge of precise boundary conditions, which are not easy to determine in compare to other input parameters. Empirical formulas are not sufficiently accurate for most non-trivial processes. Therefore, it is necessary to obtain the boundary conditions by experimental way. Boundary conditions can not be measured directly. The boundary conditions are determined by solving inverse heat conduction problem based on the measured temperature records. This doctoral thesis focuses on two types of the inverse heat conduction problems, which are poorly solved by existing methods. The first type are tasks that contains sharp increase/decrease in the values of the boundary conditions. Two new approaches are proposed and compared in this thesis for this type of tasks. The second type are tasks with non-stationary and non-homogeneous cooling. Three new methods were developed for this case. They are applied for the case of water cooling of vertical aluminum sample. The base characteristics of the current task is inhomogeneous cooling. One part of the surface is cooled intensively by flowing water in contrast to the other part of surface which is cooled only with low intensity since it is protected from direct contact with water by the vapor layer (Leidenfrost effect). The positions of these two part of surface are not stationary (they change during the experiment). The newly developed methods are compared to each other.
|
|
|
Development of inverse sub-domain method for boundary conditions computation of heat conduction
Hřibová, Veronika ; Klimeš, Lubomír (referee) ; Pohanka, Michal (advisor)
It is very important to develop efficient but still accurate and stable numerical methods for solving heat and mass transfer processes in many industrial applications. The thesis deals with an inverse heat conduction problem which is used to compute boundary conditions (temperatures, heat flux or heat transfer coefficient). Nowadays, two approaches are often used for inverse task - sequential estimation and whole domain estimation. The main goal of this work is to develop a new approach, the so-called sub-domain method, which emphasizes advantages just as reduce disadvantages of both methods mentioned above. This approach is then tested on generated prototypic data and on data from real experiments. All methods are compared with respect to accuracy of results as well as to computational efficiency.
|
|
|
Determination of thermal conductivity anisotropy of polymeric heatsinks for electronics
Brachna, Róbert ; Kůdelová, Tereza (referee) ; Komínek, Jan (advisor)
The master's thesis focuses on creating a numerical model of a polymeric heat sink with emphasis on its significant thermal conductivity anisotropy. This anisotropy is caused by highly thermally conductive graphite filler. Its final orientation is given by the melt flow inside the mould cavity during injection molding. The numerical model is created on the basis of a heat sink prototype subjected to experimental measurements, whose physical conditions are reliably replicated by the model. The determination of anisotropy is divided into two parts. The qualitative part is based on the fracture analysis of the heat sink prototype and determines the principal directions of the conductivity tensor in individual sections of the geometry. The computation of principal conductivities falls into the quantitative part, in which this task is formulated as an inverse heat conduction problem. The input data for the proposed task are experimentally obtained temperatures at different places of the geometry. The values of principal conductivities are optimized to minimize the difference between the measured and simulated temperatures.
|
|
|
Processing of temperature data for inverse heat conduction tasks
Brachna, Róbert ; Komínek, Jan (referee) ; Luks, Tomáš (advisor)
This bachelor's thesis deals with digital filters and noise removal from temperature measurements. The basic concept for the proper understanding of properties of filters is the discrete Fourier transform, which is illustrated on a given example. Next, the thesis considers linear filters and the design of basic types for noise reduction. An adaptive filter is designed by analyzing experimental data. This filter is subjected to further analysis using a simulated cooling process disrupted with artificially added noise and will be compared to other conventional filters. One criterion is to compare the curve of the filtered temperature to the simulated one. The second criterion is the reconstructed boundary condition, which is the output of the inverse heat conduction task.
|
|
|
Determination of thermal conductivity anisotropy of polymeric heatsinks for electronics
Brachna, Róbert ; Kůdelová, Tereza (referee) ; Komínek, Jan (advisor)
The master's thesis focuses on creating a numerical model of a polymeric heat sink with emphasis on its significant thermal conductivity anisotropy. This anisotropy is caused by highly thermally conductive graphite filler. Its final orientation is given by the melt flow inside the mould cavity during injection molding. The numerical model is created on the basis of a heat sink prototype subjected to experimental measurements, whose physical conditions are reliably replicated by the model. The determination of anisotropy is divided into two parts. The qualitative part is based on the fracture analysis of the heat sink prototype and determines the principal directions of the conductivity tensor in individual sections of the geometry. The computation of principal conductivities falls into the quantitative part, in which this task is formulated as an inverse heat conduction problem. The input data for the proposed task are experimentally obtained temperatures at different places of the geometry. The values of principal conductivities are optimized to minimize the difference between the measured and simulated temperatures.
|
|
|
Processing of temperature data for inverse heat conduction tasks
Brachna, Róbert ; Komínek, Jan (referee) ; Luks, Tomáš (advisor)
This bachelor's thesis deals with digital filters and noise removal from temperature measurements. The basic concept for the proper understanding of properties of filters is the discrete Fourier transform, which is illustrated on a given example. Next, the thesis considers linear filters and the design of basic types for noise reduction. An adaptive filter is designed by analyzing experimental data. This filter is subjected to further analysis using a simulated cooling process disrupted with artificially added noise and will be compared to other conventional filters. One criterion is to compare the curve of the filtered temperature to the simulated one. The second criterion is the reconstructed boundary condition, which is the output of the inverse heat conduction task.
|
|
|
Advanced Inverse Heat Conduction Methods
Komínek, Jan ; Čarnogurská, Mária (referee) ; Hajduk,, Daniel (referee) ; Raudenský, Miroslav (advisor)
Numerical simulations of thermal processes are based on known geometry, material properties, initial and boundaries conditions. The massive use of these simulations in the metallurgical industry (for example for simulation of heat treatment of steel) is limited by the knowledge of precise boundary conditions, which are not easy to determine in compare to other input parameters. Empirical formulas are not sufficiently accurate for most non-trivial processes. Therefore, it is necessary to obtain the boundary conditions by experimental way. Boundary conditions can not be measured directly. The boundary conditions are determined by solving inverse heat conduction problem based on the measured temperature records. This doctoral thesis focuses on two types of the inverse heat conduction problems, which are poorly solved by existing methods. The first type are tasks that contains sharp increase/decrease in the values of the boundary conditions. Two new approaches are proposed and compared in this thesis for this type of tasks. The second type are tasks with non-stationary and non-homogeneous cooling. Three new methods were developed for this case. They are applied for the case of water cooling of vertical aluminum sample. The base characteristics of the current task is inhomogeneous cooling. One part of the surface is cooled intensively by flowing water in contrast to the other part of surface which is cooled only with low intensity since it is protected from direct contact with water by the vapor layer (Leidenfrost effect). The positions of these two part of surface are not stationary (they change during the experiment). The newly developed methods are compared to each other.
|
|
|
Development of inverse sub-domain method for boundary conditions computation of heat conduction
Hřibová, Veronika ; Klimeš, Lubomír (referee) ; Pohanka, Michal (advisor)
It is very important to develop efficient but still accurate and stable numerical methods for solving heat and mass transfer processes in many industrial applications. The thesis deals with an inverse heat conduction problem which is used to compute boundary conditions (temperatures, heat flux or heat transfer coefficient). Nowadays, two approaches are often used for inverse task - sequential estimation and whole domain estimation. The main goal of this work is to develop a new approach, the so-called sub-domain method, which emphasizes advantages just as reduce disadvantages of both methods mentioned above. This approach is then tested on generated prototypic data and on data from real experiments. All methods are compared with respect to accuracy of results as well as to computational efficiency.
|
guest ::
