There are four parameters that define the event duration and time step size in a transient heat transfer analysis. These are specified in the Event section of the Analysis Parameters dialog. The analysis can be split into multiple time intervals, and each interval can use a different time step size. For example, immersing a hot piece of steel into a liquid bath results in rapid cooling at the very beginning of the analysis: use small time steps in this interval. As the part stabilizes, larger time steps can be used for faster solution.
In a transient heat transfer analysis, all the loads follow load curves. The load curves are defined in the Analysis Parameters dialog by pressing the Load Curves button. Refer to the table below to determine how the value in the Factor column of the load curve will affect each thermal load parameter.
Parameter | How Factor Affects Parameter |
---|---|
Applied temperature magnitude | See the page Applied Temperature |
Convection coefficient | Multiplies |
Convection ambient temperature | See the page Convection |
Radiation function | Multiplies |
Radiation ambient temperature | See the page Radiation |
Surface heat flux | Multiplies |
Internal heat generation | Multiplies |
Nodal heat source | Multiplies |
Body-to body radiation ambient temperature | Multiplies |
There are two methods that can be used to create the load curves. If you select the Piecewise Linear radio button in the Load Curve Form section, you will be able to define the load curve by defining several times and the corresponding factors. The factor values will be linearly interpolated between the defined times. If you select the Sinusoidal radio button in the Load Curve Form section, you will be able to define the parameters of the sinusoidal curve in the fields below.
There are four multipliers that will control the magnitudes of various thermal loads when they are applied to the model. These are located in the Multipliers tab of the Analysis Parameters dialog.
A value of zero in any of these fields except for the Boundary temperature multiplier field will disable the loads of that type in the model. A value of 0 in the Boundary temperature multiplier field will change the magnitudes of the applied temperatures to 0 degrees.
The value in the Default nodal temperatures field in the Options tab of the Analysis Parameters field will be used to define the initial temperature of any node that does not have an initial temperature defined. (Unless your transient analysis is starting at a temperature of zero, the temperature at the beginning of the analysis needs to be specified.)
The results of a previous thermal analysis can be used as the initial temperature profile for a transient heat transfer analysis. There are two methods to do this.
Select the appropriate analysis type in the Source of initial nodal temperatures drop-down box on the Options tab. Select the model in the Existing model field. If you are using another transient heat transfer analysis and want to use an intermediate time step, select the Specified option in the Time step from heat transfer analysis drop-down box and specify the time step in the Time step field. This capability provides the opportunity to do a restart analysis as follows. (See also Restarting an Analysis below for additional restart capabilities.)
In some cases, it may be desirable to split a transient analysis into separate analyses instead of trying to describe the entire load curve in one analysis. For example, several scenarios of cooling may need to be investigated after the part is heated. Thus, one file could be used to analyze the transient heating of the part, and then separate files could be used to analyze each cooling scenario. To accomplish this, the initial temperature for the second model (the restart analysis) is chosen as the final temperature from the first model. This is specified on the Options tab of the Analysis Parameters dialog.
The same procedure can be used repeatedly for a multi-phased transient analysis where parameters are changed from one analysis to the next. In this case, the procedure would be as follows:
When any of the parts have a phase change material model included, the calculation of the liquid fraction is chosen with the Liquid fraction relationship pull-down on the Options tab of the Analysis Parameters dialog. The options available are as follows:
Since the equilibrium partition coefficient and melt temperature are involved, this method is generally used only with alloys.
where
When the node is cooler than the solidus temperature, the fraction is equal to 0. When the node is warmer than the liquidus temperature, the fraction is equal to 1.
If you have previously run an electrostatic current and voltage analysis on a model with identical geometry and mesh to the model that you are currently analyzing, you can apply the electrostatic results to this model to determine the temperature effects of the current. There are a few guidelines that must be followed in this process.
The type of solver for a thermal analysis can be selected in the Type of solver drop-down box in the Solution tab of the Analysis Parameters dialog. See also Types of Solvers Available for background information. The options available are as follows:
If for some reason you want to create the solution matrix but not perform the analysis, activate the check box for Stop after stiffness calculations. The only time this would be useful is to access the equation number matrix. The stiffness matrix is always calculated when running an analysis, so there is no advantage to use this option in normal circumstances.
For the sparse and iterative solvers, the Percent memory allocation field controls how much of the available RAM is used to read the element data and to assemble the matrices. A small value is recommended. (When the value is less than or equal to 100%, the available physical memory is used. When the value of this input is greater than 100%, the memory allocation uses available physical and virtual memory.)
As listed above, some of the solvers take advantage of multiple threads/cores available on the computer. The drop-down Number of threads/cores control is enabled in such situations. You want to use all the threads/cores available for the fastest solution, but might choose to use fewer threads/cores if you need some computing power to run other applications at the same time as the analysis.
If the iterative solver is chosen, then the Iterative Solver section will be enabled. The input for this section is as follows:
If the Iterative (PBiCGStab) solver is chosen, then the following input will be enabled:
By default, models with body-to-body radiation are analyzed by coupling the solution of the temperature and radiation flux. For large models, this can result in a long analysis time. These two equations can be solved separately by selecting the Decoupled option in the Solution algorithm drop-down box. If this is selected, the solvers for each value can be selected individually using the Solver for temperature and Solver for radiation flux drop-down boxes. The Automatic option will select between the two available solvers based on model size.
If the sparse solver is chosen, then the Sparse Solver section will be enabled. The input for this section is as follows:
After the analysis is complete, the analysis results can be output to a text file. The Output tab of the Analysis Parameters dialog can be used to control the data that is output to this file.
If a transient heat transfer analysis has been previously performed on this model, you will be able to restart the analysis. This can be done if the analysis has finished or if the analysis was stopped before completion. First, you must activate the Perform restart analysis check box in the Restart tab of the Analysis Parameters dialog. To continue the analysis from the last time step available in the results, select the At last completed step option in the Begin restart analysis drop-down box. To continue the analysis from a different time step, select the At specified time option and then specify the desired time.
See also Default Nodal Temperatures above for an alternate method of performing a restart analysis.
If a model contains radiation loads or temperature dependent properties (convection, materials, and so on), the solution will involve a nonlinear iterative process to determine the appropriate temperatures. You can control this process using the Advanced tab of the Analysis Parameters dialog.
If any of the conditions mentioned above exist in your model, activate the Perform check box. You can control how many times the processor can iterate on the solution using the Maximum number of iterations field. The solution after this many iterations will be used as the analysis result. In some cases, an adequate solution will be converged upon before the maximum number of iterations.
There are five options to decide to stop the iterative process. These can be selected in the Criteria drop-down box. If the Do all N iterations option is selected, all the iterations specified in the Maximum number of iterations field will be performed. If the Stop when corrective norm < E1 (case 1)option is selected, the iterations will stop when the corrective norm is less than the value in the Corrective tolerance field. If the Stop when relative norm < E2 (case 2) option is selected, the iterations will stop when the relative norm is less than the value in the Relative tolerance field. If the Stop when either case 1 or 2 option is selected, the iterations will stop when either the corrective norm is less than the value in the Corrective tolerance field or the relative norm is less than the value in the Relative tolerance field. If the Stop when both case 1 and 2 option is selected, the iterations will stop when the corrective norm is less than the value in the Corrective tolerance field and the relative norm is less than the value in the Relative tolerance field.
There are two or four values that are calculated to determine the quality of the convergence. The first value is the corrective norm. The corrective norm is calculated as:
where numnp is the total number of nodes in the model, T old is the temperature from the previous iteration, and T new is the iteration from the current iteration. This norm is related to the temperature difference between iterations; hence the name corrective norm.
The second value for the convergence is the relative norm. This is calculated as:
This norm is similar to the relative change in temperature between iterations; hence the name relative norm. The Corrective tolerance field can be used to define the maximum value for the corrective norm. The Relative tolerance field can be used to define the maximum value for the relative norm.
When the analysis includes body-to-body radiation, convergence is also based on heat flux; when the analysis includes phase change, convergence is also based on the liquid fraction. The equations for the corrective and relative norms based on heat flux or liquid fraction are the same as the above (replacing the temperatures with the heat fluxes or liquid fraction).
The temperature of a node after an iteration, T new , could be higher or lower than the final converged value. Thus, you may not want to use this value as the input for the next iteration. The value in the Relaxation parameter field can be used to minimize these oscillations. The relaxation parameter is used as follows:
T cur = T old + (relaxation parameter)*(T new - T old )
In a graphical format, the relationship between the different results can be shown as Figure 1. T cur , and therefore the effect of the relaxation parameter, is the value output to the results file.
Figure 1: Graphical Interpretation of Relaxation Parameter
A relaxation parameter between 0.8 and 1 will provide good convergence when the nonlinear effects are small. When large nonlinear effects are present such as radiation at high temperatures the relaxation parameter may need to be on the order of 0.1 to 0.3 to smooth the oscillations. The convergence history can be checked in the log file.
Some of the analysis time in a transient heat transfer analysis is spent creating the stiffness matrix at each time step. In some cases, this matrix does not need to be recalculated at every time step; reformulating it at every other time step may be sufficient (and save some calculation time). In a few cases, it is not necessary to update the matrix at all.
How often the matrix is recalculated is done using the Number of time steps between matrix reformulation field in the Advanced tab. The following factors require the matrix to be recalculated often (every step or every other step) to get accurate results:
If set to 0, the matrix is calculated at the beginning of the analysis and not updated thereafter.
There are two methods of handling bonded connections. Which method is used depends in part on whether the nodes are matched between the two parts or not matched.
Activating the option Enable smart bonded/welded contact will use multi-point constraint equations (MPCs) when necessary to bond the nodes on part A, surface B with the nearest nodes on part C, surface D. Shape functions interpolate the temperatures at the nodes on surface B to the nodes on surface D. Therefore, the meshes do not need to match between the parts. The MPCs are used for all the nodes on the surface contact pair whenever any node does not match. If the meshes do match at all nodes, then node matching will be used to bond the contact surface. The two vertices on the adjoining parts are collapsed to one node, and MPC equations are not used for the contacting surfaces. The options for the smart bonding drop-down are as follows:
The smart bonding option applies to bonded contact and welded contact. Other types of contact (except for Free) require the nodes to be matched. (See the page Meshing Overview: Creating Contact Pairs: Types of Contact for a discussion of defining contact and using smart bonding.
By default, smart bonding uses the condensation method to solve your analysis. If you find your analysis doesn't converge or is not performing as you expect, you can try a different Solution method to use with MPC equations (see Multi-Point Constraints). Click SetupLoads
Multi-Point Constraint and choose from the Solution method options. If you use the Penalty Method, the accuracy of the solution is controlled by the Penalty multiplier field. The Penalty multiplier, times the maximum diagonal stiffness in the model, is used during the penalty solution. A value in the range of 10
4
to 10
6
is recommended.
When the option Enable smart bonded/welded contact is not activated, the parts will be bonded only if the nodes match between the parts.