The next few topics describe the tasks you perform to solve a 2D transient magnetic analysis.
To specify the analysis type, do either of the following:
Select menu path .
If this is a new analysis, issue the ANTYPE,TRANSIENT,NEW command.
To restart a previous analysis (for example, to restart an unconverged solution or to specify additional excitation), issue ANTYPE,TRANSIENT,REST. In a nonlinear transient magnetic analysis, the default behavior is multiframe restart; for more information, see the RESCONTROL command.
Next, you define which solution method and which solver you want to use.
To select a solution method, use either of the following:
Transient magnetic analyses require the full solution method.
To select an equation solver, use either of the following:
You can choose any of the following solvers:
Sparse solver (default)
Jacobi Conjugate Gradient (JCG) solver
Incomplete Cholesky Conjugate Gradient (ICCG) solver
Preconditioned Conjugate Gradient (PCG) solver
Voltage-fed models or models including velocity effects produce unsymmetric matrices, and can use only the sparse solver, the JCG solver, or the ICCG solver. Circuit-fed models can only use the sparse solver.
Load step options include the following:
Time option
This option specifies time at the end of the load step.
Command(s): TIMEGUI:
Number of substeps or time step size
The integration time step is the time increment used in the time integration scheme. You can specify it directly via the DELTIM command or its equivalent menu path, or indirectly via NSUBST or its menu path equivalent.
Time step size determines the accuracy of your solution. The smaller the time step size, the higher the accuracy. The size of the first integration time step following any large step change in loading conditions is especially critical. You can reduce inaccuracies such as thermal overshoot by reducing the integration time step size.
Caution: Avoid using extremely small time steps, especially when establishing initial conditions. Very small numbers can cause calculation errors. For instance, in a analysis with a time scale of unity, time steps smaller that 1E-10 can cause numerical errors.
If you step-apply loads, Mechanical APDL applies the entire load value at the first substep and holds it constant for the remainder of the load step. If you ramp loads (the default), the program increments the load values at each substep.
Automatic Time Stepping
Also called time step optimization in a transient analysis, automatic time stepping allows Mechanical APDL to determine the size of load increments between substeps. It also increases or decreases the time step size during solution, depending on how the model responds.
For most problems, you should turn on automatic time stepping and set upper and lower limits for the integration time step. The limits help to control how much the time step varies. However, time step optimization is not available for the CURR degree of freedom (voltage-fed conductors) or the EMF degree of freedom (circuit-fed models).
Command(s): AUTOTSGUI:
Specify nonlinear load options only if nonlinearities are present. Nonlinear options include the following:
Newton-Raphson Options
These options specify how often the tangent matrix is updated during a nonlinear solution. Available options are:
Program-chosen (default)
Full
Modified
Initial-stiffness.
For a nonlinear analysis, the full Newton-Raphson option is recommended. The adaptive descent option may help convergence in transient problems. To specify Newton-Raphson options, use either of the following:
Command(s): NROPTGUI:Number of equilibrium iterations
This option obtains a converged solution at each substep. The default is up to 25 equilibrium iterations, but you may need to increase the number depending on the degree of nonlinearity. For linear transient analysis, specify one iteration.
Command(s): NEQITGUI:
Convergence tolerances
Mechanical APDL considers a nonlinear solution to be converged whenever specified convergence criteria are met. Convergence checking may be based on magnetic potential (A), magnetic current segment (CSG), or both. You specify a typical value for the desired item (
VALUEfield) and a tolerance about the typical value (TOLERfield). The program then calculates the convergence criterion viaVALUE*TOLER. For example, it you specify 5000 as the typical value for magnetic current segment and 0.001 as the tolerance, the convergence criterion for magnetic flux would be 5.0.Ansys, Inc. recommends that
VALUEbe left to the default (program-calculated) and thatTOLERbe set to 1.0E-3.For potentials, the program compares the change in nodal potentials between successive equilibrium iterations ΔA = Ai - Ai-1 to the convergence criterion.
For a magnetic current segment, the program compares the out-of-balance load vector to the convergence criterion. If the solution does not converge within the specified number of equilibrium iterations, the program either stops or moves on to the next load step, depending on whether you activated the option to terminate an unconverged solution (see below).
Command(s): CNVTOLGUI:
Terminate an unconverged solution
If the solution does not converge within the specified number of equilibrium iterations, the program either stops the solution or moves on to the next load step, depending on what you specify as the stopping criteria.
This class of load step options enables you to control output. Output controls options are as follows:
Control printed output
This option enables you to include any results data in the printed output file (Jobname.OUT).
Command(s): OUTPRGUI:
Control database and results file output
This option controls what data goes to the results file (Jobname.rmg).
By default, the program writes only the last substep of each load step to the results file. If you want all substeps on the results file, specify a write frequency of ALL or 1 (every loadstep).
Command(s): OUTRESGUI:
Use the SAVE_DB button on the Toolbar to save a backup copy of the database. This enables you to retrieve your model should your computer fail while analysis is in progress. To retrieve a model, re-enter Mechanical APDL and use one of the following:
In this step, you initiate the solution for all loads steps using one of the following: