FloTHERM收敛控制.ppt

2010 Mentor Graphics Corp.Company Confidential,Click to edit Master text styles,Second level,Third level,Fourth level,Click to edit Master title style,Your Initials,Presentation Title,Month Year,*,专业的电子设备散热分析软件,FloTHERM,收敛控制,Your Initials,Presentation Title,Month Year,1,残差曲线,收敛,发散,高位震荡,低位震荡,高位稳定,低位稳定,10,10,10,2,小技巧,:,通常情况下,如果收敛慢就需要用户终止计算进行调整模型,检查模型错误,比如在密闭系统中装有离心风扇或进入系统的热无法向外传递,在残差大的区域检查网格,网格不足无法捕获详细信息,检查本身不稳定性,-,用监控点来追踪不稳定区域,用监控点和残差场来分析低位震荡或低位稳定,-,通常不需要再做修改,3,收敛曲线发散,第一步检查,复查,Sanity Check,结果,复查建模过程,寻找明显网格不足问题,确定问题点,采用,Monitor Points,Deactivate Objects/Assemblies,用于隔离问题点,检查问题点,?,网格是否足够描述物理现象,?,是否有荒谬的类似热源,粗糙度,表面属性等设置,?,单位是否正确,?,切记,:,如果曲线发散的很明显,结果一定是不正确的,.,对于发散问题重新求解时,一定要重新初始化,4,低位稳定和震荡,打开残差场存储设置,在,Project Manager,中,Solve/Overall Control,设置,残差存保留储每个单元格的残差参数,可以在,FLOMOTION,中检查确定最大和最小值的位置,同时打开网格显示来判断问题是否与网格有关联,如果发现残差较大的位置不在所关注的范围内,就不需要再对模型再做修改使其收敛,如果监控点已经达到稳态,残差常可以帮助发现劣质网格问题,5,Hint:,Storing the error field will slow things down.Only activate the error field storage after stopping the solution.Then,re-run the solution for 1 iteration.,The,error field can only be stored for one variable at a time.If all velocity directions or continuity are oscillating,activate the error field storage for pressure.If an individual velocity is oscillating,activate the error field storage for that velocity.If temperature is oscillating,activate the error field storage for temperature.,After solution,the Error field contains the local residual errors for the selected variable.By using this error field,the user can identify areas where the program may have difficulty in converging the equations.The user can then focus attention on the grid,project set-up and assumptions in the areas of difficulty.The quantities stored in the error field are the cell residuals divided by the termination residual.It is this normalized error distribution which may be displayed.For the pressure variable,the pressure residual is the continuity error in a grid cell,i.e.it is the difference between the mass flowing in and the mass flowing out of the cell.Display of the continuity error field will help to identify locations of the largest mass imbalances-it can sometimes be the case that all the error is located in one grid cell.Again,what is stored in the error field and what is therefore displayed is the residual errors normalized by the termination residual for pressure.,6,低位稳定和震荡,温度场残差,可以使用自动收敛设置,Solve/Overall Control,如果温度监控点在,30,步迭代保持在,0.5,度范围内波动和,如果温度收敛曲线降到,10,以下停止求解,7,改善收敛,改变求解选项,修正慢速温度收敛,松弛因子法,内部迭代,Fan,松弛控制,线性松弛控制,relaxation(,欠松弛,),Successive over,松弛控制,(,超松弛,),调整,False Time Step,松弛因子,改进高位稳定或震荡非收敛问题,8,求解器选项,求解器选择在,Solve/Overall,中更改,Segregated Conjugate Residual(,默认状态,),可以适用大多数分析状况,采用分离压力区域改善收敛问题,对于温度收敛问题,可以使用块校正法,首先查找分析具有大温度残差的固体模型,将其加入块校正组,在,Solve/Overall,中激活快校正求解,9,求解器选项,Multi Grid,求解器,在质量和动量求解收敛过程中和,Segregated Conjugate Residual,方法一样,采用,multi grid,技术加快求解温度场线性方程,可以很大的提高传热问题的收敛速度,在多求解器和局域化中,Multi Grid,求解器也可以使用,10,Notes:,There are two solver options available:,a)Segregated Conjugate Residual-generally significantly faster linear equation solver(the default)which should normally be used.For temperature convergence,block correction can be activated.Note that block correction will slow the solver down so make sure that you do not attach more than 20 block correction groups.,b)Multi Grid-is designed to choose an iteration procedure which uses multi grid acceleration to solve the linear equations for temperature.This option eliminates the necessity to use block correction in any form.For problems with conjugate heat transfer it can improve convergence and significantly reduce overall computation time.,How does Multi Grid solver speed up convergence?,The Multi Grid solver for temperature uses an algebraic multi grid method.The user defines a single block structured Cartesian grid as normal.For the temperature variable the solver considers calculating on different grids,successively coarsening the grid by a factor of 2 in each direction at every level.The solver chooses at which level it should solve the case and this changes dynamically throughout the solution.The cycle process is called a Flexible V cycle,which means that the solver chooses the solve level based on convergence performance,cycling up and down through the levels.,11,松弛控制法,内部迭代,(,在,Solve/Variable Control,中设置,),对于压力项可以采用大,(,等,),于,100,步,增加步数将会降低收敛速度,线性松弛控制,(,在,Solve/Solver Control,中设置,),通常设在,0.50.9,范围内,0:,冻结求解,;1:,无阻尼,Successive Over-relaxation(,在,Control/Solver,中设置,),很少使用,非常容易导致不稳定,慎用,建议在,1.01.5,范围,用于纯传导问题和高压系统,12,松弛控制法,Fan,松弛控制,在,Solve/Overall Control,中设置,有效值范围是,0.50.9,有助于改善工作点在风扇,PQ,中梯度较小位置的收敛速度,13,False Time Step,通过变量求解控制栏来调节,Solve/Variable Control,自动设置的值是基于时间特征常数,适用于大多数情况,用户可以针对每个变量设定不同的值,variable,使用滑标,直接输入值,如果,False Time Step,放大或缩小超过,50,倍容易导致 发散,切记要检查监控点以确认达到收敛,14,Notes:,The false time step sets a transient-like parameter which is used to damp or under-relax the solution of any variable except pressure.Set to correspond to time-scale of problem,e.g.(typical length)/(typical velocity).Larger values give light damping,small values give heavy damping.,Automatic sets False Time-steps based on fans in forced flow cases and the EFCV in natural convection cases.Press Calculate Automatic Values to see the automatic value.The slider bar can be used to quickly change the automatic setting by either dividing or multiplying it by a factor of up to 20.Press Calculate Automatic Values to see the effect.User Specified allows you to override the automatic settings manually.,Calculating the False Time Step,False-time step under-relaxation is the principal means that you will use to control convergence of the iterative algorithm.For a given variable,this under-relaxation provides an inertial effect which restrains the amount by which the value in each cell is permitted to change at each outer iteration.The inertial effect is proportional to the mass in the cell and inversely proportional to the false time-step set for the variable.Consequently,the smaller the false time-step the larger the inertial effect,and so the less the variable is permitted to change.,By default the program calculates an automatic false time-step which is used for all variables(the one exception being pressure to which the method is not applied).,15,False Time Step,调节,False Time Step,类似于调节阻尼,false time step,较大值,减少阻尼,false time step,较小值,-,增加阻尼,false time step,非常大值,无阻尼,false time step,非常小值,-,无变化,无法求解,False Time Step,举例,减少,False time Step(,提高阻尼,),到,10-50,目的在于改进高位震荡,增大,False Time Step(,减少阻尼,),到,10-50,目的在于改进高位稳态收敛问题,在复杂系统中,降低,False Time Step,有时会改善高位稳态收敛问题,16,Remember!,In general it is inadvisable to use extreme values for the false time steps.Very large values,where there is no damping,will often lead to divergence.Very small values which lock the variable should not be used.If the values can not change,the solution will not converge.,Hint:,Before making changes to the false time step to correct dramatic divergence,check the model and the grid for errors.You may also want to save the project with a different name so that the original solution remains available.It could become handy if changing the false time step makes the solution diverge!,17,18,William_he,Mobile:[phone],Tel:[phone],