关于adina计算温度场之后的应力场时间步匹配问题

LS_Terminator

问题描述：在adina-t中模拟瞬态温度场，
采用的是30步，每步长为1，然后同样采用30步，每步长为1来计算应力场，一切没有问题；
如果我改成60步，每步长0.5，然后同样采用60步，每步长0.5，计算应力场一切也没有问题；
但是我如果采用20步，每步长0.5，加上20步，每步长1，同样总时间是30，同样采用20步，每步长0.5，加上20步，每步长1计算应力场，就提示温度文件不匹配了。。 在viewoutput文件中发现我的第一步长竟然是1，可是我明明设置的是0.5啊，后来才变为1的，请问这个问题怎么解决呢？？

ps:网格一致，1，采用的是map文件,2，采用data read from file,3,read and interpolate 三种方式导入温度场，其中只有3 可以正常计算，我的时间步明明在温度场里面和结构场里面是一致的，为什么采用map 或者data read from file 不行呢？？

wild_field

也许是软件的bug？
什么版本算得？

tonnyw

1# LS_Terminator

How about uploading your model and let us take a look?

Thanks.

LS_Terminator

8.7.1 算的，我把我的温度场应力场命令流贴上，版主帮我看看。。

LS_Terminator

这是我的命令流 帮忙看看问题出在哪里？

tonnyw

1# LS_Terminator

I am surprised that you think you get correct results.
Let us look at the thermal transient analysis first. Here you define two time functions.
But I didn't see where you use time function 1. You do use time function 2 for the convection boundary condition. But at time equal to 0, the value should be 0 instead of 1.
I don't see you create the map file either.

You have similar problems for stress analysis too. It seems to me you have complete confusion about time functions and time steps. How about sorting out the thermal analysis first?

LS_Terminator

6# tonnyw 你好，第一，time function 1 是在给混凝土施加变化的内热率了。
第二，time function 2：首先，你说应该是0，我认为不对，原因，如果0时刻的值是0,1000时刻的值是1，那么在0~1000之间是线性上升的，和我施加的荷载不符，因为我施加的convection就是一个常量1*20度。
第三，我发的这个命令流里面没有采用map文件来导入温度场，因为网格一致，我采用的是data read from file的方式来导入的。

tonnyw

7# LS_Terminator
I see.

Why not choose Read and Interpolate option? This can avoid trouble.
Here you are using load steps with varying step size. It is better to interpolate.

If you really want to see if there is a bug in Adina you can use a very simple one without convection boundary conditions. For instance, a thin plate with two ends fixed subject to thermal expansion.

You also have problem in specifying initial condition. Right now you assume the initial temperature is zero. I would think the initial temperature should be the environment temperature which you specify in the convection boundary condition.

LS_Terminator

first，是的，差值是可以避免这个计算问题，但是我有一个想法就是，如果我选择差值的方法，那么这个步长的控制也就成为一个问题，1时刻的温度，2时刻的温度，1-2之间的差值为线性的 而实际上1-2之间的温度变化不一定是线性，那么这样在1-2时刻的应力场不对了1-2时刻之间的温度场肯定是有一点误差，不是偏小，就是偏大。
second，我定义初始温度为0度，也就是像混凝土为混凝土的入仓温度为0度，环境温度为20度，这个很正常啊，我看不出有什么问题，能不能说的仔细点？

tonnyw

9# LS_Terminator

I think we need to be clear about two points.

1. The real solution called exact solution is not known and the temperature most likely does not vary linearly within any time interval [t_n-1, t_n].

2. The finite element solution generated by the software does vary linearly within the time interval [t_n-1, t_n] since the transient thermal problem was solved with back-ward difference and the temperature within [t_n-1, t_n] is calculated by linear interpolation of the two solutions at t_n-1 and t_n.

It might happen that you want to calculate the stress at time instant 0.1 and the temperature is interpolated at 0.11. But I don't think it will bother you too much. You can always refine the time step size to see if there is significant change in the solution between the current step size and the previous one.

cesc315

继续关注  学到很多