几何尺寸与公差论坛------致力于产品几何量公差标准GD&T (GDT:ASME)|New GPS(ISO)研究/CAD设计/CAM加工/CMM测量  


返回   几何尺寸与公差论坛------致力于产品几何量公差标准GD&T (GDT:ASME)|New GPS(ISO)研究/CAD设计/CAM加工/CMM测量 » 三维空间:产品设计或CAX软件使用 » CAD设计 » 产品功能分析
用户名
密码
注册 帮助 会员 日历 银行 搜索 今日新帖 标记论坛为已读


回复
 
主题工具 搜索本主题 显示模式
旧 2009-09-10, 10:37 AM   #1
huangyhg
超级版主
 
huangyhg的头像
 
注册日期: 04-03
帖子: 18592
精华: 36
现金: 249466 标准币
资产: 1080358888 标准币
huangyhg 向着好的方向发展
默认 limitations of software packages

limitations of software packages
hi guys,
i ran into trouble today when i attempted to model a bridge deck. firstly, i wanted to generate a moving load for the traffic conditions that were set out in the specification, this worked, but it generated a large amount of static load cases that i thought was totally unreasonable for the design. particularly when i need to consider different traffic loading, multiple lanes and combinations of lane loading.
secondly, my cpu ran out of memory when attempting to invert the stiffness matrix, 2gb.
does anyone else run into these problems in their daily accounts?
as i don't model bridges or traffic loads, i have not had your specific problem. but i do regularly get inundated with load cases in my building modeling software (etabs), where i can frequently end up with 100+ load combinations automatically generated. it slows it down a bit, but not so much that i worry too much about it.
as for your stiffness matrix issue, i have seen stiffness matrix in etabs reach almost 2gb, so i could see how someone could have problems.
my question is how much ram does your computer have? you may need to have enough to keep the entire matrix in memory, along with everything else. if you have only 1gb or 2gb, you may simply not have enough and need to upgrade. if you have 4gb, try stopping any unneccessary applications (such as email, virus scanners, etc) to try to free up some ram, then retry your analysis. are you running windows xp? if so, by default it can only take advantage of the first 3gb out of 4gb (or more). but there is a way to force it to use the ram more efficiently, by enabling what is called the "3gb switch". do a google search and you should find info on how to enable it. this will free up more ram for applications to use.
oh, one thing i forgot to mention, if you do use the 3gb switch, you basically are setting up a different windows boot-up option. make sure you set it up to give you a selector screen, where you can select either normal mode, or with 3gb switch enabled. (you can also add a safe mode too). some apps may not react well with the 3gb switch enabled. so if you have the option to select to use it or not at boot up, you can still get into normal windows mode without problem if you have to reboot.
sorry, but i forgot one more thing. the 3gb switch only applies to windows xp. if you are using vista, it can handle more ram by default, up to 8gb i think. so if that is the case, you could easily add more ram and that should fix your stiffness matrix problem.
this is a software issue, not a ram issue. good blocked solvers and sparse solvers can solve most any size model with limited ram. btw, etabs has good solvers. adding ram, switching to vista are viable band-aid solutions to get you through this particular model (hopefully) but improving the solver is the long term fix.
good advice, now if he can write his own improved solver, he will be all set. changing software is difficult at the best of times, and may be impossible for the op for various reasons.
if their tech support can't help him through this problem, his best bet may be to upgrade his computer, or try to simplify the model if possible.
personally i feel that i have over complicated the problem, there is no need for the structure i am analysing to refine it to such the degree i have.
i ran the job on my seniors computer that is quite faster than mine and it was able to perform the analysis no problem.
certainly i'm not suggesting he write his own solver or even that he switch software. what asixth should do is contact his current vendor and press them to improve their solvers. again, this is the long term solution. highly robust solvers can make modeling much easier.
__________________
借用达朗贝尔的名言:前进吧,你会得到信心!
[url="http://www.dimcax.com"]几何尺寸与公差标准[/url]
huangyhg离线中   回复时引用此帖
GDT自动化论坛(仅游客可见)
回复


主题工具 搜索本主题
搜索本主题:

高级搜索
显示模式

发帖规则
不可以发表新主题
不可以回复主题
不可以上传附件
不可以编辑您的帖子

vB 代码开启
[IMG]代码开启
HTML代码关闭

相似的主题
主题 主题发起者 论坛 回复 最后发表
discussion - should software producers be registered huangyhg 产品功能分析 0 2009-09-08 06:56 PM
design software help huangyhg 产品功能分析 0 2009-09-08 05:59 PM
analysis software huangyhg 产品功能分析 0 2009-09-07 11:40 AM
【转帖】windows 64-bit Platform yang686526 DirectDWG 0 2009-05-07 07:17 PM
careers:Senior Software Engineer loushan PC-DMIS 0 2007-01-04 08:53 PM


所有的时间均为北京时间。 现在的时间是 06:23 PM.


于2004年创办,几何尺寸与公差论坛"致力于产品几何量公差标准GD&T | GPS研究/CAD设计/CAM加工/CMM测量"。免责声明:论坛严禁发布色情反动言论及有关违反国家法律法规内容!情节严重者提供其IP,并配合相关部门进行严厉查处,若內容有涉及侵权,请立即联系我们QQ:44671734。注:此论坛须管理员验证方可发帖。
沪ICP备06057009号-2
更多