From wangzhicheng at mail.etp.ac.cn Sat Aug 1 04:00:54 2009 From: wangzhicheng at mail.etp.ac.cn (=?gb2312?B?d2FuZ3poaWNoZW5n?=) Date: Sat, 1 Aug 2009 17:00:54 +0800 (CST) Subject: [Nek5000-users] =?gb2312?b?VGhyZWUgcXVlc3Rpb25z?= Message-ID: <4A740446.00009F.26348@app-03> Hi,developers Thank you for your Code. I'm really impressed with nek5000, although I have only known limited information as yet 1 The first question is about an error: I run the turbJet case and get this errors .......................................................................................................................................... Number of processors: 1 REAL wdsize : 8 INTEGER wdsize : 4 Beginning session: /home/wzc/Code/nek5_svn/examples/turbJet/jet.rea timer accuracy: 1.1205673E-06 sec read .rea file byte swap: F 6.543210 -2.9312772E+35 help: 128 1 600 9372 600 9372 help: 600 600 -1700420800 Error: This NEKTON Solver has been compiled for 600 10000 elements. The data file has dimensions 9372 9372. call exitt: dying ... backtrace(): obtained 9 stack frames. ./nek5000 [0x4d132a] ./nek5000 [0x520276] ./nek5000 [0x426cbc] ./nek5000 [0x4313b6] ./nek5000 [0x403aca] ./nek5000 [0x40338c] ./nek5000 [0x61eeee] /lib/libc.so.6(__libc_start_main+0xf4) [0x7fab919881c4] ./nek5000 [0x4032d9] At line 388 of file /home/wzc/Code/nek5_svn/trunk/nek/comm_mpi.f Fortran runtime error: Unexpected end of format string (3(A,1g13.5,A,/) ^ ................................................................................................................................... 2 What I intend to do is an accurate simulation of incompressible turbulent flow (with heat transfer) around cylinders, my question is :is nek5000 suitable for this simulation? If it is YES, can I use the mesh generate tools included in nek5000 to create mesh for this case? I know you are working with Cubit, but I think it is not a free mesh generator. 3 the last: Is there any theoretical manual for me to understand the code ? I have some back ground on FVM, FEM , right now I study the book "High-order method for incompressible fluid flow" Many thanks Zhicheng -------------- next part -------------- An HTML attachment was scrubbed... URL: From kerkemeier at lav.mavt.ethz.ch Sat Aug 1 09:26:52 2009 From: kerkemeier at lav.mavt.ethz.ch (Kerkemeier Stefan) Date: Sat, 1 Aug 2009 16:26:52 +0200 Subject: [Nek5000-users] Three questions Message-ID: Hi, nek tries to tell you, that the local size lelt is too small for the given number of elements and processors. If you have 9375 elements in total you probably need to run it in parallel. Your local size seems to be 600 elements so you need 16 proc to run the solver or you need to increase the local size lelt in the SIZEu file and recompile. Hth, Stefan --- original message --- From: wangzhicheng Subject: [Nek5000-users] Three questions Date: 1st August 2009 Time: 11:06:00 am From wangzhicheng at mail.etp.ac.cn Sat Aug 1 10:09:28 2009 From: wangzhicheng at mail.etp.ac.cn (=?utf-8?B?546L5b+X5oiQ?=) Date: Sat, 1 Aug 2009 23:09:28 +0800 Subject: [Nek5000-users] =?utf-8?q?Three_questions?= Message-ID: <200908012309278590249@mail.etp.ac.cn> Hi Stefan Thank you for your reply! There are two other issue in the previous email which I hope you can give me any adivice: 1 What I intend to do is an accurate simulation of incompressible turbulent flow (with heat transfer) around cylinders, my question is :is nek5000 suitable for this simulation? If it is YES, can I use the mesh generate tools included in nek5000 to create mesh for this case? I know you are working with Cubit, but I think it is not a free mesh generator. 2 Is there any theoretical manual for me to understand the code ? It seems very difficult for me to understand what exactly the program is doing without a guide. I have some back ground on FVM, FEM , right now I study the book "High-order method for incompressible fluid flow" and "Spectral/hp element methods for CFD" Regards Zhicheng 2009-08-01 ??? ???? Kerkemeier Stefan ????? 2009-08-01 22:26:26 ???? wangzhicheng ??? Nek5000-users at lists.mcs.anl.gov ??? RE: [Nek5000-users] Three questions Hi, nek tries to tell you, that the local size lelt is too small for the given number of elements and processors. If you have 9375 elements in total you probably need to run it in parallel. Your local size seems to be 600 elements so you need 16 proc to run the solver or you need to increase the local size lelt in the SIZEu file and recompile. Hth, Stefan --- original message --- From: wangzhicheng Subject: [Nek5000-users] Three questions Date: 1st August 2009 Time: 11:06:00 am -------------- next part -------------- An HTML attachment was scrubbed... URL: From kerkemeier at lav.mavt.ethz.ch Sat Aug 1 11:26:27 2009 From: kerkemeier at lav.mavt.ethz.ch (Kerkemeier Stefan) Date: Sat, 1 Aug 2009 18:26:27 +0200 Subject: [Nek5000-users] Three questions Message-ID: <013e01ca12c4$c088a86c$0d24a8c0@d.ethz.ch> Hi, 1: sure, nek5000 can do that. I cannot tell you more unless you come up with more details. Depending on the geometry you can use prenek to build a mesh. 2: I am afraid there is no code documentation but the books you were talking about cover more or less everything although not in great technical detail. You will get some informations about the used methods but when it comes to coding it will not really help you. Best, Stefan --- original message --- From: ??? Subject: Re: RE: [Nek5000-users] Three questions Date: 1st August 2009 Time: 5:06:49 pm Hi Stefan Thank you for your reply! There are two other issue in the previous email which I hope you can give me any adivice: 1 What I intend to do is an accurate simulation of incompressible turbulent flow (with heat transfer) around cylinders, my question is :is nek5000 suitable for this simulation? If it is YES, can I use the mesh generate tools included in nek5000 to create mesh for this case? I know you are working with Cubit, but I think it is not a free mesh generator. 2 Is there any theoretical manual for me to understand the code ? It seems very difficult for me to understand what exactly the program is doing without a guide. I have some back ground on FVM, FEM , right now I study the book "High-order method for incompressible fluid flow" and "Spectral/hp element methods for CFD" Regards Zhicheng 2009-08-01 ??? ???? Kerkemeier Stefan ????? 2009-08-01 22:26:26 ???? wangzhicheng ??? Nek5000-users at lists.mcs.anl.gov ??? RE: [Nek5000-users] Three questions Hi, nek tries to tell you, that the local size lelt is too small for the given number of elements and processors. If you have 9375 elements in total you probably need to run it in parallel. Your local size seems to be 600 elements so you need 16 proc to run the solver or you need to increase the local size lelt in the SIZEu file and recompile. Hth, Stefan --- original message --- From: wangzhicheng Subject: [Nek5000-users] Three questions Date: 1st August 2009 Time: 11:06:00 am