Intel® Fortran Compiler
Build applications that can scale for the future with optimized code designed for Intel® Xeon® and compatible processors.

IFX - Second Sheet

JohnNichols
Valued Contributor III
495 Views

The three programs we have been discussing are ULARC, BORR and STRAND7.  Strand 7 is a program I believe is written in Fortran and comes from Perth in Western Australia.  

The bridge I have been looking at is one common in parts of the world.  The real issue is that we want to review these bridges for safety, ULARC took 30 minutes to set up the input file and it told me that the point of concern is the intersection of the latticework and the main beam, a factor of safety of 2. Borr is a work in progress with the plates, it is progressing.  I trust ULARC because it lines up with experience and published results. 

The STRAND7 model took 8 hours to create for someone who has used STRAND since 1994.  It is running nicely. 

Strand6 in 1994 used to run on the mainframe at Newcastle University.  I modelled a masonry building on a ground block. Started the run at 6 pm, and left it running.  Strand6 grabbed the mainframe and ran all night, nothing else would run on the computer.  I was called to the Dean's office at 10 am and told I was banned from the mainframe, permanently.  I was given the printout and a stern lecture.  

I ran the same model on a 386 in the computer lab, it ran for 8 days, no one else could use the computer and no one switched it off.  The other grad students were not pleasant by the 2nd day and testy by the third, we will not discuss the rest of the week.  

STRAND7 runs in about a minute on this problem.  

The actual amount of information you need to draw these bridges is surprisingly small, we need a few numbers, maybe 30.  In Fortran I can write a program that takes the 30 data numbers and given the type of bridge draws it in AutoCAD and creates the data file for STRAND, ULARC and BORR.  This program exists, it is quite fast. The only pain is creating a new bridge-type.   FORTAN coding allows for speedy development, existing locked compiled programs do not. 

The issue now is not the coding language, it is data interchange in a quick and reliable fashion, the only languages, you would use are Fortran, C++ or C#.   C# is slow, C++ is a pain to write in and Fortran is it.  I wrote the FEM generator first in PYTHON, it was very slow and the experience was less than pleasant. 

This picture is the first mode from STRAND7.  It is very low frequency and you can see the issue identified by ULARC.  But there is a long way to go to get a safety factor from Strand.  It looks nice.  

image_2022-08-13_152809777.png

Modal participation is interesting.  

 

Screenshot 2022-08-13 150120.png

0.23 Hz as the first mode is a pain as it is buried in the ground noise where 1/(squared (freq)) dominates. 

I can run Ularc and Borr on a little NUC on the bridge with the data collector.  I cannot do that with Strand. 

We will soon have more data than we know what to do with, programs need to be much more agile and deal with the data.  Only a goose would not use Fortran. 

 

 

0 Replies
Reply