number of integration points of beam

chjjj · Post by **chjjj** » Fri May 06, 2005 7:10 pm

If a beam element has three different sections (the middle span section is different from the end sections of the beam), how many integration points will be appropriate for the beam?

If three integration points is enough for the beam and column element?

thanks

silvia · Post by **silvia** » Fri May 06, 2005 9:11 pm

while I haven't quite done it that way, with different sections along one element, you might want to test it out, compare pushover curves depending on the number of integration points.

chjjj · Post by **chjjj** » Tue May 10, 2005 4:46 am

If I use different integration points for beam or column element(both use three or both use five), the results are obviously different when the steel bar has yield.
So, how many integration points will be appropriate?

silvia · Post by **silvia** » Tue May 10, 2005 6:28 am

i usually have 4 or 5.

hachem · Post by **hachem** » Tue May 10, 2005 8:46 am

You may want to consider using 3 elements instead of one, especially if the member steps in size (not tapered).

If your element is tapered, you need to figure out the number of integration points that give you reasonable integration point "weights". You still may want to consider multiple elements, as this will give you more control over the integration scheme.

Be careful about having too many integration points (more than 4 or 5), especially if you have softening behavior. I usually try to limit the number of integration points to 3-5, but I've occasionally used 2 on small elements.

Boris · Post by **Boris** » Fri May 13, 2005 5:59 am

Hachem is right, try with three beams, it is not wrong to use one beam, provided that your centroid does not change. However if the centroid of the cross section is shifted away, and if you have measurable normal forces in the beam, you'll end up with a lot of moment where those beams meet!

Boris

Post by **fmk** » Fri May 13, 2005 6:27 am

if you go to more than 1 element and your materials degrade, i.e. concrete, you may need to switch to using the dispBeamColumn element, as the forceBeamColumn element may develop convergence problems .. as you add more dispBeamColumn elements you will get global convergence problems as well! .. be warned that the choice of elements and # of integration points for systems of degrading materials can have a big impact on the response.

bufa · Post by **bufa** » Tue Oct 25, 2005 4:43 am

[quote="hachem"]
Be careful about having too many integration points (more than 4 or 5), especially if you have softening behavior. I usually try to limit the number of integration points to 3-5, but I've occasionally used 2 on small elements.[/quote]

in effect with a degrading material - I did some analysis with concrete04 - NIP has a great influence on results inducing also, when it is high (8 or more) convegence problems. why?