comparative check calculations wanted 2

[austim]

Hi, all

I have just analysed a simple 2-span continuous beam with four point loads, using 3 different programs. 2 of the three give identical answers, but not all three.

I would welcome any additional solutions obtained from other packages.

My problem is this :

Nodes at x=0,1.1,1.7,5,6.3,6.9,9
pinned supports at nodes 1,4,7 (x=0,5,9)
point loads of -1471 at nodes 2 and 3, -429 at 5 and 6

Using my own continuous beam program I get reactions of 1852.8,1935.8,11.4 at nodes 1,4,7 respectively.

My old trusty DOS based Microsafe gives me 1853, 1936, 11.

So far, so good, BUT using RamAdvanse Student edition, I get reactions of 1871.5, 1893.7, 34.8.

I know the differences are not earth-shattering, but would have expected closer than 2% on the centre reaction for such a simple structure.

Does anyone know whether RamAdvanse makes allowance for shear deformations or other such fancy considerations? (I know that my own program does not, and neither does Microsafe)

What values are given by other software?

 

Hi
My english is not good, I'm french
But you can abandoned your third software.

Output results are :

type space frame
units meter knew degree sec
$
$
UNITS MS
JOINT COORDINATES
$
$
$
1 0.000 0.000 0.000 S
2 1.100 0.000 0.000
3 1.700 0.000 0.000
4 5.000 0.000 0.000 S
5 6.300 0.000 0.000
6 6.900 0.000 0.000
7 9.000 0.000 0.000 S
$
MULTPLE INCIDENCE
1 1 2 3 4 5 6 7
$
MEMBER PROP
1 to 6 Pipe od 0.200 wt 0.050
$
UNITS KGS
LOAD 1
JOINT LOAD
$ of -1471 at nodes 2 and 3, -429 at 5 and 6
2 3 FORCE Z -1471.
5 6 FORCE Z -429.
$
JOINT RELEASE
1 MOMENT Y Z
4 FORCE X Y MOMENT X Y Z
7 FORCE X MOMENT X Y Z
STIFFNESS ANALYSIS
LIST REACT
p.2 = NSO input file : lib-inp = 18/10/2001 , 08:39:27

*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
* Results after latest analysis *
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

Job id - NONE , Job title - NONE GIVEN
Active Units - Length: M ; Force: KG ; Angle: DEG ; Temperature: KELV; Time: SEC ; Mass: KG

Loading 1 ; Title -


Reactions at Support Joints in Joint Reference Frame
Joint Type Force-x Force-y Force-z Moment-x Moment-y Moment-z

1 0.0 0.0 1853.16882 0.0 -0.00000 0.0
4 0.0 0.0 1934.97021 0.0 -0.00000 0.0
7 0.0 0.0 11.86100 0.0 0.00000 0.0


*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
* End of results of latest analysis *
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

 

[Riz]

Remember type of analysis and member properties play a significant role in any output. I have run your problem on a grade 25 concrete beam 250mm X 600mm. Corresponding results obatined from any sw should have very little variation (less than 1%). There must be some problem in your other sw.

** austim.bmd **

BEAM LENGTH = 9.0 m

MATERIAL PROPERTIES
Modulus of elasticity = 26.0 kN/mm²

CROSS-SECTION PROPERTIES
Moment of inertia = 4500000000.0 mm^4
Top height = 300.0 mm
Bottom height = 300.0 mm
Area = 150000.0 mm²

EXTERNAL CONCENTRATED FORCES
1471.0 kN at 1.1 m
1471.0 kN at 1.7 m
429.0 kN at 6.3 m
429.0 kN at 6.9 m

SUPPORT REACTIONS ***
Simple at 0.0 m
Reaction Force =-1852.8 kN

B: Simple at 5.0 m
Reaction Force =-1935.799 kN

c: Simple at 9.0 m
Reaction Force =-11.4004 kN

MAXIMUM DEFLECTION ***
32.59417 mm at 2.002905 m
No Limit specified

MAXIMUM BENDING MOMENT ***
2267.161 kN-m at 1.7 m

MAXIMUM SHEAR FORCE ***
1852.8 kN from 0.0 m to 1.1 m

MAXIMUM STRESS ***
Tensile = 0.151144 kN/mm² No Limit specified
Compressive = 0.151144 kN/mm² No Limit specified
Shear (Avg) = 0.012352 kN/mm² No Limit specified

 

[Ingenuity]

austim,

i get the following using microSTRAN:


== S U P P O R T R E A C T I O N S ==

CASE 1:

Node Force-X Force-Y Force-Z Moment-X Moment-Y Moment-Z
kN kN kN kNm kNm kNm
1 0.000 1852.800 0.000 0.000 0.000 0.000
4 0.000 1935.799 0.000 0.000 0.000 0.000
7 0.000 11.400 0.000 0.000 0.000 0.000

SUM: 0.000 3800.000 0.000 (all nodes)

Max. residual: 2.251E-11 at DOFN: 6


HTH

 

[Ingenuity]

Riz,

Member props (E and I) will have no effect for this case (prismatic) as EI is constant for both spans.

Additionally, self weight was NOT a case that "austim" had presented so the difference here is that the latter (#3)software is WRONG.

austim,

Shear deformations should NOT effect this simple case.

If RAM is doing a non-linear analysis, section props will play a part in resulting actions, and that will depend what sort of non-linear analysis is performed, and if the solution did indeed converge within good tolerance.



HTH

 

[austim]

Good Grief !

I really didn't expect responses quite as prompt as that, but thank you all.

Three (very) independent sets of values, all within 0.02% or so of my original results gives me a pretty fair basis for ignoring RamAdvanse, don't you think.

What I didn't include in my posting were the results that I got from an old copy of Avansse for Dos, which can be even worse.

The first results I got from Avansse gave me 1984, 1640, 176!

After further review, I find that I can change those reactions quite significantly by changing the section used for the beam (and yes, I am still keeping the one section for both spans). It appears that the accuracy of the output depends in part on the magnitude of the deflections.

Worryingly, I cannot repeat my first set of values, despite restoring the section to my initial choice, and when the only change I made to the model was in the choice of section; second time around it gives me 1897, 1836, 67. With a much smaller section (deflections up to span/69 ) I get 1859, 1920, 20. Still not really very good.

It looks as if my once trusted Dos Avansse is destined for the recycle/disposal bin as well.

To all those to whom I have recommended either RamAdvanse and/or Avansse, I think I may not have given you the best of advice - sorry about that.

 

[LPPE]

I think RAM Advanse is the best and the worst of 'em. I really like the input interfaces, but I've had problems also with output, and some of the graphical depictions of loads are incorrect (global and local).

 

[austim]

HI, yet again.

Just a PS to my last posting, now that I have been able to repeat my first results from Dos Avansse.

What I had managed to do, while trying to set up my problem in metric units (standard in Aus, of course), was to confuse the input units, with the result that my maximum deflection was down to span/3*10^6! But I would still not expect that to change the maximum BM by 10%, and reduce the central reaction by 15%.

That just reinforces my contention that the accuracy of the results depends on the magnitude of the deflections. (Thankfully I have never used Avansse in my own practice, but only had it to be able to reproduce some analyses sent to me by a colleague).

Now I will definitely abandon Dos Avansse, and try the effect of varying stiffness on its Ram version.

 

[austim]

And guess what?

If I use Ram Advanse, take dimensions in feet, loads in kips, and use a stupidly stiff beam (w44*335) I get reactions of 1987.58, 1632.55, 179.87.

What was that about the "... worst of 'em" pylko?

 

[austim]

Hi, all.

Now for some humble pie (I don't do this too often, or very well )

There is a very interesting FAQ at

http://www.ramint.com/kb/4.shtml

That explains why the Ram Advanse values differ from the standard texbook answers, and it IS all to do with shear deflections.

It would appear that all 3 sets of my original reactions are 'correct' within the limits of the underlying assumptions.

I was clearly too hasty in my readiness to condemn Avansse and RamAdvanse - it almost looks as if their results could be more 'accurate' than the others!

So - please disregard all negative comments that I have made in regard to both Avansse and RamAdvanse (I have red-flagged two of my own postings on this topic!).