Stress calculation for a circular section POU_D_T with C_FLEX & I_SIGM

Ect

Hello,

Would anyone be able to confirm my understanding of how flexibility and stress intensity factors are applied for the case of circular pipes?

While recently doing pipe stress analysis it appeared that when one uses flexibility factors, the stresses SMFY and SMFZ are reduced, and trying to compensate with stress intensity factors works in an incomprehensible way.

Intuitively, and according to R3.8.01 §10.2 the flexibility factor should only reduce the section modulus, and the intensity factor should only scale the stress.

The output of SIEF makes sense, moments are not reduced by the flexibility factor, while the displacement is increased as expected with a decrease of the section modulus.

Now looking at bibfor/elements/porigi.F90, the in and out-of-plane flexibility coefficients are effectively reducing the section modulus xiy and xiz. If I'm correct the stresses are calculated in bibfor/elements/posigr.F90, and one can see that in and out-of plane flexibilities and intensities are applied in the case of a rectangular section, but the for the circular section the assumption that xiy == xiz is silently made (which makes the specification of in and out-of-plane ineffective). From the code here I also don't understand why the stress is lowered by the flexibility factor...

So here is the question: are these two files (porigi & posigr) the correct ones to understand what is going on with the flexibility and stress intensification factors, or there are other files I didn't spot?

Thanks,
Ect

Ect

Ok, after some tests I was wrong about the interpretation of what is done in posigr.F90 about the circular section: it's computing the uncorrected efforts and therefore the section modulus is effectively the same in both cross directions of the beam.

The behavior of COEF_FLEX is also correct in porigi.F90 and ptka*.F90, tested on a beam against simple beam theory.

However there is an implementation error for INDI_SIGM in posipr.F90 so the correction is applied twice for POU_D_T, here is the sequence for the y direction:

xiy = vale_cara(2)

xfly = 1.0d0;
xsiy = 1.0d0;

valp(1:4) = ['C_FLEX_Y', 'C_FLEX_Z', 'I_SIGM_Y', 'I_SIGM_Z']
call get_value_mode_local('PCAARPO', valp, valr, iret, retpara_=retp)
if (retp(1) .eq. 0) xfly = valr(1)
if (retp(3) .eq. 0) xsiy = valr(3)

! prise en compte de l'indice de flexibilité
xiy = xiy/xfly

! prise en compte de l'indice de contraintes
xxy = xsiy/xfly
xiy = xiy/xxy <---- FIRST CORRECTION

r1 = vale_cara1(5)
sipo(5) = -efge(5)/xiy*r1

! On doit corriger dans le cas du coude, si pas de coude les coeffs sont =1
xxy = xsiy/xfly
sipo(5) = sipo(5)*xxy <---- SECOND CORRECTION

xxy = xsiy / xfly is first applied to xiy, then applied to sipo(5).

The stress is therefore corrected by INDI_SIGM**2 / COEF_FLEX, which explains why the stress is reduced by the flexibility factor.

As a workaround one can get coherent stress intensification by setting INDI_SIGM = sqrt(expected_sif*COEF_FLEX).

Ect

Here is a base64 encoded zip with files to replicate and a presentation with the results:

beamzip.txt

484kB