Calculations in module RF-CONCRETE Columns

[doug]

Hello,
I'm having some problems on results obtained from the RF-CONCRETE Columns module, in particular the Eurocode 2 shear check according to formula 6.9 (using RFEM5).

Let's consider a short concrete column (square section 100x100cm, Length=100cm), restrained at the base. On the free node I applied two kinds of loads: the first produces an uniaxial bending and the second a biaxial bending, as per following figure

The combined shear on both load cases is the same (2500 kN).
Moreover there is a traction load 2000 kN.

For the uniaxial load case, the check for struts in compression limit state (VRd,max Eurocode formula 6.9), is verified for an angle 21.8° with an usage factor 0.7769.



For the biaxial case, the check is NOT fullfilled, even considering the favourable effect of having compression struts inclined at 45°.


The biaxial check with struts at 21.8° (like those adopted for the uniaxial case) is the following.


In conclusion: if we consider two load cases that lead to exacly the same resultant shear (2500 kN), the concrete struts' resistance (inclination 21.8°) calculated by the RF-CONCRETE Columns module (VRd,max Eurocode formula 6.9)

• uniaxial load struts resistance 3218 kN
• biaxial load struts resistance 1450 kN

The resistance according to formula 6.9 is reduced of 55%. I'm unable to justify such a great resistance reduction due to bi-axiality on a physical ground. According to these results the column is too small and should be redesigned with much larger section dimensions, just for the biaxial case.



If I compute the same case with another software, there is not such a huge difference in struts' resistance (VRd,max Eurocode formula 6.9): the reduction is 22%


In particular I notice great difference in the calculation of resistant section dimension (z and bw). The calculation base for the bi-axial shear on this alternative software may be found here: https://help.geostru.eu/rcsec/en/ver..._biassiale.htm

I suspect the formulas used in the RF-CONCRETE Columns may be too conservative.

Any comment is appreciated.
Thanks in advance,
Doug

[Frank Faulstich]

Hello,

we cannot comment on result differences to other programs, especially because the determination of the effective depth 'd', inner leverarm 'z' and effective width 'bw' is not so straight forward.
As you may see in the strain diagram in the result table for the member no. 2, the position of the neutral axis is not comparable to the results on member no. 1.
However, we have a documentation for the approaches how RF-CONCERTE Columns determines the dimension 'd' and 'bw' which are the most decisive factors in this result difference. You can find it in the manual for RF-CONCRETE Columns in Chapter 4.6.5 for rectangular cross-sections via the following link:


https://www.dlubal.com/-/media/Files...EC45EBAB366E0F

Frank Faulstich

[doug]

Thank you Frank,
from reading the reference manual it seems that if I consider the uni-axial bending case, and add a small lateral load (with no practical variation in the flexural state in terms of neutral axis, tension and compression zones) the difference in calculations of dimensions should be negligible.

For example, here below the difference is only a 1 kN load along Y for the second case.


I would expect NO appreciable difference between the two cases.

Case 1:


Case 2:

In particular in the second case the angle of struts had to be increased significantly.

If we compare the resistance for the same angle (21.8°, that maximizes the stirrups' resistance), the results are the following:


So the section resistance is reduced by 29%, only due to the presence of a force equal to 1 kN orthogonal to the main shear.
I'm unable to justify this huge difference, if I read the cited manual chapter 4.6.5 the bw and z dimensions are not expected to make such a "jump".

Best regards
Doug

[Frank Faulstich]

I have tried to reproduce the problem. V_Rd,max is almost the same in both members:





I have attached my file to the post for comparison.

Frank Faulstich

[doug]

Thanks Frank,
in realty the difference is relevant also in your case: the lever arm reduces from 796 mm to 695 mm. The value of VRd,max is equal on both cases just for the fact that the struts' angle is significantly increased (i.e. optimized: it is the minimal angle compatible with concrete struts resistance). The true comparison between your two case should be performed considering the shear area reduction, that in your case it is just the lever arm reduction, i.e. 695/796-1 = -12.5%.


I notice in your case the "component width" remains equal to 1000 mm. Interestingly in my case I observe a significant reduction of this parameter in the biaxial case.
But if I rotate the load (i.e. the main 3000 kN load along Y and the secondary 1kN load along X) i obtain a totally different result for the biaxial case (the uniaxial case instead is exactly the same).

case 3 - biaxial rotated


This is a further problem: why the rotation of the load produces a totally different biaxial result in the calculation of VRd,max? The section is bisymmetrical, so any difference is due to the algorithm.

Herebelow the FAIR (i.e. with the same struts' angle) comparison between the three cases:

Best,
Doug

[Frank Faulstich]

Hello Doug, we have investigated the problem in more detail. It seems that RFEM 5 is working incorrectly here. In this example, however, the cross-sectional dimensions are unrealistic and the module is actually intended for pressure force. With realistic cross-sectional dimensions and pressure force, we could not reproduce the problem.

We will fix the problem in one of the next versions.

We have also investigated the problem in RFEM 6. Here, the concept has been completely redesigned and the problem does not occur.

I recommend you to use RFEM 6 for your investigations if possible.

Frank Faulstich

[doug]

Thank you Frank,
you say that this section is not realistic: in reality short column stubs like this is more than an "academic" study.

At the base of big silo there can be short column (connected to the foundation elements).
I'm indeed working on the foundations of siloes, each weighing about 6000 tons (when filled with material). Under seismic load a lot of traction and shear is experienced by the concrete columns. Moreover in this specific project there is the need to let the passage for the truck mixer under the siloes, and the truck scale should also placed between the columns, under the siloes.
All those constraints limit the section dimensions of the columns to 100cm x 100cm, to avoid interferences.

Why you say a column section 100cm x 100cm is not realistic?

I can confirm the sections are realistic, and are actually verified with other software. I hope the fixes in RFEM will be general and allow to study also these elements (under traction: there are load case with even larger traction that 2000 kN, by the way).

An example of (smaller) columns for (much littler) siloes:


Best
Doug

[Frank Faulstich]

Hello Doug, my comment was not so much about the cross-sectional dimensions themselves, but about the ratio of the cross-sectional dimensions to the member length.

A condition for the application of the beam theory is that the cross-sectional dimensions are small compared to the beam length. That is not so in this example.

I am aware that in practice one is often forced to use the beam theory, although the application limits have actually been exceeded. In reinforced concrete construction, there are only engineering reinforcement checks for members and for surfaces, but not for solids.

If you are in such limit regions, then the algorithms used can reach their limits. You have shown this with your example.

Frank Faulstich