Dear Bentley team,
I’m performing a structural concept design of RC frame of residential development using EC2. I have models of the slab of one of the RC frame buildings in Tekla and Ram Concept. We made a comparison between the results for long-term deflections from both programmes. It appears that there a significant differences between the deflection in Tekla Structural Designer and Ram Concept.
In brief the deflections in Tekla using Slab Deflection tool with an event sequence similar to that recommended in the their Long-Term Deflection Webinar-part 2 are less than those from Long term deflections in Ram Concept.
I’ve use the followings parameters for the slab:
Flat slab 250mm thk C32/40, columns 300x600mm C32/40 pinned at both ends, 100% restraint type for slab panels (which is conservatively and is equal to 20% in Ram) and following event sequence in Tekla:
I've created equivalent Load-History in Ram concept for comparison purpose using default load-History/ECR parameters:
I received the following long-term deflections for the slab model in Tekla with max deflection of 18.4mm in the span at the upper left corner of the slab:
and the following deflection of the slab model in Ram Concept:
and a snapshot from the same critical span in the upper left corner with 28.3mm long-term deflection or with 10mm more than the Tekla model:
After receiving these results I'm really confused of the big difference between the result from the programmes. I consider to change some of my column layouts to break down some of the spans in my 3D Tekla
model even its deflection are within the limit in order to control the greater deflections which I received in Ram concept for the same spans.
I sent a query to the Tekla Support Team as they usually respond within the same day but it's fear to do the same with your team in order to find what is the cause for the different result.
Please find below the answer/comments from Tekla Support Team:
We wonder if you have used all the default settings in the other program, or have altered the input? Our expert had the following comments on some of the parameters set in the other program. In particular he thought the Creep factor was oddly high and that the Ageing coefficient is more usually associated with the US design code/ method than EC. Is this creep factor used for the entire load/sequence in the other product? Does the program have recommended settings following Eurocode guidance that you have used?
You probably saw in the webinar that you can view the Effective modulus calculations in TSD by exporting this to Excel. This gives the calculated creep factors – there are different values for each load sequence and we see they are quite a bit lower in general than the single value set in the other program:
No doubt you are similarly querying the other program vendor. We hope the above is helpful and would be interested to know the results of your further investigations.
Trimble Support Team
Tekla also provides information for the locations and the extend of cracking on the slab for each of the events:
I compared the deflections in one of the other span for each of the events. They are pretty the same for the first three events (1-3) but after applying the sensitive finishes (event 4) there is a big raise in the deflection of Ram Concept model compared to Tekla model. This event correspond to creep coefficient of 1.619 calculated by Tekla which is significant decrease from the initial value of 2.528:
Events Deflections in TDS Deflections in RAM
My logical explanation is that the creep factor in RAM doesn’t change the same way as in Tekla. Am I correct?
My question is how Ram Concept complies with EC2 requirements and recommended values for shrinkage, creep coefficient and ageing coefficient.
How the initial creep factor 3.35 differs from creep coefficient in Tekla /EC2 in physical way and how does it change in a similar way as in Tekla (e.g. from 2.528 to 0 at the final load event).
Please also find attached the model of the slab in Ram Concept:
Block C - additional YB load combinations.cpt
I'm looking forward a response from your team.
My desire is to find an explanation of the differences in order to use the both programmes in proper and effective way
Many thanks in advance.
You will find some good information about our load history analysis in the program manual, and also here:
Some of our defaults are based on ACI publications, this is true.
For Eurocode specific guidance see the attached document which one of my colleagues in the UK put together. DOCX
Our Creep factor appears large because in includes a 1.0 factor for initial deflections in addition to the traditional creep factor, so if you want a additional creep factor of 2.0, enter 3.0 in Ram Concept. 3.25 is a bit conservative, but in alignment with some ACI documents.
Our default load history steps can be conservative since they end with a full live load case. You edited the steps, but the final load event is also like this (LL = 1.0). That looks consistent with Tekla.
Normally you want all load history steps to include the balanced load case (but since this is not a pt slab, no matter).
You have a bit of an unusual reinforcement with a dense set of user bars in one region only, not sure if that is the same in Tekla, but the final reinforcement, including user bars, does matter in Ram Concept cracking calculations.
Beyond that we would probably need to review the technical documentation from Tekla regarding their methods to highlight other differences.
One of my colleagues did some review and I've included his comments below:
The biggest cause of difference is external restraint to shrinkage. Tekla’s means of handling this effect seem to be mostly insignificant. When I set Concept’s shrinkage restraint % to 0 the plots were reasonably similar despite the different ways in which the programs are performing the calculations.
Tekla does appear to be considering load history like Concept does. However, Tekla’s treatment of creep and shrinkage seems [different]. For creep they use a composite creep value that accounts for a cumulative creep value due to all the loading/unloading. It does not, however appear to consider reinforcement which would be an important omission. Concept’s creep calculations on the cross section consider the effects of tension and compression reinforcement as well as creep rate changes caused by stress redistribution caused by creep – a phenomenon I have coined “creep recursion”.
Tekla’s treatment of shrinkage and shrinkage warping is also crude, using a simple amplification factor on the calculated deflection. Concept’s rigorous cross section based approach is much more accurate – as shrinkage can either increase or decrease curvatures due to loads and can contribute in interesting ways to cracking. Tekla handles internal restraint to shrinkage by reducing the modulus of rupture. While this is a commonly used approach in the US it can be highly approximate.
Tekla’s program does appear to have a way to account for external restraint, but based upon their documentation it seems to have little influence on deflections (100% restraint causing a difference of 10% on deflections). Therefore it isn’t clear how they are handling this restraint nor does it seem to capture the potential influence of this effect. From the documentation I couldn’t see how this value is set.
As for cracking (and other stiffness parameters), Tekla works on the element basis so the peak stresses, cracking, etc. can be influenced by the mesh size which isn’t desirable. Tekla does not seem to consider the effect of axial force/stress on cracking which could be significant if present. Otherwise Tekla appears to handle cracking reasonably.
The most significant philosophical difference between the programs is that Tekla works on the element basis, while Concept works on cross sections. Tekla also seems to approximate creep and shrinkage effects whereas Concept’s approach is more rigorous.
In summary I would normally expect to see differences between Concept’s results and Tekla’s – and where differences occur I would consider Concept’s results to be more accurate due to the rigorous treatment of the influencing behaviors.
Please let me know if you have any questions.
Many thanks for your responses Seth.
Could you clarify how the creep factor value in Ram concept changes along with the model load history and is there any possibility to review these values for every load event as Tekla does this (Please see the Excel output for Composite Modulus Calculations attached above). As you could see from my comparison of the deflections at each load event the difference appears after the forth event. If I can compare the intermediate creep factor values I might be able to find the reason.
The unusual reinforcement has been introduced to control the deflections. Even I'm using less reinforcement in Tekla for this panel but after some amendments in reinforcement quantity and column position there is a still big difference in deflection; for example13.8mm in Tekla and 23.3mm in Ram.
The details of our creep implementation are explained in the original paper which our method is based on, and copied here: https://communities.bentley.com/other/old_site_image_galleries/member_folders/m/jason_colemans_files/87861
We cannot yet output the effective creep values in the stage output, however.
I checked the models with 0% shrinkage restraint in Ram and 100% restraint type in Tekla and received really the same results after Event 6 but this is but that is absolutely opposite shrinkage initial condition and shouldn’t be used for comparison. You are correct that Ram concept long-term deflections are greatly affected by changing of this parameter from 10% to 0%. Whereas in Tekla the setting of the restraint type from 50% to 100% changes slightly the results. So definitely that is one of the differences although I expected to find your shrinkage allowance to have the same approach as Tekla's.
Do you have a logical explanation of the identical results for the first three events. Do you think that in this case your 10% shrinkage allowance takes effect at the beginning of the 4th event when we have accumulated relatively high bending moments?