Using RC beam design for a Culvert queries

[Anonymous]

Hi,

I am attempting to do something a bit unusual with ASBD by trying to use the RC beam design functionality in ASBD for the design of a simple rectangular box culvert. My aim is to model & analyse the roof, base, and walls together and then extract results to their corresponding RC design beams so that I can then design them. I have a few queries with this;

1) I can produce a 2D model of a certain width of box (image 1) but to keep everything in the deck plane (so I can use RC design beams) I have to model the structure in the XY-plane, apply all my loads along the Y-axes, adjust the beam local axes accordingly, and use dummy elements for the unwanted members in the middle. This is fine, however I have to generate the applied loading in Excel because of its complexity and inputting this to ASBD manually can get quite complicated - is there a way to import loading direct from the clipboard or a CSV file etc without typing it in by hand?

2) I would also like to model in 3D to capture transverse effects, but currently I can only use an RC design beam for elements in the deck, which is always in the XY plane at z=0. Is it possible to use RC design beam elements in a 3D model in any other plane or at any other Z-value?

3) Alternatively if I use FE meshes for the four faces in a 3D model is there anyway to extract the results from these into the design side of ASBD?

Many thanks for your guidance and help,

Steve F

[dave_geeves · Accepted solution]

Hi Steve,

A very interesting query and one that I must admit needs quite a bit of thought and inventiveness.  There are some limitations and little quirks but in general I have generally found a solution to the analysis of culverts, one way or another.

You are quite right that a major limitation is that you can only use design beams, to assign section properties, to longitudinal/transverse members, and these can only exist in the xy plane.  Using design sections, to represent the section properties, removes this limitation but even if you have defined design beams to design the walls, slab  and base you have the problem of getting the analysis moments back to the design beam load effects tables.

There is a solution to this but again there are some small limitations. The way to do this is to define your box frame in the vertical YZ plane and then create four virtual members each representing the four sides of the box.  When the analysis is complete then the results of the composite members can be exported (via the transfer results form) to an .sld file which can be read directly into the design beam tables. The only limitation is that the moments transferred have to be MY moments so the orientation of the vertical virtual member  axes is important.  Unfortunately I don't believe there is a way to "twist" the virtual member axes (as you can do for ordinary beam members) so that is why I have suggested that the frame be created in the yz plane as this means the default virtual member axes are then orientated correctly.  The only other issue is that when you transfer the results of a vertical VM to an sld file the location along the length of the beam is not correct (a very small value).  This means that when you import this to the design beam you will get an warning  message saying that the length of the virtual member does not agree with the design beam length.  However, the default behaviour is the stretch the length out to make them agree and everything is then fine.

With regards to generating variable loads in a spreadsheet and importing them into the analysis loading tables,  I would agree that this is quite a limitation and I don't believe there is an easy way at present.

However, if you build a 3D model so that the deck is a grillage of FE then the vehicle load optimisation is a good way to generate the vehicle loads for you.  I generally prefer an FE model for this as any soil pressure loading can either be added as a series of hydrostatic loads on the walls to get an appropriate pressure profile or soil springs attached to each node in the walls.  Skewed culverts are also dealt with more readily  You can create virtual members of the finite element instead of beam members so the process I described earlier can be used to take the results into the design beams.  Again, the only limitation is that the culvert walls should lie in a vertical XZ plane

Below is one image from my example model - I also attach a pdf file showing some additional images with comments

I hope this is helpful and if this has answered your query then please mark this post as a solution so that others may benefit.

Kind regards

Dave Geeves

[dave_geeves]

Hi Steve,

Just wondering if you had seen my reply and whether my suggestions helped.  If you need further explanation please get back to me.  I can send you a typical data file if it would help but you will need to be running the latest 2020 version.

Kind regards

Dave Geeves

[Anonymous]

Hi Dave,

Sorry I was drafted onto another task last week but I am looking at this now and will get back to you hopefully tomorrow.

Thanks very much for your help,

Steve

[Anonymous]

Hi Dave,

My apologies for the delay in responding to you, I had been pulled off onto another project before being able to look at this again. Thanks for your detailed and thought-out response to my queries. I looked into using virtual beams (which I hadn't used before) with an FE model which allows you to import loads to the beam files and also allows us to see what happens in the transverse direction which is very useful. This overcomes most of the barriers which I had been encountering and with the vehicle load optimisation now gives us a functioning workflow.

To give you an idea of where we were going with this I thought I would add a few extra steps that we were looking to model. I appreciate that this use of ASBD is probably outside of the ordinary so mention this more for your interest rather than as a development request;

- Load distribution through fill - fill above structures allows distribution of point loads (such as wheels or axles) both longitudinally and laterally having the effect of smoothing out loading and therefore design effects. For structures like culverts this can be beneficial which we are keen to investigate. This is the area we were trying to emulate in Excel by generating loading and then importing using clipboard into ASBD.

- Envelope of virtual beam member loads - using FE allows for a metre strip width design approach which is helpful as we can model multiple lanes capturing a range of effects and then design on a per-strip basis. I wasn't sure if there was a way to envelope these strips before extracting to a typical beam for design which would be handy as it would allow us to automatically consider the worst-case strip rather than investigating each strip one-by-one (noting that engineering judgement would usually overcome such an issue).

- Crack widths in beam design mode - the beam design graphs are powerful as they at a stroke make it much easier to see what's going on in the design (am a big fan of this aspect). The crack widths are currently shown in the dialogue box as a single value for the selected section, I'm not sure if it is possible but to have this shown on a graph in a similar way to stresses etc would be interesting. (Noted that this is less of an issue for RC members due to cracking being evaluated under the quasi-permanent load combination which has fewer variants.)

If there are any workarounds for the above I would be keen to hear about them. I hope the above has been helpful, thanks for your help on this Dave.

Kind Regards,

Steve

[john_gamble]

Hello, I am relatively new to ASBD. I am trying to model a 2D culvert based on the portal frame example with spring foundations. But the model will not run. Is there any example that I could follow?

[dave_geeves]

Hi John,

I think the best thing would be to create a new post in the forum and attach your zipped up data file(s).  I could then take a look at it to see why it won't run.

The portal frame example in the examples manual is the correct thing to use here so you should get some results.  I look forward to taking a look for you

Kind regards

Dave Geeves