Simple Wood Truss Analysis in RISA3D 2

[DCStructures]

I'm checking some existing wood roof joists for added weight of solar panels. I built a simple RISA3D model, but it's not giving me good results. The solar panels are only 2.7psf and the building is fairly new construction (<2 yrs old), so I don't really think there is much concern with the added weight of the solar panels. But I still need to have accurate results. I have some existing information, but I don't know what species/grade of wood was used, so I'm assuming Southern Pine or Douglas Fir. In any case, the species/grade should be good enough for the existing design to be okay, but I'm still not getting good results even for existing loads. I've tried adjusting member ends (pin v. fixed), and I've tried adjusting unbraced lengths. I'd attach my model, but I can't figure out how to add an attachment. Any insights you might have are much appreciated.

 

[msquared48]

What is the specific problem with your results? Your complaint is too general.

Mike McCann, PE, SE (WA)

 

[DCStructures]

The results indicate the some of the members are highly overstressed for both bending and shear. This doesn't make sense to me because it's an existing truss and I'm only using 10psf dead load and 20psf snow load. Here are some of the things that I'm wondering if I need to fix in the model in order to get accurate results:
1. I'm applying the loads as distributed loads (k/ft) along the length of the top chords. Is this correct, or do I need to translate them into joint loads?
2. I'm using the standard IBC ASD load combinations. The program automatically includes Cd factors for wood design. Since I already have factors in the load combinations, can I make the Cd factors = 1.0?
3. I tried breaking the top chord apart so that it was a separate member between each pair of panel points. I did the same thing with the bottom chord. In both cases, I kept the member end releases fixed for the chords so they acted continuously. I also tried using one continuous member for each half of the top chord and one for the bottom chord. Is one way better than the other in RISA?
4. I have one support pinned and one with a roller, and I have the out of plane (Z) axis pinned at all nodes since it's a 2D model in RISA3D.
5. For the unbraced length of the top chord I have Le2 = Le_bend_top = 0 (continuously braced by sheathing/roofing) and Le1 = Le_bend_bot = segment. For the unbraced length of the bottom chord I have Le2 = Le_bend_bot = 0 (continuously braced by sheathing/attic) and Le1 = Le_bend_top = segment. Are these reasonable assumptions?
6. I have the "Cr" box checked for all members because it's a repetitive truss every 2ft.
7. I have the ends pinned for the web members and fixed for the chords.
What else am I forgetting? Are there any other "tricks" for using RISA3D for truss design?
If you can explain to me how to attach my risa file to this message, I'd be happy to.
Thanks.

 

[CELinOttawa]

Timber trusses are notoriously difficult for the average Structural Engineer to design/check, regardless of software used. They are normally built to a highly specialized design procedure out of an industry association.

You cannot just apply regular methods as listed above; That is hopeless. Note that even though there are industry "norms", most manufacturers will get even more refined by adjusting the load capacity of their tooth-plate connectors through testing.

You may want to try

http://design.medeek.com/calculator/calculator.pl

. Medeek, one of the members here, has been researching truss engineering and design. That script is one he produced and I have found it valuable. I highly doubt it comes close to simulating the level of sophistication that goes into the engineering of trusses these days.

Of your assumptions, only 7 feels outright wrong to me. Even the most flimsy of plate is considered to have some fixity by the truss guys, and the corollary to this is that (as best I understand) all but the absolute most rigid of press plate connections are never considered fully fixed. I suspect that this assumption (your #7) is responsible for a good half of your issues, with the highly specialized strength considerations and multiple exceptions-within-exceptions to specialist truss design applied by MiTek et al. explains the rest.

 

[DCStructures]

Thanks, CELinOttowa, for the information. I have been comparing mine with other models and I think you're right on #7. Even with that correction, I still have overstressed members and unbraced length warning messages. Any suggestions on how to simply verify that the addded 2.7psf from the solar panels can be supported, with the limited information that I have on the existing trusses?

 

[Ron]

Agree with CEL on #7. All connections should be pinned.

At the bottom of the "Reply to this thread" select upload to engineering.com and it will prompt you to browse your files for the file you want to upload. Follow the prompts and your attachment should magically appear.

What version of RISA 3D are you using?

 

[medeek]

Can you attach or post a picture of your truss? What kind of truss is it?

Unfortunately, my truss calculator at this stage is still very limited. I've only just finished up the plate engineering and that is only for the fink type truss (4/3). The other trusses you can select are only given a rudimentary geometrical and axial force only analysis if you can call it that.

The fink truss however is fairly complete at least for the one manual load case that you assign it. I have yet to add load cases for wind, attic loads, eave loading, unbalanced snow loads, dead load only etc... for a complete analysis.

If you check my numbers from the matrix analysis against STAAD or RISA you should get the same results, assuming the same fixed, pinned connections at the heel joint and peak joint. All webs I am considering as pinned joint truss members (axial forces only). I have checked my bending, shear and axial numbers against RISA, STRAND 7, STAAD and COSMOS (solidworks) with good agreement.

Notes that the choice of lumber in the top and bottom chords (relative stiffness compared to each other) makes a difference in the forces and deflection of the truss. You also have the choice of fixing, pinning or partially fixing the peak joint.

Again the only truss type that can be analyzed at this moment is a fink (4/3) truss:

 

[woodman88]

What lumber grades are you using? Smaller fab. may have only two grades in stock, #1 for chords and Stud/Standard for webs. The bigger one may have 2100f MSR or higher to use for chords.

Garth Dreger PE - AZ Phoenix area
As EOR's we should take the responsibility to design our structures to support the components we allow in our design per that industry standards.

 

[DCStructures]

Thanks, guys. Okay here's one of the models (attached).

 

[medeek]

As to the specialized methods used by the truss industry and big players like Simpson Strong-Tie and Mitek I agree everything is a black box and proprietary, hence the reason I have been interested in researching this particular field and attempting to expose its inner workings (engineering).

Regardless, the standard by which all MPC wood truss manufacturers must comply and the bible for MPC wood trusses is the ANSI TPI 1-2007. You will notice that most of what is contained with chapter 7 (wood member design procedures) and chapter 8 (metal connector plate joint design) parallels very closely with the standard equations from the NDS. However, there are a number of specialized equations that have been developed by further research and are unique to this publication/standard.

My analysis of the fink truss in the calculator listed above attempts to comply fully with both of these chapters from the TPI 1 document. I am hoping that peer review at some point will help confirm or deny whether I have been successful at that goal and where correction need to made I will make any adjustments as necessary.

Recently as I have been working on a different application I did find that my reference design values for Southern Pine were slightly incorrect for lumber sizes larger than 2x6. I am addressing that bug.

 

[DCStructures]

And here's the other truss (attached).

 

[CELinOttawa]

Well I still think you'll have a heck of a time getting the "true" industry results...

As for the truss guys having to comply with an industry standard, that is true only up to a point. For their analysis this is true, but where they have testing results which give them higher values they are allowed to use them and certainly do.

As to the unbraced length issues, this is another you cannot hope to address. The plates allow the truss manufacturers to have something akin to a 0.7 effective length factor since the plates "hold" the weak axis. Medeek may be able to tell you if this is in the TPI code, failing that it is still wholly proprietary information.

 

[medeek]

I apologize I don't currently have a copy of RISA. I have a copy of STAAD however its import function does not give me the option for RISA files.

 

[DCStructures]

medeek, you can download a demo version of risa3d, if you want:

http://risa.com/forms/demo_request.html

 

[medeek]

Looking through your list of options listed above, they all look pretty good to me.

For the top chord your highest axial compression will be in the panel closest to the heel joint. You will want to check the bending moment in the top chord at this panel at mid panel and at the first panel point joint where you will see reverse bending.

For the bottom chord again you will find the highest forces and moments in the panel closest to the heel joint.

You do want to apply the distributed loads to the top and bottom chords. If you resolve them into point loads at the joints you will loose all information about the bending of these members and will only have axial forces as your result.

 

[DaveAtkins]

woodman88 is correct--I think your problem may be assuming the lumber is weaker than it actually is. Most of the time, MSR lumber is used for chords in trusses. Since it is a new building, you should be able to get a copy of the truss shop drawings.

DaveAtkins

 

[KootK]

I second Medeek's comment regarding the ANSI/TPI document. You'll need that in order to get results that approximate a truss fabricator's design.

I wouldn't get too excited about analysing the members unless you've also go the means to check the plated connections. Often, it is those plated connections that are the limiting factor in the design. Truss manufactures don't design the the connections to exceed the capacity of the joined members or anything like that.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[XR250]

2.7 psf is not much. Usually, there is enough fat in the 20/10/10 (in our area) design loads for the extra weight.
Rarely is the top chord or bottom chord DL 10 psf.
The problem with solar panel racks is that they only attach at 4' centers (usually) so only every other truss gets loaded. Installing strongbacks in the attic under the top chord can fix that by distributing the load to all the trusses.
Def. wasting your time trying to analyze a MPCT.

 

[DCStructures]

So what would you guys do in this situation? The trusses and connections were designed for 30 PSF snow load and 30 PSF roof live load, assuming they were designed to code. The solar panels weight 2.9 PSF. How would you verify that the roof trusses can support the solar panels?

 

[XR250]

2.9 psf would get lost in the noise of that loading. I would add the strongbacks as I previously mentioned and not lose sleep over it.
Unless there is a tile roof or multiple layers of sheetrock on the ceiling, the 2.9 psf would be covered in your surplus dead load capacity.