Wood Truss Analysis-Modeling 1

[RFreund]

Questions on evaluating/designing a wood truss:

Assuming loads are applied as distributed loads (i.e. purlins are closely spaced or no purlins and sheathing distributes load)

Q1. Should the top chords be modeled as continuous over panel points (assuming that they are continuous) or should a "pin" be used at panel points so that the moment at the panel point is zero for the top chord.

I thought that I would get smaller (absolute value) bending moments when I modeled the truss with a continuous top chord. As opposed to modeling the top chord as simple span between panel points. However when I model the truss in Risa that is not what seems to happen. I believe what happens is that when the top chord is modeled as continuous member some the the tension or compression from the web members effects the top chord bending moments. This also gives me lower tension and compression values in my web and chord members. So my questions are:

Q2. Is the analysis wrong with continuous members? I realize that it may not be "wrong" however the moment near the base of the truss in the top chord is 3x that of the simple span and when unity is checked the member fails. So I guess a better question might be:

Q3. Can a trussed by analysed assuming the cord is simple span between panel points even if it is built with a continuous member?
Q4. What type of analysis is typical used?

Thanks

EIT

 

[XLjunkie]

In my opinion modeling top chord panels as being pinned on both their ends will yield a workable solution.
However the model you're creating mathematically does not match the behavior of the structural element in real life.

If i had a choice id model the chord continuous over panel points with the web members considered pinned to the joints.
Most industry leading programs model wood trusses as such.


In reality there is a bit of moment transfer between chords and webs by nature of the metal plates' ability to carry some moment.
Although that's generally not included in the overall design of the frame. (having said that the Canadian code permits a 20%
decrease in the effective length of a compression web due to the fact that metal truss plates can carry some moment at the joints.
I would not use this argument for plywood detailing of joints though).


BARetired, i agree with your point to a certain extent. Ive worked in this industry and can tell you the problem is not with design adequacy.

The problem is in the regulation of who is allowed to do this type of work. Generally speaking you don't need a degree or any conceivable educational background
to design and build trusses. All you need is a powerful computer program and basic understanding of required inputs. and an engineer to back your design.

Its scary that many truss designers Ive seen have no relation or interaction with the structure as a whole, nor do they care to. Where EOR's view trusses as designed solutions,
most truss fabricators view them as a commodity product.

Its scary to know that people with no education often detail trusses that are 80-100 ft. But the factors of safety in this industry are substantial enough that the industry generally has a good track record.

 

[msquared48]

The problem I have with this shift in responsiblity is that the truss designers, in order to be competitive with others in the industry, may or may not employ certain bracing assumptions or requirements in their calculations to save materials.

Under this scenario, it is NOT the responsibility of the EOR to either assume or identify these conditions, but it IS the responsibility of the truss designer to specify such. It is also the responsibility of the truss designer to calculate the load that goes into the bracing.

The conflict that is presently in the mill is "Who will do the detailing of the bracing connections?" I would not say that structural engineers do not know how to design a wood truss. That is not the problem.

Mike McCann
MMC Engineering
Motto: KISS
Motivation: Don't ask

 

[RFreund]

Even though you have answered the questions of my OP. I would like to on Monday try to post the results of different of the analysis of pinned vs continuous top/bot chord as I have some specific questions. Mostly due to the fact that it seems the top chord fails a unity check when modeled continuous and passes in when pinned. I will need to double check the model in the mean time.

Thanks again.

EIT

 

[BAretired]

As others have said, the problem is not with design adequacy. The problem is with bracing of web members and the complete lack of understanding of bracing requirements on the part of framers. Two typical scenarios follow:

(1) On a pitched roof, the compression web members are not designed to carry the calculated load over their full length. The shop drawing shows a symbol at midlength or third point signifying required brace locations. When the trusses all have similar configuation, the framer runs a horizontal ribbon (usually a 2x4)connecting the web members with two 3" nails at each connection. They invariably believe this constitutes adequate bracing, failing to recognize that all attached web members can buckle in the same direction. What they must do is provide a certain amount of 'X' or 'V' bracing at reasonable intervals.

(2) On a pitched roof with variable truss configurations, the problem becomes a good deal messier. Again, the computer spits out brace locations for each truss design but the framer doesn't know what to do because the adjacent trusses have different web configurations. He has nothing to nail his 2x4 ribbon to so he improvises, usually badly. The EOR has to specify bracing on the fly and somehow explain it to the framers. After they attempt to accommodate his instructions, he has to re-inspect the work and re-explain where bracing is still required.

In the above two scenarios, there is no drawing showing where bracing is to be placed other than a symbol on the truss elevation. Without the necessary bracing, the trusses could collapse.

BA

 

[RFreund]

BA - I understand both point completely as I remember coming across both situations as a framer.

I'm going to post the analysis results (bending, axial, and shear diagrams) of three different models of the same truss.

1. The top/bot chord is modeled as pin-pin between panel points
2. The top/bot chord is modeled as a continuous member
3. The top/bot chord is modeled as segments between panel points but it has fixed ends.
Webs are pinned for all.

After reviewing this I believe my question should really be to Risa (not sure if we renewed our license) but any input would be appreciated as to explain the results. In particular why the bending moment is so large in the continous top chord model near bearing. Also it appears that the results modeled as continuous are less conservative when unity is checked. Let me know if the results are too cluttered to read.

Thanks again.

EIT

 

[bill30206]

Long thread--I didn't read it all; however, it looks like you're talking about dimension lumber trusses that fall into the TPI Standard category, albiet with plywood gussets and loose fasteners. This type of truss is more properly seen as a frame, i.e. there is continuity across the joints on straight chord segments and you would not pin the chord segments at these locations. At pitch breaks you should model the joint as a pin. If there is a closed heel joint between top and bottom chords, there is no clear answer, and I would suggest at least considering an envelope solution to assure that you have the highest forces for design of the heel joint, and the most critical intra-chord positive moment in the top chord. Better yet would be to model the heel (assuming again a closed heel) with a three-node setup and small fictitious members in the heel area. The Canadian truss standard (TPIC) has prescriptions for these heel analogs, whereas the U.S. (TPI) does not. The Canadian standard was available online for free, last time I checked. MiTek's software doesn't use fictitous members while ITW (Alpine) software does. When you get your moments and forces of the members you can chuck them into an excel spreadsheet using the TPI formulae (which are substantially identical to NDS) for stress interaction and moment magnification. If your analysis program gives you, say 50 sections of data, then you can easily have Excel give you unity equation results. I have done this and produces nearly exact matches with commercial software (MiTek) results. Then you have to design the joints, splices, heels, etc. Make certain to check deflection. Most plywood gusset trusses are seat-of-the-pants creations; the heel joints in particular are seldom of sufficient capacity.

 

[XLjunkie]

RFReund is the hatched area in (2) and (3) your moment diagram? Doesnt look correct, there should be a stress reversal at each joint location, where the moment will be negative.

If you give me the span and pitch i can run the model in one of the truss programs i have available, and check what the bending moment diagram of the top chord looks like. The diagram should look like that of a multi-point bearing beam supporting a uniform load.



_________________
C

 

[MiketheEngineer]

Usually have a bit more luck with bolts - but that can still get messy. Also depends on chord sizes.

As for using nails - I usually stay away from them on existing trusses - too much chance that you will ruin the members and/or crack the drywall. Using a nail gun does help somewhat.

 

[BAretired]

RFreund,

I'm not quite clear on the difference between Condition 2 and 3, but there seems to be something wrong with the chord moment diagram in the vicinity of the end reactions for those conditions. It is almost as if you forgot to include the end vertical members in your geometry.

You have chosen a web configuration using tension diagonals. More common in the truss industry would be to reverse the direction of the diagonal members, i.e. compression diagonals and tension verticals.

BA

 

[RFreund]

Mike - Thanks, I'm figuring bolts.
BA- The current design shown is an existing condition. All though I thought that it was a little different.

I've attached the moment diagram for case 2 and 3 shown more clearly and I have shown the geometry. Also I try to show the difference between how I modeled case 2 (a continuous member) and case 3 (separate members that are fixed to each other). They should yield the same results. I believe your BA in that there is a problem with the last vertical because when I removed the vertical it did not change the moment diagram for case 2. When I remove them from case 3 it decreases the moment.
However I believe that you guys have answered all of my questions and know my problem is with understanding RISA and my model which is probably going to be difficult considering you don't have access to my model but if you have any suggestions as to what looks wrong or why that would be great.

Thanks.

EIT