RISA And OWSJ

[olivei34]

I am modelling an OWSJ on RISA with just a typical uniform load of 900 lbs/ft.
The joist is 12'-0 long all with 24" panels.
I have made the end pieces fixed and every other node on the program unrestrained.
After analyzing the and checking the chords I notice that the axial force in the top chord is almost half the force that is in the bottom chord.
This is obviously wrong. I am expecting a similar force in both the top and bottom chord.
Has anybody run in this type of problem. I know I am not modelling this correctly.

 

[RFreund]

I'm not exactly following your setup but a few things to check.
Supports: pin - roller?
Are the top chord and bottom chord segmented (separate members) between panel points?


EIT

http://www.HowToEngineer.com

 

[msquared48]

Depending on the joist you are using, I would pin the supports. Depending on the connection of the diagonals to the top and bottom chord and the type of diagonals used, I would consider fixing those connections. If the diagonals are just rods, I would pin those though.

Mike McCann
MMC Engineering

http://mmcengineering.tripod.com

 

[olivei34]

I am just using it for checking old open web steel joist.
So both bottom and top chords are parallel and have been modelled as continous members along the joist. Panels are at 610 mm apart.
In the end if I model with both ends as pinned restraints or as a Pin-Roller system I get different results.
What is the appropriate way to go about this situation?
Again these are just typical joist, not a arched or scissor type joist.

 

[amecENG]

I have questioned this before also. I actually think I posted a thread on this here before. Pin-roller will give you same forces in both top and bottom chords and is the way joist manufacturers analyze it but in reality the shoe connection is actually pin-pin.

I don't have RISA but in STAAD I typically design trusses or owsj with all members specified as TRUSS type (i.e. only take axial loads). I'm sure RISA has a similar command. This way you don't have to worry about changing all of the end conditions

 

[BAretired]

I have made the end pieces fixed and every other node on the program unrestrained.

The end pieces are not fixed. You should model the end supports as pin and roller because this is the way they are deemed to act.

BA

 

[olivei34]

Maybe I'm just stirring the pot, but joist are typically welded to a steel beam or wall plate cast in a wall. I understand that we should assume a typical weld as a pinned support. But joist are welded on both ends, which does not come accroos as a pin-roller condition. Can anybody share some insight on this issue?

 

[BAretired]

If you assume a pin-pin condition at the joist ends, you assume that the supports are capable of carrying a substantial horizontal reaction without deflecting horizontally. That is not the case in typical structures. While it is not precise, the best and most conservative assumption is that the supports will deflect laterally as required to render the condition essentially pin-roller.

Examine your own RISA output and tell us the magnitude of the horizontal reaction at each end of the joist. And then tell us whether the supports can supply that reaction without deflecting.

Welding a shallow joist shoe to its support does transfer some moment, but it is usual to ignore it as the shoe is not capable of carrying much moment and the weld is not designed to carry moment.

Joists may be made continuous by welding both top and bottom chords to the supports, but that is not standard practice.



BA

 

[dhengr]

BA says.... “The end pieces are not fixed. You should model the end supports as pin and roller because this is the way they are deemed to act.” There are a number of different definitions of fixity, the worst being what the software assumes when you tell it fixed/fixed. The program assumes no rotation or translation in any direction, not a likely real world condition. He says this for several reasons:

(1.) That is how we learned basic simple beam theory, one end pinned and the other on a roller. Do they still make any effort to teach that in school these days, or isn’t that theory needed any longer since the software takes care of all that stuff, in its own befuddling way. And, leaves us wondering why we get funny stresses (forces) in some of the truss members. Or worse yet, causes us to allow a significant error to go unchallenged, or corrected. But, you are questioning it, so, good on you.
(2.) It’s been done this way for years, that assumption simplifies the analysis significantly, and has not caused stl. jst. failures. It allows us to ignore the secondary axial affects, for a very good approximation of the real world conditions.
(3.) To design a single simple open web stl.. jst., we can’t afford to spend more than six months worrying about secondary affects which we have learned over time do not lead to problems with the final design. However, there have no doubt been Master’s and Ph.D. theses done on these very subjects for the SJI. If you have two angles for t&b chord members, and sq. or round bars for the diags.; the members are all pretty stiff w.r.t. axial loads, the primary force system; but the diags. are considerably less stiff (strong) w.r.t. bending than the t&b chords are, and it is tough to get our heads around exactly how the welding btwn. the two would transmit a moment anyway. Suffice-it-to-say that the welds or diags. might yield a little, over small areas, to accommodate these secondary moments. This same thinking is not true (or proper) for larger structural trusses when we are talking about W14 t&b chords and W14 diags. In these cases those secondary effects must be accounted for, in detailing, welding and analysis.

Olivei34 says.... that the welded ends don’t sound much like pinned and roller conditions. But, this assumption is required for our typical simplified analysis approach. Olivei34, answer these questions, answer your own question/problem, and post your thinking and we can talk some more. Assuming you control the stl. jst. defections to reasonable limits, what is the difference in the straight length (chord length) of the top chord, and the curved length when deflected, 16ths or .01's of an inch, if you can measure it. Are there things (details) within the roof system which can accommodate this length change without causing bldg. or jst. failure, or things which will resist it, and make it less critical? Then again, if you tell the software the ends are fixed, what does this length change mean in the way of axial strain (thus, large stresses or forces) in the top chord? End fixity will also induce bending moments which the top chord can’t really accommodate in the immediate area of the jst. seat, so the members or the numerous welds will yield a bit to tolerate some moment, so we call this condition pinned. Again, various degrees of fixity, or exact definition. There is a moment on the end of the jst. to the extent that the first diag. and the reaction point are eccentric by a few inches, and this must be accounted for in the jst. design, but it is not a fixed end moment. Compare your first run/model with one which is pinned and roller at the reactions, the t&b chords are continuous members, but the diags. are pinned/simple supports for the t&b chords at each panel point. In either case, the moments and forces should balance at each of the panel points (nodes), and you should be able to see how your fixed/fixed assumption radically effects the member forces/stresses.

 

[JoshPlumSE]

I'm a little late coming to this thread, but here is my 0.02 $.

The Pin-Pin or Pin-Roller question is one that comes up frequently here at RISA. I generally recommend a Pin-Roller assumption for trusses. But, I also tell folks that they can run some easy tests to see if that assumption is realistic.

Run the analysis with pin-pin boundary conditions. Make note of the (probably very large) horiztonal reaction at each of the two supports. This is the force that the end-support of the truss is required to resist in order for that boundary condition assumption to be valid.

Then run the analsyis with the pin-roller boundary conditions. Make note of the horizontal deflection that you get at the roller end of the truss. Generally speaking this will be a relatively small deflection. In order for the boundary condition assumption to be valid, this is the type of deflection that the supports must allow.

At that point it is usually clear that the real supports will be much closer to the pin-roller assumption than the pin-pin. If it's not entirely clear then I tell them to using the pin-pin assumption but with the horizontal restraint being given a spring constant commensurate with whatever structure is supporting the truss. At that point you will get a result in between pin-roller and pin-pin, but you will probably notice that the results will be much closer to pin-roller.

 

[MiketheEngineer]

Using wL^2/8 /d your answer should be in the neighborhood of 16,200 / depth in feet top and botto. Just a good double check.

 

[olivei34]

Thanks a lot; this has definitely made things a lot clearer. I have been using the wL^2/8 /d to back check my work for these simple load conditions.

 

[southard2]

I used to work for a bar joist manufacturer. You can call a manufacturer and I'm sure they will tell you this. But we always modeled all member to member connections as pin-pin. The end reactions were considered as resting on a hinge on one end and a roller on the other. Regardless of how it might react in real life this is how OWSJ have been designed for a long time. The only time you would model the end as fixed is if the bottom chord was extended and welded. In this case the top chord and the bottom chord are used to resist the moment at a column for example.

But for a typical joist the modeling is pin pin everywhere. The joist fabricator than makes sure that the webs have enough contact length for the welds to resist the forces properly. In general the weld all around versus specifying weld lengths. They don't want the guys in the shop having to think to much. IF more weld is needed than you add a gusset plate.

John Southard, M.S., P.E.

http://www.pdhlibrary.com/

 

[BAretired]

southard2, you are confusing the issue. The joist is not typically modeled as pin-pin. It is typically modeled as pin-roller.

BA