hypothetical modeling a 3 guywire mast in RISA

[adamewood]

Per the subject title, what we have is a 20ft tall mast on a hospital roof with 3 guy lines attached near the top. They are not spaced equilaterally, but let's assume they are, and each guy is of varying length and vertical slope since they attach to accessible column stub-ups on surrounding lower roofs (about 40ft elevation from top of mast to top of lower roof, or 20ft below the base of the mast). Technically, this mast is a vent stack, but for purposes of this problem, assume that MEP has provided you with a giant self-supporting plastic noodle that can't buckle but will tip over if someone breathes on it.

Now say you were given the task of modeling this system in RISA. You follow the proper RISA protocol for cables by setting their density to 0, making them T only members, and giving each just a little pretension... however you soon realize that it's first order statics which RISA is having trouble with: Say the wind load, acting along the building X or Y axis, happens to be within 30 degrees (say it's at 29.99) of one of your equilateral tension-only guy-lines; you end up with the two guys over 90 degrees away loaded in compression (so they fall slack), and then just the one tension guy holding everything up. And with your wind acting at 29.99 degrees out-of-plane, RISA will only allow that guy to support 0.866 of the load in-plane, so the 1/2 wind load normal to this goes unsupported, resulting in at best, RISA grounding you by locking your joints, or at worst, RISA giving you the boot for causing an "unhandled exception".

To avoid such a fate and retain your reputation as a RISA master-modeler, what do you do to get this model to behave realistically (giant noodle MEP mast notwithstanding)?

 

[3DDave]

Why only a little pretension? Guy wires should never go slack.

 

[271828]

Does RISA perform the large displacement second-order analysis required for this? Recommend working a problem with a known solution like a catenary out of a textbook first.

 

[adamewood]

No catenary effect since RISA instructs you to zero out cable weight. RISA does do second order displacements, would this not be first order though? the program hangup comes as soon as the statics reactions are attempted to be solved - in essence, I think RISA goes "total load in X direction = 1/2 x P, restraints in X direction = 0, therefore error"

3DDave - I think you may have just laid this question to waste. I did not know guys should never go slack - shows how many mast designs I have under my belt. I was actually thinking to add another cable so all are 90deg, but pretension the one to over 0.866 x P would keep the "compressive" side from going slack in theory (I had them all at 500 lbs). I think you get a gold star. Thank you

 

[dik]

I would think that you would want the pretension to be minimal. The pretension exerts a downward force on mast x3... as long as it's not 'slack', it should work as intended.

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?

-Dik

 

[adamewood]

Yes it would - the mast is "by others" though, so as long as loads aren't insanely high.

I think they all need different pretensions due to varying angles to the vertical. One is at a pretty high angle (50 deg), so that'll be interesting. Unfortunately, this roof is like the hospital crows nest.

 

[rb1957]

I don't know RISA, but the more I hear the less I like !

If there is pretension in the guys then the guys are capable of reacting some "compression" load, until the nett load in the cable is zero.
This means that initially all the guys are effective then one would become ineffective then the other, so the stiffness of the guys changes.
Maybe try different levels of pretension ... all very high so all are fully effective, then reduce one to very small, etc.

The simple analysis approach is to say zero load in compression, simple vector component gives the load in the tension cable, simple free body tells you the reaction at the antenna base.



another day in paradise, or is paradise one day closer ?

 

[JoshPlumSE]

Okay, here are my thoughts:

1) I think you should re-visit your pre-tension. Generally speaking you want to design your pre-tension to avoid this unstable situation. Is there some reason why you can't specify some more pre-tension? Is there a construction issue or such?
2) You probably already know this, but RISA is not the best program for modeling cables and cable behavior. That being said, for straight guys like you're talking about, it's probably good enough. As long as you're not worried about the non-linear geometric effects of the cable sag and such.
3) One thing that I like to do when the model is nominally unstable is add a really flexible spring. This could be a rotational spring at the base or a translational spring or such. Flexible enough that when the model is stable you don't get any force or moment in it (at least not compared to the real reactions). But, by introducing this spring you prevent any instability from side-tracking your analysis. But, you have to be careful about monitoring reactions and displacements and such.... Because you are "faking" out the analysis program. So, be careful.

Disclaimer: Many here already know this, but I am a former RISA employee and I now work for a RISA competitor (CSI / SAP). I'm not intentionally being biased in this post. However, I don't want to mislead people into thinking I'm an impartial engineer.

 

[rb1957]

why not consider the guys as struts so RISA (or other FEA) doesn't have to worry about a loss of stiffness ... a much better behaved model.

then determine the largest compression load, then set preload to be 50% higher.

then add this preload to the largest tension load, then size cable for 50% higher load.

another day in paradise, or is paradise one day closer ?

 

[JoshPlumSE]

rb1957 -

I like that idea... provided that the OP doesn't have a problem with a higher pre-tension.

 

[271828]

"No catenary effect since RISA instructs you to zero out cable weight. RISA does do second order displacements, would this not be first order though?"

I have tried using tension-only in RISA and could not convince myself it was handling the calc correctly. Thus I'm leery from the get-go.

There are lots of types of analysis.

With first order, the equilibrium equations are written for the initial geometry; with cables, the geometry can sometimes change quite a bit without much of a change in stress. That's the opposite of what's assumed with first order.

Then there's "small displacement P-Delta" which usually uses the initial geometry but adds an amplification of some form, such as with a geometric stiffness matrix, to accomplish a second order analysis.

Finally, there's large displacement second-order that typically uses an iterative scheme to take into account the changing geometry as load is increased. In general, this is more the direction to go for cables. Your application might be a specialized case that doesn't need this level of analysis, however.

 

[JoshPlumSE]

271828 -

To be clear, RISA does not account for geometric or material nonlinearity. The only caveat to this is the P-Delta analysis.... Which doesn't modify the stiffness matrix at all, but rather adds secondary shears to account for the P-Delta effect.

For non-linear elements (like tension only members, or compression only springs) the program merely iterates the analysis turning those elements on an off until it converges on a solution. You would get the same result if you used two way members and springs and manually turned them on or off based on the deflected shape. It's just challenging to manually do all that iteration.

 

[adamewood]

You guys all make good points. I should point out that I was able to get the model to behave. RISA makes you input pretension as a change in temperature... so guess that is how I will report it! one guy ended up with -18 deg F while the other two ended up with -12 deg F pretension. Total mast base reaction is 5 kips vertical, which sounds reasonable.

RB1957, that is what I gleaned from 3DDave's post about never slack cables. While it does work for one specific wind direction, I still think problems arise when tensioning all the cables for any possible wind direction because the pretension forces would completely negate each other in the model so that the wind ends up without resistance. I found that you must vary the pretension slightly to avoid this.

271 & JoshPlum - fantastic points, secondary effects are approximated by force magnification, so it makes sense that RISA would iterate in this regard rather than by adjusting geometry. Adding springs at joint locations is a technique I often use to "fake" reality, but my issue is the inclusion of a single spring in 1 direction often leads me down a slippery slope of OCD justification for springs at all similar joints & directions - so I try to treat that as my nuclear option.

I think the general takeway may be summed up, saying that RISA just isn't cut out for tower mast problems, which is surprising as, at first glance, it looks like it should be one of those problems in which RISA excels. (I'm sure JoshPlum feels differently, but despite all my griping, I still have to say that RISA is by far the best design program we use - I just think they strike a good balance between versatility, speed and user friendliness - plus its the only such program to actually look like it was developed after 2010.... if only they would just make it so the mouse wheel can zoom AND rotate the view...)

 

[rb1957]

"RB1957, that is what I gleaned from 3DDave's post about never slack cables. While it does work for one specific wind direction, I still think problems arise when tensioning all the cables for any possible wind direction because the pretension forces would completely negate each other in the model so that the wind ends up without resistance. I found that you must vary the pretension slightly to avoid this." ... I have almost no idea what this means !? My suggestion was to dump the cables, model as simple rods, then use the results to determine the design (pre-tension to avoid compression). I don't see why "problems arise when tensioning all the cables for any possible wind direction because the pretension forces would completely negate each other in the model so that the wind ends up without resistance.". Using temperature for pre-tension is reasonable, though very olde school.

another day in paradise, or is paradise one day closer ?

 

[3DDave]

"I still think problems arise when tensioning all the cables for any possible wind direction because the pretension forces would completely negate each other in the model so that the wind ends up without resistance."

All that happens is that for a non-zero wind load there is a non-zero displacement that balances a shift in guy wire tension against that wind load. This is no different than any load applied mid-span of a member under tension. There will be resistance.

I suppose one could use a very large section modulus for modeling the guy wires and a matching decrease in the elastic modulus to prevent problems from buckling that the nominal wire characteristics might produce, also avoiding the problem of evaluating a catenary.

 

[adamewood]

rb1957 - that is what I did to solve for pretension in one cable, which works just fine even after setting it to T only. However, I then set all three to that exact same pretension, and rather than solving for each individually, which resulted in the same error as when all cables are at 0 pretension (or all at 500, or all at 2000...)

3DDave - what you describe is what should happen when you click solve, but I tested setting them all at 20 kips, which is higher than the total wind force on the stack. Same error. I think perhaps RISA looks at the NET force in each direction and determines which "T-only" cables are downwind, then zeroes them before it has the chance to complete a single iteration. It's possible I just need to download the latest update... but I am curious if anyone else tried doing a similar RISA model, what their results were.

 

[3DDave]

They are at an angle - the tension load can be much higher than the wind load. If you post a diagram of all the dimensions and how the load is applied that would clarify. If they are not rotationally or mirror symmetrical, setting them to the same tension will cause a large displacement on the initial iteration; if the solver is not set to do large displacement it would likely crash.

As I wrote - increasing the section properties and decreasing the modulus allows them to act like beams with the reactions of guy wires, except that beams can accept a compression load. If that does not work you may simply have built the connections incorrectly and all this has nothing to do with the tension.

RISA is unlikely to know what "downwind" is. Most solvers create a matrix and, for T-only conditions, would set any negative solutions to zero before running the next iteration.

 

[1503-44]

Why would you not want at least some tension in all guys for all wind directions? Just from the point of providing lateral stability, it would seem advantageous. A 3 guy tower with wind in the same direction as one guy could potentially leave two guys unloaded. Any shift in wind direction would translate the tower rapidly until another guy picked up enough tension to arrest translation.. That could add significant dynamics and potential impact load.

 

[rb1957]

I have no idea why RISA should be unhappy if all the cables have the same pretension.

I think we're talking at cross purposes. I make a suggestion about modelling the cables as rods (capable of reacting compression), you say "that is what I did" then talk more about cable pre-tension ?

My thought is that RISA's modelling of cables may be "fussy", certainly fussier than modelling axial load elements. So ditch the cables, model as axial load elements, then set the pre-tension, then check the combined tension + pretension.

another day in paradise, or is paradise one day closer ?

 

[adamewood]

3DDave - I do like the idea of creating a new low modulus material to behave like a cable. I will try this when I revisit the model.

rb1957 - to clarify, in order to obtain an approximate value of the amount of pretension I would need in the cables, I temporarily allowed compression in the rods/cables and disabled slenderness checks. After obtaining the static reaction results, I re-enabled slenderness checks and switched cables back to tension only. This is the procedure I thought you were referring to... but are you saying to leave them as rods? like forever? Ain't that cheating?? - I think the kL/r police will put out a warrant on both our heads if we go down that road! dunno about you, but I'm tired of running...