Question on Vibrations 3

[Lion06]

I am going through Design Guide #11 and I have a few questions. Does everyone perform a vibrations check for every building they do by hand or do you let something like RAM do it? In reading through the manual, it seems like so many things are left to the discretion of the designer that by the time you compare your criteria to the suggested limits in the Design Guide, it doesn't seem like it would be as reliable as, say the buckling strength of a beam. Does anyone else feel (or did you when you first started out) a little overwhelmed by this subject?
Also, where do you find vibrations info for concrete, I know it has a lot of mass and as a result, the accelerations are low and likely not a problem, but I am sure it must still be checked.
I have info for wood, although it has more to do with deflection limitations to conrol vibrations.

 

[csd72]

Structural EIT,

I can remember at a conference they said that in vibration sensitive buildings you would design for vibrations first, then check for strength after, usually it was the opposite.

Normally I would only check the most flexible members in a frame.

In my experience, vibration for concrete is only checked for vibration sensitive areas such as hospitals, precision industries e.t.c. This is very complicated subject that is probably best left to specialists IMO.

I have seen a few good references on timber floor vibration, I recommend you look up the following website:

http://www.awc.org/HelpOutreach/faq/FAQfiles/Vibration.html?areaID=1

csd

 

[271828]

For steel buildings, I recommend checking every representative "typical" bay using FloorVibe. It only takes a couple of minutes per bay.

You are kinda right about the lack of reliability, for a few reasons.

It is almost impossible to estimate damping with any reasonable precision. That's the one that folks already know.

Natural frequency prediction is the easy one. However, they can only be reliably estimated within about 20% using the best finite element modeling techniques (design model, not a reseach model that has been "adjusted" or "updated" to match experimental results) that we have now, much less the simplified methods in DG11 or the UK SCI DG. Even with the best FE models, prediction is not good. I've personally done full modal tests of two real buildings and I've read papers from others who have done the same. Real buildings have multiple vibration modes, sometimes closely spaced. Finite element models get in the ballpark of the frequencies, but the modes will be shaped a little different and will be in totally the wrong order!! The FE model will also have the modes much tighter spaced than the measurements indicate. Sometimes, there will be 10 modes between 6.0 Hz and 6.5 Hz, for example. Any claim of nailing the first several modes using a FE model with any reliability casts serious doubt on the engineer's credibilty, IMO. Now go back to the simplified methods and consider the luck one will have in nailing it.

Here's the big one: Human walking forces are almost totally unpredictable. I've measured many of these myself and have studied the literature. These are always broken down into four Fourier amplitudes. For example, the first will be about 0.5*bodyweight, the second 0.2, third 0.1, and fourth 0.05--for example. The problem is that a plot of these amplitudes looks about like a shotgun blast!! There is literally NO pattern in the 3rd and 4th harmonics of the walking force and these are the ones usually used in design. For example, you might *think* the 3rd harmonic is 0.1*bodyweight, but the data shows A LOT of points anywhere from 0.01 to 0.16. It literally looks like a shotgun blast when plotted versus step frequency. Until we can get people to walk "better" there's no hope of really nailing an acceleration prediction.

All that being typed, DG11 Chapter 4 was developed by a couple of guys with 30-40 years each of experience, so has been tweaked and fudged so that it works reasonably well. You can use it with a fair amount of confidence. It won't nail the prediction, but your floor will almost certainly be ok if you correctly apply chapter 4 and it predicts the floor to be ok.

For the long haul, the most sophisticated techniques will be probabilistic, but these are several years away. For example, future criteria will allow you to tell an owner that he'll have a 90% chance of no complaints in a given area.

For concrete, you can adapt DG11 or use the new UK SCI DG General Assessment chapter which is directly applicable. It's fairly difficult and requires FE modeling and a solid grasp of vibration theory to be able to use it at all, though. Vibe is almost never an issue for concrete unless you have sensitive equipment.

 

[lkjh345]

StructuralEIT:

Concur with 271828. We also use FLOORVIBE for steel framed buildings. Very quick and easy to check out several possible layouts and combinations of floor thickness, beam spacing, beam size, etc.

We check floor vibration for every building. A few minutes a bay to check vibrations is time well spent.

Normally, I design 'typical' floor bay(s) for strength and deflection criteria, and then check vibration on FLOORVIBE before going too much further, to see if I need to tweak assumptions. Best to do this early on. Its a bummer to get a month into a project and realize that your assumed floor thickness isn't going to work.

 

[DaveAtkins]

If a structural floor is built of steel, with composite or non-composite deck and concrete topping, I check vibration. I use Design Guide # 11.

Otherwise I don't check vibration. I think, in general, experience indicates that the following floor systems will not have problems with vibration:
* PCC plank on steel beams
* reinforced concrete (flat slabs, one way joists, waffle slabs)
* wood

That is my opinion--I look forward to hearing others' comments!

DaveAtkins

 

[jike]

With most offices using the "open concept" layouts (half height partitions, etc.) today, I always do a vibration analysis for every office building we design. As stated previously, I have found some inconsistancy between the calcs and reality, but I try to muddle my way thru it. With large open floors and long spans, I often add an inch of extra concrete to help dampen the floor from transient vibration.

Hospitals with micro surgery and special production facilities require more of an expert.

 

[lkjh345]

DaveAtkins:

I concur. Precast planks, hollow cores, Double Tees and CIP concrete floor systems generally are not succeptible to vibration problems.

Generally, I see most lab buildings, reasearch centers, etc with sensative equipment in them designed with CIP concrete for the exact reason that these floors have good vibration characteristics.

My statement above should have read ' We check floor vibration for every STEEL FRAMED building...'

 

[jike]

My response was relative to steel framed buildings. I agree with Dave Atkins with one exception. I know of a building (not one of mine fortunately) that had a vibration problem with hollowcore plank but they were using it as a dance floor. I assume it was not ballroom dancing!

 

[csd72]

Jike,

Good point, Dance floors, stadiums and gymnasiums need special consideration above and beyond normal vibration checks.

csd

 

[archeng59]

I use FloorVibe for simple floors framed with steel. I use STAAD-pro for more complex configurations.

 

[271828]

archeng59, how do you use STAAD-pro for this? Just to get frequencies?

 

[archeng59]

yes, the natural frequency of the system.

 

[Lion06]

Thank you all for the responses. 271828, I know that you have extensive knowledge in this area. Is it a correct statement, generally speaking, to say that for a given span, the more a beam deflects, the smaller its frequency will be?
Is it correct to state that, generally speaking, in order to stiffen the floor system it is best to stiffen (and raise the frequency(ies) of the least stiff members (and hence member(s) with the lowest frequency(ies))?

 

[271828]

Q1: Yes, in fact if you apply the tributary mass to the beam as a weight per length and calc the deflection, you can get the frequency from that. In US Customary Units, if you calc this deflection in in., the natural frequency is fn=0.18*sqrt(g/Delta) where g=386 in/sec.^2 Fundamentally, the natural frequency is proportional to sqrt(stiffness/mass) or more brief: fn~sqrt(k/m).

Q2: Yes. Using DG11 methods, you'll calc the beam fn and the girder fn. If the beam fn is lower, increase the beam until the beam and girder fn are similar. If you need the frequency to be even higher, then increase them together after that. "Least stiff" is not as precise as "lower frequency" because a beam can be very stiff, but still have a high mass--remember fn~sqrt(k/m).

 

[Lion06]

Excellent!! thanks.
I have a situation in which there are W40x264 spanning 75 feet with W14 spanning 22' framing into them. The vibrations are really posing a problem and I can't increase the frequency of the girders enough to get it under control. Any suggestions you might have from previous experience would be greatly appreciated.

 

[271828]

Several ideas, probably none of them desirable:

Make the girders even bigger, add columns, reframe to get the girders closer together, place a "column" from the center of the girder up to the next level to make them work together. Try castra.. (oops) castellated beams to get more depth and still have room to install ductwork and pipes.

Put the girders the other way, so that you have 75 ft beams that you can make closer together. That's probably my best idea. For example, you might end up with W30x90 girders (mostly for depth--connections) and W36x150 or W40 beams at 4-5' apart.

 

[jike]

Try adding mass (extra concrete).

 

[271828]

jike's idea might help. Adding mass is helpful because a=F/m, so will decrease acceleration at for force input at a given frequency. However, it will be counterproductive if you're having trouble with your natural frequency being too low (likely in this case) because remember fn~sqrt(k/m). It's worth a try to both increae and decrease mass using thicker or thinner slab, lightweight or normal weight concrete, etc.

 

[csd72]

Creating continuity and fixity at the ends can help, but it is much harder to analyse.

csd

 

[Lion06]

Alright, I definitely can't start adding concrete because the W40 girder is working VERY hard. Unfortunately, I can't add columns either, because this is an area that MUST be column free below for a performance area.
In reading the design guide. it says to only account for continuity if the girder is physically continiuous over the columns, not for moment connections. Does this apply for calculating the girder panel weight only or also for the deflection calcs? We are providing full moment connections, but Floorvibe takes the deflection as a simple span deflection regardless of what end conditions you specify in RAM. If I can count on the moment connections for deflection, this well definitely help to raise the frequency by a pretty healthy margin since the fixed end deflection is 1/5 that of a simple span deflection.