Industrial equipment on a steel structure 3

[marwin23]

Good morning,

I am looking for a piece of advise how to solve the issue when placing industrial equipment on an existing steel structure.

In details: 4sty existing steel frame (columns, beams + metal deck w/concrete) carries currently some flour mill equipment (old, mostly to be replaced). Some units of the new equipment to be placed on the top floor (can not on ground) have an operating frequency between 7-14HZ, whereas it seems that structure natural frequency is close to 8.5HZ. As per manufacturer, I can not provide dampers in between unit and my floor - however here if someone has any experience to argue this statement, then it would be good.

My first thought is:
- Unit operating weight is roughly 5kips. If I pour a block of additional 15kips, then (since mass is in square root (Sqrt(k/m)->(Sqrt(k/(m+3m))) I may be able to reduce the system output frequency twice to 3.5-7HZ. Or maybe it does not work that way, and concrete block would gladly transfer 14HZ from equipment to my deck without damping it (that would be against newton laws probably, but just want to assure myself).

I probably can someone support this weight on the current structure or reinforce as necessary. But at the same time such block would reduce my natural frequency of whole structure (but that would be just a fraction of whole weight). Does this approach make sense?

If someone has any experience - I would appreciate your input!

 

[human909]

human909: I do not remember providing Manufacturer of it, but you are definitely experienced.

Why thank you! Probably the most generous complement that I have had on this forum! Though possibly somewhat undeserved, I'm still working hard to become 'experienced'. That said I do deal with Buhler equipment regularly.

The advice says better to oversize. It is an imprecise statement but that is because oscillating machinery in structures can quickly become a fairly imprecise science unless you have a team of a dozen vibration experts on speed-dial and a bucket load of cash. That said, the advice from Buhler might be based on the assumption of steel floors not concrete.

whereas it seems that structure natural frequency is close to 8.5HZ

That seems awfully high for the structure. Is that the local floor area, the entire floor, or the entire structure?

human909: based on your experience, would you find it reasonable to allow the natural frequency be below the equipment frequency (go through resonance), or rather I shall put all effort to increase stiffness. I was playing today with adding more bracing in floor plane (current floor framing attached), but results were very poor

Based on my limited experience I wouldn't worry too much about the equipment going through the natural frequency I'd be worried about the steady state point.

My experience is mostly with steel framed structures where the natural frequency of localised floors and floor beams with minimal live load is often in the 10-20hz range. This is a problem as unladed steel floors will quite happily resonate with this type of machinery. With a steel floor the solution is normally to make it stiffer. When you have significant dead load of concrete this might not be the best approach.

Seems that it is easier for me to drop it below 7HZ (adding mass) than to raise above 14HZ.

I tend to agree with you here. But make sure you ensure the localised area with only the dead load stays below the value. This runs contrary to Buhler's advice but their advice may be base on a steel floored structure where going stiffer is normally the easier and safer solution.


One more thing. Appropriate isolators can be extremely effective**. Some manufactures provide this as an option, though I couldn't see it available for this item. Generally a good isolator consists of a second set of springs, additional mass, and a shock absorber. Basically the 'spring and dashpot model' that most of us have seen at one point in our education.

**We recently installed such a system in a process plant that had very significant vibration issues. The end result was extremely effective given that the floor was previously teeth chattering to stand on.

 

[r13]

Excellent response that covers all angles and corners.

 

[marwin23]

human909: thank you very much.
Could you please

might be based on the assumption of steel floors not concrete

tell me what you mean by steel floors? Are those grating or something else? Since mine floor is concrete on steel deck (what I am quite used to in residential, but with shear studs) - do you classify it as steel or concrete?

The natural frequency of whole structure is 8.88HZ

The frequency of the said floor alone is 8.40HZ

when I look on deformed shape (modes) of whole structure (first mode, not visible here) than I see that mostly slabs deformations, whereas columns deformation is much less. Probably I would assume that for the first whole structure mode, mostly slab is responsible. That is probably why the difference between single floor and whole is very small.

 

[human909]

marwin23 I think this does deserve further analysis beyond just looking at the natural frequencies. Robot Structural Analysis can perform harmonic analysis with inputted loads, frequencies etc. This really should be your next step.

Also you should start by modeling only the bay(s) where the machinery is rather than the entire floor. Honestly I would still strongly consider stiffening up the bay, this is the most reliable option. Diagonal struts from above or below to halve the effective span length could be an approach. Or talk some experts in the field.

tell me what you mean by steel floors? Are those grating or something else?

Dimond tread plate floors. They such floors have lower dead weight so would have a normally have a higher first mode of vibration.

Since mine floor is concrete on steel deck (what I am quite used to in residential, but with shear studs) - do you classify it as steel or concrete?

Well its both (sorry for stating the obvious). I'm no expert (particularly in concrete) but at a guess the higher dead load will give you a slightly lower frequency, and the the concrete will also give you better dampening due to higher hysteresis losses.

The natural frequency of whole structure is 8.88HZ

Again, I'd check that figure. That seems very high for the whole structure unless it is an extremely stiff, light building. My reference point is some experience and a simple empirical formulas as given here. Although simplistic the formula do match up with most structures quite well. So I'd recheck your result. Not that it really matters the first frequency of the entire building is likely closer to 3hz so you are well under.

 

[marwin23]

human909 - I will try again to stiffen the floor, but somehow have got no success by now. Nevertheless will check again tomorrow.

On a site note, the article which you referred to. I have some minor academic experience (lat say less than 10 articles published), and would not argue with science. But finding a great formula which would calculate period/frequency of the same height residential building, and industrial structure, or any other then (let say steel towers) does not speak to me well. So the mill is 40ft, roughly 12.25m. Let say that period is 0.5s, so frequency is 2HZ. For the same sake I checked 40ft residential building I designed some time ago (did not do any modal analysis previously) - the result is 0.5HZ. Quite huge difference which is a result of many factors (building has columns spaced roughly 20ft apart, whereas mill has less than 10ft (etc).

I value your practical approach (your experience) a lot, and will try again to stiffen floor tomorrow morning.

 

[human909]

marwin23 formulas like that are used structural codes as a crude way of working out the period. They are sufficiently accurate for most building types to be codified, it is used in the AS code for example. I only presented it because it is the only tool that I have because I don't have any details of your building apart from an estimation of the height. I have doubts that that your mill has a fundamental frequency of 8.88hz. 2-3hz is more likely. I will trust good dynamic analysis over a rough formula any day. But I have my doubts about your Robot dynamic analysis number.

Eg. What are those restrained nodes doing at the far right of your structure on the 2nd, 3rd and 4th floors. If they are restraints then that is likely the reason for your high value for the whole structure.

Not that this matters much because as I've emphasised what matters is the natural response of the localised floor not the whole structure.

 

[r13]

I would at least run a model consists of the concrete floor and the steel beams directly under the equipment. IMO, you should check and minimize the effect on the immediate vicinity instead of the entire building.

 

[human909]

I would at least run a model consists of the concrete floor and the steel beams directly under the equipment. IMO, you should check and minimize the effect on the immediate vicinity instead of the entire building.

Agreed.

 

[r13]

Thanks. Your explanation has been the most helpful.

 

[marwin23]

r13 & human909: I can hear you. will try to get through that tomorrow. Thank you gentlemen - with each of your post I am mowing forward.

 

[marwin23]

Gentlemen,

I went through all your suggestions, and although still do not have final results there are some noteworthy facts:
- I was able to receive the same frequencies for both floor and structure in other software. Roughly speaking numbers are/were ok,
- I changed approach. It is enough to decouple right side of the structure from concrete walls, and natural frequency for whole structure drops to something between 1HZ to 2HZ, and floor frequency is (if I remember correctly) slightly above 7HZ. Probably if I modeled whole structure with concrete walls then my results were more accurate, but currently it is not something I am looking for
- right now will work just on ensuring that the right side can sufficiently cantilever/carry load, and will try to add some mass to reduce natural floor frequency slightly below 7HZ.

 

[human909]

Gentlemen,

I went through all your suggestions, and although still do not have final results there are some noteworthy facts:
- I was able to receive the same frequencies for both floor and structure in other software. Roughly speaking numbers are/were ok,
- I changed approach. It is enough to decouple right side of the structure from concrete walls, and natural frequency for whole structure drops to something between 1HZ to 2HZ, and floor frequency is (if I remember correctly) slightly above 7HZ. Probably if I modeled whole structure with concrete walls then my results were more accurate, but currently it is not something I am looking for
- right now will work just on ensuring that the right side can sufficiently cantilever/carry load, and will try to add some mass to reduce natural floor frequency slightly below 7HZ.

Thanks for the update. I know I'd like to hear more about this down the track, but the real answer won't come until you switch the machines on....

Speaking of which. The building is already built. Measuring the ACTUAL natural frequency of the building and the floors is going to get you more accurate answers than models.

https://www.tandfonline.com/doi/pdf/10.3130/jaabe.15.295

 

[ldeem]

I have some experience here and right now I am dealing with a dynamics issue requiring a very in-depth analysis.

I will bullet point items for you to consider based on what I read through the thread. This is in no particular order.

1. Have you compared current machine locations and operating parameters to the new machines? Compare both frequency and dynamic force. If they are the same then maybe that is sufficient.

2. Even for same frequency an increase in the load can cause problems. I have a situation now where equipment was replaced and new machine of similar frequency but dynamic loads 4 to 5 times higher. Structure is fine but equipment is breaking and vendor is saying velocities measured in the structure are too high.

3. You need to look at overall structure and local supporting beams (it sounds like you are doing this). Make sure you do your analysis such that sum of mass participation is reasonable. Consider lateral directions even if not specified by the manufacturer.

4. Regardless of what anyone says right now if the equipment is installed and there is a problem you will probably get dragged into it. Make sure you document meeting the equipment vendors acceptance criteria. If the equipment has problems they will eventually start looking at the structure.

5. Give consideration to foundation as soil springs instead of just pinned connections.

6. Your structural frequency are below the equipment frequency - which is typical in my type of work. To move the structural frequency your best option is probably to add mass and push the frequencies lower. Adding stiffness will require you to go past the equipment frequency which will be difficult in a 4 story structure. Changing the structural frequencies can be expensive; it may be a situation where you document the frequency response and let the client figure out if they want to take the risk of not meeting the vendors acceptance criteria.

7. A time history analysis would give you accelerations, velocities and displacements at the equipment. The vendor probably has no idea what the acceptable criteria is for their machine but you can look to various standards (ISO 20816-1 is one) and various literature on the subject to determine an acceptance criteria. Again document, document document and inform the client and vendor (assuming they don't give a criteria for A, V Displacement) what you came up with and if you meet the requirement. The caution here no mater how careful your analysis model is built field measurements will not match up very well. I am working on a similar problem right now and my field measurements are up to 1.5 times off calculation velocities. This is conservative because field measured velocities are very rough so typically reported as RMS where your program probably reports a very nice peak velocity.

 

[r13]

I think you'll be interested in this article. Link

 

[saplanti]

Ideem has summirised the process well. Especially the ıtem 6 is very important. They need to know the problem and contribute the solution.

If you do not find a proper solution there is one more option;

The original mill was on the ground floor and you client wanted to put it on the the floor above.
You may be able to provide independant structure for the new mill from the ground level by providing sufficient opening on the concrete floor above (this may require the floor structure modification and perhaps a static analysis if there is no other vibrating equipment on it). So they are to be two independant structures. You design the mill structure only for the given static and dynamic loading.

Just keep in mind this is the last option, perhaps the best.

 

[marwin23]

@saplanti - thank you. This option was already discussed in our office, and we see such resistance for even providing dampers, that we know that isolating equipment on a separate structure would not work.
@Ideem - thank you very much for your post.

 

[marwin23]

Gentlemen,

would appreciate you getting involved in a discussion on adding mass and a practical approach.

When analyzing a single floor, calculation shows that if I add 35psf all over (approx. 3" of concrete) to the existing floor, I will successfully reduce natural frequency slightly below 7HZ. I can reinforce steel below to cope with the weight added.
But:
1. Would you keep those two concrete layers separated (existing slab + add 3") - so there is no bond between both, and so presumably it is safe to assume that the top layer adds barely to stiffness of whole system? Any comment would be appreciated.
2. Or: In case I use bonding agent, maybe even some pins etc, and trying to achieve bonding between existing slab and new 3" layer, then I am running around the circle: added mass to reduce frequency, but at once added stiffness (+additional reinforcement to beams adds stiffness) in the way that I receive almost identical frequency (at least not something what would be a deal breaker for us).
3. Or does it make sense to add mass in strategical locations (first would have to identify them) instead of spreading it uniformly?

Any comments, thought and experience is appreciated.

 

[wrxsti]

its not clear on your model but have you included the load of the machinery in the natural frequency calculation

 

[marwin23]

@wrxsti - yes, it is added. It is impossible to see from the model screens I posted.

 

[wrxsti]

also to answer your question about the slab and adding stiffness

this thread has a nice real world example with wood XD

https://www.eng-tips.com/viewthread.cfm?qid=379229