Load factor for fluid loads seems too high for some structures 4

[Mawaca]

Regarding the load factor for fluid loads

Ref. ACI 318-11 9.2.4 and 9.2.1 Eq. (9-1)

U = 1.4F

This factor seems too high for deeply submerged structures or for retaining structures whose walls are too short to keep out 40% higher water.

For example, if I am designing a concrete immersed tube tunnel with 100 feet of water above it (plus something for high tide, wave crests, and sea level rise), this factor implies that I am planning for a case where the water level increases by 40 feet (or that undefined dynamic effects, 100 feet underwater, are equivalent to 40 feet of additional head).

Does anyone know of phenomena that justify 1.4 or a code, standard, or reference that allows a lower factor?

 

[CELinOttawa]

The factor is not only to allow for more water, although your load can go up 40% with much less than a 40% increase in depth.... The force on your wall is related to height (or depth of water) squared. Triangular distribution, as well your line of action for your resultant force changes (goes higher) with any accidental increase in water depth.

I believe the codes are also concerned with salt content and other "pollutants" that can change the density of the water.

I think 1.4 is a reasonable factor.

 

[Mawaca]

Thank you for responding, CELinOttawa.

We typically account for the the salt content explicitly when we (or the client or the governing body) choose the specific weight of water for the project. In any case, overlooking that would account for no more than a 3% increase in pressures, so it doesn't alter the question.

Your first point will alter the question for cantilevered retaining walls, but not for roofs, floors, or walls/ panels spanning horizontally, because pressure is linear with depth. So on the roof or floor of an immersed tube or (approximately) on a panel of a subdivided wall at a given height, the pressure is uniform and the additional 40% does indeed correspond to 40% deeper water (or unspecified dynamic effects).

 

[CELinOttawa]

You're right, my argument applies to walls, and in particular would be key for retaining walls...

But the possibility of contamination can go higher than 3%. Numerous factors can cause spikes into the teens; At that point, what factor do you want? 1.25 would be for known dead loads!

It is my opinion that 1.4 is rational. To answer your first question, I know of no code which uses a lower value. Some codes use 1.5 as per any live load.

 

[JedClampett]

I don't know if either of you have ever done any water bearing or water retaining structures. I've done many of them. If you use a 1.4 factor for a water bearing or water retaining structure, you're very likely to have a failure. Not an overstress, a failure. As in people dying.
The load factors are a complicated subject that account for statistical uncertainties, construction issues and analytical errors. They're not there due to salt in the water or a wave. If you have those, you use them in the loading.
We've learned over the years that building codes are not really equipped to accommodate fluid loads. They're not like furniture, merchandise or human occupancy. Every square inch of the structure has exactly the same load on it. It's always there. And it's immense. The question is not whether a 1.4 load factor is adequate. It's not. It's whether to use a 1.7 factor or a 2.21 (ACI 350) factor.

 

[Mawaca]

Thanks, JedClampett,

When we design hydraulic structures--such as concrete locks and flood protection structures--for the Army Corps of Engineers, per engineering manual EM 1110-2-2104, we end up with the 2.21 factor because the load factor is increased from 1.4 to 1.7 and the 1.7 is multiplied by a hydraulic factor of Hf = 1.3. The Hf is intended to eliminate an explicit check of cracking, deflections, and durability.

On projects outside the Corps's jurisdiction, cracking, deflections, and durability are evaluated (For an immersed tube, cracking rather than strength often controls local, flexural design.), so there is no Hf. Also, the EM itself would probably not be used to set load factors.

The projects where I question the load factor are
a. (non-Corps) immersed tubes at the bottom of natural bodies of salt- or freshwater or
b. the outer walls of cofferdams or floating structures that could not possibly hold back a 40, 70, or 121% higher stillwater level because their walls would be overtopped and they would flood or sink (or they would float and rise with the water level).

CELinOttawa,
These are projects where water hammer, surge (at valves), wave slam, hurricane/ storm surge, tsunamis, and fluids nearly as dense as uncured, lightweight concrete are not present or are explicitly assessed.

Keep in mind that the 1.4 factor was not derived for tunnels, cofferdams, or floating structures; it is based on the variation of loads on buildings, which I believe to be quite different.

(The Corps's 1.7 value on fluids was selected to scale new LRFD designs to make them as conservative as the ASD designs that had performed satisfactorily in the past. The value was selected for--presumably permanent, presumably surface--hydraulic structures, so there was a recognition that permanent, hydraulic structures penetrating the waterplane have fluid loads which are more variable than fluid loads on buildings.)

Before I (or preferable somebody smarter than I) could propose a load factor for situation a. above, marine engineers and historical data would be involved, since the L.F. would pertain to the variability of conditions in bays or straits.

For situation b., we'd need a better handle on what variations are accounted for in the +40 or +70%. New code language would relate the load factor to the impossibility of high water levels in self-limiting scenarios.

 

[CELinOttawa]

Are we discussing LRFD factors, part factors, or ASD? Jed's got me somewhat confused... I had understood this as a discussion of the live load factor appropriate for water based liquids (anything else is, by necessity, quite different).

 

[EngineeringEric]

If we are talking LRFD then the factor is not something that can be modified with engineering judgment. I would say that would be equivalent to removing your safety factor of 1.67 for an ASD approach. Technically you should always be below the actual stress, but sometimes it will go over and that is when the Load Factor or Omega(ASD) come into play... and that is assuming it is constructed perfect!

I guess 1.4D is equivalent to 1.4U...? and that is something not many people would argue to reduce ever.

 

[wadavis]

"4.1.3.2 The principal-load factor 1.5 for live load, L, in Table 4.1.3.2. may be reduced to 1.25 for liquids in tanks." - NBCC - 2005

The National Building Code of Canada allows for a live load reduction for a narrow category is where the fluid level is both limited and the the content is relatively predictable. No input on your specific design conditions, just wanted to point out the code that allows for a reduced fluid live load factor.

wadavis
E.I.T.

 

[JedClampett]

Mawaca, let me be clear. The designs I'm talking about are exactly like the ones you describe. We are designing tanks with well defined maximum levels. There's a wall or some other passive overflow controlling the level. And we still use 2.21 load factor. The 1.7 factor is not there because the levels fluctuate. Combined with the strength reduction factors, it's there to meet some industry agreed on factor of safety. As I said before, you can't compare liquid loads to other load types.
Take a look at the two threads I linked to below. The jury is still out on who's fault they were. But I bet the designers, their bosses, the owners of these plants and especially the families of the guys killed wished they would of used a higher load factor.
If Canada is using 1.25 for liquids in tanks, they're playing with fire. But let's see if the first couple of wall collapses will change their mind.

thread161-296099
thread507-299397

 

[CELinOttawa]

Comparing NBCC to the US codes is like comparing apples to starfruit... It is touch and go because the basic assumptions behind the codes differ. What factors get applied and for what reason differ. The requirements have similar if not identical goals (safety and serviceability), but get there is very different ways.

If we are talking about a single load factor (like many US codes), 1.25 or 1.4 is likely insufficient. If we are talking about limit states design (as in Canadian code, and to some extent LRFD codes) this is not the same animal.

The 1.25 factor for liquids in storage tanks is not new, and dates from prior to NBCC 1990 (oldest copy I have on hand).

Nothing's on fire up here, and I assure you that the factors in the code are appropriate as well as fit for use.

 

[Mawaca]

JedClampett,

Thank you for the links. It appears that the TN and NY tanks failed at wall intersections.

TN OSHA blamed poor construction rather than design for the TN failure. The contractor seems to have created an un-roughened cold joint where no joint was shown on the drawings.

http://www.knoxnews.com/news/2011/oct/27/deficient-wall-construction-cited-in-gatlinburg/

The NY failure appears to be a design failure with one or more design flaws, perhaps including poor detailing, lack of accounting for the reduction of shear capacity as a result of concurrent tension, designing for the specific weight of water rather than for the slurry and water that the tank was intended to contain, and insufficient rebar development. I recognize that the load factors account for some errors and omissions on the design side, but I didn't see any indication that either failure was driven by use of a load factor less than what was required in the code. If construction and design hadn't been less than standard practice, 1.4F (or 1.7 or 2.21 if that is required by ACI 350…I am away from my office for an extended period and don't have my usual references.) the tanks should have been alright.

It is an important difference that tubes and cofferdams keep water out while tanks keep water in, so the tanks have tension in their perimeters, which reduces shear capacity and makes them more likely to fail at wall intersections. Whereas, tubes and concrete cofferdam-like structures don't have any sections with net tension (I'm not talking about propped walls, of course). In tubes and cofferdams, the ever-present fluid load actually helps. Compression increases shear strength at intersections and makes joint failures less of a problem. That said, whether the structure keeps water in or out, should affect the load factor.

(For tubes, flexural design tends to be driven by serviceability, via limits on crack width or rebar stress, so the discussion of a reduced L.F. is important for shear but not for flexure. For Army Corps structures, this is messy, since there is no explicit serviceability check when the Hf that I described above is used; there, the L.F. does affect flexural design, with U = Hf x L.F. x Load = 1.3 x 1.7 Load = 2.21 x Load. In order to compare apples to apples, I should ask: Is it the case that you don't need to do crack width {or rebar service stress} checks when you are designing tanks for 2.21xLoad?)

To return to the idea of a lower load factor on tubes, cofferdam-like structures, and tanks without roofs; it's important that fluid loads are capped because they are limited by overtopping or by the rareness of a bay becoming 40% deeper. I agree that load factors for buildings and for bridges in the U.S. have been adopted by experts in each of those fields--with differences to account for differences in the how they are loaded. My argument is that structures dominated by fluid loads might be designed with too much waste if they use the same factors for ultimate strength design. Related to that, my next question is that, if the 1.7 or 1.4 (or 2.21) factor is not there mainly for excessive water level, then what are the main phenomena for which it accounts?

 

[CELinOttawa]

ALL load factors are present due to the uncertainty of the load, without exception. Material reduction (aka uncertainty) factors are present to deal with the variations in material strengths and reliability. Anyone who argues these two points simply doesn't understand Limit States Design and fell off the bandwagon of evolving Structural Engineering technology sometime in the late 1950s.

 

[CELinOttawa]

Oh, and something that I don't think I've made clear: The 1.25 live load factor for TANKS is specifically to apply to tanks designed for their overflow capacity.... It doesn't apply to this thread at all. We have been (see OP) discussing the LSD LLF for submerged structures and retaining structures.

As per Jed's concerns and Mawaca's excellent breakdown of the specifics, retaining structures are far more subject to unknowns in design than tanks.

 

[JStephen]

Note that in a typical water tank filled to the overflow, the possible variation in load is just about zero, so if a 1.25 factor is applied, there is something considerably more than load variation going into it.

 

[CELinOttawa]

JStephen: I get that this isn't your field, but you're wrong. Nothing but the load variation goes into a live load factor. Since the 1.25 factor is from the NBCC, a modern Limit States Design (LSD) code, this can only account for the variability (aka uncertainty) of the actual imposed load.

Importance factors, live load, live load reduction, and material reduction factors (etc) are all broken apart to be studies and applied individually.

 

[JStephen]

Could be, CEL. But that comes back to definitions to some extent. If the applied liquid load can vary +/- 25%, I would question if it is in fact a "well defined density and controllable maximum height".

I did notice a line in the commentary in ASCE 7-10 that says "However, statistical data on these loads are limited or non-existent, and the same procedures used to obtain load factors and load combinations in Section 2.3.2 cannot be applied at the present time. Accordingly, load factors for fluid load F...have been chosen to yield designs that would be similar to those obtained with existing specifications if appropriate adjustments" etc.

In the next paragraph in the commentary, it defines F as "structural actions in" the structure rather than loads applied to the structure. In ACI 318-05, the commentary says "The factor assigned to each load is influenced by the degree of accuracy to which the load effect usually can be calculated and the variation that might be expected in the load during the lifetime of the structure...Load factors also account for variability in the structural analysis used to compute moments and shears." IE, it seems more is included in the factor than just variations in depth or specific gravity would indicate.

 

[CELinOttawa]

No minus. The load factor is to allow for unknowns within a reasonable margin. In this case +25%, which is quite low; It is rare in the extreme to have a limit state design live load factor under 1.25 or 1.2 (1.2 beging the dead load factor in some jurisdictions), unless the load is helping your design case in some way (which results in requirements of factors under 1.0).

ASCE 7 is *not* a Limit State Design code, though I understand the code has finally included LSD to some extent with seismic and wind. It is an Allowable Stress Design code which has been re-written to look like a LSD code. It is best termed a split-factors design code, and anything the commentary says is not pertinent to a discussion on LSD.

Now, the ACI 318 code IS an LSD code, the only one out of the US to my knowledge, and you entirely correct to state that the accuracy of the analysis enters into the creation of factors. All factors, including reduction, live load, dead load, etc, with the singular exception of importance factors. However, I really don't think it is a useful point to make, since this is akin to saying that engineering includes dealing with inaccuracy. The fact remains that the value of a live load factor is set based on the uncertainty of the value of a load in a given situation. This what accounts for 1.2 versus 1.25 versus 1.4 versus 1.5 (etc).

 

[JedClampett]

I keep getting sucked into this vortex of load factors.
I would contend that the load factors currently in use are a wonderful stew of uncertainty, politics, and technical input combined with historical custom. For example, you almost never get a 100 psf live load over an entire floor. If it was people, it would be equivalent to a 200 lb. person over every 17 inch by 17 inch square on a floor. Yet we continue to multiply it by 1.6 (or 1.7). If you ever really had a 100 psf live load over an entire area, things would get scary. There would be deflection, shakiness, some odd noises and a general discomfort. The 1.6 factor has also taken into account the unlikelihood of the 100 psf load. In this factor are also material uncertainties, contractor inaccuracies, analytical problems and plain old mistakes by all involved. And note that sometimes these mistakes increase the factor of safety.
With liquids, one of the most important sources of load uncertainties is not available. The loads are not conservative in themselves. The load is not only there, but present over the entire footprint. And not just occasionally, but every day. To get an acceptable factor of safety vs. real life conditions (see those failures in the threads above), I need to multiply my liquid loads by 2.21. If there's a code backup and you're comfortable with it, multiply by 1.25 or 1.4 or whatever.

 

[Mawaca]

Noticed a typo. in my last post.
It should say "whether the structure keeps water in or out, should NOT affect the load factor."