Explain ASD vs LRFD to a dumb ME 19

[jdog1]

Can someone explain in really simple terms the difference between ASD and LRFD? ASD seems easier but is it going away?

I was reading through the new AISC Steel Construction Manual about the two methodologies and, aside from the different load combinations, was having a hard time seeing a clear distinction between the two.

 

[msquared48]

I thought the resistance factor was only applied to mules, not cows.

Mike McCann
McCann Engineering

 

[AlmostPE]

msquared.. I think you refering to importance factor. I think cows are more important than mules though.

Not almost anymore!

 

[VTPE]

There are a couple of important things to realize.
1.) The ASD Manual (the Green Book) has not been updated since 1989. The specificatyion part of it has not been updated since the 70s. All current research has been conducted using limit stares design.

2.) In ASD, the factor of safety is based on probability for (3) itesm.(A), the probaibility of errors in erection, (B) the probability of errors in manufacturing, and (C) the probability of errors in the "simplified" mathmatical equations we use. By the time the building is in service, these probablilities are have been accounted for and there is really no factor of safety any more.

3.) If you really read the Seismic manual, you have to use LRFD anyway.

4.) In reality, as stated above, ASD does not exist anymore. It is now strength design, which in essence is LRFD, just with the factors moved aroud to make the old method seem like it is still there.

Pretty soon we will be having the same conversation about concrete and the ACI's method of using compatibility instead of balanced steel. Those of you using 1.6 and 1.2 in Concrete and not performing a compatibility analysis to determine tesion / compression control are not doing your calculations correctly.

Have fun!

 

[dik]

The phi factor is typically referred to as a "strength reduction factor" in Canadian codes. I refer to it as a material reduction factor because I think of it as a factor that reduces the published strength due to uncertainties related to the material and type of loading. For example, wood, steel and concrete can have different phi factors for shear... or a different value for bolt shear as opposed to a steel section.

Dik

 

[JAE]

VTPE - I doin't believe the original source for the 0.6Fy factor in ASD really had much to do with probability. It developed over the years as a feel good level of safety, from the gut, that over time has been confirmed as a pretty good level.

Your statement number 2 implies that ASD is not as safe as LRFD, which is not true in simplistic terms. In fact, much of the statistical analysis done by Galambos and Bruce Ellingwood many times calibrates the load-factor/phi-factor level of safety to that of the traditional ASD safety factors.

With very high live load to dead load ratios, LRFD will result in heavier members than ASD and with low ratios the resulting designs are heavier with ASD.

ASD exists in the latest AISC manual and uses one [Ω] factor, replicating what the 9th edition of the AISC ASD manual used, but also incorporating the latest research.

ASD simply doesn't reflect at all the variations in loads and resistance factors and so provides a variable probability of failure while LRFD's probability of failure is much more constant with different loading conditions and with different materials.

I use LRFD all the time now (grew up on ASD).

 

[nutte]

VTPE:

The old ASD (allowable stress design) is the same as the new ASD (allowable strength design). The difference is it now has you check the strength (P, M, etc) as opposed to a stress (fb, fc, etc). That is the only difference to the approach, which really isn't a difference at all. The equations have changed because there has been much research since the last ASD specification.

I've studied all of the supposed benefits of LRFD over ASD, and I don't see it. For my money, I'm using ASD all day long. The frustrating part of LRFD is you still have to run the service load combinations to check serviceability of the structure. Why do two analyses when one will check both strength and serviceability? Maybe you can lighten up your members a little for certain cases with LRFD, but I don't design that close anyway, so where's the benefit?

 

[jdog1]

Thanks for all the replies everyone. The cow example was especially helpful.

 

[271828]

nutte, I have one proposed benefit. I'm inclined to think that Allowable Strength Design is only a temporary part of the Spec. and will vanish in a Spec. or two. The hottest research topics right now are all probabilistic and reliability-based. Researchers WANT the strength and loads separated to make future enhancements possible and/or easier.

I personally don't see the big deal in using LRFD, so might as well go that direction and get used to it. Just my $0.02

 

[frv]

271828-

You're absolutely right. The reason they included a combined specification was to try to coax everyone into LRFD. My steel professor even seemed a bit disappointed that they included ASD at all in the Specification.

 

[271828]

I'd guess that the metal bldg folks played a big part in keeping ASD around. With the L/D ratio, they'd be increasing sizes all over the place if they were forced to use LRFD.

Your professor might do some work for AISC. The inclusion of ASD and LRFD wouldn't make the Manual 2x as hard to create, but would definitely add a lot of time and effort. Remember that volunteers, for the most part, put that thing together.

I personally do not like having ASD in there simply because it muddies the water. Take a look at the hanger prying action stuff, for example. It's also VERY easy to mis-read a table and yank an ASD number when using LRFD.

 

[frv]

Yeah.

He's a task committee member.

 

[Robbiee]

Let us look at this example and see how the margin of safety could be deceiving and erratic with the ASD method:

You have a 4x12x0.25m concrete shear wall that resists a wind load applied on a 30x12 m sq. tributary area. You were told to design for a 75km/h wind speed, which applies about 0.35kPa pressure. For this wall you needed a tie down that is capable of resisting 48kN tension. You decided to apply a factor of safety of 2.0, therefore, your tie down is capable of resisting 96kN tension. Well, the building was hit by 85km/h wind (13% increase in speed), which applied 0.44 kPa pressure. The tension in your tie down is 96kN. It yielded. What happened to you safety factor of 2.0?

 

[nutte]

frv:

The reason they included a combined specification was to try to coax everyone into LRFD.

I disagree. The reason they went to LRFD back in 1986 or so, and the reason they only issued LRFD specifications with the silver and dark blue manuals, was to coax designers to use LRFD. Designers balked and wanted an ASD specification that reflected the latest research. The combined specification is to please the portions of the steel community that wanted a comparable ASD specification. In my mind, this is AISC saying "we give up, here's your ASD back."

271828:

The researchers want it, it's inevitable, so we may as well get used to it? What about the designers that use these specifications? Is their input worthless? I don't mean to belittle the researchers and the academics, but as designers, we should make our voice heard as well.

As I mentioned above, I don't get the factor of safety argument. We don't have structures failing all over the place. Why is there a push to change a system that's working fine, especially when the change is to make it more complicated for no good reason?

With computers, we can put in 100 load combinations as quickly as we can put in 50, so why not? How about keeping things simple? Does anybody do hand calculations anymore? Run one check ASD, determine strength and serviceability with the same loads, or have to do it twice with LRFD? How is this beneficial to anybody?

 

[frv]

nutte-

The 13th edition is a limit state design in either ASD or LRFD. They combined ASD and LRFD to showcase how "similar" the two could be and as a result ease the transition to LRFD for those designers who had been using ASD for many years.

AISC did capitulate a bit in a sense, since they really would rather not have included ASD, but they did so looking to eventually phase out ASD. The 13th edition can be considered a "transition" specification. It's the "rosetta stone" for ASD designers.

I guarantee you AISC is most definitely not "giving up".

 

[271828]

nutte, you make a lot of good points, and I tend to agree in general.

As a former designer, I sure hope their input is not worthless or ignored. A very sizeable portion of today's engineers learned LRFD first and won't give a rip if ASD goes away. It won't be long at all until the vast majority of designers will have learned on LRFD. Calculating and using wu to get Mu and then using w to get deflections is old hat to them.

I'm a researcher nowadays and there seems to be an inevitable march away from deterministic approaches toward reliability-based approaches. I personally HATE this because I despise probability and reliability as subject, but it seems completely inevitable at this point. Like it or not, researchers have fairly strong voices when it comes to Spec. writing, so these approaches will have a strong influence.

How is this stuff beneficial? To be honest, I cannot answer that for most situations. I think some of the worst technical authors have written most of the papers on this subject. I read papers and do research every day and my eyes glaze over when I read most of these.

For some types of checks, going probabilistic makes great sense. Take floor vibrations for example. The loading is almost totally random. Any criterion that claims "OK" or "NG" is hard to justify (other than not having anything else). Some of the newer methods will eventually lead to the ability to tell an owner that he has 2%, or whatever, probability of complaints.

 

[nutte]

271828:

I got my BS in 2001. I learned LRFD in school, but as a designer, I use almost strictly ASD. This is common for many of my peers that went through school about the same time, who never once studied ASD in school.

frv:

Sure, the unified specification is based on limit state design, just as the old ASD was. Further, the unification of the specification shows how the equations are the same for a given limit check. The changes between the 9th edition ASD and the 13th edition ASD are due to research that has taken place, not a change in the philosophy of any of the checks.

 

[frv]

nutte-

This argument can go on ad infinitum.

The equations are the same, but the the safety factor is not.

 

[271828]

nutte, that is not my experience. At the last two firms I worked for (retired from design, LOL, and went back to school 3 years ago after working 9 years), the younger engineers all learned LRFD in school and have never used ASD on any job.

I did work for one firm, however, in which 89 ASD was used. The younger guys there hated it and saw the firm as backward for not letting them use what they learned in school.

I do not claim to know which is more typical, my experience or your experience. As time goes on, however, I have to think more and more folks will come out of school thinking LRFD = Steel Design, so won't give a rip about ASD.

 

[DRC1]

LRFD confirms that the capacity of the beam is less than an inflated load that is never actually realized in the field.
ASD actual stress are computed for actual loads and held below a maximum allowable stress.
Which method gives you more control over your deign?
(Hint: With ASD you know stresses at every point in your cross section and can compute actual deflections)

 

[JAE]

The ASD method does NOT give you more control, nor does it offer you any sort of consistent, measurable probability of failure.

Cripes, the only difference is in the fact that LRFD weighs the safety factors on each load and the member strength to give you a much more rational level of safety.

I grew up on ASD - don't have anything against it. It is fine, easy, straightforward, etc. But it s*cks when it comes to measuring safety consistently.

DRC1, ASD's stress is "held below a maximum allowable stress" as you say. The problem with it is that the amount you hold it down is totally arbitrary - a historic one-shot safety factor that has no rationale behind it.

You say you know the stress at every point for "actual loads", but those "actual" loads are still just estimates of variable loads. Your actual stress depends on a live load that has been measured in the field as having incredibly high variability....so it's not anymore actual than the factored loads of LRFD.

I've learned both methods. Both are fine.

Yes you can argue that the extra value in knowing you have a more rational, consistent safety factor in LRFD is of questionable value in most structures. I won't argue that.

It's just that I can't believe how many engineers I've heard bashing LRFD simply because of a few extra steps - and then try to insinuate that there is some lack of rationale or intelligence in LRFD.

I've converted to LRFD (after using ASD more than 15 years) and it's not that big of a deal to use nor that cumbersome once you start using it.

Both methods have been set up to give similar results and both methods are "good engineering practice." But I can't agree with the implication that ASD is somehow superior or offers more control to LRFD.