Explain ASD vs LRFD to a dumb ME 20

[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.

 

[Lion06]

The traditional difference is using service load stresses against an Allowable Stress for ASD or using factored load forces against a maximum Strength for LRFD.
ASD has changed from Allowable Stress Design to Allowable Strength Design. There used to be different equations for ASD and LRFD (they were completely different specs), but now there is one equation for design strengths that are divided by a safety factor, omega, for ASD and multiplied by a strength reduction factor, phi, for LRFD.
The only real difference now is that ASD provides a constant factor of safety for all designs regardless of load types while LRFD provides a higher factor of safety on the loads that are less well defined (1.6 LL factor compared to a 1.2 DL factor). At a LL/DL ratio < 3, ASD is more conservative, at a LL/DL ration = 3, it is a wash, and at a LL/DL ratio > 3 LRFD is more conservative.
Any additional questions, please ask.

 

[JAE]

It really boils down to how you apply a safety factor to the design of members.

OK - my stupid analogy:

You have a small stream and need a log to drop over the stream so your ballet cow can dance across it. The log needs to be strong enough not to break under the weight of your cow and it needs to be stiff enough so the log won't sag too much under her weight.

Cow weighs 500 lbs. Log spans 15 feet.

[blue]ASD[/blue]
You know that the wood of your log can be stressed to 1000 psi before it breaks. So you need to find a log large enough so that the wood is not stressed to 1000 psi. In fact, you want to limit the stress with some safety factor to make sure that you are OK.

So you find a log large enough such that the stress is limited to only 600 psi. (i.e. your maximum [red]allowable[/red] stress is 0.6 x 1000 = 600 psi.)

So you have a safety factor against failure of 1/0.6 = 1.67.

The equation looks like this:

f_b = Stress from weight of cow
R = Strength of log
0.6 = safety factor

0.6 x R > f_b

[blue]LRFD[/blue]

You still know that the log can be stressed to 1000 psi before it breaks. You also know that the weight of the cow used to find the stress in the log might be plus or minus a few pounds. So because of this uncertainty you apply a safety factor to the weight of the cow -say 1.6. Also, you know that the 1000 psi limit varies a bit with different logs so you apply a little safety factor to it - say reduce the 1000 psi to 900 psi just to be safe. (i.e. 0.9 factor)

So you find a log big enough so that it will theoretically break when a 1600 lb cow is placed on it and you only count on 90% of its strength.

So the equation looks like this:

f_b = Stress from weight of cow
R = Strength of log
0.9 = resistance factor
1.6 = load factor

0.9 x R > 1.6 x f_b

What LRFD gives you that ASD doesn't is the ability to use different load factors (the 1.6) depending on the variability of the load. Dead loads you usually know pretty well. Live loads vary all over the place.

Also, the resistance factor can be varied depending on the variability of the strength estimates that exist with different materials and different failure modes. Abrupt, highly variable failures need lower resistance factors
while flexible, ductile failure modes can have higher factors.

I've used stress in my LRFD example, but most LRFD specifications use moment, shear, or force variables instead of stress.

LRFD allows proper, balanced statistical probabilities of failure while ASD has a single safety factor independent of how well you know your cow.

 

[frv]

It really boils down more to a statistical probability of failure.

In ASD you treat dead and live loads equally. For example, if you have a case where your dead to live load is 1 to 1 (say 200 kips dead and 200 kips live) the safety factor would be exactly the same as if your dead to live ratio was 1 to 3 (say 100 kips dead and 300 kips live) given the same total load.

LRFD recognizes the inherent unpredictability of loads and assigns a much higher "factor of safety" to live loads (we increase them by 1.6), whereas it recognizes that dead loads are most likely much closer to what you calculate (we only multiply dead loads by 1.2).

By the same token, LRFD also recognizes the uncertainty of different failure modes. For example, flexural capacity of a concrete beam is fairly predictable; therefore we count on 90% of the theoretical value. Shear in concrete, on the other hand is much less predictable; therefore, we only count on 70% of the value. I believe ASD also recognizes this to some extent, by applying different factors of safety to different failure modes, but I only really worked ASD for a few months, so I don't remember.

When all is said and done, you design so that the decreased member capacity with which you rely exceeds the capacity required by the factored loads.

In ASD you take the total loads that you expect on the structure and apply a single factor of safety to the members (say 2) regardless of the nature of the load.

It's worth mentioning that in the steel code (AISC), the publication of the 13th edition of the code attempts to unify the two. The way they did this was to calibrate load factors and resistance factors so that ASD and LRFD would yield the same factor of safety at a certain dead to live load ratio (don't quote me on this, but I believe it's 1 to 2).

My apologies for the length of this post.

 

[frv]

As JAE mentioned above, another reason that certain failure modes are "penalized" more harshly than others is because we want failures to be ductile. A beam that's failing due to excessive flexural stress tends to give a lot of warning (you see walls cracking and doors not closing, etc.)

A shear failure on the other hand is brittle. Unless you actually see the beam and see some shear cracks forming, you really wouldn't be able to see anything going on in your structure before it fails (assuming it was over loaded). As a result, you go the extra mile to avoid these failures and you do so by reducing their theoretical capacity by a higher amount than you do for ductile modes of failure.

 

[dik]

A significant difference is that you can look at a structure as a complete item and apply a more uniform factor of safety (not to be confused with a safety factor) to the overall structure. This you can do with LRFD. LRFD gives a better appreciation of the manner of failure and loading.

Dik

 

[abusementpark]

I hate how they cram LRFD down your throat in school. ASD is easier to use and generally more conservative.

 

[frv]

abusemantpark-

They "cram" LRFD down your throat because all the major codes are moving this way. As dik mentioned, it provides a more uniform (and consequently reliable) factor of safety.

As for more conservative, that entirely depends on the dead to live ratio. As I mentioned in my previous post, ASD and LRFD were calibrated at a specific dead to live ratio (I believe 1 to 2). At this ratio the true factor of safety is the same. If your dead to live ratio is lower than 1 to 2 (that is, more dead load) then yes, ASD is more conservative. However, at anything above this ratio, LRFD is more conservative (as it places more uncertainty on live load).

I'm not knocking ASD- I know it's been used for many years quite successfully. But LRFD is not entirely without merit.

 

[frv]

I hadn't read Structural EIT's post- I'll take his word for it and say the ratio is 3 to 1.

 

[271828]

frv, StrlEIT is correct about the breakeven being at 3. I think this is in Part 2 of the Manual.

I like to think of it as large L/D --> use ASD because it's more economical. Then again, I'm a metal bldg systems guy at heart, LOL.

 

[msquared48]

JAE:

I take it that the .9 resistance factor allows for the milked cow condition?

Mike McCann
McCann Engineering

 

[AlmostPE]

No.. the .9 is for the rotten wood. The 1.6 is for the miled cow Define the milked cow.. the milk has been milked out or is it full of milk? The 1.6 is just in case they have a pregnant cow.

Not almost anymore!

 

[dik]

We generally refer to the 0.9 as the material property factor; it varies from a low of about 0.65.

It depends on the type of material and the type of force and is lower for materials and the type of forces that have a greater 'scatter' of test results. It's to make up for uncertainties of how things 'behave'.

In 1965, we were the first engineering class at the U of Manitoba to deal exclusively with 'Limit States' design, the precursor to LRFD and consequentially 'Working Stress' design (ASD) was only touched on for historic reasons. So, it's been around for a while (40 years).

Dik

 

[frv]

The 0.9 is the "R" in LRFD (Resistance Factor); the 1.6 is "L" (Load Factor)

 

[JAE]

Yes, the 0.9 ( [&Phi;] factor is to account for the following variations in the log:

Varation in:
Depth of log
Width of log
Strength of the wood
Whether there were termites in the log
The span length
The accuracy of the equations that tell us what the log will support.

The 1.6 load factor accounts for variations in:
The cows weight - I guess it depends on milked/not-milked and pregnant/not-pregnant but let's not forget the steroid question.

 

[hokie66]

Is terminology being fiddled with again? Dik, you said the phi factor is "material property factor", but it has been "capacity reduction factor" to me since 1963.

 

[frv]

hokie66-

I believe it actually stands for "resistance factor" as in Load and Resistance Factor Design.

 

[JAE]

I think dik is in Canada and the term used up there may be different than the AISC term "resistance" factor or capacity reduction factor.

The term "material property" factor is a bit misleading because it not only responds to variability in material properties but also geometric property variations (i.e. variability in depth, width, location of rebar, diameter of bars, thickness of flanges).

 

[hokie66]

Oh well, I am in Australia, and now that I think of it, the term here is supposed to be "strength reduction factor" now, but most of us old hands still call it "capacity reduction factor".

 

[JAE]

a rose by any other name...