Earthquake loading and dead load factors (AS 1170 vs AS 5100) 1

[bugbus]

This is a topic of confusion in my office because of the way the two codes are written. Hopefully someone can enlighten me...

AS 1170.0
In AS 1170.0, the earthquake load combination is:
1.0 x EQ + 1.0 x G + ψ_E x Q

To my mind, the above equation makes complete sense. The EQ force is based on the unfactored mass of the structure (allowing for some permanent component of live load).
The 'load factor' for the EQ loading is basically built into probability factor k_p in AS 1170.4, and is calculated based on the unfactored mass.
The important thing to note in my opinion is that this 'load factor' scales up the magnitude of the EQ acceleration, but not the mass of the structure.

It then makes sense that the dead load factor should also be 1.0 because we are dealing with acceleration effects on the nominal (unfactored) mass of the structure. It would seem wrong to consider the unfactored mass of the structure for EQ effects, but then to artificially scale the mass up or down for gravity effects.

But --- this is exactly what AS 5100.2 seems to suggest.

AS 5100.2
Let's take a mostly concrete bridge (for simplicity), so the EQ load combination according to AS 5100.2 is:
1.0 x EQ + [0.85 or 1.2] x Concrete Dead Load

At first glance, this would suggest that we have to calculate the EQ force based on the nominal mass of the structure (hence the 1.0 factor), and then scale up or down the gravity effects (by a factor of 0.85 or 1.2). This approach has the obvious problem that only one of the mass-derived loads is scaled while the other remains unchanged, which is inconsistent. Unlike the probability factor k_p, which is effectively a ULS factor applied to the EQ acceleration, the 0.85 or 1.2 dead load factors instead scale up the mass of the structure (not the gravitational acceleration).

Or - is it saying that we should first scale the mass of the structure (by a 0.85 or 1.2 factor, say), and then apply both the EQ and gravity effects on this scaled mass? This approach is more consistent in that both of the mass-derived loads are scaled by the same amount, but we are now effectively applying a second load factor onto the EQ effect (i.e. on top of the probability factor k_p in AS 1170.4).


My personal opinion is that AS 5100.2 is simply wrong and that this particular load combination has been overlooked. There are plenty of other examples throughout the AS 5100 series where obvious errors have been overlooked and unchanged for several years without revision. The AS 1170.0 approach, in my mind, is consistent and logical.

 

[Just Some Nerd]

I don't think the intent is to scale gravity effects up and down without scaling the EQ effect. EQ load should really always match the vertical mass being included in the load combination imo


----------------------------------------------------------------------

Why yes, I do in fact have no idea what I'm talking about

 

[bugbus]

Just Some Nerd, I agree with you 100% that the EQ load should be based on the vertical mass being included in the load combination.

Are you saying that your approach would be to scale both (i.e. effectively scale the mass and apply both gravity & EQ effects to the scaled mass) or to scale neither (load factors of 1.0, a.k.a. the AS 1170 method)?

 

[Just Some Nerd]

I'd do my design to the case of 0.85G + the earthquake demand for a mass source of 0.85G, then also check 1.2G + the earthquake demand for a mass source of 1.2G.

5100 load combos tend to deviate very heavily from 1170, which is why I wouldn't stick to the "good enough" combination of G+0.3Q+EQ we use for normal structures. I imagine they put the different dead load combination factors for the earthquake for a reason. There's also a very notable absence of live loads in a lot of the ULS combinations, not particularly sure why, but I'd also personally like to capture some of that missing load in the 1.2G case.

----------------------------------------------------------------------

Why yes, I do in fact have no idea what I'm talking about

 

[IDS]

The ULS factor on dead loads is set at a level such that at any given section the design actions are not less than the actual actions at the section, with an acceptable level of confidence. It therefore includes allowances for many variations other than the actual mass of the structure, including:
- Inaccuracies in the analysis.
- Differences in load distribution.
- Differences in flexural stiffness to that assumed in the design.
- Load redistribution due to differential settlement or differential creep and shrinkage.

It is therefore quite possible, for example, that the actual axial force due to dead load at a column will be significantly lower than the design value, without a significant variation in the actual mass of the structure.

Applying the standard dead load ULS factors in conjunction with seismic loads is therefore conservative and gives a better estimate of the worst-case load conditions.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[bugbus]

IDS, thanks - that's a great explanation, I hadn't thought of that earlier.

So possibly it's the case the AS 5100.2 deliberately left in the dead load factors for the EQ load combination. I could see that for the design of bearings, for example, this would provide an extra margin of safety against bridges becoming dislodged from their supports or some other catastrophic type failure where vertical loading is important.

 

[Agent666]

I definitely would not be doing what Just some nerd is proposing. That is not the intent of most codes. NZS1170.5 does not implement this, and many other international standards do not implement an approach like that. In some international standards there is definitely a load factor on 'E', that is how it is handled typically in my experience if a different level of seismic load is to be applied in combination with a specific gravity combination. It is not by factoring the seismic mass and determining multiple earthquake loads based on a variable mass situation.

The seismic mass is worked out based on the actual mass, or at least a realistic estimate allowed for in codes (i.e. G and some proportion of the live load). This determines the lateral loads, this is a fixed value.

The axial loads present at the same time as these seismic loads can vary on an individual element (i.e. as per IDS's explanation), it is not always going to be the average determined based on the overall seismic mass.

It must also be remembered the load combinations are not 'plus' they are 'and', this means you as the designer must determine the worst design action effect. If for example a member is in tension/uplift and the live load part of any load combination were to act to reduce that tension design action effect, then you leave it out and only apply the permanent gravity load portion of the load combination when assessing the critical load case vs strength. You can of course also evaluate the case with the live load, but it will obviously not be the critical case.

https://engineervsheep.com

 

[rapt]

The combinations are similar to a lot of other variable load ultimate strength combinations,in that if a variable load increases the overall moment then the DL has a factor > 1 and if it reduces it, then DL has a factor <1.

If the Eq load is same sign as the DL then a higher DL factor is applied so 1.2DL + Eq

If the Eq Load is the opposite sign to the DL, then a reducing DL factor is applied so .85DL + Eq.

This has nothing to do with calculating Eq.

 

[Just Some Nerd]

Well it turns out it's a very good thing I don't work on any bridges cause clearly I didn't figure this one out Definitely exposed my lack of experience, I think. The explanation from IDS seems pretty straight forward tbh so a bit ashamed to have not known that. My understanding of how 1170 arrives at its load combinations probably hasn't progressed past the education I got at uni, which is to say my understanding hardly exists, evidently.


----------------------------------------------------------------------

Why yes, I do in fact have no idea what I'm talking about

 

[bugbus]

Agent666 & rapt, thanks for your input. I'm fairly well convinced now that there's a good reason to include the 0.85 and 1.2 dead load factors with the nominal EQ load.

Even for a relatively simple single-span bridge, especially where there is a skew, there is some uncertainty as to how the superstructure dead load is distributed to each of the bearings and piles, for example (depending on factors such as torsional stiffness of girders, deck stiffness, settlement, etc.). 0.85~1.2 seems like a reasonable range of certainty.

Why, then (in your opinion), does AS 1170.0 not do the same?

 

[rapt]

Or, you might ask, AS3600 also!

I have asked both 1170and 3600 to clarify.

 

[bugbus]

Good to know, would love to hear what they say

 

[rapt]

AS1170.4 says they are correct and there is no need for a variable factor on DL for the earthquake combinations.

 

[IDS]

AS1170.4 says they are correct and there is no need for a variable factor on DL for the earthquake combinations.

It seems that AS 5100.2 begs to differ.

The next draft amendment to AS 5100 has recently been released for public comment, and has no change to the dead load factors to be used in conjunction with earthquake loads.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[rscassar]

If performing a dynamic analysis, don't forget to take into account the vertical actions from earthquake

thread744-490596

 

[rapt]

IDS, I realize that, and my thought was the same as yours.

But I asked the AS3600 Seismic Subcommittee which includes the Chairman of AS1170.4 and they said 1170.4 is correct.

ACI also uses .9 or 1.2 for D so similar to AS5100.

I will ask again!