Wood (1973) Seismic At Rest Pressures - Too Conservative? 2

[strucbells]

I'm a structural engineer who has recently gotten more involved in designing large underground structures in seismic areas. These often involve thick basement walls that are subject to at rest pressures for static soil loading provided to us by our geotechnical consultants.

I've noticed that our geotechnical consultants often rely on Wood's approach for determining the seismic earth pressures they provide in their report. I have read some recent literature (mostly from Prof. Sitar at Berkeley) that describes this approach as an extremely conservative overestimation of loading.

https://www.ce.berkeley.edu/sites/default/files/assets/users/sitar/Wagner and Sitar - ASCE 2016.pdf

https://ce.berkeley.edu/sites/defau.../GT 13-01 - Geraili and Sitar (corrected).pdf

My approach thus far has been to bite my tongue and just design for the loading provided by our geotech, but I am wondering what standard current practice is in the geotechnical realm these days. Are there circumstances where the Wood approach is justified? Is this outdated and should I push back or search for another consultant for future work?

The net result is more concrete and rebar for new construction and more retrofits when checking existing facilities against modern codes when modifications are made and upgrades are triggered. It's potentially costing our clients extra money and if our competitors are using a less conservative approach we could lose future business as well.

 

[EireChch]

https://youtu.be/KOrOnFqw2wQ

This lecture explains it better than I could, in summary it seems all methods are conservative

 

[ATSE]

Sitar (UCB), Shamsabadi (Caltrans / UCLA), Stewart (UCLA) and others have been leading this research for the last decade via running scaled lab testing and numerical modeling for validation.
I would be careful to say that "... all methods are conservative..." All of the methods available to practicing engineers are super simplified.
I believe it is more appropriate to say, most of the time Wood and Ostadan methods are overly conservative for plain vanilla retaining walls; if:
- wall height below 10 ft,
- low to moderate seismic shaking,
- not saturated
- flexible wall with no adverse SSI
- predominately low free field frequency content (i.e. long wavelength relative to wall height)
Better, the Sitar solution is valid most of the time. The better question is: does the geotech engineer on your team know how to screen and differentiate the problematic ("Wood") walls from the plain vanilla ("Sitar") walls.

 

[strucbells]

I will watch that lecture EireChch for more background info.

most of the time Wood and Ostadan methods are overly conservative for plain vanilla retaining walls

To confirm, are you referring to cantilevered retaining walls, or would your comments also apply to basement walls that are restrained at the top?

 

[r13]

Using this depth-averaged acceleration measure as a correlation parameter,
the experimental and numerical results for shallow basement walls and deep, stiff
basement walls agree much more closely with each other, with the M-O and Seed &
Whitman (1970) methods providing a reasonable average estimate for the seismic
earth pressure resultant.

Excerpt from P504 of the linked article Link

 

[EireChch]

ASTE - so in summary again, most are conservative. Joking aside, I agree with your further clarifications, but I think for most applications it is conservative. Where you start getting into larger walls, with SSI requirements from adjacent structures too, big PGAs, then I would start looking at PLAXIS/FLAC. (I have never had too, so maybe I am wrong but I would definitely give FEM (with appropriate time histories) some consideration).

Here is another question, what Kh do you apply? Full PGA, 50%, 30%, something else ??

 

[ATSE]

Regarding different wall types, the true SSI for cantilever retaining walls is a bit different than for a soil nail wall. My comments above are for cantilever retaining walls (without top restraint).

Regarding FEA / FDM, there are many more complications to address, so I would not consider those results any more accurate than a spreadsheet wedge model without also seeing a detailed description of the numerical model input, and multiple runs to bound the solution results.
One of many challenges is how big to make your model ("jello problem"), where you will get different results depending on how large your model is (horizontal length of backfill). True for time history and pseudo-static, but especially pseudo-static. As you change the "idealized" damping and model boundary conditions at model perimeter, your results change as well (of course). And keep in mind that it is very difficult to provide a detailed check on the results - a basic 2D model can easily have 10,000 nodes.

Regarding kh, note that the different publications use different variable names (don't these guys talk to each other?).
USACE recommends pseudo-acceleration = 2/3 * EPGA for preliminary investigation of slope stability. However, there is a wide range of kh formulas available, from 1/3 * PGA to 100% PGA. It would be nice if this got condensed down to only 2 or 3 variables for simple walls (say, less than 15' tall).
If you're ambitious, SP117 gives more detailed and less crisp guidance (with kh a function of threshold displacement, magnitude).

 

[strucbells]

Great lecture EireChch. Pointed me to this paper which is really helpful for my condition (basement walls), and gives recommendations about the Kh value mentioned by ATSE in Figure 9 and Table 1 towards the end:

https://calgeo.memberclicks.net/assets/docs/2010-technical-paper-seaoc-lew-et-al-final-rev1.pdf

It appears to me that my geotech on this project did grab the delta Kae from the chart provided at the end of that paper, but their recommendations of where and how to apply the loading don't seem to align with the rest of the paper's recommendations. Anyways, I am going to take a conservative approach for this project and it will be on my radar to address on future similar projects.

In regards to SP117 and the threshold displacement, don't have time to review all that right now, but Sitar's centrifuge testing showed that the soil and wall move out of phase with each other and therefore a displacement-based approach rather than an acceleration-based approach will probably end up being the most accurate in the long term. In the meantime, the paper above includes some "patches" to modify the acceleration-based approach to make it more reasonable for the observed behavior in centrifuge testing and also in real world earthquakes.

Will be interesting to follow how the code evolves to address these research findings over the next few cycles. Hopefully in the effort to create a more accurate model of this behavior things don't become even more complicated