see asce7 table 20.3-1 for help. you can also fall back on ibc ('00 or '03) table 1615.1.1 for a similar table. be sure to see the notes about plasticity, moisture content, and su for class E.
i don't use deep foundations to start the analysis deeper. those upper soils still influence the structure and should be accounted for. this is a gray area but is pretty much stated that you evaluate it as needed. some folks use the deep foundations but i believe it is an incorrect approach. the only time i might start my analysis at the bottom of the deep foundation would be if it's a "short" caisson through pretty good soil and embedded in rock and i'd still try to look at the effects of the soil. but that's just me. as i said...gray area...i've met many that would do more or less than what i describe.

Site class is based on the top 100 feet of soil and the shear velocity of this soil.  As a first estimate,(until data arives with blow counts or shear velocity), use Site Class D.

if it were my stamp on the report, i would double check the note i previously mention before i come to that conclusion. but that's just me...

40 KPa is about 900 psf, which is just not that stiff (in my book).  The bedrock conditions below the depth of 45 ft would likley be suited for site class "C", but the surface layer of clay is more suited for site class "E".

What method did you use to get the undrained shear strength for the clay?

I'd likely use site class "D" also. . . .

f-d

¡papá gordo ain’t no madre flaca!

The su is based on pocket pen tests (not so reliable).
The table in IBC is useful, however it does not have any specific comment for my case.  The National Building Code of Canada (NBC) has a comment that we should not infer shear wave velocities from SPT or su values. If ignoring that comment I can infer the shear wave velocities in the upper 30 m including rock and based on average velocity determine the site class.  However, I am reluctant to use this method having seen the sentence in NBC.

there is verbage in asce7 (and probably ibc but don't recall since it's been a while since i've gone through chapter 16) regarding inference of velocities, so i'm 99% sure it is allowed based on the way it is phrased in the text...whether the inference is legit or not is another story. you could always just pretend the site doesn't exhibit characteristics of the notes for class E and then assume the default class D. doing this would keep me up at night...
you don't really get credit for the rock unless you evaluate the shear wave velocities...that is one of the conservatisms built in to the method since there can be so much variance due simply to the field practices of the drilling crew. plus, stb's are used an indirect indicator (via correlations) of the soil's response to seismic activity. the shear wave analysis is more of a direct measurement of how the soil will react.
as far as deep foundations, nehrp provides comments (although rather vague to some degree) about the need to evaluate the effect of the soils that the foundations are embedded through.
http://www.eng-tips.com/viewthread.cfm?qid=204075&page=1

if there's a market for it, you could always purchase the equipment and software to do the field work yourself. on another thread, you mention using rayleigh waves to evaluate cavities...they can also be used to evaluate seismic site class due to the relationship of Vr to Vs. it would likely depend on which method you used since SASW/MASW likely would not see deep enough. however, ReMi is a widely accepted method for such a task and can easily see 100+ feet with a modest investment in equipment (or using existing certain seismic survey equipment--such as refraction equipment).

From 2000 IBC - Table 1615.1.1, "Site Class Definitions"

"C" - Very Dense Soil and Soft Rock, Vs=1200 to 2500, N>50, Su>2000

"D" - Stiff soil Profile, Vs=600 to 1200, N=15 to 50, Su= 1000 to 2000

"E" - Soft soil profile, Vs<600, N<15, Su<1000

From my review of this table there is an accepted correlation between density (for the case of coarse-grained soils using N-value) and consistency (for the case of fine-grained soils using undrained shear strength - i.e., pocket penetrometer data) and shear wave velocity.

When faced with broad catigories like these, I would not pursue a highly refined field method to see whether you are just barely on the side of "D" or "E".  If there was any doubt, I'd just go with "E".

Hope this helps and contributes to the discussion.  Now I'll go have coffee - ha.

f-d

¡papá gordo ain’t no madre flaca!

In accordance with MSUCOG, here in North Carolina, there is a blurb in the building code that states the default Site Class is a D if subsurface information is not available.  What is alarming is that some firms who do subsurface borings and find the Site Class to be an E will state in their report that a D is suitable because it is the default site class.  Scary.


I stray from assumptions and go with the additional testing required to determine the Site Class in accordance with building code.  Check with the structural engineer to see if there is a significant cost to the project based on Site Class.  If so, recommend the additional testing required to determine the Site Class.

fatdad said:
'When faced with broad catigories like these, I would not pursue a highly refined field method to see whether you are just barely on the side of "D" or "E".  If there was any doubt, I'd just go with "E".'

I agree you re: defining site categories with the respect to accuracy of correlations used to in the table. However, depending on seismic zone and the response spectra of the structure, isnt there situations where going with E is less conservation then D (rigid short structures in zone 4)?

Quote:

However, depending on seismic zone and the response spectra of the structure, isnt there situations where going with E is less conservation then D (rigid short structures in zone 4)?

I took a class several decades ago (graduate school, Wayne Clough) pertaining to earthquake engineering.  We learned about response spectra, period of structures, liquifaction, etc.  If I recall correctly, there may have even been a program like "Shake" or something (DOS) that integrated all this along with low-strain shear modulus to give you a ground response for a design earthquake.

For some reason, I don't think that broad site classes like "E" or "D" and their associated refinement has too much to do with what I learned.  That said, I would be interested in feedback from a structural engineer on the point that you raised.  I'll also ask in the office for their feedback.

f-d

¡papá gordo ain’t no madre flaca!

he's my two cents...the whole procedure has been compacted down in to a nice and neat 5 lines table so the simplification and built in conservatism has already occurred across the entire process (in a sense). however, it cannot possibly take in to account for localized site effects therefore we have a site class determination thrown in to help. so when soil test borings are used, there is some amount (likely not an extrodinary amount) of safety factor built in to the correlations between blow count and shear velocity...this due to material effects like rocky soil or variability in the rope and cathead procedures. that's all fine. however, when they refer to "soil" it is not just the soil me and you automatically identify with as engineers. it also includes part of the rock layer (hence, this is why it says "soil profile" on the table but also includes rock in there). with soil test borings, you do not get "credit" for the rock layer...it simply is limited on the upper bound to 100bpf. by using shear velocity assessments, you are able to take advantage of the "unused" rock properties from the soil test borings.
another point is this: if you are in high seismic zones, trying to fine tune the analysis to just make it to the other side of the line may not be appropriate but there again, a site specific analysis would likely be required anyway. if i had performed shear wave analysis and it was marginal, i would make sure i'd used a "true" model in the sense that i will often use additional conservatism in my models where i live since it still usually provides a jump in site class. if it's marginal between say D and C, i'll sharpen my pencil, refine my assumptions and see how my initial "quick and dirty" assessment compares to a more refined model. if, with my sharpened Vs model, the numbers are still marginal (to either side of the line), i would go with the more conservative possibility. if i had used stb and the numbers were marginal, i'd refine my model and look at my N-value adjustments (usually using auto hammer). if it's still marginal, i'll recommended either a site specific assessment or Vs testing. the site specific can reduce the Ss and S1 by as much as 20% and because of where i live, i typically see 15-18% reduction (and as it happens, an 8% reduction will usually kick the SDC up by one category due to my locale). the Vs has provided a jump in site class on each and every project i've used it on (and thus gives a jump in SDC).
by the way, 100' is not the absolute number to use...if the soil/dirt profile is more than 100', that depth of analysis should be increased accordingly.

jmgray,

Yes, at least with the 1997 UBC, a site class E will give you lower short period accelerations compared to a site class D, which will be applicable to a relatively short rigid structure(less than 5 or so stories).  Site class E will give higher accelerations for longer period structures (taller structures).  So you should look at the period of interest of your particular structure in order to determine which site category will result in more conservative values.  I suspect the same is true with the 2006 IBC.

jmgray, i thought that's what you might have been eluding to where the max accel is lower for class E but it just last through much long periods. if i'm not mistaken, this idea is combined in to the current method but in a different way. while the max at short periods is less for a class E material, the amplification factor is higher because you should account for the entire short period and not just a spike. for example, see the following (topic 15-3 slide 30):
http://www.bssconline.org/FEMA451B/451Bchapters.htm

so under current code, the design parameters SDs and SD1 for E will always be higher than for D.

i'm not completely familiar with the older codes, but it appears to me that this particular issue might be why the code is using the current system so that it does look at the enitre long and short period accelerations instead of just a single "maximum" number that was supposed to cover all the bases. to be more accurate, i suppose there is actually a third parameter with the very long period accelerations that showed up with the latest version of the codes (i don't have the older IBC codes with me but i don't recall seeing the very longer periods until the '06 release).

I had a chance to have a look at some references and some codes (ASCE-07, IBC 2006, NBC and OBC).  A good reference is NEHRP (2000) recommendations which can be downloaded from: http://www.bssconline.org/NEHRP2000/comments/provisions/ (Chapter 4). The latter is a reference of all the codes that I have reviewed.
 
As I understand the previous versions of the codes were developed based on the relative impedance between the overlying soil layer and the underlying rock mass. This is why in the previous versions the amplification factor for a site consisting of Rock was considered to be 1.  The new versions are developed around the theories presented in Borcherdt (1992, 1994) which consider the average shear wave velocity in the upper 30 m of ground profile. Basically it needs the time required for the shear waves to traverse the upper 30 m of ground. This is why in the new versions of the codes Site class C or D is considered as base and all other seismic factors are calculated based on that, i.e. factors for site class s C or D are 1 and for better sites are less than one and for poorer sites are larger than 1.
 
Based on the above, following are some thoughts:
 
1 -  the important parameter is the soil condition in the upper 30 m below the foundation. So, if rock happens to fall within this depth it should be considered in the site evaluation.
 
2 - In all the codes the basic study method is evaluation of the shear wave velocity.  Other methods like SPT 'N' Val;use or Su can be used for this evaluation as well. Only Nib's commentary has the comment that shear wave velocities can not be inferred from SPT or Su. If you look at the attached reference (which is a main reference for NBC) page 275 second column 1st paragraph, it clearly says that if shear wave velocities measurements are not available Su or SPT 'N' values can be used to estimate shear wave velocities and get the average over the upper 30 m of soil profile. Further, ASCE-7 has a sentence that in the presence of rock the SPT 'N' value can be assumed to be 100.  NEHRP also states that SPT 'N' values should not be considered more than 100.  Therefore, I think either estimating shear wave velocities from SPT or Su, or assuming an SPT of 100 for rock and using the equations presented in all the codes for averaging is a rational method in cases that mixed soil conditions are encountered.      
 
 3 - When we have piles going to competent bearing layers I think again the important factor is the soil profile in the upper 30 m of soil-rock measured from bottom of the pile.  However, special attention should be paid to the effect of liquefiable soil or soft soils on the piles during earthquake. If the surrounding soil loose its shearing capacity the pile will buckle and the whole structure will be in danger. I have seen in some cases people assume that the piles have some improving effect on the soil and depending on the relative stiffness of the piles and the surrounding soil a site class is chosen which is poorer that Rock and better than the original soil condition.
  
I appreciate your further thoughts and comments.  

regarding deep foundations: keep in mind that in an earthquake, the soils are going to amplify the motion. so there will be more motion at the ground surface than at the bedrock. so assume you've got 50' long piles through 5 blow soil and end bearing (even socketed) in to rock. so you're saying you should calculated Nave from bottom of pile down? that would give you Nave=100/(100/100)=100 which would give you site class c...to get better, you'd then have to assume shear velocities. however, "site classes A and B shall not be assigned to a site if there is more than 10 ft of soil between the rock surface and the bottom of the spread footing or mat foundation."...from ASCE7 (notice they didn't say "the bottom of the foundation")
if you calculate Nave from the bottom of the pile cap down, you'd then have Nave=100/[(50/5)+(50/100)]=10 with a site class E.
here's my point: if the soil at the ground surface is moving more than the soil 10' deep where that soil is moving more than the soil 10' below it...so on and so on...the piles will essentially experience bending due to the seismic movement at the ground surface (and along the length of the pile). and this hasn't even included anything to do with possible liquefaction or combination of effects.
i could be completely wrong...however, i suggest that you review FEMA 450 commentary page 30 (at bottom). it states:

"The site class should reflect the soil conditions that will affect the ground motion input to the structure
or a significant portion of the structure. For structures receiving substantial ground motion input from
shallow soils (e.g. structures with shallow spread footings, laterally flexible piles, or structures with
basements where it is judged that substantial ground motion input to the structure may come through
the side walls), it is reasonable to classify the site on the basis of the top 100 ft (30 m) of soils below
the ground surface. Conversely, for structures with basements supported on firm soils or rock below
soft soils, it is reasonable to classify the site on the basis of the soils or rock below the mat, if it can be
justified that the soft soils contribute very little to the response of the structure."
i interpret that to mean as i attempted to describe above...if the pile cap is at the ground surface, calculate the average from the pile cap going 100' down...not from the bottom of the pile going down.

Getting back to the site class, in my area of Southern California, the Design Response Spectra for a site class D is higher in the short period range than for a site class E.  This is similar to the 1997 UBC.