Cost of CPT vs. SPT 3

[ncarolinageo]

I am trying to get my bosses to approve some CPT work at a project site where we have concerns over luiquefiable soils. Wwe have plenty of SPT data, as well as shear wave measurements. We feel that we have characterized the site in general and now we need to fill in the gaps. I am proposing that we use CPT to so that we can more accurately evaluate the liquefaction potential of the soils. I have been asked to provide a ballpark cost estimate. The borings would likely need to be accessed using a tracked rig - though some may be truck accessable - and would need to extend no more than 50 or 60 feet to the top of rock. I am not sure how many borings we will need - but I would estimate it on the order of 12 to 18 locations.

 

[GeoPaveTraffic]

Are CPT rigs available in the area? If they are call them and either get a quote to drill the holes or get a standard rate sheet so you can estimate the costs. If CPT rigs are not available in th area, find one nearby and get a quote to mobilize to the site and do the work.

 

[BigH]

Also, I would use the seismic cone rather than a standard CPT. Google Paul Mayne and look up some of his papers he has put on the web -

 

[ncarolinageo]

It's funny you should mention Paul Mayne - I met Dr. Mayne last year and I have already been in discussions with him via email regarding the project. I would agree with your thoughts on the seismic cone.

 

[BigH]

I count myself fortunate to have been a classmate! I was introduced to the piezocone, etc. by Campanella of University of British Columbia.

 

[dgillette]

The CPT is IMO a very good tool for getting a lot of data quickly to fill in between SPTs, and it's probably a better test for liquefaction potential anyway. If, as BigH suggests, you can get SCPT, that will give you quick and cheap SWVs to the depth you can push the CPT, for both response analysis and liquefaction potential. Even better if you have a piezo measurement to help spot contractive zones.

There are not a heckuva lot of tracked CPT rigs out there in the US, so timely availability could be an issue. Back in the '80s I logged well over a mile of CPT soundings, mostly in very soft mine spoil and tailings, but also in some naturally filled karst features full of clay and peat. For almost all of that, we pushed it with CME 35 drill rigs, one on a fairly small truck and the other on a Bombardier crawler. Sometimes, we used flight augers as anchors to hold the rig down, but it worked as long as we were in soft stuff. We had an adapter to connect the CPT to the driller's A rod and ran the cables through the rods and did 5-foot strokes instead of 1-meter strokes. Something like that might work for you.

 

[KVgeo]

CPTs are my tool of choice for assessing liquefaction potential of soils when used in association with sufficient energy calibrated SPT measurements to confirm the assumed qt - N160 relationship.

Agree with dgillete and BigH use a SCPT to get shear wave velocities. Shear wave velocities take extra time and cost more in reporting. Make sure you discuss with the CPT operator the costs associated with performing numerous shear wave tests or you might get a nasty shock.

Costs usually vary depending on the type of CPT rig you choose. You can get portable rigs which are pushed by a drill rig or self contained units which drive onto the site. The self contained rigs are usually more convenient and probably a little bit more cost effective.

Generally speaking in Ontario portable rigs can cost up CA$3000 per day and mobilisation can cost upwards of CA$4000 depending on where you are. Pushing rates, including shear wave testing and dissipation tests, can be 4m to 6m per hour depending on number of setups, depth of hole and access.

The above figures don't include reporting and you will need a drill rig to push the cone.

Hope this helps....

As a side note, I don't recommend using the published CRR charts for CPT probes. There are issues associated with compressibility that still haven't been resolved yet. Converting to N160s is a much safer route and you can compare to the existing SPT data that you already have.

 

[dgillette]

Hello, KVgeo.

I would have thought that the additional dose of uncertainty introduced by converting CPT to equiv. SPT is at least as bad what might be caused by compressibility effects in the CPT triggering curve, especially when you consider all the corrections that went into locating the dots on the SPT triggering plots. Wouldn't the compressibility issues affect the CPT-SPT conversion factors? What do you do for the fines content to assume in the SPT correlation, estimate "equivalent fines content" via Robertson's Ic?

My understanding is that certain micaceous soils and hard rock tailings tend to be more compressible than typical soils, and therefore give tip resistance that is "low," making for a liquefaction assessment that is unnecessarily conservative. Do I understand it right? If we are using correlations (Robertson and Wride, Moss and Seed, etc.) developed from typical natural deposits in similar soils, shouldn't we get about as good of an answer as we can hope for from shoving a steel rod into the ground? In the recent EERI monograph, Idriss and Boulanger don't mention concern about this issue. Has Peter Robertson published anything on compressibility effects, post-NCEER?

 

[KVgeo]

Hello dgillette,

Firstly, I apologise for misspelling your name previously.

Any good liquefaction program includes at least a couple of independent in situ assessment methods to assess liquefaction potential. As I understand the current order of confidence in insitu testing to CRR relationships is SPT, CPT, Shear Wave Velocities and Becker Hammer/other.

The use of CPTs should be calibrated with energy measured SPTs and SPT results should be verified with other in situ penetration tests.

Compressibility does affect the CPT to SPT conversion factors. Issues which can determine the compressibility of a soil include fines content, angularity of the particles and geochemical issues.

Both Davies and Lunne present methods for estimating N60 values from qt using Ic. In my experience this usually provides very reasonable results. Some issues which may affect the relationship include the angualrity of rock tailings deposits and the current overburden correction assumptions.

In summary the use of any CPT to SPT relationship should be verified to the designers satisfaction with on site in situ comparisons or demonstrated comparisons on similar previous sites.

The effect of fines on the current CPT - CRR relationship is not well understood. I direct you to Boulanger and Idriss' recent monograph. In my opinion these charts have not been developed enough for general use and are in their preliminary stages. I also understood that the Robertson Wride CRR curves are generally non-conservative and should be used with caution.

My recommendation as stated above, is to convert CPT results to SPT values. The CPT provides consistent repeatable results that when converted and interpreted provide reasonable estimates of N160 values.

It should be noted that the simplified method is one way of assessing the liquefaction potential of soils. It is empirical in its approach and does not provide a thorough understanding of soil behaviour during and after cyclic loading.

At the moment shoving a steel rod in the ground seems to be the generally accepted standard of practice for assessing liquefaction potential. Until somebody finds a way to get undisturbed samples of granular soils at a reasonable cost I think we are stuck with empirical in situ testing relationships.

As a secondary issue, I am not aware that there is an overwhelming consensus regarding the general applicability of fines correction for all types of soils in liquefaction assessments. Which is another issue associated with the CPT - CRR relationships.

Anyway that's my two cents hope it adds some value to the discussion.

 

[KVgeo]

Before I forget one more plus to CPT testing is you don't need to worry about soil disturbance and balancing water pressures in artesian conditions.

In soft to firm non-plastic silts and very loose to loose sands beneath the water table getting reliable SPT results is very difficult without weighted drilling mud to prevent heave and blow out. And by weighted drilling mud I mean something in the order of 1.2 times the density of water.

Not sure of your site conditions ncarolinageo but are you confident your SPT results are reasonable? Another good reason to do some CPT testing.

 

[ncarolinageo]

Well - If this is of any interest to you (meaning dgillette and KVgeo), my hope in having CPT (or SCPT) performed was so that I could have multiple data sets to compare. I have a good ammount of SPT data using calibrated hammers from borings performed using drilling mud - but I do have concerns over heave and blow out (or what a driller I used to know called blow-back) - and I agree that is a good reason to perform CPT (or SCPT).

Another point I think is interesteg, my current liquefaction analysis based on the simplified method using N60 values derived from SPT data indicates zones of potentially liquefiable soils. However, I have some shear wave data and using the correlations presented in the NCEER paper (and well documented by R. Andrus and Ken Stokoe)and the measured shear wave velocities in these layers are well above those where liquefaction is thought to occur. One arguement I have heard against the shear wave correlations (most recently to me by Paul Rizzo) is that the shear wave method is not good at identifying pockets of liquefiable soils because the shear waves will travel best through the higher density soils found above and below the layers of concern. I can see that point, but couldn't you argue that any seismic-induced shear stresses would do the same?

So for my sute - I feel that CPT (or hopefully SCPT) will confirm that the shear wave data I have so far is better suited to evaluating the liquefaction potential of these soils that the SPT data correltaions, for the same reasons as those stated by KVgeo.

Thanks for the interesting discussion.

 

[dgillette]

KVgeo - I wasn't too concerned about the spelling of my "handle," but I do have some points of concern about CPT vs SPT.

You are the only one that I have read or corresponded with who, in 2009, prefers to convert CPT to SPT, then use the SPT CRR curves. This was pretty common pre-NCEER, but today, more typical practice is to use Robertson and Wride, with drilling to confirm fines contents and check qc-1N against N1-60. I don't believe that the I&B monograph recommends going through SPT (or at least I can't find the page where it does), nor do Robertson, or Moss and Seed, or the consensus report of the NCEER working group. Sure, you can convert to SPT, but if it involves Ic and empirical conversions, that introduces just as much uncertainty as using Ic in a direct CPT-CRR correlation.

Is there any recent research that you can cite to support going through SPT as the better way to go? I'm now at a crossroads in establishing my agency's methodology (along with Robert Johnson), so this is the time I can change our path if I have a research basis to justify bucking mainstream practice.

I&B recommend adjusting qc-1N by a fn. of fines content, rather than any property measured by the CPT (like Robertson does, using sleeve to get Ic), which makes it quite labor intensive and introduces another big source of uncertainty (and subjectivity) in inferring fines contents for CPT holes w/o companion drill holes, on a variable site (most alluvial sites), undoing much of the advantage of CPT in dirty materials.

You mention several times that the effect of fines on CRR for a given qc-1 is not well understood, but it's really understood no better for the SPT. Witness: the very large difference between the amount of benefit they provide in the pseudodeterministic curves (Idriss & Boulanger, Seed and Idriss, NCEER), and the amount found in the statistical analysis by R. Seed, Cetin, et multum.

[Note that there are rather substantial differences in the cyclic shear stresses used in the older and newer studies, so the Seed-Cetin and Seed-Moss CRR curves are not directly comparable with the older ones, a point that gets missed with some frequency.]

As many case history data as are available nowadays to support a CPT-CRR relationship, I think we need to stop viewing SPT as the gold standard for these things. The SPT is just too sensitive to too many variables, and that includes the SPTs used to develop the correlations. Literally and figuratively, it is a blunt instrument. Shoving a rod in the ground (with instruments at its tip) can't be any worse than pounding a rod into the ground, for which there is not even consensus on such things as the short-rod correction, which is VERY important because of the shallow depth of so many of the case histories that went into the SPT-CRR correlation.

I gotta get back to my billable work.

 

[KVgeo]

dgillette,

Agreed the effect of fines on liquefaction resistance in general is not well understood. Especially non-plastic fines.

Part of my issue with the existing CPT charts is there is no seperate factors to account for the effect of compressibility and the effect of fines on the soil. This is applicable when you have non-plastic silty sands and decide not to apply the fines correction. The existing CPT charts do not allow this without being overly conservative by assuming the base clean sand curve CRR values.

Boulanger and Idriss' CPT curve accounts for fines by using their assumed relative density/critical state models . The SPT fines corrections are converted to equivalent CPT qt values and added on to the clean sand base curve. Strictly speaking, this is not the case because the fines will also affect the relative density assumptions used in the correlation and will alter the base curve. Accordingly, there should be variable base curves for different materials (i.e. sand wi 5% fines, 15% fines and 35% fines etc).

Robertson and Wride's CPT curve shows a much more significant increase in CRR values with increase in Ic (fines content). However, I think Moss' and B&I's research with additional CPT liquefaction data indicated that R&W's curves were most likley non-conservative. I should be able to dig up the references after lunch if you are interested.

By converting to CPT data to equivalent SPT values I can demonstrate that I get similar CRR values from side by side CPTs and SPTs. I have not had the same luck when using CPT CRR charts, which seems illogical.

Anyway, as mentioned above, it is generally recommended that liquefaction screening assessments should be performed with at least two alternative in situ testing methods to reinforce the results from each other.

I would recommend reviewing the assumptions that B&I used to develop their CPT CRR chart and see if they apply in you situation. I haven't reviewed the assumptions used by Seed and Cetin for their CPT chart so I can't really comment on it.

Regards

KV

CAVEAT: it is not possible to mix and match CRR charts with different methods. If you are using B&Is CRR charts then you must use the B&I empirical rd method to calculate CSR values. If you use the Seed and Cetin method then use a site response program to calculate CSR values.

 

[dgillette]

Actually, it was Moss and R. Seed, rather than Cetin and R. Seed involved in the CPT; Cetin did the analogous work for SPT. Moss and Seed minimized the use of assumptions like fines adjustment, MSF, and K-sigma by approaching it statistically, using independent variables qc1N, friction ratio, eff overburden, CSR, M, and maybe something else. (Cetin and Seed used N1-60, and % fines in place of the CPT data.) There were still some assumptions about Cn, but the effect might not be very large due to most of the case-history sites being fairly shallow (not 50 feet down).

If you don't have something else to support going through SPT, other than the ones we've mentioned here, I probably have all the refs in my heap of paper here. I've had mixed results on conclusions of CPT vs SPT, although I don't have very many sites where I can use both. Most have too much rolled fill over the critical areas, or too much coarse gravel and cobbles, or some such. (Yes, we've tried the trick of predrilling and backfilling w/ pea gravel, also with mixed results.)

R&W consider Ic>2.6 to indicate non-liq material, whereas Moss found a couple of liq'd sites with Ic>2.6, so it's best not to rely on that criterion. Note, however, that they were sites with quite low qc1N, not somewhere you would be likely to build anything.

Implicit in converting CPT and using the SPT charts (instead of the reverse, or treating the two as independent indices) is the assumption that the SPT is inherently better. That is hard to justify, given all the corrections and adjustments (short rods, hammer energy - is adjustment really directly propotional? - borehole diameter, liner yes/no, rod diameter, layering thinner than 1' test interval, etc.). Not sure we should look too hard for consistency.

Your caveat is what I was referring to in [square brackets]. Based on my own zillion SHAKE runs, the huge range in R. Seed, Cetin, and Moss's Rd curves is to be expected, but unfortunately, it makes one wonder whether Rd (anybody's Rd) should be used for a major project. I&B use depth and M as the only variables, whereas velocity profile and freq content of ground motions make a big difference. Haven't seen any specifics, but I've heard rumblings that the CSRs used by R. Seed et al having been too low (causing P(liq) to be overestimated by their model, and CRR to be underestimated, even when used with SHAKE stresses). I can't comment w/o knowing details, which I haven't tried to track down, but should.

 

[KVgeo]

dgillette,

You're right Moss and Seed (2006), thanks for pointing that out.

This is turning out to be an interesting discussion.

Your point about empirical rd and shake values is spot on. It always makes me feel a little bit awkward writing liquefaction reports and explaining how we did this site response analysis only to throw away the CSR values in favour of an empirical CSR relationship that more often than not bears no relationship to the shake CSR values.

When comparing I&B to S&C I have found some pretty similar FOSliq. S&C generally always estimates slightly more liquefaction than B&I, but all in all the two methods (in my cases) are generally comparable.

I have heard that some think that the S&C approach is flawed due to the lack of time histories used by S&C when performing their site response analyses.

This just highlights the fact that the approach is a Simplified Method and is not really meant to be anything other than a guide to assessing liquefaction potential. I think that well thought out, considered screening penetration values might do just as good a job as the Simplified Method in most cases.

I suppose that until somebody comes up with a workable critical state theory we are all stuck with it.

Regards

KV

 

[dgillette]

Even when we have that workable critical-state model, it still won't mean anything unless we can put in the right parameters for a material that may have vertical variability on a scale of inches and horizontal variability on a scale of feet to tens of feet. We'll still have a whole lot more theory than data to use it with.