Hi,

I am a junior engineer working on a project for a public client and the EPA. The client is used to dealing with environmental data, contamination, remediation goals etc. Now that some soils testing is needed they are asking for data quality objectives (DQOs), a "geotech SAPP/QAPP", and limits for Type I/II decision errors to determine the number of geotech test. Not having a lot of experience, I am unsure how typical these practices are for geotechnical engineering. Seems like geotech testing is too expensive to select a "sample size" based on confidence intervals etc. The number of samples should be based on intuitive estimates of the expected variability in the soils and the different strata/parameters needed, correct?

What are your experiences/thoughts?

The investigation should be planned by a geotechnical engineer to identify the relavent geotechnical issues. The engineer selects the investigation methods, sampling intervals, etc. The owner can retain a geotechnical consultant to review the adequacy of the proposed investigation. If they owner does not know how to "deal" with the data then they probably shouldn't.

Thanks molerat! I agree. I think the basic question has more to do with assuring that quality data is obtained from money spent on the field investigation. In other words how can the geotechnical engineer quantifiably demonstrate that the needed information will be obtained other than saying "trust me I know what I'm doing".

Also, incases with a public client, there may be a need to defend decisions or money spend on testing to public. Statistical validity can justify a sampling budget for environmental analysis. But as geotechnical engineers, I believe engineering judement is equally credible for defending testing programs, right?

No analysis of geotech properties will be done by anyone other than the geotechnical engineer, that's for sure! People might question the sampling budget though :).

Another fine example of bean counters and non-technical QA people trying to direct, dictate and supplant the professional judgment of licensed professional engineers. Bull$hit.

Molerat has it right.

Scott,

I think you are asking two questions: 1) how does the engineer justify adequacy of the investigation program, 2) how does the engineer justify costs. The answers would get long and complicated if I had enough time.

Public works projects are usually assigned to the engineer through a basic ordering agreement (BOA) or similar, or the engineer could be part of a design-build team. In each case, the engineer is required to prove that they are qualified by a process of quality-based selection (QBS) process. So, there is a big element of trust but only after the qualifications have been proven. The qualifed consultant would develop the investigation program after qualifications have been established.

We don't talk about costs until after each consultant has proven their capabilities. It is rare for a public client to take issue with the cost since this could comprise the integrity of the program. The more common problem is not proposing enough investigation (i.e., the engineer is too shy to ask for what the job really takes).

The quantity of borings and samples is usually justifed by established guidelines. For example, AWWA has boring density guidelines for water tanks or state agencies have established boring density for bridges. Sample density is also often prescribed. Additional sampling is often required is specific structures are being constructed such as a tunnel or a certain geologic formation is encountered (bedrock, soft clay, peat, etc.). Other investigations may be warranted such as geophysical if the engineer determines that its worth it and makes his case to the owner.

Most public agencies have reviewers whose only purpose is to review the adequacy of geotechnical work. I do sometimes see statistical analysis used to evaluate test data but never to review the adequacy of a proposed test program. When statisical analysis was used to evaluate test data, it was only for the largest projects where there was a lot of controls in-place for QA/QC of the geotechnical work. I have done confidence intervals on smaller projects but only to amuse the client after they requested it. It had no impact on the outcome for the foundations and earthworks.

Quote (scottm8)

I am unsure how typical these practices are for geotechnical engineering.

I daresay they are very atypical

Quote (scottm8)

The number of samples should be based on intuitive estimates of the expected variability in the soils and the different strata/parameters needed, correct?

Basically yes, this aspect has also been discussed in the other thread about # borings. Moreover, since lab test are almost always too few and far between to be statistically representative, usually field tests wich can yield numerous data such as the various CPTs, the SPTs the DTMs and all dynamic penetrometers are calibrated aginst such lab tests. Usually.
In geotechnical engineering probability analysis is used in reliability determinations, variability and epistemic uncertainty of parameters, error propagation, spatial mean and variance, random processes and random fields (cyclical fluctuations of parameters), point estimates of cautious representative values (characteristic values according to the Eurocodes). And something more.

Sorry but I forgot about the thread title: "Role of hypothesis testing"

That might be used to determine if for example, two different borings went thru the same statistical homogeneous layer or not, I used ANOVA a few times to compare CPT signals in determined intervals but I find that a trained eye is far more effective than statistical testing.