MaterialSample

[mbeaufils]

Source definition : ISO 19156 "Observations, Measurements and Samples"

A MaterialSample is a physical, tangible Sample.

[IHalfon]

In eurocode 7- part2 : the term "sample" is defined as "portion of soil or rock recovered from the ground from sampling techniques"

[dponti]

It seems to me that the generic term "sample" should incorporate anything collected or created (and ultimately tested or observed) regardless of state or material type - eg. soil, rock, fluid, gas, grout would all be samples.

[dponti]

Consider SamplingActivity as a separate class of object? Too often in the geotechnical community we conflate the act of obtaining a sample with the definition of a sample itself. This can cause problems in certain use cases. For example, when a core run recovers no sample or partial recovery. For partial recovery, the depth interval of the run (important to record) is not the same as the depth range for the recovered sample. Where there is no recovery, the activity occurred but there is no sample. Similarly, subsampling and aggregation of samples are activities that can create samples from others and may have different location relevance from the original sample (or the original activity).

SamplingActivity can create 0..unbounded MaterialSamples. All MaterialSamples derive from a SamplingActivity.

[Didymograptus]

IMHO it is not necessary to agree a definition or location in the ontology at this stage (if at all). This is metadata, which can be determined by viewing the factual source data.

[mbeaufils]

Based on March 24 discussion:

• Sample: Collect directly from the field
• Specimen: A sub-sample (on which tests are performed)

[dponti]

If we want to try to follow O&M's semantics, we should only use the term MaterialSample in the context of something tangible, since O&M reserves the term Sample as a more generic object that is representative of a concept, real-world object, or phenomenon, and that also includes MaterialSample as a member along with StatisticalSample and SpatialSample.

I think O&M's definition of MaterialSample, that @mbeaufils first posted is perfectly fine. In fact, Note 3 of the standard on page 97 I think is even better:

To restrict it further conceptually by implying that a sample (MaterialSample) is "collected directly from the field" is problematic because it conflates the sample with the sampling process. Nor should we restrict the definition to soil or rock.

Common in the geotechnical domain are samples not collected directly from the field:

1. Samples can be aggregated and subsequently specimens created from them are tested.
2. Samples can be split up (subsampled).
3. Grout mixes are made in a laboratory and specimens from them subsequently tested to determine if they meet desired strength characteristics.
4. Although more common in the environmental domain samples are created as standards or blanks from which specimens are tested.

These are all MaterialSamples, in that they are all obtained for observation.

[dponti]

Suggested definitions:

Subsample - a MaterialSample derived from another MaterialSample as a direct result of a SamplingActivity.

Specimen - a MaterialSample derived from another MaterialSample as part of the process of an Observation performed in order to obtain the value of some characteristic of the MaterialSample .

SamplingActivity (or SamplingProcedure) - the act of obtaining or creating a MaterialSample. The activity is not directly associated with an Observation.

Examples: 1) Splitting a core (a MaterialSample) longitudinally to produce "working" and "archive" halves of the original core sample. The action of creating the two halves is a SamplingActivity (of type "subsampling") and the resultant working and archive halves (the MaterialSamples created by the SamplingActivity) are each SUBSAMPLEs of the original core. There is no direct Observation associated with the SamplingActivity to produce these two MaterialSamples.

2) A piece of the working half of the original core sample is extracted and otherwise prepared for use in a particle site test to determine d60 and coefficient of uniformity results. The action of producing the MaterialSample for testing is part of the test procedure (directly tied to an Observation) and is therefore not a SamplingActivity. The piece extracted from the working half of the original core sample would be a SPECIMEN.

Conceptually, the working and archive halves of the original core and the specimens obtained for testing are all subsamples of the original core. The distinction is that the act of producing the specimen is directly tied to the observation.

[mbeaufils]

An illustration of Sample and its attributes based on OMS:

[neilchadwick-dg]

Notes on Sample vs specimen

As presented at meeting 12/1/23

From EC7 part 2 (new draft) (but not changed much if at all)

3.1.3.6 sample defined amount of rock, soil, or groundwater recovered from recorded depth [SOURCE: EN ISO 22475-1:2021]

3.1.3.7 specimen part of the sample taken for laboratory testing

Also note the following use of terms:

8.2.1 Direct determination of soil strength (4) The peak shear strength of clays, silts, and organic soils shall be determined using specimens prepared from samples in Quality Class 1. (5) The critical state and residual shear strengths of clays, silts, and organic soils may be determined using reconstituted specimens from samples in Quality Class 4 or above. (6) The shear strength of coarse soils should be determined from reconstituted specimens from samples in Quality Class 4 or above. (7) If reconstituted specimens are employed to determine shear strength, the method employed for specimen formation as well as the composition and state (stress, density, saturation) of the specimen shall be specified before testing and reported with the test results. (8) Specimens should be reconstituted at state conditions close to those existing in-situ. NOTE Methods of preparing reconstituted specimens are given in EN ISO 17892-9.

However, lab test codes, e.g. BS 17892-12, tend to refer to 'sample' as the input, then talk about creating (potentially multiple) specimens for the test itself!

Lets overlook that! The rest of the EC7 is consistent with itself.

My summary thoughts

So this is how I see it.

Samples are taken in the field. Procedure from codes related to sampling (activity).

Specimens are created from samples. Lab tests done on specimens.

Specimens may re-constituted or recompacted or subject to some other process. Covered by lab testing codes. Details of what was done to be reported with lab test results.

In some tests, the 'specimen' may be divided up again as part of the test procesure, e.g. for a compaction test, or the liquid limit test. However, these are considered and reported as part of the overall test. For the dsicussion, lets call these lab sub-specimens (a made up term).

The AGS data structure is consistent with the definitions of sample and specimen above. AGS does not have a specimen 'Group' as such - the specimens are defined as part of the test result, including any preparation information if required, e.g. in LNMC.

The way AGS does it is consistent with the codes, i.e specimen preparation is part of the test (as it is in practice, if you think about it)

It is also worth noting that, both in AGS and in practice, lab test scheduling is done on samples, not specimens.

Outlier?

Amalgamated 'samples' (as referred to in AGS), i.e sample from combination of samples, e.g. for tests that need a lot of material, should perhaps be called amalgamated specimens. (Update - it was changed to amalgamated specimens in recent update!) The AGS methodology for these involves creating a new sample (from which the specimen is then taken). However, I am not convinced that this is a good model, and it only existing in AGS as a compromise due to its strict heirachial structure. My opinion is that the amalgamation can be part of the specimen preparation (as it will be in practice)