sampling: theory and practice
TRANSCRIPT
![Page 1: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/1.jpg)
7th LBMA Assaying and Refining Conference
Keynote Speech:
Sampling: Theory and Practice
Professor Kim H. Esbensen Danish Geological Survey & Aalborg University
KHE Consulting, Copenhagen [email protected]
![Page 2: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/2.jpg)
The primary objective of the 2001 AMIRA P754 project, ‘Metal Accounting and Reconciliation’, was to improve the auditability and transparency of Metal Accounting from mine to product and thus to assist in good corporate governance. Project areas had broad coverage from precious metals to commodities and from mine to product. One of the major deliverables was the development of a Code of Practice for Metal Accounting for the Mining and Metallurgical Industry. The aim of the Code was to provide standard generic procedures and guidelines for obtaining credible figures of metal quantities processed and produced, and methods for obtaining a Metal Balance, and to be recognised and accepted as the Industry standard for best practice in this area.
![Page 3: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/3.jpg)
A major point of the conclusions was: Metal balancing and reconciliation techniques can be combined with the CI/CO approach, provided they are adapted to suit the special requirements of aligning commercial and technical estimation. Simulation techniques are recommended for both metal balancing and reconciliation. However …
at least some knowledge of the errors associated with each measurement
(including sampling and assaying) is essential.
![Page 4: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/4.jpg)
In this context, and with respect to subsequent implementation of results and recommendations, it was felt that the time had come to take stock over how efficient these objectives have been met …. - and an invitation to the present lecturer resulted. This keynote lecture deals exclusively with the sampling issues involved in this endeavor.
![Page 5: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/5.jpg)
It is generally assumed that the met is well mixed due to stirring caused by the elecmagnetic fields in induction furnaces ...
This is a crucial assumption
How well is this verified?
![Page 6: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/6.jpg)
A quotation (from earlier LBMA conf.)
” ... avoid the uncertainty related to sysetmatic errors by proper process concept and by appropriate sampling; when imhomogeneity of the sampled material can be presumed, sample the material where its fineness can be expected to be lower” ...
1. So heterogeneity does occur ... ... 2. All the world’s precious metals do not crop out of the ground as material ‘ready-for-the-crucible’ ... ;-) 3. Where is the authoritative manual/standard/GLP?
![Page 7: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/7.jpg)
Manual sampling - by “professionals” …
A grab-sample, representative of 54 ton material?
![Page 8: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/8.jpg)
Sampling is not so much material (lot) dependent --- sampling is much more up against lot *heterogeneity*
![Page 9: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/9.jpg)
DS 3077 (2013). A minimum sampling competence encompasses FSP, TOS’ paradigm of sampling correctness, five sampling errors (CSE/ISE) and four Sampling Unit Operations (SUO)
![Page 10: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/10.jpg)
The keynote invitation turns out to be timely - since all stakeholders, scientific and regulating agencies and authorities today actually have access precisely to the desired standardized procedures for sampling – but how well is this known? 2013 saw the publication of the world’s first universal standard for representative sampling, DS 3077 (2013), called a “horizontal standard” since it describes all generic principles and procedures with which to be able to guarantee representative sampling at all stages, and indeed at all scales from “mine to product” (“from lot to analysis” in the standard’s parlance). This is good stuff!
![Page 11: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/11.jpg)
The horizontal nature of this standard also means it is designed to be matrix-independent, i.e. the principles can be applied to all types of commodities, materials and process irrespective of composition. It turns out that it is highly advantageous to focus on the heterogeneity (only), which brings about a dramatic reduction of the complexities that is normally envisaged when addressing sampling. It is no longer necessary to ask: representative sampling? Well how in the view of so many different materials … …
(this must be complex!)
The Theory of Sampling (TOS) to the fore …
![Page 13: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/13.jpg)
1. FSP: Fundamental Sampling Principle 2. SSI: Sampling Scale Invariance 3. PSC: Sampling Correctness (bias-free sampling) 4. PSS: Sampling Simplicity (primary sampling + mass-reduction) 5. LDT: Lot Dimensionality Transformation 6. LHC: Lot Heterogeneity Characterization (0-D, 1-D) 7. SUO: Composite Sampling 8. SUO: Comminution 9. SUO: Mixing / Blending 10. SUO: Representative Mass Reduction (Sub-sampling)
Sneak preview: Theory of Sampling (TOS)
Rules for representative sampling in practise
![Page 14: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/14.jpg)
A fundamental insight … where all sampling starts
Princple of Sampling Simplicity (PSI)
1. Sampling is practically never a one-shot operation 2. Sampling is always a multi-stage process: i) Primary sampling + ii) Representative Mass-reduction
![Page 15: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/15.jpg)
Starting lot LOT (aL)
↓ Primary sample S1 Primary reject
↓ Primary preparation S’1
↓ Secondary sample S2 Secondary reject
↓ Secondary preparation S’2
↓ Laboratory sample S3 Tertiary reject
↓ 3’ preparation/presentation S’3
↓ Analytical sub-sample S4 (Analytical aliquot) Final reject
↓ Analysis Analytical result: aS
Sampling errors at all stages
The analytical result – is but an estimate of aL
Who’s responsible …?
![Page 16: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/16.jpg)
µ (σ) x
LOT
Primary Secondary Analysis Result sample sample
s1 s2 s3 sx
Propagation of errors: Example: GOAL: x = m
∑= 2ix ss
%5.5(%)30%)1(%)2(%)5( 2222 ==++=xs
“Analytical process” always contains several sampling & preparation steps – usually primary sampling dominates!
(55%)2 + (35%) 2 + (1.5%)2 (65%)2
Primary sampling representativity: - a key issue in all sampling situations
Which is always followed by: secondary sampling / mass-reduction/ handling & sample preparation(s) … Cannot be neglected either
![Page 17: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/17.jpg)
0 10 20 30 40 50 60 70 80 90 1000
5
10
15
20
25
30
35
40
45
50
The arch-enemy: HETEROGENEITY!
Identical issue – at all scales !!!
![Page 18: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/18.jpg)
0 10 20 30 40 50 60 70 80 90 1000
5
10
15
20
25
30
35
40
45
50
The arch-enemy: HETEROGENEITY!
Identical issue – at all scales !!!
![Page 19: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/19.jpg)
25% 1% 10%
1000 ppm 100 ppm 500 ppm
Analyte concentrations: 25% .... 1% .... 100 ppm
![Page 20: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/20.jpg)
CHL and DHL constutes the operative handles on heterogeneity – allowing to treat all types of lots (materials /forms/compositions in a complete unified manner. The general rules are codified in TOS.
![Page 21: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/21.jpg)
![Page 22: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/22.jpg)
Fundamental Sampling Principle (FSP):
All increments must have the same (non-zero) probability of ending up in the sample
Spear sampler core
Lot (3-D)
Heterogeneity (hidden)
FSP
![Page 23: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/23.jpg)
Courtesy: HERZOG Maschinenfabrik
![Page 24: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/24.jpg)
Courtesy: HERZOG Maschinenfabrik
![Page 25: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/25.jpg)
Sampling in the laboratory: What’s the difference w.r.t. the field/plant etc?
….. ONLY the SCALE ...... !!!
Sampling at all scales - N.B. Identical problems N.B.
![Page 26: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/26.jpg)
1. FSP: Fundamental Sampling Principle 2. SSI: Sampling Scale Invariance 3. PSC: Sampling Correctness (bias-free sampling) 4. PSS: Sampling Simplicity (primary sampling + mass-reduction) 5. LDT: Lot Dimensionality Transformation 6. LHC: Lot Heterogeneity Characterization (0-D, 1-D) 7. SUO: Composite Sampling 8. SUO: Comminution 9. SUO: Mixing / Blending 10. SUO: Representative Mass Reduction (Sub-sampling)
Representative Sampling: Theory of Sampling (TOS)
Governing Principles (sampling in practise)
![Page 27: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/27.jpg)
- three sources of errors
”Correct” sampling errors
”Incorrect” sampling errors
Pierre Gy
![Page 28: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/28.jpg)
![Page 30: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/30.jpg)
1. FSP: Fundamental Sampling Principle 2. SSI: Sampling Scale Invariance 3. PSC: Sampling Correctness (bias-free sampling) 4. PSS: Sampling Simplicity (primary sampling + mass-reduction) 5. LDT: Lot Dimensionality Transformation 6. LHC: Lot Heterogeneity Characterization (0-D, 1-D) 7. SUO: Composite Sampling 8. SUO: Comminution 9. SUO: Mixing / Blending 10. SUO: Representative Mass Reduction (sub-sampling/splitting)
Representative Sampling: Theory of Sampling (TOS)
Governing principles (GP) & Sampling Unit Operations (SUO)
![Page 31: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/31.jpg)
Four practical Sampling Unit Operations (SUO) 1. Composite Sampling 2. Mixing / blending 3. Particle Size Reduction (comminution) 4. Representative Mass Reduction (- sample preparation)
Used as active steps in the sampling process (often used several times, in combination)
Theory & Practise of Sampling (TOS): Practical procedures … … 4 Sampling Unit Operations (SUO):
![Page 32: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/32.jpg)
An analogy:
Maxwell's Equations describe the world of electromagnetics. The four equations describe how electric and magnetic fields propagate, interact, and how they are influenced by objects.
![Page 33: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/33.jpg)
1. FSP: Fundamental Se Invariance 2. PSC: Sampling Correctness (bias-free sampling) 3. PSS: Sampling Simplicityampling Principle 4. SSI: Sampling Scal (primary sampling + mass-reduction) 5. LDT: Lot Dimensionality Transformation 6. LHC: Lot Heterogeneity Characterization (0-D, 1-D) 7. SUO: Composite Sampling 8. SUO: Comminution 9. SUO: Mixing / Blending 10. SUO: Representative Mass Reduction (sub-sampling/splitting)
Governing principles (GP) & Sampling Unit Operations (SUO)
TOS’ six Governing Principles describe how to conduct representative sampling of heterogeneous materials. The four SUO are the only active agents at disposition.
All GP’s & SUO’s are not involved in all sampling tasks. The analysis & the sampling objective determines which GP’s and SUO’s to use. The Theory of Sampling (TOS) to the fore ... DS 3077: First horisontal standard (2013).
![Page 34: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/34.jpg)
More to the Principle of Simplicity
Focus on representative mass-reduction ...
1. Sampling is never a one-shot operation 2. Sampling is always a multi-stage process: i) Primary sampling + ii) Representative Mass-reduction
![Page 35: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/35.jpg)
Representative mass reduction Device and method examples:
Grab Sampling
Fractional shoveling
Alternate shoveling
Spoon method
Rotational dividers Boerner
divider
Riffle splitters
Representative Mass Reduction
![Page 36: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/36.jpg)
REPRESENTATIVE MASS REDUCTION IN SAMPLING - a critical review of techniques and hardware
Lars Petersen, Casper K. Dahl & Kim H. Esbensen
Chemometrics & Intelligent Laboratory Systems, 74 (2004) 95-114
![Page 37: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/37.jpg)
1. FSP: Fundamental Sampling Principle 2. SSI: Sampling Scale Invariance 3. PSC: Sampling Correctness (bias-free sampling) 4. PSS: Sampling Simplicity (primary sampling + mass-reduction) 5. LDT: Lot Dimensionality Transformation 6. LHC: Lot Heterogeneity Characterization (0-D, 1-D) 7. SUO: Composite Sampling 8. SUO: Comminution 9. SUO: Mixing / Blending 10. SUO: Representative Mass Reduction (Sub-sampling)
Representative Sampling: Theory of Sampling (TOS)
Governing Principles (sampling in practise)
![Page 38: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/38.jpg)
Incorrect sampling errors (ICS)
IDE
IEE
IPE
IWE
FSE
GSE
Correct sampling errors (CSE) Process sampling errors (PSE)
PIE2
PIE3
PIE1
Theory of Sampling (TOS)
Measurement Uncertainty (MU)
Theory of Sampling (TOS)
Total Analytical Error (TAE) Total Sampling Error (TSE)
![Page 39: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/39.jpg)
Incorrect sampling errors (ICS)
IDE
IEE
IPE
IWE
FSE
GSE
Correct sampling errors (CSE) Process sampling errors (PSE)
PIE2
PIE3
PIE1
Theory of Sampling (TOS)
Measurement Uncertainty (MU)
![Page 40: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/40.jpg)
Important terminology and understanding for analysis - and for sampling_plus_analysis
Sampling process: BIAS + imprecision (TOS: a varying, an inconstant BIAS)!
Analytical process: BIAS + imprecision (statistical concept: a constant BIAS)
![Page 41: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/41.jpg)
The Practical Summary:
A set of only 6 + 4 unit operations can be used to handle (allmost) all sampling problems, always respecting TOS
Usually only a sub-set of the 10 SUO’s is necessary It is strongly recommended to become “TOS-literate” …
10 SUO’s can be implemented for all 3-dim & 1-dim sampling – always …
10 SUO’s can usually be implemented for 2-dim sampling (creative experience)
10 SUO’s can only be implemented for 3-dim sampling with great difficulty …
Representative Sampling: Theory of Sampling (TOS)
![Page 43: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/43.jpg)
Low priority problems
Top priority problems
Problems that are not worth
too much attention
Problems to be Solved after Performing a
Feasibility study
Effect of problem
Cost of fixing problem
Courtesy: Francis Pitard Sampling Consultants
![Page 44: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/44.jpg)
(trends)
Causes: sampling variation
FSE GSE IME IPE
IWE TAE
Bias Test(s) Round Robin test(s) “Correction factors”
“Effects estimation”
IDE (cyclicity)
Effect of problem
Cost of fixing problem
Note: Elimination of (all) Incorrect Sampling Errors (ISE) TOS’ Preventive Paradigm (TPP) Principle of Sampling Correctness (PSC)
![Page 45: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/45.jpg)
References Esbensen, K.H. & Wagner, C. (2014) Why we need the Theory of Sampling. The Analytical Scientist, vol 21 (Oct. 2014) pp. 30-38* DS 3077 (9 page free preview)* http://webshop.ds.dk/product/M278012/standard.aspx http://www.impublications.com/tos-forum http://www.spectroscopyeurope.com/articles/sampling *) Attachments to conference handouts
![Page 46: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/46.jpg)
Never hesitate to contact experts, colleagues, friends re. TOS
![Page 48: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/48.jpg)
TOS – the critical success factor ... Automatic sampling Sample transport to the laboratory Sample handling Laboratory control system Sample preparation Automation of analysis
![Page 49: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/49.jpg)
![Page 50: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/50.jpg)
![Page 51: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/51.jpg)
![Page 52: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/52.jpg)
![Page 53: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/53.jpg)
![Page 54: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/54.jpg)
1. FSP: Fundamental Sampling Principle 2. SSI: Sampling Scale Invariance 3. PSC: Sampling Correctness (bias-free sampling) 4. PSS: Sampling Simplicity (primary sampling + mass-reduction) 5. LDT: Lot Dimensionality Transformation 6. LHC: Lot Heterogeneity Characterization (3-D, 1-D) 7. SUO: Composite Sampling 8. SUO: Comminution 9. SUO: Mixing / Blending 10. SUO: Representative Mass Reduction (Sub-sampling)
Representative Sampling: Theory of Sampling (TOS)
Governing Principles (sampling in practise)
![Page 55: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/55.jpg)
GP no 6: Heterogeneity characterisation
![Page 56: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/56.jpg)
No Name Inc
Lesson learned: There is no such thing as one approach that ”fits-all-sizes” (i.e. covers all types of heterogeneity). !! Heterogeneity characterisation essential !!
![Page 57: Sampling: Theory and Practice](https://reader035.vdocument.in/reader035/viewer/2022070320/62bfcf31294b8a46f82127dc/html5/thumbnails/57.jpg)
Flow
Flow
Flow