KHE Consulting

The universal power of the

Theory of Sampling (TOS)
Why is sampling the critical success factor before analysis?
 for the seller; for the buyer ; for the middleman, for the arbiter ?
 for the company; for the customer; for the scientist; for the technician?
 for science, technology, industry; for compliance; for safety; for society?

Keynote copyright: KHE Consulting

 KHE Consulting

Theory and Practice of TOS

Representative Sampling
A minimum understanding of governing principles and sampling unit operations
- all types of materials (all degrees of heterogeneity: low – intermediate – high
- at all scales (for all lot sizes: small – intermediate – big – extreme)
- unifying principles of representative sampling: field/plant/laboratory/process

Keynote copyright: KHE Consulting

Kim H. Esbensen - who / what / when ….

Geological Surveys of Denmark & Greenland (GEUS)

research professor (chemometrics and sampling) 2010

Aalborg University, prof. (chemometrics & sampling) 2001

5
01
Telemark University of Process Technol. (HIT), prof. 1991

-2
79
Norwegian Computing Center (NCC) & SINTEF 1985
19
Terra Swede (exploration) 1982

Technical University of Denmark (DTH), Ph.D. 1981

Århus University, Denmark: M.Sc. (geology), 1979

 1991

1981

19
8 0
-2
01
5

2001

2011
1980 – 2015: professor (3 universities/gov. R&D institutions)

2015  consultant, independent researcher

(assoc., guest and affiliated professor at 4 universities)

[email protected]
 How to take a
representative sample
of all this XXXX .... ?

Well, one fine day I

will know ... hopefully!
Lappeenranta University of Technology (LUT) - 1999

2000  Fast forward ...

Theory of Sampling (TOS)
– everything in a glance
The
anal
ytica
l lab
orato
ry
Incorrect sampling errors (ICS)

IDE
Measurement Uncertainty (MU)
IWE

IEE

IPE
”an
ya
naly
tica
Theory of Sampling (TOS) lm
o da
lity
”
FSE PIE2

PIE1

GSE PIE3

Correct sampling errors (CSE) Process sampling errors (PSE)

MUtotal = MUsampling + MUanalysis

Representative Sampling: Theory of Sampling (TOS)
TOS - Axiomatic exposé

Governing principles (GP) – Sampling unit operations (SUO)

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)
Composite sampling employs Q increment extractions
with the aim to ‘cover’ the lot volume (only Q = 4
increments shown in this principal illustration).
Proportional to the heterogeneity encountered, a higher
number of increments will be required. Comp samp.
must always respect FSP !!!

GP (6) Lot Heterogeneity Characterisation (LHC)

guarantees that no sampling plan, sampling procedure
nor sampling equipment is employed without a
mandatory heterogeneity characterisation of the lot
material.

Composite sampling is specifically demanding that grab

sampling (extraction of one single increment only) is
never invoked, unless thoroughly tested and accepted
by either a Replication Experiment (RE) or by
variographics.
Crushing (comminution) is a sampling unit operation
which is only brought to bear when necessary, i.e. when
the top particle size is contrasting too much with
respect to smaller size ranges in order for sampling to
be effective and representative. Comminution is the
technical process in which the top particle sizes is
preferentially crushed first.

A consequence of crushing/comminution is that the

majority of particle sizes tend to become more similar,
with the further advantage that mixing becomes more
effective.

Maceration, crushing or shredding in the presense of a

facilitating liquid (often used for selective extraction), as
applied to biological materials also lead to reduced
general particle sizes.
Mixing is a forced mechanical process designed to
reduce the distributional heterogeneity (DH) of a
material system.

It is always advantageous to mix the results of a

sampling or a sub-sampling process before further
processing (sub-sampling or a next stage mass
reduction).

Blending is mixing under stoichiometric constraints, i.e.

the final mixing product, a blend, must satisfy
compositional constraints e.g. tea, tobacco, cement,
pharmaceutical drugs.

Mixing / blending can be applied to both polyphase dry

systems (aggregates) and to slurries (solid – liquid
systems).
Representative Mass Reduction (RMR) is the key
sampling unit operation connecting all sampling stages.
Often the terms mass reduction and sub-sampling are
used inter alia. There are very many sub-sampling
procedures and types of equipment offered on the
market, but far from all deliver representative solutions.

For stationary lots, the benchmark study by Petersen et

al. (2004) showed conslusively that only the riffle-
splitting principle lead to Representative Mass
Reduction (RMR). Riffle splitters have different physical
manifestations; both stationary and roraty solutions
exist.

For dynamic lots, lots in movement, the Vezin sampler

is by far the most effective, fully representative RMR
equipment in existence. The Vezin sampler is also
superior regarding slurries a.o.
Representative Sampling: Theory of Sampling (TOS)
TOS - Axiomatic exposé

Governing principles (GP) – Sampling unit operations (SUO)

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)
Sampling Unit Operations: Composite Sampling

”Grab Sampling” - ”not thinking”

… … vs. Composite Sampling … …

 51.02%

RSV 07%

47.62%

a m p ling
p osite s
t co m
n
-in c reme
42

5.95%

RSV 17%

4.96%
 Replication Experiment (10-
Concentration
fold) of grab vs. composite
sampling
Single grab sampling variance

Composite sampling variance

aL

True lot concentration

Proper sub-sampling … proper ??
Probably the world’s worst manual ”composite sample”
It so easy to do it WRONG – and so easy to do it RIGHT (representatively)

Teaching TOS by analogy ...

Model photo: with permission

Representative Sampling: Theory of Sampling (TOS)
TOS - Axiomatic exposé

Governing principles (GP) – Sampling unit operations (SUO)

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)
TOS - Axiomatic exposé

Governing principles (GP) – Sampling unit operations (SUO)

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)
 A grab sample ... Fundamental Sampling Error (FSE) [+ GSE]
Several increments ... ... Trying (very well) to deal with (GSE) [+ FSE]

Heterogeneity (hidden)

Lot (1-, 2 & 3-D)

Fundamental Sampling Principle (FSP):

All increments must have the same (non-zero)
probability of ending up in the sample (non-neg)
FSP
 A dve r
se cha
Heter racter
og e ne istic:
ity Sampling rate
(typical)

103 – 106

Primary sample

101 – 102
Secondary sample

101 – 102
Tertiary sample / aliquot

103 –3 106 -6 109 9

Traditional laboratory domain 10 –3 10 -6 10 9
10 – 10 - 10
This is too Well, well ...
expensive!
Fundamental insights - I

 The empowering role of universal principles & SUO’s

 Governing principles (GP) & Sampling Unit Operations (SUO))

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)

TOS’ six &

All GP’s Governing
SUO’s are Principles
not involveddescribe how to conduct
in all sampling tasks.
representative sampling of heterogeneous materials.
The analysis & the sampling objective determines which
The
GP’sfour
andSUO’s
SUO’s are the only
to use. active agents
The Theory at disposition.
of Sampling (TOS) to
the fore ... DS 3077: First horisontal standard (2013)
www.ds.dk
DS-3077
 TOS’ six Governing Principles describe how to conduct
An important analogy
representative sampling of heterogeneous materials.
(for some ..)
The four SUO’s are the only active agents at disposition.

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
Fundamental insights - II

 The role of statistics – in sampling and analysis

 - very difficult to avoid conventional
-statistical
because population thinking
the analytical … depend
results … on
the sampling procedures used .... !!! !!!

xāavr

Empirical distribution of anal. results

Heterogeneity is different in nature ..... (this is difficult – at first)

Heterogeneity is different in nature ..... (this is difficult – at first)

The lot: a population of anal. results

The
anal
ytica
l lab
orato
ry
 Theory of Sampling (TOS)
All four– everything
SUO’s inanalytical
in action in the a glance laboratory ...
Work effort considerations vs. REPRESENTATIVITY

There is no case – Representativity must come first – always!

12 kg – fully crushed (TOS)

compared to
20 g (grab) – ratio 600: 1
 Representative mass reduction ---- ?? Alternate
shoveling

Rotational
dividers

Boerner
divider

Fractional shoveling

Grab Sampling
Riffle splitting
The “spoon method”
The ”Japanese slab cake” approach:
The ”Spoon method”
Open splitter

Closed splitter
A pro pos ”rotating splitters ...”

”subsequent scraping off ...???”

”radial segment”
Coning & quartering is a popular mass reduction approach
A hidden elephant in the room ....

VERY in-effcient mixing is the real reason that coning &

quartering is one of the worst mass reduction approaches
 ”Sampling Hall of Shame – BIG TIME
”Coning & Quatering”

An often overlooked factor: HIGHLY

INSUFFICIENT lot pre-MIXING !!!
 Representativeness (pooled sum for wheat, rape seed and glass)

0,600

0,500

0,400
Representativeness

0,300

0,200

0,100

0,000

g
in
t er r r er al g
ng 5

te r
or de e 1: 5 d g g lin
Lo id i itt itt m m m 1: m ho lin lin
Sh iv iv Sp
l + l m or m + m e t e e p
Sp m

ar
6 0 0 v v
34 34 D rD ed 1:
4
f1 N f2 1:
2
f3 M ho ho Sa
32 d 4 n

qu
RK RK ne e i o
Fe
e
es
o 3
es
o +
es
o o S
al
S
ra
b
er S r
RK 2 o te
Bo Va al ut ut 1: ut Sp na on G

&
m h h o h r t i
ni C C ri C lte ac
18 10 Va 10 Fr

ng
A A
RK RK RK

ni
Co
 - the analytical result depends on the sampling procedures used ....

NB. Lot sampling & subsampling error can be

up to 10-25-50 times higher (dep. heterog.)

GMO lab-sampling and

analysis error ~15-20%
Theory of Sampling
Is there an(TOS) in to
alterntive the laboratory
spatula sampling?
Theory of Sampling (TOS) in the laboratory
Not just sampling – but HOW TO sample??

??
?
TOS in the analytical laboratory

The ”Ingamells splitter”

 The an
alytical
laborat
ory
Primary sampling stage ... Secondary ... Tertiary stage ... ...

Primary sampling error ...

10 X

Secondary sampling error ...

10 X

10 X
Tertiary sampling error ...

Total analytical error ... f.ex. 1% (rel)

 Lot Lot

Primary sampling Primary sampling

Secondary sampling Secondary sampling

Tertiary sampling Tertiary sampling

Laboratory B Laboratory A

Laboratory B - the client’s only guarantee

against non-representative analytical results
 Buyer Seller
vs.
Client Laboratory

Representative sampling

Contractual uncertainty interval

Non-representative sampling
Non-representative sampling
Jazz Hall of Fame
(Downbeat)

Baseball Hall of Fame

Sampling Hall of Fame Sampling Hall of Shame

”Sample” extraction valves
 ”On-line” mass reduction equipment
”Inverse flap valve” configuration

Primary sampling collector bin

Flow conduit, toggling

Reject outlet i.e.

process stream
Sample outlet i.e.
mass reduction
It so easy to do it WRONG – and so easy to do it RIGHT (representatively)

Model photo – with permission

All delegates to SAMPLING 2018, Lima have recieved a digital copy of

TOS Forum, Issue 8

Teaching by analogy or example

 Barefeet sampling – can be fit-for-purpose

 Process engineer: NEED a sample !!

Lightweight Expanded Clay (LECA) production kiln: ~975 C

Sampling Hall of Fame
TOS Forum

Free Sampling
Resources from
IM Publications

impopen.com/tosf
Spectroscopy Europe

Free Sampling
Resources from
IM Publications

spectroscopyeurope.com/sampling
WCSB7 Proceedings

Free Sampling
Resources from
IM Publications

impopen.com/wcsb7
Minerals — Open Access Journal of Mining & Mineral Processing

Special Issue "Sampling across the Mine Value Chain"

Minerals (ISSN 2075-163X; CODEN: MBSIBI) is an international peer-reviewed open

access journal of natural mineral systems, mineral resources, mining, and mineral
processing. Minerals is published monthly online by MDPI.
Open Access - free for readers, with article processing charges (APC) paid by
authors or their institutions.
High visibility: indexed by the Science Citation Index Expanded (Web of
Science), Chemical Abstracts, INSPEC and GeoRef.
Rapid publication: manuscripts are peer-reviewed and a first decision provided to
authors approximately 17 days after submission; acceptance to publication is
undertaken in 5.4 days (median values for papers published in the first six months
of 2018).
Impact Factor: 1.835 (2017) ; 5-Year Impact Factor: 2.037 (2017)
Special Issue "Sampling across the Mine Value Chain"
Special Issue Editors
Guest Editor Deadline: December 1.st 2018
Dr. Simon Dominy
Camborne School of Mines, University of Exeter, Penryn, Cornwall
TR10 9FE, UK
Website | E-Mail
Interests: economic geology; sampling/theory of sampling;
reserve estimation/evaluation; geometallurgy; mining geology;
narrow vein mining; mineral processing

Guest Editor
Prof. Hylke Glass
Camborne School of Mines, University of Exeter, Penryn, Cornwall
TR10 9FE, UK
Website | E-Mail
Paper
Interests: sampling/theory of sampling; geometallurgy;
s from S
geostatistics; sensor-based mineral sorting; selective leaching;
A MPLIN
metal bio-accessibility; post-mining revegetation and terra-
greening G 20 18 –
are we
lcome
Guest Editor in this SI
Prof. Kim Esbensen
KHE Consult, Aldersrogade 8, 2. sal, 2100 Copenhagen Ø,
Denmark / Geological Survey of Denmark and Greenland (GEUS),
Oester Voldgade 10, 1350 Copenhagen C, Denmark
Website | E-Mail
Interests: sampling/theory of sampling;
geochemistry/geoanalysis; process analytical technology;
multivariate data analysis; chemometrics
Novel application: oil pollution monitoring
The ”BIOTA GUARD” patented approach
 NIR (Near Infra-Red) sensor

Bivalve opening caliper (sensor)

PAT Sampling Hall of Fame
Biosensor time series’ (heart beat &
gap opening)
7

0
7 0 5 03 0 1 9 9 97 9 5 93 9 1 89 8 7 85 83 81 7 9 77 75 7 3 71 69 67 65 63 61 59 57 55 53 51 49 47 45
1 2 3 3 4 5 6 7 8 9 10 1 1 12 13 1 4 15 16 17 18 1 9 20 2 1 2 2 23 2 4 25 26 2 7 28 2 9 30
Biosensor time series: ”Let the good times roll ..”

2000

1500

1000

500

0
1 12 23 3 4 45 5 6 67 78 8 9 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9
1 1 1 1 1 1 1 1 1 1
-500

-1000

-1500

Measurement variable: Heart beat

Variogram characteristics – no great surprise (yet)

700.000

600.000

500.000

400.000
V(j)

300.000

200.000

100.000

0.000

Lag (j)

Measurement variable: Heart beat

Biosensor time series: ”Let the good times roll ..”
7

0
7 0 5 03 0 1 9 9 97 9 5 93 9 1 89 8 7 85 83 81 7 9 77 75 7 3 71 69 67 65 63 61 59 57 55 53 51 49 47 45
1 2 3 3 4 5 6 7 8 9 10 1 1 12 13 1 4 15 16 17 18 1 9 20 2 1 2 2 23 2 4 25 26 2 7 28 2 9 30

Measurement variable: Bivalve gap opening

Measurement series (blow-up)
Analytical result

1.7

1.7

1.7

1.7

1.6

1.6

1.6

1.6

1.6

1.5

Increment number

Measurement variable: Bivalve gap opening

Variogram characteristics – no great surprise (yet)
0.000

0.000

0.000

0.000
V(j)

0.000

0.000

0.000

0.000

0.000

Lag (j)

Measurement variable: Bivalve gap opening

Variogram characteristics – sensitivity, specificity

Sa
mp
li ng
Ha
ll o Bivalve sensors 10,000 more
fF sensitive to trace oil conc. levels
am
e
NB Dep. upon realistic empirical
seawater-oil calibration/validation!

Selective dose-response calibration

Interferrant/matrix reponse
calib. !!
Multivariate data modelling: chemometrics

X Y
PLS-regression

PLS-regression: Y-data guided projection X-space

Multivariate calibration – powerful, generalized regression:

X-data: Y-data:
Geochemistry (descriptors)
Spectra (absorption, emission Geochem. (dependent
Concentration [ppm,var.s)
%, g/ml]
Geophysics (descriptors) Petrophys. (dependent var.s)
Instrumental
Mineralogy measurements
(modal prop.s) ... ”Wet-chem.
Rock mechanicsLaboratory”
Lithology Mineralisation/ore charac’s
”Easy-to-measure”
Heavy data
mineral characteristics ”Difficult,
Grindability laborious ...” ... ...
Mineralisation/ore
Descriptors charac.s Flotability
Functional properties
…… Leach recovery
”Input data
Variogram ”Output
…… data”
Geometalurgical X data Spillage – oil concentration

Variogram Spectral data

Geometallurgy test params
Geometallurgy desc params
Teaching by analogy or example

 The historical perspective (BIG surprise)

SAMPLING OF
MATERIAL
OBJECTS
Theory and Practice
 Pierre M. GY
Paris School of Physics & Chemistry, Ph.D
Sampling Consultant & Expert
E-mail : gy @ pierregy.com
---------------
Founder & Chairman
International Sampling Institute
Institut International de L’Echantillonnage (ISI /
IIE)
E-mail : samp.echant @ wanadoo.fr
97
 FIRST PART

PLAN OF THE
COURSE
STAP 2000

HIT, Porsgrunn, Norway, November 2000

 Historical snapshot

HIT, Porsgrunn, Norway, November 2000

 FACTORS AT STAKE IN
SAMPLING
 SAMPLING is the first, the most risky,
the most neglected and ignored link of …
QUALITY CONTROL

t IN INDUSTRY and TRADE : the

factor at stake is especially the …
ECONOMICAL FACTOR

QUALITY IS MONEY
100
 SAMPLING
IN THE LITERATURE
Some pertinent, some dubious and
some rather surprising remarks …
1930 : Grummel & Dunningham :
« Those whose interest in sampling is
recent will not easily understand how
difficult it has been for the authors to
have a new idea adopted » …
This sad remark is still valid in 2000 !
101
 SAMPLING AND STANDARDS
Standards are usually naïve and written
by unqualified people without any regard
to scientific considerations.
Example of the sampling scoop (ISO) :
the handle length alone is optional !

102
 Standards impose arbitrary rules to
users devoid of any critical formation !
103
 REPRESENTATIVITY
All analytical standards state that assays
must be carried out on …
« REPRESENTATIVE SAMPLES »
but these standards …
 Fail to give any scientific definition of a
« Representative sample »,
 Fail to say « what should or should not
be done to obtain one ».
 A scientific definition of a Representative
sample is given in Part 3 « Definitions ».
104
 PHILOSOPHY OF
STANDARDIZATION
 According to an ISO officer, the role of the
technical committees, ISO/TC, should be « to
describe the practices on which trade has
been based for a long time ».
 No qualification required from TC members
 Decisions are taken after a vote !
 Practically no standard on sampling is
based on scientific considerations. This
dubious philosophy encourages lobbying.
105
 ROLE OF UNIVERSITY AND
STANDARDS ORGANIZATIONS
 This role should be major BUT in 2000
University and Standardization go on ignoring
the mere existence of a theory which has
NEVER been contested by any- one.
Exceptions FINLAND and NORWAY
 Equipment manufacturers follow stan-
dards that their clients are liable to know.
 Lobbying in ISO Technical Committees !
106
 NINTH PART

BED-BLENDING
Theory and Practice
 PURPOSE AND PRINCIPLE OF
BED-BLENDING
Many plants or devices operate much more
efficiently or safely when they are fed with a
material of quality « as uniform as possible ».
This point has been first understood by Lafarge
Cements due to the fact that cement kilns are
dangerously sensitive to quality fluctuations of
their feed. So are metallurgical furnaces.
Most transformation processes would benefit
from a uniform feed. Bed-blending is the key.
108
 The output of the proportioning system has an
AVERAGE composition P near the REQUIRED
composition C but is absolutely NOT UNIFORM.
Its variability is characterized by its variogram…
Variograms for CaO Actual variogram of
Proportioning output

Heming Cement Works (France)

Flat variogram characterizing a uniform material

v(0) > 0
Required variogram of the Bed-blending output

The purpose of Bed-Blending is to transform

the RED variogram into a BLUE one (flat). 10
9
 BED-BLENDING SYSTEM
TwoSTAGE
FROM PROPORTIONING alternating
phases
Bed-Blending
INPUT One way Reclaimer Travel /
speed

STACKER 2 cm/mn
Harrow-type
Reclaimer
Stacker
reciprocating travel 30 m/mn
speed Main Reclaiming
Belt

PHASE 1 t PHASE 2 t PILE

Phase 1
PILE At
STACKING
Phase 2
B t RECLAIMING Paddle-chain
Conveyor

UNIFORM COMPOSITION BED-BLENDING OUTPUT 11

 BED-BLENDING t STACKING
Stacker speed along the pile  20 / 30 m/mn
Photos, courtesy of BMH, Mulhouse, France
Feeding belt Stacker Pile being built

Boom Falling
Material

Reclaimer in idle position

11
 BLENDING THEORY t STACKING
Stacker Reciprocating stacker travel

Stacker speed  30 m/mn

Future Layer 8
Future Layer 7
Future Layer 6
Future Layer 5
Layer 4
Layer 3 DEAD
Layer 2 STOCK
Layer 1

0 Stacking time of one layer T0

Vertical Cross-section of the pile 11
 BLENDING THEORY t RECLAIMING
Many types of reclaiming devices. After
studying the performances of all of them,
the « Harrow-type Scraper Reclaimer »
appears as the MOST EFFICIENT and
can be used as a MODEL FOR THE
THEORY OF BED-BLENDING.
Its major property is to reclaim a THIN
SLICE of the whole pile cross-section
SIMULTANEOUSLY (e.g.  2 cm/mn).
113
BLENDING THEORY t RECLAIMING
Harrow-type Slice S reclaimed between t and t + Dt. This slice may
Reclaimer be regarded as a … Sample S made
of 8 « increments »
Scraping
cycle ½ to
1 mn re 8
7
o
efb

Pikes 6
e t ed
tim aim

5
cl

4
re

DEAD
e

3
m
lu

2
STOCK
Vo

1
t t + Dt Reclaimer speed  2 cm/mn

Paddle-chain conveyor TO Main reclaiming belt

114
 BED-BLENDING t RECLAIMING
Harrow-type Reclaimer-Scraper in idle position
BMH – Mulhouse, France t Side view
Natural angle of repose of the Jig-saw transversal motion Period 
reclaimed material. Can / must be ½ to 1 mn
adjusted.

Paddle-chain conveyor. To
main reclaiming conveyor (in a
trench – not shown here)

Reclaims a thin slice (0.02 – 0.05 m/mn) 115

PROBLEM POSED BY THE PILE ENDS
Experiment carried out by BMH, Mulhouse, France

Uniform Composition required

15 to 20 % of the pile must be recycled

HAS BEEN IMPROVED BY MANUFACTURERS

Hours
116
 VARIOGRAPHIC ANALYSIS
OF A BED-BLENDING UNIT
The variogram is a mathematical tool that charac-
terizes the variability of a one-dimensional flow.
To figure out the efficiency of a bed-blending unit
we compare the INPUT and OUTPUT variograms.
Variograms for CaO Heming Cement Works (France)

Input Sill
Input Interval = 90 mn

INPUT Variogram
Input
Intercept OUTPUT Variogram
vI(0) Output Interval = 120 mn

Output Intercept and Sill 18 hours 11

7
 UNEXPLORED POSSIBILITIES
t SCALING DOWN : So far, bed-blending has
been implemented only on the scale of heavy
industries such as Cement or Metallurgy. But it
can very well be scaled down from hundreds of
tons to hundreds of kg per hour.
t GENERALIZATION : Many industries would
increase the quality of their products and their
profits if they were fed with a uniform material.
t EVEN IN HEAVY INDUSTRIES, the best way
to operate a bed-blending system is NOT
PROPERLY UNDERSTOOD. 118