MAD 347

Process Mineralogy
Lecture II-Sampling

Prof. Dr. İlkay Bengü CAN

Sampling is;

“taking small amounts from the

bulk material which represent
the whole”

İlkay B. CAN, Hacettepe University

Sampling;

Seems very easy but it is extremely

difficult!

Requires experience and expertise...

İlkay B. CAN, Hacettepe University

Aim 1
The samples taken from an ore deposit;

•can describe the ore deposit and mineral accumulation,

•can prepare the required data for plant design

These samples can be taken from drill cores, the galleries

inside deposit and the surface

İlkay B. CAN, Hacettepe University

What kind of information is provided from
samples?
By the tests performed on the samples taken from the ore
deposit;

• The minerals constitute the sample,

• The particle-grain size,
• The amount of these minerals,
• Textures and liberation or locking type of these minerals,
• The amount of alteration and clay content,
• Metal content,
• All details about possible concentration method,
• Moisture content

İlkay B. CAN, Hacettepe University

Aim 2

The samples taken from an existing plant;

•can be used for performance evaluation of a single equipment

or the whole plant,

•can be used for examining the properties of product and tails

(for environment consideration)

İlkay B. CAN, Hacettepe University

What kind of information is provided from
samples?

By the tests performed on the samples taken from

the plant;

• Particle size distributions,

• Moisture content,
• % solid content of pulp samples,
• Chemical composition,
• Mineralogical data on size fraction basis,

İlkay B. CAN, Hacettepe University

Why sampling is so important?
Because by sampling...
• Quantitative information is provided
• Metallurgical balance is improved
• Quality and process control is improved
• Resource utilization can be optimized...

No point in accurate analysis if sample is not

representative in the first place !!!

İlkay B. CAN, Hacettepe University

Sampling Methods

Sampling from the ore deposit

Grab samples (by hand)

Surface samples

Deep samples

İlkay B. CAN, Hacettepe University

Sampling from ore deposit

~40-50 tons of material should be digged by shovel from

every pit. Shovel should take rocks from 3 or 5 points from
the surface by scanning from bottom to top.

if deposit is horizantal if deposit is inclined

İlkay B. CAN, Hacettepe University

 Coarse Material Problem

These big lumps can be crushed down by a crusher in

the pit and then its amount can be reduced

İlkay B. CAN, Hacettepe University

Sampling period from ore deposit

İlkay B. CAN, Hacettepe University

Sampling from a stockpile

fines

medium

Coarse

İlkay B. CAN, Hacettepe University

Sampling from the Plant
➢ From the trucks, belt conveyors, and railway cars which carrying ore to
stockpiles,

➢ From stockpiles,

➢ From crushing stage; from feeding and discharging points of every crusher
and mills,

➢ From feed and products of screens and classifiers,

➢ From feed and products of washing equipment,

➢ From feeding and discharging points of dewatering

➢ equipment,

➢ From feeding and discharging points of drying

➢ equipment,

İlkay B. CAN, Hacettepe University

Sampling from Packaged Material

If material is in a bag or in a plastic

container sampling tubes can be used in
ramdom selected of bags/containers.

İlkay B. CAN, Hacettepe University

 Sampling from Bucket Elevators

1 bucket or more buckets

selected in a specific
range can be taken as
samples.

İlkay B. CAN, Hacettepe University

Sampling Methods from Conveyor Belts and Flowing
Streams

İlkay B. CAN, Hacettepe University

İlkay B. CAN, Hacettepe University
Example of linear sampler for solids: (a) Conveyor transfer/discharge
point, and (b) Vertically discharging slurry pipe (Courtesy Heath and
Sherwood).
İlkay B. CAN, Hacettepe University
Example of pressure pipe sampler

Example of a two-cutter Vezin sampler showing the

sector design of the cutters (Courtesy Heath and
Sherwood).
İlkay B. CAN, Hacettepe University
Example of gravity slot sampler

a) Example of a moving inlet sampler, and

b) Interior view of slurry flow and cutter
(Courtesy Heath and Sherwood).

İlkay B. CAN, Hacettepe University

Sampling from Pulp
Samples are taken by a syringe from a pulp stirred in a tank
(Conditioner, Flotation Cell etc.).

Wire for pulling up

the lid for opening

Handle for submerging the

Syringe
box into pulp
Lid conntected to box by
string
Sampling box

İlkay B. CAN, Hacettepe University

Cutter used in pulp sampling

Dimensions
in mm

Flowing pulp
Cutter
movement

İlkay B. CAN, Hacettepe University

 Sampler (scoop) used for pulp flowing in open
conduits (kanal)

Such kind of scoops are not used for cyclone/mill

discharge as the flows may hit severely!

İlkay B. CAN, Hacettepe University

Plant Sampling

• The discharge point of a conveyor belt or

process stream is the most suitable
sampling location

• Must take a complete cross section of the

ore or concentrate stream

• Sampling devices that take only part of

the stream must be avoided at all cost

İlkay B. CAN, Hacettepe University

Correct Cutting of Stream

Cutter Direction
t

Stream

İlkay B. CAN, Hacettepe University

Assay Sampling (for Chemical Analysis)
• A small amount of material which is taken from the main
bulk must be representative of that larger amount. Very
small sample has great responsibility

• The max. particle size of the processes are important!

For instance, sampling from the ground ore pulp is

easy, and it gives more accurate results than the feed to
the primary crusher.

İlkay B. CAN, Hacettepe University

Moisture Sampling
• All metallurgical accounting requires accurate knowledge of
the dry weights of solids handled.

• In practice grab sampling is used for moisture determination.

• Grab samples for moisture determination are frequently

taken from the end of a conveyor belt after material has
passed over the weighing device. The samples are
immediately weighed wet, dried at a suitable temperature
(not above 105ºC) until all hygroscopic water is driven off,
and then weighed again. The difference in weight represents
moisture and is expressed as:

𝑤𝑒𝑡 𝑤𝑒𝑖𝑔ℎ𝑡−𝑑𝑟𝑦 𝑤𝑒𝑖𝑔ℎ𝑡

• % moisture= ×100
𝑤𝑒𝑡 𝑤𝑒𝑖𝑔ℎ𝑡

İlkay B. CAN, Hacettepe University

İlkay B. CAN, Hacettepe University
İlkay B. CAN, Hacettepe University
İlkay B. CAN, Hacettepe University
İlkay B. CAN, Hacettepe University
İlkay B. CAN, Hacettepe University
İlkay B. CAN, Hacettepe University
Sampling is a hard job!
Before After

İlkay B. CAN, Hacettepe University

Design Criteria for Sample Cutters
It is important that:
➢ The face of the collecting device or cutter is presented at right angles to
the stream. The cutter must collect a complete cross-section of the stream
without sample loss. Extraneous materials must not enter the sample cutter.
➢ The cutter covers the whole stream.
➢ The cutter must travel through the stream at uniform speed
➢ The cutter is large enough to pass the sample.
➢ The sample cutter must be non-restrictive and self-clearing, discharging
each increment without any reflux, overflow
➢ The cutting time at each point in the stream must be equal
➢ The cutter aperture must be at least 3d, where d is the particle size, with a
minimum of 30 mm for dry materials

İlkay B. CAN, Hacettepe University

Losing Material

İlkay B. CAN, Hacettepe University

Design Criteria for Sample Cutters (cont.)

• For wet materials, a minimum cutter aperture of 50 mm

is recommended to stop bridging

• The cutter capacity must be sufficient to accommodate

the entire increment

• Ore removed by belt scrapers should be intersected by

the sample cutter

• When sampling slurries, loss of sample due to dribbles

must be prevented

İlkay B. CAN, Hacettepe University

Slurry Sampling from Pipe

dribble

İlkay B. CAN, Hacettepe University

Design Criteria for Sample Cutters (cont.)

• The cutter speed should not exceed 0.6 m/s for a

cutter aperture of 3d
• If the cutter aperture (w) exceeds 3d, the cutter
speed (vc) can be increased as follows:
w
vC = 0.3 ( 1 + )
3d
• The recommended maximum cutter speed is 1.5
m/s

İlkay B. CAN, Hacettepe University

 Sampling Schemes

• Systematic sampling – increments (samples) are taken

at the same points in a determined time period

• Random Sampling – increments are taken at

completely random points and times

• Stratified random sampling – increments are taken at

random points but in equal time periods or equal
amounts

İlkay B. CAN, Hacettepe University

Mass-basis and Time-basis Sampling

• Mass-basis sampling
• Take increments within strata of fixed tonnage
• Extract increments of almost uniform mass
• Variable speed cutter preferred

• Time-basis sampling
• Take increments within strata of fixed time
• Extract increments of mass proportional to the ore
or concentrate flow rate
• Fixed speed cutter

İlkay B. CAN, Hacettepe University

 CAUTION !

➢ More amount of samples (more number of

samples) increase the reliability.

➢ Even the amount of sample is high, reliability is

not guaranteed if only 1 sample is taken

➢ If the sample does not represent the whole, the

correct analysis does not make any sense!

İlkay B. CAN, Hacettepe University

Fundamental Rule for Correct Sampling
• All parts of the ore or concentrate being
sampled must have an equal probability of being
collected and becoming part of the final sample
for analysis

• If this rule is not respected, then bias is easily

introduced, and the sample is not representative

İlkay B. CAN, Hacettepe University

Precision, Bias and Accuracy

Estimates of D 50 Precision
D 50
Accuracy Bias
True value of D 50

Precision (Hassasiyet) A measure of the random variations

between replicate measurements

Bias (Yanlılık) The difference between the average of a series of

measurements and the true value

Accuracy (Doğruluk) A measure of the closeness of agreement

between a measurement and its true value

İlkay B. CAN, Hacettepe University

 Accuracy and Precision
Feed Conveyor
• Take samples at regular intervals
around the stack
• Find feed size distribution
• Is this sampling true??

✓ Precision can be increased by

taking more samples and bigger
samples
✓ But surface of a stack is coarser X X X X X X X X X X

than the whole stack

Sample points
✓ Hence just get a better estimate
of the wrong answer! Consider sampling of
Bias is a very serious problem a stack of ore

İlkay B. CAN, Hacettepe University

Minimizing Bias
• More important to minimize bias than improve
precision, as bias cannot be reduced by replicate
measurements

Parameters that can be eliminated:

• Sample spillage and contamination
• Incorrect extraction of increments

Parameters that need to be minimized:

• Dust loss and particle degradation
• Changes in moisture content

İlkay B. CAN, Hacettepe University

 Number of the samples is dependent to
the heterogenity of the ore.

More heterogenous is the ore, more

samples are necessary!

The number of the samples are also related to these

parameters:

o Statement of the ore (eg: moving stream or stockpile),

o Size of the ore,
o Required precision
o The total amount of ore.

İlkay B. CAN, Hacettepe University

Sample Division Methods
Samples can be cut by either one of two methods; hand sampling and
machine sampling.
Hand sampling is carried out by the following methods:
• Grab Sampling: Applicable to any type of granular materials, ores,
concentrates in railway cars, stockpiles, trucks.
• Pipe Sampling: Samples are cut by inserting a pointed slotted pipe into
the pile, rotating the pipe to cut out a sample and removing some. The
method is applied to fine granular materials, wet or dry, stockpiles, bins,
railway cars or trucks.
• Shovel Sampling: During manual transfer of material of material each
shovel at a specified interval (2nd, 5th, 10th, 20th) is removed as a sample.
The method is best applied to fine granular solids.
• Coning & Quartering: Material is first carefully piled in a cone and then
flattened, is divided into four quarters. Two opposite quarters are
rejected and the remaining quarters are kept as sample for further
coning and division to eventually produce a final sample.

İlkay B. CAN, Hacettepe University

Sample Division Methods
Machine or automatic sampling is done by a mechanically driven sampling
cutter. Most automatic samplers operate by moving a collecting device
through the material as it falls from a conveyor or a pipe.

Linear traversing sampler

İlkay B. CAN, Hacettepe University
Increment (Sample) Mass
• Cutter aperture (A) and cutter speed (vC) determine
the increment mass mI (kg) taken by a cross-stream
cutter
• If the stream flow rate is G (t/h), the increment mass
is given by:

GA
mI =
3 . 6 vC

• A useful test of cutter operation is to compare actual

and calculated masses

İlkay B. CAN, Hacettepe University

Conning and Quartering Method

D
D/4

İlkay B. CAN, Hacettepe University

Sample Division Methods

Table sampler

Jones Riffle

İlkay B. CAN, Hacettepe University

Rotary Sampler

İlkay B. CAN, Hacettepe University

 Standard Deviation
Sampling Method
of Samples, %
Coning-Quartering 6.81

Hand sampling 5.14

Riffle 1.01

Rotary Sampler 0.125

İlkay B. CAN, Hacettepe University

The source of ERRORS in sampling

Being representative of a sample is a statictical problem as it

depends on the nature and total errors done during the sampling
studies and analysis. The source of the errors can be:
• Dynamic conditions in the plant,
• Design of the sampler,
• Reducing the amount of sample,
• Analytical errors such as wrong weighing, wrong sieve selection,
insufficient sieving time, wrong selection of the constants and
calibration errors,
• Calculation errors,
• Errors based on humans/operator.

İlkay B. CAN, Hacettepe University

Fundamental Error Variance
The fundamental error variance  2FE identified by Gy is given
by:
Cd 3 𝑒𝑟𝑟𝑜𝑟%
 2
FE = 𝜎=
MS 𝑎𝑠𝑠𝑎𝑦%

where C = the sampling constant

d = nominal top size (taken as 95%passing size) (cm)
MS = divided sample mass (g)

As ± 2 represents the probability of events when 95 out of 100 assays would

be within the true assay value, 2 is the acceptable probability value of the
sample.

İlkay B. CAN, Hacettepe University

Fundamental Error Variance (cont.)
The sampling constant C is given by:

C=mlfg
Where;
m = mineralogical composition factor
l = liberation factor which has values between 0 and 1
f = shape factor relating volume to the diameter of the particles
(0.5 but for gold ores 0.2)
g = size range factor (usually 0.25 but it highest for closely sized
materials as 0.5)

İlkay B. CAN, Hacettepe University

Fundamental Error Variance (cont.)
The mineralogical composition factor is given by:

(1 - a)[(1 - a)r + at]

m=
a
Where;
r = density of particles of component of interest (g/cm3)
t = density of the gangue particles (g/cm3)
a = % mineral assay

İlkay B. CAN, Hacettepe University

Values of l can be estimated from the table below or
can be calculated as:
L d: size of largest particle
l= L: liberation size
d
Practical values for liberation factor

d95/L l
1 1
1-4 0.8
4-10 0.4
10-40 0.2
40-100 0.1
100-400 0.05
400 0.02

İlkay B. CAN, Hacettepe University

Fundamental Error Variance (cont.)
The size range factor g can be estimated from
the ratio of the nominal top size d to the lower
size limit dl (about 5% undersize) as follows:

Large size range (d/dl > 4) g = 0.25

Medium size range (2 < d/dl < 4) g = 0.50
Small size range (d/dl < 2) g = 0.75
Uniform size (d/dl = 1) g = 1.00

İlkay B. CAN, Hacettepe University

Properties of Fundamental Error Variance

• Decreases when the nominal top size is decreased

• Decreases when the sample mass is increased

• Can never be eliminated no matter what crushing

and homogenization takes place

İlkay B. CAN, Hacettepe University

Minimum Sample Mass
• Rearranging the equation for the fundamental
error variance gives:
3
Cd
MS =
 FE
2

• The sample mass cannot be reduced below this

minimum until the sample is crushed
• This is one of the most important sampling
rules!

İlkay B. CAN, Hacettepe University

Example: Calculation of Minimum Sample Mass
Take a porphyry copper ore with the following
characteristics:

• Top particle size of the ore is 2.5 cm (d)

• Lower size is estimated as 0.1 cm, so large particle size
range (g)
• Liberation size of 200 m (L)
• Particle shape factor = 0.5 (f)
• Chalcopyrite (CuFeS2) density of 4.2 gcm-3
• Gangue density of 2.6 gcm-3
• Grade of 0.35% Cu, (the mineral assay will be 1% CuFeS2)
(a -% mineral assay= 0.01)
• Required fundamental error of 0.02% Cu, i.e, 0.06% CuFeS2
• Calculation of FE = 0.06 𝑒𝑟𝑟𝑜𝑟% 0.02 0.06
𝜎= = or
𝑎𝑠𝑠𝑎𝑦% 0.35 1

İlkay B. CAN, Hacettepe University

Minimum Sample Mass-Example (cont.)
Thus,
m = (1-0.01)[(1-0.01)x4.2 + 0.01x2.6]  0.01
= 414.2
l = (0.02  2.5) = 0.089
C = 414.2 x 0.089 x 0.5 x 0.25 = 4.61

=0.02/0.35=0.06 (for 100 confidence limit)

Hence,
mS = (4.61 x 2.53 )  0.062
= 20,009 g
= 20 kg

İlkay B. CAN, Hacettepe University

Practical Values
The amount of samples according to particle size
(d: particle size, mm)

d<15 : 30 kg
15<d<30 : 60 kg
30<d<50 : 90 kg
50<d<100 : 160 kg
d>100 : 240 kg

İlkay B. CAN, Hacettepe University

Why we are taking samples from the streams of the
plant?

• The performance evaluation of the mineral processing

circuits is highly dependent on the quality of data
taken from sampling study. This is important for
modelling and simulation!
▪ Sampling should be done while the plant is operating in
steady state condition and all necessary analyses
should be done
▪ The aim is to determine the performance of the
circuit correctly and define as quantitatively

İlkay B. CAN, Hacettepe University

The reasons for sampling?
• How metal lost is reduced?
• The fineness for grinding is enough?
• Capacity is enough or can be increased?
• The addition of flotation chemicals is true?
• Are the hydrocyclones working efficiently?
• How can we improve the quality of final
concentrate?

İlkay B. CAN, Hacettepe University

Flowsheet
• The first step in sampling from a plant is to
determine the flowsheet and sampling points
on the streams!

İlkay B. CAN, Hacettepe University

İlkay B. CAN, Hacettepe University
Plant Sampling or Surveying

• Sampling points on flowsheet

• Can these points be physically reached and sampled easily?
• Selection of appropriate samplers and buckets
• Run the plant at steady state condition

Questions should be asked;

• Which samples are more important?

• Which samples are similar?
• Can tonnage be estimated?
• Does regrinding exist in the circuit?
• Have enough samples were taken to identify the circuit?

İlkay B. CAN, Hacettepe University

The steps of sampling in a plant
1. Taking moisture sample from the plant feed
2. Stopping of feed belt conveyor and taking all
samples from a certain length of the conveyor
3. Taking and accumulating of the samples
periodically during a few hours of stable
operation time of the plant
4. Wet weighing of the samples instantly
5. Filtrating and drying the samples
6. Dry weighing the samples

İlkay B. CAN, Hacettepe University

Works for taken samples

• Some amount of samples are separated for basic

analyses (head assays)

• Control of sampling accuracy and mass balance by

using analysis data

• If you have any problem in mass balance then go

and repeat the sampling!

İlkay B. CAN, Hacettepe University

What kind of analyses should be done to the
samples?

If sampling is proved to be correct by mass balancing;

• Determination of particle size distribution

• Chemical and mineralogical analyses of the particle
size fractions
• Mass balance based on size fractions
• Mineralogical evaluation of the streams

İlkay B. CAN, Hacettepe University

 Typical Sampling Study for a Grinding Circuit
• Feeding tonnage should be kept constant as far as possible.
• Waiting for the stable condition (steady state) of the circuit for
a while.

İlkay B. CAN, Hacettepe University

Typical Sampling Study for a Grinding Circuit

Rod Mill Ball Mill Hydrocyclone

Time Discharge
Feed Water Power Power Water Pressure % solid in
of feed
(t/h) (m3/h) (kW) (kW) (m3/h) (kPa) feed
pump (%)

7:00 105 59 848 1864 396 123 75.3 31.5

7:10 115 58 847 1866 396 118 73.8 31.8
7:20 116 59 850 1874 398 119 73.4 31.3
7:30 116 58 827 1863 395 110 70.2 30.4
7:40 117 58 840 1862 398 115 72.3 30.9
7:50 118 59 873 1867 400 115 73.1 31.0

İlkay B. CAN, Hacettepe University

From a sampling survey study in Çayeli Copper Works......

İlkay B. CAN, Hacettepe University

Summary

• Correct sampling practices are critical to obtaining

reliable quality information and metallurgical balances

• Elimination of bias is more important than improving

precision

• Process streams should be sampled at the discharge

end by taking a complete cross-section of the stream

İlkay B. CAN, Hacettepe University

Summary (continued)

• Sampling systems should be checked regularly to

eliminate obvious faults

• Minimum sample mass requirements after division must

be respected

• Special requirements for measurement in section

• Great care needed with rare phases

İlkay B. CAN, Hacettepe University