Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Environ. Sci. Technol.

1993, 27, 2035-2044

Sampling of Contaminated Soil: Sampling Error in Relation to Sample Size


and Segregation
Frank P. J. Lam6'vt and Peter R. Defize*
Institute of Environmental Sciences IMW-TNO and Institute of Applied Physics, Centre of Applied Statistics, The Netherlands
Organization for Applied Scientific Research TNO, P.O. Box 601 1, 2600 JA Delft, The Netherlands
When comparing these publications with the problem
When sampling a mixture of particles of different nature of sampling contaminated soil, two major differences were
like soil, there will always be a minimal amount of variation observed: (a) The concentrations are very high (% level)
between the samples due to the fact that the particle compared with concentrations in contaminated soil (ppm
composition of each sample will differ. This minimal
level) and (b) for most materials, the particles are much
variation in sampling a particulate material is known as
larger than soil particles. Due to these two major
the fundamental error. It was previously unknown to what
differences, it was unknown if these theories could be used
extent this fundamental error could influence the variance
to predict sampling error in relation to sample size in
in a series of soil samples taken for analysis. On the basis
of experiments with soil contaminated with cyanide, we contaminated soil. Therefore, two experiments were
conclude that the fundamental error only influences the performed to investigate to what extent the existing
variance when samples smaller than 10g are taken. When theories are applicable to contaminated soil.
using larger samples, the segregation error determines the In this paper we will give a short overview of the sampling
variance. However, there is no practically applicable theories, focusing on the two basic causes of sampling error
method available for estimating the segregation error when handling a particulate material: the fundamental
before sampling and analysis. Therefore, we developed a error and the segregation error. The experimental results
method for estimating the segregation error using a will be discussed and compared with the existing sampling
sampling board. This resulted in segregation plots which theories. Based on these results, a supplementary ap-
can be compared to the lot to be sampled. proach will be presented to tackle the problem of sampling
a particulate material, like contaminated soil, using an
appropriate sample size.
Introduction
In quality control of cleaned soil, both the mean Theory
concentration and the variance in this concentration are
important testing criteria (1). In order to check on both The theory of sampling particulate materials is vast and
criteria, samples of the cleaned soil must be taken and complicated. It has been widely accepted that the most
analyzed. complete theory has been developed by Gy (2). Other
Publications on the sampling of particulate materials, references are Harnby (9), Ingamells (10,11),and Visman
however, indicate that due to the fact that the samples (3). As our main point of interest was to investigate the
consist of individual particles with different characteristics relation between sample size and sampling error, we
the variance is influenced by the size of the sample. As compared experimental results with theoretical ones. The
the variance is one of the quality control criteria for cleaned theories of Gy and Visman appeared to be the most useful.
soil, it is essential to know to what extent the sample size Gy's Theory. As mentioned by Pitard (12,131, the
affects the variance in concentration between samples. publications by Gy are hard to digest and difficult to read.
For this reason, we studied the effect of sample size on Gy introduces a whole set of new terminology and concepts
sampling error. which are the building blocks of his sound and thorough
In a strongly simplified model of the sampling of a theory. Here we shall only discuss in short those parts of
particulate material, namely, the sampling of a mixture his theory which are important for our study.
of red and white balls, the effect of sample size on sampling
error can be calculated using simple binomial statistics. Cy defines two forms of heterogeneity: constitution
In the case of the sampling of a complex (and even partially heterogeneity (CHL)and distribution heterogeneity (DHL).
unknown) mixture of different kinds of particles, like soil, CHL is the heterogeneity that is inherent to the compo-
the effect of sample size on sampling error cannot be sition of each particle making up the lot. The greater the
predicted with simple statistics. More precise theories difference in compositionbetween each particle, the greater
are needed to describe the relationship between sample the constitution heterogeneity. The DHL of a lot is the
size and sampling error. In a review of literature in heterogeneity that is inherent to the manner in which the
Geophysical and Chemical Abstracts data bases, as well individual fragments are distributed over the lot. Dis-
as in literature on this subject which we already possessed, tribution heterogeneity is directly connected with segre-
we found only a relatively small number of publications gation (a lot is segregated when the critical component is
in which the effects of sample size on sampling error have concentrated in certain parts of the lot). The distribution
been described. The most relevant publications on this heterogeneity can be reduced by reducing the segregation,
subject are shown in the reference list to this paper (2- e.g., by mixing the lot.
14). The mathematical definition of CHL is as follows:
* Author to whom correspondence should be addressed.
t Institute of Environmental Sciences IMW-TNO.
t Institute of Applied Physics, Centre of Applied Statistics.
0013-936X/93/0927-2035$04.00/0 0 1993 American Chemlcal Society Envlron. Scl. Technol., Vol. 27, No. 10, 1993 2095
with bottlenecks in applications):
when d 5 dl then 1=1
when d > dl then 1 = (dl/d)l/2
S h a p e Factor f. This factor is related to the shape of
where N is the number of particles in the lot, ai is the the particles. It is defined as the ratio of the volume of
concentration of the critical component in particle i, U L is a particle passing a certain sieve to the volume of a cube
the concentration of the critical component in the lot, Mi passing the same sieve [Minkkinen (14)l. Gy stated that
is the weight of the particle i, and is the mean weight a value of 0.5 for the shape factor is a good estimate for
of the particles; hi is called the standardized deviation of most materials. For particles with a coin-shape, however,
particle i. avalue in the range of 0.1-0.2 should be taken. Pitard (ref
The definition of DHL is similar to that of CHL.Instead 12, p 160) described a method to estimate f by means of
of dividing the lot into particles, the lot is divided into sieve experiments.
groups of particles. The variance of the standardized
deviation of the groups is defined as DHL.The magnitude S i z e Distribution Factorg. Gy (2)gave a rule of thumb
of DHL for a lot is not a fixed value but depends on the on how to calculate g. Let d be the maximum diameter
grouping that has been chosen. and d’ give the minimum diameter of the particles:
The fundamental sampling error plays an important if did‘ > 4 then g = 0.25
role in the theory of Gy. This is the error that ‘remains’ if 2 < did’ 5 4 then g = 0.50
under ideal sampling conditions due to the different if 1 < did’ 5 2 then g = 0.75
characteristics of the particles. These conditions are met if d/d‘ = 1 then g = 1.00
if the sampling procedure can be thought of as a Bernoulli Thus, as the variation in size diameter increases, the size
experiment where each particle of the lot has an equal distribution factor decreases. From eq 4 it follows that
probability to be part of the sample. the relative variance of the fundamental error is inversely
Gy showed that the relative variance a 2 of~ the funda- proportional to the sample weight. It has been found by
mental error is related to CHL as follows: others [Ingamells (IO), Harnby (91,Visman (311 that such
a relationship holds in many practical situations. It was
(3) Gy’s merit that he gave a method to estimate the constant
C , purely on the basis of material properties and an initial
where P is the selection probability of a particle, and a,
estimate of the concentration of the critical component.
is the concentration of the critical component in the
sample. Segregation. When sampling in the field or from a lot,
This theoretical formula cannot be applied in practice. segregation is an important factor which affects the
The formula, however, is a basis for the derivation of sampling error. Usually, it cannot be assumed that lots
formulas that can be used. are perfectly mixed. This section is therefore concerned
In fact, Gy did this and derived the following formula with quantifying the effect of segregation on sampling
under certain assumptions: error in relation to sample size.
Consider a lot of N F particles to be divided into NG
Cd3
~ ~ ~ (=a-with
, ) C = clfg (4) subsamples. A sample is obtained by a Bernoulli exper-
M.9 iment with selection probability P for each subsample.
where M, is the weight of the sample, c is the composition The selected subsamples are combined in one final sample.
factor, 1 is the liberation factor, f is the shape factor, g is Then the relative variance of the final sample can be shown
the size distribution factor, and d is the maximum diameter to be equal to
of particles. These parameters and the maximum diameter
can be estimated from the characteristics of the particles.
The quantity C is called the sampling constant and is the
product of 4 parameters: c, 1,f, andg. Foreachparameter, This relative variance can be compared with the relative
we give a short description., variance of the fundamental error as follows.
C o m p o s i t i o n Factor c. The following formula holds for
this factor: Define the grouping factor as

(5) (7)

where U L is the concentration of the critical component in This factor is almost equal to the average number of
the lot, A, is the density of the critical component, and A, particles in a subsample when the sample size is small
is the density of the matrix. relative to the size of the lot. Gy showed that the following
Liberation Factor 1. The liberation factor has a value inequality between CHL and DHL holds:
between 0 and 1 and defines to what extent the critical
component is ‘liberated’ from the matrix (complete
liberation: 1 = 1). Gy (2) gave a rule of thumb on how 1
can be calculated from the values of the maximum diameter By introducing a segregation factor [ with 0 5 5 5 1,one
d and the liberation diameter di. The latter parameter is can write
defined as ‘the particle size of a comminution that would
ensure the complete liberation of the constituent of
interest’ (this definition is not very precise and one of the

2038 Environ. Sci. Technol., Vol. 27, No. IO, 1993


From eqs 3 and 6 it follows that ground and divided with the aid of a rotating divider,
without affecting the concentration. In this way it is
azo = (1 + y[)a$ (10) possible to make an analytical sample of the normal size
Logically, 620 is always larger than U ~ F .How much larger (50g) without allowing alarge error during sample division.
depends on the value of 75. Gy stated that ‘when designing Of course, this assumption must be checked by duplicate
a sampling plant, it is always safe to assume that 75 = 1’ subsampling and analysis.
(ref 2, p 343). This was also confirmed by Pitard, who To confirm the results of the first experiment, it was
mentioned: ‘Taking into account the large amount of followed by a second one. In this second experiment, a
variographic experiments he performed between 1960 and soil which was very heterogeneously contaminated with
1975, Gy reached the pragmatic conclusion that the cyanide was sampled and analyzed. The results of these
product of yt stays slightly smaller than 1’ (ref 12, p 188). two experiments were compared with each other as well
In most cases, however, the sampling of contaminated as with the theoretical relation between sample size and
soil will deviate strongly from the ideal sampling that is sampling error.
possible in a specially designed sampling plant. In the First Experiment. During the thermal cleaning process
majority of cases, soil contamination will have a strongly of soil originating from a former gasworks site and
heterogeneous nature, and therefore the amount of seg- contaminated with cyanide and PAHs, 60 bulk samples
regation within a lot will be high. This is illustrated by were taken from the conveyor belt which entered the plant.
Figure 2. The sampling of contaminated soil, therefore, It was preferred to sample the contaminated soil because
is one of the cases in which the assumption that 75 I 1 sampling of the cleaned soil would lead to a relatively
is not correct. large number of samples with concentrations below the
A source of inspiration for our own research was the detection limit, which could make it impossible to study
work of Visman (3). He is one of the few who did practical the effect of the sample size on sampling error.
research into the effect of segregation on sampling error. From each bulk sample, subsamples of 0.1, 1, 50,400,
Visman did hypothetical experiments with a ‘sampling and 6500 g were taken without any further homogenizing
board’. This sampling board consisted of 10 000 positions, of the bulk samples. In this way, the subsamples taken
2500 of which were occupied by lead pellets. The sample were comparable to samples of these sizes taken directly
taken from the board consisted of n neighboring positions. from the conveyor belt.
The sample result was the fraction of positions containing The subsamples were dried at 40 “C and, for samples
a pellet. Visman studied the variance of this fraction (a2) larger than the normal analytical sample size of 50 g, ground
in relation to the distribution of the pellets over the over a 1-mm sieve and divided with a rotating divider.
sampling board. From these empirical experiments, he After the original sampling from bulk samples, duplicate
concluded that the following relationship between variance samples were taken during pretreatment and analysis. As
and sample size approximately holds: a result, it was possible to estimate the different variance
components due to sampling (a2(,)), due to sample pre-
a’= (A/M,) + B (11) treatment ( Q ~ ( ~ ) )and
, due to analysis (a2(*)), together
In this formula, A and B are parameters and M , is the resulting in the total variance (a2(t)= a2(s)+ a2(p) +
sample size. Visman called the quantity AIM, the random (the number of duplicate samples and analyses was based
variance and B the segregation variance. The parameters upon earlier studies with this type of contamination). By
A andB should be estimated from available data. Visman’s means of an unbalanced hierarchical analysis of variance
formula was discussed by Duncan (41, who also suggested (161,we were able to give unbiased estimates of the variance
some modifications. components, without analyzing a very large number of
Equations 10and 11are identical if the segregationfactor subsamples. An example of the resulting sampling
5 is taken as a constant. The second term in eq 10 (75 a 2 ~ ) schemes is shown in Figure 1. Table I shows estimates of
is then independent of sample size because y is directly the variance components and the total variance.
proportional to the sample size. By taking the product of An intensive study of the soil and its cyanide contam-
y and &, the influences cancel each other. The segre- ination was made in order to gain information about the
gation factor is not a constant by definition, but distribution of the cyanide in the soil. This included the
simulation results of Duncan (4) and ourselves with a determination of the particle size distribution of the soil,
sampling board give some confidence that it is a fairly the distribution of cyanide over the particle size classes,
good approximation. a microscopic study of the contaminated particles using
an optical microscope as well as a scanning electron
Experimental Section microscope (SEMI, astudyof the form in which the cyanide
is present by means of infrared microscopicanalysis, X-ray
Introduction. In view of the sampling theory by Gy microanalysis (XRMA), and specific analytical methods
(2), a first experiment was carried out with soil contam- to determine the amount of complex cyanide (Prussian
inated with cyanide. Cyanide, mainly present as ferric Blue), free cyanide (CN-), and thiocyanides (SCN-). As
hexacyanoferrate(II), is a stable salt (15);it is present in a result of this part of the study, it was possible to obtain
the soil in the form of small particles (ferric hexacyano- estimates of the parameters which were used in the formula
ferrate(II1, Fe4[Fe(CN)&, better known as Prussian Blue; for the fundamental error (eq 4). A direct comparison of
it is a cyanide-iron complex which is formed during the theoretical and practical results could therefore be made.
dry cleaning of coal gas and one of the major contaminants Second Experiment. To ensure a strongly heteroge-
found on former gasworks sites). It may therefore be neous distribution of cyanide, we used a contaminated
assumed that this type of contamination behaves as a site for the second experiment. The top soil layer of a
particulate material. Furthermore, the contaminated soil small part of the site was carefullyremoved before sampling
can be dried at low temperature and, subsequently, be in order to make the contamination visible. The exper-

Environ. Sci. Technol., Vol. 27, No. 10, 1993 2037


Sampling etc. (60samples)

Pretreatment etc. (105 su bsamples)

Analysis

Flgure 1. Samplingscheme for the sample pretreatment of 400-g samples. The original number of samples (60) was based on previousexperlence
with the analysls of cyanide and the variance components involved (sampling, sample pretreatment, analyses). Knowing the order of magnitude
of the different variance components,a hierarchicalanalysis of variance could be designed which would give the deslred power of dlscrlmlnatlon.
For each series of sample sizes a dlfferent pretreatment was necessary, resulting from the size of the original sample and the amount of soil
used for the analysis (=50 9). Samples smaller than 50 g were dried and, excludlngthe 0.1- and 1-g samples, ground over a 1-mm sleve; samples
larger than 50 g were also dlvlded In subsamples by means of rotating dividers. Therefore, each sample slze had its own pretreatment pattern
and analysls of variance. For the 4 0 0 9 samples, this resulted in 105 subsamples and 120 analyses.

Table I. Mean Concentration, Total Variance, and Variance Contributions for First Experiment
sample size (9)
0.1 1 50 400 6500
no. of analysed 60 72 90 120 50
mean concentration (mg/kg) 1194 501. 480 506 529
total variance (&)) 3.09 X 10’ 2.87 x 104 8.04 x 103 2.73 x 104 5.46 X 109
Variance Components
absolute
sampling ( u 2 d 2.72 x 104 7.93 x 103 2.48 x 104 5.16 x 103
pretreatment ( U Z ( ~ ) ) 2.37 x 103 Ob
analysis 1.51 x 103 110 90 300
relative ( % )
sampling (&) 94.7 98.6 91.0 94.5
pretreatment ( u ~ ( ~ ) ) 8.7 Ob
analysis (a!$*)) 5.3 1.6 0.3 5.5
The number of analyses depends upon the pretreatment scheme used; 0.1 g, no pretreatment, no duplicate analyses (no analysis of
variance); 1 g, no pretreatment, some duplicate analyses; 50 g, drying and grinding, duplicates of pretreatment and analyses; 400 g, see Figure
1;6500 g, drying, grinding, subsampling, duplicates of pretreatment, and analyses. In the 6500-gseries, only half of the original bulk samples
were used. Negative estimates, which are theoretically possible due to random variation, have been replaced by zero estimates.

imental area was then divided into 60 squares of 0.25 m2. how, the results of this second experiment made a
Starting from the center of each square, samples were comparison irrelevant, as will be shown in the discussion.
taken from a surface of 0.0001, 0.01,0.09, and 0.25 m2and Discussion of the Results. By knowing the distri-
a depth of approximately 1cm. With each sampling step, bution and the nature of the cyanide contamination in the
the center which had already been sampled was not first experiment, we were able to make good estimates of
sampled again. This resulted in 60 samples of approxi- the shape factor f (=0.5), the size distribution factor g
mately 1, 400, 2500, and 6000 g. The sampling of 0.1-g (=0.25), the composition factor c (=3600 g/cm3), and the
samples took place in the laboratory. Subsampling from maximum particle size d (=0.8 cm). As already mentioned,
both the 2500- and 6000-g samples resulted in two series the estimate of the liberation factor 1 is more difficult.
of 0.1-g samples. Calculation of 1 from d and dl gave a very low value for 1.
Sample pretreatment and analysis comparable to the In line with Gy’s recommendation (ref 2, p 2621, a minimum
first experiment were carried out. Because of the heter- value of 0.03 for 1 was chosen to calculate the fundamental
ogeneity within the lot, see Figure 2, the resulting error.
distribution of the analytical data was strongly skewed, as In Figure 3, the coefficient of variation according to
shown in Table 11. The unbalanced hierarchical analysis Gy’s formula (= CJF) is plotted against the sample size,
of variance used in the first experiment resulted therefore together with the coefficient of variation due to the
in highly unreliable estimates of the variance components. sampling calculated from the data of the first experiment.
Therefore, only an estimate of the totalvariance was made. When the experimental variance for the 0.1-g samples is
By neglecting the variance caused by sample pretreatment not taken into account, there seems to be no significant
and analysis (these components proved to be less than difference between the variances found for the different
10% of the total variance in the first experiment), the sample sizes. Furthermore, the shape of the experimental
total variance slightly overestimates the sampling variance. relationship does not support the theoretical relationship
In contrast to the first experiment, the distribution and based upon Gy’s theory with respect to the fundamental
particle characteristics of the cyanide contamination were error.
not investigated in the second experiment. Therefore, it The fundamental error is, for at least part of the
was not possible to give estimates for the parameters in investigated sample size range, larger than the experi-
the formula of the fundamental error (eq 4)and to compare mental sampling error. As the fundamental error is the
the experimental results with theoretical estimates. Any- smallest possible sampling error, it is expected that the

2038 Envlron. Scl. Technol., Vol. 27, No. 10, 1993


8
4
R
C
0
.J

4
a3
c
8
h

Q)
-0
.J
8
u,
c
a3

0 m
e

Flgure 2. Cyanide distribution on the sampling location for the second experiment.

Table 11. Mean Concentration (mg/kg) and Total Variance for Second Experiment
samde size (a)
0.1 0.1 1 400 2500 6000
no. of analysesn 72 72 72 84 84 84
mean 99.6 118 267 116 126 200
total variance 1.16x 105 2.70 X lo6 5.55 x 105 0.64 x 104 0.81 x 104 3.01 X 106
median 19.5 21.5 72.0 28.8 30.0 36.0
lower quartile 6.5 3.0 27.5 13.8 16.0 19.0
upper quartile 92.0 57.0 217 88.5 96.5 147
minimum 2.0 2.0 1.0 2.0 5.0 7
maximum 1690 6460 2770 3320 2430 2280
Variance is estimated on the basis of 60 samples. When more than one concentration was available due to duplicate sampling, the variance
was calculated using the mean concentration for a sample which was analyzed in duplicate.

fundamental error is always smaller than the experimental 1g, the experimental variance increased with decreasing
sampling error. Therefore, neither the shape nor the sample size. In the second experiment, with a much more
position of the experimental relationship supports the strongly segregated lot, there did not seem to be a direct
relationship based upon the fundamental error. relationship between sample size and sampling error for
In the first experiment, the sampling error seemed to the whole range of investigated sample sizes (0.1-6000 g).
be more or less independent of the sample size for samples Therefore, it must be concluded that the fundamental
larger than approximately 1g. For samples smaller than error alone is of no use for predicting the sampling error.

Envlron. Scl. Technol., Vol. 27, No. 10, 1993 2099


- mxpmrinrntrl - rxpmriinrntrl

+ rsoording Oy + ass. Uirmmn

1000

‘+\ i
‘t

0.1 1 10 ice ieee ieeee 8.1 1 ie $00 ieee iewe

rmmplm miz. (grrmr) m r m p l m r i z m (grmrnm)

Figure 3. Coefficient of variation in relation to sample size. Plotted Figure 4. Coefficient of variation In relation to sample size. Plotted
are the experimentalresultsand the relationshipbetweenthe coefficient are the experimentalresultsand the relationship betweenthe coefficient
of variation and the sample size based on the formula of the fundamen- of variation and the sample size based on Visman’s theory. Exper-
tal error given by Gy. Experimental results: [(experimental imental results: [(experimental variance)”2/mean concentration],
variance)l/*/mean concentration]. According to fundamental error: According to Visman’s theory: ( A M , i-
fJF .
1-and 6500-g samples, this resulted in A = 2.2035 X 10-8
The experimental results clearly showed that the and B = 5.1536 X (M, in grams). The relationship is
sampling error is dominated by the heterogeneity within
the lot. Only for very small samples is the fundamental
shown in Figure 4 . With these estimates, and the
assumption that f is constant, the product -yf in eq 10 can
error a substantial part of the sampling error. The results be estimated. For 1-g samples the product equals 0.23,
of the first experiment made clear that a minimum sample and for 6500-g samples -yf = 1520. The last value is much
size of approximately 10g must be used for the investigated greater than l!
lot to obtain a good estimate of the heterogeneity within
the lot. Using smaller samples, the particulate hetsro- It must be stated that by using the 0.1-gsamples instead
geneity started to have an important effect on total of the 1-g samples the ‘Visman curve’ would, within the
variance, and the sampling error increased due to the investigated sample size range, resemble much more the
fundamental error. No significant increase of the sampling ‘Gy curve’ which is shown in Figure 3. Only for samples
error was found in the sample size range of the second larger than approximately 1000 g would the relationship
experiment. Compared to the first experiment, the flatten. We used the estimated variance of the 1-gsamples
heterogeneity of the second lot was much greater, even for because this estimate is more accurate than the estimated
small samples, resulting in a relatively small contribution variance of the 0.1-g samples (because the variance itself
of the fundamental error to the total sampling error. is larger).
Although Gy’s sampling theory (2) also introduced the When comparing the sampling error in the first exper-
segregation error, which describes the variance caused by iment with the theoretical prediction according to Visman,
heterogeneity within the lot, Gy himself gave no practical there seems to be a large amount of similarity between the
method for calculating this sampling error. In fact, due two. The more or less constant variance found for the
to the use of composite samples, Gy stated that the highly segregated second lot can be explained by the fact
segregationerror is normally smaller than the fundamental that the segregation error is independent of the sample
error (ref 2, p 343). In environmental sampling and size. Therefore, Visman’s formula could be useful in
specifically in quality control, the use of composite samples predicting the minimum sample size to be used. However,
is, however, limited, and mostly individual samples are the practical applicability of this formula is limited because
analyzed. Therefore, the segregation error is one of the the variance between samples must be known for two
main sources of sampling error. sample sizes. Therefore, we used the basic ideas of Visman
When the importance of the segregation error is rec- to develop an alternative method to predict the sampling
ognized, it is important to know that Visman (3) did take error in relation to sample size and segregation patterns.
this sampling error into account. In order to apply eq 11, Application of Geostatistics. Although based on a
it is necessary to know the variances which occur when different concept, segregation can also be analyzed by using
sampling and analyzing large and small samples. By geostatistics. Geostatistics is based on the assumption
knowing these variances, it is possible to calculate the that data are correlated in space and/or time. When this
values of A and B which are used in eq 11. Based on the correlation exists, it can be used to predict data at locations

2040 Environ. Sci. Technol., Vol. 27, No. 10, 1993


gamma
322
12000 *
10500 178 288 284
w s
216
90000 *
146
75000 * 118
*
60000

45000

30000
80
15000 * 32
*
0
I I I I I I 1 I I I I
0 0.500 1.000 1.500 2.000 2.500 3.000 3.500 4.000 4.500 5.000

d istance

mode 1twpe : spher ica 1


direction: total
cO= 8.39€+0004 e = 1.72€+0004 a = 1.07€+0000
S S D / S S T = 7.494

gamma
322

24000 284
*
*
20000

216
16000
*
146
&

12000

80000 /
/
118
40000 *
80 32
0
I I I I I I I I
* 1
* I

d istance
inputfile: cn2500s.rem
node 1 type : spher ica 1
direction: total
cO= 5.43€+0004 e = 1.55E+0005 a= 1.82€+0000
SSD/SST=469.416
~~~~ ~~

Flgure 5. Semlvarlogramsfor the first series of 0. Ig samples of the second experiment (a, top) and the 25004 samples of the second experiment
(b , bottom).

that have not been sampled (17). The existence of Spatial and/or temporal correlation can be measured
segregation implies the existence of spatial correlation. by means of a semivariogram. In a semivariogram, the
The fundamental error, as described in this paper, has no variance of differences between samples is plotted against
relation with spatial or temporal correlation. So in the distance (in time or space) between the sample
principle geostatistics is a useful method to investigate locations.
whether the segregation error is predominant over the The semivariograms of the first experiment (measure-
fundamental error. mente in time) show no temporal correlation. Apparently,

Envlron. Scl. Technol., Vol. 27, No. 10, lQQ3 2041


64.14

54.18 64.19

0:

54.21 64.2

SEGREGATION PLOTS

Coefficient of variation
Segregation factor : given in upper right-hand corner

Flgure 8. Segregation plots, N = 6561, n = 81, u y = 0.4.

the scale of segregation is much smaller than the scale on iments (approximately 10-15 kg each), the sampling
which the temporal correlation can be studied. This can interval (between 10 and 60 min), and the amount of soil
be explained by the amount of soil taken for the exper- passing through the soil cleaning plant in this period (20-

2042 Environ. Scl. Technol.. Vol. 27, No. 10, 1993


25 tons per hour). values can be changed and will lead to different plots which
For the second experiment, two semivariograms are in fact visualize the lot and samples at different scales.
given in Figure 5a,b. Figure 5a is the semivariogram of Appropriate scales should be chosen for specific appli-
the first series of 0.1-g samples which were taken in the cations.
laboratory from the 2500-g samples. Figure 5b shows the We attempted to perform such an exercise for contam-
semivariogram of the 2500-g samples. In both plots the inated soil. The sampling constant C in eq 4 was estimated
semivariogram values drop for distances larger than from the first experiment: C = 13.5 g/cm3 cf = 0.5, g =
approximately 2.5 m. This is caused by the fact that high 0.25, 1 = 0.03, c = 3600 g/cm3,aL = O.O005g/g). For samples
values are concentrated in a relative small area of the of 400 g, this resulted in a coefficient of variation for the
investigation site, see Figure 2. Only Figure 5b does show .
fundamentalerror of 13.1% Suppose that this is the ‘true’
some correlation for data at distances less than 1.5 m. value of UF in the second experiment. From the data of
Semivariograms of other sample sizes are comparable to experiment 2, a coefficient of variation of 268% has been
the ones given, showing no spatial correlation for the 0.1-g calculated.
samples and some correlation for larger sample sizes.These The size of the experimental area in experiment 2 was
results are in line with our conclusion that the segregation 15 m2. A sample of 400 g took about 0.01 m2. In terms
error is of much more importance than the fundamental of the sampling board, this gave a value of 15/0.01= 1500
error in sampling (contaminated) soil. for the ratio N / n . Substituting an arbitrary value of 0.1
for p in eq 13 gave n = 524 and consequently N = 786 000.
Segregation Plots From eq 10 with y = 524, UF = 0.131, and UG = 2.68, it
follows that 4 = 0.80. Unfortunately such a large degree
We used Visman’s sampling board to apply Gy’s theory. of segregation can only be visualized by means of the
Suppose one takes a sample of n neighboring locations sampling board by simulating 78 600 pellets (10%)con-
from a sampling board with N locations on which a fraction centrated in a subarea of a size just over 10% of the size
of p pellets are placed at random. The relative variance of the totalarea (say 12-15 % ). Suchapicture willstrongly
of the fraction locations filled with pellets ( & I ) in the deviate from a picture with 78 600 pellets distributed
sample, disregarding the final population correction, is completely at random over the complete area.
Exercising with data, as has been done above, does give
a rough idea of how much the pattern of contamination
in experiment 2 is due to segregation, compared with the
or pattern that would arise if the distribution was purely
random and only heterogeneous because the material
1
P= (13) under study consists of particles.
(nUzv+ 1)
With the aid of this relationship, it is possible to visualize Conclusions
the constitution heterogeneity (CHL). Suppose N = 6561 In the experimental part of this study, two lots of
(812),n = 81, and UV =.0.4. Then it follows from eq 13 that contaminated soil have been sampled. The first lot was
p = 0.0717. A picture of a sampling board with 6561 sampled from a conveyor belt, while the second lot was a
locations with a fraction of 0.0717 pellets randomly highly segregated part of a former gasworks site.
distributed over the sampling board is shown in the top
From both experiments it was difficult to reach quan-
left corner of Figure 6. titative conclusions due to the uncertain assumptions
The heterogeneity of the picture is completely due to which had to be made. Nevertheless, we can still obviously
the random distribution of the pellets. It cannot be conclude that the distributional pattern of experiment 2
reduced. It can be enlarged, however, by distributing the is highly due to segregation and that the influence of the
pellets not randomly over the board but in a segregated fundamental error can be ignored. This can also be
pattern, as shown in the other pictures of Figure 6. The concluded for the first experiment in spite of the fact that
pellets are concentrated around certain locations (hot the sampled lot of the first experiment was much more
spots). Tne sampling board can be divided into 81 homogeneous.
nonoverlapping squares of 81 positions. Then, by calcu-
Sample sizes for the analysis of cyanide in soil should
lating the relative variance of these squares and using eq
not be smaller than approximately 10 g to avoid a
10 with y = 81, one can estimate the segregation factor 4.
significant contribution of the fundamental error on the
The estimates of 5 are printed at the top right corner of
sampling error. By making segregation plots, it is possible
each picture. The value of 5 is 0.18 in the third picture
to get an idea of the segregation in a lot to be sampled.
of Figure 6. Since y = 81, it follows from eq 10 that UG
When the estimated segregation of a lot is compared to
= 3.95UF in this case.
a segregation plot with the same kind of segregation, an
The numerical exercises as performed above can help estimation of the segregation factor is available prior to
to get a feeling for the sampling error in relation to sample sampling. This can be used to estimate a minimum sample
size and degree of segregation. The idea is first to calculate size, which results in a negligible contribution of the
the relative variance of the fundamental error with eq 4; fundamental error on sampling error.
to then simulate different kinds of segregation plots using
Visman’s sampling board, with the first plot representing
only constitution heterogeneity; and then to calculate the Acknowledgments
variance in relation to sample size. An important point We performed part of this study for the Netherlands
of consideration is the choice of the sample size on the Integrated Soil Research Programme and were financially
sampling board (we used 81 in our example) and the size supported by the Netherlands Ministry of Housing,
of the sampling board itself (we used 6561). All these Physical Planning, and the Environment. We thank

Envlron. Sci. Technol., Vol. 27. No. 10, 1993 2049


Afvalstoffen Terminal Moerdijk B.V. and t h e City of T h e segregation factor (dimensionless)
Hague for their cooperation in t h e sampling, a n d Mr. M. relative variance of the fundamental error (% )
M. Albert for carrying out t h e larger part of t h e practical relative variance of the final sample (% )
work. variance due to analysis
variance due to sample pretreatment
Nomenclature variance due to sampling
concentration of the critical component in particle total variance
i (kg/kg) relative variance of the fraction locations (% )
concentration of the critical component in the lot
(kg/kg)
concentration of the critical component in the Literature Cited
sample (kg/kg)
sampling constant determined by random variance (1) Bosman, R.; De Kwaadsteniet, J. W.; Lamb, F. P. J.;
Versluijs, C. W. A quality control method for decontam-
sampling constant determined by segregation inated soil; Soil Protection Series, Part 79; Sdu Uitgenerij:
variance Den Haag, 1989; ISBN 90 120 6466 x (in Dutch).
composition factor (g/cm3) (2) Gy, P. M. Sampling of particulate materials, Theory and
sampling constant (g/cm3) practice; Developmentsin Geomathematics 4; Amsterdam,
constitution heterogeneity; the heterogeneity that 1982; ISBN 0 444 42079 7.
is inherent to the composition of each particle (3) Visman, J. Mater. Res. Stands. 1969, 9, 8-66.
making up the lot (dimensionless) (4) Duncan, A. J. Mater. Res. Stands. 1971, 11, 25.
(5) Visman, J.; Lerner, M. Mater. Res. Stands. 1971,11,32-37.
maximum diameter of particles (m)
( 6 ) Stange, K. Chem.-Ing.-Tech. 1954,26,150-155 (inGerman).
liberation diameter (m) (7) Stange, K. Chem-Ing.-Tech. 1954,26,331-337 (inGerman).
distribution heterogeneity; the heterogeneity that (8) Stange, K. Chem.-Zng.-Tech. 1963,35,580-582 (inGerman).
is inherent to the manner in which the individual (9) Harnby, N.; Edwards, M. F.; Nienow, A. W. Mixing i n the
fragments are distributed over the lot (dimen- process industries; Butterworths Series in Chemical En-
sionless) gineering;Butterworths: London, 1985; ISBN 0 408 11574
shape factor (dimensionless) 2.
size distribution factor (dimensionless) (10) Ingamells, C. 0.;Switzer, P. Talanta 1973, 20, 547-568.
(11) Ingarnells, C. 0. Talanta 1978,25,731-732.
standardized deviation of particle i (76) (12) Pitard, F. F. Pierre Gy’s sampling theory and sampling
mean weight of the particles (kg) practice, volume Z Heterogeneity and sampling; CRC
liberation factor (dimensionless) Press: Boca Raton, 1989; ISBN 0 8493 6658 5.
(13) Pitard, F. F. Pierre Gy’s sampling theory and sampling
weight of particle i (kg) practice, volume IZ Sampling Correctness and sampling
weight of sample (kg) practice; CRC Press: Boca Raton, 1989; ISBN 0 8493 6659
number of neighboring locations in one sample 3.
(dimensionless) (14) Minkkinen, P. Anal. Chim. Acta 1987, 196, 237-245.
number of particles in the lot; number of locations (15) Meeussen,J. C. L.; Keizer,M. G.; de Haan, F. A.M. Environ.
on a sampling board (dimensionless) Sci. Technol. 1992, 26, 511-516.
number of particles in a lot (16) Gower, J. C. Biometrics 1962, 18, 537.
(17) Journell, A. G.; Huijbregts, Ch. J. Mining geostatistics;
number of subsamples in a lot Academic Press: London, 1978.
selection probability of a particle or subsample
(dimensionless) Received for review September 15, 1992. Revised manuscript
grouping factor (dimensionless) received June 2, 1993.’
density of the critical component (g/cm3) ~~ ~~~~

density of the matrix or gangue (g/cm3) 0 Abstract published in Advance ACS Abstracts, August 15,1993.

2044 Envlron. Scl. Technol., Vol. 27, No. 10, 1993

You might also like