Marine and Petroleum Geology 98 (2018) 384–396

Contents lists available at ScienceDirect

Marine and Petroleum Geology

journal homepage: www.elsevier.com/locate/marpetgeo

Research paper

A new method for calculating the primary porosity of unconsolidated sands T

based on packing texture: Application to modern beach sand
Yiming Yana,b, Liqiang Zhanga,b,∗, Xiaorong Luoc, Chao Lic, Fanjun Hua,b
a
China University of Petroleum (East China), Qingdao, 266580, China
b
Qingdao National Laboratory for Marine Science and Technology, Qingdao, 266071, China
c
Key Laboratory of Petroleum Resource Research, Institute of Geology and Geophysic Chinese Academy of Science, Beijing, 100029, China

A R T I C LE I N FO A B S T R A C T

Keywords: The primary porosity is one of the fundamental parameters for quantitatively predicting reservoir quality and
Primary porosity studying diagenetic evolution. Primary porosity is usually estimated by empirical approaches or empirical for-
Mathematical model mula, through using a single parameter, such as sorting or grain size. However, these methods have some
Packing texture disadvantages which lead to obvious errors. Primary porosity can be affected by multiple factors, such as particle
Grain size distribution
size, sorting, grain shape and packing texture, therefor, one empirical formula is not practical for different type
Beach sands
of sands. A new mathematical model is proposed in this paper for calculating the primary porosity of un-
consolidated sands based on the concept of packing texture. The grain size, grain size distribution curve and
packing texture are considered in this mathematical model. This paper tests the practicability of the mathe-
matical model through using 40 modern beach sands sample collected from eastern China and 37 samples from
literature. These modern beach sands in the study area can be divided into three types based on the particle size
distribution curve. Each type sample is further divided into several units according to the particle size ratio. The
mathematical model firstly calculates the primary porosity of two adjacent units by using the binary packing
model, and then the primary porosity of the entire sand sample. The error of model-predicted primary porosity
ranges between −2% and 2%, and the correlation coefficients are all above 0.9. As a comparison, the errors of
empirical formula-predicted results range between −6% and 6%, and the correlation coefficient is 0.63. The
newly-proposed mathematical model for the primary porosity based on grain size distribution and packing
texture is more reliable compared to the traditional empirical formula.

1. Introduction
Sandstone consists of a framework of grains, interstitial detrital silt
and clay, chemical cements and an interconnecting network of void
space or pores (Pettijohn et al., 1972). Pores are areas of sandstone that
can store and transmit fluid, such as ground water, oil, gas, and mi-
neralizing solutions. Porosity in sand sediments may be altered during
the diagenetic processes (Pan et al., 2011; Mai, 2013; Wang et al, 2013,
2017). The primary porosity are described quantitative and semi-
quantitative for calculating quantitatively the effects of compaction,
resolution, and cementation on the porosity of sandstone in some stu-
dies, such as porosity-depth trends (Taylor et al., 2010; Ajdukiewicz
and Lander, 2010), porosity evolution curve (Liu et al., 2007; Morad
et al., 2010), hydrocarbon filling history (Bloch and Helmold, 1995;
Guo et al., 2016), and reservoir quality prediction (Taylor et al., 2010).
Past studies show that primary porosity can be expressed in terms of
five parameters. They are grain size, sorting, sphericity, roundness, and
packing texture (King, 1989; Ellis and Lee, 1919; Terzaghi, 1925; Trask,
1930, 1931; Fraser, 1935; Graton and Fraser, 1935; Folk and Ward,
1957; Pryor, 1972; Beard and Weyl, 1973; Scherer, 1987). Theoreti-
cally, the grain size has no influence on the primary porosity of uniform
spheres (Fraser, 1935). And the primary porosity is only influenced by
the packing texture. The tightest packing of spheres is rhombohedral
with a primary porosity of 25.9% (Fraser, 1935; Beard and Weyl, 1973);
the porosity of cubic packing is 47.6% (Fraser, 1935; Beard and Weyl,
1973).
Two methods are commonly used for acquiring primary porosity.
(1) Primary porosity can be acquired by empirical approach, which is
measured by physical experiment. Pryor et al. (1972) measured 992
sand samples, and pointed out that the average porosity of river
point bar sands is 41%, the average porosity of beach sands and

∗
Corresponding author. China University of Petroleum(East China), Qingdao, 266580, China.
E-mail address: liqiangzhangwxm@163.com (L. Zhang).

https://doi.org/10.1016/j.marpetgeo.2018.08.033
Received 5 June 2018; Received in revised form 20 August 2018; Accepted 29 August 2018
Available online 30 August 2018
0264-8172/ © 2018 Elsevier Ltd. All rights reserved.
Y. Yan et al.
dune sands are 49%. Wilson and Mcbride (1988) pointed out that
the porosity of river sands is 45%, and the porosity of beach sands is
43%. This method assumes that the sands have the same primary
porosity which are deposited in the same depositional environment.
This method has a large error when the hydrodynamic forces are
different in the same depositional environment or sedimentary fa-
cies. The porosity of river point bar sands predicted by empirical
approach are 41% (Beard and Weyl, 1973), 43% (Wilson and
Mcbride, 1988), and 48% (Scheidegger, 1958), in different studies.
The maximum error for river point bar sands is 7%.
(2) The formula of Scherer (1987) can also be used to calculate the
primary porosity of sands, as formula (1). It was established based
on the data of wet sands from the experiment of Beard and Weyl
(1973).
Primary porosity = 20.91+(22.9/sorting) (1)
Where primary porosity is the void space as a percentage of bulk vo-
lume and sorting is the Trask sorting coefficient. The primary porosity
of sands is influenced by packing texture, and this single empirical
formula is not practical for all types of sands.
This paper aims to propose a mathematical model through using
packing texture. The model is established on the basis of the mathe-
matical model between the void ratio and the primary porosity, which
is widely used in the high-density packing of ceramics, powder me-
tallurgy, concrete and metal particles (Mcgeary, 1961; Ridgway and
Tarbuck, 1968; Larrard, 2005; Yilmaz, 2009; Chang et al., 2013, 2015,
2016). The calculated results from the proposed model are very close to
the measured primary porosities of sandy samples from modern beach
and literature.
2. The influence of packing texture on primary porosity of
unconsolidated sands
The packing texture of sands refers to the contact relations among
each size particles, the spatial overlay and filling relationships
(Terzaghi, 1925; Trask, 1931; Fraser, 1935; Graton and Fraser, 1935;
Beard and Weyl, 1973; Scherer, 1987). The packing of particles has
been attracting the interest of researchers for over a century. The
packing texture is influenced by the grain size distribution. When the
packing grains are the same size, this packing model is named as one-
component packing; When the packing grains only have two size grains,
this packing model is named as binary packing (The smaller size grains
called fine grains, and the larger size grains called coarse grains); And
when the packing grains have three or more size grains, the packing
model will be names as ternary packing or multiple packing (The grains
will be called fine grain, medium grains and coarse grains by the size of
the grains) (Terzaghi, 1925; Trask, 1931; Fraser, 1935; Graton and
Fraser, 1935).
2.1. One-component packing
For one-component packing, published studies have shown that the
porosity of natural well-sorted sands ranges from 37.2% to 43.7% for
random packing (Trask, 1930; Fraser, 1935; Graton and Fraser, 1935;
Folk and Ward, 1957; Pryor, 1972; Beard and Weyl, 1973; Scherer,
1987).
The theoretical model is relatively simple (Fraser, 1935), and is
shown in Fig. 1 (a, b, c), and the porosities are 46.64%, 30.19%, and
25.95% (Mcgeary, 1961). Based on previous studies, the packing tex-
ture of sand is neither rhombohedral (Fig. 1c) nor cubic (Fig. 1 a) in
nature (Ellis and Lee, 1919; Terzaghi, 1925; Trask, 1931; Graton and
Fraser, 1935; Fraser, 1935; Scheidegger, 1958; Beard and Weyl, 1973;
Scherer, 1987; Wilson and Mcbride, 1988; King, 1989; Atkins and
Mcbride, 1991; Berger et al., 2009; Taylor et al., 2010; Wang et al,
385
Marine and Petroleum Geology 98 (2018) 384–396
2013, 2017; Mai, 2013). Experiment results show that the porosity of
single size spheres is approximately 40.0% in nature (Graton and
Fraser, 1935; Fraser, 1935; Beard and Weyl, 1973; Scherer, 1987; King,
1989). When spheres are piled, any horizontal layer is rhombohedral
(Fig. 1c), and any vertical section is cubic; thus, the theoretical porosity
is 39.5% (Westman and Hugill, 1930). A vertically cubic and horizon-
tally hexagonal packing is the prevailing tendency of one-component
packing in nature.
2.2. Binary packing
For binary packing, Mcgeary (1961) and Chang et al. (2013, 2015)
point out that there are only two types of packing textures (Fig. 1d and
e). When the content of fine grains is low, fine grains fill the voids
among the coarse grains and decrease the primary porosity; this case
has a dominant network of coarse grains (Fig. 1d). When the content of
fine grains is higher, this case has a dominant network of fine grains,
and larger particles are present as embedded, isolated inclusions
(Fig. 1e). The critical condition is that fine particles just fill the voids
space among the coarse particles.
In order to analyze the effect of packing texture on the primary
porosity in the binary packing, this paper calculated the trends which
are acquired by the experiment data from Mcgeary (1961) and
Westman and Hugill (1930). The data in Fig. 2a marked “M” is the
trend line predicted from the experiment by Mcgeary (1961), and the
data in Fig. 2a marked “A” is the trend line predicted from the ex-
periment by Westman and Hugill (1930). R is the diameter of coarse
grains; r is the diameter of fine grains; R/r is the grain size ratio be-
tween coarse grain and fine grain. The results show that the primary
porosity is closely related to its packing texture in the binary packing.
(1) The data in Fig. 2a indicates that the content of fine grains is an
important factor influencing the primary porosity. When the grain sizes
of two spheres are constant, the trend line for primary porosity is a
broken line. Upon an increase in the content of fine grains, the primary
porosity decreases first, and then increases. The essence of the change
in the primary porosity is the packing texture change from Fig. 1d to e.
(2) The data in Fig. 2a also shows that R/r is an important factor in-
fluencing the primary porosity. Upon an increase in R/r (diameter of
large spheres/diameter of small spheres), the primary porosity tends to
decrease.
The experimental data for binary packing are plotted in Fig. 2b (Ye
and Zhang, 1990; Mcgeary, 1961; Yilmaz, 2009; Westman and Hugill,
1930; S. Thevanayagam, 2007; Lade, 1998). The vertical axis is the
minimum primary porosity of the binary mixture, and the horizontal
axis is R/r. When R/r is between 1 and 2, the minimum primary por-
osity decreases slowly. When R/r is between 2 and 7, the minimum
primary porosity decreases significantly. When the R/r is between 7
and 32, the minimum primary porosity decreases slowly. When the R/r
is more than 32, the minimum primary porosity does not change.
Fig. 2b also shows that for a spheres packing, the size of the small
particles should be at least 6.46 times smaller than that of larger par-
ticles in order to fit in the tetrahedral cavities of the packed spheres. In
addition, this relation is the same as the conclusion that when R/r is
greater than 7, the primary porosity decreases slowly. Therefore, to
calculate the primary porosity of a binary mixture, R/r should be
considered, in addition to the sorting coefficient, sphericity and
roundness.
2.3. Ternary packing
The packing texture of ternary packing and multiple packing is
complicated and difficult to characterize, and this is why the primary
porosity of sands is difficult to calculate. However, it can be divided it
into multiple binary packing in three-dimensional space.
Ternary packing is equivalent to a three-phase system and can be
represented by a concentration triangle, such as Fig. 3. The data for
Y. Yan et al. Marine and Petroleum Geology 98 (2018) 384–396

Fig. 1. Packing Structures (modified from Mcgeary (1961) and Chang et al. (2015)) (Fig. 1a, b and c are modified from Mcgeary (1961); Fig. 1e and f are modified
from Chang et al. (2015). The porosity of single size spheres packing is 46.64% for cubic packing, 30.19% for double-nested packing, and 25.95% for closely packed
(rhombohedral packing). Fig. 1d shows the dominant network of coarse grains, and Fig. 1e shows the dominant network of fine grains).

analyzing are from the experiment of Westman and Hugill (1930); and
the data in Fig. 3b are from the experiment of Ridgway and Tarbuck
(1968). Fig. 3 shows that the primary porosity has significant correla-
tions with the grain size distribution and diameter ratio. The diameter
ratio is smaller, and the variation range in the primary porosity is
smaller; the diameter ratio is larger, and the variation in the primary
porosity is larger. The trends of the primary porosity contour line are
similar in Fig. 3a and b, although the diameters and diameter ratios of
particles are different in the two experiments. When the content of any
size particle is more than 90%, the primary porosity of the mixture is
high. The higher the degree of mixing, and the larger the diameter ratio
of the particles, the lower the original porosity is. Therefore, the pri-
mary porosity of ternary sphere packing is affected by the packing
texture; the similarity of primary porosity trends is the best proof.
These experiments which discuss the relationships between one-
component, binary, ternary packing textures and the primary porosity
lay a good foundation for studying the packing texture in the sands.
However, the grain size distribution is continuous in nature. Therefore,
the continuous grain size distribution packing texture should be dis-
cussed to calculating the primary porosity of sands.

Fig. 2. Primary Porosity of Binary Mixture (Fig. 2a Data From McGeary, Westman and Hugill. The trends line marked “M” which is binary mixtures of coarse spheres
(3.15 mm) with six different sizes of smaller spheres (0.914 mm, 0.660 mm, 0.483 mm, 0.279 mm, 0.191 mm, 0.165 mm) is plotted in Fig. 2a. The primary porosity
marked “A” for binary mixtures is the trend line form Westman and Hugill (1930). Fig. 2b Data from Experiments in Six Published Literature).

386
Y. Yan et al.
Fig. 3. Primary Porosity of Three Size Spheres Mixture (The data in Fig. 3a is from the experiment by Westman and Hugill (1930), The data in Fig. 3b is from the
experiment by Ridgway and Tarbuck (1968)).
3. Method for calculating primary porosity
3.1. Method for calculating primary porosity of binary mixture
Past studies have established a mathematical model for calculating
the primary porosity of a binary mixture; the mathematic model shows
that the primary porosity is affected by the packing texture, and the
diameter ratio of the particles, and has no connection with the grain
size (Westman and Hugill, 1930; Mcgeary, 1961; Ridgway and Tarbuck,
1968; Lade et al., 1998; Thevanayagam, 2007; Yilmaz, 2009; Chang
et al., 2013, 2015, 2016).
(1) Considering the simplest case of a binary mixture, the larger voids
can be filled with smaller particles (R/r ≥ 7). This situation can be
divided into two types; the dominant network composed by fine
grains (Fig. 4a) and the dominant network composed by coarse
grains (Fig. 4b).
① The dominant network composed by fine grains
The porosity volume of one component is defined as Vφ, the change
of porosity volume is defined as ΔVφ; the volume of fine particles is
defined as Vf, the volume of coarse particles is defined as Vc, and the
primary porosity of the binary mixture is defined as φ.
The phase diagram for pure fine packing is shown in Fig. 4a. Then,
we consider a mixture of fine and coarse grains. When the content of
coarse particles is low, the coarse particles are present as isolated in-
clusions embedded within the network of fine particles. When the
Fig. 4. Primary porosity calculating theoretical model for binary mixture (modified from Chang et al., 2013, 2015, 2016).
387
Marine and Petroleum Geology 98 (2018) 384–396
content of coarse grains increase, the volume of the mixture increase,
but the volume of pore space do not change. The solid volume of the
mixture is increases to Vf + Vc + Vφ (Fig. 4b). Formula (2) can be used
to calculate the primary porosity of the mixture (Chang et al., 2013,
2015; 2016; Yilmaz, 2009; Ridgway and Tarbuck, 1968).
φ= Vφ/(Vf + Vφ + Vc) (2)
Fig. 4b illustrates the limiting situation when there is no change in
pore volume for the minimum porosity of the binary mixture. When the
content of coarse particles is large, the pore space changes (Fig. 4c). In
addition, Formula (3) can be used to calculate the primary porosity of
the binary mixture (Chang et al., 2013, 2015; 2016; Yilmaz, 2009;
Ridgway and Tarbuck, 1968).
Vφ + ΔVφ
φ=
Vc + Vf + Vφ + ΔVφ (3)
② The dominant network composed of coarse grains
The phase diagram of pure coarse particle packing is shown in
Fig. 4d. Then, we consider a mixture of coarse and fine grains. In the
limiting case, all the added fine particles fill the pore spaces among the
coarse grains without altering the network of coarse grains (Fig. 4e).
The volume of fine grains in this case is defined as Vf0. The volume of
the binary mixture and the volume of coarse particles do not change,
but the volume of the pore space decreases (Vφ’ = Vφ-Vf0). Formula (4)
can be used to calculate the primary porosity of the binary mixture.
Y. Yan et al.
Vφ − Vf 0
φ=
Vφ + Vc (4)
When the content of fine particles is large, the added solid volume of
fine particles (Vf0’) occupy a space in the void volume (ΔVφ), and the
volume of pore space are Vφ+ΔVφ. The volumes of the binary mixture
and the pore space all change. Formula (5) can be used to calculated the
primary porosity of this binary mixture (Fig. 4f).
Vφ − Vf 0 + ΔVφ
φ=
Vφ + Vc + Vf0′ + ΔVφ (5)
(2) Considering the general case of a binary mixture, larger voids
cannot be filled with fine grains (7 ≥ R/r ≥ 2). In this case, the
minimum primary porosity of the mixture is reduced with the in-
crease in the particle size ratio of the two spheres (Fig. 2b).
The results of the experiment by Westman and Hugill (1930) and
the mathematical model formulated by Chang et al. (2013, 2015, 2016)
all indicate that the functional relationship between the minimum
primary porosity and the content of fine particles is similar. Fig. 2a
show that when R/r = 2, 3.45, 4.76 and 6.54, the trends of the change
in the primary porosity are similar. In addition, the mathematical
model of Chang et al. (2013, 2015,2016) also shows that the trend of
the change in the primary porosity is similar, but the slope of change in
the primary porosity is different. Therefore, the model for calculating
the primary porosity when the condition is R/r ≥ 7 can be used to
calculate the primary porosity when the condition is 7 ≥ R/r ≥ 2. And
this type is transitional, can also be divided into two dominant network
types.
(3) In the special case of a binary mixture with 2 > R/r ≥ 1, larger
voids cannot be filled with smaller particles. The experiment results
(Ye et al., 1990) show that the primary porosity changes less in this
case, because the packing texture is similar when the contents of
different size particles change. The primary porosity in this case is
same as the single size packing. Formula (6) can be used to calcu-
late the primary porosity for this case.
φ= Vφ/(Vc + Vf + Vφ)
(6)
For the first two cases (7 ≥ R/r ≥ 2 and R/r ≥ 7), the ΔVφ is needed
to calculated the primary porosity, but ΔVφ is difficult to acquiring. The
mathematical models by Chang et al. (2013, 2015, 2016) show the
Fig. 5. Primary porosity calculating theoretical model.
388
Marine and Petroleum Geology 98 (2018) 384–396
relationship between the void ratio (e) and different size particle con-
tents for a mixture. The void ratio is defined as the ratio of the void
volume to the solid volume. Formula (7) is the relationship between the
void ratio and the primary porosity.
φ= e/(1 + e) (7)
The mathematical model for calculated the void ratio can be ex-
pressed as formula (8) and formula (9).
e= ec Vc + ef Vf − a(1 + ef )Vf
(8)
For coarse-grain dominant packing texture.
e= ec Vc + ef Vf − bec Vc (9)
For fine-grain dominant packing texture.
Where e is the void ratio for a binary mixture, ec is the void ratio for
pure coarse particle packing; ef is the void ratio for pure fine particle
packing. The parameter a is termed the filling coefficient. And the value
range is between 0 and 1. The condition a = 1 corresponds to the
limiting situation where no change in the total volume occurs, and the
small particles perfectly fill the voids among the large particles. The
parameter b is termed the embedment coefficient. And the value range
is also between 0 and 1. The condition b = 1 corresponds to the change
in the void volume ΔVφ = 0, where large particles are perfectly em-
bedded in the small particles. The condition of a = 0 and b = 0 cor-
responds to the maximum change in the void volume in the mixture as
the different size particle contents change. For the same grains mixture,
the coefficients a and b are constant.
In regard to a polydisperse system (sands), the void ratio (e) would
also be governed by the particle size distribution and packing texture
(Cumberland and Crawford, 1987). However, this mathematical model
must be modified for calculating the primary porosity of modern sands.
3.2. A new mathematical model for calculating the primary porosity of
modern sands
The shape of the cumulative grain size distribution curve for
modern sands is “S”, and the grain size distribution curve is often a
Gaussian distribution, or two or more Gaussian distributions. The for-
mula for calculating the primary porosity should be established based
on the packing texture and the character of the grain size distribution
curve.
Y. Yan et al.
(1) When the hydrodynamic conditions in the depositional environ-
ment are relatively stable and flow is one-way, the grain size dis-
tribution is a Gaussian distribution as shown in Fig. 5a. This type of
sand is named type A. R is the particle diameter at the peak of the
grain size distribution curve. Based on the discussion in Section 2
and Section 3.1, the results show that when 2 > R/r ≥ 1, the
packing texture is similar when the content of different size parti-
cles changes; when r < R/2 and r > 2R, the primary porosity can be
calculated by formula (7), formula (8) and formula (9). To calculate
the primary porosity in Fig. 5a, the mixture can be divided into
three parts (0∼R/2, R/2∼2R and 2R∼+∞). Therefore, the mix-
ture can be thought of as a combination of the three parts. There-
fore, the packing texture is describes as follows: particles with
diameters between 0 and R/2 fill the void spaces among the par-
ticles with diameters between R/2 and 2R, and then, the particles
between 2R and +∞ are embedded in the mixtures with diameters
less than 2R. The formula for calculating the primary porosity for
the type A sands can be expressed as formula (10) and formula (7)
(ϕ = e/(1 + e) ) based on formulas (8) and (9).
e= e0 −a1 (1 + e0)vf − b1e0 Vc (10)
Where Vf is the content of particles with diameters between 0 and R/2;
Vm is the content of particles with diameters between R/2 and R; Vc is
the content of particles with diameters between 2R and +∞. e0 is the
void radio of single-size particle packing; a1 is the filling coefficient for
parts (particles sizes between 0 and R/2) and parts (particle sizes be-
tween R/2 and 2R); and b1 is the embedment coefficient for parts
(particle sizes between R/2 and 2R) and parts (particle sizes between2R
and +∞).
(2) When the hydrodynamic conditions in the depositional environ-
ment are relatively stable and flow is two-way, the grain size dis-
tribution is a mixture of two Gaussian distributions, as shown in
Fig. 5b and c. R1 is the particle diameter at the first peak of the
grain size distribution curve; R2 is the particle diameter at the
second peak of the grain size distribution curve. The value is
R2 > R1. Sands can be understood as a result of two Gaussian dis-
tribution mixtures. The sands can be divided in to four parts (par-
ticle sizes between 0 and R1/2, particle sizes between R1/2 and
(R1+ R2)/2, particle sizes between (R1+ R2)/2 and 2R2, and par-
ticle sizes between 2R2 and +∞). Vf1 is the content of particles
between 0 and R1/2; Vm1 is the content of particles between R1/2
and (R1+ R2)/2; Vm2 is the content of particles between (R1+ R2)/
2 and 2R2; Vc2 is the content of particles between 2R2 and +∞; Vφ
is the void volume of single-size particle packing; ϕ0 is porosity of
the sands; e0 is the void radio of single-size particles packing.
When Vm1 > Vm2 (Fig. 5b), the primary porosity calculation can be
divided into three steps which are binary mixture. This type sands is
named type B. First, the porosity of mixtures that are composed of
particles between R1/2 and 2R2 is calculated by formula (7) and for-
mula (8) as fine grain-dominant packing structures. Second, the por-
osity of mixtures that are composed of particles between 0 and 2R2 is
calculated by formula (9) and formula (7) as coarse grain-dominant
packing structures. Third, the primary porosity of the sand is calculated
based on the results of the first two steps and formula (9) as fine grain-
dominant packing structures.
Formulas, Formulas (7)–(9) lead to formula (11), and the primary
porosity can be calculated by formula (11) and formula (7):
e= e0 −a2 (1 + e0)Vf1 − b1e0 Vm2 − b2 e0 Vc2 (11)
When Vm1 < Vm2 (Fig. 5c), the primary porosity calculation can also
be divided into three steps which are binary mixture. This type of sands
is named type C. First, the porosity of mixtures that are composited of
particles between R1/2 and 2R2 is calculated by formula (7) and for-
mula (8) as coarse grain-dominant packing structures. Second, the
389
Marine and Petroleum Geology 98 (2018) 384–396
porosity of mixtures that are composed of particles between 0 and 2R2
is calculated by formula (9) and formula (7) as coarse grain-dominant
packing structures. Third, the primary porosity of sands is calculated
based on the result of the first two steps and formula (9) as fine grain-
dominant packing structures.
Formulas (7)–(9) lead to formula (12), and the primary porosity can
be calculated by formula (11) and formula (7):
e= e0 −a1 (1 + e0)Vm1 − a2(1 + e0 )Vf1 − b2 e0 Vc2 (12)
Where a1 is the filling coefficient for particles size between R1/2 and
(R1+ R2)/2 and particles size between (R1+ R2)/2 and 2R2; a2 is the
filling coefficient for particles size between 0 and R1/2 and particles
size between R1/2 and 2R2; b1 is the embedment coefficient for parti-
cles size between 0 and 2R2 and particles size between 2R2 and +∞.
(3) When the hydrodynamic conditions in the depositional environ-
ment are relatively complex and involve multidirectional flow, the
grain size distribution is a mixture of multiple-Gaussian distribu-
tions, as shown in Fig. 5d. R1 is the particle diameter at the first
peak of the grain size distribution curve; R2 is the particle diameter
at the second peak of the grain size distribution curve. R3 is the
particle diameter at the third peak of the grain size distribution
curve. The method for calculating the primary porosity are com-
plex, but it can be divided into more operable steps such as those for
type A, type B and type C.
4. Application of the primary porosity calculation model
4.1. Sample and primary porosity measurements
Forty samples from beach were obtained in eastern of Qingdao
(N35.890556°–35.901389°; E120.138889°–120.166667°), China. The
weight of each sample is 1 kg. The particles in the beach sands are
subrounded and well sorting. This study uses comparators provided by
Rittenhouse (1943) and Powers (1953) to obtain this textural data. The
shapes and grain sizes of particles in the beach sands were measured by
a BT-2900 Image particle size and shape analysis system made by
Bettersize Instruments Ltd (Shimizu, 2002). The particle mean size
(Md), median size (Mz), standard deviation (σ1), skewness (SK1) and
Trask sorting coefficient (S0) can be calculated from the grain size
probability curve measured by the BT-2900 instrument. The porosity of
the sands was measured by the Diagenetic Analog instrument. A dia-
gram of the apparatus, which was developed by the China University of
Petroleum Machinery Plant, is shown in Fig. 6. The principle for mea-
suring the porosity is the Boyle law (P
(Vpore + VCyliner) = P′(Vpore + VCyliner + Vcompare chamber)) (Bonnar,
1956). The result is shown in Table 1 φs is the measured porosity by the
Diagenetic Analog Meter, and φe is the porosity calculated by the em-
pirical formula of Scherer (1987).
4.2. Parameters and results
Based on the grain size distribution curve, 23 samples are classified
as type A sands; 9 samples are classified as type B sands; 8 samples are
classified as type C sands. The characteristic parameters (R1, R2, Vf1,
Vm1, Vm2 and Vc2) are obtained from the grain size distribution curve
measured by the BT-2900 instrument. φ is the primary porosity cal-
culated by the new method, and e is the void ratio for the mixture. The
relationship between e and φ is formula (7). The error is equal to φs-φ.
There are some unknown constants (e0, a1, b1, a2 and b2) in formula
(10), formula (11) and formula (12). Based on the experimental data,
the unknown constants can be obtained by Matlab data fitting. To
calculate the primary porosity of type A sands, we must know the
constants e0, a1 and b1. We can obtain particles with diameter are be-
tween 70 μm and 140 μm by the sieve method, and the primary porosity
of the mixture (particle sizes between 70 μm and 140 μm) is 44%, as
Y. Yan et al.
Fig. 6. The diagenetic analog meter.
measured by the Diagenetic Analog instrument. The parameters e0 is
0.79, as calculated from the primary porosity. Based on the quadratic
fitting, the parameters a1 and b1 can be calculated using samples2, 14,
15, 16, 18, 19, 20, 29 and 37, which have the same R1. The results show
that a1 is equal to 0.45, and b1 is equal to 0.70. The parameters a1, b1
and e0 are substituted into formula (10), and the void ratio (e) and
primary porosity (φ) can be calculated. The results are showing in
Table 2 and Table 3. The results show that the measured porosity (φs) is
close to the calculated porosity (φ). The maximum error is 1.84%
(Fig. 7).
To calculate the primary porosity of type B sands, we must know the
constants e0, a2, b1 and b2. The parameters e0 is the same as that for
type A sands. Using the Vf1, Vm1, Vm2, Vc2 and φs of type B sands in
Table 3, we can obtain the parameters a2, b1 and b2. The results show
that b1 = 0.29, a2 = 0.50, and b2 = 0.37. The parameters a1, b1 and e0
are substituted into formula (11), and the void ratio (e) and primary
porosity (φ) can be calculated. The results show that the measured
porosity (φs) is close to the calculated porosity (φ). The maximum error
is 1.84% (Fig. 7).
For type C sands, the constants e0, b2, a1 and a2 are also be acquired.
The parameters e0 is the same as that for type A sands. Using the data of
type C sands (Vf1, Vm1, Vm2, Vc2, φs) in Table 2, we can obtain the
parameters b2, a1 and a2. The results show that a1 = 0.45, a2 = 0.16,
and b2 = 0.90. The parameters a1, b1 and e0 are substituted into for-
mula (12), and the void ratio (e) and primary porosity (φ) can be cal-
culated. The results show that the measured porosity (φs) is close to the
calculated porosity (φ). The maximum error is 1.96% (Fig. 7).
5. Discussion
5.1. The practicability of empirical formula and new mathematical model
5.1.1. The published experiment data for verification the practicability
37 samples from the experiment of Westman and Hugill (1930) are
selected in order to verify the practicability of the new method in this
390
Marine and Petroleum Geology 98 (2018) 384–396
paper. The mixtures are composed of three grain sizes with mean par-
ticle sizes (4.476 mm (0.176 inch), 0.0711 mm (0.028 inch), 0.089 mm
(0.0035 inch)) that were artificially graded using a variety of sieves.
The mixtures are three series of binary mixtures and one series of
ternary mixtures. The data and results are shown in Table 4 and
Table 5. The results and data for the three series of binary mixtures are
given in Table 4. Using formula (7), formula (8), and formula (9)
combined with Matlab numerical simulation, the filling coefficient (a)
and the embedment coefficient (b) can be obtained. The results and
data for the ternary mixtures are shown in Table 5. The content of
medium particles is equal to or greater than 50%; the grain distribution
curve is close to the grain distribution curve for modern sands and is
approximated a Gaussian distribution. For coarse-fine mixtures, the
coefficient (a) is equal to 0.88, and the coefficient (b) is equal to 0.80;
for coarse-medium mixtures, the coefficient (a) is equal to 0.76, and the
coefficient (b) is equal to 0.68; for medium-fine mixtures, the coeffi-
cient (a) is equal to 0.75, and the coefficient (b) is equal to 0.76. The
diameter ratio of these three grain sizes is 50.5:8:1, and the values of a
and b for the coarse-fine mixtures are the largest, and these results
conform to the basic physical meaning of parameters a and b (see
Table 6).
5.1.2. The comparison of the calculated results by two methods
Firstly, the new mathematical model proposed in this paper is used
to calculating the primary porosity. The content of fine grains is greater
than the void space between the medium particles in the first three lines
of data in Table 5. Thus, this is the fine grain-dominant packing tex-
tures, and b1 is equal to 0.76, which can be obtained from the binary
mixtures in Table 4. When the mixtures of medium and fine grains are
mixed with the coarse grains, the material has a fine grain-dominant
texture. The coefficient b2 is equal to 0.68, which can be obtained by
the binary mixture in Table 4. Then, the porosity can be calculated by
formula (7) and formula (9). The content of fine grains is less than the
void space between the medium particles in the last five lines of data in
Table 5, so these mixtures have coarse grain-dominant packing
Y. Yan et al. Marine and Petroleum Geology 98 (2018) 384–396

Table 1
The test results of modern sandstone porosity and particle size parameters (40 samples).
Sample φ5 φ16 φ25 φ50 φ75 φ84 φ95 Md MZ σ1 SK1 KG S0 φs/% φe/%

17 −1.05 −0.99 −0.84 0.82 1.83 2.23 3.12 0.82 0.68 1.44 −0.01 0.64 2.52 29.9 29.99
17+ −0.61 0.52 0.98 1.7 2.35 2.79 3.53 1.7 1.67 1.19 −0.08 1.24 1.61 38.04 35.15
18 −1.33 −1.22 −1.07 −0.46 0.22 0.8 2.4 −0.46 −0.29 1.07 0.39 1.19 1.56 31.44 35.58
18+ −1.24 −1.13 −0.98 −0.15 1.13 2 3.05 −0.15 0.24 1.43 0.43 0.83 2.08 29.9 31.92
19 −1.31 −1.09 −0.85 −0.15 0.98 1.48 2.4 −0.15 0.08 1.2 0.32 0.83 1.88 30.93 33.06
19+ −1.11 −0.96 −0.72 0.04 1.11 1.61 2.5 0.04 0.23 1.19 0.29 0.81 1.88 34.02 33.06
20 −1.35 −1.26 −1.13 −0.46 0.72 1.68 2.48 −0.46 −0.02 1.31 0.49 0.85 1.9 34.54 32.97
20+ −1.11 −0.92 −0.7 −0.02 1.2 1.81 2.57 −0.02 0.29 1.24 0.38 0.8 1.93 31.44 32.79
21 −1.02 −0.43 −0.07 0.67 1.5 1.81 2.35 0.67 0.68 1.07 0 0.88 1.72 36.19 34.21
22 −0.61 0 0.28 0.94 1.55 1.81 2.29 0.94 0.91 0.89 −0.05 0.94 1.55 40.1 35.69
22+ −0.98 −0.43 0 1 1.76 2.02 2.53 1 0.86 1.15 −0.15 0.82 1.84 37.11 33.35
23 −0.07 0.57 0.87 1.37 1.85 2.09 2.57 1.37 1.34 0.78 −0.08 1.1 1.41 40.1 37.21
24 0.22 0.76 0.98 1.42 1.83 2.02 2.4 1.42 1.4 0.65 −0.07 1.05 1.34 41.24 37.97
25 −0.87 0.65 1.05 1.55 1.98 2.18 2.59 1.55 1.46 0.91 −0.28 1.52 1.38 38.14 37.47
26 0.5 1.11 1.33 1.72 2.11 2.33 2.7 1.72 1.72 0.64 −0.05 1.15 1.31 42.58 38.36
27 0.37 1 1.22 1.68 2.09 2.29 2.7 1.68 1.65 0.67 −0.09 1.1 1.35 43.2 37.84
28 −0.26 0.5 0.8 1.35 1.85 2.09 2.53 1.35 1.31 0.82 −0.11 1.09 1.44 40.31 36.84
29 −0.92 0.2 0.83 1.52 1.98 2.18 2.53 1.52 1.3 1.02 −0.38 1.22 1.49 40.21 36.26
30 −0.59 0.43 0.85 1.46 1.89 2.09 2.5 1.46 1.33 0.88 −0.28 1.21 1.44 40.72 36.84
31 0.39 0.98 1.2 1.68 2.11 2.33 2.81 1.68 1.66 0.7 −0.05 1.08 1.37 42.37 37.59
32 −0.74 0.02 0.33 1.02 1.65 1.92 2.46 1.02 0.99 0.96 −0.08 0.99 1.58 39.18 35.36
33 −0.89 0.28 0.65 1.46 2.09 2.4 3 1.46 1.38 1.12 −0.16 1.11 1.65 38.66 34.83
34 −0.46 0.2 0.5 1.09 1.68 1.92 2.46 1.09 1.07 0.87 −0.05 1.02 1.5 38.97 36.15
35 −0.7 −0.02 0.37 1.2 1.83 2.09 2.59 1.2 1.09 1.03 −0.15 0.92 1.66 40 34.71
37 −0.65 0.17 0.48 1.24 1.92 2.22 2.79 1.24 1.21 1.03 −0.07 0.98 1.65 36.19 34.83
38 −1.05 −0.11 0.28 0.98 1.68 1.98 2.63 0.98 0.95 1.08 −0.07 1.08 1.62 38.25 35.04
41 −1.02 −0.39 −0.11 0.7 1.65 1.98 2.63 0.7 0.76 1.15 0.07 0.85 1.84 37.63 33.34
42 −0.92 0 0.39 1.22 1.87 2.15 2.72 1.22 1.12 1.09 −0.15 1.01 1.67 36.19 34.61
43 −1 −0.22 0.28 1.31 1.96 2.24 2.74 1.31 1.11 1.18 −0.24 0.92 1.79 38.14 33.72
44 −1.15 −0.81 −0.15 0.74 1.39 1.72 2.37 0.74 0.55 1.17 −0.15 0.93 1.71 36.08 34.31
45 −0.83 −0.11 0.2 0.85 1.74 2.15 2.9 0.85 0.96 1.13 0.13 0.99 1.71 35.98 34.32
46 −1.13 −0.89 −0.52 0.24 1.61 2.2 3.03 0.24 0.52 1.4 0.3 0.8 2.1 31.03 31.84
48 −0.94 −0.11 0.17 0.65 1.18 1.48 2.22 0.65 0.67 0.88 0.02 1.29 1.41 37.22 37.10
49 −1.37 −0.09 0.3 0.94 1.55 1.87 2.61 0.94 0.91 1.09 −0.1 1.32 1.54 40.1 35.80
50 −1.06 −0.48 0.05 0.72 1.47 1.93 3.11 0.72 0.72 1.23 0.07 1.2 1.64 36.08 34.87
56 −0.98 −0.08 0.37 1.05 1.75 2.2 3.93 1.05 1.06 1.32 0.09 1.46 1.61 31.03 35.10
57 −0.83 0.17 0.65 1.55 2.25 2.65 3.86 1.55 1.46 1.33 −0.06 1.2 1.74 35.88 34.04
58 −0.64 0.4 0.86 1.7 2.69 3.18 4.51 1.7 1.76 1.48 0.08 1.16 1.88 33.92 33.08
60 −0.99 0.35 0.79 1.53 2.41 3.1 5.35 1.53 1.66 1.65 0.17 1.61 1.75 34.12 33.99
61 −0.52 0.49 0.93 1.7 2.83 4.17 5.74 1.7 2.12 1.87 0.32 1.36 1.93 35.88 32.79

Table 2
The test results of modern sandstone porosity and particle size parameters (Type A).
Sample Type R1/mm Vf1/% Vm1/% Vc1/% a1 b1 φs/% φ/% Error e

17+ A 0.28 12.85 69.68 17.47 0.45 0.70 38.04 37.15 −0.89 0.59
18 A 0.92 22.00 49.91 28.09 0.45 0.70 31.44 31.48 0.04 0.46
23 A 0.40 8.13 84.90 6.97 0.45 0.70 40.10 40.71 0.61 0.69
24 A 0.40 5.41 90.66 3.93 0.45 0.70 41.24 42.03 0.80 0.73
25 A 0.28 4.62 79.59 15.79 0.45 0.70 38.14 39.98 1.84 0.67
26 A 0.28 2.67 90.60 6.73 0.45 0.70 42.58 42.25 −0.33 0.73
27 A 0.28 1.96 93.74 4.30 0.45 0.70 43.20 42.88 −0.32 0.75
28 A 0.39 7.07 82.47 10.46 0.45 0.70 40.31 40.33 0.02 0.68
29 A 0.28 1.33 77.07 21.60 0.45 0.70 40.21 39.78 −0.43 0.66
30 A 0.28 1.20 79.15 19.65 0.45 0.70 40.72 40.20 −0.52 0.67
31 A 0.28 3.86 88.28 7.86 0.45 0.70 42.37 41.72 −0.65 0.72
32 A 0.39 6.07 75.08 18.85 0.45 0.70 39.18 38.94 −0.23 0.64
33 A 0.33 3.85 75.17 20.98 0.45 0.70 38.66 39.17 0.51 0.64
34 A 0.40 10.64 80.66 8.70 0.45 0.70 38.97 39.65 0.68 0.66
38 A 0.46 12.31 71.31 16.38 0.45 0.70 38.25 37.56 −0.69 0.60
43 A 0.28 6.52 66.92 26.56 0.45 0.70 38.14 37.18 −0.97 0.59
44 A 0.46 7.76 66.19 26.05 0.45 0.70 36.08 36.89 0.81 0.58
45 A 0.33 27.84 65.17 6.99 0.45 0.70 35.98 34.59 −1.39 0.53
48 A 0.65 12.28 79.98 7.74 0.45 0.70 39.36 39.36 0.00 0.65
49 A 0.46 9.65 83.61 6.74 0.45 0.70 40.10 40.32 0.22 0.68
50 A 0.65 20.12 64.32 15.56 0.45 0.70 36.08 35.21 −0.87 0.54
57 A 0.28 12.54 67.17 20.29 0.45 0.70 35.88 36.63 0.76 0.58
58 A 0.39 25.92 61.56 12.52 0.45 0.70 33.92 33.94 0.02 0.51

391
Y. Yan et al. Marine and Petroleum Geology 98 (2018) 384–396

Table 3
The test results of modern sandstone porosity and particle size parameters (Type B and TypeC).
Sample Type R1/mm R2/mm Vf1/% Vm1/% Vm2/% Vc2/% a1 b1 a2 b2 φs/% φ/% Error e

21 B 0.39 0.65 3.66 41.31 35.55 16.48 0.45 0.29 0.50 0.37 36.19 37.48 1.29 0.60
22 B 0.33 0.55 3.80 48.22 43.38 4.60 0.45 0.29 0.50 0.37 40.10 38.61 −1.50 0.63
22+ B 0.28 0.90 2.67 43.45 34.65 19.23 0.45 0.29 0.50 0.37 37.11 37.56 0.45 0.60
35 B 0.28 0.65 2.16 58.79 33.39 5.66 0.45 0.29 0.50 0.37 40.00 39.97 −0.03 0.67
37 B 0.28 0.65 4.12 58.25 32.15 5.21 0.45 0.29 0.50 0.37 37.98 39.34 1.36 0.65
41 B 0.28 0.78 4.68 47.38 35.57 12.37 0.45 0.29 0.50 0.37 37.63 37.69 0.06 0.60
42 B 0.28 0.39 3.07 40.69 37.56 18.68 0.45 0.29 0.50 0.37 36.19 37.21 1.02 0.59
60 B 0.28 0.38 10.03 41.47 34.01 14.49 0.45 0.29 0.50 0.37 34.12 35.04 0.92 0.54
61 B 0.28 0.38 12.14 43.01 32.39 12.46 0.45 0.29 0.50 0.37 35.88 34.54 −1.33 0.53
17 C 0.23 1.50 7.15 44.98 46.75 1.12 0.45 0.29 0.16 0.90 29.90 28.53 −1.36 0.40
18+ C 0.23 1.10 5.94 24.28 54.32 15.46 0.45 0.29 0.16 0.90 29.90 31.86 1.96 0.47
19 C 0.33 0.92 5.20 24.01 57.03 13.76 0.45 0.29 0.16 0.90 30.93 32.61 1.68 0.48
19+ C 0.38 1.10 5.75 29.16 57.26 7.83 0.45 0.29 0.16 0.90 34.02 32.57 −1.45 0.48
20 C 0.28 1.30 1.79 23.56 62.03 12.62 0.45 0.29 0.16 0.90 34.54 33.57 −0.96 0.51
20+ C 0.28 1.55 2.09 29.97 61.71 6.23 0.45 0.29 0.16 0.90 31.44 33.26 1.81 0.50
46 C 0.28 1.11 6.27 33.22 51.84 8.67 0.45 0.29 0.16 0.90 31.03 30.69 −0.34 0.44
56 C 0.33 0.55 6.45 26.95 50.71 15.89 0.45 0.29 0.16 0.90 31.03 30.63 −0.40 0.44

Fig. 7. The correlation of measured porosity (φs) and calculating porosity (φ) (The correlation coefficients between the measured porosity (φs) by the Diagenetic
Analog Meter and the calculated porosity (φ) by the new mathematical model is 0.93; The maximum error is 1.96%).

textures, and a1 is equal to 0.75, which is can be obtained from the is 0.7 by Rittenhouse (1943) chart; but the grains in these experi-
binary mixtures in Table 4. When the mixtures of medium and fine ments are spherical particle.
grains are mixed with the coarse grains, the material has a fine grain-
dominant texture. The coefficient b2 is equal to 0.68, which can be
5.2. Verification of empirical formula and new mathematical model in the
obtained from the binary mixture in Table 4. The results for calculated
modern beach sands
porosity (φ) are shown in Table 5. The measured porosity (φs) and
calculated porosity (φ) are plated in Fig. 8a. The calculated porosity
5.2.1. Analysis of the influencing factors of primary porosity in modern
(φs) correlates well with the measured result. The correlation coeffi-
beach sands
cient is 0.95, and the maximum error is only 1.41%. These results show
In this paper, the application of two primary porosity calculation
that the new method is reasonable.
methods in well-studied modern beach sands is discussed. The results
Then, the primary porosity is calculated by the empirical formula
show that the primary porosity has linear relationships with mean grain
(Scherer, 1987). The results for calculated porosity (φe) by empirical
size (Mz), skewness (SK1), standard deviation (σ1) and Trask sorting
formula (Scherer, 1987) are shown in Table 5. The measured porosity
coefficient (S0) (Fig. 9). However, the correlation coefficient is low
(φs) and calculated porosity (φe) are plated in Fig. 8b. The correlation
value. The SK1, S0, and σ1 are calculated by the characteristic para-
coefficient is 0.3, and the maximum error is only 20.84%. The corre-
meters in the grain size distribution curve. S0 is a function of P25 and
lation between the φs and φ is not well, and this method is unreliable.
P75. SK1, σ1 and Mz is the functions of φ5, φ16, φ25, φ50, φ75, φ84 and
The reasons for empirical formula unreasonable are as follows:
φ95. In other words, the primary porosity is a function of the char-
acteristic parameters in the grain size distribution curve. Therefore,
(1) The grains size in this experiment are non-continuous, and the grain
primary porosity cannot be estimated by any single parameter, such as
size distribution cannot be characterized by the sorting coefficient
S0, Mz, SK1.
(S0). The mixtures can only be divided into well-sorting mixture and
bad-sorting mixture by the sorting coefficient (S0). And this is
contrary to the fact. 5.2.2. Verification of empirical formula and new mathematical model in the
(2) The sphericity of samples in the Beard and Weyl (1973) experiment modern beach sands
The porosity of modern sands can be calculated by empirical

392
Y. Yan et al. Marine and Petroleum Geology 98 (2018) 384–396

Table 4 Table 6
Data from Westman and Hugill Binary mixture experiment. The appeared parameters in this paper.
C/% M/% F/% φs/% es a b φ% e0 Parameters Definition

100.00 – 0.00 37.70 0.61 0.88 0.80 37.70 0.61 R the diameter of coarse grains
90.00 – 10.00 31.50 0.46 0.88 0.80 31.75 0.47 r the diameter of fine grains
80.00 – 20.00 25.10 0.34 0.88 0.80 24.54 0.33 R/r the grain size ratio between coarse grain and fine grain
70.00 – 30.00 18.50 0.23 0.88 0.80 18.79 0.16 R1 the particle diameter at the first peak of the grain size distribution
60.00 – 40.00 23.60 0.31 0.88 0.80 23.31 0.24 curve
50.00 – 50.00 27.90 0.39 0.88 0.80 27.35 0.33 R2 the particle diameter at the second peak of the grain size
40.00 – 60.00 31.60 0.46 0.88 0.80 30.99 0.41 distribution curve.
20.00 – 80.00 37.80 0.61 0.88 0.80 37.27 0.57 R3 the particle diameter at the third peak of the grain size distribution
0.00 – 100.00 42.50 0.74 0.88 0.80 42.50 0.74 curve
100.00 0.00 – 37.70 0.61 0.76 0.68 37.70 0.61 Vφ The volume of porosity
90.00 10.00 – 32.00 0.47 0.76 0.68 32.61 0.48 ΔVφ the change of porosity volume
80.00 20.00 – 27.80 0.39 0.76 0.68 26.60 0.36 Vf the volume of fine particles
70.00 30.00 – 26.20 0.36 0.76 0.68 23.92 0.31 Vc the volume of coarse particles
60.00 40.00 – 27.00 0.37 0.76 0.68 26.35 0.36 Vf the content of particles with diameters between 0 and R/2
50.00 50.00 – 28.20 0.39 0.76 0.68 28.63 0.40 Vm the content of particles with diameters between R/2 and 2R
40.00 60.00 – 30.20 0.43 0.76 0.68 30.77 0.44 Vc the content of particles with diameters between 2R and +∞
20.00 80.00 – 34.80 0.53 0.76 0.68 34.70 0.53 Vf1 the content of particles between 0 and R1/2
0.00 100.00 – 38.20 0.62 0.76 0.68 38.20 0.62 Vm1 the content of particles between R1/2 and (R1+ R2)/2
– 100.00 0.00 38.20 0.62 0.75 0.76 38.20 0.62 Vm2 the content of particles between (R1+R2)/2 and 2R2
– 90.00 10.00 32.80 0.49 0.75 0.76 33.36 0.50 Vc2 the content of particles between 2R2 and +∞
– 80.00 20.00 29.00 0.41 0.75 0.76 27.70 0.38 Vf0 In the limiting case, all the added fine particles fill the pore spaces
– 70.00 30.00 26.70 0.36 0.75 0.76 24.49 0.32 among the coarse grains without altering the network of coarse
– 60.00 40.00 27.60 0.38 0.75 0.76 27.72 0.38 grains. The volume of the fine grains is Vf0
– 50.00 50.00 30.20 0.43 0.75 0.76 30.69 0.44 Vf0′ When the content of fine particles is large, the added solid volume
– 40.00 60.00 33.00 0.49 0.75 0.76 33.43 0.50 of fine particles is Vf0′
– 30.00 70.00 35.60 0.55 0.75 0.76 35.95 0.56 Vφ′ Vφ’ = Vφ-Vf0
– 20.00 80.00 38.60 0.63 0.75 0.76 38.29 0.62 φ/% the model-calculated primary porosity of the mixtures
– 10.00 90.00 40.40 0.68 0.75 0.76 40.47 0.68 φ0/% the primary porosity of single-size particle packing
– 0.00 100.00 42.50 0.74 0.75 0.76 42.50 0.74 φs/% the measured primary porosity
φe/% the calculated porosity by empirical formula
e the void ratio (or the packing density)
ec the void ratio for coarse particle packing
Table 5
ef the void ratio for pure fine particle packing
Data from Westman and Hugill three size spheres mixture experiment.
e0 the void radio of single-size particle packing
C/% M/% F/% a1 b1 b2 S0 φs/% φ/% φe/% a,a1,a2 the filling coefficient
b,b1,b2 the embedment coefficient
30.00 50.00 20.00 – 0.76 0.68 2.51 22.4 21.81 30.04 Md The particle mean size
20.00 50.00 30.00 – 0.76 0.68 2.83 23.6 25.01 29.01 Mz median size
10.00 60.00 30.00 – 0.76 0.68 2.83 24.6 24.75 29.01 σ1 standard deviation
30.00 60.00 10.00 0.75 – 0.68 2.51 27 27.17 30.04 SK1 skewness
20.00 60.00 20.00 0.75 – 0.68 1.00 23.8 22.97 43.81 S0 Trask sorting coefficient
20.00 70.00 10.00 0.75 – 0.68 1.00 29.4 29.36 43.81
10.00 70.00 20.00 0.75 – 0.68 1.00 25.8 25.41 43.81
10.00 80.00 10.00 0.75 – 0.68 1.00 31.2 31.42 43.81 25% and 75%, the results calculated by empirical formula (Scherer,
1987) are the same, which is a clear contradiction between the
calculated result and previous studies in Fig. 2. In the binary mix-
formula (Scherer, 1987) through using S0 in Table 1. The correlation
ture (Fig. 2), the primary porosity is lowest when the fine particle
between the result by empirical formula and the new proposed math-
content is approximately 30%. However, empirical formula
ematical model is showing in Fig. 10. The correlation coefficient is only
(Scherer, 1987) shows that the primary porosity is the same. The
0.63. The error for most samples is more than 3%. The maximum error
primary porosity increases, as the content of smaller grains in-
is 6%. The porosity of type A sands calculated by empirical formula is
creases (the content of smaller particles is over 30%). When the
lower than the measured porosity; the porosity of type B sands calcu-
content of smaller particles is the same in the binary mixture
lated by empirical formula is greater than the measured porosity; the
(Fig. 2b), the minimum primary porosity don't change when R/r is
calculated porosity of type C sands is the largest discrete degree relative
greater than 7. However, empirical formula (Scherer, 1987) shows
to the measured porosity. Fig. 7 is the correlation between measured
that the primary porosity decreases as R/r increases. The limit of
porosity (φs) and calculating porosity (φ) by the new proposed math-
the primary porosity by empirical formula (Scherer, 1987) is
ematical model. The correlations presented in Fig. 7 are valuable since
20.91%, but the experimental result is 14% (Fig. 2b). Based on
the regression is a good fit due to very high value of R-square. These
these analyses, empirical formula (Scherer, 1987) cannot show the
results also show that the new proposed-method is reasonable.
relation between the primary porosity and the grain size distribu-
The empirical formula (Scherer, 1987) is not well application in the
tion curve and primary porosity or the relationship between the
modern beach sands, the reasons include follows:
grain packing texture and primary porosity.
(3) The sphericity and content of the samples are difference. The sand
(1) The empirical formula (Scherer, 1987) only shows the relation
and silt in the Beard and Weyl (1973) were washed thoroughly to
between primary porosity and sorting coefficient (S0). Fig. 10 shows
remove all clay minerals. Silica flour, ranging in grain size from less
that the primary porosity is influenced by multiple factors (Mz, SK1,
than 0.044 mm to about 0.002 mm, was used for the finer fractions.
σ1 and S0). And the value of the correlation coefficient between
And The sphericity of samples is 0.7. The sphericity of samples in
primary porosity and S0 is low.
this study is 0.85 by Rittenhouse (1943) chart, and the sands and
(2) The grain size distribution curve cannot be characterized by the
silt were not washed but the original components were retained
sorting coefficient (S0), which is calculated by the P25 and P75. For
(Fig. 11). The silica flour is spherical, but the clay is tabular.
the binary mixture, when the content of fine particles is between

393
Y. Yan et al. Marine and Petroleum Geology 98 (2018) 384–396

Fig. 8. The calculated porosity and experimental porosity of Westman and Hugill Experiment (The calculated results by empirical formula (Scherer, 1987) are
showing in Fig. 8a; and the calculated results by proposed model in this paper are showing in Fig. 8b).

Fig. 9. The correlation between the experimental porosity and grain size parameters.

In this paper, the new method is based on the grain size distribution,
packing texture and their effects on the primary porosity. In the past
studies about binary and ternary packing shown in Figs. 2 and 3, the
primary porosity of the mixture is influenced by the grain size ratio and
grain composition. Fig. 2 shows that the primary porosity of the binary
mixture is influenced by the content of fine grains: if the fine grains
content does not change, the primary porosity is influenced by the grain
size ratio. Fig. 3 also shows that the primary porosity of a ternary
mixture is influenced by the content of different size particles and the
particle size ratio. Based on this conclusion, the new method is estab-
lished by the technique of controlling variables. When R/r is less than 2,
the content of grain size has little effect on the primary porosity of the
mixture. For calculating the primary porosity of sands, we can divide
the composition of sands into three parts, R, R/2 and 2R. Then, the
primary porosity can be calculated by the method of calculating the
primary porosity of a binary mixture. The advantage of the new method
is that the influencing factors (grain distribution and packing texture)
are considered. The effect of roundness and sphericity on the primary
porosity is transformed into systematic error. In addition, the accuracy
of calculating the primary porosity is improved.
In summary, when the sphericity of grains is 0.70 by Rittenhouse
(1943) chart, and the sorting is well, the empirical formula is reason-
able. This study only compared the application in the modern beach
sands and theoretical experiments. The application of the mathematical
model in the fluvial facies and deltaic facies is under study and will
continue to be published.
6. Conclusion
(1) The primary porosity of sands is mainly influenced by its sorting
coefficient, grains composition, packing texture, the roundness and
sphericity of grains. For one-component packing, the packing tex-
ture is vertically cubic and horizontally hexagonal packing; For
binary packing, the packing texture is the dominant network
composed by fine grains or the dominant network composed by
coarse grains; For ternary packing and multiple packing, the pri-
mary porosity is also affected by the packing texture.
(2) The newly-proposed mathematical method is based on the grain

394
Y. Yan et al.
Fig. 10. The correlation of measured porosity and calculating porosity by empirical Formula (Scherer, 1987) (The correlation coefficient between the measured
porosity and calculated porosity is only 0.63. The maximum error is 7%.).
Fig. 11. Sandstone sample particle morphology.
size distribution, packing texture and their effects on the primary
porosity. And, the accuracy of calculating the primary porosity is
improved. Based on the grains size distribution, the modern beach
sands in the study area are divided into three types. Each type
sample is further divided into several units according to the particle
size ratio. The mathematical model firstly calculates the primary
porosity of two adjacent units by using the binary packing model,
and then the primary porosity of the entire sand sample. The error
of model-predicted primary porosity ranges between −2% and 2%,
and the correlation coefficients are all above 0.9.
(3) The errors of empirical value and empirical formula are larger. The
empirical formula only shows that the single factor influencing on
primary porosity is sorting coefficient (S0) instead of sorting coef-
ficient (S0), mean grain size (Mz), skewness (SK1), standard devia-
tion (σ1), packing texture, the roundness and sphericity of grains.
One empirical formula is not practical for all types of sands. When
the sphericity of grains is closed to 0.70, and the sorting is well, the
empirical formula is reasonable. The newly-proposed mathematical
model for the primary porosity based on grain size distribution and
packing texture is more reasonable, and the applicable conditions
are more extensive.
Acknowledgements
This study is supported by the Strategic Priority Research Program
of the Chinese Academy of Sciences (XDA14010202), National Science
and Technology Major Project (2017ZX05008-004), and Graduate in-
novation fund of China University of Petroleum(R1701014A). We are
also thankful to four anonymous reviewers and associate editor for their
insightful comments and suggestions to improve the manuscript.
395
Marine and Petroleum Geology 98 (2018) 384–396
References
Atkins, J.E., Mcbride, E.F., 1991. Porosity and packing of Holocene river, dune, and beach
sands. AAPG Bull. 75, 339–355.
Ajdukiewicz, J.M., Lander, R.H., 2010. Sandstone reservoir quality prediction: the state of
the art. AAPG Bull. 94, 1083–1091.
Beard, D.C., Weyl, P.K., 1973. Influence of texture on porosity and permeability of un-
consolidated sand. AAPG Bull. 57, 349–369.
Berger, A., Gier, S., Krois, P., 2009. Porosity-preserving chlorite cements in shallow-
marine volcaniclastic sandstones: evidence from Cretaceous sandstones of the Sawan
gas field, Pakistan. AAPG Bull. 93, 595–615.
Bloch, S., Helmold, K.P., 1995. Approaches to predicting reservoir quality in sandstones.
AAPG Bull. 79, 97–115.
Bonnar, W.B., 1956. Boyle's Law and gravitational instability. Mon. Not. Roy. Astron. Soc.
116, 351–359.
Chang, C.S., Meidani, M., 2013. Dominant grains network and behavior of sand–silt
mixtures: stress–strain modeling. Int. J. Numer. Anal. Meth. GeoMech. 37,
2563–2589.
Chang, C.S., Wang, J.Y., Ge, L., 2015. Modeling of minimum void ratio for sand–silt
mixtures. Eng. Geol. 196, 293–304.
Chang, C.S., Wang, J.Y., Ge, L., 2016. Maximum and minimum void ratios for sand-silt
mixtures. Eng. Geol. 211, 7–18.
Cumberland, D.J., Crawford, R.J., 1987. The Packing of Particles. Elsevier, Amsterdam.
Ellis, A.J., Lee, C.H., 1919. Geology and ground waters of the western part of San Diego
County, California. Water Supply Paper 446, 121–123.
Folk, R.L., Ward, W.C., 1957. Brazos River bar, a study in the significance of grain size
parameters. J. Sediment. Res. 27, 3–26.
Fraser, H.J., 1935. Experimental study of the porosity and permeability of clastic sedi-
ments. J. Geol. 43, 910–1010.
Graton, L.C., Fraser, H.J., 1935. Systematic packing of spheres: with particular relation to
porosity and permeability. J. Geol. 43, 785–909.
Guo, X., Liu, K., Jia, C., Song, Y., Zhao, M., Lu, X., 2016. Effects of early petroleum charge
and overpressure on reservoir porosity preservation in the giant Kela-2 gas field,
Kuqa depression, Tarim Basin, northwest China. AAPG Bull. 100, 191–212.
King, F.H., 1989. Principles and conditions of the movements of groundwater. In: 19th
Auunual Report. U.S. Geol. Survey. Government Publishing, Washington, pp. 59–294.
Lade, P.V., Liggio, C.D., Yamamuro, J.A., 1998. Effects of non-plastic fines on minimum
and maximum void ratios of sand. Geotech. Test J. 21, 336–347.
Larrard, F.D., 2005. Concrete mixture proportioning: a scientific approach. Beijing Rev.
Y. Yan et al.
55 2-2.
Liu, Z., Shao, X., Jin, B., Li, H., Xu, X., Liang, Q.S., 2007. Co-effect of depth and burial
time on the evolution of porosity for classic rocks during the stage of compaction.
Geoscience 21, 125–132.
Mai, B.E.M., 2013. Diagenesis and quartz cement distribution of low-permeability Upper
Triassic–Middle Jurassic reservoir sandstones, Longyearbyen CO2 lab well site in
Svalbard, Norway. AAPG Bull. 97, 577–596.
Mcgeary, R.K., 1961. Mechanical packing of spherical particles. J. Am. Ceram. Soc. 44,
513–522.
Morad, S., Al-Ramadan, K., Ketzer, J.M., De Ros, L.F., 2010. The impact of diagenesis on
the heterogeneity of sandstone reservoirs: a review of the role of depositional fades
and sequence stratigraphy. AAPG Bull. 94, 1267–1309.
Pan, G., Liu, Z., Zhao, S., Huang, Z., Huang, X., 2011. Quantitative simulation of sand-
stone porosity evolution: a case from yanchang formation of the zhenjing area, ordos
basin. Acta Pet. Sin. 23, 249–256.
Pettijohn, F.J., Potter, P.E., Siever, R., 1972. Sand and Sandstone. Springer Publishing,
New York, pp. 618.
Powers, M.C., 1953. A new roundness scale for sedimentary particles. J. Sediment. Res.
23, 117–119.
Pryor, W.A., 1972. Reservoir inhomogeneities of some recent sand bodies. Soc. Petrol.
Eng. J. 12, 229–245.
Ridgway, K., Tarbuck, K.J., 1968. Voidage fluctuations in randomly- packed beds of
spheres adjacent to a containing wall. Chem. Eng. Sci. 23, 1147–1155.
Rittenhouse, G.A., 1943. Visual method of estimating two-dimensional sphericity. J.
Sediment. Res. 13, 79–81.
Scheidegger, A.E., 1958. The physics of flow through porous media. Soil Sci. 86, 355.
Scherer, M., 1987. Parameters influencing porosity in sandstones: a model for sandstone
porosity prediction. AAPG Bull. 71, 485–491.
396
Marine and Petroleum Geology 98 (2018) 384–396
Shimizu, I., 2002. Optical analogue image processing of shapes and/or sizes of particles
distributed three-dimensionally. J. Flow Visual. Image Process. 9, 2–3.
Taylor, T.R., Giles, M.R., Hathon, L.A., Diggs, T.N., Braunsdorf, N.R., Birbiglia, G.V.,
Espejo, I.S., 2010. Sandstone diagenesis and reservoir quality prediction: models,
Myths, and reality. AAPG Bull. 94, 1093–1132.
Terzaghi, C., 1925. Principles of soil mechanics: V-physical differences between sand and
clay. Eng. News Rec. 96, 912–915.
Thevanayagam, S., 2007. Intergrain contact density indices for granular mixes-
Ⅱ:Liquefaction resistance. Earthq. Eng. Eng. Vib. 6, 135–146.
Trask, P.D., 1930. Mechanical analyses of sediments by centrifuge. Econ. Geol. 25,
581–599.
Trask, P.D., 1931. Compaction of sediments. AAPG Bull. 15, 271–276.
Wang, J., Cao, Y., Liu, K., Liu, J., Kashid, M., 2017. Identification of sedimentary-di-
agenetic facies and reservoir porosity and permeability prediction: an example from
the Eocene beach-bar sandstone in the Dongying Depression, China. Mar. Petrol.
Geol. 82, 69–84.
Wang, Y., Cao, Y.C., Xi, K.L., Song, G., Liu, H.M., 2013. A recovery method for porosity
evolution of clastic reservoirs with geological time: a case study from the upper
submember of ES4 in the Dongying depression, Jiyang Subbasin. Acta Pet. Sin. 34,
1100–1111.
Westman, A.E.R., Hugill, H.R., 1930. The packing of particles. J. Am. Ceram. Soc. 13,
767–779.
Wilson, J.C., Mcbride, E.F., 1988. Compaction and porosity evolution of Pliocene sand-
stones, Ventura basin, California. AAPG Bull. 72, 664–681.
Ye, D., Zhang, J., 1990. Haphazard packing of spheres. Scientia Geologica Sinaca 2,
127–136.
Yilmaz, Y., 2009. A study on the limit void ratio characteristics of medium to fine mixed
graded sands. Eng. Geol. 104, 290–294.