Determination of pore size distribution
by suction porosimetry
G. F A G E R L U N D (1)
The theory and practice of suction porosimetry methods
(suction plate, pressure membrane and centrifuge) are reviewed
together with related methods based on the rate of penetration
of wetting liquids, and on static and dynamic fluid/fluid displace-
ment measurements.
1 - INTRODUCTION
The concept of " suction porosimetry " is used to
describe all methods where a wetting liquid is involved
in the measurements. Consequently not only pure
suction experiments are taken into account but also
methods based upon centrifugation and displacement
from the pores of a wetting liquid by another liquid
or a gas.
The basic equation is the same as for mercury
porosimetry viz the Laplace equation for the pressure
difference across the curved interface liquid-gas or
liquid-liquid.
(1 1) - - 2 a~jcos 0 (I)
AP = - - a~j + -- r However suction porosimetry brings certain addi-
where rl and r2 are the principal radii of curvature
of the meniscus and ~r~jis the surface tension between
fluids i and j. I f the i - - j interface meets the wall of
(1) The Lund Institute of Technology, Division of Building
Materials, Lund, Sweden.
a cylindrical pore of radius r with a contact angle 0,
then the relationship between the pore radius r
and the radii of interracial curvature rl and r2 is as
shown in eq. (1).
The negative value of Ap indicates that the pres-
sure is lower on the convex side of the meniscus.
The principal difference between suction porosimetry
and mercury porosimetry is the effect of the contact
angle between the liquid and the pore-wall, 0.
Mercury is almost nonwetting for which reason
it must be forced into the poles. In suction porosi-
metry the liquid is wetting and therefore penetrates
the pore system itself, and consequently must be
forced out of the specimen.
Suction porosimetry is applicable in the same
pore-size range as mercury porosimetry, that is, in the
intermediate range between very coarse pores where
visual phase integration is the only possible method
and very small pores where the Kelvin equation for
capillary condensation is suitable.
tional advantages :
- - The surface tension of a liquid like water is sub-
stantially lower than for mercury. Hence the
pressures exerted on the specimen during the
determination are considerably smaller.
- - Methods can easily be designed where anisotrcpy
of a specimen is taken into account.
191
VOL. 6 - N ~ 33 - 1973 . M A T i ~ R I A U X ET CONSTRUCTIONS

Watel can be used as measuring fluid, which from
some points of view is a great advantage, since the
interaction between water and pores influences
most materials properties.
Information on the pore shape can in principle
be obtained by determination of pore-size distri-
bution from desorption or drainage as well as
from absorption or imbibition curves.
Hence the capillary rise is a measure of the pressure
difference across the curved interface and conse-
quently of the pore radius.
The interracial curvature is also responsible for a
change in equilibrium vapour pressure of the pore
liquid, which is thus also related to the pore radius.
This fact forms the basis of Kelvin methods of pore
structural analysis (see p. 209).

The materials often exhibit considerable hysteresis
in calculation of pore-size distributions from wetting
of a material or from drying of the same material.
N o final answer can be given as to whether the " wett-
ing curve " or the " drying curve " gives the best
information on the pore structure. Hence it is re-
commended to use both types of methods at the
determination.
General works dealing with determination of pore-
size distributions where also the problem of wetting-
drying hysteresis is discussed are Dullien & Batra [10]
(with a comprehensive bibliography), SeweU & Wat-
son [24], Minnen & Vos [19], Vos & Tammes [26],
Bomberg [5-7] (also with a comprehensive biblio-
graphy) and de Boer [4].
2 - DIRECT SUCTION MEASUREMENTS
2.1 - Theory
As a consequence of the occurrence of a pressure
difference across a curved interface according to
eq. (1), liquid will rise in a capillary in contact with
bulk liquid as long as the contact angle between the
meniscus and the capillary wall is less than 90 ~, the
decreased pressure under the meniscus being balanced
by the liquid column. At equilibrium,
2 a,j cos 0
h -
(6 - - o~o)gr
where h is the capillary rise, g is the gravitational
constant and 60 is the density of the surrounding
medium. When this is a vapour 6o is negligible compar-
ed with 6, the density of the liquid.
2 a,s cos 0 (for instance by Hansen [19]) that at equilibrium
h --
6gr
Pressure difference A p is called the " s u c t i o n " o f
the capillary or the material. Schofield [22] introduc-
ed the concept p F as a measure of the suction of
water of a material, p F is defined as the Briggsian
logarithm of the possible capillary rise in a capillary
or a material, expressed in cm of water. Thus for h in
metres,
p F = 2 + log10 h (4)
The principle of the pF-concept may of course also
be used for other types of liquids, but then expressed
in cm or m of possible capillary rise of the actual
liquid.
Eq. (1) expresses the " capillary potential " at the
interface. For a horizontal capillary at rest the same
potential is acting throughout the filled part of the
capillary. For a vertical capillary at rest the potential
gradually decreases downwards from the interface.
W = -- Ap-- g x ~ (5)
where W is the capillary potential (N/m 2) and x is the
vertical distance from the meniscus to a point in a
capillary (m). However this effect is of importance
only if the capillary potential is small, e.g. in coarse
capilIaries. According to Luikow [17] it is negligible
if the pores are smaller than 0.01 mm, when water is
the measuring fluid. Normally, the specimen is so
small that it can be assumed that at equilibrium all
pore liquid has the same capillary pressure, and all
interfaces the same curvature. In centrifugation, on
(2)
the other hand, the additional effect of external
forces is not negligible (see section 2.4).
So far all equations are derived for single capillaries.
A porous material however consists of a complex
system of capillaries of a wide range of sizes, inter-
connected in a complicated way. It can be proved
(3)
all pores larger than that corresponding to the

TABLE I

AP h PIPs Methods
( N / m 2) pF (m) (%) of
A measurements

--0 - - oo 0 I00
- - 9.8.10 z 0 10-2 , ~ 100 1,530
-- 9.8.10 z 1 10-i ~-~ I00 153
-- 9.8.103 2 1 99.99 15.3
-- 9.8.104 3 10z 99.92 1.53
suction
- - 9.8.10 s 4 102 99.27 1,530 porosimetry
-- 9.8.106 5 1 oa 93.00 153
-- 9.8.107 6 104 48.43 15.3 Kelvin's
-- 9.8.108 7 105 0.07 t .53 law

192

\4oir conlent, "[o by weigk~
Z•0 I I 1 I ~ I
2ZO ~ ~ _ r , . j
zoo Lng_
180
160
140
IZO
B0
60
40
7_0
0 I 1 i i i
0 ~ Z 5 4 5 6 7
Fig. 1. m Suction-water content
pF curve for spruce at 20 oC [21].
applied suction value are empty, while all pores
smaller than the " equilibrium pore " are filled with
liquid of the same pressure. Hence the pore liquid
pressure at equilibrium with a certain suction value
is the same whatever the pore radius, and determined
by the radius of the largest filled pore.
It is now easy to calculate corresponding values
of Ap, pF, h, PIP, (from the Kelvin equation), and r
for any given measuring fluid. This has been done
for water at + 20 ~ in table I.
Similar tables can be calculated for other fluids or
for combinations of immiscible fluids.
In practice, the amount of liquid retained in the
specimen is measured at each suction value, expressed
as a pressure, Ap, as a capillary rise, h, or a relative
vapour pressure, PIP,. The relationship between
suction and liquid content at a given temperature is
called an adsorption or desorption isotherm when the
suction is expressed as a relative vapour pressure, and
a capillary equilibrium curve or moisture characteris-
tic, when the suction is expressed in pF-degrees or as
capillary rise. By use of the connection between
capillary pressure, Ap, and pore radius, eq. (1), the
pore-size distribution can be calculated from the
equilibrium curve.
Evidently calculation of pore-size distribution
from the Kelvin equation (and also calculation from
freezing point depression) is a sort of suction porosi-
metry. However the two types of porosimetry differ in
the following ways.
- - Methods of determination; calculation by Kelvin's
law is based upon the sorption isotherm. Suction
porosimetry is based upon measurements of capill-
ary pressure, that is capillary equilibrium curves.
q In suction measurements no consideration is
given to the adsorbed molecules on the pore wall.
With Kelvin's law this consideration is essential.
However the two methods overlap at the upper part
of the relative vapour pressure range which is shown
in table I. In this pore-size range the two methods
therefore give exactly the same results, if consideration
is taken to adsorbed molecules even in suction porosi-
metry.
G. F A G E R L U N D
line
Fig. 2. - - Suction apparatus. Range p F l - p F 3 [9].
The hysteresis between wetting and drying can be
extremely great. Figure 1 shows an example of this.
The hysteresis is to a great extent an effect of the pore
shape. In many cases, the filling of an interconnected
pore system may be determined by the radius of the
widest parts, whilst its emptying may be determined
by the constrictions or " necks " in the pores. Hence,
information on the shape of the pore system can
in principle be obtained by determination of wetting
curves as well as of drying curves.
The fact that hysteresis occurs must be taken into
careful consideration in the suction tests described
below. If a wetting curve is to be obtained, the speci-
men must always have a greater suction than the
porous plate on which it is put. I f not the specimen
will have a certain drying out before the test starts
which means that the measured curve will be a scann-
ing curve between the wetting and drying curves.
An excellent review of traditional methods of suction
measurements is given by Croney, Coleman &
Bridge [9].
2.2 - Suction Plate, pF0-pF3
Normally water is used as measuring fluid (see
Croney et al. [9].
Range pFl-pF3 :figure 2
A typical apparatus is shown in figure 2. The porou.,
specimen with or without water is placed on a porous
plate, a sort of filter. This contains moisture, since it is
in direct contact with bulk water. If the suction of the
porous plate is not equal to that of the material,
water is transferred until equilibrium conditions are
attained.
The suction is regulated by a vacuum pump. When
the air pressure in the flask is reduced, an equal
reduction of the suction of the porous plate occurs.
The resulting suction can be expressed
pF = loglo (z 9 102 + 1 354 Z) (6)
where Z is the mercury manometer height (m) and z is
the distance between the upper surface of the porous
plate and the water surface in the flask (m). Maxi-
m u m p F i s 3 since this corresponds to complete vacuum.
Range pFOopF1 to 2 : figure 3
In this low range a simplified version of the method
can be used. The suction is obtained by variation of
the depth z of the water surface below the specimen.
193
VOL. 6 - N * 33 - 1973 - M A T I ~ R I A U X ET CONSTRUCTIONS
Z
. . . . X
Fig. 3. ~ Suction apparatus. Range pF0-pF1 to 2 [9].
At these small suctions a certain effect of the specimen
thickness will occur, cf eq. (5). The suction will be
p F -= log~0 (z- 102)
Both methods mentioned are relatively slow.
Measurements of water content (by removing and
weighing the specimen) can only be made when
equilibrium is reached. A complete capillary equilibri-
um curve can be an affair of weeks or months.
Modified suction plate apparatus for rapid measure-
ments. Range pFO-pF3 :
In figure 4, a modified suction plate apparatus is
shown.
The specimen is brought to a certain water content
and placed on the porous plate. The specimen will
draw water from the water reservoir in contact with
the flow-tube. By decreasing the pressure in the flow-
tube the meniscus in this is kept fixed. The suction is
obtained from eq. (6). The method is rapid since
there is almost no moisture transport involved.
Some practical aspects of the method are given in
Croney et al. [9].
General discussion
The porous plate must fulfil certain requirements.
- - The pore size must be so small that air cannot pass
through when the maximum pressure difference is
reached (cf eq. (1) which gives a pore diame-
ter < 3 r for pF = 3).
- - The external surface must be plane so that intimate
contact is achieved between the specimen and the
plate.
Changes of temperature and atmospheric pressure
influence the measurements especially at small
suctions.
2.3 - Pressure membrane : pFO-pF4 to 6 material and a length of the cylinder o f 6 cm.
The suction plate method is limited to pF = 3 at
which suction water boils at room-temperature.
The method can be extended to pF = 6.2 by
putting the specimen chamber under pressures to
1.5 x 108 N/m 2 (1 500 atms). This pressure membrane
method has been described by Croney et al. [9],
Penner [21] and Bomberg [5].
An apparatus according to Penner is shown in
figure 5. The porous plate consists of a fine porous
pressure membrane which can be supported by a
coarse porous metal filter. The maximum pore-size
of the membrane can be calculated from eq. (1).
The suction as in an ordinaiy suction plate apparatus,
is now obtained as the pressure difference across the
194
Fig. 4. - - Modified suction apparatus for rapid measurements.
pF0-3 [9].
pressure membrane. Water expulsion and water
intake are measured by the change of the meniscus
in the pipette. Calibrations must be made o f the
(7) pressure-expulsion curve o f the porous membrane.
2.4 - Centrifuge : pF3-pF4.5
The method, shown in figure 3, is limited to pF = 1
to 2 which corresponds to a water column of 0.1 to
1 m. The suction is determined by the gravitational
field. By increasing the external force field, for instance
by centrifugation, a certain water column signifies a
greater suction. Eq. (3), (1) and (4) are transformed to
2 ao cos O
h -- (8)
~ngr
AP = - - o n g h (9)
pF=loglo(nh • 10e) (10)
where n is the ratio of the centrifugal field to the gravit-
ational field.
The centrifuge method has been described by
Croney et al. [9] and by Ksenzhek et al. [16]. The appa-
ratus used by Croney et al. [9] is shown in figure 6.
The specimen is supported b y a porous cylinder
which is in contact with bulk water. It is essential to
keep the water level constant. This is done by means
of a drain hole in the specimen jacket.
Normally only desorption tests are performed.
This means that the water is d r a w n from the specimen
into the porous cylinder. This is a relatively slow
process, and in the case of a coarse-porous cylinder
may involve merely a vapour phase transport since the
ambient medium is almost dr2,.'. On the other hand,
a fine porous cylinder may remain wet, but its perme-
ability is low. Measurements o f moisture content cannot
be done until the moisture d i g r i b u t i o n throughout
the cylinder corresponds to equilibrium conditions.
The authors recommend hard chalk as a suitable
When the angular velocity is <,~the mean acceleration
at the distance R2 q- R1 from the centre of rotation is,
2
Rz + Ri
ng= 2 - o~ (11)
and the height of capillary rise, h, is,
h = R2 - - R1 (12)
Hence the suction is
pF = logio ( R ~ - - 2 / ~ r,~ZgY. 10z) (13)
 G. FAGERLUND

The capillary equilibrium curve is obtained by
weighing the specimen after each increase in suction.
Drying curves as well as wetting curves can be obtained.
The suction at rest is about I pF. Hence a specimen
having initially a greater suction will increase its
moisture content when it is placed upon the cylinder.
This water is removed during acceleration but it may
have a considerable effect on the results because of the
hysteresis between wetting and drying. In the same
way a change in equilibrium may occur when the
centrifuge is stopped. The specimen thickness has an
influence on the results, cf. eq. (5). Hence it must not
exceed 5 mm.
Ksenzhek et al. [16] have refined the centrifugal
method considerably. In their method consideration
can also be taken of the distribution of liquid through
the thickness of the specimen. Besides the method is
speeded up considerably. Only desorption curves can
be obtained, however.
The authors use a cuvette as specimen support in
which the specimen is not in direct contact with a
water column. The water displaced from the specimen
is collected in the cuvette where it is measured after
stopping the centrifuge. It can also be measured during
the run in a very elegant way : The specimen is
impregnated with an electrically conducting liquid and
the amount of liquid entering the ring-formed cuvette
is measured by the electric resistance between two
platinum electrodes.
Since the specimen thickness is small the acceler-
ation can be assumed to be the same in all cross-
sections of the specimen. All pores are assumed to be
cylindrical and parallel to the radius of rotation.
At saturation all pores are completely filled. When
the centrifugal acceleration is increased the capillary
rise in the coarsest pore is gradually decreased accord-
ing to eq. (8). A certain acceleration corresponds to a
certain pore radius, rn, where the capillary rise h is
equal to the specimen thickness, H.
The radius r~ is obtained from eq. (14)
2 cru cos 0
r, --
0ngH
Pores of radius r > r, are partly filled to an extent
obtained from eq. (8). However the authors show
theoretically that the pore-size distribution func-
tion, f(r), is independent of the thickness of the
specimen.
f(r)= (r~) d' e~
- - "dr I
where e0 is the pore-volume (m 3) and e, is the liquid
retained in the specimen at a certain acceleration (m3).
Hence, by determination of e~ at a certain acceler-
ation, n 9 g, as function of the pore-radius, r~, calcul-
ated from eq. (14), the pore-size distribution can
easily be determined by eq. (15) for instance by a
simple graphical method.
The method has been compared with mercury
porosimetry. The agreement was satisfactory.
3- RATE OF PENETRATION OF WETTING
LIQUIDS
The pore-size and, subject to certain assumptions,
its distribution can also be determined by measure-
ments of penetration of liquids in a porous specimen.
The connection between time, t, and linear e:;tent,
l, of capillary uptake of water of viscosity /z, in a
straight cylindrical capillary of radius r is, for laminar
flOW,
2/zl 2
t -- (16)
(r O r c o s 0
Hence by measurements of time, t, and depth of
penetration, l, the pore-size can be determined.
The complications in application to real materials
are :
- - T h e pores are not straight cylinders but each
capillary has a random variation of pore radius
along its axis.
- - The pores have an axial " tortuosity " or " mean-
dering ".
- - The pores are interconnected.
- - T h e method is accompanied by difficulties of
measurements of penetration.
Astbury [2, 3] has elaborated a methed where
consideration is taken of all complications listed above
except the interconnection of individual pores.
The principle of the apparatus is as follows. The
(14) dried prismatic specimen is placed in a plastic box.
The outer surface of the specimen is painted with a
silver paint which creates a porous electrode. Another
electrode is placed in the box. The resistivity of the
specimen is measured after which an electrolyte
(e.g. decinormal aqueous solution of Ca Clz or
(15)
--~F~F ~ Cenlr%oi_r_oJo!io_n_
i ,
I

Moisture block Pressure r

!i j I
=
r
i
Rz A1 i
I
o ....... l, J'l II :i i So~p~
--4-~" Porous c@nder
Is
\Porou~ plole holder
Fig. 5. - - Pressure membrane apparatus [21]. Fig. 6. - - Centrifuge according to Croney et al. [9].

195
VOL. 6 - N ~ 33 - 1973 - M A T # R I A U X ET CONSTRUCTIONS
2. Xo
14
Fig. 7. - - Astbury's pore model [3].
K2 SO4) is introduced into the box. According to
eq. (16) the rate of capillary rise is higher the coarser
the capillary. Hence when the electrolyte in the coarsest
connection through the specimen has reached the
silver electrode the resistivity is changed. A curve of
conductance against time is obtained. The amount of
electrolyte absorbed is measured by maintaining the
electrolyte at a fixed level by filling from a burette.
The measurements are continued until a steady state
is reached. The apparatus is shown in detail in Ast-
bury [3].
If all capillaries were straight cylinders it would be
possible to calculate the pore-size distribution from
eq. (16) and the conductance time curve. However
Astbury [2] shows that this assumption is erroneous.
Instead he evaluates a pore-size distribution from his
own pore model shown in figure 7. Its main character-
istics are :
- - All pores are geometrically alike. However they
differ in size.
- - All pores are meandering. Their tortuosity (effective
length) is ft. H where fl > 1 and H is the thickness
of the specimen.
- - A l l pores are cylindrical. The radius is varying
along the pore axis.
- - The appearance of radius variation is the same in
all random pore sections.
- - The pore-walis are completely wetted by the fluid
and the Hagen-Poiseulle law is valid.
The basis of calculation is three theoretical express-
ions describing in terms of the pore model, the time
to fill a single capillary, the total accessible pore
volume of the specimen and the permeability at
saturation given by the Hagen-Poiseulle law.
Thus two measurements are needed :
- - The time function of capillary uptake of liquid.
The permeability.
As mentioned earlier all measurements are made as
conductivity measurements. The working equations
must therefore be transformed to measured quantities.
It is also necessary to make an explicit assumption
about the way in which the radial variations are
distributed along the capillary. Astbury [3] suggests
a rectangular distribution which means that any
radius between the smallest and largest is equally
probable. The evaluation of parameters is discussed
more in detail in [3]. The result is two radii distribut-
ions which define an area within which the real pore-
size distribution is situated, see figure 8. The real
pore-size distribution can also be calculated [3].
It lies closer to the coarsest of the two distributions
196
Poro~ds (%) <a or Za
60 ~ I o l.z~ [
0.o~ IO~i 1 ~0 ~OOaorZa
o z~ (~)
Fig. 8. - - D i s t r i b u t i o n o f smallest and coarsest accessible radii [3].
which is reasonable since the coarser pores represent
more pore volume (see fig. 9). It is of course also
possible to calculate the specific surface from the same
measurements.
The method is applicable to pore radii down to
0.05/~m according to Astbury [3].
The method has been compared with the water
expulsion method and with mercury porosimetry
by Astbury [2, 3] and Clements & Vyse [8]. In general
it indicates a coarser pore-size distribution than do the
other methods. This is logical since in the other two
methods the pore-size is determined by the most
narrow passage into the pore. It is interesting to notice
that the mercury porosimeter distribution often
agrees very well with Astbury's distribution of smallest
accessible radii [3]. Astbury's method on the other
hand, gives a more true pore-size distribution since
" radial tortuosity " is considered.
Asbury writes : " The present model is no easier
to accept than others which have preceded it. It does,
however, introduce a new dimension into the problem
and it does predict the existence of large cavities not
shown by the water-expulsion and mercury-penetrat-
ion models "
The penetration of the water front can also be
detected by other methods, for instance by radio-
isotopes, G~mesi [13].
4 - DISPLACEMENT METHODS
4.1 - Introduction
Eq. (1) also indicates the pressure needed in order
to displace the liquid in a cylindrical capillary by
another liquid or a gas.
-- 2 a~j c o s 0
Ap -- (1)
r
Hence the pore-size distribution can be obtained
by successive displacement of liquid from a saturated
material. This is the basis of the so-called displacement
methods9
Normally the medium used for displacement is a
gas (air) but it is also possible to use a liquid which
is immiscible with the liquid in the pores. The flow
must be uni-directional. Hence the faces of the
specimen parallel to the flow must be sealed. The
method is especially suitable for indicating anisotropy
of the pore system by determination of pore-size
distributions in three perpendicular directions of the
same specimen.
The displacement methods are by nature desorption
methods only9

Volume traction
/ ,./~z,, Poa_o{.-,~_.,.L~,~ pg,:. ~_i.. ~,.,-,,<
L_ _/ .y__,l,... ,
r/Z. r Rodiu~
Fig. 9. -- Real pore size distribution and dlstribution of smallest
and coarsest accessible radii [3].
4.2 - Static methods
Clements & Vyse [8] report a method where water
is expelled by air frcm an initially saturated specimen.
The amount of water expelled after each increase in
air-pressure is measured by weighing. The method is
compared to determination by mercury porosimeter.
The two methods give roughly similar results in the
pore diameter range 20 # m - - 85 ffm, when applied to
refractories. However the mercury porosimeter under-
estimates the amount of pores < 20 fire. This is cer-
tainly an effect of wetting-drying hysteresis. Hence the
two methods supplement each other.
The authors also discuss the effect of duration of
pressurisation on the pore-size distribution. " Since
the rate of water movement through a pore is infinitely
small when the applied pressure is just equal to the
surface tension force, water will exude from a test
specimen for infinite time " '
The authors find that pressurisation for 1 minute
gives an error in pore radius of as much as 0.3 ffm
for a specimen 2.5 cm thick. The effect of duration of
pressurisation has been theoretically studied by
Manegold et al. [18]. Some experimental data shown
in figure 15 will be discussed later.
4.3 - Dynamic methods
The amount of fluid expelled from the specimen and
thus the pore-size distribution can also be obtained
indirectly by measurements of change in permeability.
Such methods have been suggested and described by
Erbe [ 11], Grabar & Nikitine [ 14], Manegold et al. [ 18],
Zagar [27, 28, 29], Schwiete & Ludwig [23], Alviset
& Liger [1] and Ohnemfiller [20]. Since the calculation
of pore-sizes is based upon data of permeability the
structure of the pore-system has a great effect on the
Wo~'ee
--\
pre~m~
~pec,men /~bululakohol
Fig. 10. - - Apparatus for displacement liquid-liquid [18].
G. F A G E R L U N D
resulting pore-size distribution. A comprehensive
discussion of permeability of porous materials together
with 134 literature references is provided by Timo-
feev [25].
The methods can be divided in two groups according
to whether the displacement medium is a liquid or
a gas.
4.3.1 - Liquid - Liquid displacement
The method is described in detail by Manegold
et al. [18].
The principle is simple. Consider a material with
parallel cylindrical capillaries of three separate sizes,
radii rl > r2 > r3. The material is saturated with a
liquid, which will be displaced by another liquid,
immiscible with the first. Suitable liquids are for
instance water + isobutylalcohoI or water -f- methyl-
isobutylalcohol (Manegold et al. [18]). A suitable
apparatus is shown in figure 10.
A hypothetical diagram of the relationship between
flow rate and pressure is shown in figure 11. As the
pressure of the displacement medium is gradually
increased, the permeability will remain zero until a
pressure AP1, is reached, corresponding to pore
radius rl according to eq. (1). Then the pores of
radius r~ are emptied and there is a sudden increase
in permeability. The coefficient of permeability is
then constant until the pressure AP2 is reached corres-
ponding to pores of radius r2, and there is a new sudden
increase in permeability. Between each jump in perme-
ability and the next the connection between flow rate
and pressure is a line through the origin. The slope
of the line, or coefficient of permeability, is determined
by the Hagen-Poiseulle law :
N~
i 8 H-~-~ " AP (17)
or
v = B~ Ap (18)
where v is the flow rate, N is the number of capillaries,
H is the specimen thickness, Ap is the pressure
difference across the specimen and B is the coefficient
of permeability.
V
// /~"
/ /
/, !()i
oir"
"- ,, p ( N / g )
P, ~ Pz /" P~
r q rz r3
Fig. 11. - - Flow rate pressure curve at displacement liquid-liquid
in a material with three pores of different size.
197
vot.. 6 - N o 33 - 1973 - MATERIAUX ET CONSTRUCTIONS

V / hensive information on the length of the " waiting

period " which can be very long in the case of long
capillaries and low pressures.
/ ,
;2 i
4.3.2 - Liquid-gas displacement
Consider the displacement of water by air. I f the
displaced water is removed from the specimen surface,
the curve of flow rate against pressure will be of the
same type as when a liquid is displaced by a liquid,
that is an S-shaped curve tending at high pressures
~<"~ /</--%,r.
/. " ~- ! ] I
to a line intersecting the origin. If however the free
,/'/J-2 / i '~
specimen surface is covered by a layer of water the
A Pmin ~ ~ &Pi*l '~ Pmo~ flow rate-pressure curve will have a different shape,
since the air cannot leave the specimen until a bubble
r rm~ x ri rt,1 rmin can be formed at the end of the capillary. The necess-
ary bubble pressure, APm is obtained from eq. (21).
Fig. 12. -- Flow rate-pressure curve at displacement I/quid-
liquid in a real material 2 eu cos 0
Ap~ -- (21)
r

This pressure is the same as that necessary to

In a real material the pores are of all sizes and certain- displace the liquid from the pore.
ly not parallel. In this case a continuous curve will be Consider again the material represented in figure 11.
obtained (fig. 12) being typically S-shaped. At high The specimen is completely sealed by the water until
pressures the curve is linear and parallel to the curve the pressure APz is reached. At a pressure A p > API
of an initially dry material. The point where this line air flows through the pore of radius rz. However
begins corresponds to the smallest permeable capillary because of the necessary bubble pressure A p a the
and the intersection with the AP-axis corresponds to flow rate pressure line intersects the pressure axis
the coarsest permeable pore. at A p = Apz. When the pressure is increased further
The real curve can be transformed to a stair-formed until Ap -----A P z air begins to flow through the capill-
curve, as in the figure. The coefficient of permeability aries with radius r2. The flow rate pressure line through
at each step is calculated from the slope of a line from those capillaries intersects the pressure axis at
the origin to the actual point of the curve. The pores Ap _=--APe. When the flows in both types of capillaries
are divided in pressure classes Ap~ - - AP~+I, etc. are added the result is an abrupt increase in slope
By transforming the pressure into a radius accord- at A p = AP2 (see fig. 13).
ing to eq. (1) the amount of pores of each size class The result of a test on a real material will be a
r~ - - n+l, etc., can be calculated from the coefficients curve which is concave (see fig. I4). At A p > Ap,,~,~
of permeability. If all pores are assumed to be cylind- the curve will be linear with slope equal to the
rical, parallel to the flow and of length equal to the air permeability of the dry specimen. A p = APm,~
specimen thickness, eq. (19) con be used for calculation corresponds to the smallest permeable pore while
of cumulative pore size distribution. It is assumed that A p = APmi, (intersection with the A p _-- axis)
the difference r~ n+~ ~s small.
- -

corresponds to the coarsest permeable pore.

1 8 H z The intersection of the linear part of the curve
1 Ee~ -- Z ( B ~ Bi-0 ff (19) (Ap > &p,~,~) on the &P-axis signifies an " e q u i -
E0 EO r~
valent pore-size ", Zagar [27].
where eo is the pore volume.
The calculation starts with the coarsest pore.
The amount of pores of class r~ - - r,+~ can also be V
obtained from the change of the slope of the line /
between segments r~-z - - r~ and r~ r~+~. This change - -

is Bi ~ B~-~ and when the interval n - - r,+~ is small

the amount of pores is :
/
/ Flow lhrough copillorie~
(B~ - - B~-I) 8 Iz H 2
/ o, o79
e~ = (20)
d I //f / " r~
The foregoing discussion implies that the interface
liquid-liquid has penetrated the entire pore. During
the penetrating period the coefficient of permeability
is determined by viscosities of both liquids. Reliable / ',-/ ~J - ~ p (,s/m)
measurements cannot be made until this first period P~ Pz P~
has passed. Even at a constant pressure difference ! I i
across the specimen the flow rate is never quite constant. r G rz r3
It will only approach a limiting value asymptotically.
The significance of this phenomenon is discussed in Fig. 13. - - Flow rate-pressure curve at displacement liquid-gas
detail by Manegold et al. [18], who gives compre- in a materiel with three pores of different size.

198
 G. FAGERLU ND

Zagar [28] gives a profound discussion of the method Flow ro~e- cons~ .perm~bildL3
showing theoretically that rm,~ is equal to the " effect- - ~--.--o & P - 3 Z S 10 N/m
s

ive pore radius ", r ~ , which in turn is a pore radius

that in a hypothetical equal-pore-sized material /
/
gives the same permeability as in the real material /
when consideration is taken of the fact that only a /
part, 8elf, of the total porosity, 8o, is permeable.
He shows how the pore system may be characterized
by calculation of permeable and non-permeable I ,- Conlinou~ lncreo~in [:x'e,bsure.
porosity, the " pore shape factor " and the effective .... f.orlier pressure,~i~,, relieved
pore length. ___._~!ore. o_s n.e~. p~s,sur~

The calculation of pore-size distribution from the 9 ~ P - Z.0O- 10s N / m z

flow rate pressure curve can be made graphically S Z

---t Ap--1.7S'10 N/m
(~agar [28]) or analytically (Alviset & Liger [1]).
The principle is essentially the same as for calculation - _- II z~p=l.50.~0 ~ N / m 2
from liquid-liquid displacement except that it is - - 4 Lx p = I. 25. I 0 ~ N/m ~ ~ P - Io0.10 N/mz
s
necessary to take molecular slip flow (Knudsen flow) 0 ~.~----I.--'1-'------ I 4 Ill ~ p = 0 3 5 . 1 0 5 N/m ~"

into consideration for pores > 1/zm. 0 1 Z i kJ~ h m e (h)

~'\This ~or}mognilied below
The Hagen-Poiseuille law corrected for slip takes
the form Flow role. consl 9 per~obihl~

--~r4N -p- 9 ~ p ( I (22) • x --x-- ~ P=1.25qO s N/mz

v = 8fill Po
..x/
/
where ( is the coefficient of slip. This coefficient is /•
dependent on the gas used, its pressure and temper- x/
ature, since it is = 2/3 of the mean free path of gas /
molecules. Timofeev [25] gives further information P - 1 0 0 . 1 0 ~ N/m ~

of the value of ~. Eq. (22) shows that the slip correct- ~'P'O35"los N/m z

ion is considerable when the pores are small. Po is the

pressure under which the volume of the penetrating
gas is measured,/~ is the mean pressure in the specimen
(eq. (23)) and Ap is the pressure difference across the I Z Voil,n9 time(h)
specimen. The factor P/Po is introduced to correct
for change in density of the gas within the capillary Fig. 15. - - Effect on p e r m e a b i l i t y o f w a i t i n g - t i m e at w a t e r - a i r
Ap displacement. Clay brick, size 30 cm ~ thickness 30 m m ,
P = Po + -~- (23)

The coefficient of permeability in a small interval

r~ - - r~+~ can now be expressed If all capillaries are cylindrical and have a length
equal to the thickness of the specimen, the number
of capillaries N~ can be changed to a pore-volume e~
8,uH P-o 1 + 4 (24)
8~
N, = ~ (25)
=r~H
B~ is measured from the curve as the slope of the
V tangent at i, and r~ is calculated from eq. (1). Hence
1
the cumulative pore-size d i s t r i b u t i o n , - Z el, can be
~0
calculated by insertion of eq. (25) in eq. (24).
/i

1 Ze~ 1 Z(Bi+I--B 0 8 f i l l zPo 1 (26)

The volume of pores, 8~, with pore radius (n)eq~ can

be calculated by inserting in eq. (24) the difference in
,
,
i /
/ I
i
slope of the tangents at i and i + 1.
Hence :
e~-- ( B i + t - - B 0 8 # H 2 Po 1
I I I ! I (rOZequ "~" ( 1 + 4 ~ ) ( 2 7 )
r rrnox ri (ri)eq'2"_ req~ rmin =rMi~

Fig. 14. - - Flow rate pressure curve at displacement liquid-gas Zagar [28] takes no account of Knudsen flow in his
in a real material. graphical treatment; this can be reasonable as long as

199
VOL. 6 - N* 33 - 1973 - M A T I ~ R I A U X ET C O N S T R U C T I O N S
the measured pores are relatively coarse ( ~ I /~m).
The effect on flow rate of change in density of the
gas is neglected by both Zagar [28] and Alviset &
Liger [1].
Zagar [28], [29] discusses reproducibility and the
effect of specimen thickness. The reproducibility was
satisfactory. The measured rmln was smaller the
thinner the specimens. However the effect was less
than 4 ~ .
The waiting period between change of pressure and
attainment of constant flow-rate was determined
theoretically by Manegold et al. [18]. They found that
the waiting period in displacement of water by air is
considerably shorter than for liquid-liquid displace-
ment. On the other hand the high surface tension
between water and air makes the pressures needed
very high. The present author has determined absolute
values of the waiting period. Some results are shown
in figure 15. The material is a well burnt clay brick
with thickness 30 m m and surface area 30 cm 2.
Normally the pressure difference, AP, is increased
without relief of the earlier pressure difference. Despite
this a reasonable waiting period at low pressures
RI~SUMI~.
D~termination de la distribution du diam~tre des
pores par la m~thode de la succion. - - L a porosimdtrie
par succion repose sur la mesure de la dJpression
capillaire que suscite un liquide mouillant retenu clans
un milieu poreux. C'est donc une mdthode compHmen-
taire d celle de la porosimJtrie par pression (par ex.
pJndtration de mercure) oit l'on opdre avec des [iquides
mouillants. On expose la thJorie et la pratique des
REFERENCES
[1] ALVISET L., LIGER C. La mesure de la rdpartition
- -
des capillaires dans un tesson de terre cuite et les
applications que l'on peut en faire. Proc. VI, Int.
Ceram Congr., Wiesbaden, 1958, 401.
[2] ASTBURY N.F. - - A model f o r connected porosity
in ceramic bodies. Trans Brit Ceram, Soc., 67,
Aug. 1968, 319.
[3] ASTBURYN.F. - - Neues Verfahren zur Untersuchung
der Poren-GrOssen-verteilung. Ber Deuts Keram
Ges, 49 (1972)/2/, 52.
[4] de BOER J.H. - - The structure and properties o f
porous materials, Everett & Stone (Editors). Tenth
Colston Syrup., Bristol 1958, Butterworth, London,
1958.
[5] BOMBERGM. - - Water flow through porous materials.
Part I : Methods o f water transport measurements.
Report 19, Division of Build. Techn. Lund Inst.
of Techn. 1971.
[6] BOMBERG M. - - Part II : Relative suction model.
Report 20, Div. of Build. Techn Lund Inst. of
Techn., 1971.
200
might be as long as half an hour, shorter at higher
pressures. Equilibrium conditions are not assured
even after 2 hours.
In three cases the earlier pressure was relieved for
some hours before application of the new pressure
(broken lines). In these cases the waiting periods are
very long. This method must therefore be avoided.
The fact that the waiting times are quite large means
that it takes a long time to obtain a complete curve.
This is a disadvantage of the method.
The author has also made experiments in order to
find out the effect of degree of saturation of the
specimen. Such experiments make it possible to
find out ia which pores water is located at different
degrees of saturation. This an advantage of the method.
Suitable apparatus is described by Zagar [27, 29],
Schwiete & Ludwig [23] and Alviset & Liger [1].
An apparatus built at our institution is shown in
Fagerlund [12]. The specimen holder is adapted
to angular specimens. The apparatus can also be
used for pure permeability measurements of liquids
or gases.
m~thodes par succion, entre autres les m~thodes de
la plaque, de la membrane et de la centrifugation.
Un groupe de m~thodes voisines ddcoule de la
vitesse de pdn&ration des liquides mouillants sous
l'action des f o r c e s capillaires (par ex. la mdthode
d'Astbury), et de l'observation clans les conditions
statique et dynamique des mouvements gaz/liquide et
liquide/liquide.
[7] BOMBERGM. - - Part III : Applications o f the relative
suction model. Report 2l, Div. of Build. Techn.
Lund Inst. of Techn., 1972.
[8] CLEMENTSJ.F., VYSE J. - - A comparison o f methods
o f measuring pore-size distribution in refractories.
Trans Brit Ceram Soc, 67 (1968), 285.
[9] CRONEY D., COLEMAN J.D., BRIDGE P.M. - - The
suction of moisture held in soil and other porous
materials. Road Res. Techn. Paper No. 24, Her
Majesty's Stationary Office, London, 1952.
[10] DULLIEN F.A.L., BATRA V.K. - - Determination oJ"
the structure o f porous media. Ind. Eng. Chem., 62
(1970), 25.
[11] ERBE F. - - Die Bestimmung der Porenverteilung
nach ihrer Grrsse in Filtern und Ultrafiltern. Kolloid
Z 63 (1933), 277.
[12] FAGrRLUND G. - - Frost resistance o f porous and
brittle materials in relation to their structure-theory
and experiments. Report 30, Div. of Build Techn.
at Lurid Inst. of Techn., 1972 (in Swedish).

[13] G~MESIJ. - - Untersuchung der Kapillarerscheinungen
in Beton mit Zerst6rungsfreien Methoden. Wissensch
Zeitschr Hochsch f Bauwesen, Leipzig H3 (1969),
175.
[14] GRABAR P., NIKITINE S. - Sur le diambtre despores
- [22] SCHOFIELD R.K. - - The pF o f water in soil. Trans.
des membranes en collodion utilis~es en ultrafiltration.
J. Chim. Phys., 33 (1936), 721.
[15] HANSEN T.C. - - Drying shrinkage o f concrete due
to capillary action. Techn. Rep. No. 1. Techn. Univ.
of Denmark, Dept Civil Eng., 1967.
[16] KSENSHEK O.S., KAIANOVSK[[E.A., PETROVA S.A.,
LtTV[NOVA V.J. - - Centrifugal method for finding
pore-size distribution in porous media. Russian J. of
Phys. Chem., 41 (1967), 856.
[17] LUtKOW A.V. - - Heat and mass transfer in capillary
porous bodies. Pergamon Press Ltd., Oxford, 1966.
[18] MANEGOLD E., KOMAGATA S., ALBRECHT E. - - t)ber
Kapillarsysteme, XIX. Kolloid Zeitschr., 93 (1940),
166.
[19] VAN M[NNEN J., VOS B.H. - - Distribution and
transport o f water in porous materials. Report
No. II-8 TNO, Delft, 1965.
[20] OHNEMULLER W . - - Die Bestimmung der Frostemp-
findlichkeit porOser keramischer Werkstoffe mit
Hilfe physikalischen MessgrOssen. Diss Fakult/it fiir
Bergbau u n d Hiittenwesen der Bergakademie Claus-
thal, 1960.
G. F A G E R L U N D
[21] PENNER E. - - Suction and its use as measure of
moisture contents and potentials in porous materials.
Humidity & Moisture, Vol. 4, p. 245. Reinhold,
New York, 1965.
3rd Int. Congr. of Soil Sci. No. 2, Oxford, 1935.
[23] SCHWIETEH.E., LUDWIG U. - - Ober die Bestimmung
der offenen Porositdt im Zementstein. Tonind Ztg.,
90, 1966, 562.
[24] SEWELL E . C . , WATSON E . W . - - Hysteresis in the
moisture characteristics o f ideal bodies. Bull Rilem,
29, 1965, 125.
[25] TIMOFEEV D.P. - - The mechanism o f transport o f
matter in porous sorbents. Russ. Chem. Rev., 29
(1960), 180.
[26] Vos B.H., TAMNES E. - - Moisture and moisture
transfer in porous materials. Report No. B 1-69-96
TNO, Delft, 1969.
[27] ff~AGAR L. - - Ermitthmg der Gr6ssenverteihmg yon
Poren in Feuernfesten Baustoffen. Archiv f. das
Eisenh/.ittenwesen, 26 (1955)/H 9/, 561.
[28] ZAGAR L. - - Ermitthmg der Gr6ssenverteihmg yon
Poren in Feuernfesten Baustoffen und Glasnutschen.
Teil II, Archly. f. das Eisenhi.ittenwesen, 27 (1956)/H
10/, 657.
[29] ~AGAR L. - - ~)ber die Textur yon Feuernfesten
Baust-offen. Ber Deut Keram Ges. 35 (1958), 294.
201