LIMESTONE CONTACTORS: STEADY-STATE

DESIGN RELATIONSHIPS
By Raymond D. Letterman, 1 Member, ASCE, Marwan Hadad, 2
and Charles T. Driscoll3

ABSTRACT: Limestone contactors can mitigate corrosion in small water-supply

Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

systems that use dilute, acidic water. As water is transported through a packed bed
of crushed limestone, CaC0 3 dissolves and the pH, calcium-ion concentration, and
alkalinity increase. Operation of a contactor can be effectively modeled by con-
sidering the rate of dissolution and interfacial transport of calcium ions. The steady-
state model developed and tested in this study relates the depth of limestone
required in the contactor to the desired effluent water chemistry, influent water
chemistry, limestone-particle size and shape, bed porosity, water temperature,
and superficial velocity. The magnitude of the rate constant that describes the re-
lease of calcium ions from the calcite surface varies with the pH at the particle
surface. When this pH is less than about 9.5, the rate constant for the surface
reaction becomes large, and the rate of dissolution tends to be controlled solely
by the transport of calcium ions away from the interface.

INTRODUCTION

Throughout the United States, home owners and small public and private
water-supply systems use water that is potentially corrosive to metallic ma-
terials in the distribution system. Corrosion may be a concern in these sys-
tems because of the potential health problems associated with ingesting cor-
rosion by-products, degradation of the water's aesthetic quality, and the
significant economic consequences of deteriorating the piping system.
Corrosion and water contamination by corrosion by-products may be caused
by the use of dilute (low ionic strength) acidic waters that generally have
low pH, alkalinity, and concentrations of dissolved solids. Dilute acidic ground
and surface waters are found in many regions of the country, particularly in
areas underlaid by siliceous bedrock. The waters naturally have low buff-
ering capacity, and they are corrosive. They are also prone to acidification
by atmospheric deposition of strong acids (acid precipitation) or other fac-
tors, such as changes in land use. In some areas (for example, the Adiron-
dack region of New York), acidic deposition may increase the water's cor-
rosivity.
Corrosion-mitigation techniques for small water-supply systems should be
inexpensive; the maintenance required should be minimal; and the technique
should present a low potential for public health hazard if the device is im-
properly constructed, installed, or maintained. Limestone contactors can meet
these requirements.
In a limestone contactor, water flows through, under a closed-to-the-at-
mosphere condition, and dissolves calcium carbonate from a packed bed of
crushed limestone or another source of CaC0 3 . The water's chemistry is
'Prof., Dept. of Civ. and Envir. Engrg., Syracuse Univ., Syracuse, NY 13244.
2
Prof., Dept. of Civ. Engrg., An-Najah Univ., Nablus, West Bank, Israel.
3
Prof., Dept. of Civ. and Envir. Engrg., Syracuse Univ., Syracuse, NY.
Note. Discussion open until November 1, 1991. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on May
31, 1988. This paper is part of the Journal of Environmental Engineering, Vol.
117, No. 3, May/June, 1991. ©ASCE, ISSN 0733-9372/91/0003-0339/$1.00 +
$.15 per page. Paper No. 25927.

339

J. Environ. Eng., 1991, 117(3): 339-358

 altered as the CaC0 3 dissolves. Limestone contactors are used to neutralize
acid mine drainage, acidic industrial wastes, and dilute acidic surface waters.

BACKGROUND
Limited literature exists on the kinetics of calcium carbonate dissolution
from packed beds of crushed limestone. Recent investigators (Pearson and
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

McDonnell 1975a, 1975b; Barton and Vatanatham 1976; Vaillencourt 1981)

assume that the CaC0 3 dissolution rate depends on the transport of hydrogen
ions from the bulk solution to the limestone surface. Each of these studies
used a film-transport—type model based on the difference between the hy-
drogen-ion concentration at the surface and the concentration in the bulk
solution to describe the kinetics of the dissolution process.
Recent work on the kinetics of CaC0 3 (calcite) dissolution by Rickard and
Sjoberg (Sjoberg and Rickard 1983, 1984; Rickard and Sjoberg 1983) has
shown that the CaC0 3 dissolution rate in slightly acid to alkaline solutions
is controlled by an interfacial calcium-ion mass-transfer resistance and a first-
order surface reaction acting in series. The equations presented in the fol-
lowing assume that the rate of CaC0 3 dissolution within a limestone con-
tactor can be modeled using three resistances acting in series, i.e., the two
described by Rickard and Sjoberg (liquid film transport and surface reaction)
and a third for the transport of calcium ions across a thin porous layer at
the particle surface. We assume that the porous layer is formed by insoluble
alumino-silicate (clay) impurities that remain on the particle surface as the
CaC0 3 dissolves from the limestone matrix. Using a scanning electron mi-
croscope with an X-ray energy spectrometer attachment, we gathered evi-
dence for this layer's formation through long-term experiments with labo-
ratory and field contactors (Letterman et al. 1986).
The mathematical model described here can be used for contactor design
under a steady-state operating condition and for relatively high-purity, high-
calcium limestone. An approximately steady-state condition can be assumed
when the influent water flow rate, chemistry, and temperature are reasonably
constant and when two effects are insignificant: that of CaC0 3 's dissolution
on limestone-particle size in the packed bed and that of the porous layer
formed by insoluble impurities in the limestone matrix on mass transfer to
and from the limestone surface. A subsequent paper will discuss the effects
of magnesium, aluminum, and iron content on the rate of dissolution and
on the tendency for a residue layer to form. The formation of a residue layer
is likely to be negligible, in any case, when the operating period, which
starts with fresh limestone, is relatively short (e.g., less than several years
under typical operating conditions).

Chemical Equilibrium Calculations

In the dissolution-rate model described in the following, the rate of cal-
cium-ion transport from the surface to the bulk solution is assumed to depend
on the equilibrium calcium-ion concentration Ceq. The magnitude of Ceq and
the corresponding equilibrium pH, pHe?, and the effluent pH as a function
of the effluent calcium-ion concentration CbL were determined using the fol-
lowing relationships.
For a dilute acidic water with dissolving CaC0 3 and a closed-to-atmo-
spheric-carbon-dioxide condition, the charge balance relationship is
340

J. Environ. Eng., 1991, 117(3): 339-358

 2{Cbo + S) + Cc + [H+] = [(DIC + 5)(a, + 2a 2 )] + Ca + —?- (1)
W J
where Cbo = the molar concentration of calcium in the influent; S = the
molar concentration of calcium carbonate dissolved; Kw = the activity-coef-
ficient-corrected ion product for water; DIC„ = the influent-dissolved in-
organic-carbon concentration; and Cc and Ca = the influent concentrations
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

(in equivalents per liter) of cations excluding calcium and hydrogen, and
anions excluding inorganic-carbon species and hydroxide, respectively. The
quantities a, and a 2 = the ionization fractions for the deprotonation of car-
bonic acid and are calculated using the pH, activity-coefficient-corrected-
ionization constants, and standard relationships used in aquatic chemistry
(Snoeyink and Jenkins 1980).
When the solution is in equilibrium with solid CaC0 3 , the following sol-
ubility product expression is included in the calculations
(Cbo + S')(a2)(DIC0 + S') = Ksp (2)
where Ksp = the effective, activity-coefficient-corrected solubility product
for C a C 0 3 in limestone; and S' = the molar concentration of C a C 0 3 dis-
solved at equilibrium. The calcium concentration at any point in the con-
tactor CbL is given by
CbL = Cbo + S (3)
where S < S'.
The magnitudes of 5" and pHc<7 were determined by solving (1) and (2)
simultaneously. The equilibrium calcium concentration is given by Ceq =
Cbo + S'. When the contactor effluent is not in equilibrium with solid CaC0 3
(S < 5" and CbL < Ceq), the effluent pH is determined by solving (1) for
[H + ] using S = CbL - Cbo.
We corrected for temperature equilibrium constants in the chemical model
using equations from Plummer and Bussenburg (1982). Activity coefficients
used in correcting for ionic strength and calculating the pH were determined
using the Davies equation.

DISSOLUTION-RATE MODEL

The dispersed-plug flow model for steady flow in a packed-bed reactor is

given by the basic differential equation (Levenspiel 1972)
d2C dC
N +r (4)
°^-Tz <=°
where
Fe
ND = (5)
(V.L)
is the dimensionless axial dispersion number; C = the reactant concentration;
Z = the dimensionless axial distance (Z = z/L); L = the overall depth of
limestone in the column; t = the mean fluid residence time (i = Le/Us); Us
= the superficial or empty column fluid velocity; F = the dispersion coef-
341

J. Environ. Eng., 1991, 117(3): 339-358

 ficient; e = the bed porosity; and r = the reaction rate expression.
For the dissolution and interfacial transport of calcium ions, the reaction
rate expression is given by the film-transfer relationship
r = kDa{Ceq -C) (6)
where k0 = the overall dissolution-rate constant; Ceq = the calcium concen-
tration in the fluid when the calcium carbonate in the limestone is in equi-
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

librium with the influent flow; and C = the bulk-fluid calcium-ion concen-
tration. The interfacial area of limestone particles per unit volume of interstitial
fluid a is related to the volume-mean limestone-particle diameter d and
sphericity ^ by
6(1 - e)
a =— - (7)

The overall dissolution-rate constant is related to the liquid and residue-

layer film-transport coefficients (kL and kf) and the surface reaction rate con-
stant for calcium dissolution kc. This relationship is expressed by

\kf kc kj
The two coefficients and the constant in (8) are discussed in the following.

Liquid-Film Mass-Transfer Coefficient kL

Letterman et al. (1986) evaluated a number of correlations of dimension-
less film-transfer parameters for determining kL. A set of expressions by Chu
et al. (1953) was found to predict contactor performance under a useful range
of conditions. The expressions in this set include, for low modified Reyn-
old's numbers
kL = 5.7t/5(MR)-°- 87 (Scr 2/3 , 1 £ MR < 30 (9)
and for high modified Reynolds numbers
kL = l.Ws(MRT0A4(Sc)-2/3, 30 < MR < 10,000 (10)
where the modified Reynolds number MR is given by
dVs
MR = — (11)
M l - e)]
the Schmidt number, Sc by

Sc = - (12)
D
and v = the kinematic viscosity; and D = the calcium-ion diffusivity.

Surface-Reaction Rate Constant kc

The dissolution rate of low-solubility minerals is often controlled by re-
actions at the interface between the solid phase and water (Stumm and Wol-
last 1990). When surface protonation is a controlling surface reaction, the
342

J. Environ. Eng., 1991, 117(3): 339-358

 TABLE 1. Summary of Surface Chemical Reaction Rate Constant Calculations
Ceq x 10 7a J0 x 10 9b kc x 10 3 0
3 2
pH„ (mole/cm ) pH e , a (mole/cm /s) (cm/s)
(1) (2) (3) (4) (5)
3.00 11.00 8.50 20.00 681.00
3.20 8.02 8.80 12.60 98.00
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

3.50 5.00 9.28 5.62 27.90

4.00 3.80 9.66 2.66 11.20
4.50 3.50 9.78 1.19 4.20
5.00 3.50 9.82 0.79 2.60
5.50 3.50 9.82 0.71 2.30
6.00 3.50 9.83 0.71 2.30
"Calculated using the chemical equilibrium model described in the text.
b
From Sjoberg and Rickard (1984) by interpolation.
'Calculated using (15) and kL = 1.9 X 10" 2 cm/s.

pH of the solution in the interfacial region is an important rate-determining

parameter.
Data obtained by Sjoberg and Rickard (1984) were used to derive an em-
pirical relationship between the equilibrium interfacial pH, pHe?, and the
magnitude of kc. In Sjoberg and Rickard's experimental system (pH-stat-type
batch experiments with an initial bulk solution calcium concentration Cb equal
to zero), the initial rate of increase of the calcium-ion concentration in the
bulk solution J0 is given by
k,kn
Jn = (13)
(kL + U
Sjoberg and Rickard measured /„ as a function of solution pH. The dis-
solving CaC0 3 was mounted on a rotating disk operating at 1,000 rpm. The
water temperature was 25° C, and the C0 2 partial pressure was essentially
zero. The measured values of J0 are listed in Table 1 for each bulk solution
pH, P H 0 .
Fig. 1 is a plot of J0/Ceq versus pH„ for Sjoberg and Rickard's results.
The values of Ceq and pHe? listed in Table 1 were calculated for each bulk
solution pH using the pertinent experimental conditions and the chemical
equilibrium model described previously. An effective CaC0 3 solubility prod-
uct of 3.55 X 10"9 (25° C) was used (Herman 1982). In general, as pH„
increases, pHe9 increases and the quantity J0/Ceq decreases.
To develop a relationship between the chemistry of the bulk solution and
kc, it was assumed that the magnitude of kc increases as pHe9 decreases. This
assumption is consistent with a mechanism in which the release of calcium
and carbonate ions from the CaC0 3 matrix is preceded by the attachment of
protons at the interface, possibly to form a surface species that is a precursor
to the solution HCOJ ion.
Given the assumption that at low pH, kc is large and significantly greater
than kL, then, according to Fig. 1 and (13)
kLkc
(14)
Sh + K)_
343

J. Environ. Eng., 1991, 117(3): 339-358

Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. JJCeq Plotted as Function of Bulk Solution pH for Results Obtained by

Sjoberg and Rickard (1984)

The asymptote in Fig. 1 at solution pH values less than three corresponds

to kL « (J„/C«,)(pH<3) = 1.9 X 10" 2 c m / s .
Eq. (13) was rearranged and used with the experimental values of J0 listed
in Table T and the estimated value of kL (1.9 X 1CT2 cm/s) to determine
values of kc, i.e.

kc — kL (15)
J,
The results of this calculation are listed in Table 1. Fig. 2 is a semi-log plot
of kc versus pH,,?.
The data plotted in Fig. 2 were fitted with a linear equation by the method
of least squares, with the following result
logkc 14.2 - 1.7 pH s? , (r 2 = 0.986) (16)
or
K = 1.6 x 10u{Heq}iJ , (17)
where {Heq} = the equilibrium interfacial hydrogen-ion activity, and kc is in
cm/s.

Residue-Layer Mass-Transfer Coefficient kf

Using film theory (Satterfield 1970), we can model the transport of cal-
cium ions through the porous layer formed by impurities in the limestone.
In other words

344

J. Environ. Eng., 1991, 117(3): 339-358

 1000
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

pHeq

FIG. 2. Surface-Reaction Rate Constant Plotted as Function of Equilibrium pH;

Points Were Obtained Using Data from Sjoberg and Rickard (1984) and Chemical
Equilibrium Model; Line Was Fitted to Data by Method of Least Squares

kf- (18)
(§T r )'

where er and 8 = the porosity and thickness of the layer; D = the diffusivity
of the calcium ion in the bulk solution; and Tr = the pore length (tortuosity)
factor. The magnitude of kf decreases as calcium carbonate dissolves and the
thickness of the residue layer increases.

Steady-State Relationship
When the limestone contains small amounts of alumino-silicate impurities,
a porous layer occurs very slowly or not at all. In this instance or when
impure CaC0 3 is replaced frequently with fresh stone, we can assume that
8 in (18) is essentially zero; kfis large; and the overall dissolution-rate con-
stant is given by
-i

(19)
kn =
kj + W,
345

J. Environ. Eng., 1991, 117(3): 339-358

 If, in addition, we assume that the effect of CaC0 3 dissolution on the vol-
ume-mean diameter of the limestone particles is negligible throughout the
design period, then k„ and the quantity a [see (7)] are essentially constant,
and (4) can be integrated (Levenspiel 1972) to yield the following expression
for the bulk solution calcium concentration CbL as a function of contactor
depth L
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

-knaLe\ (k„aLz
CbL = Cm+ \ exp ND {Ch0 - C„) (20)
u. \ v,
In integrating (4), we substituted (6) for the reaction rate term r, and the
reactant concentration C was set equal to the bulk solution concentration of
the calcium ion at axial location z, or CbI. It was also assumed that axial
dispersion in the packed bed is minimal (ND < 0.01). Using tracer experi-
ments and packed beds of limestone, Hadad (1986) determined that when
values of ND are low, they are approximated by

ND = 2\~\ (21)

EXPERIMENTAL MATERIALS AND PROCEDURES

Contactors
The laboratory study involved using four down-flow, packed-column con-
tactors, with each column containing a different limestone-particle size. We
chose column diameters that would yield a column-to-particle diameter ratio
of at least 10 to minimize the effect of the higher porosity at the wall on
the flow through the bed. The four columns and the water feed system are
shown in Fig. 3.
The four columns shown in Fig. 3 were equipped with through-the-wall,
0.6-cm-diameter sampling tubes spaced in the axial direction at 15.2-cm in-
tervals at the influent end and at 30.4-cm intervals over the remaining por-
tion of each column. Fig. 3 incorporates a drawing of a typical sampling
tube.
The water-supply and flow-control system used with the four laboratory
columns is shown in Fig. 3. The raw water was pumped from a 200-L plastic
tank to a constant-head tank situated above column A. Overflow from the
constant-head tank returned to the plastic tank.
The limestone was washed with tap water and placed in each column layer
by layer. When the depth of stone reached the level of a sampling tube hole,
the tube was inserted and stone was gently installed around it. Gentle tapping
and shaking of the column consolidated the bed as it was installed.

Limestone Characteristics
The limestone used throughout the study came from a quarry in Boonville,
New York. The cation content of this limestone is (by mass) 85% calcium,
12.3% aluminum and 2.4% magnesium. Iron, Mn, Zn, Cu, and Cd are pres-
ent at less than 0.1% each; no Pb, K or Na is evident. Thus, Boonville is
a high-calcium (and low-magnesium) limestone. (Evidence exists that the
rate of dissolution decreases as the magnesium content of the stone increases.
The results presented in this study pertain only to high calcium stone.) A
346

J. Environ. Eng., 1991, 117(3): 339-358

Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

FIG. 3. Laboratory Columns with Water-Supply and Flow-Control System; Insert

Is Drawing of Typical Through-Wall Sampling Tube; Inside Diameters, Length, and
Stone-Particle Sizes (Volume-Mean Diameter) for Each Column Are as Follows:
Column A = 15.2 cm, 3.5 m and 0.96 cm; Column B = 15.2 cm, 2.1 m, and 0.54
cm; Column C = 15.2 cm, 2.1 m, and 1.5 cm; Column D = 38.1 cm, 2.1 m, and
3.2 cm

supplemental experiment involving acid digestion and measuring the amount

of C0 2 released, combined with the calcium measurement, indicated that the
Boonville stone contains 79% CaC0 3 by mass. Batch dissolution experi-
ments used in conjunction with the program MINEQL (Westall et al. 1976)
showed that the limestone CaC03 effective solubility product is lCT8,71 (20° C
and 7 = 0). This value is somewhat lower than the range (10~8,5-10~8'3)
typically associated with calcium carbonate in limestone (Herman 1982). These
findings are detailed in Letterman et al. (1986).
The four size fractions of our limestone particles were analyzed to deter-
mine particle volume-mean diameter, sphericity, and mass density (see Table
2). To detennine the mass-median particle size for each size fraction, we
used a standard ASTM sieve analysis procedure (ASTM 1972). In the case
of the 0.96-cm median-size fraction, 90% of the particles were between 0.7
and 1.3 cm.
347

J. Environ. Eng., 1991, 117(3): 339-358

 TABLE 2. Limestone-Particle Size and Sphericity Analysis Results
Limestone-particle
Mean surface area per
diameter Particle unit volume of
Size sieve analysis sphericity interstitial water Bed
fraction (cm) (dimensioniess) (cm"1) porosity
(D (2) (3) (4) (5)
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

I 3.20 0.83 2.6 0.49

II 1.45 0.78 5.3 0.49
m 1.01 0.79 11.4 0.41
IV 0.55 0.81 18.2 0.43

We determined the volume-weighted mean particle diameter by using a

fluid displacement technique (Hadad 1986) to measure the volume of at least
1,200 particles in each size fraction. Our results agreed closely with the sieve
analysis results.
The planimetrically measured surface area of the particles was used with
the volume-weighted mean diameter to calculate particle sphericity. The av-
erage sphericity ranged from 0.83 for the 3.2-cm size fraction to 0.78 for
the 1.5-cm fraction. In the case of the 0.96-cm fraction, the measured spher-
icities ranged from 0.50 to 0.98, with an average value of 0.79.
Each column's porosity was determined by measuring the volume of fluid
required to displace all the air from the bed. Special tests (Letterman et al.
1986) indicated that the effect of the column wall on the bed porosity was
negligible.
Using (7) and the measured porosity, volume-mean particle diameter, and
sphericity, we calculated the limestone-particle surface area per unit volume
of interstitial water for each column. The calculated values ranged from 2.6
cm 2 /cm 3 for the 3.2-cm stone to 18.2 cm2/cm3 for the 0.54-cm stone.

Sampling and Analytical Procedures

Starting at the top of the column and moving down, we collected samples
from all the ports. Samples were withdrawn by gravity flow and collected
in 500-mL polyethylene bottles. To minimize the effect of C0 2 exchange,
we measured the pH at each port by attaching a device that allowed the
sample to flow past a combination pH electrode under a closed-to-the-at-
mosphere condition. To minimize disturbing the flow in the column during
sampling, we allowed some time to elapse between taking each sample. The
"waiting period" we chose equalled twice the distance between two sampling
ports divided by the interstitial flow velocity.
The column experiments were conducted at room temperature (9°-22° C).
To minimize microbial growth, the columns were first rinsed with chlori-
nated water and then with deionized water. The clear acrylic column (column
A) was covered with black plastic sheets to reduce exposure to light. The
pH meter was calibrated with pH-4, -7, and -10 buffers using a procedure
given in Standard Methods (1985). The alkalinity was measured by strong
acid titration and Gran plot analysis (Gran 1952). Concentrations of Ca2+,
Mg 2+ , Na + , and K + were determined by atomic absorption spectrophotome-
try (Slavin 1968).
348

J. Environ. Eng., 1991, 117(3): 339-358

 The raw-water solution consisted of HC1, CaCl2, and NaCl in deionized
water. The raw-water pH ranged from 2.3 to 6.1, and the calcium concen-
tration from 0 to 28 mg Ca/L. In all experiments, the background electrolyte
was 20 mg NaCl/L.

RESULTS
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

Chemical Equilibrium Model

Results obtained in a special series of column experiments are in reason-
able agreement with the predictions of the chemical equilibrium model. The
results of selected column experiments in this series are plotted in Fig. 4.
Under these experimental conditions (very low flow rate, Us = 5.5 cm/min,
and small limestone-particle diameter, 0.54 cm), an equilibrium condition,
as evidenced by a calcium concentration that did not vary significantly with
column depth, was apparent in at least the bottom half of the column. The

CURVE PH0
200 A 4.50
ec 4.10
3.60
0 3.20
180- E 3.00
F 2.90
6 2.70
H 2.60
I 2.30

4 6 8 10 12 14
BED DEPTH ,L(ft )

FIG. 4. Calcium Concentration Plotted as Function of Axial Distance to Sampling

Port and Influent pH, pH,; Results Were Obtained Using Column A (Fig. 3); Cal-
cium Concentration in mols/L = 2.5 x 10~S Times Concentration in mg Ca/L

349

J. Environ. Eng., 1991, 117(3): 339-358

 too i p—, r ,.
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

J I i L _fc i—-x.
20 3.0 4.0 50 6.0
initial pH,pH 0

FIG. 5. Equilibrium pH, Dissolved Inorganic Carbon and Calcium Concentrations

Plotted as Function of Influent pH and Following Influent Calcium Ion Concentra-
tions; Curve A = Cbo = 0; Curve B = cba = 28 mg Ca/L; Lines Plotted Are from
Equilibrium Model [see (1)-(3)]; Calcium Concentration in mg Ca/L = 40,000 times
Molar Concentration

350

J. Environ. Eng., 1991, 117(3): 339-358

 sampling time in the case of Fig. 4 was approximately one hour after start-
up.
The calcium concentration plotted as a function of the distance to the sam-
pling port (Fig. 4), and similar plots for pH and the dissolved inorganic
carbon concentration were used to estimate the equilibrium value of the mea-
sured parameter. The estimated equilibrium concentrations of the calcium
concentration, pH, and DIC are plotted as a function of the raw-water pH
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

in Fig. 5.
The lines in Fig. 5 derive from points plotted from calculations using the
chemical equilibrium model. The magnitude of the effective solubility prod-
uct for CaC0 3 , determined in the batch experiments to bepKsp = 8.71 (20° C),
was found to give a reasonable fit of the experimental results; this value
was, therefore, used in the remaining analyses.
As shown in Fig. 5, the agreement between the equilibrium model pre-
dictions and the column data is reasonable for pH and the calcium concen-
tration. In the case of DIC, for pH values > 3 . 5 the measured values of DIC
are somewhat greater than those predicted by the model. Although we took
care to avoid the possibility, some C 0 2 may have entered the solutions after
the samples had been drawn from the column. In general, for negligible
influent calcium and DIC and when the influent pH is < 4 . 2 , as the raw-
water pH decreases, the maximum effluent pH falls from a value ==9.5 to
< 8 . As the influent pH decreases to < 4 , the amount of C a C 0 3 dissolved at
equilibrium increases exponentially.
The presence of calcium in the raw water tends to reduce the dissolution
of C a C 0 3 by the common ion effect. For raw water with pH„ = 4, the
maximum effluent pH decreased from 9.4, when no Ca was present, to about
8.9, when 28 mg C a / L was added. For raw-water pH values greater than
about four, the amount of DIC at equilibrium also declined. Use of the chemical
equilibrium model (Letterman et al. 1986) can illustrate that increasing the
influent DIC concentration has an effect on the effluent chemistry similar to
that of increasing the influent Ca concentration.

Calibration of Dissolution-Rate Model

Experimental Values of k„
The data on calcium concentration versus column depth (e.g., Fig. 4) were
used to determine a best-fit experimental value of k0 (k0 exp) for each of the
64 column experiments. To accomplish this, we determined values of the
calcium concentration as a function of the column depth by using (20) and
the experimental conditions of that run. The magnitude of k0 was varied
systematically to ultimately minimize the quantity X, given by

X = 2_j (CbL calculated - CbL experimental)? (22)

In (22), n = the number of sampling ports in the column. Ceq in (20) was
again determined using the chemical equilibrium model [(l)-(3)].

Model Predicted Values of k0

Values of k0 predicted by the model {k0 model) were determined using
(19), and values of kL calculated from (9)-(12). The effect of the interfacial
351

J. Environ. Eng., 1991, 117(3): 339-358

 TABLE 3. Ranges of Experimental Conditions in Runs Used to Determine k0

Parameter Range
(D (2)
Superficial velocity V, 0.9-72 cm/min
Influent pH, pH„ 3.4-6.0
Influent calcium concentration Cbo 0-5.2 mg Ca/L (0-1.3 x 10"4 mols/L)
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

Temperature T 9°-22° C
Volume-mean particle diameter d 0.55-3.20 cm
bed porosity E 0.41-0.49

pH on kc was included in the calculations using the result from Sjoberg and
Rickard (1984), i.e., (17), which yields

(23)
lo^j
where k* = the surface-reaction constant at pH = 9.5. The temperature
corrected value of the calcium-ion diffusivity Dx was determined using the
calcium-ion difusivity at 20° C, D20, and
(T + 273)
Dx = D* (24)
293
where T = the temperature in degrees Celsius; and vx and v20 = the kinematic
viscosity of water at temperatures of x and 20° C, respectively.
Using an optimization procedure [Powell's search in conjugate directions
(Powell 1965)], we identified the values of D20 [in (24)] and k* [in (23)]
that give the best agreement between model-predicted and experimental val-
ues of k0. The following minimum-sum-of-squares objective function was
used

Y = 2^ (k0 exp - k0 model)? = minimum. (25)

The analysis included 64 values of k0 exp. The ranges of experimental con-

ditions used to determine these values are listed in Table 3. Details of each
experimental run appear in Letterman et al. (1986).
The best-fit value of D20 obtained in the optimization procedure (1.2 X
10~3 cm2/s) is within the literature's range of values. Published values of
the calcium-ion diffUsivity for dilute solutions at 20° C fall in the range of
0.79 to 1.50 X 10"5 cm 2 /s (Sherwood et al. 1975; Cussler 1984; Robinson
and Stokes 1959). Differences among reported values may be attributable,
in part, to the presence of different anions.
It should be noted that the assumption of a constant calcium-ion diffusivity
is an approximation. The diffusivity of a cation such as calcium depends on
the associated anion and the presence of other electrolytes. Diffusion to or
from dissolving calcium carbonate in limestone is a complicated process that
may involve changes in ion speciation (particularly the inorganic carbon spe-
cies) as these move from the surface to the bulk solution. This change in
speciation may affect the diffusivity of the calcium ion.
The best-fit value of the surface-reaction rate constant at pH = 9.5 (fc* =
352

J. Environ. Eng., 1991, 117(3): 339-358

Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

0.001 0.01 0.1 1.0

k0 (cm/min,experimental)

FIG. 6. Correlation between Model Calculated and Best-Fit Experimental Values

of Overall Dissolution-Rate Constant k„

0.85 cm/min) agrees approximately with the value (0.7 cm/min) obtained
by analyzing Sjoberg and Rickard's (1984) calcite dissolution data [(17) with
{Heq} = 10~9,5)]. The disparity can be attributed to a number of factors,
including the difference between the ionic strength in these and in Sjoberg
and Rickard's experiments (7 = 0.7) and to the fact that calcium carbonate
in limestone is being compared with essentially pure calcite. Also, it was
impossible to adjust the surface-reaction rate constant for the effect of tem-
perature.
In Fig. 6, values of ka calculated using the best-fit values of D20 and k*
are compared with the values obtained by fitting the experimental data (k0
exp). The correlation coefficient is 0.73.
Past investigations of the kinetics of limestone dissolution in packed beds
concluded that the dissolution rate is controlled solely by a mass-transfer
resistance at the limestone-solution interface, i.e., k0 = kL. Our study results
suggest that the validity of this assumption depends on the interfacial pH
and the magnitude of kc.
For the ranges of experimental conditions used in this study (see Table
3), the maximum calculated [(9)-(12)] value .of kL is 0.15 cm/min. This
value corresponds to the highest superficial velocity and temperature and the
smallest limestone-particle size. According to (23), with k* = 0.85 cm/min,
353

J. Environ. Eng., 1991, 117(3): 339-358

Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

u
z
g

1 5—
o
<
u
o
S!
K
Wg
u>
ffl
o

0 5 10 15
M00EL CALCULATED CALCIUM CONCENTRATION, C 5 L ' m 9 C o / L >

FIG. 7. Observed Calcium Concentration as Plotted Function of Model-Predicted

Value; Calcium Concentration in mols/L = 2.5 x 10"5 Times Concentration in mg
Ca/L

when the equilibrium pH is less than 9.5, the magnitude of kc will be greater
than 0.85 cm/min. Therefore, when the operating conditions are similar to
those used in this study, the validity of the assumption that k0 » kL improves
as pH„, decreases below about 9.5. Between pHe? = 9.5 and its upper limit
of approximately 9.8 (see Fig. 5), the surface reaction may significantly
affect both the dissolution rate and the contactor effluent's chemistry.
In Fig. 7, the measured calcium-ion concentrations are plotted versus the
model-predicted values for all 64 column experiments. The points plotted
represent measurements and calculations for all sampling port locations. The
agreement between the measured and model-predicted calcium-ion concen-
trations is good except at high calcium concentrations (>8 mg Ca/L) where
all the calculated values tend to be larger than the measured concentrations.
Calculated and experimental values of the change in alkalinity between
the influent and each sampling port are plotted in Fig. 8. The model-cal-
culated calcium concentrations were used to determine the alkalinity for each
sampling port location using
Alkalinity (|xeq/L) = Influent Alkalinity Oeq/L) + 50 x ChL, mg/L (26)
The agreement obtained between the model-calculated and measured al-
354

J. Environ. Eng., 1991, 117(3): 339-358

 4O0F 1 | | -
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

0 100 200 300

A ALKALINITY MODEL CALCULATEO^eq/L)

FIG. 8. Measured Change In Alkalinity within Laboratory Contactors Plotted as

Function of Model-Predicted Change; Model Equations Were Used to Determine
k„ for Each Run

kalinities in Fig. 8 supports using the model as a predictive tool under steady-
state conditions. The alkalinity results also suggest that the poor agreement
between the measured and model-predicted calcium concentrations at high
calcium concentrations (Fig. 7) is probably attributable to an analytical prob-
lem and not a model deficiency.
The model described has been used to prepare a computer program called
DESCON and a set of graphs (nomographs) that can be used for contactor
design. DESCON calculates, as a function of the influent water chemistry
and design and operating variables, the depth of contactor required to yield
a specified effluent chemistry. The program and nomographs are available
from the first writer.

CONCLUSIONS

The dissolution of calcium carbonate in a packed bed of crushed limestone

can be modeled, under conditions that approximate steady state, using the
assumption that an interfacial calcium-ion transport resistance acting in series
with a first-order surface-chemical reaction controls the kinetics of disso-
lution. Under a condition of high superficial velocity, high temperature, small
limestone-particle size, and a dilute influent solution containing negligible
calcium and dissolved inorganic carbon, the surface-chemical reaction may
significantly affect effluent quality. When the equilibrium (interfacial) pH is
355

J. Environ. Eng., 1991, 117(3): 339-358

 depressed by the presence of significant Ca or DIC in the influent, the sur-
face-chemical-reaction rate constant increases, and the mass-transfer resis-
tance tends to control.

ACKNOWLEDGMENT

This study was supported by the U.S. Environmental Protection Agency,

Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

Water Engineering Research Laboratory, Cincinnati, Ohio, through coop-

erative agreement CR-809979-01-3. We greatly appreciate the help of our
project officer, Gary S. Logdson.

APPENDIX I. REFERENCES

Barton, P., and Vatanatham, T. (1976). "Kinetics of limestone neutralization of acid

waters." Envir. Sci. Tech., 10(3), 262-266.
Chu, I., Kaill, J., and Wetteroth, W. A. (1953). "Mass transfer in fluidized beds."
Chem. Engrg. Prog., 49(3), 141-149.
Cussler, E. L. (1984). Diffusion: Mass transfer in fluid systems. Cambridge Uni-
versity Press, Cambridge, U.K., 101-105.
Gran, G. (1952). "Determination of the equivalence point in potentiometric titra-
tions." Int. Cong. Anal. Chem., 77(1), 661-671.
Hadad, M. (1986). "Modeling of limestone dissolution in packed-bed contactors treating
dilute acidic water," thesis presented to Syracuse University, at Syracuse, N.Y.,
in partial fulfillment of the requirements for the degree of Doctor of Philosophy.
Herman, J. S. (1982). "The dissolution kinetics of calcite, dolomite, and dolomitic
rocks in the C0 2 -water system," thesis presented to Pennsylvania State University,
at University Park, Pa., in partial fulfillment of the requirements for the degree
of Doctor of Philosophy.
Letterman, R. D., Driscoll, C. T., Hadad, M., and Hsu, H. A. (1986). "Limestone
bed contactors for control of corrosion at small water utilities." Water Engineering
Research Laboratory, Office of Research and Development, U.S. Environmental
Protection Agency, Cincinnati, Ohio.
Levenspiel, O. (1972). Chemical reaction engineering. John Wiley & Sons, New
York, N.Y.
"Manual on test sieving methods." (1972). ASTM-STP447A, ASTM, Philadelphia,
Pa.
Pearson, F. H., and McDonnell, A. J. (1975a). "Limestone barriers to neutralize
acidic streams." J. Envir. Engrg. Div., ASCE, 101(3), 425-440.
Pearson, F. H., and McDonnell, A. J. (1975b). "Use of crushed limestone to neu-
tralize acid wastes." J. Envir. Engrg. Div., ASCE, 101(1), 139-158.
Plummer, L. N., and Bussenberg, E. (1982). "The solubility of calcite, aragonite,
and veterite in C0 2 -H 2 0 solutions between 0 and 90° C and an evaluation of the
aqueous model for the system CaC0 3 -C0 2 -H 2 0." Geochim. Cosmochim. Acta, 46,
1011-1040.
Powell, M. J. D. (1965). "A method for minimizing a sum of squares of non-linear
functions without calculating derivatives." The Comput. J.,1, 303-307.
Rickard, D., and Sjoberg, E. L. (1983). "Mixed kinetic control of calcite dissolution
rates." Am. J. Sci., 283, 815-830.
Robinson, R. A., and Stokes, R. H. (1959). Electrolyte solutions. Butterworth, Lon-
don, U.K.
Satterfield, C. (1980). Heterogeneous catalysis in practice. McGraw-Hill Book Co.,
New York, N.Y.
Sherwood, T. K., Pigford, R. L., and Wilke, C. R. (1975). Mass transfer. McGraw
Hill, New York, N.Y.
Sjoberg, E. L., and Rickard, D. (1983). "The influence of experimental design on
the rate of calcite dissolution." Geochim. Cosmochim. Acta, 47, 2281-2285.
Sjoberg, E. L., and Rickard, D. (1984). "Calcite dissolution kinetics: Surface spe-

356

J. Environ. Eng., 1991, 117(3): 339-358

 ciation and the origin of variable pH dependence." Chem. Geol., 42, 119-136.
Slavin, W. (1968). Atomic absorption spectroscopy. John Wiley Interscience, New
York, N.Y.
Snoeyink, V. L., and Jenkins, D. (1980). Water chemistry. John Wiley & Sons,
New York, N.Y.
Standard methods for the examination of water and wastewater. (1985). 16th Ed.,
American Public Health Association, Washington, D.C.
Stumm, W., and Wollast, R. (1990). "Coordination chemistry of weathering: Ki-
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

netics of the surface-controlled dissolution of oxide minerals." Rev. Geophys.,

28(1), 53-69.
Vaillencourt, G. W. (1981). "Crushed limestone neutralization of dilute acidified
Adirondack surface waters," thesis presented to Cornell University, at Ithaca, N.Y.,
in partial fulfillment of the requirements for the degree of Master of Science.
Westall, J. C , Zachary, J. L., and Morel, F. M. (1976). "MINEQL: A computer
program for the calculation of chemical equilibrium composition of aqueous sys-
tems." TN-18 Parson Laboratory, Massachusetts Institute of Technology, Cam-
bridge, Mass.

APPENDIX II. NOTATION

The following symbols are used in this paper:

a = interfacial area of limestone particles per unit volume of in-

terstitial fluid;
C = reactant concentration and bulk-fluid calcium concentration;
Ca = influent concentrations of anions excluding inorganic carbon
species and hydroxide;
Cb = initial bulk solution calcium concentration;
Cc = influent concentration of cations excluding calcium and hy-
drogen;
CbL = calcium concentration at any point in contactor;
Cb0 = molar concentration of calcium in influent flow;
Ceq = calcium concentration in fluid when calcium carbonate in
limestone is in equilibrium with influent flow;
D = calcium-ion diffusivity;
Dx = temperature corrected value of the calcium-ion diffusivity;
D20 = calcium-ion diffusivity at 20° C;
DIQ = influent dissolved inorganic carbon concentration;
d = volume-mean limestone-particle diameter;
F = dispersion coefficient;
{Heq} = equilibrium interfacial hydrogen-ion activity;
I = ionic strength;
Jo = initial rate of increase of calcium-ion concentration in bulk
solution;
Ksp = effective, activity-coefficient-corrected solubility product for
calcium carbonate in limestone;
Ky, = activity-coefficient-corrected ion product for water;
k0 = overall dissolution-rate constant;
kc = surface-reaction rate constant for calcium dissolution;
kf = surface-reaction constant at pH = 9.5;
kL = liquid film-transport coefficients;
kf = residue-layer film-transport coefficient;
ka exp = best-fit experimental value of ka\

357

J. Environ. Eng., 1991, 117(3): 339-358

 k0 model = k0 predicted by model;
L — overall depth of limestone in column;
MR = modified Reynold's number;
ND = dimensionless axial dispersion number product;
n — number of sampling ports in column;
pHe, = equilibrium interfacial pH;
pH0 = bulk solution pH;
Downloaded from ascelibrary.org by Cardiff University on 02/17/23. Copyright ASCE. For personal use only; all rights reserved.

r = reaction rate expression;

S = molar concentration of calcium carbonate dissolved;
S' = molar concentration of CaC0 3 dissolved at equilibrium;
Sc = Schmidt number;
T = temperature in degrees Celsius;
i = mean fluid residence time;
Us = superficial or empty-column fluid velocity;
Z = dimensionless axial distance;
a!,a 2 = ionization fractions for deprotonation of carbonic acid;
8 = thickness of residue layer;
e = bed porosity;
er = porosity of residue layer;
v = kinematic viscosity;
Vx,v20 = kinematic viscosity of water at temperatures of x and 20° C,
respectively;
Tr = pore length (tortuosity) factor for residue layer; and
*9 = limestone-particle sphericity.

358

J. Environ. Eng., 1991, 117(3): 339-358