Document Sample
					                         ASSESSING THE SUITABILITY OF LIMESTONE
                          FOR USE AS WATER STABILISATION MEDIA

                                    G J DU PLESSIS and G S MACKINTOSH

                Cape Water Programme, CSIR Western Cape, P O Box 320, Stellenbosch, 7599


A number of limestone mediated stabilisation processes have been developed for the treatment of aggressive and
corrosive waters, to prevent attack of potable water distribution networks. Common to all of these processes is the
need for the identification and commercial sourcing of a limestone aggregate suitable for stabilisation purposes,
both in terms of mineral composition and dissolution characteristics.

This paper describes the modelling of the calcite dissolution process in fixed-bed reactors, and presents results
of mineral analyses and kinetics investigations of four different calcite deposits. The deposits investigated include
those of a porous, friable sedimentary nature, grey-coloured calcitic marble and a white-coloured crystalline
deposit. Calibration of the dissolution model is shown for the different types of calcite pebbles. Finally, the
dissolution rate constants of the four types of pebbles are used as a basis for comparison of their dissolution


The first attempts to stabilize municipal water were made using limestone. The first documented application
occurred in 1906 in Frankfurt, Germany, when domestic water was treated by contacting it with a marble bed. It
was shown that with such treatment, corrosion of the distribution network was significantly reduced. However,
the availability of the more readily soluble lime resulted in the use of limestone falling from favour. Although there
are few documented cases of full scale use of the technology, it would appear that there has been recent renewed
interest in limestone mediated stabilisation and that a number of small plants have been installed in the USA and

In South Africa, Cape Water Programme, CSIR, reinitiated research into limestone stabilisation in the early 1990's 3.
The associated development process has included: identification and experimental assessment of potentially
suitable limestone deposits in Southern Africa; interaction with limestone mines and chemical suppliers to ensure
commercial supply of water treatment grade limestone pebbles; kinetic modelling of limestone dissolution; pilot
plant and prototype full scale plant assessments; and commissioning and trouble shooting of full scale limestone
contactor units. To-date, limestone contactor units have been installed for stabilisation of municipal drinking water
at nine locations in the Western Cape, their capacities ranging between 2 and 13 ML/d. Similar implementation
of limestone mediated stabilisation in other regions requires the sourcing of suitable limestone aggregates outside
the Western Cape. This paper assesses four deposits from different regions in Southern Africa in terms of their
suitability for stabilisation purposes, both in terms of mineral composition and dissolution characteristics.


The rate at which a mineral dissolves in aqueous solution is influenced by a number of factors such as
temperature, chemical composition, physical and chemical properties of the dissolving mineral and the manner
in which the mineral is brought into contact with the solution.
An assessment of the rate controlling reaction (surface reaction vs diffusion) was carried out by Kornmáller and
Loewenthal2, using an upflow reactor system and limestone granules from Bredasdorp in the Western Cape. The
effects of mixing energy on the observed rates of dissolution were investigated by varying the flow characteristics
in the pH region 5 # pH # 9 . Their observations indicated that the calcite dissolution rate is controlled by
transport/diffusion phenomena between the surface of the dissolving mineral and the bulk solution. The chemical
reaction rate at the surface of the solid calcium carbonate is sufficiently high to maintain a thin mono-layer located
immediately adjacent to the solid in an equilibrium condition with respect to the mineral. Transport of dissolution
products from this layer to the turbulent bulk liquid is by means of diffusion through the boundary layer.

The difference between the equilibrium activity in the mono-layer and the active concentration (activity) of the
corresponding species in the bulk solution is the force driving the process.

               kd S
  Rate '            aeq & a
               ä V
                                     1              1            1
         ' kd        (hyd)   S       2
                                   ksp c & ( Ca 2% ) 2 ( CO32& ) 2

where       aeq , a          =   equilibrium activity and activity of species
            ()               =   activity of species, moles/R
            ä                =   thickness of boundary layer
            S                =   surface area of calcium carbonate crystals per unit volume in the solution
            kd               =   dissolution rate constant
            k d (hyd)        =   dissolution rate constant including hydrodynamic effects
            k sp c           =   solubility product for CaCO3

This equation is identical in form to the equation proposed by Sj" berg4 based on observations of calcium carbonate
dissolution in solution in the region 5 # pH # 9, ie.

                                          1              1           1
    d [ Ca 2% ]           )                       2&
                ' k(S) S ksp c 2 & [Ca 2% ] 2 [CO3 ] 2

where       k (S)            = dissolution rate constant used in the Sj" berg equation (ms-1)
            kNsp c           = apparent solubility product for CaCO3

The above equation appears to have obtained wide acceptance as a model describing calcium carbonate
dissolution kinetics in the pH region 5 # pH # 9.

In their study of the dissolution kinetics of Bredasdorp limestone, Kornmáller and Loewenthal2 replaced k(S) S in
the above equation with a single constant, kDC compound), to obtain the following equation:

                                                                 1         1         1
                   d [ Ca 2% ]                       )                       2&
                               ' k D.C. compound fd ksp c 2 & [Ca 2% ] 2 [CO 3 ] 2

 k D.C. compound             =           compound rate constant for a diffusion controlled reaction, which varies with temperature,
                                         ionic strength, physico-chemical properties of the mineral and hydraulic characteristics
                                         (flow rate), and includes a surface area term.
 [Ca2+ ]                     =           concentration of Ca2+ in the bulk solution
 [CO32- ]                    =           concentration of CO32- in the bulk solution
 fd                          =           activity coefficient for the divalent ion
 k'sp c                      =           apparent solubility product for CaCO3
Experimental data was obtained by analysing
water samples from a series of sampling ports, in
a tubular upflow reactor packed with limestone
granules. The above rate equation was used in a
computer simulation of the reactor, assuming plug-
flow conditions. The reactor was simulated as a
number of small reactors in series along the
longitudinal axis. The compound rate constant,
k DC compound was varied to obtain a best least squares
fit of the model to the experimental data.

In Figure 1* is shown plotted the compound rate
constant versus the loading rate for the three
granule sizes 3.55 mm, 5.7 mm and 8.05 mm.
This plot indicates a closely linear relationship
between the compound rate constant and the
loading rate for each nominal mean granule size.           Figure 1 Compound rate constant versus loading rate plot for
Linear regressions were carried out on this data in                 3.55 mm, 5.7 mm and 8.05 mm mean size
order to obtain the best fit line representing each of              Bredasdorp limestone granules in acidified tap water
the data sets and are plotted in the diagram.

Experimental data gives a compound rate constant
which increases with decrease in granule size.A
plot of kDC compound versus the ratio of loading rate to
mean granule diameter is shown in Figure 2. In this
form experimental data of all three granule sizes is
closely linear. A linear regression was carried out
to obtain an equation representing the effects of
both granule size and loading rate on the                                                               8.05 mm
compound rate constant giving,
                                                                                                        5.7 mm

                                                                                                        3.55 mm
  k D.C. compound   (ö )
                           ' 0.0021      % 0.0044
                                       ö                                                                Regression line


                                                           Figure 2 Compound rate constant versus LR/ö for all
ö          = nominal mean granule size (mm)                         limestone granules investigated in acidified tap
LR         = loading rate (m hr-1)                                  water

               Figures 1 and 2 duplicated from Kornmüller and Loewenthal, 19952

It is important to note that this equation is valid only for granules obtained from the Bredasdorp deposit, and within
the mean nominal granule size range of 8.05 mm to 3.55 mm.

As far as the physical characteristics of the mineral are concerned, calcium carbonate occurring in natural deposits
may vary from a dense metamorphosed marble through a pure non metamorphosed chalk to a friable impure
permeable form. In general each calcium carbonate deposit will give rise to different rate constants. A model
based on the rate equation of Kornmüller and Loewenthal2 was used to assess the dissolution kinetics of a number
of different limestones. The model was shown to be suitable for use in the design of upflow reactors treating
terrestrial waters in the pH range of 5 to 9, under normal temperatures and at atmospheric pressure.

The following four limestones were included in this investigation:

A:     (calcrete from Bredasdorp)- primary limestone, poorly indurated with a creamy-brown appearance. The
       particles are porous and friable, and therefore tends to form many fines. However, the porosity is regarded
       to be an asset w r t dissolution, because it increases the exposed dissolution surface.
B:     Calcitic marble) - marble-like grey appearance, very hard and hardly produce any fines.
C:     a grey coloured calcitic marble, similar in appearance to B, and hardly produces any fines.
D:     visible crystalline structure with a “soapy” feel when wet. It too does not produce much fines.

The mineral composition of the limestones are shown in Table 1 below:

Table 1: Mineral composition of various limestones (reported in mg/g on a dry basis)
                                A                    B                     C                     D

    Calcium                    370                  381                   357                  213

    Magnesium                   4.6                 8.56                  19                   129

    Iron                        0.7                 0.11                  0.05                  1.4

    Acid Insolubles             14                  50.2                  46.0                 24.0

The amount of acid insolubles gave an indication of the degree of solubility in a concentrated acid solution. A high
value for this parameter indicates that the limestone is highly insoluble in acid, and would therefore be very
insoluble in water. This fraction probably mainly comprises of silica. A high acid insoluble content is undesirable,
because it indicates the probability of the formation of a large percentage of undissolved residue in a fixed
limestone bed. This may result in either blockage of the bed, or carry-over of insolubles into the product stream,
which may settle in pipes, eventually restricting flow. Evidence exists that the rate of dissolution decreases as
the magnesium content of the stone increases (Letterman et al, 1991). From this initial investigation indications
were that Deposits A, B, C, but not D, would be suitable.


Fixed-bed kinetics tests were carried out on Deposits A (Bredasdorp) and B. Comparison of the dissolution rate
constant of Deposit B with that of the Bredasdorp deposit —which has been shown to be effective as stabilisation
media— would give an indication of its suitability as stabilisation media.

Fixed bed tests were performed using a test rig consisting of clear PVC columns of height 2000 mm and diameter
160 mm. Each column had three sampling valves along its side, at relative heights above the bed supporting base
of 370 mm, 735mm and 1240 mm respectively. Two columns were filled with limestones A and B respectively.
Particle sizes ranged between 12 and 15 mm. The heights of the fixed beds were 1600 mm above the supporting
base. The final product outlet was situated just above the top of the beds.
The bed loading rate affects the compound rate constant (see paragraph 2). To determine the effect of loading rate,
loading rates of 3.13 m/h and 11.32 m/h respectively were therefore used.

A 200 R drum was used to make up a batch of acidified (to pH ± 4.00) Stellenbosch tap water. The drum was filled
after each run, and the pH adjusted. A period of three times the retention time of the water in the entire bed was
allowed, after which steady state conditions were assumed. After that, pH was measured at the various sample
points. A top-down sampling sequence was followed, to minimise interference with conditions inside the bed. A
special sampling device was constructed, allowing in-stream pH measurement, whilst minimising contact between
the measured water and the atmosphere. The Samples were then filtered and submitted for calcium, total alkalinity
and electrical conductivity analyses. The total acidity for each sample was determined using Stasoft 1. The total
acidity was assumed to stay constant throughout the reactor during each run. The total acidities of all samples
taken during a run were averaged to determine the total acidity for that run.

A computer model based on Kornmüller and Loewenthal’s calcite dissolution rate equation was set up to simulate
the change in water quality through a plug-flow reactor, at time-of-travel intervals of 3 s. For each interval, new
calcium and alkalinity values were calculated from the above rate equation, and a new pH value was determined
using an iterative calculation method, assuming the total acidity stayed constant throughout the reactor. By trial-
and-error adjustment of the compound rate constant and comparison of the modelled and experimentally
determined water quality determinants, compound rate constants were estimated for each run. Best-fit plots are
shown in Figures 3 to 6. The simulations are shown as lines, and the experimentally obtained values as markers.

        9                                                        50                     9                                                                   50

        8                                                        40                     8                                                                   40
                                                                      mg/L CaCO3

                                                                                                                                                                 mg/L CaCO3
        7                                                        30                     7                                                                   30


        6                                                        20                     6                                                                   20

        5                                                        10                     5                                                                   10

        4                                                        0                      4                                                                   0
            0        1             2              3          4                              0   2        4   6         8        10    12          14   16
                         Time-of-travel (min)                                                                Time-of-travel (min)

                pH              Calcium         Alkalinity                                          pH              Calcium          Alkalinity

Figure 3 Deposit A: fixed bed kinetics tests with loading Figure 4 Deposit A: fixed bed kinetics tests with loading
         rate 11.3 m/h; compound rate constant for best-           rate 3.1 m/h; compound rate constant for best-
         fit: 0.007 s-1.                                           fit: 0.008 s-1.

It is evident from Figures 4 and 6 that the simulated pH values deviated from the experimentally measured pH values.
These deviations can be ascribed to error in the pH measurements. The top of the limestone columns were open to the
atmosphere, allowing carbon dioxide exchange between the sampled water and the atmosphere. Such interaction
explains the lower than expected pH values of the experimentally determined values.
       9                                                                           60                       9                                                                              60

                                                                                   50                                                                                                      50
       8                                                                                                    8

                                                                                   40                                                                                                      40

                                                                                        mg/L CaCO3

                                                                                                                                                                                                mg/L CaCO3
       7                                                                                                    7

                                                                                   30                                                                                                      30
       6                                                                                                    6
                                                                                   20                                                                                                      20

       5                                                                                                    5
                                                                                   10                                                                                                      10

       4                                                                           0                        4                                                                              0
           0                1             2                   3               4                                 0       2        4        6         8        10     12           14   16
                                Time-of-travel (min)                                                                                      Time-of-travel (min)

                       pH              Calcium              Alkalinity                                                               pH           Calcium         Alkalinity

Figure 5 Deposit B: fixed bed kinetics tests with loading Figure 6 Deposit B: fixed bed kinetics tests with loading
         rate 11.3 m/h; compound rate constant for                 rate 3.1 m/h; compound rate constant for best-
         best-fit: 0.007 s-1.                                      fit: 0.007 s -1.

   The above fixed bed tests required approximately 100 kg of stone, and a test rig complete with pumping
   equipment. Dissolution tests in stirred beakers were carried out with the same deposits, determining the
   same water quality determinants and following the same methodology for determining the compound rate
   constants as was used in the fixed bed experiments. The compound rate constants determined via the
   beaker tests were similar to those determined via the fixed bed tests (see Table 2: Summary of rate
   constant results). Stirred beaker tests were subsequently carried out on Deposits C and D, of which too
   small sample sizes were available for the fixed bed tests.

   Dissolution tests in stirred beakers and simulations with the calcite dissolution model yielded the following
   results (see Figures 7 and 8). The lines are representing the model simulated values, and the markers the
   experimentally determined values.
       9.5                                                                         30                       7.5                                                                            20

                                                                                   25                           7
           9                                                                                                                                                                               15
                                                                                        mg/L CaCO3

                                                                                                                                                                                                mg/L CaCO3
                                                                                   20                       6.5


       8.5                                                                                                                                                                                 10
                                                                                   15                           6

           8                                                                                                                                                                               5
                                                                                   10                       5.5

       7.5                                                                         5                            5                                                                          0
               0   2        4   6      8          10   12         14     16   18                                    0       2        4     6         8       10      12          14   16
                                            min                                                                                            Contact time (min)

                        pH                 Calcium          Alkalinity                                                          pH                Calcium           Alkalinity

Figure 7 Deposit C: stirred beaker kinetics tests;                                                   Figure 8 Deposit D: stirred beaker kinetics tests;
         compound rate constant for best-fit: 0.008 s-1.                                                      compound rate constant for best-fit: 0.002 s-1.

   From the results and the determined compound rate constant it can be seen that Deposit C’s dissolution
   characteristics are very similar to that of Deposits A and B. It should therefore be suitable as stabilisation
   media. Deposit D, however, with the visibly crystalline structure and relatively high magnesium content, had
   a much lower compound rate contstant (0.002 s-1, compared to >0.007 s-1 for the other deposits). It can be
   seen from the experimentally obtained calcium, total alkalinity and pH levels for Deposit D in Figure 8, that
   after five minutes contact time no more calcium carbonate went into solution. This clearly shows that
   Deposit D will not be suitable as stabilisation media.

A summary is given below of the results of the different rate constant determinations.

Table 2: Summary of rate constant results (kD.C. COMPOUND, in s-1)
                        Deposit                             A               B               C              D

                        Loading rate 3.1 m/h               0.008          0.007             -              -
    Fixed bed tests
                        Loading rate 11.3 m/h              0.007          0.007             -              -

                        Tap water                          0.008             -              -              -
    Stirred beaker
                        Distilled water                    0.01           0.006           0.008          0.002
                        Simplified procedure               0.011          0.006           0.007          0.002

The compound rate constant is expected to increase with loading rate1. However, no significant influence
of loading rate on the rate constant can be observed from the fixed bed test results. It is suspected that the
experimental and/or data-analysis procedures were not sensitive enough to reflect the effect of loading rate
on the rate constant.


There is reasonable consensus in the literature that limestone dissolution kinetics in the pH region of 5.0
to 9.0 can be modelled using a diffusion controlled reaction rate equation, confirmed by Kornmüller and

CSIR investigation of the dissolution kinetics of two different limestones showed that a model based on the
above mentioned rate equation effectively predicts calcium carbonate uptake in fixed limestone bed reactors.

Stirred open-beaker tests may be used to obtain rough estimates of calcium carbonate dissolution rate
constants for different limestones.

Calcium carbonate dissolution kinetics tests have shown four limestones from different regions in Southern
Africa to be suitable for use in fixed-bed limestone contactors as stabilisation media for potable water.
However, the long term practical feasibility of using Deposits B to D should be confirmed with pilot-scale


[1]    Friend JFC and Loewenthal RE (1992) Chemical Conditioning of Low and Medium Salinity Waters: STASOFT
       Version 3.0. Water Research Commission of South Africa, PO Box 824, Pretoria, 0001.

[2]           "
       Kornmuller U.R and Loewenthal RE (1995) Limestone dissolution kinetics in upflow systems, MSc
       Dissertation, Dept Civil Engineering, Univ Cape Town.

[3]    Mackintosh G.S., De Villers H.A., Du Plessis G.J., Loewenthal R.E., Kornmüller U.R. (1998). Stabilisation of
       soft acidic waters with limestone. Report to the Water Research Commission, WRC K5/613.

[4]    Sjöberg E. L. (1976) A fundamental equation for calcite dissolution kinetics. Geochimica et Cosmochimica
       Acta vol. 40, pp 441-447.

Shared By: