The kinetics of carbonate scaling application for the prediction of downhole carbonate scaling.pdf

Embed Size (px)

Citation preview

  • .Journal of Petroleum Science and Engineering 29 2001 8595www.elsevier.nlrlocaterjpetscieng

    The kinetics of carbonate scalingapplication for the predictionof downhole carbonate scaling

    Yuping Zhang a,), Harry Shaw b, Rod Farquhar a, Richard Dawe ca Read Well Serices Ltd., Offshore Technology Park 1, Claymore Aenue, Aberdeen AB23 8GW, UK

    b Centre for Petroleum Studies, T.H. Huxley School, Imperial College, London SW7 2BP, UKc Department of Petroleum Engineering, Uniersity of West Indies, St. Augustine, Trinidad and Tobago

    Received 10 January 2000; accepted 31 October 2000

    Abstract

    Reliable prediction of calcium carbonate, CaCO , scaling for estimating scale production oilfield production wells and3surface facilities requires both thermodynamic models to indicate the tendency for scaling from solution and kinetic modelsto predict the rate of scaling and thus the time required to cause blockage. The application of such models could contributeto field scale management and in the development of more effective treatments of carbonate scale during oilfield production.The performance of a kinetic model for calcium carbonate scaling rate based on the measurements of individual calcitecrystal growth rates is tested against scale deposition in tube blocking experiments using synthetic formation brines underpressure and temperature levels. The profile of the scale formed on the internal surface of the tube during the experimentswas measured and compared with the predicted results from the kinetic model. The application of the kinetic model to theprediction of carbonate scaling in a North Sea well has also been tested. The results from the model are in good agreement

    .with the actual scale profile recorded in the tubing by a downhole multi-finger caliper MFC tool. q 2001 Elsevier ScienceB.V. All rights reserved.

    Keywords: Scale prediction; Scaling kinetics; Carbonate scale; Crystal growth; Nucleation

    1. Introduction

    Chemical scale formation associated with the pro-duction of formation and injection waters can resultin oilfield production problems which are frequentlyexpensive to remedy. Calcium carbonate, CaCO , is3one of the most common scale components found inoilfield production wells and surface facilities. Car-

    ) Corresponding author. .E-mail address: [email protected] Y. Zhang .

    bonate scale formation can impair production byblockage of tubing and flowlines, fouling of equip-ment and concealment of corrosion. In order tomanage a potential scale problem, it is important toknow where and how much CaCO scale will be3deposited during oil and water production. Effectiveprediction of scaling requires a reliable thermody-namic model for the prediction of the scaling ten-dency and a kinetic model for the prediction ofscaling rate. Normally, the field operator does nothave control over all the variables that determinescale deposition, particularly the water composition,

    0920-4105r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. .PII: S0920-4105 00 00095-4

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 859586

    but the application of these models could help iden-tify some production parameters that could be con-trolled. This could then minimise the scale deposi-tion as well as facilitate the design of more effectivetreatment of carbonate scale during oilfield produc-tion.

    The scale tendencies of the solutions used in ourexperiments were calculated by a thermodynamic

    .modelSolmineq 88 Kharaka et al., 1988 . Thekinetic model has been developed from measure-ments of the growth rate of individual calcite crystals

    in a dynamic flowing system Zhang, 1997; Zhang.and Dawe, 1998 . The rate equation and rate con-

    stant obtained from our previous experiments are:20.5R sk S y1 1 . .L p0.5log k s0.126 IS y2400r Tq273 y2.11 . .p

    2 .

    where R and k are the crystal surface linearL pgrowth rate and rate constant in metre per second .mrs , respectively, IS is the ionic strength in mole

    .per kg molrkg water, T is temperature in centi-grade and S is the saturation ratio; for CaCO , S can3be expressed as:Ssa 2qa 2yrK 3 .Ca CO sp3where a is the activity of the ith item for a givenisolution and K is the solubility product of thesp

    .scale forming mineral. The saturation index SI ,equal to logS, is also commonly used to expressscaling tendency. To validate the kinetic model forCaCO scaling rate prediction under oilfield down-3hole conditions, tube blocking experiments have beenconducted under controlled saturation states and tem-perature and pressure conditions. The scale profile

    built-up along the length of the tube predicted by thekinetic model has been compared with the measuredscale thickness in the tubes produced by the experi-ments. The kinetic model has also been used toanalyse carbonate scaling in a North Sea well andthe performance of the model compared against theactual scale profile measured by a downhole Multi-

    . .Finger Caliper MFC survey Desbrandes, 1985 .Finally, the general application of the kinetic modelto the management of scale downhole is also dis-cussed.

    2. Experimental

    Tube blocking experiments were conducted in adynamic flowing system to model the actual scalingprocess. Two different solution systems were usedfor the experiments: the first is a simple NaClCaCO H O system, and the second is a synthetic3 2formation water which simulates a typical oilfieldformation water. Table 1 gives the water composi-tions used for the tube blocking experiments and thecalculated saturation indices at ambient conditions.To maintain a constant pH, CO saturated solutions2at ambient conditions were used in most of ourexperiments. The composition and pH of the solu-tions were monitored throughout the experimentsand no change was recorded during two days stand-ing time for each batch of solution. The saturation

    indices of the solutions at room temperature around.208C and experimental conditions were calculated

    using Solmineq 88. Large errors in the calculation ofthe saturation states of calcium carbonate solutions at

    .high pHs )7 from Solmineq 88 have been noticed .previously Zhang, 1997 . Corrections have been

    Table 1 .The chemical compositions molrkg of the brines used in tube blocking, their calcite saturation indexes and pH at the ambient conditions

    q q 2q 2q y y 2yNo. Na K Ca Mg Cl HCO SO pH SI3 4 20 201 1.00 0 0.00305 0 1.00 0.00613 0 7.98 0.392 1.03 0 0.01720 0 1.03 0.03442 0 6.07 0.073 1.04 0 0.01851 0 1.04 0.03628 0 6.10 0.154 1.28 0.05 0.02482 0.011 1.37 0.03276 0.0011 6.01 0.105 1.28 0.05 0.02412 0.012 1.37 0.03288 0.0011 6.00 0.106 1.29 0.05 0.02708 0.013 1.38 0.03615 0.0011 6.05 0.23

    pH and SI are the brine pH and SI at 208C and 1 bar pressure.20 20

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 8595 87

    made by changing the two complex constants KCaCO 3.yand K according to the experimental dataNaCO 3

    .obtained from previous experiments Zhang, 1997 .A schematic representation of the tube blocking

    experiment apparatus is presented in Fig. 1. A solu-tion containing both calcium and carbonate ions waspumped through a stainless steel tube placed in awater bath heated to the required experimental tem-perature. The tube length was about 1 m and internaldiameter was 3.04 except for experiment No. 3 inwhich the tube internal diameter was 2.28 mm. Thepumping rate for all the tests was fixed at 90 mlrh.A back pressure regulator was installed at the end ofthe test tube to set a required system pressure. Thetemperature profile inside the tube was measuredwhile the solution was being pumped by using atemperature sensor inserted into the tube. At 708Cand a flow rate of 90 mlrh, the pumped solutionreached the water bath temperature within 3 cm.Water samples were collected at the tube outlet atregular time intervals to monitor the Ca2q ion con-centration and alkalinity. As the carbonate scaleformed in the tube, there were pressure drops overthe test tubes. When the pressure drop increasedabove 1 bar, the pump was stopped and the experi-ment ceased. The blocked tube was washed with aCaCO saturated solution to remove soluble salts3and then cut into small pieces to inspect the scaleprofile. The thickness of the carbonate scale withineach cross-section of tube was measured under amicroscope. Some scale deposits were also examined

    .using scanning electronic microscopy SEM and .X-ray diffraction XRD techniques to determine the

    scale morphologies and mineralogies.

    3. The calculations of CaCO scale profile along a3tube

    . .According to Eqs. 1 3 , CaCO scaling rate is3a function of its saturation index. When scale isdeposited inside a tube, the concentration of scale-forming ions decreases, resulting in a change insaturation index as well as scale thickness down-stream. The amount of scale deposited inside thetube is a function of the crystal growth rate, the totalcrystal surface area and solution flow rate. If it isassumed that the total surface area for scaling isequal to the tube internal surface area and the tube

    .length of L m is uniformly sliced into n sections,the scale thickness in one section, d in meter, canxbe obtained from:d s R t 4 . .x L xwhere x gives the positions of the tube, xs1, 2,

    .3 . . . n, R is the scale growth rate in x sectionL xand t is the total scale forming time in second. Theamount of ions precipitated from one kg of waterwithin one section length of the tube can be calcu-lated from:

    22543p r y ryd r L .V r . x ss sxPPT s s . . x V M nMQ tw w5 .

    .PPT is the total precipitation within the x sectionx .in molrl, V is the total scale volume formed in xs x

    section when a V volume of solution flowingwthrough the tube, r is the density of the scales 3.kgrm , M is the molecular weight of the scale

    Fig. 1. A schematic representation of the tube block experimental system.

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 859588

    . .component kgrmol , r is the radius of the tube mand Q is the water production rate in barrels perwday. The scale forming ions at the end of the xsection can be then calculated from:

    C 2q s C 2q y PPT 6 . . . . . xxq1Ca Ca xAlk sAlk y2 PPT 7 . .x xq1. xwhere C refers to the concentration of ith item, Alkiis the alkalinity of the water system. For oilfieldproduced water, the alkalinity is roughly equal to theconcentration of HCOy if there is no organic acid3presence in the water system. The CO2y concentra-3tion can be calculated from solution pH and alkalin-ity. The saturation index at the following section . 2qxq1 can be then obtained from the Ca andCO2y concentrations.3

    An oilfield production system is more compli-cated than the system used in our experiments, nor-mally including water, oil and gas. The water pHvaries with the amount of CO dissolved in the water2phase, which is influenced by the partitioning of CO2between the water, oil and gas phases. An estimateof downhole pH can be made from the water chem-istry and the CO partial pressure in the gas phase.2The CO partial pressure can be obtained from:2P sPf X 8 .CO g CO2 2where P is the CO partial pressure, P is theCO 22total pressure inside the tubing, f is the CO gasg 2fugacity coefficient and X is the mole fraction ofCO 2CO in the gas phase. Both f and X are func-2 g CO 2tions of pressure and temperature. According to Oddo

    .and Tomson 1994 , f and X under oilfieldg CO 2downhole conditions can be calculated from:

    2.05y4f sexp P 4.12=10 y 9 .g / /Tq273

    and:

    4 pf Q q2Q =10y4 .g w ooX sX 1.0qCO CO2 2 /Q Tq273 .g10 .

    where X o is the mole fractions of CO in gasCO 22phase at standard conditions, respectively, Q is oilo

    .production rate barrels per day , Q is the total gasg .production rate million standard cubic feet per day ,

    and P is the pressure in bar. NB the constants havebeen changed as the units of P and T used here are

    .different in the Oddo and Tomson, 1994 paper.

    4. Results and discussions

    Table 2 lists the calculated solution pH values andsaturation indices under the test temperatures andpressures. The measured tube blocking time andCa2q concentration at the outlet of the tubes for all

    . .six tests are given in Table 2. If Eqs. 1 and 2were used to calculate blocking times for all the tubeblocking experiments, only experiment No. 1 givesgood agreement with the measured blocking time.For the other five tests, the calculated blocking timesare about three times longer than the measured val-ues. As we can see, the significant difference be-

    Table 2Prediction of CaCO scaling rate using our kinetic model3

    2q 9 . . . . .No. T 8C P bar pH SI Block time day Ca k mrs=10TP TP out pM C M C

    1 90 10 7.47 0.68 6.0 6.0 0.0026 0.0025 2.52 90 10 5.96 0.64 2.1 1.9 0.0121 0.0114 8.63 70 10 5.98 0.52 5.0 5.9 0.0141 0.0141 3.54 90 100 5.87 0.58 3.5 3.4 0.0207 0.0193 5.85 90 15 5.88 0.62 3.3 3.0 0.0190 0.0184 5.76 70 15 5.92 0.58 8.2 9.0 0.0230 0.0224 2.2

    2q. 2qpH and SI are the pH and SI at the experimental pressure and temperature. Ca : Ca concentration at the out let,PT PT outCcalculated, Mmeasured.

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 8595 89

    tween experiment No. 1 and the others is the solutionpH, suggesting that the CaCO scaling rate is strongly3influenced by solution pH. For constant saturationindex, the lower the solution pH, the higher thecalcite growth rate. The rate constant given in Eq. .2 was obtained from test solutions with relatively

    . .high pH approximately 8 Dawe and Zhang, 1997 .The influence of pH on calcite growth rate has also

    been reported in other studies Dreybrodt et al.,1997; Przybylinski, 1985; Zhang and Dawe, 1998;

    .Zhang and Grattoni, 1998 . Therefore, the influenceof pH on calcite growth rates has now been includedin the rate constant equation:

    0.5log k s0.126 IS y2400r Tq273 . .py0.34pHq0.444. 11 .

    For the synthetic formation brine solution contain-ing Mg2q, the inhibition effect of Mg2q on calcitegrowth rate also needs to be considered. For MgrCamolar ratios between 0.1 and 0.5, the growth rate is

    reduced by 40% as found in previous studies Dawe.and Zhang, 1997 . The rate constant for the solutions

    containing low concentration of Mg2q should becalculated from:

    0.5log k s0.126 IS y2400r Tq273 . .py0.34pHq0.222. 12 .

    The tube blocking time for each experiment, cal- . .culated using Eqs. 11 and 12 , are listed in Table

    2. It seems that although many other ions were

    Fig. 2. The measured Ca2q concentrations at the outlet of thetube.

    present in the complex water systems, the calcitegrowth rate can be well described by consideringonly the inhibition from Mg2q.

    Variations in the analysed Ca2q concentrationswith time at the outlet of the tube for test No. 6,together with the calculated Ca2q concentrations atthe outlet and at the equilibrium state are shown inFig. 2. As we can see, the Ca2q concentrations at theoutlet gradually decreased with time. At the end ofthe experiment, the measured outlet Ca2q concentra-tion was very close to the calculated value.

    Most natural scaling processes starts by heteroge-neous nucleation with many isolated and uncon-

    nected crystals formed on the metal surface Nancol-.las and Sawada, 1980; Sohnel and Garside, 1992 .

    To simulate the natural scaling process, the satura-tion indexes of all the solutions used in our experi-ments are designed to be below the levels at whichhomogeneous nucleation1 under the test conditionsmay occur. Therefore, the reduction of scale formingions through the tube can be considered as due to theformation of scale. As the surface area of the scalecrystals is initially smaller than the total tube internalsurface area, the concentration of scale forming ionsat the outlet of the tube will be higher than thecalculated values at the initial stage of scale forming.As the size of the nuclei increase with time, theCa2q concentration will gradually decrease until thewhole surface of the tube is covered with crystals. Aconstant Ca2q concentration at the tube outlet can beachieved or the Ca2q concentration slightly increasesas a result of the tube inner diameter decreasing.

    .The SEM photomicrographs Fig. 3 of carbonatescales show that the inner surface of the tube near

    .the inlet Fig. 3a has been covered by scale, whereas .the surface near the outlet Fig. 3b has not been

    completely covered. This may be the reason why ourmeasured Ca2q concentration did not reach the sta-ble stage for this test.

    The scale developed along the tube for test No. 6is shown in Fig. 4. As we can see, the highestscaling rate is located at the inlet of the tube where

    1 The critical saturation indexes for homogeneous nucleation,SI , can be calculated from the critical free energy required tocform a solid phase in a solution, for calcite SI s0.75 at 908Cc .Mullin, 1993 .

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 859590

    the temperature reaches the water bath temperature.The scale profile along the tube for test No. 6 hasalso been calculated and compared with the mea-

    .sured scale profile Fig. 5 . The predicted scaleprofile matches the measured values very well at theupstream end of the tube, but are lower than themeasured values near the outlet of the tube. This

    may be because before the tube surface is completelycovered with crystals, the saturation index at the endof the tube is higher than the calculated value, whichprovided a higher scaling rate than the calculatedvalues.

    The XRD analysis shows that almost all the pre-cipitated scale is calcite when simple water systems

    . . .Fig. 3. The CaCO scales and the morphologies SEM photomicrographs . a across length of the picture represents 2.3 mm the scale3 . .formed near the inlet of the tube, calcite is the main mineral of the scale; b across length 1.7 mm the scale formed near the outlet of the

    . .tube, both calcite rhombohedron and aragonite hexagon are found.

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 8595 91

    .Fig. 3 continued .

    were used in the experiments, but in the experimentsusing synthetic formation water initial MgrCa ratio

    .0.5 , the precipitated scale is composed of 90%calcite and 10% aragonite. The SEM photomicro-graphs of scale from these experiments show evi-dence of the characteristic morphologies of aragonite .Fig. 3b . Calcite is the most common form of the

    three CaCO mineral polymorphs calcite, aragonite3.and vaterite observed in downhole carbonate scales

    although other carbonate minerals have also been .reported Nancollas and Sawada, 1980 . The miner-

    alogy of the precipitates depends mainly on thesaturation index, temperature and the presence of

    2q.inhibitors e.g. Mg . Many studies have shownthat Mg2q inhibits the growth rate of calcite but not

    aragonite Akin and Lagerwerff, 1965; Morse andMackenzie, 1990; Mucci, 1987; Paquette et al.,

    .1996 . In general, the higher the MgrCa ratio, tem-

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 859592

    .Fig. 4. The photographs of the scale formed at different positions along the tube exp. No. 6 .

    perature and saturation index, the higher the percent- .age of aragonite precipitated Zhang, 1997 . The

    saturation index under downhole conditions is nor-

    mally much lower than that used in our experiment .demonstrated later and the MgrCa ratios of most

    formation waters are below 0.5 Warren and Smal-

    .Fig. 5. The measured scale profile along a metal tube compared with the calculated scale profile No. 6 .

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 8595 93

    .ley, 1994 . From such solutions with a low saturationindex and low MgrCa ratios, it is more likely for

    .calcite to form Zhang and Dawe, 1998 . Therefore,our kinetic model obtained from the growth rate ofcalcite can be generally used for the prediction ofCaCO scaling rates.3

    5. Application for the prediction of oilfield down-hole scaling

    Analysis of the scale profile logged in a well froma North Sea field with a recognised carbonate scaleproblem has been used to evaluate the potential fieldapplication of the kinetic model. A MFC tool wasused to generate a profile of the internal geometry ofa production tubing string in which scale build-upwas suspected. The MFC tool is run downhole onwireline and a number of sensor arms arrangedaround the circumference of the tool measure changesin the radial dimensions of the internal profile. Inthis case, 40 sensors in 5.5 in. tubing results in ahigh density image from sensors spaced only 0.38 in.apart. These types of tools have proved to be very

    useful for monitoring scale deposition downhole. Inthis well, a MFC survey conducted 11 months previ-ously did not indicate any presence of scale sinceinhibitor squeezes had been conducted regularly tocontrol the scaling. However, in the intervening pe-riod, the operator became aware that the inhibitorconcentration had fallen far below the required level,which is presumably the main cause for the scaledeposition. The information provided by the MFCtool combined with advanced data processing andinterpretation software, not only provides a reliableimage of the actual scale profile downhole, but alsofacilitates the determination of scale volumes. In thisgas lifted well, the scale found at gas entry points isshown in Fig. 6 and similar profiles were found tohave developed in the well at each gas entry point.Generally, the greatest scale thickness appeared atthe gas entry point and gradually declined to almostzero. The scale formed in this way is most likely tobe composed of carbonate minerals whose precipita-tion is strongly influenced by pressure change. Asthe pressure decreases, dissolved CO in the water2phase will escape and raise the pH and subsequentlyincrease the saturation index. The turbulent flow at

    Fig. 6. The calcium carbonate scale profile in an oil production well detected by an MFC logging tool with 40 sensor arms and calculated .from the tubing mean internal diameter ID and minimum ID.

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 859594

    the gas entry points can increase the speed of CO2coming out of solution. This may be one reason whyhigher scaling rates occurred at gas entry points. It isalso possible that the CO content in the lift gas was2lower than the original produced gas, thus when thelift gas entered the tubing and mixed with the pro-duced gas, the pCO reduced sharply and caused the2saturation index to rise.

    To model the scale profile for this field scalingcase, the temperature and pressure profiles werecalculated from the reported downhole and wellheaddata assuming linear changes with well depth. Thereported CO content in the produced gas at the2wellhead was reported to vary from 4.3% to 7.3%which corresponds to SIs0.28 to 0.08. These satu-ration index values are much lower than the SIvalues used in our tube blocking experiments. Asdiscussed earlier, with lower SI and lower MgrCaratios, the mineralogy of the scale can be expected tobe calcite. Our kinetic model has been used tocalculate the CaCO scaling rate at downhole tem-3peratures and pressures to fit the detected scaleprofile from the MFC survey by modifying slightlythe input X . The simulated scale profile is shownCO 2in Fig. 7, which shows very good agreement with theactual logged profile. To form CaCO scale at a rate3

    of less than 8 mm in 11 months, the saturation indexrequired should be below 0.15 with X 6.0%,CO 2which is consistent with the range of the valuesmeasured at the wellhead. It is also possible that thescale formed in this well was at a much higher rateover a shorter period when the lower X lift gasCO 2was being supplied. If the X in the lift gas hadCO 2been the same as in the produced gas, the amount ofscale would have been much smaller. The opportu-nity for the operator to adjust such production pa-rameters will vary from case to case, however, theexample considered here shows that the kinetic modelcould contribute to the identification of some possi-ble parameters which the operator could control. Forexample, by an appropriate change in the lift gascomposition, significant scale problems could possi-bly be avoided or at least reduced. The cost ofcontrolling the lift gas composition may prove to beless in the long run than the costs of lost oil produc-tion and well workovers.

    More work will be required to compare the pre-diction from the kinetic model with actual scaleprofiles recorded downhole by MFC tools in otherwells to confirm the reliability of the model. How-ever, the initial results presented here are promisingand suggests that the model could also be used to

    Fig. 7. Calcium carbonate scale profile calculated using kinetic model and compared with the detected scale profile.

  • ( )Y. Zhang et al.rJournal of Petroleum Science and Engineering 29 2001 8595 95

    optimise downhole monitoring by increasing theconfidence in targeting wells vulnerable to scaling.

    6. Conclusions

    1. The most probable mineral in downhole cal-cium carbonate scales under oilfield reservoirconditions is calcite.

    2. A kinetic model derived from the growth rateof calcite can be generally used for the predic-tion of downhole CaCO scaling rates.3

    3. As far as formation brines are concerned, thecalcite growth rate can be well described byconsidering only the inhibition by Mg2q.

    4. The scale profile predicted by the kinetic modelagrees well with a measured downhole carbon-ate scale profile.

    5. The application of the kinetic model couldcontribute to the management of downhole car-bonate scaling problems.

    Acknowledgements

    The authors are grateful to Bryan Clarke, CarlosGrattoni and Sean Mulshaw of Imperial College fortechnical assistance, Garry Williams and KeithYoung of Read Well Services for their help inpreparing the paper. The laboratory work was finan-cially supported by the EU Joule III QC-Scale pro-ject.

    References

    Akin, G.W., Lagerwerff, J.V., 1965. Calcium carbonate equilibriain aqueous solutions open to the air: II. Enhanced solubility ofCaCO in the presence of Mg2q and SO2y. Geochim. Cos-3 4mochim. Acta 29, 353360.

    Dawe, R.A., Zhang, Y., 1997. Kinetics of calcium carbonatescaling-observations from glass micromodels. J. Pet. Sci. Eng.18, 179187.

    Desbrandes, R., 1985. Encyclopaedia of Well Logging. IFP Publi-cations Graham and Trotman, London, 584 pp.

    Dreybrodt, W., Eisenlohr, B., Madry, B., Ringer, S., 1997. Precip-itation kinetics of calcite in the system CaCO H OCO :3 2 2the conversion to CO by the slow process HqHCOyCO2 3 2qH O as a rate limiting step. Geochim. Cosmochim. Acta 61,238973904.

    Kharaka, Y.K., Gunter, W.D., et al., 1988. Solmineq 88: acomputer program for geochemical modelling of waterrockinteractions. U.S. Geological Survey, Water-Resources Inves-tigations Report, California, pp. 884221.

    Morse, J.W., Mackenzie, F.T., 1990. Geochemistry of Sedimen-tary Carbonates. Oxford, Elsevier, Amsterdam.

    Mucci, A., 1987. Influence of temperature on the composition ofmagnesian calcite overgrowths precipitated from seawater.Geochim. Cosmochim. Acta 51, 19771984.

    Mullin, J.W., 1993. Crystallisation. Butterworth Heinenann, Ox-ford, pp. 173175.

    Nancollas, G.H., Sawada, K., 1980. The formation of scales ofcalcium carbonate polymorphs. The influence of magnesiumion and inhibitors. SPE 8992, Oilfield Geothermal Chemistry,Stanford, California, May 2830.

    Oddo, J.E., Tomson, M.B., 1994. Why scale forms and how topredict it. SPEPF, 4753, Feb.

    Paquette, J., Vali, H., Mucci, A., 1996. TEM study of PtCreplicas of calcite overgrowths precipitated from electrolytesolutions. Geochim. Cosmochim. Acta 60, 46894699.

    Przybylinski, J.L., 1985. The role of bicarbonate ion in calcitescale formation. SPE 13547, Oilfield Geothermal Chemistry,Arizona, April, 911.

    Sohnel, O., Garside, J., 1992. PrecipitationBasic Principles andIndustrial Applications. Butterworth Heinenann, Oxford, pp.309339.

    Warren, E.A., Smalley, P.C., 1994. North Sea Formation WatersAtlas, Geological Society, Memoir No. 15.

    Zhang, Y., 1997. The kinetics of calcium carbonate precipitationand the application to oilfield scaling problems. PhD thesis,Imperial College, University of London.

    Zhang, Y., Dawe, R.A., 1998. The kinetics of carbonate scaling inhigh salinity water systems. Appl. Geochem. 13, 177184.

    Zhang, Y., Grattoni, C.A., 1998. Comments on APrecipitationkinetics of calcite in the system CaCO H OCO : the3 2 2conversion to CO by the slow process HqqHCOyCO2 3 2qH O as a rate limiting stepB by Dreybrodt et al. Geochim.2Cosmochim. Acta 62, 37893790.