The pH Values for Cooling Water Systems

By Dan Vanderpool, Laurel Functional Chemicals

Abstract
There are several important pH values for cooling water systems. These pHs are used to understand the potential for scale formation. They are also used to assess a pilot cooling towers ability to a model an actual cooling tower and to design research and laboratory evaluations. This article explains each of these pHs and the calculations used to derive them. Key Words: Ion !uilibrium "odel# $angelier %aturation Index# &y'nar %tability Index# (ractical %caling Index# &egression "ethod# )tmospheric !uilibrium# Temperature effect on pH# pHs.

Introduction
pH has a pivotal role in the chemistry of water. It influences the driving forces for scale and corrosion. *et# a measured pH value becomes meaningful only when it is put into context with the other important pH values that are characteristic of the water. These characteristic pH values are derived from basic ionic e!uilibria and from empirically established relationships. The commonly used pH relationships in Table + are important benchmar,s for cooling water. The example water shown in Table - will be used to compute these pH values.
Table 1: Characteristic pH Relationships for Cooling Water Kun': pH . +./ $og 0)l, as mg1$ 2a2345 6 7.7 2aplan: pH . +.8 $og 0)l, as mg1$ 2a2345 6 7.9 $%I . pH : pHs $%I . approximate $og02a234 %aturation5 &%I . - pHs : pH (%I . - pHs : pHe pHe . +.7/ $og 0)l, as mg1$ 2a2345 6 7.;7 $angelier: pHs . =$og >H? . $og 0Ksp1K-5 6 p>2a? 6 p0)l,5 $arson @ Auswell: pHs . $og 0Ksp1K-5 6 p>2a? 6 p0)l,5 6 >-.;0I5+1-?1>+6;.40I5+16 ;.;I? )l, at pHs . =>H? 6 Kw1>H? 6 >H?0Ksp1K->2a?5 6 - Ksp1>2a? )tmospheric !uilibrium pH . 9.B- $og 0)l, as mg1$ 2a2345 6 /./ 0+5 0-5 045 075 0;5 0/5 0<5 085 0B5 0+95 0++5

Table 2: Composition of Example Water Used for Calculation Comparisons 2alcium as 2a234 )l,alinity as 2a234 2hloride as 2l= %ulfate as %37. 2onductivity Temperature pHs E << oD pHe )tmospheric !uilibrium 0)tm !uil5 4<; mg1$ /99 mg1$ -// mg1$ +7 mg1$ +7;; Cmho << oD /.;B 8./9 pH B.-

The Typical Cooling Tower pH

The operating pH of a cooling tower unfortunately cannot be accurately forecasted from e!uilibrium e!uations. The reason is that typical cooling water has a lower pH than expected due to excess carbon dioxide arising from neutrali'ation of al,alinity and biological respiration. Dorecasting the operating pH re!uires another approach. In the mid=+B<9s# &. F. Kun' too, samples from operating cooling tower waters to determine an empirical relationship for forecasting the operating pH.+ 2aplan updated this e!uation by averaging nine other empirical relationships.- !uations + and - show the two relationships. !uations + and - are used for estimating the al,alinity re!uired achieving a target pH in cycled=up cooling water# and for calculating the acid needed to neutrali'e excess al,alinity in the ma,e=up. They are also used for comparing the operation of a cooling= tower with Gtypical towers# and in evaluating pilot cooling=towers as mimics of real towers.

pHs and Scaling Indexes

The pHs is the pH at which water has a calcium carbonate saturation of one and it is commonly used to !uantify calcium carbonate scaling tendency. $angelier defined the scaling tendency for calcium carbonate# the $%I# as the difference between the operating pH and pHs# as shown by e!uation 4. This value is approximately e!ual to the $og of 2a234 saturation as shown by e!uation 7. )n important limitation of the $%I is the non=consideration of the chemical complexes of calcium# and approximation of the carbonate1bicarbonate ratio inherent in the pH difference. The $%I begins to significantly deviate from the true 2a234 saturation above pH 8.9H and above pH +9# where the 2a3H6 complex becomes important. The $%I becomes meaningless with regard to 2a234 saturation. Table 4 shows the deviation of the $%I saturation from the true saturation.

)nother measure of scaling tendency based on pHs was developed by &y'nar.4 He recogni'ed that above pH +9 the $%I would falsely predict scaling where# in actuality# the water was corrosive# i.e.# the water would dissolve cement rather than deposit 2a234. &y'nar apparently# however# did not recogni'e that the failure of the $%I at high pH is due to the neglect of the complexes of calcium with al,alinity and hydroxide. &y'nar developed an empirical index# ,nown as the &y'nar %tability Index 0&%I5# as shown by e!uation ;. )lthough this index has no physical meaning# it is fre!uently used to !uantify calcium carbonate scaling and corrosivity to concrete. (uc,orius @ Aroo,e introduced a modification of the &y'nar %tability Index# ,nown as the (ractical %caling Index 0(%I5# as shown by e!uation /. &ather than using the measured pH as in the &y'nar %tability Index# (uc,orius @ Aroo,e used a formula to calculate a Gpractical pH based on their experience of treating 2a234 scaling# they called their pH# pHe# as shown by e!uation <. How do the %caling Indexes compareI Table 7 compares the True 2a234 %aturation# $%I# &%I# and (%I for the example water# whose pHs is /.;B.
Table 3: Comparison of the aturation Calculated from the ! " #ith the True Calcium Carbonate aturation pH of the Water: caling "ndex $%I 0. pH : pHsH pHs . /.;B5 )pprox 2a234 %aturation . +9$%I True 2a234%aturation $%& 9.B+ 8 8 '%( +.+7 -/ -4 '%& +.B+ 8// )%( -.7+ -;< +;B )%2 -./+ 79< -+9

Table *: +ariation of caling "ndexes #ith pH pH of the Water: aturation "ndex True 2a234 %aturation $og 0True 2a234 %aturation5 $%I . pH : pHsH pHs . /.;B &%I . - pHs : pH (%I . - pHs : pHeH pHe . 8./9 $%& 8 9.88 9.B+ ;./8 7.;8 '%( -4 +.4/ +.7+ ;.+8 7.;8 '%& // +.8+.B+ 7./8 7.;8 )%( +;B -.-9 -.7+ 7.+8 7.;8 )%2 -+9 -.4-./+ 4.B8 7.;8

Jotice that the (%I scaling tendency remains the same# indicating no change in scaling tendency even though the pH and True 2a234 saturation change: this results from the use of pHe rather than the actual pH.

Calculation of pHs
There are several formulas used to calculate pHs. They differ by the approximations used in the calculations. They share the common definition of al,alinity# defined by $angelier as consisting only of hydrogen# hydroxide# bicarbonate and carbonate ions. 0ionic charges are not shown for clarity5: )l, . =>H? 6 >3H?6 >H234?6 ->234?. They also share the definition of unit saturation for calcium carbonate according to the e!uation 2a234 solid 6 H6 . 2a66 6 H234=# namely# 0>2a?>H234?1>H?510Ksp1K-5 . +H where# Ksp . >2a?>234? for calcite# and K- . >H?>234?1 >H234? for the second dissociation constant of carbonic acid. $angelier derived his formula for pHs by rearranging this e!uation to give >H? . 0K-1Ksp5>2a?>H234?. )nd thus# pHs . =$og >H? . $og 0Ksp1K-5 6 p>2a? 6 p>H234?. )ll the terms on the right side are ,nown except >H234?# so $angelier used the definition of )l,alinity to calculate >H234? as follows: )l, . =>H? 6 Kw1>H? 6 >H234? 6 -K->H234?1 >H?# where# >3H? . Kw1>H?H and >234? . K->H234?1>H?. Kpon rearrangement one obtains >H234? . 0)l, 6 >H? : Kw1>H?510+ 6 -K-1>H?5H or# p>H234? . p0)l, 6 >H? : Kw1>H?5 = p0+ 6 -K-1>H?5. %ince )l, LL >H? and Kw1>H?H and + L -K-1>H?# where K- is approximately +9=+9.4 and >H? is approximately +9=<# then p>H234? . approximately p0)l,5. Thus# pHs is given by $angelier as pHs . =$og >H? . $og 0Ksp1K-5 6 p>2a? 6 p0)l,5# as shown by e!uation 8. $angelier therefore used an approximate value of >H234?H moreover# he did not correct the values of Ksp# K-# and >H? for dissolved salts# although the values of Ksp and K- were corrected for temperature. To improve the accuracy for cooling water# $arson @ Auswell added a new term to approximately correct for dissolved salts to give: pHs . $og 0Ksp1K-5 6 p>2a? 6 p0)l,5 6 >-.;0I5+1- ?1>+6 ;.40I5+1- 6 ;.;I?# as shown by e!uation B. In contrast to these approximate approaches a precise way to calculate pHs is the regression method# which rigorously calculates the bicarbonate concentration and incorporates dissolved salts corrections. It starts by combining the e!uation for unit saturation and the definition of al,alinity to give an e!uation for the un,nown >H? 0where >2a? is e!ual to the calcium hardness5: )l, . =>H? 6 Kw1>H? 6 >H?0Ksp1K->2a?5 6 -Ksp1 >2a?# as shown by e!uation +9. The solution to this e!uation using the Jewton=&alphson method is given in )ppendix +. Then# >H? is corrected by its activity coefficient to yield pHs# i.e.# pHs . =$og 0>H?fh5. ) 2omparison of these three ways of calculating pHs is shown in Table ; for the example water at << oD.
Table &: Comparison of Three ,ethods of Calculating pHs ,ethod $angelier original $arson @ Auswell &egression pHs /.-B /.7B /.;B Comment Jo ionic strengthH approximate >H234? Kses approximate ionic strengthH approximate >H234? Kses exact ionic strengthH exact >H234?

Jotice that pHs varies by up to 9.4 units between the calculations. This translates to a two= fold variation in terms of 2a234 saturation according to the relation 02a234 saturation5 . approximately +9$%I.

The pH in the Heat Exchanger

pH decreases when temperature is raised in a confined space where 23- cannot escape. $angelier believed it was important to account for this effect.7 The true scaling tendency in the heat exchanger# where scaling is usually most problematic# re!uires the use of this lower pH. The conversion of the pH of the basin water to the pH in the heat exchanger re!uires two separate calculations. Here is how $angelier approached the two calculations. Dirst# the dissolved carbon dioxide concentration is determined from the conditions at the measured pH using the $angelier definition of al,alinity: )l, . =>H? 6 Kw1>H? 6 K+>23-M?1>H? 6 -K-K+>23-M?1 >H?-# where# >3H ? . Kw1>H?H >H234? . K+ >23-M?1>H?H >234? . K->H234?1>H?H and# the e!uilibrium constants are for the temperature of pH measurement. %ince >H? is ,nown from the measured pH# this e!uation is now a function of >23-M?. >23-M? can be found by the regression e!uations given in )ppendix -. The results yield the total carbon dioxide at the measured pH as: Tco- . >23-M? 6 >H234? 6 >234?. %ince Tco- stays the same while the water is inside the heat exchanger# it is used along with the al,alinity e!uation# to calculate the redistribution of carbonate species and pH at the second temperature. The calculation uses the e!uilibrium constants for the second temperature# Kw# K+# and K-. %ee the spreadsheet on the )WT website www.awt.orgNmembersNpublicationsNanalystN-997NspringN$angelierO%at2alc+.htm for a wor,ing layout of the calculations. Table / shows the effect of temperature adPustment for the example water in Table -# with measured pH of 8.9 at << oD# and heat exchanger temperature of ++- oD.
Table -: Effect of Temperature Correction of pH on Calcium Carbonate aturation +alue Uncorrected for Temp pH pHs for ++- oD $%I . pH : pHs for ++- oD )pprox 2a234 %aturation . +9$%I 8.9 0measured at << oD5 /.77 +.;/ 4/ Corrected for Temp <.B4 0at ++- oD5 /.77 +.7B 4+

In this example# the temperature adPustment of the pH has a relatively small effect of only 9.9< $%I units# which arises from the lower pH in the heat exchangerH this results in a +; Q difference in the 2a234 saturation values.

The Atmospheric E uilibrium pH

The atmospheric e!uilibrium pH is the pH achievable when pH=control is not practiced and provided that no calcium carbonate precipitation occurs. Knowledge of the atmospheric e!uilibrium pH will define how closely a cooling system is operating to its

Gnatural pH limit and may help to obtain more cycles from a given ma,e=up water. Dor example# typical municipal potable water used as ma,eup has a higher pH than expected because it contains too little carbon dioxide. The pH can be lowered to the atmospheric e!uilibrium pH by adding a small amount of carbon dioxide# potentially allowing greater cycling before calcium carbonate scale becomes limiting. The atmospheric e!uilibrium pH is calculated from Henrys 2onstant# Kg# and the partial pressure of carbon dioxide in the atmosphere# (co-# which is typically between +90=4.4B5and +90=4./+5 atmospheres. In order to calculate the atmospheric e!uilibrium pH# $angelier used the following e!uilibria to define the species: 23- gas . 23- a!# where >23- a!? . Kg (co-H 23- a! 6 H-3 . H-234# where Kh . >H-234?1>23- a!?H 23-M . >23- a!? 6 >H-234?# so >23-M? . >23- a!?0+ 6 Kh5 . approximately >23- a!? since Kh . approximately +90=-.85# also >23-M? is used to define the first dissociation constant for carbonic acid# thus >H234=? . K+>23-M?1>H?. The $angelier definition of al,alinity then becomes on expansion: )l, . =H 6 Kw1>H? 6 K+0Kg(co-51>H? 6 -K-K+0Kg(co-51>H?-. The $angelier )l, e!uation can be solved for >H? by the regression method as in )ppendix 4. %ee the spreadsheet on the )WT website www.awt.orgNmembersNpublicationsNanalystN-997NspringN )tm !uilOpHO$angelier.htm for wor,ing spreadsheet. The calculation can be made more precise by adding the chemical species for hardness and its complexes with carbonates# hydroxide and sulfate. This re!uires adding the mass balance e!uations for hardness 0>"? . sum of hardness ions5 and Total %ulfate: )l, . = >H? 6 >3H? 6 >H234? 6 ->234? 6 >"H234? 6 ->"234? 6 >"3H?H Total Hardness . >"? 6 >"H234? 6 >"234? 6 >"3H? 6 >"%37?H and Total %37 . >%37? 6>"%37?. The solution of this expanded model can be achieved with the regression method for simultaneous e!uations as demonstrated in the spreadsheet# )tm !uilOpHO xpanded. )lternatively# a much easier way to calculate of the atmospheric e!uilibrium pH is the al,alinity1pH relation e!uation ++.; How do these calculations of atmospheric e!uilibrium pH compareI Table < shows the results for the example water.
Table $: Three ,ethods of Calculating .tmospheric E/uilibrium pH Calculated .tmospheric E/uilibrium pH: $og (co- M $angelier model xpanded model Ruic, DormulaMM =4.; B.+7 B.+B B.+/ =4./+ B.-4 B.-< B.+/ =4.4B B.9; B.+9 B.+/

M(co- is the carbon dioxide partial pressure in units of atmospheres. MM)tmospheric !uilibrium pH . 9.B- $og 0)l, as mg1$ 2a2345 6 /./# for (co- . +9=4.;H << oD.

The simple $angelier model and the Ruic, Dormula give the e!uilibrium pH within 9.9; of a pH unit of the more rigorous xpanded model. In contrast# variation in the normal atmospheric carbon dioxide partial pressure of between +9=4.4B and +9=4./+ atmospheres causes greater uncertainty of 9.+< pH units.

Conclusions
The various pH values used to characteri'e the calcium carbonate scale potential and tower operation have been described# along with their method of calculation. The pH of unit saturation# pHs# is important as it is used in several scaling indexesH the most widely used being the $%I. The $%I is important because it is approximately e!ual to the logarithm of the true calcium carbonate saturation. Fepending on the level of approximation in the calculation of the pHs# pHs can vary by several tenths of a pH unit# or up to two=fold variation in terms of 2a234 saturation values. The influence of temperature on pH is described: the pH decreases as the temperature is raised 0in a confined space where 23- cannot escape5. The temperature affect typically changes the $%I by about a tenth of a pH unit# or approximately +; Q variation in terms of 2a234 saturation values. Three ways of calculating the atmospheric e!uilibrium pH are compared. The atmospheric e!uilibrium pH is the pH attainable when acid is not fed to the cooling water and provided no 2a234 precipitation occurs. The rudimentary $angelier model and the GRuic, Dormula give values of atmospheric e!uilibrium pH within a few hundredth pH units of the more rigorous xpanded model. In contrast# the choice of atmospheric carbon dioxide partial pressure causes a greater uncertainty of almost 9.- pH units. The various pH values can be calculated using the regression formulas in this article with common spreadsheet computer programs li,e "icrosoft xcelS# $otus +-4S# etc. In setting up even the simplest ionic e!uilibrium model# one sees the paradoxical nature of dissolved carbon dioxideH namely# its concentration can affect the pH but not the al,alinity.

Appendixes
.ppendix 1% !uations used in the &egression "ethod for pHs. 0%ee $angelierO%at2alc for a wor,ing example5. The regression e!uation for )l,alinity# )l, . =>H? 6 Kw1>H? 6 >H?0Ksp1K->2a?5 6 -Ksp1 >2a?# is solved using an initial guess value for >H?. The difference between the calculated value of )l, and the true )l, is used to find an improvement to the initial guess of >H?. The improvement is found with the formulas below: )l, . 0)l, true: )l, calc5 >H? . >H guess? )l,10=>H guess? : Kw1>H guess? 6 >H guess? Ksp10K->2a?5 6 - Ksp1>2a?5 >H? better . >H? guess 6 >H?

)l, is repeatedly calculated with >H? 6 >H? so that )l, is very small. Tenerally# 7 repetitions give the correct value of >H? so that )l, U +9=-9. pHs is calculated as :$og 0>H?fh5# where fh is the activity coefficient for >H?. .ppendix 2% &egression formulas for solving for >23-M? concentration. 0%ee $angelierO%at2alc for a wor,ing example5. When the pH is given# the un,nown in the al,alinity e!uation# )l, . =>H? 6 Kw1>H? 6 K+>23-M?1>H? 6 - K-K+>23-?1>H?-# is the dissolved carbon dioxide since >H? . +9=pH1fh. These are the e!uations used in the Jewton=&alphson method: )l, . )l, true : )l, calc 23-M.>23-M? )l,10=>H? 6 Kw1>H? 6 K+>23-M?1>H? 6 -K-K+>23-M?1>H?-5 >23-M? better . >23-M? guess 6 23-M .ppendix 3% &egression e!uations to calculate the $angelier )tmospheric !uilibrium pH. 0%ee )tm !uilOpHO$angelier for a wor,ing example5. %olve the )l, e!uation for atmospheric e!uilibrium# )l, . =H 6 Kw1>H? 6 K+0Kg(co-51 >H? 6 -K-K+0Kg(co-51>H?-# by ma,ing a guess value of >H?/#< then calculate improvements with the formula: >H? . >H?0 )l,510=>H? : Kw1>H? = K+Kg(co-1>H? = 7K-K+Kg(co-1>H?-5. The correct concentration of >H? is converted into activity to give pH# i.e.# )tmospheric !uilibrium pH . =$og 0>H?fh5. Web lin0s to spreadsheets for doing the calculations described in this article: True 2a234 %aturation 0%ee xpandedO%at2alc-.htm5 pHs &egression method Fissolved 2arbon Fioxide# >23-M? Temperature 2orrection of pH 0%ee $angelierO%at2alc+.htm5 )tmospheric !uilibrium pH#$angelier "odel 0%ee )tm !uilOpHO$angelier.htm5 )tmospheric !uilibrium pH# xpanded "odel 0%ee )tm !uilOpHO xpanded.htm5

!eferences"
+. &.F. Kun'# ).D. *en# and T.2. Hess# V2ooling Water 2alculations#W 2hemical ngineering# 0)ugust +B<<5 pp /-=<+. -. Tary 2aplan# V2ooling Water 2omputer 2alculations: Fo They 2ompareIW 2orrosion 1B9# J)2 International 2orrosion Dorum# $as Xegas# JX 0)pril +BB95

4. Yohn W. &y'nar# V) Jew Index for Fetermining )mount of 2alcium 2arbonate %cale Dormed by a Water#W Y )mer Water Wor,s )ssoc 0)pril +B<<# vol 4/5 pp 7<-=78/ 7. W.D. $angelier# V ffect of Temperature on the pH of Jatural Waters#W Y )mer Water Wor,s )ssociation# vol 48 0+B7/5 pp +<B=+8; ;. Fan Xanderpool# VHidden )ssumptions in %aturation 2alculations#W The )nalyst# vol IZ# Jo. 4# -99-# p. -;. /. H. "argenau and &. ". "urphy# V"athematics of (hysics and 2hemistry#W 0Xan Jostrand 2o.# J* +B745 p. 7</ <. The )tmospheric !uilibrium pH e!uation is a !uadratic with respect to the variable >H?H there are mathematically two correct solutions. To avoid obtaining the inappropriate solution# the initial guess of >H? must be higher than the correct value# while not being too high. The author uses the formula pH . 9.Blog0)l, as mg1$ 2a2345 6 <.-. >H? is approximately e!ual to +9=pH. If a low initial guess is used# the convergence finds a physically meaningless solution 0e!ual to the al,alinity.5 If too high a guess is used# the first iteration over corrects and Pumps to the convergence of the physically meaningless solution. .bout the .uthor: Dr. Dan Vanderpool is president of Laurel Functional Chemicals, 2925 32nd Street, orthport, !L 35"#$, %hone&'2(5)*339*5#+,.