Journal of Chemical Technology and Biotechnology

J Chem Technol Biotechnol 79:13311341 (online: 2004) DOI: 10.1002/jctb.1143

Review Titration methodologies for monitoring of anaerobic digestion in developing countriesa review
O Lahav1 and BE Morgan2
1 Faculty 2 Department

of Civil and Environmental Engineering, TechnionIsrael Institute of Technology, Haifa, 32000, Israel of Civil Engineering, University of Cape Town, Rondebosch 7700, South Africa

Abstract: An increase in volatile fatty acids (VFA) concentration (or the proportional decrease in carbonate alkalinity concentration) is the rst practical measurable indication that an anaerobic treatment system is in a state of stress. If the system is not rectied at this early stage, failure is likely. Current methods for VFA measurement include distillation, colorimetry, gas chromatography and various titration techniques. In terms of simplicity, speed and cost-effectiveness it is generally accepted that titration methods are superior for the purpose of on-site routine monitoring and control, particularly in developing countries. This paper reviews the methods published in the last four decades concerning on-site titration measurement of VFA and carbonate alkalinity concentrations. The review encompasses the following: aquatic chemistry related to the theory on which most of the methods are based, and a detailed description of each of the principal methods published followed by critical and comparative evaluation. 2004 Society of Chemical Industry

Keywords: volatile fatty acids; titration methods; anaerobic digester monitoring; developing countries

INTRODUCTION Start-up and successful operation of anaerobic treatment facilities is a difcult and delicate process, requiring reasonably accurate and rapid monitoring techniques. The control strategy is based on maintaining a low concentration of volatile fatty acids (VFA) and a pH in the range 6.6 < pH < 7.4. Normally in anaerobic reactors the carbonate system forms the main weak-acid system responsible for maintaining the pH around neutral, while the VFAs (mainly acetic, propionic, and butyric acids) are the major cause for a decline, in pH. Under stable operating conditions, the H2 and acetic acid formed by acidogenic and acetogenic bacterial activity are utilized immediately by the methanogens and converted to methane. Consequently, the VFA concentration in properly running anaerobic digesters is typically fairly stable and low (typically 0.52.0 mmol dm3 ),1 carbonate alkalinity is not consumed in excess and the pH is stable. In contrast, under overload conditions or in the presence of toxins or inhibitory substances, the activity of the sensitive methanogenic and acetogenic populations is reduced, causing an accumulation of VFA which

in turn increases the total acidity in the digester, thus reducing the pH (the term total acidity is used here to dene the total proton-donating capacity of a solution, including the contribution of all weak-acid subsystems present). The onset of reactor failure can have a spiraling effect on the methanogenic population,2 where the buffering capacity cannot keep up with the increasing production of VFAs, causing further, and ultimately, complete, failure. The extent of the pH drop depends primarily on the H2 CO3 alkalinity concentration. The term H2 CO3 alkalinity is used here to dene the total proton-accepting capacity of the carbonate weakacid subsystem combined with the proton-accepting capacity of the water system3 (ie H2 CO3 alkalinity = 2[CO3 2 ] + [HCO3 ] + [OH ] [H+ ]). In terms of routine monitoring, pH measurement cannot form the sole indication of imminent failure, because in medium or well-buffered waters high VFA concentration would have to form in order to cause a detectable drop in pH, by which time failure would already occur. Consequently, direct measurement of either (or both) VFA or H2 CO3 alkalinity concentration is necessary. Measurement of H2 CO3 alkalinity in a mixture of weak-acid subsystems cannot be carried out via

Correspondence to: O Lahav, Faculty of Civil and Environmental Engineering, TechnionIsrael Institute of Technology, Haifa, 32000, Israel E-mail: agori@techunix.technion.ac.il (Received 23 June 2004; revised version received 9 July 2004; accepted 19 July 2004) Published online 14 September 2004

2004 Society of Chemical Industry. J Chem Technol Biotechnol 02682575/2004/$30.00

1331

O Lahav, BE Morgan

the standard titration procedure to the H2 CO3 equivalence point (pH 4.5) because (i) the point is not dened sharply and (ii) titration to 4.5 does not account for all the proton-accepting capacity of the VFA system (ie the non-protonated forms of acetate, butyrate and propionate). Characterization of the carbonate subsystem can be carried out using an inorganic carbon analyzer; however, this instrument, apart from not being generally available on-site, is prone to gross inaccuracy due to CO2 loss occurring between sampling and measurement. Therefore, an increase in VFA concentration is the rst practical measurable indication that an anaerobic treatment system is in a state of stress. If the system is not rectied at this early stage, failure is likely. The demand for reliable VFA measurement has increased in recent years due to the introduction and widespread use of high-rate anaerobic treatment processes, where more rigorous control is needed.4 In addition to conventional anaerobic digesters, other treatment systems such as biological sulfate removal reactors and hydrolysis reactors (prefermenters) depend on VFA measurement as a principal means of monitoring reactor performance. Furthermore, anaerobic treatment of municipal sewage has gained popularity recently as evidenced by the increasing introduction of full-scale UASB reactors,5 particularly in tropical areasthe majority of which are in developing countries where sophisticated technology tends to be unsuccessful. Currently, VFA can be measured using straight distillation, steam distillation, a colorimetric technique, gas chromatography, and titration techniques. Some of these methods are time consuming, others require expensive equipment and a dedicated operator, and often, in particular in developing countries, the equipment is not available on-site. Combining such factors as simplicity, speed and cost-effectiveness it is generally accepted that titrative methods are superior for the purpose of routine monitoring and control.6 While difcult to verify, it would appear that in developing countries the vast majority of anaerobic digesters are monitored by various titration techniques. This also holds true for many treatment plants in the developed world. During the last four decades a considerable number of quantitative and semi-quantitative titration methods have been proposed for the measurement of either VFA or H2 CO3 concentrations or both. These titrative methods can be roughly divided into three categories of approaches: 1 Approximation of VFA concentration alone or approximation of both VFA and H2 CO3 alkalinity, both by titration techniques.7 11 2 Measurement of H2 CO3 alkalinity only by direct titration, with or without an external measurement of VFA concentration using a different analytical approach.12 15 3 Accurate measurement of both VFA and H2 CO3 alkalinity with differing levels of complexity and
1332

accuracy using a titration technique followed by a mathematical algorithm.16 19 In addition to these, automated, in-line methods based on one of the above are also found in the literature.20,21 The multiplicity of methods available in the literature and the large difference in approach and in the results obtained from each method emphasizes the need for a comprehensive review of the subject, extending the review published by Moosbrugger et al in 1993.6 The current review encompasses the following: aquatic chemistry theory on which most of the methods are based, and a detailed description of each of the principal methods published followed by critical and comparative evaluation. It is hoped that the review will provide assistance to researchers, engineers, and laboratory technicians in their quest for the most appropriate method for the control of anaerobic processes. A list of the main methods covered, including characterization, ease of execution, and suitability for use as the monitoring technique is given in Table 1.

THEORY OF WEAK-ACID SYSTEMS AND CORRESPONDING BUFFERING INTENSITY IN ANAEROBIC REACTORS All the titrative procedures proposed for VFA and H2 CO3 determination stem from classical aqueous solution weak-acid equilibrium theory. Thus, in order to evaluate the various methods using common grounds, a brief review of fundamental aquatic chemistry principles is given. Acids or bases that dissociate only partially in solution are dened as weak. The principal weak-acid subsystems commonly found in anaerobic reactors are the carbonate, ammonium, phosphate, VFAs (namely acetic, propionic and butyric acids), and sulde subsystems. The various species of these subsystems can be represented as a function of the total species concentration of a particular weak-acid subsystem and its apparent equilibrium constant adjusted for temperature and DebyeHuckel effects. An example of such representation is given below for the carbonate and VFA subsystems. The equilibrium and mass balance equations for the carbonate subsystem are: (H+ ) [HCO3 ]/[H2 CO3 ] = K (H+ ) [CO3 2 ]/[HCO3 ] = K
C1

(1) (2)
2

C2

CT = [H2 CO3 ] + [HCO3 ] + [CO3

(3)

Where () denotes activity, [] denotes molarity, K equals apparent equilibrium constant after adjustment for DebyeHuckel effects, and CT = total inorganic carbon concentration (mol dm3 ). For the VFA subsystem (VFAs are commonly considered to constitute a single weak-acid system
J Chem Technol Biotechnol 79:13311341 (online: 2004)

Table 1. Summary and comparison of principal methods (+ = low, +++++ = very high)

Authors Removal of CO2(g) at pH 3.5 and back titration to pH 7 Titration between pH 5 and 4 Basic laboratory equipment ++++ + Basic laboratory equipment ++ +

Parameter measured Characterization Equipment required

Ease of execution Accuracy Suitability Gives very rough approximation; appropriate only for well-buffered reactors. Gives the change in VFA concentration rather than a value. Can be used in conjunction with a more accurate method. Can give good indication of trends, especially in the cases when the ratio H2 CO3 alkalinity to VFA concentrations is lower than 10 to 1. +++ +++

DiLallo and Albertson (1961)7 Pauss et al (1990)10

VFA

McGhee (1968)8

VFA

J Chem Technol Biotechnol 79:13311341 (online: 2004)

Titration to pH 5.75 as indication of H2 CO3 alkalinity and between pH 5.75 and 4.3 for VFA determination +++++ +++ Basic laboratory equipment ++++ ++ Direct linkage of reactor pH and VFA pH meter concentration Titration to pH <4 and measurement of Gas collection apparatus volume or pressure of CO2(g) emitted Suitable only for prefermenters operated at low pH and high VFA concentrations. Suitable particularly for low-buffered waters. The major disadvantage is the choice of H2 CO3 alkalinity as the sole control parameter. +++ ++++ Accurate when applied to high strength anaerobic digesters with low concentrations of other weak acid systems. Otherwise, the application depends on the similarity of operational conditions to those under which the empiric relations were derived. ++ +++++ Can be applied generally provided that CT > 2 AT . Basic equipment + computer + phosphate, ammonium and sulde analysis Four-point titration. Algorithm based on Basic laboratory equipment empirical relations Five-point titration, includes other weak-acid systems, EC, and temperature in algorithm Eight-point titration, includes other weak-acid systems, EC, and temperature in algorithm Basic equipment + computer + phosphate, ammonium and sulde analysis

Ripley et al (1986)9 Jenkins et al (1983)13

VFA and H2 CO3 alkalinity

Munch and Greeneld (1998)11

VFA

Rozzi and Brunetti (1981)12 Di Pinto et al (1990)14

H2 CO3 alkalinity

Kapp (1984)16

VFA

Moosbrugger et al (1993)17,18

VFA and H2 CO3 alkalinity

Monitoring of anaerobic digestiona review

Lahav et al (2002)19

VFA and H2 CO3 alkalinity

1333

O Lahav, BE Morgan

with equilibrium constant K a because of the similarity of their pK values):6 (H ) [A ]/[HA] = K AT = [HA] + [A ]
+ a

where: Ma , Mb = concentration of strong acid or strong base, respectively, added to 1 dm3 of solution (mol dm3 solution). = buffer intensity index (mol dm3 solution/ pH). The equation for the calculation of the buffer intensity index for monoprotic weak-acid subsystems and for diprotic weak-acid subsystems with dissociation constants differing by four pH units or more is given by the following term:22 = 2.303 [AT Ka (H+ )]/[Ka + (H+ )]2 (12)

(4) (5)

where: AT = total VFA species concentration (mol dm3 ), HA represents the acidic, protonated species and A the ionized form of each acid. Representing the individual species of the carbonate and VFA subsystems as a function of CT , AT , and equilibrium constants: [H2 CO3 ] = CT /{1 + K +K
C1 + C1 /(H ) + 2 C2 /((H )) } C2 /(H + +

(6)
+ C1 }

[HCO3 ] = CT /{1 + K [CO3 2 ] = CT K

+ 2

) + (H )/K
C2

(7)

For the water subsystem the buffer intensity index is given by: = 2.303{(H+ ) + KW /(H+ )} (13)

C2 /{(H

)+K

+ ((H )) /K

C1 }

(8) (9) (10) The overall buffer intensity index of a solution composed of a number of weak-acid subsystems is the sum of the buffer intensities of all the weak-acid subsystems including the water subsystem. pHlog species and pHbuffer intensity index diagrams Using equations describing individual weak-acid species concentration (such as eqns (6)(10)) and buffer intensities (eqns (12) and (13)), pHlog species and pHbuffer intensity index diagrams can be plotted. In Fig 1 an example of such plot is given for a typical species concentration distribution encountered in anaerobic digesters. Using Fig 1 as it applies to titration methods, a number of points can be made:

[A ] = AT K a /{(H+ ) + Ka } [HA] = AT /{1 + Ka /(H+ )}

Similar equations can be developed for the phosphate, sulde, and ammonium proton-accepting species. Buffer intensity The buffering contribution of each subsystem can be calculated through a parameter called buffer intensity, dened as the slope of a titration curve plotted from the cumulative mass of strong acid (or base) added to a sample vs the change in pH: = dMa /dpH = dMb /dpH (11)

0 0.1 0.001

8 9 HCO3

10

11

12

13

14 0.03

1E-05 H2PO4 1E-07 Log Species 1E-09 1E-11 1E-13 1E-15 1E-17 1E-19 1E-21 1E-23 1E-25

HPO4
NH3 CO3 PO4
2

Ac

H2CO3 HAc H3PO4 Actual buffer intensity VFA buffer intensity

0.025 0.02 0.015 0.01

Carbonate buffer intensity

0.005 0

pH
Figure 1. pHlog species and buffer intensity index diagrams in a typical anaerobic digestion sample (CT = 1000 mg dm3 as CaCO3 , VFA = 100 mg dm3 as HAc, total phosphate concentration (PT ) = 50 mg dm3 as P, total sulde concentration (ST ) = 20 mg dm3 as S, and total aqueous ammonium concentration (NT ) = 50 mg dm3 as N, temperature = 22 C, TDS = 3000 mg dm3 ). Actual buffer intensity is the sum of buffer intensity curves of all subsystems.

1334

J Chem Technol Biotechnol 79:13311341 (online: 2004)

Buffer intensity (mol/(l pH))

Monitoring of anaerobic digestiona review

The magnitude of the (cumulative) curve at a particular pH species the buffering capacity of the solution at that pH or in other words, its ability to minimize a change in pH when strong acid or strong base is added. For a titration between any two pH points, the area under the buffer intensity curve equals the mass of strong acid (or base) that has to be added to bring about the pH change. For a particular weak-acid subsystem, the buffering intensity is maximal at the pK (pK = a pH value where two species of a weak-acid subsystem are equal in concentration). On either side of the pK the buffer intensity decreases sharply, becoming only 1% of its maximal value within two pH units. The shape of the actual buffer intensity curve (ie the cumulative curve) depends on both the concentrations and the respective apparent dissociation constants of the weak-acid subsystems present in the water. At the normal pH range maintained in anaerobic reactors, the carbonate, sulde and phosphate subsystems can have a signicant contribution to the cumulative curve as their respective pK values are close to the range 6.7 > pH <7.4. Typically, the carbonate subsystem is present at high concentrations and thus its component has the most signicant effect on the cumulative buffer intensity curve. The sulde and phosphate concentrations are usually much lower compared with the carbonate subsystem, but nevertheless neglecting their effect may lead to an erroneous measurement. The pK of the ammonium subsystem (pKN 9.4) is such that its contribution to the buffer intensity curve at the relevant pH range is very small up to relatively high concentrations (around 1000 mg N dm3 ). The VFA subsystem, which is invariably represented by the acetic acid subsystem, has a pK of 4.75 and under normal operating conditions is present at relatively low concentrations (typically AT < 2 mmol dm3 ). Accordingly it has only a small effect on the cumulative buffer intensity curve at 6.7 > pH <7.4 but a much larger effect is apparent at 4.25 < pH <5.25. On the other hand, given that the carbonate subsystem is present at high concentrations relative to the VFA subsystem, its contribution to the cumulative curve at this pH range (4.25 < pH <5.25) may also be relatively high. The overlap between the buffering intensity curves of the carbonate and VFA subsystems precludes the use of the standard measurement of carbonate alkalinity (ie titration to an end point near pH 4.5) as a meaningful means of control for anaerobic reactors. In addition, acid-titration to two pH points around the pK of the VFA system will produce the sum of the proton-accepting capacity of both the VFA and carbonate subsystems between the two points, reducing the ability of this approach to serve as means of direct measurement of the VFA concentration, particularly in cases where the ratio CT to AT is large (ie where the area under the buffer intensity curves of the carbonate and VFA
J Chem Technol Biotechnol 79:13311341 (online: 2004)

subsystems between two pH values are of the same order of magnitude). From a technical standpoint, obtaining accurate titrative results depends on the stability of the pH readings. The stability of a reading at a given pH, using an appropriate pH probe, depends on the magnitude of the cumulative buffer intensity curve at that pH and to a lesser degree on the mixing conditions governing the exchange of volatile species such as CO2 and H2 S with the atmosphere (such volatilization might introduce errors in lowpH titration point measurements). The magnitude of the cumulative buffer capacity at any pH depends on the total species concentration and the proximity to a relevant pK value. As a rule, the buffer intensity is highest close to the pK values and lowest in between them. In a mixture of weak-acid systems of unknown concentrations it is impossible to predict a priori the exact shape of the cumulative curve, but it is safe to assume that titration to pH values close to the known pK values would increase the readings stability and result in more accurate observations. The effect of the ionic strength and temperature of the tested solution on the pK values of the weakacid systems, and thus on the shape of the buffer intensity curve is also noteworthy. Changes in the salt composition and concentration (and to a lesser degree in temperature) may shift the pK values by up to 0.5 pH units, changing the cumulative curve signicantly. Therefore, neglecting these parameters can lead to a large error in interpretation of experimental results. There are particular waters (eg agro-industrial waters, distilleries, paper mills, landll drainage) in which not all proton-accepting species can be readily identied (eg lignin fractions). If these species are present in signicant concentrations, an assessment is essential for correct interpretation of the titration data.

PUBLISHED METHODS Acknowledgment that monitoring of the carbonate alkalinity or VFA concentration (or preferably both) is crucial for control of anaerobic reactors has led, since the early 1960s, to the publication of a variety of practical procedures based upon titration techniques. In all cases the incentive was to develop a cheap, simple and rapid method to measure at least one of the two parameters, H2 CO3 alkalinity or VFA. In the following, these methods are grouped according to the three categories outlined in the introduction. Within the groups, the methods are by and large presented in chronological order of their appearance in the literature. Approximate measurement of VFA alone or both VFA and H2 CO3 alkalinity The rst to propose a titration method for VFA measurement were DiLallo and Albertson.7 Their goal
1335

O Lahav, BE Morgan

was to develop a reasonably accurate method for VFA determination using only equipment normally found in a treatment plant laboratory in order to decrease the time to obtain results and to gain better reactor control. The authors developed a technique aimed at detecting the change in VFA concentrations rather than measuring accurately their absolute value. Their fundamental idea was to circumvent the overlap between the buffer intensity curves of the carbonate and VFA systems by removing the inorganic carbon concentration as CO2 , thus isolating the VFA system so that it can be measured directly through titration. Total alkalinity in their method was dened as the proton-accepting capacity of the solution titrated down to pH 4, at which pH it was assumed that all carbonate species are in the form of CO2 . After recording the amount of standard acid added to pH 4, the pH is lowered to between 3.3 and 3.5 and the sample is boiled lightly for 3 min to completely remove CO2 . Thereafter, the amount of standard base required to elevate the pH from 4 to 7 is recorded. This value is considered in the method to consist 80% of the VFA alkalinity (irrespective of the VFA concentration). Because the acid titration at this pH range also includes a minor contribution of what is termed base alkalinity (referring to proton-accepting capacity of subsystems such as the phosphate and sulde subsystems), the VFA concentration is attained by multiplying the titration results by a factor of 1.5 when the method yields a VFA concentration above 180 mg dm3 . Below this value a factor of 1.0 is used (ie no correction factor is applied). The carbonate alkalinity is calculated by subtracting the VFA alkalinity from the total alkalinity. The method proposed by DiLallo and Albertson, although having the credit of being the rst titrative method, suffers from a number of shortcomings. The method requires the compulsory addition of both standard acid and base and the boiling of the sample, a step that tends to be cumbersome. More importantly, several steps in the procedure are prone to gross inaccuracy: First, boiling of the sample to remove CO2 can result in a loss of a fraction of the VFA due to stripping that will depend on the VFA concentration and composition, and on the type of boiling. The authors suggest 3min gentle boiling, but such procedure can hardly be standardized. Also, an unknown volume of water is vaporized in the boiling procedure. Second, the back-titration between pH 4 and pH 7 is assumed in the method to incorporate 80% of the VFA alkalinity. This value is a not a bad approximation since its magnitude is relatively insensitive to such factors as the VFA concentration, the composition of the acids, and the ionic strength and temperature of the sample which affect the dissociation equilibrium constant. However, VFA concentration is calculated in the method directly from the titration results for values lower than 180 mg dm3 VFA as CH3COOH (HAc ) and multiplied by a factor of 1.5 above 180 mg dm3 . This approach almost invariably results in a large error:
1336

for example, for a VFA concentration of 200 mg dm3 as HAc with total phosphate of 150 mg dm3 as P, the approximation results in 23% error in the VFA concentration (0.016 mg dm3 instead of 0.013). Despite its faults, and considering that the procedure can be modied to include externally measured weak-acid subsystems (phosphate, sulde, ammonia, etc), the method may be used to detect a large upsurge in VFA concentrations, as intended by the authors. Indeed, it is the most popular method in Israel and it is also practiced in many other places. However, it should be noted that the method is practical only where relatively large changes in VFA concentration are not detrimental to the process, as might sometimes be the case in well-buffered reactors. Pauss et al 10 proposed a similar back-titration method in which bicarbonate alkalinity is the monitored parameter rather than VFA. In their method, the solution is rst titrated from the initial pH to pH 4.54.0 and CO2(g) is removed by vacuum boiling. Subsequently, the solution is back-titrated to the initial pH and the bicarbonate concentration is calculated by the difference between the acid and base titration. McGhee8 presented a different approach to approximating VFA concentrations. He suggested determining the slope of the titration curve between pH 5 and pH 4 as a simple and rapid means of estimating VFA concentrations. The sample is titrated rapidly to pH 5.5, a short delay is given to allow CO2 to reach equilibrium with the atmosphere, and thereafter the titration is continued drop-wise to a pH slightly above 5. From this point one records the amount of additional acid required and pH attained. The values attained are then plotted, and the reciprocal of the slope is calculated. The method is based on the idea that for a given reactor, with a high and therefore fairly constant carbonate alkalinity concentration, the amount of standard acid added to effect a pH change between 5 and 4 reects the change in VFA concentration. The author intended the approach to supplement but not replace the more accurate methods for VFA determination such as chromatographic techniques. As such it has value, however, as a more general tool the approach has several faults. First, it cannot serve as a tool for determining VFA concentration but simply to detect large changes in concentration. Second, even within this scope its application is limited because the titration between pH 5 and 4 accounts for only about 50% of the VFA alkalinity, and the effect of the carbonate system on the proton-accepting capacity in this pH range is not negligible, especially when the CT to AT ratio is high (above ten, as in most anaerobic reactors). Furthermore, in anaerobic reactors any increase in VFA alkalinity is accompanied by a similar decrease in carbonate alkalinity (and/or in the alkalinity of other proton-accepting species such as the phosphate subsystem). As a result, an increase in VFA concentration will not be represented
J Chem Technol Biotechnol 79:13311341 (online: 2004)

Monitoring of anaerobic digestiona review

proportionally in the amount of acid added between pH 5 and 4. Ripley et al 9 working on poultry manure treatment, suggested another way to monitor the biological stability of a high-strength anaerobic digester. They recommended titration to two end points and the use of the ratio PA (volume of strong acid required to titrate the solution down to pH 5.75) to IA (volume of strong acid required to titrate the solution from pH 5.75 to pH 4.3) as a means of rapid detection of possible stress (they suggested a value exceeding 0.3 is indicative of stress). PA relates roughly to bicarbonate alkalinity and the titration from 5.75 to 4.3 (IA) approximates VFA alkalinity. The concept of titration to pH 5.75 as means of estimating bicarbonate alkalinity was rst introduced by Jenkins et al 13 They claimed that 80% of the bicarbonate is converted to CO2 at pH 5.75 while at the same pH only around 20% of the VFA will have contributed to the alkalinity. Therefore, for the high alkalinity concentrations encountered in high strength reactors, the effect of VFAs on the bicarbonate alkalinity (PA) value would be minor, even if the VFA reached high concentrations. Ripley et al 9 added that titration between 5.75 and 4.3 gives roughly the VFA concentration, and so they assert that the ratio between the two values is analogous to the ratio of VFA to carbonate alkalinity. The clear advantages of this method are simplicity, cost effectiveness, and rapidity. As the ratio PA to IA is dimensionless it does not require titrant standardization nor sample volume measurement. On the other hand, because only about 65% of the VFA is represented in the titration between 5.75 and 4.25 and in reality less than 70% of the carbonate alkalinity is titrated at pH 5.75 (value calculated for samples with TDS higher than 3000 mg dm3 ) the method lacks accuracy and is somewhat insensitive to increase in VFA concentrations, especially in the case of high CT to VFA ratios. A concept for approximation of VFA concentrations in prefermenters (hydrolysis reactors), based on pH reading only, was developed by Munch and Greeneld.11 In prefermenters, unlike typical anaerobic reactors, the VFA are the desired products of the anaerobic activity and their build-up is a sign of a healthy process. The typical operational pH range is 56. The authors developed a mathematical function relating the VFA concentration to pH using a set of simplied assumptions. The value of this method is that without any additional work a simple pH reading from the working reactor can give a good indication of the VFA concentration. The method is only suitable for reactors working with high VFA concentrations and low pH and providing that the simplifying assumptions on which the model is based are met. The major disadvantage of this approach appears to be the possible lack of reliable pH measurements emanating either from unaccounted CO2 supersaturation or from other pH probe inaccuracies.
J Chem Technol Biotechnol 79:13311341 (online: 2004)

It is therefore recommended to use the method in conjunction with (at least) a weekly analytical VFA measurement. Measurement of H2 CO3 alkalinity only by direct titration, with or without an external measurement of VFA concentration To overcome the shortcomings of using the concept of total alkalinity (ie titration to pH 4.5) for anaerobic reactor monitoring, a number of modied alkalinity procedures have been proposed. Hattingth et al 23 advocated the use of an alkalinity proton accepting capacity (PAC) value titrated down to pH 6 and expressed as HCO3 alkalinity as a realistic measure of the available buffering capacity of an anaerobic digester. Rozzi and Brunetti12 proposed a method where a digester sample is saturated with CO2 (to yield PCO2 = 1 bar) and subsequently the pH is reduced to 3.7 by the addition of standard acid. Such addition of CO2 does not alter the bicarbonate alkalinity. The volume of CO2(g) released from solution at pH 3.7 is then measured by a gas meter. As the loss of CO2(g) during titration is negligible and assuming that the original CO3 2 concentration at the operational pH is very low, this measured volume of CO2(g) is proportional to the mass of HCO3 converted. Bicarbonate alkalinity (BA) is thus determined by: BA(in mg dm3 as CaCO3 ) = (VCO2 Vacid )/Vsample 50 000/22.4 C) (14) where: VCO2 volume of gaseous CO2 released at pH 3.7 (dm3 ) Vacid volume of standard acid from initial pH to pH 3.7 (dm3 ) C correction factor, which adjusts for temperature (T) and pressure (P) effects in the vessel: C= T0 P PV (T0 = 273.16 and T P0 P0 = 1.013 bar)

(15)

Results from the proposed method were compared with results derived from total alkalinity minus the VFA concentration (calculated via chromatographic techniques) and found to be very accurate and reproducible in the presence and absence of VFA (3%). It was therefore concluded that the initial CO2 bubbling does not cause volatile acid stripping. According to the authors, this method, in addition to the parallel method in which the CO2(g) pressure change (rather than the volume released) is measured following titration to pH <4, is suitable for automated control of anaerobic reactors.24 It appears that from a theoretical standpoint both methods are robust and sound. Because the calculation
1337

O Lahav, BE Morgan

in the procedure is independent of equilibrium constants and pH measurements, and no problematic simplifying assumptions are made, the method should give accurate results. However, on the negative side, the procedure is relatively complex and specic equipment (pressure vessel, gas ow meter or pressure gauge, CO2 bottle) is required in addition to the standard laboratory equipment. On a more general note, concern for the usage of these methods (and similar procedures) may stem from the very choice of bicarbonate alkalinity as the principal control parameter in anaerobic reactors, especially where the total inorganic carbon concentration is much higher than VFA concentration. In such cases a small change in the carbonate alkalinity could mean a large increase (percentage wise) in VFA concentration. Such small change may fall within the accuracy of the method and go unnoticed, detracting from the effectiveness of the concept. Accurate measurement of both VFA and H2 CO3 alkalinity with differing levels of complexity and accuracy This group of methods is composed of more complex titrative methods typically requiring several titration points and computerized data interpretation. Colin15 suggested an automated method based on acid and base titration to three endpoints. In this method, after measurement of the initial pH, the sample is divided in two. One part is acid-titrated to pH 2 and the other is base-titrated to pH >10. The sample that was acid-titrated to pH 2 was further base-titrated up to pH >10. This procedure yields three pairs of pH and Vx (acid dosage). Using these three pairs and the initial pH, CT , AT and NT (total ammonium concentration) are obtained using equilibrium equations and a computer program. The author reports good accuracy as compared with VFA and CT values determined by analytical methods. However, proper evaluation of the advantages and disadvantages of the method was not possible because the method lacks a sufciently detailed description. A different empirical-theoretical approach extending the method suggested by McGhee was developed by Kapp.16,25 McGhee originally proposed that titration from pH 5 to pH 4 can be considered proportional to the VFA concentration. Kapp accepted this approach but considered the carbonate subsystem to also have PAC in the pH range between pH 5 and 4, neglecting the sulde, phosphate, and ammonium subsystems. Accordingly, the following equality holds: VA5 4, VFA = VA5 4, measured VA5 4, H2 CO3 where: VA5 4, VFA = volume of acid required to titrate from pH 5 to pH 4 due to the VFA PAC VA5 4, measured = volume of acid required to titrate from pH 5 to 4
1338

VA5 4, H2 CO3 = volume of acid required to titrate from pH 5 to pH 4 due to the carbonate PAC To develop an explicit and simple mathematical expression linking VFA concentration to VA5 4, measured Kapp conducted titration experiments for a broad VFA and CT concentration range and derived the following empiric equation linking volume of titrant to VFA and H2 CO3 alkalinity concentrations: VA5 4, VFA = 0.1/N (0.0283 + 0.09418 VFA/60) Vs /20 VA5 4, H2 CO3 = 0.005 (0.044875 + 0.00469 [Alkmeasured ]) Vs /N where: N = titrant concentration (eq dm3 ) Vs = volume of sample (cm3 ) [Alkmeasured ] = total PAC as titrated to pH = 4.3 (eq dm3 ) Insertion of eqns (17) and (18) into eqn (16) and rearranging yields Kapps rst approximation: [VFA] = 127 416 N VA5 4, measured /Vs 2.99 [Alkmeasured ] 10.6 (19) (18) (17)

where: [VFA] = VFA concentration (mg dm3 as HAc) Using a further two assumptions, eqn (19) transforms slightly yielding: [VFA] = 131 340 N VA5 4, measured /Vs 3.08 [Alkmeasured ] 10.9 (20)

(16)

Kapps approach involves three pH titration set points (pH 5.0, 4.3 and 4.0), in addition to the initial pH. Working on samples of digested sludge, Kapp reported an accuracy of 10% for VFA concentrations above 20 mg dm3 as HAc. Baucher reported a similar accuracy for samples of raw wastewater, primary sludge, and high- and low-load activated sludge.25 The major drawback of Kapps approach is that it is based on empirical mathematical relationships that were developed under unique conditions (ionic strength, temperature, absence of other weak-acid systems) that are not necessarily generally applicable. Its advantage stems from its relative simplicitythe VFA concentration is calculated using a single equation, the apparatus needed is simple and laboratory execution is easy and quick. The method is also suitable for automated execution. The most general method to date was presented by Moosbrugger et al 17,18 The authors developed a model based almost solely on aquatic chemistry considerations, with very few simplifying assumptions.
J Chem Technol Biotechnol 79:13311341 (online: 2004)

Monitoring of anaerobic digestiona review

Because of its importance the method is described in detail. The authors developed a ve-point method involving equating a mass balance for alkalinity in terms of volume of titrant added (eqn (21)) to a mass balance of alkalinity in terms of species concentration (eqn (22)). Mtotal alk(x) = Ve Ca Vx Ca (21)

where: Mtotal alk(x) = total mass of alkalinity after the addition of Vx cm3 of standard strong acid (mol), Ve = the unknown volume of standard strong acid to be added to the alkalimetric end point (dm3 ), Vx = the volume of standard strong acid added to a point x with pH equal to pHx (dm3 ), and Ca = concentration of standard strong acid titrant (mol dm3 ). Mtotal alk(x) = {[HCO3 ]x + 2[CO3 2 ]x + [A ]x + [HS ]x + 2[S2 ]x + [NH3 ]x + 3[PO4 3 ]x + 2[HPO4
2

where: PT , ST , and NT represent the total phosphate, sulde and ammonium concentrations, fm = monovalent activity coefcient, and Fn1 to Fn5 are functions of pHx and equilibrium constants for the carbonate, acetate, phosphate, sulde and ammonium subsystems respectively. Equating eqns (21) and (26) gives the desired equation linking the mass of alkalinity based on acid added to the mass of alkalinity based on species concentrations: (Ve Vx ) Ca = {CT Vs /(Vs + Vx ) Fn1 (pH)x + AT Vs /(Vs + Vx ) Fn2 (pH)x + PT Vs /(Vs + Vx ) Fn3 (pH)x + ST Vs /(Vs + Vx ) Fn4 (pH)x + NT Vs /(Vs + Vx ) Fn5 (pH)x + 10(14pHx ) /fm 10pHx /fm } (Vs + Vx ) (27) At each point in the titration (ie for each Vx and corresponding pHx ), eqn (27) includes three unknowns: Ve , AT and CT . Thus, to solve for Ve , AT and CT only three data pairs (ie three values for corresponding Vx and pHx pairs) need to be known. This was found to lead to poor prediction. Moosbrugger et al found that the best rst estimate for AT and CT and can be obtained from four titration data points (ie two pairs of data points), each pair symmetrical about pKC1 and pKa (they suggested approximately half a pH unit to either side of the respective pK values).17 When inserted into eqn (27), the data from the four titration points give four equations. The pair of observations around pKa (ie the third and the fourth points) is in a region where the buffer capacity of the VFA system dominates that of the carbonate system and vice versa for the rst and second titration points. Consequently, subtracting the equation formed from the fourth data point from that derived from the third, gives an equation in terms of CT and AT in which the VFA alkalinity term, ie the species [A ]x , dominates. Similarly, subtracting the equation formed from the second data point from that derived from the rst, gives an equation in terms of CT and AT in which the H2 CO3 alkalinity term, namely [HCO3 ]x , signicantly dominates. This technique enables a relative separation between the two subsystems in which the third and the fourth point are mainly responsible for the VFA derivation. Therefore, an error in the rst two pH observations would be largely absorbed by the carbonate subsystem, minimizing the effect on the VFA calculation. The two new equations are solved to produce the rst estimate of AT and CT . Modication of the rst estimate in this approach is carried out as follows: a second estimate of AT and CT is calculated by again taking two pH pairs: one symmetrical about pKa (ie pH3 ; pH4 ) but the other asymmetrical about pKC1 (ie pH1 ; pH4 ). Subsequently, these two CT values (calculated in
1339

]x + [H2 PO4 ]x (22)

+ [OH ]x [H+ ]x } (Vx + Vs )

Where [y]x indicates molar concentration of species y after addition of x cm3 of standard acid (mol dm3 ), [A ] = dissociated short chain VFA species concentration (mol dm3 ) and Vs = volume of sample (dm3 ). Equation (22) can be reformulated in terms of total weak-acid species concentrations using equilibrium equations for the weak-acid systems and mass balance equations for each of the weak-acid systems as represented in eqns (23)(25) (For brevity only the VFA and carbonate subsystems are given. The other subsystems follow the same approach.) [HCO3 ]x = CT Vs /(Vx + Vs )/{1 + K + (H )x /K
+ 2 + C1 } C2 /{(H + C2 /(H +

)x (23)

[CO3 2 ]x = CT Vs /(Vx + Vs ) K + ((H )x ) /K + K a}

C1 }

)x + K

C2

(24)

[A ]x = AT Vs /(Vx + Vs ) K a /{(H+ )x (25)

Similar equations can be developed for the phosphate, sulde and ammonium proton-accepting species. Substituting the equations for each species concentration into eqn (22) (for example as given in eqns (23)(25) for the carbonate and VFA subsystems) gives an equation for total mass of alkalinity in terms of AT , CT , PT , NT , ST and pH: Mtotal alk(x) = {CT Vs /(Vs + Vx ) Fn1 (pH)x + AT Vs /(Vs + Vx ) Fn2 (pH)x + PT Vs /(Vs + Vx ) Fn3 (pH)x + ST Vs /(Vs + Vx ) Fn4 (pH)x + NT Vs /(Vs + Vx ) Fn5 (pH)x + 10(14pHx ) /fm 10pHx /fm } (Vs + Vx ) (26)

J Chem Technol Biotechnol 79:13311341 (online: 2004)

O Lahav, BE Morgan

the rst and second estimate) are compared, and if different, all pH observations are adjusted by pH and the calculation procedure is repeated (by changing pH) until the difference between the two CT values is negligible. The DOS computer program required for this method can be purchased from the Water Research Commission of South Africa at a cost of approximately $50. The necessity for correcting observed pH values was attributed by the authors to either a residual liquid junction potential (error in pH measurements caused by differences in the dissolved solids concentration between the buffer solution used to calibrate the probe and the test solution), or from poor pH meter calibration.18 Execution of the ve-point method is facilitated by a computer program supplied by the authors. The method was tested with results typically within 2% accuracy.18,26,27 Application of the method is restricted to cases where the CT is greater than twice AT , otherwise the systematic pH correction does not converge. As mentioned above, the ve-point method is the most ambitious effort thus far to include all system components in a single sophisticated, yet simple model (titration to ve points is carried out rapidly and with ease, because the selected pH points do not have to be reached accurately and the selected pH areas are very stable). However, the nal step (correction of pH observations) tends to detract from the theoretical justication of the method, and from its general applicability. With regard to errors emanating from residual liquid junction potential: the total dissolved solids concentrations in the samples tested in the vepoint approach varied between 500 and 1000 mg dm3 (after dilution), which is very close to the dissolved solids concentration in standard buffer solutions. For this reason and others it is very difcult to ascribe a systematic pH error correction to liquid junction effects.27 Lahav et al proposed a new eight-point titration method for measuring both VFA and H2 CO3 alkalinity.19 As in the ve-point method, execution is facilitated by a computer program (available free of charge from the author). The model extends the ve-point method by resolving the mathematical and analytical problems which gave rise to the systematic pH error of the Moosbrugger approach. In this modication, total alkalinity (PAC of all species in the sample) is measured accurately using the Gran titration technique,28 and this value is used, in addition to the rst estimate of AT and CT , to give the nal result. The Gran procedure requires a further three pairs of (Vx , pH) points taken in the pH range of 2.4 < pH <2.7. The assessment of the rst estimate of AT and CT is effected as follows: AT and CT , determined from the rst estimate, and Ve , determined from the Gran function analysis, are inserted in eqn (27) together with the initial pH value (ie where Vx = 0). Both AT and CT are now multiplied by a proportional term x, to account for inconsistencies in
1340

pH observations yielding: Ve Ca = {x CT Fn1 (pH0 ) + x AT Fn2 (pH0 ) + Const} Vs (28)

where: Const = a constant representing the protonaccepting term for the phosphate, sulde, ammonium and water subsystems at the initial pH (pH0 ). Solving for x gives an assessment of the rst estimate for AT and CT . The closer x is to unity, the better the rst estimate conforms to the accurately measured Ve . An acceptable value for x is a relative error (|(x 1)| 100) of less than 5%. The improved values for AT and CT are then obtained by multiplying each of the two parameters by x to conform to the accurately measured Ve using the initial pH. For the nal output of the algorithm, the improved AT gives the nal value for VFA concentration and the improved CT is used to calculate the nal value for H2 CO3 alkalinity using the initial pH. Typical results obtained using the method on simulative and industrial efuents are similar to the ve-point method, ie less than 2% error, but in contrast to the Moosbrugger approach, this method can be applied for any CT to AT ratio.19 It should be noted that in both the ve-point and the eight-point methods, when the initial pH is lower than about 6.85, a known volume of standard base should be added to the sample to allow acid titration to the prescribed pH points. In such cases the algorithm is changed as follows: (i) Vs is modied by the volume of NaOH addition; (ii) Ve is derived as before and then modied giving: Ve (nal) = (Ve (Gran function) Ca VNaOH CNaOH )/Ca (29)

where: VNaOH = volume of standard NaOH solution added to lift pH above 6.85 dm3 and CNaOH = concentration of standard NaOH solution (mol dm3 ). The disadvantages of Lahav et als method include a more tedious titration procedure, the need for measuring all weak-acid subsystems in addition to EC and temperature, and the need for using base titrant for samples with initial pH lower than 6.85. Advantages include high accuracy, and general applicability.

SUMMARY This review examines multiple on-site, titrative measurement approaches to monitor anaerobic processes. The main objective is to clarify the theoretical basis on which the methods were developed, to discuss their advantages and disadvantages, and to allow the operators to choose a method that is suitable for their process needs. Multipoint titration methods that take into account the various weak-acid subsystems and the important parameters of EC and temperature are generic and
J Chem Technol Biotechnol 79:13311341 (online: 2004)

Monitoring of anaerobic digestiona review

highly accurate. It appears that there is little reason to choose a more rudimentary approach with the current availability of relatively inexpensive computerized and programmable titration equipment. However, even the more simple methods can be applied successfully provided that the operators understand the assumptions and simplications behind the method, and provided that the accuracy level is such that it allows for proper detection of changes in VFA concentrations and rapid intervention.

15

16

17

REFERENCES
1 Van Haandel AC and Lettinga G, Anaerobic Sewage Treatment. John Wiley & Sons Ltd, London (1994). 2 Florencio L, Field JA and Lettinga G, Substrate competition between methanogens and acetogens during the degradation of methanol in UASB. Wat Res 29:915922 (1995). 3 Loewenthal RE, Wentzel MC, Ekama GA and Marais GvR, Mixed weak acid systems. Part 2: Dosing estimation, aqueous phase. Water SA 17:107122 (1991). 4 Pretorious WA, pH controlled feed-on-demand for high rate anaerobic systems. Wat Sci Tech 30:18 (1994). 5 Seghezzo L, Zeeamn B, van Lier JB, Hamelers HVM and Lettinga G, A review: the anaerobic treatment of sewage in UASB and EGSB reactors. Bioresource Technology 65:175190 (1998). 6 Moosbrugger RE, Wentzel MC, Ekama GA and Marais GvR, Weak acid/bases and pH control in anaerobic systemsa review. Water SA 19:110 (1993). 7 DiLallo R and Albertson OE, Volatile acids by direct titration. Journal WPCF 33:356365 (1961). 8 McGhee J, A method for approximation of volatile acid concentrations in anaerobic digesters. Water & Sewage Works 115:162166 (1968). 9 Ripley LE, Boyle WC and Converse JC, Improved alkalimetric monitoring for anaerobic digestion of high strength wastes. Journal WPCF 58:406411 (1986). 10 Pauss A, Rozzi A, Ledrut MJ, Naveau H and Nyns EJ, Bicarbonate determination in complex acidbase solutions by a back-titration method. Environmental Technology 11:469476 (1990). 11 Munch E and Greeneld PF, Estimating VFA concentrations in prefermenters by measuring pH. Wat Res 32:24312441 (1998). 12 Rozzi A and Brunetti A, Direct bicarbonate determination in anaerobic digester liquors. Environmental Technology Letters 2:385392 (1981). 13 Jenkins SR, Morgan JM and Sawyer CL, Measuring anaerobic sludge digestion and growth by a simple alkalimetric titration. Journal WPCF 55:448453 (1983). 14 Di Pinto AC, Limoni N, Passino R, Rozzi A and Tomei MC, Anaerobic process control by bicarbonate monitoring. 18

19

20

21

22

23

24

25

26

27

28

Abstract, 5th IAWPRC workshop on Instrumentation, Control, and Automation of Water and Wastewater Treatment. Yokahoma, Kyoto, August (1990). Colin F, Development of an automatic equipment for the study of acid-base equilibria for the control of anaerobic digestion, in Anaerobic Digestion and Carbohydrate Hydrolysis of Waste, ed by Ferrero GL, Ferraut MP and Naveau H. Elsevier Science, London, pp 381393 (1984). Kapp H, Schlammfaulung mit hohem Feststoffgehalt. Stuttgarter Berichte zur Siedlungswasserwirtscaft, Band 86, Oldenbourg Verlag, Munchen, 300 pp (1984). Moosbrugger RE, Wentzel MC, Loewenthal RE, Ekama GA and Marais GvR, A 5-point titration method to determine the carbonate and SCFA weak acid/bases in aqueous solution containing also known concentrations of other weak acid/bases. Water SA 19:2939 (1993). Moosbrugger RE, Wentzel MC, Ekama GA and Marais GvR, A 5-point titration method for determining the carbonate and SCFA weak acid/bases in anaerobic systems. Wat Sci Tech 28:237245 (1993). Lahav O, Morgan BE and Loewenthal RE, A rapid, simple and accurate method for measurement of VFA and carbonate alkalinity in anaerobic reactors. Env Sci & Tech 36:27362741 (2002). Powell GE and Archer DB, On-line titration method for monitoring buffer capacity and total volatile fatty acids levels in anaerobic digesters. Biotechnology and Bioengineering 33:570577 (1988). Feitkenhauer H, von Sachs J and Meyer U, On-line titration of volatile fatty acids for the process control of anaerobic digestion plants. Wat Res 36:212218 (2002). Loewenthal RE and Marais GvR, Carbonate chemistry of high salinity waters. Water Research Commission SA, Research Report No W46 (1983). Hattingth WHJ, Kotze JP, Thiel PG, Toerien DF and Siebert ML, Biological changes during the adaptation of an anaerobic digester to a synthetic substrate. Wat Res 1:255277 (1967). Rozzi A, Di Pinto AC and Brunetti A, Anaerobic process control by bicarbonate monitoring. Environmental Technology Letters 6:594601 (1985). Baucher K, A comparison of two simple titration procedures to determine volatile fatty acids in inuents to wastewater and sludge treatment processes. Water SA 24:4956 (1998). De Hass DW and Asam N, Use of simple titration procedure to determine H2 CO3 alkalinity and volatile fatty acids for process control in wastewater treatment. Water SA 21:307318 (1995). Lahav O and Loewenthal RE, Measurement of VFA in anaerobic digestion: the ve point titration method revisited. Water SA 26:389391 (2000). Gran G, Determination of the equivalence point in potentiometric titrations. The Analyst 77:661671 (1952).

J Chem Technol Biotechnol 79:13311341 (online: 2004)

1341