An automated spectrophotometric system for monitoring buffer capacity in anaerobic digestion processes

Download An automated spectrophotometric system for monitoring buffer capacity in anaerobic digestion processes

Post on 30-Oct-2016

215 views

Category:

Documents

2 download

Embed Size (px)

TRANSCRIPT

<ul><li><p>Water Research 38 (2004) 364</p><p>mro</p><p>sch,</p><p>Engin</p><p>ised f</p><p>Anaerobic biological treatment of wastewater has</p><p>degradation chain. This implies that unfavourable</p><p>conditions such as organic overload, unfavourable pH</p><p>ity problem. Upon instability, the loading to the process</p><p>ARTICLE IN PRESSmay have to be stopped for prolonged periods of time to</p><p>facilitate degradation of the VFAs and recovery of the</p><p>metabolic balance in the reactor. In severe cases the</p><p>instability may cause acidication of the reactor with</p><p>complete inhibition of the methanogenic activity. Such a</p><p>*Corresponding author. Tel.: +46-46-222-8264; fax: +46-</p><p>46-222-4713.</p><p>E-mail address: bo.mattiasson@biotek.lu.se</p><p>(B. Mattiasson).</p><p>0043-1354/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.wseveral advantages over aerobic treatment processes</p><p>such as lower sludge production and a valuable by-</p><p>product; methane. Despite these obvious advantages,</p><p>the use of anaerobic treatment is not as widespread as it</p><p>could be and one reason for this is its reputation of</p><p>instability during start-up and operation.</p><p>The anaerobic biodegradation of organic matter to</p><p>biogas is conducted by a consortium of microbial groups</p><p>with a high degree of interdependence. The methano-</p><p>genic organisms (pH optimum: 6.5 to 7.5) are considered</p><p>to be the most sensitive organisms in this consortium</p><p>or presence of toxic compounds will cause instability</p><p>and be observed as elevated concentrations of metabo-</p><p>lites in the metabolic chain preceding the action of the</p><p>methanogens. One group of metabolic intermediates</p><p>that is especially of high signicance is the volatile</p><p>fatty acids (VFAs). Increased concentration of VFAs</p><p>may lead to a decrease in the buffering bicarbonate</p><p>concentration, as the bicarbonate becomes protonated</p><p>and is released as carbondioxide gas. The pH of the</p><p>reactor may decrease to levels below the optimal for the</p><p>methanogenic organisms, thereby escalating the instabil-useful by-product. A common way of preventing instability problems and avoiding acidication in anaerobic digesters</p><p>is to keep the organic load to the digester far below its maximum capacity. An improved way of operating digesters</p><p>would be to use monitoring and control systems for increased organic load under controlled conditions such that the</p><p>digester performance is improved. The partial alkalinity, which indicates the bicarbonate concentration, has in many</p><p>cases been found to be a suitable parameter to monitor. Here, an automated monitoring system for alkalinity</p><p>measurements is described. It is shown to be applicable for measuring a wide range of bicarbonate concentrations. The</p><p>system shows potential for monitoring anaerobic digesters as it responds to the alkalinity of digester efuent, as well as</p><p>being stable over a relatively long time span with few maintenance requirements.</p><p>r 2004 Elsevier Ltd. All rights reserved.</p><p>Keywords: Automated monitoring; Buffer capacity; Alkalinity; Spectrophotometric; Anaerobic digestion</p><p>1. Introduction and are also the slowest growing organisms in theAnaerobic digestion is a suitable method for the treatment of wastewater and organic wastes, yielding biogas as aAn automated spectrophotobuffer capacity in anae</p><p>Tor Gunnar Jant</p><p>Department of Biotechnology, Center for Chemistry and Chemical</p><p>Received 7 October 2003; received in rev</p><p>Abstractatres.2004.05.01053650</p><p>etric system for monitoringbic digestion processes</p><p>Bo Mattiasson*</p><p>eering, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden</p><p>orm 7 May 2004; accepted 12 May 2004</p></li><li><p>and the acid is compared to mixing ratios of the</p><p>ARTICLE IN PRESST.G. Jantsch, B. Mattiasson / Water Research 38 (2004) 364536503646situation makes it necessary to remove the reactor</p><p>content and to start up the reactor again.</p><p>A common way of avoiding instability problems is to</p><p>operate the anaerobic digestion process at an organic</p><p>loading rate which is far below the maximum capacity of</p><p>the system. A more economically favourable way of</p><p>operating a digester and avoiding instability problems</p><p>would be to monitor and control the process. An ideal</p><p>monitoring and control system would detect instability</p><p>and imply countermeasures to compensate for the</p><p>instability. The monitoring system should be online,</p><p>automated, robust and give early indications of in-</p><p>stability in the process. Some process indicators, which</p><p>traditionally have been used for monitoring, are the gas</p><p>production rate, gas composition, pH, alkalinity and</p><p>concentrations of VFAs. Liquid-phase parameters (pH,</p><p>alkalinity, VFA) reect the environment of the micro-</p><p>organisms and therefore, often give a faster response</p><p>than gas-phase parameters (gas composition and pro-</p><p>duction rate). pH is commonly used as a process</p><p>indicator, but the effectiveness of using this as a control</p><p>parameter is strongly dependent on the alkalinity, i.e.</p><p>buffering capacity of the process (Ahring et al., 1995).</p><p>The alkalinity is mainly dependent on the bicarbonate</p><p>and VFA-concentrations (log[acid dissociation con-stant] (pKa) values of 6.35 and 4.75, respectively), and in</p><p>some processes the ammonium concentration (pKa-</p><p>value of 9.4). The most important buffering species</p><p>within the optimal pH for the methanogenic organisms</p><p>is the bicarbonate and therefore it is of interest to</p><p>develop methods to monitor it.</p><p>The total alkalinity (TA) (alkalinity to a pH of 4.3)</p><p>has been used for monitoring anaerobic processes but is</p><p>considered as an insensitive indicator of process</p><p>instability. As shown by the pKa-values of bicarbonate</p><p>and VFAs, measurements of TA will reect both the</p><p>levels of VFAs and bicarbonate. Upon instability, the</p><p>increase in VFA concentrations will cause a decrease in</p><p>bicarbonate concentration resulting in a constant TA-</p><p>value (Jenkins et al., 1991). Partial alkalinity (PA)</p><p>(alkalinity to a pH of 5.75) reects mostly the</p><p>bicarbonate concentration and has been found to be a</p><p>valuable tool for process monitoring (Bjornsson, 2000).</p><p>More sophisticated titration techniques have been</p><p>Nomenclature</p><p>Abbreviations</p><p>VFA volatile fatty acidsdeveloped for the determination of VFA and bicarbo-</p><p>nate concentrations (Moosbrugger et al., 1993). On-line</p><p>methods for monitoring of bicarbonate concentrations</p><p>have been developed based on sample saturation with</p><p>carbon dioxide and acidication to release bicarbonate</p><p>as carbon dioxide with subsequent gas-pressure orstandard samples and used to determine the PA of</p><p>the sample. Upon changes in bicarbonate concentrations</p><p>of the samples, this is reected in a change in the mixing</p><p>ratio between the sample and the acid.</p><p>The monitoring system was based on a method</p><p>described by Jantsch and Mattiasson (2003) and is</p><p>schematically presented in Fig. 1. The spectrophot-</p><p>ometer and the pumps were controlled and data were</p><p>collected by Fieldpoint modules (National Instruments,</p><p>USA) connected via ethernet to a PC. The PC program</p><p>for control, data collection and data handling was</p><p>developed in the Labview program (National Instru-</p><p>ments, USA).gas-ow determination (Hawkes et al., 1993; Rozzi</p><p>and DiPinto, 1994). Also online methods based on other</p><p>principles such as titration have been developed (Powell</p><p>and Archer, 1989; Almeida et al., 2001).</p><p>Previously, a new spectrophotometric method for</p><p>alkalinity measurements based on pH indicators has</p><p>been used for process monitoring of different anaerobic</p><p>digester systems (Bjornsson et al., 2001; Jantsch and</p><p>Mattiasson, 2003). The aim of this investigation was to</p><p>construct and characterise a fully automated system for</p><p>monitoring of anaerobic digestion processes based on</p><p>the new spectrophotometric method.</p><p>2. Materials and methods</p><p>2.1. Titrimetric alkalinity</p><p>The titrimetric alkalinity was measured as PA</p><p>by titration to pH 5.75 with standardised 0.1M HCl</p><p>by using a TitraLabTM 80 titrator (Radiometer,</p><p>Copenhagen, Denmark). The results are reported as g</p><p>CaCO3 11.</p><p>2.2. Automated alkalinity monitoring system</p><p>The principle for the monitoring system is that a pH</p><p>indicator is used to spectrometrically determine the pH</p><p>of a mixture of the sample and an acid. When the pH of</p><p>the mixture is 5.75, the mixing ratio between the sample</p><p>pKa log[acid dissociation constant]TA total alkalinity</p><p>PA partial alkalinityThe solutions of sample and acid were delivered by</p><p>peristaltic pumps (Alitea, Sweden) tted with Tygon</p><p>tubings. The peristaltic pump caused uctuations of the</p><p>ow and these uctuations were dampened by an air</p><p>lled syringe tted into the line just after the pump.</p><p>Teon tubing (outer diameter lmm, inner diameter</p></li><li><p>The response of the monitoring system towards</p><p>ARTICLE IN PRESST.G. Jantsch, B. Mattiasson / Water Research 38 (2004) 36453650 36470.75mm) was used for connecting tubes and mixing coil.</p><p>The conuence point was constructed of a Teon</p><p>polymer Tee (Scantech, Sweden). The tube connections</p><p>were made of silicone rubber tubing. An Ultrospec 1000</p><p>spectrophotometer (Pharmacia Biotech, Sweden)</p><p>equipped with a SOG 1.0 ow cell (Starna) was used</p><p>for detection.</p><p>A ow of the sample was continuously mixed with a</p><p>ow of degassed acid (20 or 25mM HCl) containing</p><p>methylred pH-indicator (5mg l1). The mixture was</p><p>pumped continuously through the ow cell in the</p><p>spectrophotometer.</p><p>Determination of PA by the system was based on the</p><p>measurement of the spectral properties of the indicator,</p><p>that is, the absorbance of the yellow and red colours of</p><p>the pH indicator Methylred. The absorbance maxima</p><p>for the protonated (red) and unprotonated (yellow) were</p><p>438 and 516 nm, respectively, and these were used as the</p><p>wavelengths for measurement.</p><p>For each wavelength the absorbance was monitored</p><p>for l50 s (ve measurements per second) as the mixture</p><p>owed through the ow cell. The absorbance value was</p><p>averaged for the 150 s. The total measurement time for</p><p>each sample was two times 150 s; 5min.</p><p>The ratio between the absorbances at the two</p><p>wavelengths was used to indicate the pH of the mixture.</p><p>A ratio of 2.30 indicated a pH of 5.75 in the mixture.</p><p>The ratios 2.20 and 2.40 indicated pHs of 5.7 and 5.9,</p><p>respectively. For a measurement to be valid, a pH range</p><p>of the mixture of 5.7 to 5.9 was set. If the pH of the</p><p>mixture went outside this range, the measurement was</p><p>considered to be invalid because it deviated too far from</p><p>WasteMixing coil Spectro-</p><p>photometerSample</p><p>Acid</p><p>Pump</p><p>Pump</p><p>PC</p><p>Fig. 1. Automated system for monitoring buffering capacity.the desired value of 5.75. The mixing ratio was recorded</p><p>as sensor response.</p><p>To make the pH-value approach the desired pH, the</p><p>mixing ratio for the subsequent sample was adjusted by</p><p>controlling the speed of the pumps. The size of</p><p>adjustment depended on the deviation from the desired</p><p>pH-value.</p><p>pH pKaHCO3 =H2CO3: 1</p><p>From Eq. (1), the ratio HCO3/H2CO3 can be calculated.</p><p>For the ideal mixture of sample and acid, the pH is</p><p>5.75 which gives a HCO3/H2CO3-ratio of 0.251. At thesamples of Na2CO3 (5, 10, 15, 20 and 25mM) was</p><p>tested.</p><p>Anaerobic digester samples were ltered online from</p><p>a full-scale anaerobic digester treating municipal</p><p>sludge. The lter was a nylon cloth (gap size: 20 mm)inserted into a pipeline where the reactor content were</p><p>circulated from the bottom of the reactor, through a</p><p>heat exchanger and to the top of the reactor. The</p><p>samples were diluted with distilled water or spiked with</p><p>Na2CO3-solution and measured using the monitoring</p><p>system.</p><p>The stability of the system during a 5-day period was</p><p>investigated. Different standard samples were monitored</p><p>continuously in the system during the period.</p><p>The system was applied to monitor a full-scale</p><p>anaerobic digester. The feed to the digester consisted</p><p>of a mixture of municipal wastewater sludge and potato-</p><p>processing waste. An overload situation was induced by</p><p>overloading the digester with a pulse of potato-proces-</p><p>sing waste that corresponded to 20% of the normal daily</p><p>load of the digester (Jantsch, 2003).</p><p>3. Results and discussion</p><p>The linear relationship between the concentration of</p><p>Na2CO3 in standard solutions and the response of the</p><p>monitoring system is illustrated in Fig. 2. In the</p><p>concentration range 525mM Na2CO3 the monitoring</p><p>system response had a regression analysis correlation</p><p>coefcient (R2) of 0.9965. The resolution of the system</p><p>was dened as two times the standard deviation for the</p><p>sample series that had the highest standard deviation,</p><p>and when inserted in the regression analysis linearity</p><p>formulae, this gave a resolution of 1.44mM Na2CO3.</p><p>Guwy et al. (1994) found the range 5 to 50mM</p><p>NaHCO3 with an accuracy of 7.5% (0.383.8mM) in</p><p>a response time of 30min in an automated system based</p><p>on gas ow measurements from an acidied sample.same mixing ratio, the pH of the mixture will decrease to</p><p>5.7 if the bicarbonate concentration of the sample</p><p>decreases by 2.5%. Likewise, at the same mixing ratio</p><p>the pH will increase to 5.9 if the bicarbonate concentra-</p><p>tion of the sample increases by 8.3%. This implies that if</p><p>the bicarbonate concentration in the sample is decreas-</p><p>ing by 2.5% or increasing by 8.3% before the next</p><p>sample is taken, the next measurement will be valid, that</p><p>is, within the range. If the changes are higher than the</p><p>limits, then the measurement will be out of range and the</p><p>system will require one or more measurements with</p><p>adjustment of the mixing ratio for the pH of the mixture</p><p>to be within the range again.</p><p>2.3. Characterisation of the system</p></li><li><p>ARTICLE IN PRESS</p><p>0</p><p>2</p><p>4</p><p>6</p><p>Titrimetric partial alkalinity [g CaCO3 l-1]</p><p>Mon</p><p>itorin</p><p>g </p><p>0 1 2 3 4</p><p>Fig. 3. Monitoring system response as a function of titrimetric</p><p>partial alkalinity of diluted and spiked digester efuent sample</p><p>with linear regression analysis (y 5:4068x 1:0508;R2 0:9863). X-error bars depict 95% condence intervalsfor titrimetric measurements in duplicates; y-error bars depict</p><p>95% condence interval for system response in triplicates.</p><p>6</p><p>8</p><p>10</p><p>12</p><p>ing </p><p>syst</p><p>em re</p><p>spon</p><p>se</p><p>10 15</p><p>20 25</p><p>30</p><p>T.G. Jantsch, B. Mattiasson / Water Research 38 (2004) 364536503648In this system the monitored range of alkalinity of the</p><p>sample was dependent on the concentration of the acid</p><p>titrant. By changing the concentration of the acid</p><p>titrant, the range of sample alkalinity could be shifted,</p><p>thereby expanding the range where the system is useful.</p><p>This can also be used to make the concentration of the</p><p>digester sample more or less diluted in the mixture. By</p><p>diluting the sample in acid and using the ratio between</p><p>the absorbance at two wavelengths, the inuence of the</p><p>intrinsic absorbance of the digester sample is minimised.</p><p>0</p><p>2</p><p>4</p><p>6</p><p>8</p><p>10</p><p>12</p><p>14</p><p>0 5 10 15 20 25 30</p><p>Na2CO3 [mM]</p><p>Mon</p><p>itorin</p><p>g sy</p><p>stem</p><p> resp</p><p>onse</p><p>Fig. 2. Monitoring system response as a function of Na2CO3concentration in standard solutions with linear regression</p><p>analysis (y 05008x 0:3213; R2 0:9863). Error bars depict95% condence interval for the system response (3 to 7</p><p>measurements).The linear relationship between the response of</p><p>the monitoring system and the titrimetric PA of diluted</p><p>or spiked samples of anaerobic digester efuent, in</p><p>the concentration range of 1.12.8 g CaCO3 11 (11</p><p>28mM), is illustrated in Fig. 3. A correlation study</p><p>showed good agreement between the response of the</p><p>monitori...</p></li></ul>