During calibration, the instrument constants are determined from measurements made when the U-tube is filled with either air or water. The numerical values for A and B are retained in the memory of the instrument allowing the microprocessor to calculate the density of unknown samples. In operation, the tube is oscillated to its natural frequency. The time period of the frequency is determined by comparison of the number of oscillations with an internal quartz clock. From this value, the density of the liquid in the U-tube may be calculated.

Very accurate measurements may be obtained with this instrument; however, some sample preparation is necessary. Yeast cells and trub must be removed, either by centrifugation or by filtration. In addition, where necessary the sample must be degassed. This is conveniently achieved by brief immersion in an ultrasonic bath. Automatic measurement of gravity. Several methods for measuring wort gravity automatically during fermentation have been developed. These have the advantage that off-line sampling is not required and gravity is measured continuously. This provides early warning of non-standard behaviour and rapid identification of the achievement of racking gravity. Output from automatic gravity measuring systems can be used in control regimes. For example, regulation of attemperation in response to changes in wort gravity. On the other hand, automatic gravity monitoring systems are expensive and obviously must be fitted to each fermenter. Bearing costs in mind, they are appropriate for use only with high capacity vessels.

Gravity sensing devices may be sited within the fermenting vessel. Alternatively, they may be located within a loop of main through which the wort is circulated, out of and back into, the vessel. Sensors must be located in such a way that a representative reading of the whole of the contents of the fermenter is obtained. As with manual sampling this can be a problem where a single point sensor is used, particularly towards the end of fermentation when layering may occur in large vessels. Locating sensors in a circulation loop is an advantage in this respect since continuous pumping is an aid to good mixing. However, some gravity sensors are prone to errors due to the presence of gas bubbles. Gas breakout can be a problem in external loops and this must be controlled by fitting appropriate valves to maintain sufficient back-pressure to keep carbon dioxide in solution. Invasive devices such as probes introduce a potential source of contamination and a cleaning problem. Perhaps in this regard, external loop systems present a greater problem than in situ systems. Whichever method is adopted, hygiene is a critical design parameter.

The oscillating U-tube method of gravity determination, as described for analysis of off-line samples (Section can also be used to make in-line measurements (Jiggens, 1987). In this case, the measuring U-tube is located in an external pressurised loop. For improved robustness the U-tube is made from stainless steel as opposed to glass in the laboratory instrument and is of a wider bore to accommodate high flow rates. Since it is not possible to make measurements at 20°C a temperature sensor is fitted to the U-tube to allow appropriate compensation to be made. Output from the instrument can be fed to a recorder, visual display of gravity or as a voltage or current for input to a control device.

A few devices have been developed in which the sensors are located within fermenting vessels. In the Gravibeam system (Dutton, 1990), a displacer of known volume is immersed in the fermenting wort. The displacer experiences an upthrust, which is detected by a highly accurate load cell. The value of the upthrust is proportional to the density of the wort. After calibration, output from the load cell is expressed directly as specific gravity. The device is intrusive. However, it is constructed entirely from stainless steel and is designed to be cleaned with the vessel CiP. The system can be retrofitted to existing fermenters but it is necessary to break into the wall of the vessels to do this. It is reportedly unaffected by gas breakout, the presence of yeast or trub and pressure variations. Temperature compensation is provided by a platinum resistance thermometer fitted to the displacer. Output from several sensors, each located in a separate fermenter, is fed, via a serial data link, to a microprocessor for visual display and/or use in a control system.

Density may be computed from differences in pressure measured simultaneously at different depths within vessels. Thus, at a constant vertical distance between two points the differential in pressure is a direct function of the density of the liquid phase. The Platometer (Fig. 6.15) described by Moller (1975) uses two stainless steel diaphragms separated by 45 cm and located close to the inner wall of the fermenting vessel. Pressure is sensed by the diaphragms and transmitted to a transducer, which converts the pressure differential into an electrical signal. The pressure sensors and transducer are linked by water-filled columns. This arrangement means that the diaphragms and transducers are balanced by equal sized columns of both wort and water, at the same temperature. Therefore, the electrical output from the transducer provides a temperature-compensated indication of specific gravity.

The transducer consists of a housing made from iron, which is divided into three compartments by two transmission diaphragms. The two outer compartments terminate the water columns and the inner compartment is filled with an incompressible

Wort flow

Wort flow

Fig. 6.15 Platometer automatic gravity sensor.

fluid. The centres of the transmission diaphragms are linked by a rigid connecting arm. The latter is also attached to a leaf spring. Movement of this arm provides a measure of the pressure differential of the system. An electrical signal is generated by allowing movement of the spring to actuate moving plate condensers. The resultant capacitance is then converted to a signal that is directly proportional to specific gravity.

The pressure sensing diaphragms are shrouded in a stainless steel casing provided with apertures at top and bottom to allow access to wort. The casing gives protection to the device and more importantly minimises errors due to short-term variations in pressure due to random convection currents. An armature serves as the method of attaching the casing to the vessel wall. This also provides a duct for electrical wiring and an entry point for dedicated cleaning in place. The system reportedly gave an accuracy of +0.1 "Plato.

Gravity determination by differential pressure measurement has been described by others, for example, Cumberland et al. (1984). This report described a device that used pressure cells separated by 3 m of vertical tank wall. Other sensors measured temperature and the tank pressure (difference between top pressure and that of the atmosphere) and the difference between the tank bottom and atmospheric pressure. The vessel level contents could be deduced from the difference between the tank and bottom pressures. Output from the gravity meter was in the form of a 4-20 mA current suitable for input to the vessel temperature controller.

Sugden (1993) reported use of an automated fermentation monitoring and control system termed FerMAC, which has been installed in several UK breweries. This used three pneumatic pressure balance transmitters located in the vessel as shown in Fig. 6.16. The transmitters comprised stainless steel diaphragms attached to the inner wall of the vessel. Gas, either nitrogen or carbon dioxide, is fed to the back of the

Fig. 6.16 FerMAC system for automatic monitoring of fermentation (from Sugden, 1993).

diaphragms via flow regulators. When the gas flow rate balanced the force exerted by the fermenting wort, the return pressure from the diaphragm provided a measure of the hydrostatic head above the transmitter. Comparison of output from the top and bottom sensors (1 and 3) provided a level reading. Similarly, the difference in output from the two lower sensors (2 and 3) could be related to wort gravity. It was reported that careful positioning of the sensors was required. The top one should be located such that CiP sprays did not cause damage. The lower transmitter was best placed at the top of the cone of the vessel to avoid false readings due to the diaphragm becoming covered with yeast.

The data acquisition module received output from sensors fitted to up to 15 fermenters. It converted the transmitted pressure data into digital readings of level, gravity, volume and top pressure. Together with output from a thermometer, the data was transferred to a display and control PC. This provided a graphical representation of change in gravity and temperature with time for each fermentation, and tabulated data describing tank volumes, yeast pitching details and records of CiP.

The system was shown capable of measuring gravity to an accuracy of 1 "saccharin or 0.5°Plato. It was proposed that the system could be used to control fermentation rate by relating the gravity output to the attemperation system. In addition, the processor could be interfaced with the CiP and tank filling and emptying systems to provide a complete fermentation management system.

Hees and Amlung (1997) described yet another differential pressure system for monitoring gravity in large cylindroconical fermenters. This system uses three pressure sensors, one located in the cone, the second just below the surface of the wort and the third in the headspace. Output from these is used to calculate wort specific gravity, tank volume and headspace pressure. Output from the top transmitter allows compensation for changes in headspace pressure. The authors claimed that this could affect the accuracy of gravity measurement. This system provided continuous gravity output, which correlated closely with that obtained from off-line measurements and with a precision of + 0.2%. As with other pressure systems, the output could be used for automatic control. In addition, the headspace sensor was used to provide a signal for automatic control of a carbon dioxide recovery system.

On-line gravity measurement can be made using an ultrasonic sensor, as described by Forrest and Cuthbertson (1986). The sensor measures the time taken for an ultrasonic signal to pass between two fixed points where the intervening space is filled with the liquid which is to be analysed. The time, or sound velocity, is related to the density of the liquid. Gas bubbles interfere and it is necessary to locate the sensors within an external loop. The loop, through which the wort is circulated, is operated under back pressure to prevent gas break-out. An ultrasonic transmitter and receiver are located at one end of the pipe such that the signal passes through the liquid, is reflected off the other end of the pipe and then detected on its return. The method of measurement is affected by temperature and a compensating thermometer is located close to the sensors. Calibration is empirically derived from comparison of output with off-line measurements.

The ultrasonic sensor is relatively inexpensive. It was shown to have an accuracy of + 0.5°saccharin and has the advantage of being totally non-invasive. However, this has to be balanced against the requirement to use an external loop. Probably this sensor is more suited for application to measurement of the gravity of bright beer. For example, it is particularly suited to on-line dilution of high-gravity beers prior to packaging (Forrest, 1987; Forrest et al., 1989). These authors described an in-line sensor which used a combination of ultrasonic and refractometric measurement to measure beer original gravity.

6.3.3 Monitoring C02 evolution rate

Formation of carbon dioxide during fermentation is stoichiometric with ethanol formation and sugar utilisation. Therefore, the profile of carbon dioxide evolution can be used to monitor fermentation progress. Daoud and Searle (1990) studied patterns of carbon dioxide evolution in trial fermentations from laboratory (1.5 litre) to pilot scale (100 hi). At laboratory scale, the authors demonstrated correlation coefficients of 0.9944 between C02 evolved and ethanol production and 0.99 between C02 evolved and carbohydrate utilisation. In 100 hi fermentations, no gas evolution was observed until the wort became saturated. After this time, rates of approximately 1.0 g C02 per litre per degree gravity drop were measured.

Stassi et al. (1987, 1991) used thermal mass flow meters to measure C02 evolution rates at both laboratory and production scale. They also noted a high correlation between C02 formation and decline in gravity. In addition, the former also correlated with ethanol formation, the extent of yeast growth, decline in wort pH and the concentration of dissolved sulphur dioxide. At laboratory scale, an automatic control system was established in which a set-point rate of C02 evolution was maintained in a feedback loop in which temperature was variable.

Eyben (1989) described a totally automated fermentation monitoring and control system used at the Sebastien Artois Brewery in France. This was also based on the correlation between C02 evolution and gravity decrease. A single tank fermentation and conditioning process was controlled by measuring total carbon dioxide evolution using a vortex flow-meter. The fermentation, which lasted for a total of 12 days, had three distinct phases. In the first period of active primary fermentation the temperature was held at 11°C and the top pressure 20 g cm 2. In the second phase of warm conditioning, the temperature was allowed to increase to 16°C and the top pressure to 700gem 2. During the second phase, wort attenuation and C02 evolution were completed. The third phase commenced when the vessel contents were cooled to 0°C whilst maintaining the same high top pressure. This allowed an appropriate level of carbonation to be achieved. After 12 days, the yeast was harvested and the vessel was racked. Whilst the vessel was emptied the C02 top pressure was maintained to ensure anaerobiosis during transfer of beer to filtration.

During primary fermentation, the evolved C02 was directed through the flow cell then collected via a triple exhaust-gas network, operating at atmospheric, 20 and 700 g cm 2, respectively. The triple arrangement allowed maintenance of the same pressure at the inlet and outlet of the flow meter. Entry into individual legs of the C02 network was via automatic valves. In the flow cell, vortices proportional to gas flow rates were detected by a piezoelectric sensor and converter. This was mounted on the outside of the flow cell and provided a 4-20 mA output. The flow meter had an operating range of 0-30 m3 h 1 with an accuracy of + 1 %. The flow meter was not affected by the presence of moisture in the gas phase or by variations in temperature.

The system was used in the following manner. During early primary fermentation, C02 was vented to atmosphere. When a pre-determined volume of gas had passed through the flow meter (c. 12 hours after collection), the atmospheric valve was closed and flow directed to the C02 recovery system working at a top pressure of 20 g cm 2 The fermentation was allowed to proceed at a temperature of 11°C until a second predetermined total volume of C02 was recovered. At this time the exit gas flow was switched, again automatically, to the C02 recovery system operating at a top pressure of 700gem 2 and the temperature allowed to increase to 16°C. After a further five days, when the wort was completely attenuated and C02 evolution had ceased, cooling was applied and the fermenter contents were chilled to 0°C.

Determination of the volumes of C02 required to trigger the changes described above were based on theoretical calculations. These took into account the capacity of the vessel, the volume and fermentability of the wort, the beer type and the growth characteristics of the yeast strain. Measurements from flow cells were made every 20 minutes, the total evolved gas being calculated by a microprocessor based on extrapolation. This system allowed a single flow meter to monitor gas evolution simultaneously from eight vessels such that 24 measurements were made per hour per flow meter. A single supervisory computer collected and recorded data from subordinate microprocessors as well as calculating the set-point gas volumes on which process steps were based.

Monitoring fermentation progress via profiles of C02 evolution is an attractive option. The sensors are non-invasive, relatively inexpensive, and, as in the example described above, several vessels can be serviced by a single device. An output is provided for use in feedback control loop systems. It has the major advantage that the status of the entire body of fermenting wort is assessed as opposed to single point measurements such as may be made with an automatic gravity sensor. Thus, problems associated with wort heterogeneity are obviated. The major potential problem is that there is little or no opportunity to gather data during early fermentation. In the first few hours of fermentation, the critical processes of oxygen assimilation and yeast sterol synthesis take place. In this phase, little or no C02 formation occurs and, as discussed earlier, even when gas evolution begins there is the period of inertia due to saturation of the wort. In the event of inadequate control of wort oxygenation or pitching rate, non-standard performance must be identified as quickly as possible. Thus, there is but a short period in which appropriate corrective action can be taken (see Section 6.5.4). By the time C02 evolution is detectable, this window of opportunity is probably closed.

Annemuller and Manger (1997) addressed this problem, suggesting that it would be more appropriate to monitor the dissolved carbon dioxide concentration in very early fermentation in the period when the wort has not had a chance to become saturated. Undoubtedly this is true but it adds a further level of complexity to the control system.

6.3.4 Monitoring exothermy

Glycolysis is an exothermic process, and therefore the profile of heat generation during fermentation can be used to monitor progress (Mou & Cooney, 1976). Ruocco et al. (1980) developed such a system in which exothermy was determined by periodically measuring the increase in temperature of wort when the coolant supply was switched off. The numerical value of the exotherm measured at any particular time was based on calculations that took into account the observed increase in temperature as a function of time and the total quantity of wort present. Exotherm measurements made throughout fermentation were put together to form a profile, which showed good correlation with the profiles of gravity attenuation and yeast growth. These profiles were specific for individual fermenters and wort qualities. Averaging many sets of data allowed an 'ideal' exotherm profile to be constructed which could be used as a standard for comparison with new data.

Exothermy measured at any time during fermentation provides an instantaneous indication of fermentation rate (attenuation rate). Parameters which influence fermentation rate have a concomitant effect on exothermy. For example, increasing the temperature set-point increases the value of exothermy. The total exotherm, which may be determined by calculating the area encompassed by the profile of exothermy versus time, is a function of the total quantity of fermentable sugar present in the wort. It follows, therefore, that measurement of exothermy can be used to control fermentation by comparing the measured profile with a standard. In the event of under-achievement, the set-point temperature may be allowed to rise until correspondence is achieved. Second, through automatic integration of the exotherm/time profile, achievement of the desired attenuation gravity can be inferred.

As with measurement of rates of C02 evolution, monitoring fermentation through determination of exothermy has the advantage of being non-invasive and provides an output suitable for use in automated control systems. It does have some disadvantages. If temperature readings are made at single specific locations within the vessel there is an assumption that the wort is perfectly mixed. This is not so at all times and clearly it would be necessary to take an averaged output from several thermometers. The nature of the coolant is also critical. The trials reported by Ruocco et al. (1980) used vessels cooled by direct ammonia expansion. This system has little residual effect when the coolant valves are closed. With refrigerants such as glycol there is a considerable residual effect which would tend to underestimate the true value for exothermy.

Like on-line measurement of gravity or rate of C02 evolution, there is apparently little exothermy during the critical early phase of fermentation, which perhaps again mitigates against this approach to monitoring for use in interactive control strategies. However, this is not necessarily strictly true. Ruocco et al. (1980) reported that in some trials an early and transient peak in exothermy was sometimes but not always observed. They speculated that this burst of exothermy could be associated with sterol synthesis during the aerobic phase of fermentation. If this observation was verified it could perhaps be used as a probe to provide confirmation that the conditions established at the completion of wort collection were appropriate. If not, corrective action could be initiated during early fermentation and the exothermy profile obtained thereafter could be used simply to monitor and confirm satisfactory progress. It is certainly true that when pitching yeast is exposed to oxygen under non-growing conditions, sterol synthesis and glycogen dissimilation proceed. This is an exothermic process (see Section and Fig. 6.31).

6.3.5 Monitoring pH

The transformation from wort to beer is accompanied by a decline in pH, typically from just over pH 5.0 to around pH 4.0. This change is a consequence of yeast metabolism, involving excretion of several organic acids and proton extrusion in response to assimilation of sugars (see Section 3.7.1). The pattern of pH change is characteristic for a given fermentation. Therefore, on-line measurement could be used to monitor fermentation progress. This is particularly so since pH electrodes capable of withstanding the rigours of the production environment are now readily available.

The most dramatic changes in pH occur during early fermentation and the minimum value is achieved before wort attenuation is complete (see Section 3.2). Often, there is a modest increase in pH from the mid-point onwards. In this regard, therefore, pH is not a particularly useful monitor of overall fermentation performance, and certainly it is of no value in identifying the end-point. However, the rapid decrease, which occurs in the first few hours after pitching, presents early identification of nonideal performance.

Leedham (1983) described a control regime in which initial rates of pH decline were measured immediately after the completion of collection and compared with stored profiles pre-determined for particular fermentation types. The system was designed to optimise wort oxygenation. During collection, the wort was oxygenated to a less than optimum concentration. At the end of collection, the pH of the wort was measured at intervals of 15 minutes. After each measurement was made, aliquots of oxygen were added to the vessel if the rate of pH decline was less than the stored control profile. No action was taken if the pH value was at, or below the stored figure. If, after 10 cycles, the pH reading was still too high, the temperature set-point was raised progressively, up to a maximum of 2°C above the normal value.

The method has utility; however, it is not capable of responding to a situation where the pH decline is greater than the standard. This could occur in the event of accidental over-pitching/wort oxygenation, or more likely where yeast had a less than usual requirement for wort oxygenation because of exposure of pitching yeast to air during storage and handling (see Section 6.3.6).

6.3.6 Monitoring rate of oxygen assimilation

Oxygen added with wort at the beginning of fermentation is used by yeast to syn-thesise lipids, principally sterols and unsaturated fatty acids, which are essential to proper membrane function (see Section 3.5.1). The quantity of sterol synthesised by individual yeast cells is governed by availability of oxygen and the pitching rate. The quantity of sterol synthesised per yeast cell controls the extent of yeast growth during fermentation.

Oxygen falls to an undetectable concentration a few hours after the completion of wort collection. Since conditions are anaerobic during most of fermentation, monitoring oxygen concentration provides little useful information. Thus, as with monitoring of pH, it is not possible to correlate oxygen concentration with wort attenuation. However, also in common with pH decline, measurement of oxygen consumption by yeast affords a method for early prediction of non-ideal behaviour and an opportunity for remedial action to be taken.

It has been demonstrated that the rate at which yeast assimilates exogenous oxygen gives an indication of its physiological condition. In particular, it can be related to the size of the sterol pool (Kara et al., 1987; Boulton et al., 1991). Measurement of the rate at which oxygen is taken up during the aerobic phase of fermentation can be used to assess the sterol content of the yeast at pitch (providing the yeast concentration and temperature are defined). The appropriate quantity of oxygen may then be added to ensure that additional sterol synthesis brings the total concentration up to an optimum value for that fermentation.

In practice, this is a difficult undertaking. When wort collection is finished, the yeast concentration may be unknown since cell proliferation may have already commenced. Dosing oxygen into fermenter is inefficient and difficult to quantify. The requirement to adjust conditions post-collection prevents making a rapid start to the fermentation. A potential approach is to assess the yeast oxygen uptake rate in-line during wort collection, as shown in Fig. 6.17. This system consists of a loop of coiled stainless steel tubing attached to the wort main at a point after addition of oxygen and yeast. At each terminal of the loop an in-line dissolved oxygen meter is provided. The coil, which is attemperated, must be long enough to provide sufficient distance between the two dissolved oxygen meters to allow significant uptake of oxygen by the yeast.

In operation, wort flow commences at a known rate and oxygenation and yeast pitching are initiated. Yeast is dosed accurately into the wort at a rate controlled by a biomass meter (see Section When steady flow rates have been established, the flow through the coil is initiated by automatically opening valves Vi and V2. Measurement of the decrease in dissolved oxygen concentration between the meters A and B allows a microprocessor to compute the oxygen uptake rate. With input from the wort flow meter and the biomass meter, this may be converted into a specific oxygen uptake rate. The specific rate of oxygen uptake may be compared with reference data for that particular combination of yeast strain and wort type. With

Fig. 6.17 System for automatic monitoring and control of wort oxygenation (from Boulton & Quain, 1987).

knowledge of the total volume of wort to be transferred, the controller directs a mass flow meter to dose in the appropriate quantity of oxygen.

This approach has not yet been implemented other than in laboratory simulation trials. It offers several advantages. A single unit may service several vessels, no modifications to fermenters are needed and control is exerted during collection such that there is no delay in starting the fermentation.

6.3.7 Monitoring yeast growth

Monitoring yeast growth is an obvious 'direct' method of assessing fermentation progress and overall performance. In laboratory or small pilot scale fermenters which have mechanically driven agitators, off-line analysis of samples for yeast count or by measurement of dry or wet weight are suitable methods which supply reliable data. In production scale fermenters, heterogeneity of vessel contents makes analysis of offline samples taken from a single point of dubious value. Therefore, it is not usual to monitor yeast concentration during commercial fermentation. It is however, normal practice to assess the magnitude and quality of the yeast crop.

The quantity of yeast transferred to storage vessels gives a rough indication of the extent of growth during the fermentation from which it was taken. With any given fermentation, this should be relatively constant. Abrupt changes, particularly if carried forward into subsequent fermentations, should be taken as evidence of process or raw material changes, which require investigation (see Section 6.5.4). Yeast quality, usually assessed via a viability measurement, provides information of the preceding fermentation only. Thus, if the yeast crop is small and the viability is low it is likely that the fermentation performance would have been less than ideal. Unfortunately, this is unlikely to be a surprising correlation.

Automatic monitoring of yeast concentration during fermentation has been proposed but hampered by a lack of suitable sensors. An automatic sampling system linked to an electronic particle counter (Moll et al., 1978; Lehoel & Moll, 1987). To overcome the problem of heterogeneity in vessels, samples were removed from three locations at various heights in the vessel and analysed in sequence. Output from the counter provided cell number, size distribution and cell size. At present, this information is perhaps of academic interest only since it is unclear how it could be used.

There are no methods currently in use for automatic measurement of yeast concentration in vessel during fermentation. Turbidometric sensing would not be possible and in any case would not distinguish between viable and dead cells. Ultrasonic sensing of suspended solids has been proposed as a method of detecting yeast cells inline (Behrman & Larson, 1987). Unfortunately, gas bubbles interfere and therefore this approach would be unsuitable.

The biomass meter described in Section would perhaps be of use in this regard. It gives an instantaneous measure of viable yeast concentration and it is relatively unaffected by gas bubbles and non-yeast solids. Individual instruments can be multiplexed with several probes, and therefore it would be possible to locate these at various heights within a fermenting vessel and obtain a real-time profile of the distribution of yeast cells throughout the vessel. This could be of utility in the case of large cylindroconical fermenters. It would be possible to identify the point in fermentation when the bulk of the viable yeast was in the cone and cropping could proceed. Early removal of yeast crops is beneficial to beer quality, by minimising the adverse effects due to yeast autolysis. In addition, it benefits crop quality, by removing yeast from the stressful conditions experienced in the depths of fermenters. However, the biomass meters are relatively expensive, more so when multiplexed, and installation into every fermenting vessel is not likely to be a financially viable proposition for the near future.

6.3.8 Monitoring ethanol formation

Measurement of ethanol formation during fermentation provides a direct indication of process efficiency. Monitoring changes in the concentration of this metabolite should be a prime candidate for assessing progress of fermentation as a whole. Ethanol concentration may be measured on line using refractometry (Forrest et al., 1989). Automatic on-line sampling followed by gas chromatographic analysis for ethanol determination has also been proposed (Klopper et al., 1987; Pfisterer et al., 1988). In both of these cases, the intended application was automatic control of ethanol concentration during dilution of high-gravity beers.

Knol et al. (1988) investigated, at pilot scale, monitoring ethanol concentration during primary fermentation using on-line sampling and analysis by either HPLC or gas chromatography. Two sampling devices were tested. First, a membrane device located in a by-pass loop through which volatile components of the wort, including ethanol, diffused into a flow of nitrogen carrier gas and thence to a gas chromatograph for analysis. Second, a membrane sampling device in which liquid was separated from solids. A sample of the clarified liquid injected automatically onto a HPLC analyser. The HPLC approach showed a good correlation with off-line analysis and the liquid sampling and filtration device provided satisfactory performance. The diffusion sampling approach was less satisfactory. Diffusion is temperature dependent and it was necessary to include accurate compensation for this. The gas chromatographic analysis of ethanol showed significant deviation from offline results. It was concluded that this was due to baseline drift and correction required frequent re-calibration of the instrument.

Anderson (1990) described results of preliminary trials in which ethanol was monitored by automated sampling of the fermenter head-space and subsequent gas chromatographic analysis. Using stream-switching 16 fermenters could be monitored simultaneously, each being sampled hourly. The results were fed to a supervisory microprocessor. This provided data logging facilities and the possibility of using the ethanol input in a feed-back control loop linked to the attemperation system. During the first 10 hours of fermentation, a good correlation was demonstrated between ethanol formation and decrease in wort gravity. Over a longer time-scale the correlation became increasingly poor.

Anderson (1990) concluded that, at least for control purposes, a parameter that showed significant change during very early fermentation would be more appropriate and that ethanol did not fit into this category. In the real commercial world, it would be difficult to justify to accountants the cost of on-line chromatographs dedicated to every fermenter. In this respect, use of a single instrument to monitor several fermenters via sampling head-space gases is a much more attractive proposition. In addition, this approach obviates the need for intrusive sensors or the use of external loops.

In addition to measurement of ethanol, having a facility for on-line chromatography offers further interesting possibilities. These techniques are used already for laboratory determination of essential flavour metabolites such as esters, higher alcohols and vicinal diketones. The concentrations of these compounds in beer are much influenced by fermentation performance (see Section 3.7). At present, it is assumed that if the rate and extent of wort attenuation are within specification then so will be the resultant beer. This assumption has to be lived with. However, a control regime based on monitoring the formation of essential flavour components of beer, as well as some aspect of wort attenuation, may well offer new opportunities. For example, it would offer the promise of fermenting at ultrahigh gravity without the unbalanced formation flavour compound. Nevertheless, until more robust and economic sensors, capable of measuring the concentrations of flavour compounds, are developed, such applications at commercial scale remain unlikely.

6.3.9 Monitoring vicinal diketone concentration

For many lager-type fermentations, the process is judged complete when the concentration of the vicinal diketones, diacetyl and its precursor a-acetolactate, have fallen below a pre-determined threshold level. Thus, diacetyl has a strong butterscotch flavour, considered undesirable in some beers. The precursor, a-acetolactate, is derived from pyruvate and is excreted by yeast during primary fermentation. Sub sequently it undergoes spontaneous decomposition to form diacetyl. The diacetyl is then reduced by yeast to the much less flavour-active acetoin and 2,3-butanediol (see Section

Vicinal diketone (VDK) concentrations are monitored during the latter part of primary fermentation. Chilling cannot be applied until a desired maximum threshold VDK concentration. Analyses are usually performed on off-line samples removed from fermenters at regular intervals, typically every 8 hours. VDK concentration is determined using either spectrophotometric or gas chromatographic procedures (EBC Analytica, 1987; American Society of Brewing Chemists, 1992; IOB Recommended Methods of Analysis, 1991). Recently a differential-pulse polarographic method has been proposed as a more accurate substitute for the colorometric approaches (Rodrigues et al., 1997).

Whichever method is used, it is essential that samples are subjected to a heat treatment prior to analysis to ensure that all the precursor a-acetolactate is converted to diacetyl. Thus, the spontaneous decomposition of a-acetolactate to diacetyl is the rate-determining step in the sequence of reactions. Subsequent reduction of diacetyl by yeast is rapid, and therefore in the presence of yeast the concentration of free diacetyl is always low. Should the yeast be removed from the beer prematurely the pool of a-acetolactate may subsequently decompose to diacetyl and persist because further metabolism is not possible. Hence, the requirement to monitor total VDK, as opposed to free diacetyl.

Off-line analysis of vicinal diketones is time-consuming and requires skilled personnel. In most breweries, several hours might elapse between sampling and obtaining a result. Clearly, it would be advantageous if on-line analysis was available. No such methods are currently in use although automatic sampling coupled with a heat treatment and gas chromatography is feasible although complex. An alternative approach has been described by Denk (1997). This author described a method in which the pattern of VDK formation and reduction could be predicted using a software modelling approach based on an artificial neural network (see Section The model builds a simulation of fermentation held in the software of a computer. The underlying premise is that by inputting several sets of real data into a model system the program is capable of undergoing a learning process. Subsequent input of a partial set of data allows output of realistic profiles of missing data. Thus, provided some easily measurable parameters are monitored in early fermentation, the real change in concentration of another parameter, which might be much more difficult to measure, may be predicted with accuracy. The greater the number of sets of data used to develop the model, the greater the precision of the model.

In initial experiments performed at pilot scale (150 litres) input parameters were temperature, pressure, gravity, turbidity, pH and ethanol. Of these, temperature and pressure were both monitored and controlled at set-points; the remainder were simply monitored. In-line data were monitored during the first 15 hours of fermentation after which time the profiles of each parameter, including VDK, could be predicted with accuracy. In later trials, the ethanol input was dispensed with because of the difficulties and cost of off-line measurement of this metabolite. Ultimately it was intended to apply the system to 2500 hi fermenters, with one system servicing several vessels. At the time of writing there have been no reports of this application in brewing at commercial scale.

6.3.10 Miscellaneous

A few methods have been developed which utilise sophisticated optical devices for monitoring closed fermentation vessels. Hosokawa et al. (1999) described a portable camera system which could be used to examine the inner surfaces of large vessels. It consisted of a video camera attached to a controllable boom which, together with a monitor and dedicated lighting system, allowed direct visual examination of all parts of the empty vessel. This was considered essential as a means of checking the efficacy of CiP.

Wasmuht and Weinzart (1999) also recommended the use of a camera located inside the fermenting vessel, in this case to monitor fermentation progress. The system was called TopScan. The camera, together with suitable lighting, was mounted in the top plate of the vessel and directed vertically downwards. A visual image of the surface of the fermenting wort was channelled to a monitor in the control room. This allowed direct viewing of the stages in traditional lager fermentation, such as low and high krausen. In this way, the advantages of closed vessels and ability to judge fermentation progress by direct observation could be combined. As in the previous example, the system also allowed vessel CiP to be assessed.

Excessive foaming in production-scale fermenters can be a problem, which if uncontrolled can result in product loss. Ogane et al. (1999) described an in-tank system which allowed automatic measurement of the foam depth in fermenter. This used a laser-based sensor mounted in the top-plate of the vessel. Output from the sensor was directed to a computer via an RS232 interface, hence providing a detailed record of foaming throughout fermentation. The claimed advantages were that improved control of foaming by early detection of problems allowed vessels to be operated with a minimum free-board.

Was this article helpful?

0 0
Brew Your Own Beer

Brew Your Own Beer

Discover How To Become Your Own Brew Master, With Brew Your Own Beer. It takes more than a recipe to make a great beer. Just using the right ingredients doesn't mean your beer will taste like it was meant to. Most of the time it’s the way a beer is made and served that makes it either an exceptional beer or one that gets dumped into the nearest flower pot.

Get My Free Ebook

Post a comment