Control of anaerobic digestion is crucial in order to secure or even to maximise the performance of the process. In order to develop a control scheme the following steps should be considered:

• Definition of the control objective: The objective could be as simple as the pH stabilisation or more complicated involving stabilisation and optimisation of the bioreactor operation in terms of biogas production or chemical oxygen demand removal. Since optimisation and stabilisation are conflicting objectives, the control law should be sophisticated enough to meet these targets in the best way.

• Selection of the suitable measurements: The properties of a suitable measurement to be used in a control scheme are the ability to reflect the process state and its changes due to disturbances (sensitivity), as well as the time response and the simplicity of the measurement method. The most common measurements in anaerobic digesters (Table 12.2) are:

— Biogas flow: The biogas production rate and especially the performance in methane is the most commonly used measurement to detect the process stability. A reduction in the biogas production rate usually suggests that the volatile fatty acids have been accumulated as a result of overloading or presence of a toxicant. However, any change in this parameter is caused by process instability and cannot be an early warning, that is, it is not sensitive enough.

— Biogas composition: The principle gases in the headspace of an anaerobic digester are CO2 and CH4. When CO2 increases relatively in proportion to CH4, process imbalance has already evolved and, consequently, this index cannot be used as an early indicator. On the other hand, CO2 in the gas phase is influenced by changes in alkalinity and pH in the bioreactor, and as a result when pH control is applied in low buffered systems, changes in its value do not reflect process instability (Ryhiner et al., 1992). Another important gas found at very low concentrations is hydrogen. Hydrogen has been suggested as an early reliable measurement for early detection of an imminent imbalance (Archer et al., 1986, Molina et al., 2009). Hydrogen is a significant intermediate compound regulating the performance of the acetogens. Accumulation of hydrogen entails accumulation of volatile fatty acids due to thermodynamic limitations of acetogenesis. It should be kept lower than 40 nM (which corresponds to a partial pressure less than 6 Pa at 35°C). Archer et al., (1986) monitoring hydrogen partial pressure in the headspace predicted an accumulation of volatile fatty acids 3-6 h before it happened. However, the changes in the hydrogen concentration cannot be correlated necessarily with imbalance (Guwy et al., 1997). Measuring hydrogen in the headspace does not correspond to the actual concentration sensed by the microorganisms which are in the aqueous phase. This is why measurement of dissolved hydrogen is suggested as a more reliable index (Pauss et al., 1990; Frigon and Guiot, 1995). Hydrogen sulphide and carbon monoxide can also be detected but they are not important for control purposes.

— Volatile fatty acids: They are the most important intermediate compounds in anaerobic digestion since their accumulation leads to pH decrease, stressing the methanogens further. The increase in acetate concentration under overload conditions does not indicate necessarily process imbalance if the biogas production rate has also increased. In this case, the system may operate at a higher acetate concentration at a new steady state, without rejecting the possibility of process failure. However, propionate and butyrate accumulation denote signs of imbalance since it usually happens when hydrogen concentration increases. Propionate is accumulated first, since its conversion requires six times lower concentration of hydrogen than butyrate (Ozturk, 1991). Therefore propionate has been suggested as a suitable indicator for process imbalance (Pullammanappallil et al., 1998; Boe et al., 2008), along with butyrate (Renard et al., 1991), the ratio of propionate to butyrate (Hill, 1982), and the iso forms of butyrate and valerate (Cobb and Hill, 1991; Ahring et al., 1995). Depending on the metabolic pathways prevailing in an anerobic bioreactor, volatile fatty acids may be formed at various concentrations and there cannot be a rule of thumb for a ‘safe’ level of volatile fatty acids securing stable operation. For example, Pullammanappallil et al., (2001) found that operation of a controlled, glucose fed bioreactor in the presence of phenol remained stable at a high propionate concentration (2750 mg/L). Moreover, the inhibition of volatile fatty acids is pH dependant and their inhibitory effect increases at pH values ranging from 6 to 7.5.

— pH: Monitoring pH is very important since it affects the microorganisms activity and can be correlated with changes in acids and bases as well as anions and cations produced or consumed as a result of the metabolic activity. However, it cannot be used to evaluate the state of the system since it is affected by the buffer capacity of the liquid (determined mostly by the bicarbonate, ammonia, volatile fatty acids).

— Alkalinity: It is distinguished in total and bicarbonate alkalinity. Total alkalinity is measured through titration to pH 3.7 (Powel and Archer, 1989) and expresses the capacity of an anaerobic system to maintain the pH under acidification. However, total alkalinity increases as the volatile fatty acid concentration increases. Therefore, the bicarbonate alkalinity, measured through titration to 5.75, can reflect the effective buffer capacity of the system. Various methods have been developed for the on-line measurement of the bicarbonate alkalinity (Table 12.2).

— Organic matter: Common parameters such as the total and volatile solids, chemical oxygen demand, total organic matter and biochemical methane potential (preferable to biological oxygen demand in the case of anaerobic systems) express the aggregate organic matter present in a digester and, correlated with the organic matter of the influent, give an accurate estimate of the organic matter removal. However, these are time consuming, off­line measurements, except from the total organic carbon method which can be applied on-line in the case of anaerobic systems with low

solid content (Table 12.2). Therefore, they are not suitable for on-line controllers.

— Metabolic activity: The physicochemical parameters available for measurement respond to changes in the metabolic activity of the anaerobic microorganisms, but the correlation is not always direct. Since the success of a control scheme applied on anaerobic systems is based on directing the microbial activity to the desired performance, its assessment is very important. The microbial activity can be evaluated through measurement of the specific methanogenic activity (Ince et al., 1995; Garcia-Morales et al., 1996; Fountoulakis et al., 2004; Dong et al., 2009; Montero et al., 2009), application of molecular techniques (for the qualitative and quantitative detection of specific microorganisms based on the DNA and RNA probing (Macario and de Macario, 1993; Macario et al., 1989; Raskin et al., 1994; Montero et al., 2009) and detection of changes in cellular components such as enzymes (NADH and coenzyme F420) (Perk and Chynoweth, 1991; Amann et al., 1998), ATP (Chung and Neethling, 1988) and phospholipid fatty acids (Nordberg et al., 2000). Moreover measurement of the activity of certain enzymes and application of microcalorimetry (heat released in an anaerobic ecosystem which can be correlated to the size of the microbial population, the metabolic state and activity) have also been used for monitoring (Switzenbaum et al., 1990). Since most of the analytical procedures required for assessing the metabolic activity are elaborate and time consuming or require samples of low solid content, the utilisation of these measurements is limited for on-line control, but can be used off-line to give a better insight of the system status.

• Manipulated variables: The manipulated variables are operating parameters through which the process state can be affected and led to the satisfaction of the control objective according to the applied control law.

The most common manipulated variable is the dilution rate, or equivalently, the hydraulic retention time (inverse of the dilution rate). The dilution rate should generally be lower than the maximum specific growth rate constant of the slowest growing microorganisms group to avoid wash out in a continuously stirred tank reactor. In such type of bioreactor, the sludge (solids) retention time coincides with the hydraulic retention time. In order to increase the conversion rate, recirculation of the sludge is often applied to increase the biomass concentration. In systems fed with waste of high solid content, the liquid effluent stream is recirculated to provide it with nutrients and microorganisms. In both cases, the hydraulic and sludge retention times are separated and can be manipulated independently. The extent of manipulation of the hydraulic retention time is restricted in practice given the waste storage capacity of the treatment plants (a few hours to a few days). The hydraulic retention time in thermophilic conditions can be as low as 4-6 d, while in mesophilic conditions it is 10-15 d, although higher values of the hydraulic retention time result in more stable operation (Pind et al, 2001).

The organic loading rate, influenced by the organic content of the waste at a given hydraulic retention time, is another manipulated parameter, but since the organic content of the waste does not vary, its use is rather restricted.

In the case of more than one waste stream being commonly digested (codigestion), the composition of the waste mixture is another manipulated variable. In codigestion, wastes can be combined to make up for nutrient deficiencies, dilute the inhibitory compounds of waste stream, enhance the process yield of low potential waste (Angelidaki and Ahring 1997; Gavala et al., 1999; Angelidaki and Ellegaard 2003; Alatriste-Mondragon et al., 2006; Nielsen and Angelidaki 2008; Dareioti et al, 2009; Li et al., 2009; Shanmugam and Horan, 2009).

Other manipulated variables are the acid, base or bicarbonate addition rates to control the pH or alkalinity in the bioreactor or the feed (Pind et al., 2001). pH and alkalinity control require the addition of chemicals, which raise the cost of the process. An alternative is to recycle the CO2 produced in order to increase the alkalinity, but this is not effective in case the bioreactor pH is lower than 6.5 (Romli et al., 1994).

• The control law: It is the information flow structure through which the manipulated variables are handled based on the measurements. The complexity of the control law is determined by the diversity of the control objective. As a result, the controller can be simple (on-off, proportional, proportional-integrated — differential), more complicated adaptive model-based, empirical (expert systems), fuzzy or neural network-based. Detailed references on the various control systems having applied on anaerobic digesters can be found in Pind et al. (2003), Liu (2003) and Boe (2006).