niveus novobiocin fermentation and demonstrated that high dissolved oxygen levels (60-80%) occurred in oxygen-limited cultures. It was concluded that, although oxygen was being transferred into solution, the dissolved gas was not reaching a large proportion of the biomass. Thus, as well as KLa being affected adversely by a high viscocity broth, efficient mixing also becomes extremely important in these systems. These workers also demonstrated that, at constant power input, small impellers were superior to large impellers in transferring oxygen from the gas phase to the microbial cells. Wang and Fewkes (1977) confirmed Steel and Maxon's work by demonstrating that the critical dissolved oxygen concentration (Ccrit) for S. niveus in a fermentation varied depending on the degree of agitation and the size of the impeller. Remember that Ccrit is the dissolved oxygen concentration below which oxygen uptake is limited, i.e. it is a physiological characteristic of the organism. It was concluded that the limiting factor was the diffusion of oxygen to the cell surface through a dense mycelial mass. At higher agitation rates biomass within clumps would be receiving oxygen and would thus contribute to the measured respiration rate whereas at low agitation rates such mycelium would be oxygen limited, i.e the heterogeneity of the system increased at low agitation rates. Wang and Fewkes examined their results in terms of the impeller's ability to produce turbulent shear stress (oxygen transfer into solution) and pumping power (mixing). Turbulent shear stress is proportional to N2D2 and impeller pumping power is proportional to A©3 (where N is the impeller rotational speed and D is the impeller diame-
Fig. 9.18. The effect of shear to flow ratio on the observed critical oxygen concentration of S. niveus (Wang and Fewkes, 1977).
ter). Thus, the ratio of impeller turbulent shear stress to impeller pumping is proportional to:
It was demonstrated that the observed critical dis-solved-oxygen concentration decreased exponentially as the shear stress to pumping ratio increased, over the range 0.2 to 1.0 (cm sec)"1, as shown in Fig. 9.18. Thus, an increase in the ratio of impeller shear stress to impeller pumping decreases the transport resistance of oxygen to the cell surface resulting in a lower dissolved oxygen concentration maintaining a higher respiration rate. Wang and Fewkes' analysis quantifies Steel and Maxon's observation that smaller impellers gave better oxygen transfer to the cells of S. niveus, in that the smaller impeller would have a larger shear stress to impeller pumping power ratio.
Wang and Fewkes' correlations are particularly relevant when it is considered that many mycelial broths are pseudoplastic. The viscosity of a pseudoplastic broth will decrease with increasing shear stress so that viscosity increases with increasing distance from the agitator. Air introduced into the fermenter tends to rise through the vessel by the route of least resistance, that is, through the well-stirred, less viscous central zone. Thus, stagnant zones, receiving little oxygen, may occur in the vessel. Therefore, it is essential that the agitation regime employed creates the correct balance of turbu-
Fig. 9.18. The effect of shear to flow ratio on the observed critical oxygen concentration of S. niveus (Wang and Fewkes, 1977).
leu«; (and hence the transfer of oxygen into solution) and pumping power (mixing) to circulate the broth through the region of high shear.
The quantification of the problem of oxygen transfer and mixing is also considered by Van't Riet and Van Sonsberg (1992) in the context of the critical time for mass transfer. It is assumed that oxygen transfer into solution in a stirred, aerated reactor takes place only in the stirrer region. If one considers an aliquot of aerated broth leaving the agitator zone, it will be circulated through the vessel and eventually return to the agitator. The dissolved oxygen imparted to the broth should sustain the respiration of the organisms in that aliquot during the circulation. The time it takes for the oxygen in the aliquot to be exhausted will be:
where tcro is the time for oxygen to be exhausted, CL(ag) is the dissolved oxygen concentration in the zone of the agitator, OUR is the oxygen uptake rate. If the circulation time for the vessel exceeds tcro then oxygen starvation will occur in the aliquot before it returns to the agitator. To prevent this occurring the dissolved oxygen concentration at the agitator should be high, but this would reduce the driving force of oxygen into solution and the oxygen transfer rate would decrease. The alternative approach is to achieve the balance of mass transfer and pumping power (broth circulation) already discussed.
As discussed in Chapter 7 the most widely used fermenter agitator is a disc turbine (Rushton turbine). Van't Riet (1979) and Chapman et al. (1983) demonstrated that for non-viscous broths the KLa is dependent only on the power dissipated in the vessel and is independent of impeller type (at least those impellers included in the study). However, it is obvious from the foregoing discussion that impeller type is particularly relevant for viscous, non-Newtonian fermentations and this realization has resulted in the development of a range of agitators which address the dual problems of oxygen transfer and mixing in viscous fermentations.
Legrys and Solomons (1977) approached the problem of combining adequate pumping power and mass transfer in mycelial fermentations by using two impellers, a bottom-mounted disc turbine and a top-mounted curled-blade (hydrofoil) impeller. The bottom turbine produced a high degree of turbulence and radial mixing while the top-mounted impeller produced axial mixing with a high flow velocity, resulting in the circulation of one tank volume in 20-30 seconds. Thus, the mycelium was re-circulated through the oxygenation zone of the vessel before it became oxygen limited. Cooke et al. (1988) extended Legrys and Solomon's approach using a combination of radial flow and axial flow agitators in a 60-dm3 fermenter intended for non-Newtonian fermentations. The radial flow agitator was an ICI Gasfoil which is similar to the Scaba SRGT illustrated in Fig. 9.19, being a disc turbine with concave blades. However, in this case the combination was not successful due to minimal fluid movement at the vessel walls which would have created significant cooling problems.
Gbewonyo et al. (1986) evaluated the performance of a hydrofoil impeller, the Prochem Maxflo (Fig. 9.19) in the avermectin fermentation employing Streptomyces
Fig. 9.19. Agitators used in filamentous fermentations:
(a) Scaba agitator;
Fig. 9.19. Agitators used in filamentous fermentations:
(a) Scaba agitator;
avermitilis in a 600-dm3 working volume vessel. The avermectin process is challenging because the broth is extremely viscous, the fermentation requires fairly high oxygen-transfer rates and the situation is complicated by the shear sensitivity of the mycelium. The results of. this investigation may be summarized as follows:
(i) The impeller pumped the broth axially, that is from the top to the bottom of the fermenter, which is very different from the Rushton turbine which pumped radially, outwards from the agitator.
(ii) The Prochem agitator supported a significantly higher oxygen uptake rate than did the Rush-ton turbine.
(iii) The power number of the Prochem was 1.1 compared with 6.5 for the Rushton turbine. The equation (9.16) for power number was given previously as:
where Np is the power number,
P is the external power from the agitator, p is the liquid density, N is the impeller rotational speed, D is the impeller diameter. Thus, a low power number indicates a low power draw and, hence, the Prochem agitator drew significantly less power than did the Rushton turbine, making the former far more economical to operate. This observation is strengthened by Nienow's work on a similar hydrofoil impeller, the Lightnin' A315, which gave a power number of 0.75 compared with 5.2 for a Rushton turbine.
(iv) The relationships between KLa and power consumption per unit volume at a viscosity of 700 cp were as follows:
Rushton KLa = 51 (P/V)05S, Prochem KLa = 129(P/V)039.
These figures reinforce the previous point, demonstrating that the power requirement for the Prochem agitator is approximately 50% of that for the Rushton.
(v) Raising the power of the Prochem had a greater effect on oxygen transfer at high viscosity than it did at low viscosity. This points to the key role that bulk mixing plays in a viscous fermentation and suggests that it is at least as impor tant as bubble breakup (at which the Prochem is mediocre).
(vi) Unlike a Rushton turbine, the Prochem agitator did not generate high shear forces, which is advantageous for a shear sensitive organism.
(vii) The avermectin yields were slightly better in the Prochem fermenter, but these were achieved with approximately 40% less power consumption.
The same group (Buckland et al, 1988, 1989) performed similar experiments on viscous fungal fermentations in 800-dm3 and 19-m3 vesssels and came to the same basic conclusions that bulk mixing is extremely important in viscous fermentations and that an axial flow hydrofoil impeller results in lower power costs. Data generated from non-Newtonian polysaccharide fermentations using hydrofoil impellers are considered in a subsequent section of this chapter.
(ii) The manipulation of mycelial morphology
The previous section considered engineering solutions to the problem of oxygen transfer in mycelial fermentations. However, this is not the only approach to improve oxygen transfer in such processes; it is possible to modify the morphology of the process organism. As discussed in Chapter 6, the biomass of mycelial organisms grown in submerged culture may vary from the filamentous type, in which the hyphae form a homogeneous suspension dispersed through the medium, to the 'pellet' type consisting of compact, discrete masses of hyphae. The filamentous form tends to give rise to a highly viscous, non-Newtonian broth whereas the pellet form tends to produce an essentially Newtonian system with a much lower viscosity making oxygen transfer much easier. Buckland (1993) reported that the K, a attained in the lovastatin Aspergillus terreus fermentation was 20 h"1 with a filamentous culture and 80 fC1 with a pelleted one at the same power input. Not all pelleted cultures are Newtonian: Metz et al. (1979) demonstrated that pellet suspensions could be non-Newtonian but confirmed that they did give rise to low viscosity broths. Also, it should be appreciated that the terms 'filamentous' and 'pelleted' each describe a range of morphology and the form of filamentous or pelleted growth may be affected by both the genetic makeup of the organism and the environment. Thus, the morphological form of a mycelial organism in submerged culture has a major effect on the broth rheology and may, therefore, be expected to influence aeration efficiency.
Carilli et al.'s work (1961) provides a good example of both the effect of different filamentous form on process performance and the behaviour of filamentous and pelleted cultures in 3000-dm3 fermenters. Two strains of P. chrysogenum were employed, one which grew as short, highly branched hyphae and the other as long, relatively unbranched hyphae. The short, branched hyphae gave rise to a relatively low viscosity broth in which the oxygen transfer rate was approximately twice that achieved with the more viscous broth of the unbranched form. By manipulating the cultural conditions of A. niger, Carilli et al. were able to produce the fungus in either filamentous or pellet form and demonstrated that the pellet form gave rise to a broth exhibiting half the viscosity of the filamentous broth. Also, oxygen limitation occurred far earlier in the fermentation when the organism grew in the filamentous form.
Although the pellet type of growth tends to produce a low viscosity Newtonian broth in which turbulent flow conditions may be achieved, it may also give rise to problems of oxygen availability if the pellets become too large. A large pellet may be so compact that its centre may be unaffected by the turbulent forces occurring in the bulk of the fermentation broth so that the passage of oxygen within the pellet is dependent on simple diffusion; this may result in the centre of the pellet being oxygen limited. Thus, to maintain the intra-pellet oxygen concentration at an adequate level it would be necessary to maintain a high dissolved oxygen concentration to ensure an effective diffusion gradient. A similar situation was described by Steel and Maxon (1966) and Wang and Fewkes (1977). Kobayashi et al. (1973) demonstrated this phenomenon in pellets of A. niger where large pellets required a higher dissolved oxygen concentration to maintain the same specific oxygen-uptake rate as smaller pellets. If oxygen limitation does occur within a pellet then only its outer layer would contribute to its growth and the centre may autolyse. The diffusion of oxygen into the centre of a pellet will be influenced by the size of the pellet, and thus it is important to control pellet size.
Schugerl et al. (1988) monitored the dissolved oxygen concentration within pellets of P. chrysogenum and demonstrated that, provided they were smaller than 400 fim in diameter, the oxygen concentration in the centre of the pellet was not limiting. Similarly, Buckland (1993) reported that pellets of Aspergillus terreus in the lovastatin fermentation had to be smaller than 180 /xm in diameter to avoid oxygen limitation in the centre of the pellet. It should be appreciated that the pellet sizes recommended by both groups are very small and it is possible to obtain fungal pellets which are at least 1 cm in diameter. Pellet size may be influenced by the inoculum, the medium and the cultural conditions. As discussed in Chapter 6, pellet size is reduced at high spore inoculum concentrations, but it is unlikely that this alone would produce pellets of less than 400 /¿m in diameter. Schugerl et al. (1988) controlled the pellet size of the inoculum by physical means by either incorporating glass beads in inoculum shake flasks or using high agitator speeds in seed fermenters. Metz and Kossen (1977) also claimed that, once pellets are formed, strong agitation tends to give rise to smaller, more compact pellets. Buckland (1993) reported that the conditions of the lovastatin fermentation are carefully controlled to maintain the optimum pellet size but the cultural conditions used to achieve this end were not revealed.
Righelato (1979) discussed the effects of mycelium morphology on culture rheology and oxygen transfer and came to the conclusion that the most desirable way for a mycelium to grow in submerged culture is in the form of short, hyphal fragments which would produce a broth less susceptible to diffusion limitation than a pelleted one, and less viscous than one containing long filaments. However, attempts to encourage the formation of the desirable short hyphal fragment morphology (as compared with the long filaments) by increasing the shear stress on the mycelium has met with only limited success. Even if a less viscous broth is obtained the damage done to the mycelium may well be counterproductive. Dion et al. (1954) showed that the morphology of P. chrysogenum was influenced by the degree of agitation in that short, branched mycelium was produced at high agitation rates compared with long hyphae produced at low agitation rates. Lilly et al. (1992) extended this observation at 10 dm3 and 100 dm3 scales and related the mean main hyphal length and the penicillin specific production rate (qp) to the term P/Df tc, where P is the agitator power, l)l is the impeller diameter and tc is the calculated circulation time. This term is a measure of the maximum shear stress due to agitator power dissipation and the frequency with which mycelia pass through the high shear region. Both mean hyphal length and q decreased with increasing P/Df tc, implying that increased shear is disadvantageous at this scale. At 1000 dm3 it was not possible to introduce enough power into the fermenter to decrease qp, which suggests that it is very difficult to disrupt the mycelium at this scale, thus confirming Van Suijdam and Metz's (1981) observation that an enormous amount of energy is required to reduce the hyphal length of P. chrysogenum. Righelato (1979) also claimed that it is unlikely that shear forces could ac count for the break up of mycelia and that autolysis and lysis of some hyphal compartments may be more important controlling factors, perhaps implying that the phenomenon may be more under genetic, rather than physical, control. This leads us on to strain improvement of morphologically favourable strains, as discussed in more detail in Chapter 3. However, Belmar-Beiny and Thomas (1990) demonstrated in 9-dm3 fermenters that increased stirrer speed did result in the production of shorter, less branched hyphal fragments of Streptomyces clavuligerus and clavulanic acid synthesis was unaffected. This suggests that this approach may be used to influence rheological properties in clavulanic acid fermentations.
Other cultural conditions which have been claimed to influence mycelial morphology include medium composition (see Chapter 5), growth rate, dissolved oxygen concentration, polymer additives and temperature. Kuenzi (1978) reported that the viscosity of a Cephalo-sporium broth was considerably reduced by growing the organism at 27° rather than 25°C. Olsvik and Kristian-sen (1992) investigated the influence of specific growth rate and dissolved oxygen concentration on the viscosity of Aspergillus niger in continuous culture. K, the consistency index (indicative of apparent viscosity, see equation (9.9)) was measured over a range of conditions. At dissolved oxygen (DO) concentrations above 10% saturation, K increased with increasing dilution rate whereas at DO concentrations below 10% saturation, K decreased with increasing dilution rate. The effect of DO on K was particularly evident at low DO values and at low growth rates where a 2% change in the DO could give a 25% change in K. These observations may be particularly relevant in the late stages of batch or fed-batch processes where low growth rates, nutrient limitation, high biomass levels and low oxygen concentrations occur, all contributing to complex changes in morphology, viscosity and oxygen transfer rate.
Dispersed growth can be encouraged in certain organisms by incorporating polymeric compounds into the medium. Such anionic polymers include Junlon PW110 and Junlon 111 (cross-linked polyacrylic acids) and Carbopol-934 (carboxypolymethylene). It is claimed that these polymers modify the electrical charges on the spore surface and thus prevent the aggregation of spores into clumps, thus preventing the initiation of pellet formation. These agents have been used to in-cease the homogeneity of both fungal (Trinci, 1983) and streptomycete (Hobbs et al., 1989) broths. Although these agents would not be practical to use on a large scale they may be useful in the early stages of an inoculum development programme if a dispersed morphology is desirable.
Several workers have discussed the possible advantages of reducing the viscosity of a mycelial fermentation, in its later stages, by diluting the broth with either water or fresh medium. Sato (1961) increased the yield of a kanamycin fermentation, displaying Bingham plastic rheology, by 20% by diluting the broth 5% by volume with sterile water. Taguchi (1971) achieved a 50% reduction in the viscosity of an Etidotnyces broth by diluting 10% with water or fresh medium. A scheme has been put forward for the control of viscosity and dissolved oxygen concentration in a hypothetical fermentation. These workers proposed that, as the critical dissolved oxygen concentration is approached, a set volume of broth could be removed from the fermenter and replaced with fresh medium. The process could be repeated in a step-wise manner as the system became oxygen limited, which could be determined by dissolved oxygen concentration or viscosity measurements. Thus, by using such techniques the viscosity may be controlled and maintained below the level which may cause oxygen limitation. Kuenzi (1978) reported an instance where the very slow feeding of medium to a Cephalosporium culture resulted in the organism growing in the form of long filaments which produced a highly viscous culture which could not be adequately aerated. The design of fed-batch processes such that efficient control may be achieved over the process is discussed in a subsequent section of this chapter and in Chapter 2.
The production of Fusarium graminae biomass for human food in the ICI-RHM mycoprotein (Quorn®) fermentation (see Chapter 1) presents a very different problem from those of most other fungal fermentations. It is essential that the organism grows as long hyphae so that the biomass can be processed into a textured food product. Long hyphae are susceptible to shear forces, so to maintain the morphological form of the organism an air-lift reactor is used, despite the fact that the viscous broth severely limits the attainable oxygen transfer rate. This limitation of the air-lift fermenter means that only a relatively low biomass concentration may be maintained in the vessel compared with that in a stirred system, but this is an acceptable penalty to pay for the correct morphological form.
Was this article helpful?