2.4. INCIDENCIA DEL COMPORTAMIENTO DEL PÚBLICO EN LA OFERTA
2.4.2. Variables determinantes del Coeficiente de Preferencia por Efectivo
In terms of allowing the viscosity of one of the processes to effectively mimic that of the other, the addition of viscosity altering compounds can be considered successful
Substantially altering the viscosity of fermentation broth does, however, cause concerns. This is particularly so when increasing broth viscosity. High broth viscosity, especially for broth exhibiting non-Newtonian rheological characteristics, has negative implications for transport processes within the fermentation, such as heat, mass and momentum transfer (Olsvik and Kristiansen, 1992). In aerobic fermentations, the most significant mass transfer process is that of oxygen fi*om the gas bubbles into the aqueous phase. Due to the low solubility of oxygen in water, the transfer of oxygen to the cells can become limited, particularly at elevated viscosity (see Section 1.7.). Indeed, several researchers have reported a direct relationship between increased viscosity leading to lower mass transfer of oxygen (McNeil and Harvey, 1993). McNeil and Harvey (1993) also
comment that in a stirred system the mass transfer of oxygen is likely to vary depending on the position in the reactor, with the level of transfer falling as the distance fi*om the impeller increases. Similarly, the concentration of dissolved oxygen will also vary across the fermenter. As a consequence, although DOT limitation may not be apparent at the position of a DO probe, it is possible that certain regions of a fermenter can suffer periods of DOT limitation.
The effect of decreased DOT at higher viscosity was observed with the M4018 process in this study. The fermentations with increased viscosity had an average minimum DOT of 20 %; this was approximately half of that observed in the control culture. This is of concern because, in addition to possible DO limitation, experiments that have a profound affect on the DOT of a fermenter, are often criticised for an inability to distinguish the effects due to the experimental change and those due to changes in the DOT (Large et al., 1998). The results of increasing broth viscosity could therefore be questioned. However, with little effect on either OUR or product formation (Figures 5.7., 5.8., & 5.9.) it was very unlikely that this lowering o f DOT had any significant effect on the fermentation. Furthermore, due to the level o f agitation and the scale of operation used, it was unlikely that stagnant, oxygen depleted, zones were numerous within the
fermentation. It is, therefore, not unreasonable to suggest that any effects observed in the fermentation could be attributed to changes in the broth viscosity rather than any
consequential effects of reduced DOT. To eliminate possible effects of consequential changes in DOT, an improvement to the experimental protocol would be to maintain a minimum DOT by gas mixing and oxygen enrichment; unfortunately these facilities were unavailable when the experiments were performed.
In reducing the viscosity of the M38182 process, the problem of reduced DOT was not an issue. The higher viscosity of the M3 8182 process when compared to the M4018 is primarily due to a change in the interaction of the biomass with the media components (Chapter 3.). The zeolite type compound added to the M38182 fermentation reduces viscosity by entrapping the media components that cause the increase in viscosity. As with the M4018 process, lowering the viscosity of the M3 8182 process had little effect on OUR or antibiotic production (Figures 5.2. and 5.3.).
Regarding the utilisation of oil in the two processes. In particulate streptomycete broth, oil has been observed to exist as droplets (Ohta et al., 1995; Mitsuru et al., 1997). If it is also assumed that the oil was in the form of droplets in these processes, the degree of oil utilisation can be related to the size of the oil droplets. Chatzi et ût/. (1991) state that in liquid-liquid dispersions the physical and chemical phenomena taking place in a vessel largely depends upon the size of the dispersed droplets. In the case of mycelial fermentations, Ohta et al. (1995) state that in general, the consumption rate of a hydrophobic substrate (such as oil) by mycelia is dependent on the specific substrate- mycelial interface area. Indeed, if the residual oil problem is due to a physical mass transfer limitation, by decreasing oil droplet size, the increase in surface area to volume ratio may increase the accessibility of the oil to the organism, thereby decreasing the residual oil concentration. It is for this reason that many o f the approaches to decreasing oil residual levels in streptomycete fermentation have centred on methods of decreasing the oil droplet size. The droplet size will, to a certain extent, depend on the turbulence within the stirred vessel.
Stirred fermentation vessels are generally under conditions of turbulent flow. Under such conditions, the kinetic energy supplied by agitation of the fluid is dissipated into the vessel via liquid eddies (see Section 1.6.). During turbulent flow numerous eddies of a variety of sizes co-exist within the broth. The smallest of these eddies determine the degree of homogeneity in a system being mixed. In a bioreactor, dispersion of the media components into eddies of decreasing size will allow rapid transport of these components
C h apter 5: Effect o f Viscosity on O il U tilisation in S. rim osus Ferm entation
throughout the vessel. For example in the context of this work, if oil droplets have a diameter greater than that of a particular eddy, they are likely to be broken into smaller droplets, since the interfacial tension forces holding the drop together are likely to be exceeded by the deforming fluid dynamic forces. This process will continue until the oil droplets have a diameter similar to or less than that of the smallest eddies. In summary, for oil droplets under the influence of a turbulent flow field, their size is determined by that of the smallest eddies. The size of the smallest, or terminal, eddies in the vessel is estimated using the microscale of turbulence. The microscale of turbulence can be estimated using Equation 1.1. The microscale of turbulence is dependent on the physical properties of the liquid being mixed and the energy dissipation in the liquid. Energy dissipation is related to power input per unit volume, and is typically calculated using the whole liquid volume. However, recent work suggests that the estimation of Kolmogorov microscale is often inaccurate when the energy dissipation is calculated over large
volumes (Aloi and Cherry, 1996). These workers therefore estimated energy dissipation using only the liquid volume of the highly turbulent impeller region, and in particular the trailing vortices. The rational for this being that a disproportionate amount of energy was dissipated in this region. Zhou and Kresta (1998), comment that drop breakup mainly occurs in the strongest turbulence energy dissipation region, and it is this that dominates the deformation of droplets. Energy dissipation in this work was, therefore, calculated using the Aloi and Cherry method. An example of the estimation of terminal eddy size is shown in Appendix III.
If the equation for the calculation of the microscale of turbulence is considered, it is found that the kinematic viscosity function is highly significant due to the fact that it is to the power of three. The difference in the viscosity of the two processes under
investigation would therefore have a significant effect on the size of the terminal eddies vsdthin each fermentation broth. Indeed, if the size of the terminal eddies in the control processes is estimated, the maximum terminal eddy diameter within the M4018 process was approximately a third of that of the M38182 process (Figure 5.11.). For oil droplets under the influence of turbulent forces this would have an affect on their diameter and thus utilisation. Upon alteration of the broth viscosity as described in Sections 5.1. and 5.2., it was possible to minimise the difference in terminal eddy diameter between the two processes (Figures 5.12. and 5.13.). This, however, had little effect on residual oil
Utilisation. Indeed, the final residual oil level was found to be independent of the viscosity of the process (see for example Figure 5.10.).
2000 1800 1600 1400 xg 1200 3 1000 ^ 800 600 400 200 0 20 40 60 80 100 120 140 160 180 200 220 240 Time (h)
Figure 5.11. Predicted terminal eddy diameter (A,) throughout M4018 (lEl) and M38182 (■) control fermentations. Viscosity data taken from Figure 5.1. X calculated according to Appendix III.
I
100 120 140 160 180 200 220 240
Time (h)
Figure 5.12. Predicted terminal eddy diameter (A.) throughout control
fermentations of M4018 (13), M38182 (■) and M38182 fermentation containing 2.5 gL'* zeolite: duplicate fermentations “a” (—O —) and “b” (—A—). Viscosity data taken from Figure 5.1. A, calculated according to Appendix III.
C h apter 5: E jfect o f Viscosity on O il U tilisation in S. rim osus Ferm entation 2000 1800 1600
>
1400 - I 1200 - 1000A 800 600 400 200 0 20 40 60 80 100 120 140 160 180 200 220 240 Time (h)Figure 5.13. Predicted terminal eddy diameter (A.) throughout control fermentations of M4018 (lEI), M38182 (■) and M4018 fermentations with Thixagum additions at: 0.75 % [w/AIv] (—# —), 1.0 % [w/AIv] (--A--) and 1.0 % [w/AIv] “shot” (—♦ —). Viscosity data taken from Figure 5.6. X calculated according to Appendix III
If the oil droplets have a diameter much less than the terminal eddy size, the viscosity of the broth may not affect the oil droplet size and, thus, residual oil utilisation. Since, in such a case the oil droplets would not be subject to the forces imparted by the turbulent eddies. Mitsuru et al. (1997), present photographic evidence of oil droplets in particulate Streptomycete fermentation broth. Although they present no extensive data regarding oil droplet size, the droplets within this photograph were stained and measure 5 pm or less. By applying the Kolmogorov microscale to their data, the terminal eddy size was estimated to be around 250 pm. This suggested that in this process the oil droplet diameter was influenced by mechanisms other than the turbulent flow field. The Mitusru process has many similarities to the processes under investigation here, so it is reasonable to suggest that the oil droplets in both the M4018 and M38182 processes were also considerably smaller than the microscale of turbulence. The size of the oil droplets, the distribution of oil, and whether the oil is indeed present as droplets is investigated in Chapter 6.
The difference in the residual oil utilisation o f the two processes was therefore unlikely to be a consequence of the variation in broth viscosity. This is supported by evidence
from investigations exchanging strains and process conditions (Section 3.3.) In these investigations it was found that the utilisation, and final concentration, of residual oil was very similar irrespective of the process conditions. This suggests that the difference in the residual oil utilisation observed between the two processes was more dependent on the S. rimosus strain than process conditions. It is therefore, possible that through the mutation of this organism, a change in the microbial physiology has occurred that has attenuated the ability of the M38182 strain to utilise residual oil.