Solid-state fermentation (SSF) involves the growth of microorganisms on moist solids in the absence of free water. Unlike submerged liquid fermentation (SLF), SSF is not used extensively at large scales, despite potential advantages of SSF over SLF for the production of some biotechnology products. This is because quantitative rules for the design, operation and scale-up of SSF are currently lacking. The development of such rules has been hampered by difficulties in studying SSF systems and by the poor understanding of the effects of bioreactor design and operation on SSF performance. Several bioreactor types have been used for SSF at small scales. One that has potential at large scales is the rotating drum bioreactor (ROB).
This thesis investigates the operation of small-scale ROBs for SSF with the filamentous fungus, Aspergillus oryzae. It aims to improve understanding of the effects of operational variables on SSF performance, and thereby provide a basis for further research into the design and operation of ROBs for SSF at large scales.
The work presented in this thesis consisted of several steps:
1. Development of experimental systems. Two identical, glass 18.7 ℓ RDBs were constructed with length to diameter ratios (L/D) of 4.4. The ends were fined with metal caps which permitted gas flow from one end of the bioreactor to the other. The bioreactors were able to be rotated at a wide range of speeds in an enclosed unit. The air temperature in the unit was controlled at 32°C during all fermentations.
Two substrate types were used during SSFs conducted in the 18.7 ℓ RDBs: an artificial gel based substrate and steamed wheat bran. The gel substrate consisted of 6 mm cubical particles, had a starch concentration of 4.4 % (w/w) and contained only inorganic nitrogen. Biomass of A. orvzae during gel SSFs was estimated by protein. Growth of A. oryzae during wheat bran fermentations was monitored by oxygen uptake.
2. Experimental investigations of the effects of operational variables on growth of A. oryzae during SSFs in small-scale RDBs. Gel substrate fermentations were conducted using 1.5 or 3.0 kg quantities in the 18.7 ℓ RDBs. Under stationary conditions, the fungus grew well and evenly throughout the substrate bed. A. oryzae also grew well during all rolled fermentations, although shear forces caused by drum rotation reduced overall protein production and sporulation. The highest recorded substrate temperature was 35.5°C which occurred during a stationary 3 kg SSF. The highest oxygen uptake rate (OUR) recorded during a gel fermentation was 38 µmole. (g gel substrate)-1.h-1.
Fermentations of 2 kg wheat bran were also conducted in the 18.7 ℓ RDBs. The fungus grew poorly and unevenly during stationary fermentations. Drum rotation improved overall growth of A. oryzae compared to static fermentations. Rotational speed had no apparent effect on fungal growth during rolled fermentations. Any adverse effects of shear caused by agitation of the substrate, were outweighed by improvements in mass and heat transfer. Peak OURs in excess of 250 µmole.(g wheat bran)-1.h-1 were recorded during rolled fermentations. In all fermentations, maximum substrate temperatures were greater than 40°C. The fungus also grew well during rolled, 10 kg wheat bran SSFs in a 200 ℓ RDB system.
During rolled SSFs in the 18.7 ℓ RDBs, the substrate appeared homogeneous in the radial direction (ie. within the substrate depth) but not in the axial direction (ie. along the bioreactor length). Substrate near the aeration inlet end appeared drier and was cooler than substrate near the gas exit end of the bioreactors.
3. Experimental investigations of the effects of operational variables on gas flow behaviour in small-scale RDBs. Gas flow behaviour in the 18.7 ℓ RDBs was investigated under similar conditions to those used in wheat bran fermentations. Gas was fed continuously into the bioreactors during residence time distribution experiments. Air was replaced by pure nitrogen for a 5 min period and the oxygen depletion response was measured in the gas stream leaving the bioreactor. The response curves for various operating conditions showed that high rotational speeds and low gas flow rates promoted axial dispersion of the gas molecules as they passed through the bioreactors.
Overall, the experiment studies (steps 2 and 3) showed that:
• Optimal aeration characteristics are more likely to be determined by metabolic heat removal than by oxygen supply requirements.
• Shear forces may adversely affect fungal growth, but their effect is not critical since the fungus may still grow well under high shear conditions.
• Substrate characteristics, such as particle size and nutrient concentrations are important for SSFs conducted in RDBs since they affect the total metabolic activity in the bioreactor and therefore, operating requirements.
• Substrate homogeneity in the radial direction may be achieved by agitation of the substrate during drum rotation. However, substrate homogeneity in the axial direction is difficult to achieve in RDB designs with end-to-end aeration.
4. Proposal of design and operation features of RDBs for large-scale SSF. Heat removal, oxygen supply, and homogeneous conditions in the substrate (in both the radial and axial directions) will be important during fungal pure-culture, batch SSFs conducted in RDBs at large scales. Baffles and large L/Ds could be useful for such operations since these will enhance interactions between the solid and gas phases in the bioreactor and thereby improve the rates of heat and mass transfer. The design of aeration systems for large-scale RDBs should attempt to achieve gas phase behaviour analogous to a number of ideal stirred tanks in parallel. This will result in homogeneous conditions in the bioreactor headspace in the axial direction. Since the rates of heat transfer and mass transfer between the substrate and the headspace will then be equal along the bioreactor length, substrate homogeneity in the axial direction should also be achieved.
5. Development of rule-of-thumb scale-up criteria for SSF in RDBs. Simple rule-of-thumb scale-up criteria, including constant L/D, constant solids loading, constant aeration rate (in vvm) and constant fraction of the drum critical speed of rotation, were explored. However, scale-up based on these criteria would lead to overheating problems.
6. Proposal of a mathematical model which mechanistically describes heat transfer during SSF in RDBs. A mathematical model was proposed which mechanistically describes heat transfer during SSF of A. oryzae in a RDB. The model incorporates several mechanisms for heat transfer between the substrate, bioreactor headspace, bioreactor wall and the surroundings. These mechanisms include conduction, convection, evaporative cooling and energy transfer related to gas entering and leaving the RDB system.
The model was used to predict SSF performance at a number of scales and under various design and operating conditions. The small scale predictions compared well with experimental results. The predictions of large scale systems showed that aeration requirements (in vvm) increased significantly with scale. These predictions also showed that water loss from the substrate was the dominant mechanism of heat removal at large scales. Although not validated, the model was shown to be a useful tool in exploring the design and operation of RDBs for SSF. It represents a first step towards developing semi-fundamental scale-up criteria for SSF in RDBs.
Overall, the experimental studies conducted as part of this research have contributed significantly to the understanding of the operation of RDBs for SSF. However, further work and different experimental approaches are needed for continued research in this area. The holistic approach (as used in step 2) will be useful for characterising the overall effects of bioreactor design and operation on SSF performance. However such studies will need to be complemented by other experiments where transport and microbial phenomena are characterised independently of each other (such as in step 3). This should lead to better understanding of SSF performance in RDBs and permit the development of accurate mathematical models incorporating important microbial and transport phenomena. Mathematical models based on mechanistic descriptions of processes occurring during SSF in RDBs will facilitate the development of useful scale-up strategies.