As mentioned in the section ’’Influence of the raw material” it is important for the economic viability of the process to reach high yields and to utilise the raw material efficiently. In the following example, the effect of recycling liquid streams on the various unit operations was investigated through a combination of simulations and experiments in the bench-scale unit in order to reduce the use of fresh water in the process. In this case, it is also important to examine the build-up of inhibitory substances in the conversion steps, such as hydrolysis and fermentation, as they may affect the yield and productivity in these steps. One of the simulated process alternatives was then experimentally verified in the bench-scale process development unit. A base case was established (Figure 4) which serves as a reference case. The base case simulates a plant where fresh water is added if needed to modify the concentration of dry matter in a step, and also for washing steps. Obviously, this is not realistic for an industrial process, but it reflects the normal procedure used in most lab-scale experiments. A number of different process configurations were simulated to establish the concentration levels of soluble components at various locations in the process, but in the following, two examples are reviewed in more detail. The flowsheets shown have been simplified to speed up calculations and to make the results easier to interpret.

The simulations are based on a feed capacity of 20,000 kg/h wood chips with a moisture content of 50%. This gives the minimum amount of water that enters the system and which must be disposed of. The dry matter is assumed to consist of 36% cellulose, 24% hemicellulose, 21% lignin, and 19% solubles, which is the composition of willow (55). It is assumed that the conversion in the hydrolysis stage is 90%, in the fermentation step 95%, and the recovery in the distillation step is 99%.

The outputs from the process include ethanol as the major product, lignin as the major by-product, resulting from the filtering of the hydrolysis residue, and the stillage waste stream. In the process configuration for the base case, large volumes of fresh water are used. This will result in a very dilute distillation feed, containing around 2.5% (wt/wt) ethanol. Distillation of such a dilute feed makes the operation cost-sensitive for changes in the feed concentration for nearly all distillation technologies (56, 57). The base case yields a liquid waste stream of about 38 tonnes for every tonne of ethanol produced. This results in high costs for fresh water and for waste treatment. The main and probably the only advantage of this process alternative is the low concentration of inhibitors in all the reaction steps, due to the dilution with fresh water. One reason to recycle liquid process streams is to decrease the fresh water demand and to considerably reduce the amount of waste water. Other reasons for recycling are the opportunities of increasing the concentrations of glucose in the fermenter and ethanol in the distillation column.

Recycling of Distillation Stillage. Several recycling options are possible. The simulations were performed to investigate how various components were distributed

Figure 4. The base case flowsheet. 1: Pretreatment; 2: Filtration; 3: Enzyme production; 4: Hydrolysis; 5: Fermentation; 6: Distillation.

in the process assuming that they would not alter the yields in the various reaction steps. A number of components were chosen to represent very volatile (furfural), moderately volatile (acetic acid) and non-volatile (’’soluble”) components. Figure 5 shows a configuration where the stillage resulting from the distillation column is recycled to the washing steps. This reduces the amount of fresh water required in the washing steps to virtually zero. In this case, the stillage stream is reduced to about 7 tonnes/h but the concentration of by-products is much higher. The total amount of by­products is, of course, unaltered. Through this change in the original configuration, it is possible to increase the ethanol concentration to approximately 4.5% (wt/wt), which reduces the energy consumption in the distillation step with about 40%. However, the concentration of by-products is greatly increased and in the hydrolysis step is up to 18 times higher than that obtained in the base case for acetic acid, furfural, and non-volatile components (Figure 6). It can also be seen in the same figure that there is a 9-10 times. higher concentration of acetic acid in the fermentation step compared with the base case, about 1-1.5% (wt/wt). This may cause a negative effect on the yield and on the productivity in the fermentation.

The results obtained from the simulation were examined experimentally in the bench-scale unit using willow as raw material (36). The experiments comprised pretreatment, enzymatic hydrolysis, fermentation, and distillation. The overall ethanol yield in this experiment was only 65% of the theoretical, which was lower than previously obtained in lab-scale investigations (14, 58). But enhanced yields are expected when, for example, hydrolysis is performed in fed-batch mode and fermentation is run continuously. To evaluate the effect of recycling, the stillage stream was fractionated into several volatile fractions and one non-volatile fraction by evaporation, thus simulating a multi-effect evaporation unit. The inhibitory effects of the various fractions were assessed by fermentation using S. cerevisisae after the addition of glucose. The non-volatile residue of the stillage was found to be inhibitory to fermentation already at a concentration five times higher than in the original stillage. The ethanol yield decreased from 0.37 g/g in a pure sugar reference to 0.31 g/g in the residue and the average fermentation rate decreased from 6.3 g/(L h) to 2.7 g/(L h). The acetic acid concentration in the residue was 9.2 g/L, a concentration previously found not to inhibit S. cerevisiae significantly (36), but it is more likely that lignin-degradation products are responsible for the inhibitory action. The evaporation condensates, containing the volatile components, showed no negative effects on fermentation.

The COD and the BOD7 in the stillage stream, the volatile fractions, and the non-volatile residue were used to estimate the environmental impact of disposal. The most volatile part and the non-volatile residue exhibited a considerably higher COD and BOD7 than the intermediate fraction (Figure 7). This indicates that the stream most suited for disposal is the intermediate part of the stillage stream and the parts most suited for recycling are the more volatile fractions. These results show that although the simulation indicates that a high degree of recirculation is an attractive option, in practice it is not possible to achieve an optimised process without the removal of non-volatile compounds. Incorporation of an evaporation unit is one way of removing the non-volatile residue from the process. Since this residue contains high amounts of organic compounds it is well suited to be used for steam generation.

3: Enzyme production; 4: Hydrolysis; 5: Fermentation; 6: Distillation.

Furfural HAc Glycerol Ethanol Solubles Figure 6. Concentration of ethanol and by-products in the hydrolysis. 1: Base case; 2: Recycling of the stillage; 3: Recycling of the distillation feed.

It is also possible to recycle part of the distillation feed back to the washing stage of the pretreated material. This configuration is shown in Figure 8. This alternative gives the same distribution of by-products as the former example (see Figure 6). But the ethanol concentration of the distillation feed could be increased to about 7% (wt/wt) (59), which is comparable to the concentration obtained when ethanol is produced from corn. This reduces the energy demand in the distillation by approximately 30%. However, in this particular case, the ethanol concentration in the hydrolysis reactor is increased to roughly 4% (wt/wt), which could yield a negative effect on the enzymatic hydrolysis. This configuration is presently being investigated using the bench-scale unit.

There are several other ways of increasing the ethanol concentration in the feed to the distillation unit (59). One of these is to recycle part of the stream from the hydrolysis tank. This will increase the ethanol concentration in the distillation feed, but it will also increase the risk of infection, since the recycling of sugar-containing liquids is involved.