Modelos estadísticos de estimación de lluvias

The laboratory trials performed as part of this work were terminated due to time limitations, and it was therefore not possible to further investigate the maximum loading rate, kinetic of operation at multiple loadings rates and under other filtration regimes. Extension of this work could provide further understanding and insight into the two-stage system developed. Specific suggestions of how the work could be extended are given below.

Confirmation of Maximum OLR and Kinetic Investigation

The operating time at the maximum loading rate of 15 g VS l-1 d-1 applied to the RDMBRs was insufficient to confirm stable operation of the two-stage system. After two weeks the system seemed to be recovering from the shock increase in loading. It would be of interest to determine the operational characteristics at a steady state condition at this loading rate and above if possible. At a higher OLR the feed may be need to be dried in order that its volume is not larger than the amount of digestate removed. Also given more time (and more reactors) a kinetic study into the response of the RDMBR/AF system to various OLR and more importantly HRT would be of interest. In particular the response of the system to lower flush rates (longer 1st stage HRT) would be important since the HRT determines the limit to the scale of the process (see section 6.6.1).

The Effect of Inoculum

As noted in the literature review, the activity of rumen hydrolytic microorganisms is up to twice that of those in sewage (Song et al. 2005). The successful

application of rumen inocula was also shown in the RUDAD studies (Gijzen et al. 1987a), although ciliate numbers always were lower in reactors than in the rumen and decreased with time. The effect of rumen inoculum in the RDMBR/AF system would be interesting since the comparatively long SRT (compared with RUDAD) may allow ciliate numbers to remain high. Key points would be the ability to sustain the high hydrolysis rates provided by the rumen inoculum and comparison with the sewage inoculated equivalent. Another interesting study in this and other anaerobic digestion processes would be co- digestion with rumen contents from a slaughterhouse, which may provide enhanced hydrolysis without the need for concern about the sustainability of ciliate populations since these would be added with the feed.

Use of Appropriately sized AF reactors and Continuous 2nd Stage Feeding

The AF reactors used in this work were oversized for their application, since the 0.8-1.5 litre 1st stage reactors provided neither a large volume nor a high strength effluent. This meant the OLR and HRT applied to the AF reactors was low compared with other applications in the literature. Furthermore the 4-litre volume of these reactors more than tripled the total volume of the two-stage systems.

It is possible that the use of these AF reactors allowed the system to operate more stably for two reasons: firstly there was spare capacity in the AF reactors to deal with the higher strength 1st stage effluent during the start up phase or after a temperature shock. Secondly the increased system volume meant that the build-up of VFA or other intermediates occurred more slowly, allowing the biomass extra time to deal with the change. In a larger-scale implementation it would be financially unviable to over-engineer the 2nd stage reactors and so some testing of a two-stage system with appropriately sized reactors is required. The method of feeding the 2nd stage reactors could become more important with appropriately sized reactors, since the reactors will have less spare capacity to deal with changes in conditions such as those experienced under batch feeding. With continuous feeding the inlet conditions remain steady over time, which may allow AF reactors to operate under conditions that would cause failure with batch wise transfer of the 1st stage effluent.

Mesh Filtration Monitoring and Flux Optimisation

So far only a superficial understanding of the flux characteristics of the MSW digestate has been gained. Given that the flux effectively limits the maximum possible size of the system as described in section 6.6.1, greater understanding and optimisation of this part of the process is required. TMP is an important parameter in filtration systems which can be used to monitor the build-up of a fouling layer on the surface of the filtration media, and therefore reactors with pressure sensors on both sides of the mesh surface would give some insight into this. Maximum flux in the RDMBRs was lower than in the MFBRs, probably because there was no applied TMP to encourage filtration or fouling removal in the former. Increasing the flux in RDMBRs requires that a method of direct flux control and maintenance be introduced. Flux could be maintained by a number of methods including liquid and gas backwash, although at laboratory scale all of these are rather difficult to implement.

A simple filtration test rig could bypass the difficulty of laboratory scale in situ testing of flux maintenance/optimisation options. This would allow trials of a number of different methods and regimes under controlled and monitored conditions, which would ideally lead to the development of routines to detect excessive fouling. Actual BMW digestate from a RDMBR could be used to ensure representative filtration characteristics.

Digestion of Other (Problematic) Feed Materials

The process stability demonstrated by the RDMBR/AF system may allow the digestion of potentially problematic feed materials such as those that undergo rapid breakdown (e.g. FVW, FW) or are naturally low in buffering, which may otherwise produce an unstable AD process at high organic loading rate. This hypothesis would need to be tested on a case to case basis but the FW used in the recipe for SBMW in this work was shown to cause process failure in single- stage digestion (Climenhaga and Banks 2007) and thus would be a good example of a problematic feed material.

Three-Stage System

As was described in section 6.6.4, the potential of a high solids post-RDMBR stage to obtain the residual methane should be explored. Of prime importance

is that this system provides high throughput of waste and also that the volumetric production and specific methane are both high. Therefore emphasis should be placed on finding the minimum size for the 3rd stage (and AF) reactors. One possible method for this would be to de-water the removed digestate, keeping the liquid in the RDMBR/AF system to moisten the feed, while allowing the partially stabilised solids to degrade in a high solids environment. As discussed earlier, since the rapidly degradable fractions of the BMW will have been removed in the 1st stage, the digestate is unlikely to undergo acidification when digested in a high solids reactor.

8 Appendices