Demonstrating Commercial Hollow Fibre Membrane Contactor Performance at Industrial Scale for Biogas Upgrading at a Sewage Treatment Works

Reviewer 1

The work deal with biogas upgrading by membrane separation. The object was not clearly defined.

We have revised the aim, in accordance with the reviewers request,” This study proposes to complement and build upon the limited existing literature on real gases, through implementing a demonstration scale HFMC system within a full-scale anaerobic digestion facility in order to determine technology robustness for CO₂ separation on real biogas”

Please rewrite the abstract to convey clear information.

This has been rewritten to rebalance the abstract in accordance with request

Fig. 1 should be in 2.1.

This has been moved.

The conclusion was way too long.

This has been cut down significantly in accordance with the reviewers instruction

Figure 3a is not so clear for the reader. I guess vertical and horizontal error bars have different meaning. What is the meaning of "cumulative hours of biogas upgrading"? I suggest the Authors to re-design the figure or at least to clarify in the figure caption the meaning of vertical and horizontal bars

The reviewer is correct, we have revised the figure caption for clarification as,”For (a) x-axis error bars are the time period over which the data is collected; y-axis is the mean and standard deviation of η_CO2 over discrete period measured. For (b) box edges, 25^th and 75^th percentile; centre line, median; whiskers, min. and max.; n = 17).”

Figure 5 remove the table borders. Change "more focused image" to "high magnification image".

These changes have been made

Figure 6 and related text: large part of the discussion concern peaks at wavenumbers between 1800 and 800 cm-1. It would be great to have a second figure or an insert, where this area of the graph is expanded.  

As recommended, an inset of this region has now been included

Figure 8 and related text: why the authors fitted experimental data with a straight line? Authors should explain this choice in the text.

We have inserted a comment as requested by the reviewer, ”Straight lines represent synthetic data from virgin membrane for comparison, deviations plotted from this data to indicate the impact of fouling, and subsequently cleaning of the module”.

My suggestions/corrections/comments to improve the manuscript are:

I think that the main flaw of the manuscript is that it seem that the work could have been more carefully planned. Probably it is a consequence of implementing a method (already ensemble in a lab) to real onsite operation conditions with all the contingencies associated with that. This comment is related with the data presented in Figures 2 and 4.

Thanks for this feedback. The reviewer is correct that work in field conditions is quite different from the laboratory, which arose in a key distinction in the data. Whilst we commented on this in the conclusions, we probably didn’t underline sufficiently in the main body of the text in retrospect – and so we can see the reviewers perspective. For clarification, the key distinction between the data from the synthetic (lab data) and real data (onsite) is that whilst the feed pressure was boosted from the anaerobic digester, this was only pressurised to around 60 mbar – sufficient pre-pressure for a CHP engine. Whilst the pressure drop across the actual module is quite low on the gas-side, the entrance effects created a pressure drop which reduced the flow rates that we could achieve because of this low feed pressure. The extent of this limitation was not foreseen and due to the zoning (Health & Safety convention for risk of explosion), there was little that we could do to implement a change.  This made it impossible to reach the higher flow rates evidenced in the lab. in addition to the expected CO2 capture ratios as the operational L/G ratios were then sorely limited.

Two important comments – we still believe that this is a genuinely unique dataset, that is of significance as there is almost no data on real gas, largely because it is so difficult to obtain! We would also like to appease the reviewers concerns that we were able to , we did record comparable flow rates, and were able to compare L/G ratios within these discrete regions.

We therefore reflected on how best to plot this data and felt that it might be best to demonstrate that high efficiency CO2 capture ratios are achievable with the lab data, and overlay the data onsite within the relevant L/G range and using comparable flow rates, to illustrate the relative parity and similarity in trajectory between these different conditions.   

Concerning Figure 2:

1. Unless I am not “reading” the figure correctly, for onsite data (red) you go until an L/G around 0.6. For Lab data, you have for Gas Flow Rates 0.2 and 0.3 data until L/G value of 2.5, for Gas Flow Rate 0.6 until L/G around 1.2, for Gas Flow Rate 0.4 until L/G around 1.7, for Gas Flow Rate 0.8 until L/G around 0.9 and for the others until L/G around 0.7. It looks to random to make sense. Since the focus is the evaluation of the onsite procedure, I recommend that you change the graphic and use only data for L/G until 0.6 for all Gas Flow Rates. This change, will not only overcome this seemingly random planning, but it will also evidence that you have an even better CO2 capture ratio with real biogas then you have with synthetic gas for L/G bellow 0.6. You may present in a different graphic the lab data until L/G equal to 2.5 as a Lab study of the behavior of the HFMC behavior for higher L/G ratios.

We appreciate your perspective on this. Please see the detailed response above which corresponds to this point and explains why feed gas pressure influence this effect.

As stated above, we were in two minds as to which route to take for this plot, but the reviewers advice has helped determine which is more apprapriate. We have therefore reduced the range to make it clearer. The reviewers is also correct that real data looks comparable if not better – for which there are reasons, that are outlined within the text.

There is also an important feature that you must address: the tendency of the CO2 capture ratio onsite (red) seems to be starting to cross with the lab data around the maximum L/G ratio you measured. It might not follow the same tendency of the lab data and future works must study L/G ratios over 0.6 to know how the CO2 capture ratios really evolves.

Whilst it is not exact, the trend and approximation are very close given the plethora of variables which become important and are too difficult to reconcile between the two datasets through modelling, e.g. viscosity (gas,liquid), diffusivity (gas,liquid), density (gas,liquid), to name but a few. We have inserted a comment in the discussion to compensate for the distinction in gradient – however, we are confident that the trend will ultimately hold enabling exceedingly high CO2 capture ratios for an exceptionally strong economic return

See discussion comment, “Slight discrimination in responsiveness (the slope identified between L/G ratio and η_CO2) was identified between the synthetic and real data which can be accounted for by the complex and dynamic physical chemistry of both phases in an industrial environment (e.g. density, viscosity, diffusivity).”

1. Concerning the inset, why was the lab data not collected until L/G ratios at least equal to the ones collected onsite? It’s a lab study it’s not related to onsite constraints. You have data for L/G ratios bellow the ones collected onsite. It’s another thing that deserves better planning and if you have the apparatus still assembled you should get the data at least until L/G = 0.7. In the main graphic you went until L/G = 2.5…

The reviewer is correct. This would have been ideal. However, as with most studies, data were collected in series. We worked in the lab to develop this process, which was really quite new and untested. We then installed this in the field – while this appears here as 6 months works, I can assure you that it took a lot longer! As such, we had to diagnose the data, with data that we had available from earlier trials in the laboratory as we could not go back into the lab due to time constraints. To reduce costs in the lab, in the earlier trials, we used a smaller module – this lowered cost of synthetic gas (particularly methane which is expensive), and as such we had less data on this demonstration scale module with which to compare. We elected to use this type of module onsite because its modular design is similar to larger modules which would be used in the scale-up of this process.

Concerning Figure 4: Here is the same problem as the inset of Figure 2. Like it is, it shows lack of planning, you want to compare the onsite results with the results obtained in the lab but you didn’t measure in the lab until the same L/G ratios you measured onsite.

This was limited due to the pumping scale in the laboratory environment. However, two comments: (i) both datasets overlap very well; and (ii) there are ten data points in this overlap which we feel is pretty convincing – even if the alignment is not perfect! Such discrimination is inevitable when undertaking such complex work and we do not feel that it undermines the message that the dataset creates.

The results presented in the manuscript (concerning pre-filtration, type of water, temperature, fouling diagnosis, recovery procedures, gas flow rates) are interesting and deserve publication but these discrepancies take some merit off of the work.

We hope that the above commentary now satisfies the reviewer. Whilst hindsight is 20:20, the industrial environment is somewhat unforgiving, and whilst this industrial innovation is significant, the science is complex and as such, some imperfection is almost inevitable. This is the reason why there is so little data out there, making this data presented only more important. To compensate for this comment, we have of course also altered both the text and the figures as the reviewer has recommended which we feel softens this perspective, and improves the overall impact.

 In the Discussion section, first phrase, you state that during a period of steady-state operation, the CO2 capture ratio onsite was comparable to what was predicted with synthetic biogas of equivalent composition. Like I said above, I have many doubts on this point. From the data presented in Figure 2, it seems that the CO2 capture ratio is starting to cross with the lab data around the maximum L/G ratio you measured. You can keep the phrase but at his point you should make some reference at this tendency.

We have changed the structure of this sentence and added a further sentence to explain. This is simply a function of the complexity of the system, the physical chemistry is influenced by temperature on both the gas and liquid phase, which will introduce some error, which cannot be compensate through models. However, as a system well defined by physical chemistry, we also know that we will ultimately achieve the same trend and high separation efficiency, were the two systems isothermal – which of course the on-site system cannot be due to the fact that it was installed outside and exposed to the elements. The additive explanation and corrections are included as follows,” Slight discrimination in responsiveness (the slope identified between L/G ratio and η_CO2) was identified between the synthetic and real data which can be accounted for by the complex and dynamic physical chemistry of both phases in an industrial environment (e.g. density, viscosity, diffusivity and solubility).”

In line 234 you mention Figure 4 but it has nothing to do with what your saying. Just remove it.

We agree with the reviewer, this has now been removed.

References 2, 5 and 19. Remove when you last accessed it, the information the information is not updated so the year of its publication is enough.

In accordance with the request, these have since been removed.