the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Dec 2022
Towards real-time seismic monitoring of a geothermal plant using Distributed Acoustic Sensing
Abstract. Distributed Acoustic Sensing (DAS) is an emerging technology for acquiring seismic data on virtual sensors densely distributed along an optical fiber. The broadband response of the sensors, associated with the possibility of deploying fiber optic cables in harsh conditions and the relatively moderate cost of this sensing element gives clear perspectives for DAS in geothermal wells to contribute to the monitoring operations of geothermal plants. However, the technical feasibility of managing the large flow of data generated by the DAS and the suitability of the strain-rate acquisitions to monitor locally induced seismicity was yet to be assessed.
We propose a monitoring concept establishing DAS as an effective component of the seismic monitoring of the Schäftlarnstraße geothermal plant (Munich, Germany). The underlying data management system links the existing on-site infrastructure, including the fiber optic cable deployed in one of the site’s injection wells and the associated DAS recorder, to a cloud Internet-of-Things (IoT) platform designed to deliver both a secure storage environment for the DAS acquisitions and optimized computing resources for their processing. The proposed solution was tested over a period of six months and showed the feasibility of efficiently acquiring and processing the large flow of continuous DAS data. For seismic risk mitigation purposes, we additionally investigate the potential of the monitoring concept to tend towards real-time monitoring. The processing outcomes, focusing especially on two detected local seismic events, demonstrates the relevance of DAS from geothermal wells for the (micro)seismic monitoring of the geothermal site. Despite the noisy operational conditions, the applied processing workflow takes advantage of the sensors’ high spatial density for data denoising and event triggering and highlights that higher detection sensitivity than conventional seismometers can be achieved. From a different perspective, further analyses of the DAS records confirm the logging capabilities of the technology, especially regarding well completion integrity.
The 6-months test period shows that permanent DAS can be integrated as a routine seismic monitoring component of geothermal plants and advantageously complement surface seismometer-based networks, especially in urban environments.
- Preprint
(3288 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2022-1417', Ariel Lellouch, 28 Dec 2022
Dear authors and editor,
The manuscript "Towards real-time seismic monitoring of a geothermal plant using Distributed Acoustic Sensing" describes the implementation of a DAS-based seismic monitoring system in a geothermal environment. While more complete workflows using downhole DAS data for seismic monitoring have already been published, I think that it is interesting to see the operation of a closed-loop system from raw data to actionable alerts to the operator in a functioning geothermal project. There is also very limited seismic activity, which hinders the possible analysis. Overall, I think the paper should be published, but I have some comments about the writing/focus of the paper. I also attach the PDF with more technical issues marked down.
First of all, the data processing sequence is eventually quite limited in extent – a temporal filter, applying an F-K filter for upgoing waves, and an STA/LTA on individual DAS traces. I think that it takes an overly large portion of the paper. In addition, this approach does not make full use of the spatial continuity of DAS, in my opinion, because picking individual (and noisy) DAS channels is problematic for low SNR. Finally, from an actionable point of view, estimating at least to some extent the location and magnitude of the detected events are, as far as I know, critical. I understand that the seismic activity was practically non-existent and it is a challenging task, but I wonder whether detection only is sufficient for operational decisions. Anyhow, it would be good to discuss it in the paper.
I also think the computational infrastructure is the most interesting part of the paper, and I would further emphasize it. The current data pipeline is manageable – processing 5Gb per hour is doable (as the authors mention, they have about a 5X margin with the current system). It would be very interesting to discuss significant scaling up to longer/multiple wells. Having said that, my problem with that part of the paper is that it is overly tied to Azure (specifically), to the point it almost sounds commercial. I think some form of abstraction would be better tailored.
In terms of the alert system fed by the detections, it would also be very interesting to hear about the thresholding decisions and approach for automatically alerting the operator. Going to shorter than 1-hour windows (± 1m 20 as the authors report) is a useful direction, but I wonder how realistic it would be to use the same system not only for traffic-light but maybe also for early warning and automatic shutdown of operations.
Thanks for the interesting read and happy new year,
Ariel Lellouch
-
AC1: 'Reply on RC1', Jérôme Azzola, 25 Feb 2023
Dear authors and editor,
The manuscript "Towards real-time seismic monitoring of a geothermal plant using Distributed Acoustic Sensing" describes the implementation of a DAS-based seismic monitoring system in a geothermal environment. While more complete workflows using downhole DAS data for seismic monitoring have already been published, I think that it is interesting to see the operation of a closed-loop system from raw data to actionable alerts to the operator in a functioning geothermal project. There is also very limited seismic activity, which hinders the possible analysis. Overall, I think the paper should be published, but I have some comments about the writing/focus of the paper. I also attach the PDF with more technical issues marked down.
Dear Ariel Lellouch,
We thank you very much for your detailed review of the manuscript and for your constructive comments. Below is a response to the main remarks formulated in the evaluation letter. In addition, we attach the pdf file you provided with answers to each of your comment.
- RC1: First of all, the data processing sequence is eventually quite limited in extent – a temporal
filter, applying an F-K filter for upgoing waves, and an STA/LTA on individual DAS traces. I think that it takes an overly large portion of the paper.JA et al.: This manuscript is not intended to seismologists only, but also to geothermal field operators or mining authorities, in general, to geoscientists. This explains why the processing sequence is not the main focus of the manuscript (nor the induced seismicity detailed characterization) but a component of the infrastructure developed around the DAS monitoring. However, sections 3.2 and 3.3 are of interest for two main reasons. First, it enables to describe the recording conditions, especially regarding the anthropogenic noises one has to account for in a real operational environment. Second, it enables to discuss what is a minimum requirement in terms of seismic processing (i.e. identification of seismic event candidates) that will need to run in near real-time. Moreover, we note that both subsections represent about 10% of the whole paper content (~2.5 pages over 23 in total).
-In addition, this approach does not make full use of the spatial continuity of DAS, in my opinion, because picking individual (and noisy) DAS channels is problematic for low SNR.
Indeed, we are not exploiting all the resources brought by a dense antenna for which many different approaches could be applied, like those using trace similarities as you suggest. Again, this is motivated by the fact that we do not focus the paper on data processing and it is clear that there could be many ways to improve the processing sequence, triggering/detection included, that are well documented in many papers. Despite the simplicity of the considered approach (using the sum of coincidences on single channels), it does not require any a priori knowledge, especially regarding the velocity model and therefore may avoid systematic bias. It also allowed identifying a small magnitude event not visible from other sensors deployed in the area.
- Finally, from an actionable point of view, estimating at least to some extent the location and magnitude of the detected events are, as far as I know, critical. I understand that the seismic activity was practically non-existent and it is a challenging task, but I wonder whether detection only is sufficient for operational decisions. Anyhow, it would be good to discuss it in the paper.
In the Munich area, the seismic rate is low, as you noticed. Both local events have nevertheless been characterized by their origin time and distance to TH3 well and depth (see section 4.2.3). With one single vertical antenna, it is difficult to do more for the hypocentre determination. However, that could provide a first set of parameters needed for decision-making. It is correct that the magnitude (or PGV at surface) would constitute the other important parameter. Without any empirical approach calibrated at the site, the magnitude determination from DAS recordings is challenging because strain-rate on a single vertical component are recorded whereas velocity measurements are expected in source models. The transformation from one to the other is not straightforward.
In order to improve the processing sequence in the overall integration concept, we are currently working on the automatic picking of the arrival times using wave cross-correlation, the automatic inversion of the source parameters (hypocentre, magnitude and stress drop). This goes beyond the current manuscript presenting the integration concept and the associated data management structure.
- I also think the computational infrastructure is the most interesting part of the paper, and I would further emphasize it. The current data pipeline is manageable – processing 5Gb per hour is doable (as the authors mention, they have about a 5X margin with the current system). It would be very interesting to discuss significant scaling up to longer/multiple wells.
Thank you for the suggestion that we accounted for in the new manuscript. Section 5.3, which discusses perspectives, elaborates on this aspect.
- Having said that, my problem with that part of the paper is that it is overly tied to Azure (specifically), to the point it almost sounds commercial. I think some form of abstraction would be better tailored.
We recognize that the original manuscript associated the IoT cloud platform solely to Azure. The original constraint comes from the fact that we deliberately chose to keep the cloud services used by the geothermal field operator to fit with the resources which are already in place. In the new manuscript, this point is discussed in Section 2.2.1. We also minimized the references to proprietary products and we used generic terminology applying to cloud-based services as much as possible.
- In terms of the alert system fed by the detections, it would also be very interesting to hear about the thresholding decisions and approach for automatically alerting the operator. Going to shorter than 1-hour windows (± 1m 20 as the authors report) is a useful direction, but I wonder how realistic it would be to use the same system not only for traffic-light but maybe also for early warning and automatic shutdown of operations.
This is a very interesting point that we develop end of Section 5.2 and in Section 5.3. Indeed, one task of the INSIDE project, in which the present work has been carried out, consists in proposing a concept and a prototype for a reservoir management system (RMS). The objective of this RMS is to help the field operator in decision-making and this topic is developed in Gaucher et al. (2022). We show here that DAS recordings could feed such an RMS by delivering observed seismicity results and “boundary” conditions for seismicity forecast models. The latter could be run on the same cloud infrastructure.
Thanks for the interesting read and happy new year,
Ariel Lellouch
Thank you for your careful reading and fruitful comments. We hope that the revised manuscript will reflect them.
Best regards,
Jérôme Azzola and co-authors.
-
AC1: 'Reply on RC1', Jérôme Azzola, 25 Feb 2023
-
RC2: 'Comment on egusphere-2022-1417', Anonymous Referee #2, 31 Dec 2022
The authors provide a description of a fibre optic installation for seismological observations of a geothermal site in southern Germany.
Although this is an ineresting subject and could justify sharing information to the scientific community, to be frank, I struggle to see scientific novelty in the current manuscript. At times it reads more like a marketing brochure or a proposal, with all the emphasis on "novelty" of the approach. There are numerous installations world-wide in the oil and gas industry where a similar data-management approach is applied routinely.
Furthermore, the long delay caused by 1h data files upload to the cloud system hardly make it suitable for real-time monitoring and hazard evaluation. The authors describe in the later part of the paper a 20sec delay system, but it is unclear to me from the current manuscript how this is realised.
I feel the manuscript would benefit from a comparison with the mentioned seismometer network. At the moment, only two events seem to be observed. if that is the case, it should be stated clearly. Otherwise some statistics and comparsion with the surface network would show scientific rigour.
Also some analysis of the noise of the installation: is there difference between day and night; between different months; public holidays?
Finally if would seem appropriate here to clearly state the benefits and down-sides of a fibre-optic system for such cases. Additionally some discussion on the design: would there be benefits in using deviated parts of the well; how could the second fibre be integrated in the workflow and what would the add
Please see attached PDF with comments for more detailed comments.
-
AC2: 'Reply on RC2', Jérôme Azzola, 25 Feb 2023
Dear Sir, dear Madam,
We thank you for your review and hope that the revised manuscript will be an important improvement that will answer the concerns you raised. In addition to our answers below, we are attaching the pdf file with replies to each of your comments.
- RC2: Although this is an interesting subject and could justify sharing information to the scientific community, to be frank, I struggle to see scientific novelty in the current manuscript. At times it reads more like a marketing brochure or a proposal, with all the emphasis on "novelty" of the approach. There are numerous installations world-wide in the oil and gas industry where a similar data-management approach is applied routinely.
JA et al.: Regarding the Solid Earth guidelines and the scopes addressed by the journal, we consider that our manuscript covers interdisciplinary topics, which are relevant to the journal, including seismic instrumentation / infrastructure, time dependent seismology and monitoring related to geo-energy applications. The particularity and the value of the study lies in the concept and infrastructure proposed to make DAS continuous acquisitions and associated results one component of a georeservoir risk mitigation system (see Introduction and Section 5.3). It is described and tested under real operating conditions, those of the Schäftlarnstraße geothermal plant. To fit with the infrastructure in place, it was necessary to base it on the Microsoft cloud services used by the geothermal field operator. We recognize that the reference to this provider in the original manuscript may sound like an advertisement. Consequently, we minimized all references to proprietary products and we used generic terminology applying to cloud-based services as much as possible. We also discuss the possible adaptation to other cloud infrastructures.
Compared to the O&G industry, the deep geothermal community is applying fiber-optic sensing for a short while only. One possible explanation is the perspective of larger associated developments (linked to energy transition) that come together with increased investments. Thus, new perspectives and challenges emerge for the involved actors that are not necessarily familiar with this technology. Besides, the technology may also have to be adapted to the geothermal field characteristics that can differ from those of the O&G fields.
Finally, the abstract and the introduction of the manuscript have been revised to better describe the field of application of the study and its objectives.
-Furthermore, the long delay caused by 1h data files upload to the cloud system hardly make it suitable for real-time monitoring and hazard evaluation. The authors describe in the later part of the paper a 20sec delay system, but it is unclear to me from the current manuscript how this is realised.
Apparently, we have not been clear enough and Sect. 5.2 has been misunderstood, hence the potential confusion caused by the 1h-files mentioned earlier in the manuscript, in Sect. 2 and Sect. 3.
The monitoring system presented in the manuscript was initially tested on 1-hour long files, which leads to a potentially relatively long delay between an event occurrence and its detection: at worst, one hour plus data transfer duration plus processing duration. Therefore, we proposed in Section 5.2 to assess the potential for (near) real-time monitoring and discuss the use of 1-minute long files. With these 1-minute long files, the data processing lasts 10 seconds, and therefore, the durations indicated in section 5.2, counting also for the transfer of the data. With the current configuration and the DAS interrogator characteristics, it is not possible to stream the DAS data in real-time (Section 5.2).
For more clarity, we only refer to the initially tested file durations at the end of Section 3.3 in the revised manuscript. Section 5.2 then discusses the time-dependent processing of the data files..
-I feel the manuscript would benefit from a comparison with the mentioned seismometer network. At the moment, only two events seem to be observed. if that is the case, it should be stated clearly. Otherwise some statistics and comparsion with the surface network would show scientific rigour.
During the 6-months monitoring period, only two local events (within 10 km radius from the well TH3) have been detected (Section 4.2.1). They were not listed in the catalogue of the Bavarian seismological services and no other event is listed in this catalogue during the period near the site. This clarification was added at the beginning of Section 4.2.1. Thus, they are the only two events that can be analysed for local seismicity monitoring perspective. As mentioned, the April 22 event is seen by the DAS only, not by any seismometer (Section 4.2.1). The February 22 event, larger, is visible on surface seismometers. A proper characterization of this specific event, with the use – and comparison – of the surface network seismograms is ongoing. At the light of the current seismicity, it is not possible to perform any statistical analysis.
-Also some analysis of the noise of the installation: is there difference between day and night; between different months; public holidays?Thank you for this suggestion. In the revised manuscript, last paragraph of Section 4.1, we analyse the evolution in time and in depth of the energy contained in the target frequency band. The associated Figure A1 (in appendix A) complements Figure 4 and highlights the anthropogenic origin of the noise visible up to 50 m depth, from the temporal variations observed in the spectral energy. Figure A1 shows night/day variations and changes between working and non-working days.
-Finally if would seem appropriate here to clearly state the benefits and down-sides of a fibre-optic system for such cases.
Unfortunately, there was no seismometer, at the well site, at surface or downhole, to enable any noise or signal to noise ratio comparison between DAS and seismometer recordings.
From another point of view, various advantages associated with fibre optic cables deployed in geothermal wells are discussed in Introduction, in particular for the availability of multiple sensing capabilities. On the other hand, we clearly describe the challenges associated with fibre optic sensing, in connection with the amounts of acquired data, the equipment of geothermal wells and the coupling of the fibre (Section 2.1.2), and the limitations related to the quality of the acquisitions (first paragraph of Section 3.2) or the broadside insensitivity (last paragraph of Section 5.1).
-Additionally some discussion on the design: would there be benefits in using deviated parts of the well; how could the second fibre be integrated in the workflow and what would the add
In the revised manuscript, we further developed Section 5.3 to account for your suggestion, which was also recommended by A. Lellouch (the other reviewer). In the before-last paragraph of Section 5.3, we discuss possible up-scaling of the infrastructure: the acquisition on multiple DAS cables, e.g. in TH3 and TH4 at the SLS site, as well as the increase of the processing and storage resources. We discuss the benefit of deviated and instrumented wells in Section 5.1 (last paragraph).
We hope that the revised manuscript will answer your concerns.
Best regards,
Jérôme Azzola and co-authors.
-
AC2: 'Reply on RC2', Jérôme Azzola, 25 Feb 2023
-
EC1: 'Editor comment on egusphere-2022-1417', Michal Malinowski, 30 Jan 2023
Dear authors,
I think that both Reviewers (experts in your field!) raised several important issues related to your study. In my opinion, addressing them might be beyond what is normally considered a 'Major Revision' and I would be more happy to suggest 'Revise and Resubmit'. I expect a substantial revision of your manuscript following the reviewers' comments. If you feel that this task takes more time than allowed for normal revision, please consider resubmission (after informing the EO).
Best regards,
Editor
Citation: https://doi.org/10.5194/egusphere-2022-1417-EC1 -
AC3: 'Reply on EC1', Jérôme Azzola, 25 Feb 2023
Dear Michal Malinowski,
We have carefully addressed all comments of both referees and we believe that the revised manuscript greatly benefits from their remarks. We provided them answers to their general comments but also to their detailed review of the original manuscript.
The main concern was related to the focus of the paper that was not clear enough to both reviewers. Therefore, we significantly modified the abstract and the introduction to better emphasize the scope of the manuscript. To remain consistent with it, we also reworked the rest of the manuscript and especially the conclusion. As stated in the title, the particularity of the study lies in the infrastructure proposed to make DAS continuous acquisitions and associated results one component of a risk mitigation system applied to an operating geothermal site.
As suggested by the reviewers, the Discussion section has been broaden to the possible upscaling of the infrastructure and its integration into a reservoir management system. We also minimized the wordiness of some sentences and, for more neutrality, explicit references to proprietary services. The sections 2, 3 and 4, however, have not been fundamentally modified, although the noise analysis has been developed as proposed by Reviewer 2.
We consider that we have made a substantial revision of the manuscript, hence the submission of the revised manuscript.
Best regards,
Jérôme Azzola, Emmanuel Gaucher and Katja Thiemann
Citation: https://doi.org/10.5194/egusphere-2022-1417-AC3
-
AC3: 'Reply on EC1', Jérôme Azzola, 25 Feb 2023
Status: closed
-
RC1: 'Comment on egusphere-2022-1417', Ariel Lellouch, 28 Dec 2022
Dear authors and editor,
The manuscript "Towards real-time seismic monitoring of a geothermal plant using Distributed Acoustic Sensing" describes the implementation of a DAS-based seismic monitoring system in a geothermal environment. While more complete workflows using downhole DAS data for seismic monitoring have already been published, I think that it is interesting to see the operation of a closed-loop system from raw data to actionable alerts to the operator in a functioning geothermal project. There is also very limited seismic activity, which hinders the possible analysis. Overall, I think the paper should be published, but I have some comments about the writing/focus of the paper. I also attach the PDF with more technical issues marked down.
First of all, the data processing sequence is eventually quite limited in extent – a temporal filter, applying an F-K filter for upgoing waves, and an STA/LTA on individual DAS traces. I think that it takes an overly large portion of the paper. In addition, this approach does not make full use of the spatial continuity of DAS, in my opinion, because picking individual (and noisy) DAS channels is problematic for low SNR. Finally, from an actionable point of view, estimating at least to some extent the location and magnitude of the detected events are, as far as I know, critical. I understand that the seismic activity was practically non-existent and it is a challenging task, but I wonder whether detection only is sufficient for operational decisions. Anyhow, it would be good to discuss it in the paper.
I also think the computational infrastructure is the most interesting part of the paper, and I would further emphasize it. The current data pipeline is manageable – processing 5Gb per hour is doable (as the authors mention, they have about a 5X margin with the current system). It would be very interesting to discuss significant scaling up to longer/multiple wells. Having said that, my problem with that part of the paper is that it is overly tied to Azure (specifically), to the point it almost sounds commercial. I think some form of abstraction would be better tailored.
In terms of the alert system fed by the detections, it would also be very interesting to hear about the thresholding decisions and approach for automatically alerting the operator. Going to shorter than 1-hour windows (± 1m 20 as the authors report) is a useful direction, but I wonder how realistic it would be to use the same system not only for traffic-light but maybe also for early warning and automatic shutdown of operations.
Thanks for the interesting read and happy new year,
Ariel Lellouch
-
AC1: 'Reply on RC1', Jérôme Azzola, 25 Feb 2023
Dear authors and editor,
The manuscript "Towards real-time seismic monitoring of a geothermal plant using Distributed Acoustic Sensing" describes the implementation of a DAS-based seismic monitoring system in a geothermal environment. While more complete workflows using downhole DAS data for seismic monitoring have already been published, I think that it is interesting to see the operation of a closed-loop system from raw data to actionable alerts to the operator in a functioning geothermal project. There is also very limited seismic activity, which hinders the possible analysis. Overall, I think the paper should be published, but I have some comments about the writing/focus of the paper. I also attach the PDF with more technical issues marked down.
Dear Ariel Lellouch,
We thank you very much for your detailed review of the manuscript and for your constructive comments. Below is a response to the main remarks formulated in the evaluation letter. In addition, we attach the pdf file you provided with answers to each of your comment.
- RC1: First of all, the data processing sequence is eventually quite limited in extent – a temporal
filter, applying an F-K filter for upgoing waves, and an STA/LTA on individual DAS traces. I think that it takes an overly large portion of the paper.JA et al.: This manuscript is not intended to seismologists only, but also to geothermal field operators or mining authorities, in general, to geoscientists. This explains why the processing sequence is not the main focus of the manuscript (nor the induced seismicity detailed characterization) but a component of the infrastructure developed around the DAS monitoring. However, sections 3.2 and 3.3 are of interest for two main reasons. First, it enables to describe the recording conditions, especially regarding the anthropogenic noises one has to account for in a real operational environment. Second, it enables to discuss what is a minimum requirement in terms of seismic processing (i.e. identification of seismic event candidates) that will need to run in near real-time. Moreover, we note that both subsections represent about 10% of the whole paper content (~2.5 pages over 23 in total).
-In addition, this approach does not make full use of the spatial continuity of DAS, in my opinion, because picking individual (and noisy) DAS channels is problematic for low SNR.
Indeed, we are not exploiting all the resources brought by a dense antenna for which many different approaches could be applied, like those using trace similarities as you suggest. Again, this is motivated by the fact that we do not focus the paper on data processing and it is clear that there could be many ways to improve the processing sequence, triggering/detection included, that are well documented in many papers. Despite the simplicity of the considered approach (using the sum of coincidences on single channels), it does not require any a priori knowledge, especially regarding the velocity model and therefore may avoid systematic bias. It also allowed identifying a small magnitude event not visible from other sensors deployed in the area.
- Finally, from an actionable point of view, estimating at least to some extent the location and magnitude of the detected events are, as far as I know, critical. I understand that the seismic activity was practically non-existent and it is a challenging task, but I wonder whether detection only is sufficient for operational decisions. Anyhow, it would be good to discuss it in the paper.
In the Munich area, the seismic rate is low, as you noticed. Both local events have nevertheless been characterized by their origin time and distance to TH3 well and depth (see section 4.2.3). With one single vertical antenna, it is difficult to do more for the hypocentre determination. However, that could provide a first set of parameters needed for decision-making. It is correct that the magnitude (or PGV at surface) would constitute the other important parameter. Without any empirical approach calibrated at the site, the magnitude determination from DAS recordings is challenging because strain-rate on a single vertical component are recorded whereas velocity measurements are expected in source models. The transformation from one to the other is not straightforward.
In order to improve the processing sequence in the overall integration concept, we are currently working on the automatic picking of the arrival times using wave cross-correlation, the automatic inversion of the source parameters (hypocentre, magnitude and stress drop). This goes beyond the current manuscript presenting the integration concept and the associated data management structure.
- I also think the computational infrastructure is the most interesting part of the paper, and I would further emphasize it. The current data pipeline is manageable – processing 5Gb per hour is doable (as the authors mention, they have about a 5X margin with the current system). It would be very interesting to discuss significant scaling up to longer/multiple wells.
Thank you for the suggestion that we accounted for in the new manuscript. Section 5.3, which discusses perspectives, elaborates on this aspect.
- Having said that, my problem with that part of the paper is that it is overly tied to Azure (specifically), to the point it almost sounds commercial. I think some form of abstraction would be better tailored.
We recognize that the original manuscript associated the IoT cloud platform solely to Azure. The original constraint comes from the fact that we deliberately chose to keep the cloud services used by the geothermal field operator to fit with the resources which are already in place. In the new manuscript, this point is discussed in Section 2.2.1. We also minimized the references to proprietary products and we used generic terminology applying to cloud-based services as much as possible.
- In terms of the alert system fed by the detections, it would also be very interesting to hear about the thresholding decisions and approach for automatically alerting the operator. Going to shorter than 1-hour windows (± 1m 20 as the authors report) is a useful direction, but I wonder how realistic it would be to use the same system not only for traffic-light but maybe also for early warning and automatic shutdown of operations.
This is a very interesting point that we develop end of Section 5.2 and in Section 5.3. Indeed, one task of the INSIDE project, in which the present work has been carried out, consists in proposing a concept and a prototype for a reservoir management system (RMS). The objective of this RMS is to help the field operator in decision-making and this topic is developed in Gaucher et al. (2022). We show here that DAS recordings could feed such an RMS by delivering observed seismicity results and “boundary” conditions for seismicity forecast models. The latter could be run on the same cloud infrastructure.
Thanks for the interesting read and happy new year,
Ariel Lellouch
Thank you for your careful reading and fruitful comments. We hope that the revised manuscript will reflect them.
Best regards,
Jérôme Azzola and co-authors.
-
AC1: 'Reply on RC1', Jérôme Azzola, 25 Feb 2023
-
RC2: 'Comment on egusphere-2022-1417', Anonymous Referee #2, 31 Dec 2022
The authors provide a description of a fibre optic installation for seismological observations of a geothermal site in southern Germany.
Although this is an ineresting subject and could justify sharing information to the scientific community, to be frank, I struggle to see scientific novelty in the current manuscript. At times it reads more like a marketing brochure or a proposal, with all the emphasis on "novelty" of the approach. There are numerous installations world-wide in the oil and gas industry where a similar data-management approach is applied routinely.
Furthermore, the long delay caused by 1h data files upload to the cloud system hardly make it suitable for real-time monitoring and hazard evaluation. The authors describe in the later part of the paper a 20sec delay system, but it is unclear to me from the current manuscript how this is realised.
I feel the manuscript would benefit from a comparison with the mentioned seismometer network. At the moment, only two events seem to be observed. if that is the case, it should be stated clearly. Otherwise some statistics and comparsion with the surface network would show scientific rigour.
Also some analysis of the noise of the installation: is there difference between day and night; between different months; public holidays?
Finally if would seem appropriate here to clearly state the benefits and down-sides of a fibre-optic system for such cases. Additionally some discussion on the design: would there be benefits in using deviated parts of the well; how could the second fibre be integrated in the workflow and what would the add
Please see attached PDF with comments for more detailed comments.
-
AC2: 'Reply on RC2', Jérôme Azzola, 25 Feb 2023
Dear Sir, dear Madam,
We thank you for your review and hope that the revised manuscript will be an important improvement that will answer the concerns you raised. In addition to our answers below, we are attaching the pdf file with replies to each of your comments.
- RC2: Although this is an interesting subject and could justify sharing information to the scientific community, to be frank, I struggle to see scientific novelty in the current manuscript. At times it reads more like a marketing brochure or a proposal, with all the emphasis on "novelty" of the approach. There are numerous installations world-wide in the oil and gas industry where a similar data-management approach is applied routinely.
JA et al.: Regarding the Solid Earth guidelines and the scopes addressed by the journal, we consider that our manuscript covers interdisciplinary topics, which are relevant to the journal, including seismic instrumentation / infrastructure, time dependent seismology and monitoring related to geo-energy applications. The particularity and the value of the study lies in the concept and infrastructure proposed to make DAS continuous acquisitions and associated results one component of a georeservoir risk mitigation system (see Introduction and Section 5.3). It is described and tested under real operating conditions, those of the Schäftlarnstraße geothermal plant. To fit with the infrastructure in place, it was necessary to base it on the Microsoft cloud services used by the geothermal field operator. We recognize that the reference to this provider in the original manuscript may sound like an advertisement. Consequently, we minimized all references to proprietary products and we used generic terminology applying to cloud-based services as much as possible. We also discuss the possible adaptation to other cloud infrastructures.
Compared to the O&G industry, the deep geothermal community is applying fiber-optic sensing for a short while only. One possible explanation is the perspective of larger associated developments (linked to energy transition) that come together with increased investments. Thus, new perspectives and challenges emerge for the involved actors that are not necessarily familiar with this technology. Besides, the technology may also have to be adapted to the geothermal field characteristics that can differ from those of the O&G fields.
Finally, the abstract and the introduction of the manuscript have been revised to better describe the field of application of the study and its objectives.
-Furthermore, the long delay caused by 1h data files upload to the cloud system hardly make it suitable for real-time monitoring and hazard evaluation. The authors describe in the later part of the paper a 20sec delay system, but it is unclear to me from the current manuscript how this is realised.
Apparently, we have not been clear enough and Sect. 5.2 has been misunderstood, hence the potential confusion caused by the 1h-files mentioned earlier in the manuscript, in Sect. 2 and Sect. 3.
The monitoring system presented in the manuscript was initially tested on 1-hour long files, which leads to a potentially relatively long delay between an event occurrence and its detection: at worst, one hour plus data transfer duration plus processing duration. Therefore, we proposed in Section 5.2 to assess the potential for (near) real-time monitoring and discuss the use of 1-minute long files. With these 1-minute long files, the data processing lasts 10 seconds, and therefore, the durations indicated in section 5.2, counting also for the transfer of the data. With the current configuration and the DAS interrogator characteristics, it is not possible to stream the DAS data in real-time (Section 5.2).
For more clarity, we only refer to the initially tested file durations at the end of Section 3.3 in the revised manuscript. Section 5.2 then discusses the time-dependent processing of the data files..
-I feel the manuscript would benefit from a comparison with the mentioned seismometer network. At the moment, only two events seem to be observed. if that is the case, it should be stated clearly. Otherwise some statistics and comparsion with the surface network would show scientific rigour.
During the 6-months monitoring period, only two local events (within 10 km radius from the well TH3) have been detected (Section 4.2.1). They were not listed in the catalogue of the Bavarian seismological services and no other event is listed in this catalogue during the period near the site. This clarification was added at the beginning of Section 4.2.1. Thus, they are the only two events that can be analysed for local seismicity monitoring perspective. As mentioned, the April 22 event is seen by the DAS only, not by any seismometer (Section 4.2.1). The February 22 event, larger, is visible on surface seismometers. A proper characterization of this specific event, with the use – and comparison – of the surface network seismograms is ongoing. At the light of the current seismicity, it is not possible to perform any statistical analysis.
-Also some analysis of the noise of the installation: is there difference between day and night; between different months; public holidays?Thank you for this suggestion. In the revised manuscript, last paragraph of Section 4.1, we analyse the evolution in time and in depth of the energy contained in the target frequency band. The associated Figure A1 (in appendix A) complements Figure 4 and highlights the anthropogenic origin of the noise visible up to 50 m depth, from the temporal variations observed in the spectral energy. Figure A1 shows night/day variations and changes between working and non-working days.
-Finally if would seem appropriate here to clearly state the benefits and down-sides of a fibre-optic system for such cases.
Unfortunately, there was no seismometer, at the well site, at surface or downhole, to enable any noise or signal to noise ratio comparison between DAS and seismometer recordings.
From another point of view, various advantages associated with fibre optic cables deployed in geothermal wells are discussed in Introduction, in particular for the availability of multiple sensing capabilities. On the other hand, we clearly describe the challenges associated with fibre optic sensing, in connection with the amounts of acquired data, the equipment of geothermal wells and the coupling of the fibre (Section 2.1.2), and the limitations related to the quality of the acquisitions (first paragraph of Section 3.2) or the broadside insensitivity (last paragraph of Section 5.1).
-Additionally some discussion on the design: would there be benefits in using deviated parts of the well; how could the second fibre be integrated in the workflow and what would the add
In the revised manuscript, we further developed Section 5.3 to account for your suggestion, which was also recommended by A. Lellouch (the other reviewer). In the before-last paragraph of Section 5.3, we discuss possible up-scaling of the infrastructure: the acquisition on multiple DAS cables, e.g. in TH3 and TH4 at the SLS site, as well as the increase of the processing and storage resources. We discuss the benefit of deviated and instrumented wells in Section 5.1 (last paragraph).
We hope that the revised manuscript will answer your concerns.
Best regards,
Jérôme Azzola and co-authors.
-
AC2: 'Reply on RC2', Jérôme Azzola, 25 Feb 2023
-
EC1: 'Editor comment on egusphere-2022-1417', Michal Malinowski, 30 Jan 2023
Dear authors,
I think that both Reviewers (experts in your field!) raised several important issues related to your study. In my opinion, addressing them might be beyond what is normally considered a 'Major Revision' and I would be more happy to suggest 'Revise and Resubmit'. I expect a substantial revision of your manuscript following the reviewers' comments. If you feel that this task takes more time than allowed for normal revision, please consider resubmission (after informing the EO).
Best regards,
Editor
Citation: https://doi.org/10.5194/egusphere-2022-1417-EC1 -
AC3: 'Reply on EC1', Jérôme Azzola, 25 Feb 2023
Dear Michal Malinowski,
We have carefully addressed all comments of both referees and we believe that the revised manuscript greatly benefits from their remarks. We provided them answers to their general comments but also to their detailed review of the original manuscript.
The main concern was related to the focus of the paper that was not clear enough to both reviewers. Therefore, we significantly modified the abstract and the introduction to better emphasize the scope of the manuscript. To remain consistent with it, we also reworked the rest of the manuscript and especially the conclusion. As stated in the title, the particularity of the study lies in the infrastructure proposed to make DAS continuous acquisitions and associated results one component of a risk mitigation system applied to an operating geothermal site.
As suggested by the reviewers, the Discussion section has been broaden to the possible upscaling of the infrastructure and its integration into a reservoir management system. We also minimized the wordiness of some sentences and, for more neutrality, explicit references to proprietary services. The sections 2, 3 and 4, however, have not been fundamentally modified, although the noise analysis has been developed as proposed by Reviewer 2.
We consider that we have made a substantial revision of the manuscript, hence the submission of the revised manuscript.
Best regards,
Jérôme Azzola, Emmanuel Gaucher and Katja Thiemann
Citation: https://doi.org/10.5194/egusphere-2022-1417-AC3
-
AC3: 'Reply on EC1', Jérôme Azzola, 25 Feb 2023
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 536 | 260 | 44 | 840 | 27 | 46 |
- HTML: 536
- PDF: 260
- XML: 44
- Total: 840
- BibTeX: 27
- EndNote: 46
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1