Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Microseismic monitoring on gas storage reservoirs: a ten-year experience.

Jean-Pierre Deflandre, Institut Franais du Ptrole, France; Frdric Huguet, Gaz de France, France. Abstract
Since 1991, IFP and GDF have been improving microseismic monitoring for reservoir characterization. Using permanent downhole geophones, the location of small microseisms has been performed on two underground gas storage facilities. These experiments make it possible to learn about Passive Seismic Monitoring especially for long term applications. Instrumentation and implementation of a permanent monitoring are discussed. Considering production and microseismic data together, reservoir pressure effects are highlighted: relationships between microseisms and production data being identified. Variations over time have also been observed. The approach developed to process microseismic data on an exploitation site, during a long term microseismic monitoring experiment, is described. This approach, consisting in an evolving platform software, makes the integration of the acquired experience possible. Results obtained suggest the development of a pseudo-real-time monitoring methodology to be applied to production and fluid reinjection or storage.

Introduction
Our topic is to present some results of a ten-year research program on Passive Seismic Monitoring (PSM hereafter) associated with gas storage. This program was based on a partnership between Institut Franais du Ptrole (IFP) and Gaz de France (GDF). Measurements have been performed on two underground gas storage facilities belonging to GDF. The program benefits from the development of well instrumentation especially the permanent downhole geophones initially developed for repetitive well seismic applications. The first part of the paper briefly defines the context of PSM. It also gives some elements of the following challenges: the permanent measurement and the interpretation of microseismicity associated with reservoir exploitation. The second part briefly presents the two field experiments in terms of instrumentation and main results. The third part reports on instrumentation and presents how we benefit from this experience. Potentialities of new developments are highlighted. The fourth part deals with the processing and the interpretation of microseismic data related to reservoir exploitation. It defines the requirements for a real-time application of PSM and presents the approach finally adopted to integrate our field experience and to take into account the exploitation data in both the processing and the interpretation of measurements. The last part gives some conclusions and makes some recommendations for the setting up and use of this technique to characterize the reservoir mechanics. Although our experience mainly concerns gas storage, it is important to mention that the results, tools and methodology are suitable for oil and gas production and also for Hot Dry Rock geothermal applications.

Passive Seismic Monitoring: its contribution to reservoir characterization Context


Oil and gas production acts on the in situ state of stress because of pore pressure variations induced by the exploitation. Fluid extraction and fluid injection (water, steam, CO2, fracturing fluid, etc to maintain pressure level or to improve productivity) modify local effective stresses. This makes the occurrence of different mechanical phenomena at the reservoir level or in adjacent layers possible. Compaction, fracturing and fault displacement have been reported1,2,3. They can be compared to what occurs with extraction of solids in mining. All these phenomena can be associated with a microseismic activity that corresponds to the elastic energy release part of the phenomenon. Microseisms are analogous to earthquakes but the energy involved is very small in comparison: magnitudes (on the Gutemberg-Richter scale) often range between -3 up to 0. Taking into account the amplitude of the phenomenon we want to measure and the instrumentation that we can use (geophones, accelerometers, hydrophones), we will restrict our discussion to downhole measurements. In some particular cases, values between 0 to 4 have been recorded but often local tectonics has to be considered a fortiori if higher values are measured2,3. Reservoir geomechanics is affected by the exploitation but it is itself a key factor of the productivity as reservoir petrophysical properties depend on the mechanical state of stress. Also, the characterization of the reservoir mechanical behavior is of prime interest to understand and optimize field production. This can be based on reservoir modeling if it couples geomechanics and reservoir engineering, but it can also be based on the interpretation of field measurements such as microseismic monitoring4. To illustrate simply the potential of PSM, we suggest a series of schematic cases: A decrease in pressure due to oil or gas production induces compaction and then fracturing of the reservoir layer in some field areas, it can temporarily increase production in these areas. If this is so, an equivalent amount of oil or gas is potentially available in every other reservoir area where compaction is delayed because of the reservoir properties, the well pattern or the production history. PSM can help to map these areas if the instrumentation covers the whole field. Fault displacement is induced in such a way that fault permeability could be increased/created making water breakthroughs possible or, inversely, it could be impaired. The mapping of this fault can help to understand fluid paths and so define a different management of injection and production wells. Improvement of the reservoir permeability is scheduled using hydraulic fracturing: it is interesting to estimate the fracture extension and azimuth, during a "minifrac" operation, in order to confirm the interest in and to adapt the design of the subsequent propped fracture. Thermal fracturing due to massive water injection is suspected: PSM can contribute to map injected-fluid paths. The measurement and interpretation of the microseismic activity can then contribute to locate and identify mechanical active areas. Then it can help to understand the reservoir mechanics and can contribute to significant improvements in production. In the particular case of underground gas storage, the state of stress is also affected but we have to consider an increase of the initial pore pressure instead of a decrease. We also have to consider cyclic variations during the year period: gas is stored during Spring and Summer and withdrawn during Fall and Winter (Figure 1). Here the interest of PSM is to contribute to the characterization of the mechanical response of the structure to the loading during initial reservoir fill-up and over time after several cycling periods. It can also help to map possible displacements on sub-seismic faults and to optimize the maximum storage pressure in combination with laboratory mechanical measurements performed on cap-rocks. For underground storage performed in low permeability matrix but highly fractured reservoirs, PSM can contribute to map fluid paths and then to design the well pattern and exploitation.

Figure 1: Exploitation of an underground gas storage facility after initial reservoir fill-up (Cr-la-Ronde, France).

Short term and long term applications of PSM


To understand well the interest of our research program we can define two types of PSM experiments on the basis of their duration, which depends on the focused application itself. Short term applications of PSM correspond to fluid injection performed on one or a limited number of wells and for a short period of time. The advantage of these conditions is that a temporary measurement can be scheduled, the instrumentation can be rented and the engineering time associated with acquisition and processing of data can be easily estimated. The most representative case is hydraulic fracturing. Here, the interest of PSM is to map the fracture 5,6 created or solicited by the fluid injection . The location and time of occurrence of the microseisms are more or less known: close to the well perforations and a few hours or days after the injection. The challenge is to record the microseismicity and to process the data in order to map the fracture development and its geometry in real time. The best conditions are offered when observation wells close to the treated well are available and effectively instrumented with three-component geophones (or accelerometers). But, unfortunately, such conditions are not representative and generally just one observation well or only the treated well can be used to record the microseismicity. The instrumentation will depend on each case both having advantages and drawbacks. In the treated well, you can only operate during the "minifrac" operation with a specific slim hole acoustic tool after the injection but for the moment only one prototype tool is available (equipped with one three-component sensor level). But it is then possible to record simultaneously the downhole fracturing pressure. Long term applications of PSM (Figure 2) are another challenge. No compromise can be made with the exploitation activity; it is necessary to work with it as it is. In other words, you can sometimes stop production on a well for a few hours or a day but for no more! As measurements can be performed over several years it is necessary to manage instrumentation updates and to integrate the knowledge acquired over time. For a long term survey, in particular, you have to manage a large number of problems which can generate data files that do not correspond to seismic data. The better the background is integrated, the faster it is possible to interpret the relevant microseismic information. To benefit from microseismic data, production data and any other information about what occurs at the field exploitation level are required. A partnership between the field operators and those in charge of the data processing and interpretation is thus a key factor to success. Over a long period of time it is important to maintain such a relationship.

Figure 2: The long term Passive Seismic Monitoring concept.

Field experiments
Both underground gas storage facilities are located in the Paris Basin, France. In both cases, the reservoir is an anticline structure and it consists in a series of argillaceous and sandstone layers located between 800 and 950 m depending on the site. The reservoir temperature is close to 40C. Gas is stored in high porosity and high permeability sandstone layers. The first site had been in use for many years when measurements were performed whereas the second was not being used at the time PSM started. From the downhole instrumentation point of view, the first site corresponds to the feasibility phase and the second one to the first test application.

The Germigny-sous-Coulombs experiment


The PSM research program started in 1991 on the underground gas storage facility located at Germigny-sous-Coulombs. On this site, a feasibility experiment started on the development of permanent downhole geophones to be used for repetitive well seismic. A vertical observation well, located at the reservoir edge, was equipped with three permanent downhole geophones fixed on the tubing (2 7/8-in.) and coupled to the casing wall (7 in.). Benefiting from a completion recover, two sensors were located above the reservoir level and one below. Each sensor was pre-amplified and connected to a hepta-cable. The well head was adapted to allow for 21 connections. During Winter 1991-1992, we recorded the microseismicity associated with gas withdrawals. The main difficulties encountered, dealt with the PC hard disk autonomy and back-up performances of the acquisition system (a commercial one). At this moment, the analog to digital conversion was performed in the PC. This implies locating the PC close to the well head. Due to incompatibility between the acquisition system and the well cathodic protection instrumentation, we were only able to record two weeks per month in alternation with the well protection because of the induced noise level. Despite these precautions, a large amount of files corresponding to electrical ambient noise perturbations were stored and only a few microseisms were recorded. Nevertheless, very interesting results were obtained for both the PSM and repetitive seismic applications7. For PSM in particular, we recorded small microseisms within a radius of 75 m around the instrumented well although we were at a distance of more than 1.3 km from production wells.

The effect of the exploitation on the sites mechanical behavior was also suggested by travel time variations in the overburden between two repetitive seismic measurements (repeated VSPs). Also, taking into account the geological and petrophysical properties of this site (unconsolidated sandstone reservoir layers with up to 20 % of porosity and 1 D of permeability), we were quite confident in the potential of PSM for other types of formations, especially in the case of fractured ones and generally those with more conventional petrophysical properties. In 1992, the development of permanent downhole geophones was confirmed and it was decided to instrument a new gas storage facility. This decision made it possible to continue the studies on well repetitive seismic and PSM.

The Cer-la-Ronde experiment


With this new underground facility we had the opportunity to start our measurements before the initial gas injection began. In June 1993, two vertical observation wells (CE112 and CE4) were equipped with permanent downhole geophones supported on the 3-in. tubing and coupled to the 7 in. casing wall (Figure 3) with respectively 15 and 10 levels of geophones. The sensors were placed during the well initial completion increasing the rig time to only 12 hours. Sensors were connected in single ended mode using three hepta-cables and here the cathodic protection was modified to allow for a permanent recording. As both wells were mainly devoted to repetitive seismic applications (VSPs), the antennae of sensors were located at half the reservoir depth. Mono-component geophones were used except for two of the three deepest sensors of the top instrumented well (at 660 and 680 m depth) where three-component geophones were used. These two sensors were used for PSM but, taking into account their distance from the reservoir layer, a logging acoustic tool was added in an adjacent observation well (CE12) close to the reservoir at depths ranging from 910 m to 958 m over time. In October 1993, gas injection started on the main top and, in 1995, gas was also injected on the secondary top. PSM was operational since August: at least three sensors were connected to the same acquisition system but equipped with a new PC. The underground facility structure being a new one, technical inspections were regularly performed. This contributes to have a high functioning rate, making excellent conditions to monitor the initial reservoir fill-up. This system was operational from 1993 to 1998. Taking into account the important activity on the site because of its development and the connection mode imposed by the number of sensors per antenna, we recorded a large amount of files: up to 24,000. The major part corresponds to particular pressure effect at wellbore (up to 250 microseisms per day sometimes), another part corresponds to electrical perturbations, some data correspond to well interventions and less one percent can be considered as microseisms not located at wellbore. A maximum of two events of this type has been recorded in one day. Results as discussed later were very interesting but the position of well CE112 sensors was not ideal for PSM taking into account the geology, wave refraction and especially the bad signal to noise ratio because of the single ended connecting conditions. But on the basis of what was observed, the increase of the number of sensors devoted to PSM was decided. In 1995, a vertical production well (CE22) drilled on the secondary top was equipped with three preamplified levels of three-component downhole geophones located at 840, 888 and 915 m depth. A tubing diameter reduction on a 90-meter length was done to allow the positioning of on-tubing sensors between the 3-in. tubing and the 7 in. casing. This solution appeared as a good compromise to maintain optimal production flow rates. This well was not connected to the surface network before September 1996 and PSM started at the same time on this top. A new commercial acquisition system was used which made a remote access to the acquisition PC possible. For different reasons, a very large number of files (13,000) was recorded during 1996 and 1998 including a large proportion of non-microseismic data files due to production noise but also to the acquisition system itself. In 1998, a deviated production well (CE18) drilled on the top of the central structure was instrumented such as well CE22 was. At this moment, we extended the second acquisition system to this reservoir area and we removed the first one, unfortunately this option was not satisfactory over time (significant decrease of the running rate of the acquisition system) due to the system itself. Nowadays, we plan to

replace it with the new acquisition system developed by IFP which is able to manage a large amount of channels and to answer both passive and VSP applications with the same equipment as described hereafter. In parallel we benefit from improvements in the acquisition of production-surface data (pressure, flowrate, injected or withdrawn gas volumes, etc per well) that became available every 5 minutes instead of every hour or day.

Figure 3: Photograph and schematic drawing of a permanent downhole geophone. The schema has been updated to take into account downhole digitalization development.

Comments relative to instrumentation


Instrumentation is a key point in PSM that requires a sufficient number of sensors to ensure redundancy in the measurements and precision in microseism location. Advantage of downhole sensors is evident as they make the recording of high frequency microseismic information possible, generally not detectable at the ground surface because of attenuation. As ground is also a low-pass filter for wave propagation, a higher resolution in the description of source mechanism can be obtained. But location of sensors in the well is a compromise between completion design, well cementing, geology and, of course, budget. At least 5 wells have been instrumented with analog downhole on-tubing geophones making the observation of phenomena non-detectable using surface instrumentation possible. In terms of instrumentation time life, excellent results have been obtained. A few problems have been encountered but they were not associated with the downhole sensors themselves but mainly with already commercial complementary instrumentation. At Germigny-sous-Coulombs sensor electronic was destroyed during a summer thunder storm, also the instrumented wells of Cr-la-Ronde were systematically connected to an underground lightning protector. In 2002, all the Cr-la-Ronde sensors were still operational except for one in the CE112 antenna that had never been operational and one in CE22 because of a weak well-head connector. A good signal to noise ratio was obtained with 60 dB pre-amplified geophones using one hepta-cable per level (Figure 4) but a total gain of 111 dB was sometimes used with the initial 12 bits A/D converter.

Figure 4: Microseism recorded during initial reservoir fill-up (October 1993 March1994). In fact the main problems were due to the surface acquisition systems. Some problems were due to the limited performances of PCs at the beginning of the survey, but the main problem was in the triggering technique that was not able to make the difference between noise, electrical perturbations and a microseism. We also had to manage large numbers of files for a limited number of microseisms. This was time consuming at the processing level and constituted a risk in its evaluation but we do not loose information except in the last three years. To reduce this problem, remote access to the system is necessary but the most critical point is to improve the acquisition techniques by integrating automatic advanced detection analysis. Taking advantage of this long term experience, IFP has developed a 20-bit acquisition system designed for both PSM and VSP applications with specific software to manage: up to 24 three-component geophones per telemetry cable, 18 of these 24 levels can be recorded at a 1 ms sampling rate per seismic channel -with some restrictions for great depth; up to 10 telemetry cables making possible to work on 10 wells; additional sensors (pressure, temperature, etc). Main restrictions are only due to temperature (limited to 125C) and tubing-to-casing diameter ratio. For PSM, a library of criteria can be developed to optimize the real-time sorting, in order to reduce the number of recorded non-seismic files. Validation tests of this system were performed in December 2000 and April 2001 on sensors located in well CE18 and using a digital acoustic tool in well CE12.

Comments relative to data processing and interpretation The need of a global and structured approach
These experiments were very fruitful to learn about the processing and interpretation of data recorded during long term PSM applications. The first 3 years of acquisition were processed manually but the use of a second acquisition system in 1996 made this work time consuming. We spent a lot of time on non-seismic data and soon the main challenge was to optimize the engineering time devoted to data processing. Nevertheless it was very efficient training to be aware of the problems associated with long term PSM, we should say "a necessary step" for background acquisition. In 1998, we decided to develop a "tool" that would be able to process the amount of stored data files in order to extract automatically microseisms from them. We were also convinced of the importance of integrating the site background and feed-backs in the processing and interpretation process. But to totally benefit from the microseismic information, relationships between microseismicity and site exploitation also need to be studied. In this context, an evolving software platform has been developed to process and characterize microseismic data in relationship with exploitation data and site mechanical response. More than a software, it is in fact an open approach8 where the processing can

be improved over time and where non-seismic data such as production and exploitation ones can be easily managed and updated. Figure 5 represents the corresponding schematic workflow: PSM, production and exploitation raw data are sent to the platform software such as the results of reservoir modeling (estimated reservoir pressure) and the mechanical relationships obtained after geomechanical modeling.

Figure 5: Schematic workflow for long term PSM applications. Presently, the processing/interpretation module for microseismic data assumes: conventional signal processing, high-precision automatic first break picking and polarization analysis; source location; standard source parameters. In addition, site parameters, production data and indicators of site exploitation are also introduced (or extracted from the other input data) to complete the characterization of microseisms. In order to optimize the sorting of the files, the characterization is performed at three levels: the "seismic channel", the "sensor" and the "event" levels. For each level, a series of seismic attributes is computed if the event is qualified as a relevant one. Indicators computed from the production and some exploitation data are also associated to the event. Production data being also used for reservoir and geomechanical modeling, the corresponding results are considered in data interpretation (Figure 5). A specific data analysis module has been integrated for visualizing, analyzing and clustering microseismic data. Initial results of this work make it possible to obtain the same classification results for a series of microseisms as the one performed "manually". Main relevant attributes are the length of the signal, its dominant frequency, the energy and the wave content. Main indicators are the reservoir pressure and a parameter derived from the volume of injected gas. The platform makes possible to consider site-associated specific processing, developed on the acquired experience: for example to automatically identify signals associated with production or to compute the estimated reservoir stress variations at microseism occurrence.

Microseismic data interpretation


Our observations differ between sites and are due to different factors: site geological properties, relative well locations and the time at which we did the survey in comparison with the exploitation history. At Germigny-sous-Coulombs, the site had been operational for approximately 10 years when PSM started. Several cycles of exploitation had already been done. The instrumented well was quite distant from exploitation wells. The structure is not considered as a faulted one. Microseisms were located mainly at the top and the bottom of the pressurized reservoir layers. They suggest some communication paths between lower reservoir layers. Their magnitudes were close to -2. These microseisms were in correlation with significant variations of withdrawn volumes of gas7. On this site we consider that PSM can be used to control the pressure effect on the layers. If the survey could also be performed during the gas withdrawal period, it would be interesting to study eventual relationships between microseismicity and sand/water production. At Cr-la-Ronde, conditions were different. The geological structure is faulted, the instrumented wells are located at the top of anticlines close to injection wells and PSM started before the first injection of gas. During the initial reservoir fill-up (October 1993 - March 1994), some particular groups of microseisms were identified. Most of them show a clear correlation with the reservoir pressure evolution (Figure 6). A particular series of microseisms recorded until March 1994, seems to be associated with initial reservoir fill-up. Location of these microseisms (not located only at the reservoir level or in surrounding layers), their specific correlation with reservoir pressure and more advanced parameters, and also the fact they occur in a discrete manner over time (a maximum of 2 events per day), suggest a mechanical rearrangement of the reservoir structure induced by the initial gas injection. This could have occurred on a subseismic fault approximately in the North-South direction (Figure 7). Taking into account the very small magnitude of the events, this would not have been observed if surface instrumentation had only been used.

Figure 6: Relationship between reservoir pressure and two particular groups of microseisms: upper part: pressure variations and recording rate versus time, lower part: distribution of microseisms versus time.

Figure 7: Location of the E-group microseisms (dots) recorded during initial reservoir fill-up (October 1993 - March 1994). After March 1994, this particular group of microseisms disappeared and was replaced by other different groups (in terms of frequency content, wave content and length) whose origins have to be determined or confirmed. Some events up to 16 seconds long with frequency below 20 Hz were recorded, they do not seem to be associated with shear mechanism. Because of the limited number of sensors at that time and the fact they do not present a succession of compressional and shear waves, their location is not possible. As already mentioned, most of the events are generally correlated with significant reservoir pressure variations. The history of reservoir exploitation -especially the way wells are used over time- seems to be another significant parameter. In this work, the limited number of sensors devoted to PSM (only four on the same area) constituted a penalizing factor. A series of 6 to 12 three-component sensors distributed in a minimum of four wells to cover the area to be monitored is recommended. In 2002, the processing of the data recorded at Cr-la-Ronde will be updated taking into account the experience acquired. Utility functions have been introduced in the platform software to reduce the processing time. The analysis of the pertinent attributes will constitute an important part of this work. At the moment, it is probable that the information inside microseismicity has not completely been exploited.

Conclusion and perspectives


The field experiments reported here highlight the potential of PSM to characterize the mechanical behavior of a gas storage reservoir. The results they deliver can be considered in terms of dynamic properties of the structure but also in terms of reservoir integrity over time. PSM can be considered as an emergent technique although the mapping of microseisms has been performed for more than 25 years now, especially in the case of Hot Dry Rock geothermal experiments and hydraulic fracture mapping. Instrumentation has been seriously improved in recent years and this will continue with the arrival of new technologies such as optical sensors and microsensors. But for the oil and gas industry, the long term use of this technique is quite uncommon. For the moment, only a few small-sized companies are effectively involved in the permanent monitoring as the market is still not defined. This situation may change in the next few years with the development of the concept of the intelligent well and the need for information on deep offshore fields, but market can also grow with the necessary improvement of the production on mature fields where low cost solutions can be proposed.

For the moment, PSM may be a good way to characterize mechanical and hydraulic reservoir behavior especially in faulted/fractured reservoirs if a sufficient number of sensors are used. From the environmental point of view, PSM may also be used for waste fluid disposal or sequestration, such as re-injection of cuttings, and also for the decommissioning of wells. A combination of permanent ontubing sensors, well cemented sensors and logging acoustic tools can be an excellent solution to minimize costs while increasing the monitored area. The main objectives of future work are the validation of this approach to operate in real time in order to develop/improve methodology for other applications of PSM. We would like to contribute to a better characterization of fractured reservoirs.

Acknowledgements
The authors would like to thank the Institut Franais du Ptrole and Gaz de France for permission to publish these results. They are grateful to the teams of both companies involved in the different research programs and those in charge of the gas storage facilities of Germigny-sous-Coulombs and Cr-la-Ronde.

References
1. Maxwell S.C., Young R.P., Bossu R., Jupe A. and Dangerfield J., Microseismic Logging of the Ekofisk Reservoir, paper SPE/IRSM-47276, Eurock98, Trondheim, Norway, July 1998. 2. Grasso J.-R., Mechanics of seismic instabilities induced by the recovery of hydrocarbons, PAGEOPH, Vol.139, No.3/4, 1992. 3. Adushkin V.V., Rodionov V.N., Turuntaev S. and Yudin A.E., Seismicity in the Oil Field, Oil Field Review Summer 2000. 4. Maxwell S.C. and Urbancic T.I., The role of passive microseismic monitoring in the instrumented oil field, The Leading Edge, June 2001. 5. Bell M., Kraaijevanger H. and Maisons C., Integrated Downhole Monitoring of Hydraulically Fractured Production Wells, paper SPE-65156, SPE European Petroleum Conference, Paris, France, October 2000. 6. Rutledge J. and Phillips W.S. Reservoir, high resolution microseismic imaging of a hydraulic fracture, Carthage Cotton Valley gas field, East Texas, submitted to Geophysics, March 2001. 7. Deflandre J.-P., Laurent J., Michon D. and Blondin E., Microseismic surveying and repeated VSPs for monitoring an underground gas storage reservoir using permanent geophones. First Break, Vol.13-No4, April 1995. 8. Deflandre J.-P., Delaplace P. and Huguet F., Permanent Passive Seismic Monitoring for Reservoir Management: the SICSTM Approach, EAGE/SEG Research Workshop on Reservoir Rocks Pau, France, April-May 2001.

You might also like