SOLUTION MINING RESEARCH INSTITUTE 105 Apple Valley Circle Clarks Summit, PA 18411, USA

SOLUTION MINING RESEARCH INSTITUTE 105 Apple Valley Circle Clarks Summit, PA 18411, USA Technical Conference Paper Telephone: +1 570 585 8092 Fax: +1 570 585 8091 www.solutionmining.org Cavern Integrity and performance management at Geomethane underground storage Mehdi KARIMI-JAFARI and Arnaud REVEILLERE, GEOSTOCK, France Charles FRASSY, STORENGY, France SMRI Fall 2014 Technical Conference 29-30 September 2014 Groningen, The Netherlands

Solution Mining Research Institute Fall 2014 Technical Conference Groningen, The Netherlands, 29-30 September 2014 Cavern integrity and performance management at Geomethane underground storage Abstract Mehdi KARIMI-JAFARI and Arnaud REVEILLERE, GEOSTOCK, France Charles FRASSY, STORENGY, France Located in Manosque, France, Geomethane gas caverns are used for seasonal and peak gas storage since the early nineties. These caverns have been monitored by a comprehensive survey program including gas inventory follow-up, sonar survey, well pressure and temperature logging, cavern bottom sounding, micro-seismic monitoring and subsidence survey. The gas inventory follow-up method applied on Geomethane caverns consists in thermodynamic simulation of gas operations together with downhole measurements such as cavern sonar and pressure/temperature logging. Many advantages can be taken of this follow-up method: A more accurate continuous assessment of stored gas volume, without requiring a permanent downhole measurement tool. Prediction of cavern capacity and performance Coupling thermodynamical behavior of cavern with rock mechanical modeling to assess the cavern stability Optimization of storage performance In this paper the Geomethane caverns monitoring and surveys are presented. Thermodynamic simulation of gas operations is introduced and some results are discussed. Key words: Geomethane natural gas storage, Cavern follow-up, Monitoring program, Gas operation, Inventory verification, Thermodynamic simulation 1. Introduction Geomethane GIE, a company equally owned by Geosud (subsidiary of Total, Geostock and Ineos) and Storengy (GDF Suez), was created in 1989 for underground storage of natural gas in Manosque, France. Storengy sells storage services to third parties and operates the site. Geostock delivers the technical and administrative support services. As presented in some SMRI papers (Colin and de Laguerie, 1990; Colin and You, 1990; de Laguerie and Durup, 1994), Geomethane gas storage has a total working gas capacity of nearly 300 Mnm3 and comprises: 7 salt caverns solution mined between 900 and 1500 m below ground, operating since the early 90s. The pressure range at the casing shoe is 60 to 190 bar. surface facilities including the collection site at Gontard and compression, treatment and metering facilities at Gaude. 1

The site continues expending: 2 new gas storage caverns are currently being solution mined and the extension of the existing surface facilities is planned. The location of the caverns is illustrated in Figure 1. Géométhane. Figure 1: Location of the Geomethane caverns. GA and GB leaching is expected to be finished by 2016, other caverns are in gas operation since the early 90s. Located in a protected area of the Luberon Regional Park, the facilities received special attention regarding environmental protection, in close cooperation with the Park and local authorities. As any other industrial activity, gas storage in salt caverns comprises risks that should be managed. The risk management process aids decision making by taking account of uncertainty and the possibility of future events or circumstances (intended or unintended) and their effects on agreed objectives. The International Standard ISO 31000 (ISO, 2009) proposes a framework dedicated to the risk management. It consists of the following successive steps: setting the context, assessing the sitespecific risks (risk identification, analysis and evaluation), and implementing treatment options when the risk level is assessed as unacceptable. Throughout this logical system, a monitoring and review is conducted, as presented in Figure 2. Monitoring should enable early detection and a characterization of the potential deviations from the expected behavior. 2

Figure 2- Risk management general concept, from ISO 31000 (ISO, 2009) From this figure, also note that similarly to constant monitoring, the ISO 31000 risk management also considers constant communication during all stages. In this perspective, monitoring and third party verifications demonstrate the transparency of the site operation among stakeholders. Geomethane caverns are followed by a comprehensive monitoring and survey program in order to manage environmental and safety (stability of salt caverns and their integrity) as well as financial (inventory verification) risks, to provide requested or mandatory information to competent authorities and to ensure optimized gas operation. The monitoring (defined in this paper as continuous data gathering and analysis) and surveys (defined as periodic operations) can be grouped in three objectives: Cavern stability follow-up, which includes seismic monitoring, subsidence survey, cavern bottom sounding and sonar survey Well and completion integrity follow-up, which includes corrosion monitoring and annular tubing pressure monitoring Inventory follow-up including bottom Pressure / temperature logging, cavern thermodynamic modeling from wellhead measurement and sonar survey. The performance assessment (sustainable gas deliverability over time) and optimization of gas operations at Geomethane are annually updated for gas marketing purposes using thermodynamic follow-up results. The monitoring techniques implemented for the follow-up of Geomethane caverns are presented in section 2. The inventory verification methods are reviewed in section 3. The implementation and results of Geomethane inventory follow-up are presented in section 4. 2. Geomethane caverns follow-up 2.1. Stability follow-up 2.1.1. Micro-seismic monitoring As discussed in Renoux et al. (2013), micro-seismicity is a good indicator of rock physical and mechanical change. Setting up a permanent array of seismic sensors at shallow depth provides useful 3

data to detect underground cavern instabilities as well as external events associated to structural activities. The micro-seismic network at Manosque has been set up for monitoring both Geosel liquid storage site and Geomethane gas caverns. It was developed more than twenty years ago and includes 7 three-axis geophones cemented in shallow boreholes at about 45 m depth. The Manosque network allows detecting on-site events (mainly identified as local instability related to rock spalling at caverns wall during solution mining process and storage operations) as well as off-site seismic activities. The utility of the micro-seismic monitoring to predict a possible cavern collapse was experimentally studied by INERIS in the north of France. A high resolution micro-seismic network installed in a brine field site (at north east of France) recorded some precursory signs before the cavern collapse and highlighted the major role of overlying layers on the cavern stability (Cao et al, 2010). 2.1.2. Subsidence survey If any significant cavern closure happens due to the salt creep or mechanical instability, remarkable surface subsidence or ground deformation can be observed at ground level. Thus measuring the surface subsidence is a good indicator of caverns geometry evolution with time. At Manosque a benchmark network consisting of more than one hundred points spread over Geosel and Geomethane sites has been set up to measure vertical ground movement since 1992. Subsidence surveys at Manosque are carried out every 5 years measuring the elevation of all benchmark points. The surface subsidence is hence calculated as the difference of two subsequent campaign measurements. The surface subsidence at Manosque has been reported as small as the measurement uncertainty and no significant down ward tendency in surface level all over the area has been observed since the nineties. This result is consistent with the fact that the creep closure rate of Geomethane caverns is extremely limited. 2.1.3. Sonar Survey One sonar measurement is run every ten years in each Geomethane cavern. The periodic sonar survey aims at checking caverns contour evolution. The possible instability events at cavern wall and the rate of cavern creep closure are deduced from the comparison of successive sonars. A sonar survey in a gas cavern is always completed together with a downhole pressure/temperature measurement to assess the total gas inventory in the cavern as well as the physical cavern volume. Geomethane caverns show a very small volume loss rate about 0.1-0.2 % per year. Different reasons are at the origin of such small volume loss: Caverns have mostly operated in the upper part of the designed pressure range (operated between 120 and 190 bar at cavern mid-depth), Rock Salt at Manosque has a small to medium creep ability (Vouille s index 1 is around 1%), The salt formation has a relatively high percentage of insolubles (20 %). 2.1.4. Cavern bottom sounding This survey is run in each wellbore every 3 years on the occasion of subsurface safety valve replacement. The cavern bottom sounding measures any change in the sump level depth. Two mechanisms can be considered to be at the origin of any rise of the sump level in a gas cavern: upward displacement of the cavern bottom due to the salt creep rock slabs falling to the cavern bottom. To identify the actual cause of sump level depth variations, if any, further investigations are necessary. 1 The Vouille s index is used for a comparison of the creep potential of different salt formations. It represents the axial creep strain after one year for a 10 MPa deviatoric stress. 4

For Geomethane caverns, no significant change in the sump level depth has been observed. 2.2. Well Integrity follow-up Gas completions of Geomethane wells have been equipped with a production tubing and a packer sealing off the annular space between the last cemented casing and the inner production tubing. This well architecture provides a double protective shield against gas migration. The well integrity can be jeopardized only if both the casing and the packer are affected. Thus, the common practice to check and survey the well integrity consists in controlling the casing corrosion and lack of tightness of the packer (permanent control of the annular pressure). 2.2.1. Corrosion monitoring Well casings could be subject to external corrosion (in contact with rock formations or ground water). In general, the lack of good cementation is known as the most common cause of the external corrosion. Different tools and methods have been developed to detect any internal or external corrosion of the casing such as ultrasonic logging, visual inspection etc. A preventive and cost effective method is cathodic protection system which can be installed to protect casings from the external corrosion only. Geomethane wells are protected by an impressed current system. It consists in coupling the well to the negative pole of a direct current source, while the positive pole is coupled to an anode bed. Anodes are generally installed in vertical holes at a given distance from wells. The electric potential and current are measured at wellheads on a periodic basis to ensure the protection. 2.2.2. Annular pressure monitoring The annular pressure at wellhead is recoded on a daily basis to survey the packer tightness. A pressure build-up in the annular space is an indicator of gas migration through the packer. One of the Geomethane wells has been identified with a loose packer from the beginning of gas operations. This situation is perfectly under control and the well has been subjected to many tests and surveys. To limit the packer leakage, it has been decided to keep limited the pressure difference between the annular and tubing. This restriction implies a dynamic management of annular pressure during gas operations. 2.3. Gas inventory follow-up The first estimate of the gas inventory in a cavern is obtained by metering the quantity of gas injected in and withdrawn from the cavern. Gas metering of Geomethane storage is checked by thermodynamic follow-up that enables to detect and correct possible deviation of the gas inventory. The thermodynamic analysis of caverns is continuously updated using the surface and downhole data on the gas pressure and temperature. A brief review of inventory verification methods is discussed in section 3. The Geomethane follow-up method is described in more detail in section 4. 2.3.1. Surface measurements The following continuous surface measurements are used for inventory verification: The flow metering of the gas injected in or withdrawn from each cavern The wellhead pressure measurement, which gives a reliable estimation of the cavern pressure in both static or dynamic conditions 5

Wellhead temperature measurement, which provides an input data during injections and enables to verify the cavern temperature during withdrawals Gas composition, which is then used to derive the cavity gas properties (particularly gas compressibility factor Z) 2.3.2. Pressure/temperature logging Downhole pressure/temperature measurements are performed in two ways: Temperature survey method: Periodic temporary logging of pressure and temperature from the wellhead down to the cavern bottom every 3 years per cavern Temperature monitoring method: Permanent downhole tool continuously recording pressure and temperature at a specific depth (casing shoe depth or cavern mid-depth) over a period of time (e.g. one year) Temperature survey method provides a one-time measurement of temperature and pressure in the well and cavern, whereas the second method enables to record the downhole temperature and pressure evolution during gas operations over a period of time. Besides its use for inventory check, pressure/temperature log also indicates the gas/brine interface depth. The gas/brine interface aims at determining the gas free volume evolution in the cavern. Normally the interface depth falls down with time as the injected gas is much drier than the withdrawn gas. However, any cavern closure or rock falling, hiding this effect, can lead to the interface rise. 2.3.3. Thermodynamic analysis Thermodynamic analysis of the cavern data aims at checking the main uncertainties such as the cavern physical volume (during the calibration of the model) and the gas inventory throughout the operation of the site. Verification of gas inventory in underground storage is one of the most important economic issues. It is understood by the concern about contractual commitments on gas deliverability as well as the storage gas tightness. That is why the inventory must be periodically or even continuously rechecked. Using the above-mentioned surface measurements, temperature logs and sonar surveys as inputs, the thermodynamic simulation of a gas cavern is a constant follow-up work that enables to: Check and eventually adjust the cavern gas inventory, Provide valuable information for the operating purposes by predicting the withdrawn gas temperature (pipeline limits or hydrate formation limits), Predict the cavern temperature evolution for performance assessment (including rock mechanical aspects) 3. Review of inventory follow-up techniques The gas inventory can be either directly measured (book inventory based on flowmeters) or deduced from the Equation of State applicable to the gas in the cavern. 3.1. Book inventory deduced from surface metering The cavern inventory at a time is equal to the algebraic sum of injection/withdrawal quantities from the beginning of gas operations: I Q Equation (1) The uncertainty associated with this method is strictly increasing over time and can reach unacceptable levels even if each individual metered error may be small (less than 2%): 6

I Q Equation (2) 3.2. Equation of state methods The other alternative is to deduce the gas inventory in the cavern from the following gas equation of state, which as reasonably assumed to be verified in the cavern at any time:. Equation (3) Where I, the inventory in normal cubic meter P, cavern pressure in bar V, cavern free volume in cubic meter Z, the compressibility factor T, absolute temperature in the cavern in Kelvin P 0, atmospheric pressure at normal condition (=1.013 bar) T 0, absolute temperature at normal condition (=273.15 K) One can distinguish three applicable methods, depending on whether the parameters on the righthand side of the equation are known with reasonable accuracy or to be estimated: Volumetric method, if all parameters including cavern volume are known Depletion method, where a differential version of equation (3) is used in order to estimate the cavern volume as well as the cavern inventory Thermodynamic method, where the downhole temperature is estimated using a thermodynamic model of the gas in the cavern coupled with a model of heat exchange with the surrounding rocks since the start of the leaching. We note that contrarily to the book metering method, the equation of state methods don t rely on a sum over time and Consequently, the associated uncertainty is not increasing over time. 3.2.1. Volumetric method This method relies on measuring all the Equation 3 right-hand side parameters. Even if using accurate downhole measurements, the error on the inventory will be at least in the same order as the cavern volume error, which is usually estimated at 5% from sonar surveys. The method can also be improved by discretizing the computation on the vertical axis, as proposed by Skaug et al. (2010). This enables to consider the volume as function of depth and to use the gas properties at this depth temperature and pressure. This does not raise major practical complications for the measurement since the log can easily measure temperature and pressure from the bottom to the top of the cavern. 3.2.2. Depletion method If the cavern physical volume is not accurately known, measuring downhole parameters before (subscript 1) and after (subscript 2) a withdrawal, one can calculate the cavern free volume and initial inventory as follows: I V. 7. Equation (4)

Where Q, is the metered withdrawn gas. The uncertainty associated with this method is not only depending on the metering error, but also highly influenced by the relative quantity of withdrawn gas. The larger the quantity of withdrawn gas, the smaller is the error on inventory (Tek, 1993; Nelson & Van Sambeek, 2003). 3.2.3. Thermodynamic simulation method As mentioned, the drawback of these two previous methods is that they require cavern temperature measurements, either a one-time measurement (P, T log) or a costly downhole gage. The thermodynamic model of the gas in the cavern and in the well and the heat transfer with the surrounding rocks enables to calculate the cavern temperature based on the surface measurements, history of injections and withdrawals, and thermal properties of the system. 4. Gas inventory follow-up of Geomethane caverns The gas inventory of Geomethane caverns is evaluated on a daily basis by thermodynamic simulation method based on the wellhead pressure and temperature, calculating the cavern pressure/temperature during gas operations. Thermodynamic simulations of Geomethane caverns are carried out using SCTS software (Nieland, 2004). 4.1. Model calibration The thermodynamic model of the cavern must be calibrated before being used for the prediction of cavern pressure/temperature during gas operations. The overall idea of that calibration is that the gas equation of state (3) in the cavern should be verified at any time. The calibration aims at setting all the parameters that are used for deducing the temperature in the cavern, and other parameters of the gas equation of state to calculate the inventory. Once these parameters are precisely adjusted, the only remaining parameter that should eventually be adjusted by pressure history matching is the cavern inventory. The model calibration relies on four main parameters: Cavern free volume Initial rock temperature at cavern mid-depth Cavern temperature at the end of first gas filling Cavern heat exchange ratio (cavern volume to area ratio) 4.1.1. Cavern free volume The cavern free volume is calibrated against the cavern pressure difference prior to and after a significant gas withdrawal or injection. This value must be consistent with the sonar survey measurement. The difference between sonar volume and thermodynamically calibrated volume is generally less than 5%. The confidence level on the calibrated volume value can be higher than the sonar measurement if several calibrations have been performed upon large withdrawal/injection phases. The dewatered cavern volume at the end of first gas filling of the cavern gives an additional convenient benchmark value of the cavern free volume. 8

Geomethane caverns volumes have been thermodynamically calibrated and have been compared to sonar measurements. For six caverns, the difference between measured and calculated volume was below 5% (ranging from 1% to 4%). The 7 th cavern showed a large difference between the calculation and the sonar measurement (close to 11%) as presented below: Cavern Sonar volume (m3) Thermodynamically Calibrated volume (m3) Debrined cavern volume (m3) EJ 507 199 565 000 561 300 As shown in Figure 3, the adjusted volume greatly improves the fit, and the obtained cavern volume is closer to the debrining volume. The large discrepancy between sonar and calibrated volume may be explained by the very irregular shape of the cavern. Figure 3- Cavern volume calibration from wellhead pressure 4.1.2. Initial rock temperature The initial rock temperature is measured by a temperature log. The gas in the wellbore is in thermal equilibrium with surrounding rock and indicates the rock temperature. On contrary, the gas in the cavern is not necessarily in equilibrium with rock formation. The thermal gradient observed in the well within the salt formation can be extrapolated further down in order to estimate the rock temperature at the cavern mid-depth (cf. Figure 4). 9

Figure 4- Pressure/temperature log 4.1.3. Temperature after 1st gas fill Cavern temperature at the end of first gas filling constitutes an initial reference point in the thermal history of the cavern. This temperature is mainly influenced by the cavern leaching conditions that have cooled down the rock below its initial temperature. A temperature log must be run at the end of first gas filling to provide data for this calibration. The main parameters influencing the cavern initial temperature (at the end of gas filling) are the following: Leaching water temperature Leaching duration Stand-by period between the end of leaching and gas first filling Duration of first gas fill Varying only one of the above parameters (mainly leaching duration) could result in an accurate fitting of the initial cavern temperature. 10

4.1.4. Cavern heat exchange ratio Cavern heat exchange ratio is a coefficient which enables to adjust the heat flow rate between the gas and surrounding rock. In SCTS software, this parameter is adjustable by the cavern volume to area ratio (V/A). This parameter has a strong influence on the cavern temperature evolution during gas operation. It is therefore calibrated against all cavern temperature measurements during gas operations (see Figure 5). Figure 5- Cavern temperature adjusted against downhole measurements An initial value of V/A can be calculated from sonar data. Actually the cavern area measured by sonar has a large uncertainty because of the cavern wall irregularities. In general, the contact surface between the gas and rock salt is much larger than the cavern surface area derived from sonar and the calibrated V/A value are smaller than the sonar based V/A. The calibrated V/A ratios of Geomethane caverns are within the range of 4 to 8 m. 4.2. History matching When there is a good accordance between downhole temperature measurements and modeling predictions (discrepancy less than 2 C), the model calibration can be considered to be achieved. As shown in Figure 6, the cavern pressure and temperature calculated at the cavern mid-depth are in good accordance with continuous downhole measurements. 11

Figure 6- Comparison of simulations and downhole measurements The cavern gas inventory can be re-assessed by matching the quantity of gas injected or withdrawn against the observed wellhead pressure during gas operations. As shown in Figure 7 for a typical Geomethane cavern, the metered gas injections / withdrawals (without any correction) did not match with the observed wellhead pressure. Therefore, Injections and withdrawals have been adjusted to fit the calculated wellhead pressure with the measured one (see Figure 8). These adjustments hence result in a continuous calibration of the inventory throughout the gas operations. Figure 7- Wellhead pressure calculated using injections / withdrawals as metered (without adjustment) 12

Figure 8- Wellhead pressure calculated using adjusted injections / withdrawals If the metered inventory is smaller than the thermodynamically calibrated one (negative gap) and this difference is growing with time, the likely explanation could be a higher cavern closure rate than what expected and used into the simulations. A non-symmetrical metering error (underestimating withdrawals and overestimating injections) during operations can also be suspected. If the metered inventory is larger than the calculated inventory (positive gap) and the difference is increasing with time, this drift cannot be explained by a growing cavern volume. This result can be related to different causes and needs to be deeply investigated: gas loss (underground or aboveground leak), non-symmetrical metering error (overestimating withdrawals and underestimating injections). The implementation of this follow-up in Geomethane storage since 2010 and the analysis of all survey data have made it possible to identify a growing positive gap between the metered and calculated gas inventory over a long period of time (since the beginning of gas operations) as shown in Figure 9. 13

Figure 9 Difference between metered and adjusted gas inventory Modeling works and investigations highlighted that some anomalies in the surface facilities are at the origin of this drift. A loose valve separating the storage from the gas network has been suspected to be the main cause of the gas leakage into the network for many years. After replacing the loose valve and strengthening surface facility follow-up, the inventory drift has remained broadly stable. 5. Conclusions Geomethane caverns are monitored by a comprehensive follow-up program that aims at the early detection of any potential deviation from the expected behavior. This monitoring program follows the good industry practices and provides the required information to the competent administration. It is coherent with the risk management international standard proposed by the ISO31000. It constitutes key elements to the management of safety and environmental risks (cavern stability and tightness) and financial risk (gas inventory verification). It also participates at the transparency of the information among the stakeholders. The techniques applied in this follow-up and the objectives they serve are summarized in the following table: 14

Objective: Technique and frequency Cavern stability and closure Well integrity Inventory assessment Micro-seismic monitoring Subsidence survey Cavern bottom depth sounding Continuous 5 years 3 years local or global instability Cavern creep closure rock spalling or cavern closure Cathodic protection continuous Casing string Annulus pressure monitoring Continuous Tubing string and packer tightness Sonar survey 10 years Rock spalling and cavern closure Wellhead pressure monitoring Continuous Minimum and Maximum allowable pressure Volumetric method Thermodynamic simulation Downhole temperature measurement 3 years Depletion method Gas metering Continuous Book inventory The thermodynamic simulation involves uncertainty in its input parameters (gas pressure, temperature, injected/withdrawn quantity, composition etc.) and in model predictions. Due to these uncertainties, the absolute accuracy of calculated inventory, at best, cannot be less than 3-4%. The main purpose of the inventory follow-up is to control the relative error growing and to identify operational anomalies or gas leak, if any. This monitoring program has been successfully applied to Geomethane, a mostly seasonal gas storage site. The need for cavern monitoring and survey increases with the frequency and the amplitude of gas cycling. Large volume, high rate and frequent withdrawals result in more significant thermo-mechanical loading on rock salt and require more accurate prediction of storage capacity and performance. The need for gas inventory verification is higher due to measurement uncertainties and increasing errors of gas metering, and this follow-up program is deemed even more necessary in that context. 15

Acknowledgements The authors wish to thank Geomethane for authorizing this publication. References Colin P. and de Laguerie P. (1990), Conversion from hydrocarbons to natural gas storage at Manosque. Proc. SMRI Fall Meeting. Colin P. and You Th. (1990), Salt Geomechanics seen through 20 years experience at the Manosque facility. Proc. SMRI Fall Meeting. de Laguerie P. and Durup G. (1994), Natural gas storage facility at Manosque, France. Proc. SMRI Fall Meeting, Hannover, Germany. ISO 31000:2009 (2009), Risk management principles and Guidelines on Implementation. ISSN 0335-3931 Renoux P., Fortier E. and Maisons Ch. (2013), Microseismicity induced within Hydrocarbon storage in salt caverns, Manosque, France. Proc. SMRI Fall Meeting, Avignon, France. Cao N.T., Klein E., Contrucci I., Daupley X. and Bigarre P. (2010), Large-scale salt cavern collapse: Multi-parameter monitoring from precursor signs to general failure. Proc. ISRM-EUROCK, Lausanne, Switzerland. Skaug N., Ratigan J., Thompson M. (2010), Natural Gas cavern inventory assessment: A New Approach. Proc. SMRI Spring Meeting, Grand Junction, Colorado, USA. Tek M. R. (1993) Inventory verification in salt cavity storage. Proc. SMRI Spring Meeting, Syracuse, New York, USA. Nelson P. E. and Van Sambeek L. L. (2003), State of the art review and New techniques for Mechanical Integrity tests of (gas-filled) natural gas storage caverns. SMRI research project report 2003-2. Nieland, J., D. (2004), Salt Cavern Thermal Simulator version 2.0 User s Manual. RESPEC Topical Report RSI-1760, March 2004. 16