Mechanical Integrity Testing Using the Fibre OpticTemperature Sensing Technique

Transcription

1 SOLUTION MINING RESEARCH INSTITUTE 1219 Bittersweet Drive Richmond, Texas , USA Technical Conference Paper Telephone and Fax: Mechanical Integrity Testing Using the Fibre OpticTemperature Sensing Technique Stephan Grosswig 1 Eckart Hurtig 1 Bernhard Vogel 1 Fritz Crotogino 2 Joerg Schoenebeck 2 Ralf Riekenberg 3 Paul Groenefeld 3 T. H. von Tryller 4 Geso Jena/Germany 1 KBB Hannover/Germany 2 EWE Aktiengesellschaft, Oldenburg/Germany 3 Socon, Giesen/Germany 4 Fall 2003 Conference 5 8 October Chester, United Kingdom

2 Solution Mining Research Institute; Fall 2003 Technical Meeting Chester, United Kingdom, October 5-8, 2003 Mechanical Integrity Testing Using the Fibre Optic Temperature Sensing Technique Stephan Grosswig, Eckart Hurtig, Bernhard Vogel, GESO Jena/Germany Fritz Crotogino, Joerg Schoenebeck, KBB Hannover/Germany, Ralf Riekenberg, Paul Groenefeld, EWE Aktiengesellschaft, Oldenburg/Germany, T. H. von Tryller, SOCON, Giesen/Germany Abstract The evaluation of Mechanical Integrity Tests (MIT) to verify the tightness of cavern wells is primarily based on determining the depth of the gas/liquid interface below the casing shoe of the last cemented casing. Repeated measurements using gamma-gamma tools represent, however, a significant source of errors. The fibre optic temperature sensing technique is a viable alternative. The fibre optic temperature sensing technique was first implemented in a pilot test at the Huntorf K 5 gas cavern, EWE Aktiengesellschaft, Oldenburg, in order to confirm that temperature measurements are able to determine the position of the interface level with sufficient accuracy. The measurements took place immediately after a standard gamma-gamma interface log. The fibre optic temperature sensing cable was installed in the future 4 ½ brine displacement string to a depth approx. 30 m below the anticipated interface level. This was achieved by determining the interface level in the open well below the last cemented casing based on measurements taken from within the 4 ½ string. Temperature effects were stimulated by withdrawing approx. 400 l brine from the string at the cavern head and measuring the reestablishment of temperature equilibrium. The gas/liquid interface was proven unequivocally. The depth corresponded with that determined by the gamma-gamma log. While the temperature sensing technique offers measurement accuracy comparable with the gamma-gamma method, it also offers the advantage of a quasi-continuous measurement. This would allow early identification of trends e.g. in the development of rates of leakage, which would in turn allow the period of testing to be curtailed. Keywords: fibre optic temperature measurements, gas/liquid level determination, mechanical well integrity testing, MIT 1 Problem The testing of caverns for tightness is generally based on testing the tightness of the casing shoe of the last cemented casing to the rock. This test is known as the mechanical integrity test. This test assumes that the salt rock into which the cavern is leached is itself tight. A positive test result also verifies of tightness of the last cemented casing. This is therefore only tested separately in special cases or is indeed actually made superfluous when using for example welded strings. The tech- 1

3 niques for testing tightness are presented in detail by Crotogino (1995, 1996). In principle mechanical integrity testing takes place on liquid filled caverns (saturated brine). Three methods are adopted in practice: In-situ compensation method In-situ balance method and Surface balance method. The general principle is to inject a test gas into the annulus of a cavern filled with brine or some liquid to a level below the casing shoe of the last cemented casing. To this end the planned brine withdrawal string or an appropriate test string is run into the well. The test gas is lighter than and immiscible with the brine, or the stored medium, such that a stable interface forms in the open well below the casing shoe (uppermost part of the cavern neck) between the liquid and the test gas. This creates a defined volume filled with test gas which covers the transition between the cavern neck and the last cemented casing shoe, the annulus and up to the cavern head. The definition for tightness in the transition zone between the cavern neck and the last cemented casing shoe is the mass of gas enclosed within the defined volume. The gas volume can also alter even in a tight well as a result of the influence of temperature or the cavern, which would cause the interface level to change. If the balance of the enclosed mass of gas is constant between two measurement time points, the result of the test is determined as being positive, i.e. tight. The following data are required to determine the tightness/mass balances: Exact well geometry below the casing shoe to the level of the liquid/gas interface Test gas head pressure Depth of liquid/gas interface Temperature development between cavern head and liquid/gas interface. In the In-Situ-Compensation (ISC) method, the position of the interface is determined via a weep hole, located at the lower end of a test string, with an accuracy of 0.01 m. For a detailed description of this method see Edler et.al. (2003). This ISC method has become widely established in Germany because of its accuracy and short test time. A disadvantage compared with the In-Situ-Balance method is the higher cost for a workover unit and the installation of proprietary test equipment. At the Huntorf K 5 cavern a specific problem was the partial instability of the gas-brine interface as a result of discontinuously after-flowing nitrogen blanket residues; with the ISC method it is not possible to adjust to the interface depth after installation. equipment. The In-Situ-Balance method contrasts in that the position of the liquid/gas interface is measured by running a gamma-gamma log in the test string. The high degree of flexibility in adjusting to the current interface depth is advantageous; a drawback is the lower accuracy of approximately m and the limitation of providing only individual measurements as opposed to quasicontinuous logging in the case of the ISC method. Since temperature differences are present at the liquid/gas interface, the concept was to test whether temperature measurements represent an alternative technique to determine the level of the gas/liquid interface and any changes in depth with sufficient levels of accuracy. 2

4 It is necessary to measure the temperature simultaneously over a large depth range at the maximum possible local resolution. The application of fibre optic temperature sensing techniques was thought to be suitable for measurements of this type. This technique has the following advantages in the measuring of interface depth and temperature in mechanical well integrity testing as well as other possible application areas: No winch is required for installing the test string (if a brine withdrawal string not previously installed) Normal crane job for log snubbing assembly Sensing cable remains in well throughout test period Temperature regime and enclosed mass of gas not disturbed by run-in/pull-out of logging tool Temperature and interface measurements are quasi-continuous The establishing of equilibrium or any changes can be identified immediately and allow any trends in leakage rate reduction to be observed Test time can be reduced to absolute minimum. This technique was first implemented for test purposes at the Huntorf K 5 well to confirm the suitability of the method. The results of the tests are presented. 2 Distributed fibre optic temperature sensing method Fibre-optic distributed temperature sensing is based on optical time-domain reflectometry (OTDR). A pulsed laser is coupled to the optical fibre which is the sensing element. Owing to changes in density and composition as well as molecular and bulk vibrations, a small part of light is backscattered as the pulse propagates through the fibre. The intensity and spectral composition of the backscattered light are determined by the molecules in the optical fibre. The backscattered light includes different spectral components which are the result of differing interaction mechanisms between the propagating light pulse and the optical fibre. The Raman backscattering component is caused by thermally influenced molecular vibrations. Its intensity is dependent on temperature and can thus be used to measure the temperature along the optical fibre. The Raman backscattered light has two components: the Stokes line and the Anti-Stokes line which have different intensities. The intensity of the Stokes line is only weakly dependent on temperature while the Anti-Stokes line shows a strong relation to temperature. The basic principle of fibre-optic temperature measurement thus consists in filtering the Stokes and the Anti-Stokes lines out of the backscattering light. Using the ratio of the intensities of the Stokes and the Anti-Stokes lines eliminates external influences such as changes of the light source or the optical fibre. As the velocity of light within the fibre is known, the location of the backscattering centres along the optical fibre can be derived from the travel time of the backscattered Raman component. This makes it possible to determine both the temperature and the distance along the fibre simultaneously. The temperature is determined as an integral value for a short section of the optical fibre. As the temperature values and their location along the optical fibre are determined simultaneously, it is not necessary to record the fibre length or distance (depth) separately using a conventional borehole winch technique. The absolute temperature is determined by way of a calibration function which depends on the specific material properties of the optical fibre (geometry and chemical composition) and their temperature dependence. These properties can vary for different optical fibres. 3

5 Therefore, the calibration function must be determined for the individual fibres using laboratory temperature calibration equipment before the measurements are performed. The DTS technique makes it possible to measure temperatures over extended time periods and lengths of up to several kilometres while allowing an exact allocation of temperature data. A spatial resolution of 0.25 m and a temperature resolution of 0.05 K are possible. The fibre-optic temperature sensing system operates without any electronic circuits along the fibre. It is thus totally insensitive to electromagnetic fields and can be used even in hazardous areas subject to stringent explosion protection requirements. As the temperature is determined from the ratio of the two intensities of the Stokes and Anti-Stokes lines, age effects of the optical fibre can be neglected. The measurement principle and the principle display schematic are shown in figure 1. M easuring object Tem perature sensing cable M easuring Laserim puls device Backscattered T I a I s spectrum W eave length Travel time ns 20 ns 30 ns 40 ns ~ 1m ~ 2m ~ 3m ~ 4m Intensity PC T Tem perature Location s Fig. 1: Schematic of fibre optic sensing principle There is now a considerable data-base available on using fibre optic temperature measurements in caverns (see Großwig et.al 2001 a-c, Schneider et al. 2002). In terms of tightness tests, the fibre optic temperature sensing techniques offers primarily the following advantages: Simple installation of sensing cable in well Cable remains in well throughout entire test period Temperature field is not influenced by repeated logging runs such that measurements can take place continuously throughout the selected time window Simultaneous measurement of temperature along the entire length of the cable No electronic/electric components or modules located along the length of the temperature sensing cable. 4

6 It should be borne in mind that all fibre optic temperature measurements determine the integral temperature over a short section of the fibre optic cable. The measurement point specified refers to the centre of this section of cable. The measurement principle is described in more detail in Großwig et al. (1998). 3 Installation of sensing cable, making measurements A fibre optic temperature sensing cable was suspended in the 4 ½ string of the Huntorf K 5 well. The cable end was located at a depth of m (Huntorf K 5 well is a deviated well, therefore all depths are given in measured depths). The depth was then confirmed with a CCL at a point 3.05 m below the cable end. The spatial resolution along the fibre optic sensing cable was 0.5 m. The gas/liquid interface to be measured was located below the casing in the open cavern neck area. It was therefore necessary to determine the depth of the interface using measurements within the 4 ½ string, without the sensing cable coming into direct contact with the interface proper (see figure 2). The measurement procedure is summarised in table 1. The measurements were made in three phases: Phase 1: Phase 2: Phase 3: Adjust equilibrium after installing sensing cable and determining initial status (measurement period approx h). Measurements to determine interface depth during which the sensing cable was lifted in five steps of 0.1 m (measurements K 6 through K 11). This was intended to improve local resolution. During each step 15 to 16 individual logs were run across the entire depth. To increase temperature contrast between the nitrogen filled part of the annulus and the brine in the 4 ½ string, approx. 400 l brine were withdrawn causing colder brine from the cavern to rise up the 4 ½ string. As in phase 2, the sensing cable was then lifted in five steps of each 0.1 m (measurements K 13 through K 18). Because of the CCL measurement, at the start of phases 1 and 2 the cable end was located at a depth of m, at the start of phase 3 it was at a depth of m. 5

7 , P P test gas P fibre optic temperature sensing cable??? production string final cemented casing test gas brine brine string cavern Fig. 2: Schematic of sensing cable to determine interface level. The sensing cable is suspended in a string without direct contact to the gas/liquid interface. Table 1: Procedure for fibre optical temperature measurements in well K 5 Day Phase 1: Install sensing cable and determine depth via CCL log, measure initial situation over a 12.5 h period through to (25 individual logs) Phase 2: Temperature measurement involving raising sensing cable five times in steps of 0.1 m (measurements K 6 through K 11) Remove 400 l brine, Phase 3: Temperature measurement involving raising sensing cable five times in steps of each 0.1 m (measurements K 13 through K 18) 6

8 4 Measurement results 4.1 Fibre optical interface measurements The following data are available to determine the interface depth: measurement of initial situation (phase 1), measurements during phase 2 (lifting of sensing cable in steps of 0.1 m) and measurements during phase 3 (withdrawing 400 l brine and lifting sensing cable in steps of 0.1 m). The evaluations are based on the data records recorded during phases 1 and 3. Figure 3 shows the temperature/depth distribution in the initial situation. This curve clearly shows the boundary between the cover rock and the salt at approx. 450 m. From approx m there is a sharp drop in temperature. Temperature ( C) Depth (m) Fig. 3: Temperature-depth distribution in initial situation (average value from 21 individual measurements 7

9 The use of the fibre optic temperature sensing technique allows the entire temperature-depth distribution to be determined continuously and simultaneously. This means it is possible to directly determine the time at which equilibrium is reached in the well after injecting the test gas. Furthermore, temperature fluctuations can be determined during the tightness test with a temperature resolution of minimum 0.1 K, such that it is possible to continuously check the influence of the temperature on the gas volume in the annulus. This substantially raises the accuracy of the test. Measurements in phase 1: Fig. 4 shows the temperature-depth distribution in the initial situation for depth section 900 m through 1070 m. The interface is indicated at a depth of approx m. This is marked by a temperature step of approx. 0.5 K. Fig. 4: Temperature-depth distribution in initial situation Average value of 21 individual measurements, depth range 900 m through 1070 m 8

10 The plot shows contrasting courses of the temperature curves above and below the interface (see figure 5). Reflecting this situation separate optimised curve approximations were calculated for the separate temperature curve sections below the interface and above the interface to the casing shoe (Nawrotzki, 2003). The blue plot is the curve approximation for the depth range below the interface. Since the course of the curve of the measured temperature changes again at the cavern shoe at depth 1040 m, only the small depth range from the interface to the casing shoe depth (red curve) can be used for the curve approximation. The intersection of the two curves marks the location of the interface, at m. Fig. 5: Temperature-depth distribution in initial situation Average value of 21 individual measurements, depth range 1000 m 1070 m. Optimised approximation of measurement values. Measurements in phase 3: In order to enhance the temperature effect, approx. 400 l brine were withdrawn from the 4 ½ string at the start of the second measurement day (16.04.). As a result, the brine rose in the string by about 40 m. The rise of colder brine created an additional temperature difference between the brine-filled string and the annulus. As a result of the heat conductivity of the brine filled annulus below the interface being better than that of the nitrogen filled annulus above the interface, the temperature difference has a different rate of equalisation. Thus it was anticipated that the interface would be identifiable by an anomaly in the development over time of the temperature distribution. Figure 6 shows the temperature development for the average value of measurements K 13 through K 16 for the entire section of the Huntorf K 5 well. The rise of the 9

11 colder brine in the string can be identified up to approx. 950 m. The temperature showed a clear drop. In the upper part of the well the temperature was raised because slightly warmer brine rose in the string due to normal geothermal gradients. Temperature ( C) K13 K14 K15 K16 K17 K Depth (m) Fig. 6: Temperature-depth profile for entire well after withdrawal of approx. 400 l brine (measurements K 13 K 18) Approximately 2 hours after the brine was withdrawn, the situation had returned approximately to initial status. Figure 7 shows the depth section 1000 m through 1070 m. The plots clearly show that during removal of the brine from the 4 ½ string, the brine in the string rose by approx. 38 m 39 m. As seen previously in figure 6, the temperatures quickly equalise and return to initial status. 10

12 The sensing cable was lifted in five steps of 0.1 m during the measurement cycle, i.e. a total of 0.5 m. The plots in figure 7 have been corrected for depth. Fig. 7: Temperature-depth profile after withdrawal of approx. 400 l brine (measurements K 13 through K 18), depth range 1000 m through 1070 m A step is apparent in the plots between depth 1045 m and 1050 m, and is particularly obvious in measurement K 13. The temperature change is higher above the step than below the step. The temperature change is interpreted as marking the position of the gas/liquid interface. The minimum between 1030 m and 1035 m is the result of the rise of cooler brine after the withdrawal of approx. 400 l. In the subsequent evaluations, the K 13 measurements (logs 1 4) recorded immediately after the withdrawal of 400 l brine and the rise of the brine in the brine string are taken as reference curve. The difference between the reference curve and measurements K 14 K 18 is plotted in figure 8. The temperature step at the gas/liquid interface is clearly apparent despite the noise. The depth of the gas/liquid interface is taken as being the inflection point at which the temperature difference increases. In order to emphasise the inflection point, calculations were made of the linear trend below the interface (from 1050 m) and also of the linear trend for the area of increase of temperature difference. By correcting for the lifting of the cable (in each step 0.1 m), the inflection point should be at the same depth for all measurements. Figure 9 presents the temperature difference between the reference curve (K 13, log 1 4) and measurement K 16, at which the sensing cable was raised by a total of 0.3 m, as an example. The intersection of the two trend curves is at a depth of m. 11

13 Fig. 8: Difference in measurements K 13, Log and K 14 K 18 compared with temperature plot K 13, Log 1 4 Fig. 9: Difference curve between measurement K 13, Log 1-4 and measurement K 6 (lifting of sensing cable by a total of 0.3 m) 12

14 Depths of between m and were also derived from the other difference curves. These values fit well with the value of m, as derived from the evaluation of the temperature log for the initial situation. 4.2 Gamma-gamma interface logs The planned interface depth logs for the official test evaluations were performed by KBB using a SOCON gamma-gamma logging tool. Figure 10 shows a comparison plot of the gamma signal plot for the last test log on 8 April It is pointed out that in this case cavern K 5 had an unusually large interface surface area of approx. 12 m², and as a result the interface signal shows a particularly clear reading through the 4 ½ string. The evaluation of the log indicates an interface depth of m. According to SOCON, the accuracy of the gamma-gamma log is ± 0.1 m Depth, m /8" Casing Shoe 1045 Interface Depth 1049,4 m Counts Per Minute, x 10 E+5 Fig. 10: Gamma-gamma log dated 8 April 2003, Huntorf K 5, SOCON interface depth m after log evaluation 13

15 5 Conclusions The evaluation of the fibre optic logs shows that it is possible in principle to determine the depth of the gas/liquid interface using fibre optic temperature logs through the brine string. The depth values recorded for the logs while raising the cable in steps of 0.1 m vary between m and m. The depth determination derived from the curve of the initial situation produces a comparable value of m. These data allow the accuracy of the depth determination of the gas/liquid interface to be derived and show it is comparable with that of gammagamma log results. The gamma-gamma log results also evaluated the gas/liquid interface to be at a depth of m. The measurements within the brine withdrawal string represent the worst case for the fibre optic sensing principle. More favourable measurement conditions would apply if measurements were performed in the casing without the brine string, since this would allow the sensing cable to be in direct contact with the gas/liquid interface. In this case, the measurements should be performed as quickly as possible after injection of the nitrogen, since the temperature contrast between the gas and the brine temperature can be expected to be highest at that time. Furthermore, the fibre optic temperature sensing technique allows the temperature-depth distribution to be determined continuously throughout the tightness test period across the entire length of the well with a high level of local resolution, generating data to allow temperature-dependent changes in the test gas volume to be recorded and corrected if required. Similarly, the data allows for example incoming gas volumes to be quickly identified which is not possible when implementing the In-Situ-Compensation method. The results show that when applying the In-Situ Balance method, determination of the gas/brine interface depth using the fibre optic temperature sensing technique offers the following advantages compared with conventional gamma-gamma logs while achieving similar levels of accuracy: quasi-continuous logging shorter test times increased accuracy in determining leakage rates since repeated logging tool passes are not necessary. 6 Literature Crotogino, F. (1995): SMRI Reference for External Well Mechanical Integrity Testing / Performance, Data Evaluation and Assessment. SMRI-Research and Development Project No S. Crotogino, F. (1996): Reference Value Developed for Mechanical Integrity of Storage Caverns, Oil & Gas Journal 94 (1996), 28. Oct., S Edler, D.; Scheler, R.; Wiesner, F. (2003): Real Time Interpretation of Tests Investigating the Tightness of the Cementation of Final Cemented Casings in Gas Storage Cavern Wells, Solution Mining Research Institute (SMRI), Technical Meeting Spring 2003, Houston, Texas 14

16 Großwig, S.; Hurtig, E.; Kasch, M.; Kühn, K. (1998): Die ortsaufgelöste Temperaturmeßtechnik Leistungsfähigkeit und Anwendungsmöglichkeiten im Umwelt- und Geobereich anhand ausgewählter Beispiele. Temperatur 98, Verein Deutscher Ingenieure - VDI-Berichte Nr.1379, VDI-Verlag Düsseldorf 1998 (ISBN ): Großwig, S.; Hurtig, E.; Kühn, K.; Rudolph, F. (2001a): Distributed Fibre-optic Temperature Sensing Technique (DTS) for Surveying Caverns and Brine Pipelines. Solution Mining Research Institute (SMRI). Fall 2001 Meeting. Albuquerque, New Mexico, U.S.A., , Technical Papers, Großwig, S.; Hurtig, E.; Kühn, K.; Rudolph, F. (2001b): Distributed Fibre-optic Temperature Sensing Technique (DTS) for Surveying Underground Gas Storage Facilities. IGRC 2001 International Gas Research Congress. Amsterdam, Netherlands, , Proceedings, EPP16. Großwig, S.; Hurtig, E.; Kühn, K.; Rudolph, F. (2001c): Distributed Fibre-optic Temperature Sensing Technique (DTS) for Surveying Underground Gas Storage Facilities. OIL GAS - European Magazine 4/2001: Nawrotzki, K. (2003). Personal communication. Schneider, R.; Großwig, S.; Hurtig, E.; Mücke, L. (2002): Detection and Determination of the Fluid Level in the Annulus in the Kiel Underground Gas Storage Facility (Germany) using Fibre-optic Temperature Measurements. Solution Mining Research Institute (SMRI). Fall 2002 Meeting. Bad Ischl, Österreich, , Technical Papers,