Basin scale rock mechanics: Logs and core measurements
A.RMarsala, G.Ragazzini, O.Meazza, M.Brignoli & RISantarelli, Agip, San Donato, Italy Copyright 1994. Society of Petroleum Engineers This paper was prepared for presentation at the 1994 Eurock SPE/ISRM Rock Mechanics in Petroleum Engineering Conference held in Delft, The Netherlands, 29-31 August 1994. ABSTRACT: The Northern Adriatic basin contains a great number of developed fields at depth ranging between 1000 and 4000 meters, and a comprehensive geophysical, petrophysical and mechanical characterization of these weakly consolidated formations was carried out. For 15 wells cored in the pay zones, a systematic campaign of core measurements was undertaken and comparison between logs and cores made. Typical log programmes contained porosity, gamma ray, resistivity and sonic, including full waveforms and dipole. In the laboratory typical core measurements consisted of density, porosity, permeability, residual oil saturation, formation resistivity factor, granulometry, clay content, ultrasonic velocity measurements along with triaxial and oedometric tests. Empirical correlations and statistical analysis of this entire database at the scale of a geological unit is presented. 1. INDUSTRIAL BACKGROUND A major problem encountered by rock mechanics in all of its industrial applications, including tunnelling, mining, surface civil engineering, etc., is the scarcity of relevant experimental observations at the level of both the field and the laboratory. A direct consequence is that the needed integration of rock mechanics into day to day design engineering remains largely underdeveloped. Such an unfortunate state of affairs is nowhere more acute than in the field of petroleum engineering for obvious reasons. i. The rock of interest is usually burled at depth of several thousands of meters making any direct observation extremely difficult if not impossible. The rock mechanics characterization of formations is therefore based on the converging interpretation of a series of indirect observations such as drilling performance, cuttings characterization, logs, core measurements, etc. [1]. ii. The activities of a given company are often spread out allover the world, in vastly different geological and operational environments. This further contributes to the dilution of the already scarce information which could be useful to apply rock mechanics to solve specific problems. iii. In most corporations, rock mechanics is being handled by a small team of specialists based in central office, because it cannot be attached to any specific branch of petroleum engineering such as drilling, reservoir, production, etc., and is therefore not thought after by operating branches. This lack of awareness at the level of operations often means that rock mechanics related problems remain unheard of by those who could solve them. 105 This paper presents a pilot study aimed at addressing the problems raised in points ii. and iii., in order to provide rock mechanics based operational guidelines within a minimum time- frame to engineers without any specific formation in the field of rock mechanics. The idea was to concentrate all direct and indirect available rock mechanics information at a scale which would have both an industrial and a technical relevance. The criteria used during the selection of such a scale -i.e. a geological basin- will be presented in Section 2. Section 3 will present the geological background of the Northern Adriatic basin upon which this pilot approach was performed. Section 4 will introduce the most relevant rock mechanics and petrophysical measurements performed on cores. Finally, Section 5 will analyse the data and it will be shown how the adopted scale -i.e. a well known consistent geological environment- can be used as a frame to interpret all empirical correlations. The practical application of such an approach, in the case of sand production prediction, is illustrated in a twin paper presented at the same conference [2]. 2. SELECTION OF THE SCALE As implied by the previous section, a scale at which data would be gathered and analysed had to be chosen with the idea of optimizing the operational return derived from such an exercise. When analyzing the literature four such scales appear clearly: i. numerous in depth studies have been performed at the scale of a single well where operational problems such as wellbore stability, sand production, etc. had occurred in order to be able to identify the mechanisms at the origin of the problems such as to cure them on subsequent similar wells [3-5]; ii. studies can also be performed at the scale of an entire reservoir because it is anticipated that its development and production will trigger problems such as wellbore instabilities, subsidence, etc.; in that respect, the North Sea Field of Ekoftsk is exemplary and may be the most widely published reservoir of the entire industry -e.g. [6-8]; iii. data can also be gathered in a world wide database such as in the case of stress mapping over entire continents -e.g. [9]- or when an operator tries to identify within its own datasets correlations between rock properties -e.g. [10] in the domain of sand production- or when published data are used [11]; iv. ftnally, the domain of in situ stresses gives us an interesting example where workers -e.g. [12- 14]- try to gather data at the scale of a coherent geological unit and to establish the link between such stresses and geological features such as to derive some sort of logical pattern. The ftrst two scales (single well or reservoir) were rejected because, whilst they are the most adapted to the study of speciftc problems, they are too limited for a general exercise such as the one described here. The third scale (worldwide) usually reveals obvious general trends which have limited operational impacts. The last scale (a coherent geological unit or "geological basin") immediately appeared as the most attractive for several reasons: i. From an industrial point of view, successful exploration is often linked with the understanding of the internal geological logic of a basin which, in turn, results in a concentration of operations in such a unit when the ftelds are brought into production. ii. Such concentrated operations results in a pre- concentration of data making the gathering much easier. iii. Furthermore, a concentration of operations also results in a concentration of rock mechanics related problems and an increased need to provide satisfactory solutions to them. IV. Finally, the example of in situ stress measurements often reveals a link with geological features and it was hoped that such a pattern would repeat itself for other data -e.g. tectonic, consolidation, diagenesis, etc. 3. GEOLOGICAL DESCRIPTION OF THE SELECTED BASIN For many years now, Agip has developed a series of gas reservoirs in the Northern and Central parts of the Adriatic Sea and it has long been recognised that they all corresponded to a single geological unit which will be named in this paper the Northern Adriatic Basin. Before describing the geology of the basin let us present some industrial factors which contributed to its choice for the pilot study presented therein. A large number of ftelds has been developed in the basin at depths ranging from 1000 to 4500 m. Long 106 term experience has revealed many important rock mechanics related problems during field development such as wellbore stability (oil based mud is used on many wells with the need to ship all cuttings onshore in a zero discharge area), sand production [2] (many wells are equipped with sand control after some dramatic sand production had stopped production on entire platforms) and subsidence [15] (of particular concern for the coastal fields in the sensitive Ravenna area). As more marginal fields are brought into production [16], the need to optimize the technical solutions to handle such problems appears every day more crucial and so does a proper evaluation of the rock mechanical environment which provokes them.
+. 0 K .. 150 . Figure 1 - The Northern Adriatic basin [16]. The Adriatic basin -see map of Figure 1- is considered to be an undeformed remnant of the African promontory (Apulia plate) which acted as a foreland for both the Apennine and Dinaride thrust belts -see the geological cross section of Figure 2 [16]. The sedimentary sequence consists of Mesozoic carbonates and Tertiary terrigenous deposits. The basin itself was deposited during the Pliocene, Pleistocene and Quaternary periods in a turbiditic deltaic environment by the Po river; the hydrocarbons that are trapped, consist of pure and isotropically-light dry methane. Moreover after the sedimentation process, the basin does not seem to have been uplifted nor to have been affected by any kind of relevant tectonic event; the most characteristic geological factors for this area could be summarized as follows: Turbiditic sedimentation. No tectonized environment Cold thermal regime A typical feature of this basin is that the gas accumulations occur in mUlti-pay zone reservoirs; the pay layers consist of sandy formations without any kind of cementative materials (neither of calcareous nor of siliceous type), the coherence of this rock being due essentially to compaction. The seal above the reservoirs consist of thin beds of shale often less than 1m thick. ADRIATIC FORELAND COAST !.WE Figure 2 - A typical geological cross section of the Northern Adriatic basin [16]. The mineralogy of the reservoir rocks show a mixture of sands, silts, shaley sands and silty shales. In particular the clay fraction typically contains large proportions of smectite, illite and mixed layers; nevertheless owing to the cold thermal regime and to the insufficient in situ stress level, typical smectite/illite transition seems to have occurred before the sedimentary process. In this way any kind of in-situ cementation phenomenon related to such a transition effect has to be excluded. 4. LOGS AND CORE MEASUREMENTS In the Northern Adriatic basin more than 50 gas reservoirs were developed at different depth, giving a deep insight in the most extended italian off-shore area, and providing a wealth of data to build upon to construct the database described here. For this purpose 15 wells were selected on which a complete and systematic campaign of logs and core measurements was conducted. The selection of the wells was made in such a way so that to cover not only the entire area of the basin, but also the complete range of depth of the reservoirs. A typical log campaign, usually consisted of a y-ray (OR) along with a sonic (SLS) and a resistivity (AIT) log. Moreover a deepmeter (SHDT) was often included in the program. Cased holes were systematically loged to check the cement bond 107 (CBLNDL). In the pay zones, density (LDL), porosity (CNL) and electromagnetic propagation logs (EPT) were run to complete the petrophysical characterization of the reservoir formations. Four inches cores were taken from the pay zones of all the wells, sealed inside fibreglass tubes and shipped to the laboratories. Here a 'Y-ray log was recorded on each core, compared to the one recorded downhole for the purpose of depth correlation. However, in these particular rock formations, clay minerals may affect the results of this comparison, so porosity data were also used to improve depth determination on cores. For reservoir study purposes, porosity measurements were performed on plugs sampled every 30 cm. Typical porosity values of the sandy formations constituting the reservoirs ranged from 15% to 40%. To complete the routine petrophysical characterization of the cores, permeability measurements were conducted on plugs sampled with the same frequency both parallel and orthogonal to the core axis (typical permeability values of the reservoir rocks range from 20 to 600 mD). Residual oil saturation and formation resistivity factor were measured where it was possible, while grain density was measured every 30 cm. Because the reservoir sands are uncemented, and may have to be produced with sand control completions, systematic granulometry measurements were performed by two different methodologies: x-ray diffraction and laser intereferometry. Clay content quantification was therefore possible using data measured by the two different methodologies, keeping in mind that laser interferometry is only able to detect the radii of dispersed particles whilst x-ray diffraction cannot discriminate between dispersed and attached clay particles. In order to better characterize, from both petrophysical and mechanical standpoints, the poorly consolidated plugs and to calibrate and interpret the sonic logs (SLS) recorded in the wells, non traditional petrophysical measurements were performed on several samples. Transit times for both compressional and two shear waves were measured on dry and brine saturated plugs which had been brought back at first . to reserVoir temperature (less than 80C) with a rate of one degree per minute and then to reservoir pressures by a stepwise increase of confining and pore pressures (reservoir net pressure ranged from 5 to 30 MPa). Extreme care was taken to prepare the unconsolidated samples avoiding to damage them and thus allowing the obtention of representative results. For all measurements, a frequency of 1 MHz was used. During the ultrasonic tests the volume of the brine displaced by the stress increase from the sample was measured and the corresponding porosity reduction calculated. Typical values ranged between 5 and 15% of the initial porosity. Such data were not analyzed further in so far as the rock compressibility does not usually play a role in the recovery of gas; this would not stand for oil reservoirs. Consolidated drained triaxial tests were performed on selected sand plugs saturated with formation brine and the complete stress-strain curves were drawn, for different confining stress conditions. For each depth, four levels of confining pressure were used: namely 0.5 MPa, ~ O ' h " O'h', MO'h' (where O'h' is the minimum horizontal effective in situ stress). Uniaxial compression tests were not included in the programme because the reservoir sands were too unconsolidated. Because of the intrinsic weakness of the material, systematic unloading/reloading cycles were performed at preset levels of deviatoric stress in order to differentiate between reversible and irreversible deformation. In some cases, ultrasonic velocity measurements were conducted at the same time of the triaxial tests to allow a direct comparison between dynamic and static properties. As already mentioned, subsidence estimation is an important issue for many costal reservoirs and oedometric tests were conducted in order to quantify the compaction effect related to effective stress increase provoked by reservoir depletion. In some cases even the shaly formations of the cap-rock were tested in order to quantify possible effects on subsidence phenomena. Nevertheless in this paper the data measured during such tests will not be presented even though they have been integrated in the global database. In conclusion we can summarize that for each well in the database are allowable the following parameters: porosity, density and compressional and shear wave velocities from both cores and logs; grain size distribution., mineralogical analysis, 108 penneability and complete stress-strain curves from cores. 5. DATA ANALYSIS Figure 3 shows a typical result of a series of four triaxial test on cores: the minimum effective in-situ stress was 20 MPa and it is clear that the brittle to ductile transition takes place when the confining stress of the test overcomes this value. This almost systematic transition at the level of the minimum effective in-situ stress led to think that the basin was nonnally consolidated. w ;0 -.-:-:::;: :- -1 I :--,_<--".-.--- .. ---t-_ . -_- .. M=M'a==1l I' " ,,' 20 .. CP: 10!\Fa _ .' , " . --= ': ..' I " .' t i':' 10 .'. "'iii .. -- CP:20M'a -- CP:40M'a _ ,e:" ----- o 2 3 4 Axial SIrain, % Figure 3 - Stress-strain curves at different confining pressures (CP); brittle to ductile transition is revealed.
. . .. ,. t .
,.. .. . fi:, :. .
o 1000 4000 Ver1icaI Depb, m Figure 4 - Porosity versus vertical depth in the Northern Adriatic basin. Further indication of the nonnal consolidation of the basin was revealed by the porosity vs. depth trend of Figure 4 and the UCS vs. depth correlation (Figure 5): in both cases the direct dependence of the petrophysical and mechanical properties upon compaction is due not only to the deposition process but also to the absence of any effect of cementation phenomena. ro . R<awesy: l00"/' .
o o 1000 4000 Ver1icaI Depb, m Figure 5 - Uniaxial Compressive Strength versus depth; recovery factors are indicated. However the UCS versus depth plot of Figure 5 shows two trends with some cores having a lower value of uniaxial compressive strength with respect to depth. Because of the relatively cold thennal regime and because of the environmental conditions any kind of in-situ chemical transition between smectite and illite can be excluded, therefore the two trends of Figure 5 can not be explained assuming different quartzic cementating process linked to clay minerals transitions. Moreover, the lack of calcareous sediments in the basin, further excludes the other major cementation process and the double trend of Figure 5 must Imd another explanation. Further analysis of the data of Figure 5 reveals that for the apparently weaker rocks, the brittle to ductile transition occurs far below the minimum effective in-situ stress or, in some cases the rock may even show no transition at all and have a ductile behaviour even at the lowest confining pressures. 109 Checking the recovery factors of the cores belonging to this lower trend and exhibiting this type of abnonnally low brittle to ductile transition, it was demonstrated that they were systematically lower than the expected 100%. The two-fold trend of Figure 5 is therefore the result of core damage, which induced unreliable, apparently weaker, rock strength data. Such an explanation was further confinned when a study simulating core damage on weak artificial sandstones tailored to mimic those of the Adriatic showed similar losses of strength and lower brittle to ductile transitions in case of damaged cores [17]. Shear waves velocities (V s), measured on both dry and brine saturated plugs, were not affected -as expected- by the fluid presence. Their values ranged from 1100 to 2000 mls. A small anisotropy (a few percents) of the selected plugs was revealed by the shear waves polarized in ortogonal directions. ..!!! E 1 ... oS >- 4000 3000 2000 1000 IS . . ~ '. ~
. !
. . .
.,
. 2S g Vssat . . . . .. .
. . . . . . . . . . ..... . .' . . . 35 4S Porosity, % Figure 6 - Compressional (Vp) and Shear (Vs) ultrasonic wave velocities vs. Porosity (laboratory measurements on fully saturated samples). The compressional wave velocities (Vp) increased by about 20% reaching the value of 3600 mls for the plugs subjected to the higher net pressure -Le. these cored in the deeper zone of the basin. VpNs ratio ranged from about 1.6 in the higher part of the basin to about 1.8 in the deeper one. The quite good exponential trend with an high correlation coefficient (r=0.7) observed between velocities (compressional and shear) and porosity (see Figure 6) allowed their respective matrix values to be extrapolated at zero porosity. These values (4900 mls and 2900 mls respectively) confirmed the validity of the inverted Raymer equation applied to the experimental data. The dynamic elastic moduli, calculated at reservoir conditions, are representative of typical unconsolidated formations (Shear Modulus: 3-9 GPa, Bulk Modulus: 7-17 GPa, Young Modulus: 7-22 GPa and Poisson Ratio: 0.27-0.35 going from 1000 to 4000 m of depth in the Adriatic basin). The good trend observed between plugs permeability and elastic wave velocities (see Figure 7) and the non linear correlation between compressional velocity and bulk density at reservoir conditions permitted the petrophysical properties of the plugs to be estimated from the dynamic moduli. All these good correlations are likely to be due to 110 the simple geological history of the basin, i.e. normal consolidations, lack of tectonics and of diagenesis, as already mentioned. 4000 ..!!! 3000 E .;; - VpSllt .. :e c Vssat ... Q '---,-- 'il 2000 > 1('" . 10 .' . . . , 0 0 o on 100 ~ , m . . 0 , 1000 Figure 7 - Compressional (Vp) and Shear (Vs) ultrasonic wave velocities vs. Permeability (laboratory measurements on fully saturated samples). The critical frequency values (below one KHz), calculated according to the Biot theory considering both the fluids and the petrophysical properties, allowed to compare directly log and laboratory velocities both belonging to the same frequency limit. The correlation coefficients between the main petrophysical and mechanical parameters included in the database are given in Table 1. The cross correlation clearly shows the direct dependence of the rock strength (UCS) upon porosity (r=0.87) and the clear link between porosity and depth (r=0.86) due to the normal consolidation of the basin. Table 1 - Correlation coefficients. Vert. Depth UCS Young's Mod. Porosity lab Porosity log Clay cont lab Clay cont log Dtclog Northern Adriatic Sea Cross correlation oS -d :@ "" ~ .s "" ., CI) . ~ . ~ 0 u '" t: ~ -"" ~
e > 0 0 "'" >- 1.00 0.75 1.00 0.57 0.79 1.00 0.86 0.87 0.77 1.00 0.58 0.40 0.26 0.73 1.00 0.10 0.20 0.20 0.33 0.26 0.10 0.17 0.14 0.17 0.49 0.85 0.54 0.33 0.69 0.64 :@ ~ g 1.00 0.36 0.10 "" .s ~ "" .s ,s 1ti' CJ 1.00 0.64 1 However, when considering the log determined porosities, the correlation coefficients become quite poor (r=0.4 for the UCS - porosity log). Among several explanations for this phenomenon, it is believed that the main reason for this apparent paradox is linked to the fact that the Adriatic reservoirs are composed of thinly interbeded formations. This heterogeneity results in: the difficulty to correlate core depth and log depth with high precision; the fact that core and log measurements are made at two widely different scales. Systematic specific statistical data treatments are currently being developed by Agip to overcome these difficulties. From a rock mechanics point of view, index test are currently being developed as routine core measurements with a large sampling frequency, thus allowing the above mentioned statistical data treatment to be used. 6. CONCLUSIONS This paper has presented a pilot study to try to elaborate and interpret a rock property database at the scale of a coherent geological unit -namely the Northern Adriatic basin. The main conclusions which can be drawn from such a pilot study are the following: the geological logic of the selected basin can be used as a screen to allow the interpretation of the data within the database; for example, in the case of the Northern Adriatic basin, it has been demonstrated that depth and compaction alone controlled both porosity and strength of the rocks, thus leading to close correlation of the two parameters; specific trends valid for the entire region can be established reliably -e.g. the porosity/sonic/ permeability trends- thus allowing a reduced data acquisition campaign on future reservoirs; when confronted with operational request demanding urgent answers, the most significant analogue can easily be identified and a complete rock characterisation estall?lished within a few minutes only thus allowing meaningful engineering decision to be made in time; such a database can further be used to check the validity of models of various types -e.g. sand production prediction models; general engineering guidelines can be drawn rather easily from such a database as demonstrated by a twin paper presented at the same conference [2]. This feasibility being proven in the case of a basin with a simple history, it is intended to renew the experience in other geographical sectors where Agip has important operational activities. ACKNOWLEDGEMENT The authors would like to thank Agip S.p.A. for the permission to publish this paper and to express their sincere gratitude to all the colleagues that made this work possible. REFERENCES 1. Santarelli F.J.: "Rock mechanics characterisation of deep formations: a technico- economical overview," Paper SPEI/SRM 28021, to be published in Rock Mechanics for Petroleum Engineering, Proc. EUROCK'94, (1994). 2. Moricca G., Ripa F., Sanfilippo F. and Santarelli F.J.: "Basin scale rock mechanics: field observation of sand production" Paper SPEI/SRM 28066, to be published in Rock Mechanics for Petroleum Engineering, Proc. EUROCK'94, (1994). 111 3. Mikkelsen M. and Inderhaug O.H.: "Sand strength evaluation from logs and cores," Proc. 11th Eur. Formation Evaluation Symp., (1988), AA1-AA16. 4. Santarelli F.J., Detienne J.P. and Zundel J.P.: "Determination of the mechanical properties of deep reservoir sandstones to assess the likelyhood of sand production," Rock at Great Depth, Proc. ISRMlSPE Symp., 2, (1989), 779- 787. 5. Plumb R.A., Herron S.L. and Olsen M.P.: "Influence of composition and texture on compressive strength variations in the travis peak formation," Paper SPE 24758, Proc. 67th SPE An. Conf Exhib., n, (1992), 985-998. 6. Teeuw D.: "Prediction of formation compaction from laboratory compressibility data," SP E J., (September 1971). 7. Johnson J.P., Rhett D.W. and Siemers W.T.: "Rock mechanics of the Ekofisk reservoir in the evaluation of subsidence," Paper OTC 5621, Proc. 20th Offshore Tecnol. Corif., (1988), 39- 50. 8. Ruddy I., Andersen M.A, Pattillo P.D., Bishlawi M. and Foged N.: "Rock compressibility, compaction and subsidence in a high porosity Chalk reservoir: a case study of Valhall field," J.P.T., (July 1989),741-746. 9. Zoback M.D. and Zoback M.: "State of stress in the conterminous United State," J. of Geophys. Res., 85, (1980), 6113-6156. 10. Veeken C.AM., Davis D.R., Kenter C.J. and Kooijman A.P.: "Sand production prediction review: developping an integrated approach," Paper SPE 22792, Proc. 66th SPE An. Con! Exhib., III, (1991), 335-346. 11. Santarelli F.J., Dusseault M.B. and Yassir N.: Rock Properties for Petroleum Engineers: Problem Statement, Report of the ISRMlSPE commission, (1991), 80. 12. Bell J.S. and Gough D.I.: "Northeast-Southwest compressive stress in Alberta: evidence from oil wells," Earth Planet. Sci. Lett., 45, (1979), 475-482. 13. Gough D.1. and Bell J.S.: "Stress orientation from oil well fractures in Alberta and Texas," Can. J. Earth Sci., 18, (1981), 638-645. 14. Maury, V. and Sauzay, J.M.: "Borehole instability: case histories- rock mechanics approach and results," Paper SPElIADC 16051, (1987). 15. Chierici, G.L.: "Subsidenza naturale e da estrazione di fluidi dal sottosuolo," Agip internal report, (1992). 16. Mattavelli, L.,Novelli, L. and Anelli, L.: "Occurence of hydrocarbons in the Adriatic basin," In: Generation, accumulation, and production of Europe's Hydrocarbons, Ed. AM.Spencer (1991), 369-380. 112 17. Holt R., Brignoli M., Fjrer E., Kenter C.J. and Unander T.E.: "Core Damage Effects on Compaction Behaviour," Paper SPEIISRM, to be published in Rock Mechanics for Petroleum Engineering, Proc. EUROCK'94, (1994)