CARBONATE RESERVOIRS

Transcription

1 CARBONATE RESERVOIRS How to choose the right petrophysical evaluation method Evaluation of mineralogy, pore geometry, saturation and permeability Brian Moss Retired Steve Cannon Retired Petrophysics 101 Seminar Co-convened with PESGB Young Professionals The Geological Society, Burlington House 7 th March 2019 Brian Moss & Steve Cannon LPS March

2 What are carbonates? Rocks made up of calcareous minerals: Calcite CaCO 3 Aragonite CaCO 3 Dolomite CaMg(CO 3 ) 2 Stain in Alizarin Red to identify dolomite Fix in blue resin to show porosity Limestones are greater than 50% calcite and dolomites or dolostones are greater than 50% dolomite Brian Moss & Steve Cannon LPS March

3 Brian Moss & Steve Cannon LPS March

4 How are carbonates formed? Where? Carbonates are both allochthonous and autochthonous in origin and occur along stretches of coastline and subduction zones (form in situ or transported) Comprise atolls, carbonate ramps, rimmed shelf Geologically, although carbonate platforms may develop in a range of geotectonic settings, the majority can be found along passive continental margins and back arc basins to foreland basins Brian Moss & Steve Cannon LPS March

5 Comparison: clastics and carbonates Clastics Sandstones quartz, lithic fragments, carbonate, clays Transported from elsewhere and modified through compaction, lithification, diagenesis Principal reservoir quality controls: - Mineralogy, grain-size, sorting, texture (layering), and induration MINERALOGY AND TEXTURE Classic petrophysics geared to porosity and matrix, so are tuned to clastics Carbonates Calcite, aragonite, dolomite, with evaporites In-situ precipitation or organic growth and modified through compaction, lithification, diagenesis including dolomitization and dissolution Principal reservoir quality controls: - Pore-size distribution, pore connectivity, fracturing and dolomitization PORE CHARACTER Interpretation requires partitioning into units with similar pore characteristics, so pore-typing is paramount Brian Moss & Steve Cannon LPS March

6 Clastic vs Carbonate sequences Sediment supply mechanisms are fundamentally different giving a different response to sea level change Carbonate production is proportional to the area of flooded platform, thus greatest sedimentation occurs during a highstand Rate of carbonate production may exceed sea-level rise or subsidence leading to progradational/aggradational packages with shoal profiles even though sea-level rises When the accommodation space is filled by sediment, the excess may be shed off the slopes to deeper water Carbonate ramps tend to have smaller producing areas and behave like siliciclastic shelves Brian Moss & Steve Cannon LPS March

7 Depositional processes Tidal flat progradation - Results in the formation of upward-shallowing sequences via re-deposition of subtidal sediments over tidal flats and beach ridges during major storm periods Reef progradation - Seaward growth of the reef over the fore-reef slop at rimmed shelf margins; reefal sequences of various facies types will be produced, particularly vertically Carbonate sand migration - In high-energy locations, the migration of carbonate sand bodies is an important depositional process, particularly on ramps and sand shoals or shelf margins Offshore storm transport - Results in the deposition of shore-face carbonate sediments Slumps, slides, turbidity currents, debris flows - Are various types of re-sedimentation of previously deposited sediments. They are very common and mostly restricted to shelf margins and slopes Brian Moss & Steve Cannon LPS March

8 Carbonate sediments and controls Most carbonate sediments are produced in water depth less than 50m with little clastic input Comprise skeletal and non-skeletal components bioclasts and grains - Organic/skeletal corals, algae, molluscs bryozoa - Inorganic/non-skeletal coated grains (ooids), peloids, clasts: generally slower build up rate Main controls are water depth, temperature and photosynthetic activity in top few tens of metres At higher latitudes fewer major reef builders and increase in bio-clastic shoals and patch reefs Carbonate factories can be killed off when drowned Growth LAW OF SIGMOIDAL GROWTH Time Lag phase Log phase Keep up Catch up Start up After Neumann and Macintyre, 1985 Brian Moss & Steve Cannon LPS March

9 Skeletal components Skeletal organisms have varying mineralogy and these have varied with geological time possibly related to global climatic change and sea level Calcite seas dominate during greenhouse conditions when sea levels are high and aragonite during icehouse conditions when sea levels are lower Brian Moss & Steve Cannon LPS March

10 Non-skeletal components grains/mud Carbonate solubility in seawater (after Bathurst 1971) ooids - CO 2 +H 2 O+CaCO 3 Ca +2HCO 3 grapestone Solubility driven by: - Temperature (lower solubility in warmer water) - Acidity/alkalinity (High CO 2 = acid = dissolution) - Pressure (increases solubility) peloids, shells, forams Calcite dissolves in deep ocean water 2cm Brian Moss & Steve Cannon LPS March

11 Carbonate cycles During a sea level fall the area of carbonate production is substantially reduced forming a lowstand wedge There is little erosion during the lowstand, only chemical weathering Expect to see a different type of carbonate grain composition deposited in the lowstand wedge With a subsequent rise carbonate production increases Sea level fall Sea level rise Brian Moss & Steve Cannon LPS March

12 High resolution sequence stratigraphy Detailed sequence stratigraphy requires the identification of parasequences Core and log integration provides the key to the high resolution breakdown If pattern is regular and predictable then a deterministic approach to stratigraphy can be used Brian Moss & Steve Cannon LPS March

13 Diagenetic overprinting Basic processes to consider: - Calcite cementation - Compaction - Selective dissolution - Dolomitization - Evaporite mineralization - Massive Dissolution Each results in a specific rock-fabric; processes may overlap in time and space Timing of events usually associated with deposition, meteoric flushing, later burial and hydro-thermal events Petrography can determine the sequence of events Brian Moss & Steve Cannon LPS March

14 Generalized diagenetic sequence Brian Moss & Steve Cannon LPS March

15 Early diagenesis Marine deposition Isopachous fringes Marine vadose Meteoric flushing Limited compaction due to drusy cementation. Mouldic porosity Brian Moss & Steve Cannon LPS March

16 Burial Diagenesis Sparry calcite Non-ferroan and ferroan Saddle dolomite Non-ferroan and ferroan Fills intra-grain voids and mouldic porosity to varying degrees Brian Moss & Steve Cannon LPS March

17 Burial Diagenesis Exotic minerals fluorite, kaolinite, anhydrite, celestite. Blue John Often associated with petroleum filling but no fluid inclusion data. Late stage secondary porosity. Brian Moss & Steve Cannon LPS March

18 Comparison of Reservoir Quality Basin 1 Basin 2 Poroperm : 10-12%, md Porosity similar, perm. lower Brian Moss & Steve Cannon LPS March

19 Roof collapse after dolomitization Brian Moss & Steve Cannon LPS March

20 Core to Log Integration Use of limited core data to calibrate readily available log data Key is accurate and consistent depth matching and QC every step Core description should include: - Dominant lithology, grain types colour, texture grainsize/sorting - Interparticle porosity, dolomite crystal size - Separate and touching vug space How can basic rock-fabric characteristics be related to log data. Brian Moss & Steve Cannon LPS March

21 Core to log integration (2) Interparticle porosity is estimated by subtracting separate vug porosity, estimated from sonic logs, from total porosity from density/neutron logs Grain size and sorting can be estimated from gamma ray, but better still from interpreted resistivity logs that give water saturation Touching-vug pore-space is best estimated from high resolution image logs Exotic minerals can be identified by Spectral Gamma Ray and Photo-Electric Factor (PEF) logs; flooding surfaces, evaporites, dolomitic zones Brian Moss & Steve Cannon LPS March

22 Core/Log Calibration A two step process: 1. Describe the cores and capture the data in graphical and numerical formats 2. Accurately depth shift the cores by: - comparing core-gamma with wireline gamma - comparing core porosity with log porosity - comparing core mineralogy with logderived mineralogy Present the results in a composite 1:200 scale format including major stratigraphic/zonal markers Brian Moss & Steve Cannon LPS March

23 Will a carbonate reservoir produce hydrocarbons? Carbonates reservoirs are very difficult to evaluate because Heterogeneity many different pore types impact resistivity logs Dual permeability systems matrix, vugs, fossil moulds, fractures Diagenetic overprinting dissolution, dolomitization, micritization Wettability commonly oil to intermediate but how to measure Need to develop integrated methods using thin sections, cores, logs, well tests, seismic Archie saturation equation works in intergranular or intercrystalline porosity: S w a R w = m R t φ Brian Moss & Steve Cannon LPS March S w a Rw = m Rt φ 1 n 1 n

24 Pore space terminology Confusing number of classification schemes, many are hybrid - Archie (1952) textural, classically petrophysical but does not relate to geology - Dunham (1962) textural, useful for determining depositional energy - Folk (modified 1962) compositional and textural - Choquette and Pray (1970) pore space genesis, fabric selectivity, does not distinguish isolated and connected pores - Lucia (1983) petrophysical, distinguishes between interparticle and vuggy porespace Pore space should be defined in terms of rock fabrics, texture and petrophysics to capture geology and engineering Most schemes are based only on petrography Brian Moss & Steve Cannon LPS March

25 Dunham s 1962 Classification Varies from chalks through limestones to dolomites; muds, grains, shells, corals, crystals Can contain evaporates e.g. as disseminated anhydrite; Can contain clay minerals and other conductive material e.g. pyrite; Pore geometry and mineralogy may only be weakly correlated Brian Moss & Steve Cannon LPS March

26 Folk s 1962 Classification Varies from chalks through limestones to dolomites; muds, grains, shells, corals, crystals Can contain evaporates e.g. as disseminated anhydrite; Can contain clay minerals and other conductive material e.g. pyrite; Pore geometry and mineralogy may only be weakly correlated Brian Moss & Steve Cannon LPS March

27 Pore types Did you say porosity or porosity? Brian Moss & Steve Cannon LPS March

28 PORE TYPES ARCHIE (1952) Intergrain Intercrystal Matrix Moldic Intrafossil Shelter Visible A, B, C, and D Cavernous Fracture Solution Choquette & Pray (1970) LUCIA (1983) Interparticle Fabric Selective Separate Vuggy Non-Fabric Selective Touching Brian Moss & Steve Cannon LPS March

29 Central issue Disparate pore character at the root of interpretation difficulties Classic petrophysical methods have been developed for clastics Intergranular phi, some fractures; texture; mineralogy (e.g. clays) Carbonates are different Interparticle, intraparticle; variable interaction with secondary pore system comprising dissolution voids and fractures; mineralogy arguably simpler Heterogeneous at all scales Require delineation of pore facies principally type and connectivity For prediction: Distribution and mode of formation of phi layering, core examination, mineralogy - downhole imagery, chemical logging For reservoir quality: Volume, pore size distribution and connectivity Pc, NMR, Dielectric, classic logs (nuclear, resistivity) Brian Moss & Steve Cannon LPS March

30 Partitioning terminology - 1 Facies (Gressly, 1838) distinctive rock unit forming under certain conditions, reflecting a particular process or environment Electrofacies (Serra et al., 1982) log-identifiable rock types Lithofacies (Miall, 1990) rock types with distinct texture and mineralogy Petrofacies ( AAPG Watney et al., 1998) petrophysically distinct pore types and fluid saturations expressed on a phi-sw cross-plot (aka the Pickett plot) Rock-fabric facies (Jennings and Lucia, 2003) rock types with characteristic range of pore size (inferred from particle size and sorting), characteristic distribution of interparticle porosity and characteristic nature and size of vuggy porosity; fracture porosity and dissolution are modifiers Electroporefacies (Bust et al, 2011) rock-fabric facies partitioned in terms of pore character that are log-identifiable. Logs required to accomplish this go beyond the standard suite Partitioned units must be geologically architecturally significant correlatable and distributable through geological controls Brian Moss & Steve Cannon LPS March

31 Partitioning terminology - 2 Petrophysics is a remote-sensing discipline Outside the laboratory, downhole methods sensing formations that cannot directly be touched Often we cannot directly observe the phenomena of interest but use physical measurements that are heavily, but not exclusively, influenced by the parameters of interest such relations established through algorithms To be differentiated by petrophysical measurements, rock types require a sufficiently exclusive set of algorithms (functions); as such they become Petrofacies (Worthington, 2002). Petrofacies may equal electroporefacies, but not necessarily; former are often discarded whilst latter are predictable beyond the sample m and n may vary by porosity and saturation The estimation of fractures and fissures largely the same as for clastics Brian Moss & Steve Cannon LPS March

32 Pore volume Pore volume on its own generally can not sufficiently distinguish fabric-related reservoir character (and hence flow potential) Brian Moss & Steve Cannon LPS March

33 Beyond pore volume Typical scatter of porosity:permeability data renders pore volume on its own inadequate as a discriminant of permeability Bring fabric into the mix and order emerges Fabric is related to environment and hence amenable to geological controls Brian Moss & Steve Cannon LPS March

34 Petrophysical classification Permeability and saturation are related to pore-size distribution, which in turn is related to rock fabric Pore-size distribution is related to rock fabric through a poretype: 1. Interparticular between grains and bio-clasts 2. Separate-vug dissolved fossils, molds 3. Touching-vug inter-connected vuggy systems Each class has a different pore-size distribution and interconnection and will be represented by a different volume or proportion in the reservoir After Lucia, 1999 Brian Moss & Steve Cannon LPS March

35 Interparticle pore-space Interparticle PSD can be described in terms of particle size, sorting and porosity Particle size can be related to Pc displacement pressure through pore size Large pores are filled first then smaller pores as pressure increases After Lucia, 1999 Brian Moss & Steve Cannon LPS March

36 Porosity, particle size and capillary pressure Lucia (1983) showed the relationship between displacement pressure and particle size for MICP data on a range of porosity classes with permeability >0.1mD Suggests boundaries at 20 μm and 100μm related to different permeability fields Refined using textural information grain or mud dominated: in reality a continuum Mercury displacement (psia) After Lucia, Average particle size (μm) Brian Moss & Steve Cannon LPS March

37 Porosity/permeability relationships Non-vuggy limestones Grainstone PSD controlled by grain size and sorting Packstone PSD controlled by particle size and size of micrite particles between grains Muddy/Wackstone PSD controlled by size of micrite particles Grainstone: φ=25%, k=15d GrnDomPkst: φ=16%, k=5.2md MdDomPkst: φ=18%, k=4md After Lucia, 1999 Wackestone: φ=33%, k=9md Brian Moss & Steve Cannon LPS March

38 Porosity/permeability relationships Non-vuggy dolomites Dolomitization changes rock fabric significantly; remnant textures may be preserved Dolomite crystals range from 10 s of micron up to 200μm so that micrite particles can increase by 2 OOM In muddy fabrics permeability increases as crystals grow GrnDomDolPk: φ=9% k=1md Dolgrnst: φ=7.1%, k=7.3md After Lucia, 1999 DolWkst: φ=20%, k=4d DolWkst: φ=16%, k=30md Brian Moss & Steve Cannon LPS March

39 Texture and pore type From Lucia, 1999, 2007 Brian Moss & Steve Cannon LPS March

40 Rock fabric classes Class 1(>100μm) limestone and dolomitized grainstones; large crystalline, grain dominated dolopackstone and muddy dolostones Class 2(100-20μm) grain dominated packstones; fine-med crystalline, grain dominated dolopackstone; medium crystalline muddy dolostones Class 3(<20μm) mud dominated fabrics and fine crystalline dolostones After Lucia, 1999 Graph is both limestone and dolomite non-vuggy fabric; dotted lines RMA Brian Moss & Steve Cannon LPS March

41 Rock fabric classes RMA transforms exist for each class to predict permeability: 1. k = (45.35x10 8 )xφ ip (r=0.71) 2. k = (2.040x10 6 )xφ 6.38 ip (r=0.80) 3. k = (2.884x10 3 )xφ ip (r=0.81) In reality there is a continuum from mudstone to grainstone and from very fine dolostone to coarse crystalline dolostone Diagenesis can produce unique types of interparticle porosity that need to be considered, e.g. collapse of vuggy fabrics may result in diagenetic particles Graph is both limestone and dolomite non-vuggy fabric; dotted lines RMA After Lucia, 1999 Brian Moss & Steve Cannon LPS March

42 Capillarity gives pore geometry From Lucia, 1999, 2007 Brian Moss & Steve Cannon LPS March

43 Porosity/saturation relationships A common way to relate porosity, permeability and water saturation is to use a J-function: J = In reality Sw = f(pc, φ ip, rock-fabric) where Pc is related to reservoir height Each rock-fabric class has a defined relationship to water saturation and height After Lucia, Pc σ cosθ K φ Conversion Values 3.162Pc J = σ cosθ K φ Laboratory (Hg/air/solid) Reservoir (Oil/water/solid) σ (dynes/cm) θ (degrees) Water density (g/cc) Brian Moss & Steve Cannon LPS March

44 Porosity/permeability/saturation relationships Each rock-fabric class integrates permeability and water saturation with interparticle porosity and reservoir height: Sw = A*H B * φ ip C A B C Class Class Class After Lucia, 1999 Brian Moss & Steve Cannon LPS March

45 A word on vugs - I From Lucia, 1999, 2007 Brian Moss & Steve Cannon LPS March

46 Separate-vug pore-space Separate-vug pore-space can be connected inter-particle porosity Typically fabric selective such as intrafossil (chambers) or dissolved grains Larger examples are fossil moulds or dissolved crystals in mud-dominated lithologies Grain-dominated rocks may show composite moulds solution fabrics? After Lucia, 1999 Brian Moss & Steve Cannon LPS March

47 Touching-vug pore-space Significantly larger than particle size and forms an inter-connected pore system Usually non-fabric selective caverns, breccia, fractures and solution-enlarged features, karsts Fractures are included because of the pore-type not the way in which they formed After Lucia, 1999 Brian Moss & Steve Cannon LPS March

48 Petrophysics of vuggy pore-space Separate vugs are connected through the matrix Can significantly increase porosity but may not impact permeability Separate vugs usually considered to be oil-saturated unless interparticle porespace is impermeable Intragrain microporosity may trap water leading to high water saturation within an otherwise productive interval Touching vugs lead to greatly improved permeability but are difficult to characterise and predict e.g. fractures After Lucia, 1999 Plot shows bi-modal nature of intergrain and intragrain microporosity Brian Moss & Steve Cannon LPS March

49 A word on vugs - II From Lucia, 1999, 2007 Brian Moss & Steve Cannon LPS March

50 More on vugs From Lucia 2007: needs core calibration In Limestones Φ sv = [DT-141.Φ total ] In Dolostones Φ sv = [DT-145.Φ total ] Brian Moss & Steve Cannon LPS March

51 Lucia s workflow For irreducible hydrocarbon zone above transition zone and pre-production Calculate φ total using available logs; calibrate to core Calculate φ sv using sonic travel-time; calibrate to core Calculate interparticle porosity: φ ip = (φ total - φ sv ) Calculate petrophysical class number: Log(rfn) = [A + B.log(φ) + log(sw i )] / [C + D.log(φ)] rfn = rock-fabric number (range 0.5 to 4) (aka petrophysical class) S wi = irreducible water saturation φ = total porosity A = ; B = ; C = ; D = Calculate permeability using the global transform: Log(k) = [a - b.log(rfn)] + [c - d.log(rfn)].log(φ ip ) a = ; b = ; c = ; d = Brian Moss & Steve Cannon LPS March

52 Comparison of class schemes Industry classification schemes do not all emphasise the same attributes From Bust et al Brian Moss & Steve Cannon LPS March

53 Hydraulic Flow Units Amafule et al Amafule, J.O., Altunbay, M., Tiab, D., Kersey, D.G., and Keelan, D.R., 1993, Enhanced Reservoir Description: Using core and log data to identify hydraulic (flow) units and predict permeability in uncored intervals/wells. Proceedings of the 68 th SPE Annual Technical Conference and Exhibition, October 3-6, 1993, Houston, Texas, Society of Petroleum Engineers, pp kk ee = ee 1 1 ee FF ss ττss gggg k = permeability (md) Φ e = effective porosity F s = shape factor τ = tortuosity S gv = surface area per unit grain volume RQI = Reservoir Quality Index (µm) Φ z = pore volume to grain volume ratio FZI = Flow Zone Indicator (µm) FZI RQI φ e φ = z 1 φe = 2 2 Fsτ Sgv RQI φ Care is required in deriving HFU from log data if k = f( ) = llllll RRRRRR llllll zz k φ = e z = llllll(ffffff) Brian Moss & Steve Cannon LPS March

54 Using Pc curves - 1 WINLAND R35 Figure 4 Grouping of the MICP curves by R35 value, denoted by colour. Left: All groups. Right: selected groups illustrating that because the only criterion used in grouping MICP curves is the value of pressure at mercury saturation of 35% (red line), this approach results in a wide range of MICP curve shape, and hence pore geometry, being incorporated within each R35 group. Brian Moss & Steve Cannon LPS March 2019 From Skalinski et al

55 Using Pc curves - 2 Lower asymptote (~smallest pore system heterogeneity) Inflexion point (~effective pore radius) Slope (~overall pore system heterogeneity) Hyperbolic asymptote (~entry pressure) Higher asymptote (~closure effect magnitude) Example summarising parameters marked in red on the crossplot Figure 5 Illustrative hyperbolic tangent function for a single MICP sample, showing its summarizing parameters and their physical meaning. These parameters are calculated for each curve, and the resulting set of parameters is partitioned for effective grouping amongst them. Such grouping forms the basis for the MICP derived rock types (Figure 8). The parameter constant is not shown on this plot it is an overall scaling constant used to reflect position of samples in the axes (generally high or generally low pressures etc.). The lower asymptote is also not highlighted because it has no easily identifiable physical significance in normalized saturation data such as used in this study. From Skalinski et al Brian Moss & Steve Cannon LPS March

56 Using Pc curves - 3 Figure 6 The coherent points in the middle crossplot of pairs of summarizing parameters from the hyperbolic tangent functions for all curves reflect the single-mode capillary curves (left), with multi-mode capillary curves (right) plotting offtrend [NB. The colour in the middle and left plots indicates that between-plot selection is active; the selection shows that single-mode samples (i.e those with unimodal pore-size distributions) plotted in the left graph correspond to the red points in the middle graph that form a coherent relationship in terms of the two parameters plotted (constant and inflexion point). The scatter in the black points in the middle graph is much greater; these black points are off-trend in the middle graph; these black points are from multi-mode capillary curves and are plotted in the right-hand graph.] From Skalinski et al Brian Moss & Steve Cannon LPS March

57 Using Pc curves - 4 Figure 7 Sorted and organized Self-Organised Map, subdivided into 9 groups (the colours) by reference to the full capillary curve data as described by the MICP tanh function summary parameters. Figure 8 Capillary pressure data differentiated by summary parameter group; groups 1-5 (left plot) relate to multi-modal capillary data and groups 6-10 (right plot) relate to uni-modal capillary data. Groups are generally much more coherent than those found from R35 groupings alone. Brian Moss & Steve Cannon LPS March 2019 From Skalinski et al

58 Another approach E.A.Clerke, Saudi Aramco - I Brian Moss & Steve Cannon LPS March

59 Another approach E.A.Clerke, Saudi Aramco - II Brian Moss & Steve Cannon LPS March

60 Another approach E.A.Clerke, Saudi Aramco - III Brian Moss & Steve Cannon LPS March

61 A word on Fractures Some carbonates fracture more readily than others and even than clastics Dolomites fracture most readily because they resist compaction during burial Evaporites can provide support framework, but will fracture Faults in carbonates rarely seal Fracture porosity is very small, generally adding <1-4% pu Fracture permeability is huge, think SuperK layers in Saudi Arabia - are they fractures or karst? Expect the unexpected! From Lucia, 1999, 2007 Brian Moss & Steve Cannon LPS March

62 Core calibration of log analysis Could be its own seminar Representative core samples: Plugs may not be the requisite 2 orders of magnitude greater than the largest pore size; may need higher sampling to cope with heterogeneity Whole cores necessary, unless interparticle porosity is proved use a size equivalent to log responses X-ray CT scans and a quantitative definition of heterogeneity are useful How many samples? At 95% confidence to within p% tolerance: ss NNNN = 200. [ aa mm. pp ]2 (Ref. Hurst & Rosvoll, 1991; s = sample std.dev. and a m = sample average value) Stress sensitivity Responses of porosity and permeability to stress vary considerably with pore type and particle type May lead to a concept of a stress facies Brian Moss & Steve Cannon LPS March

63 Heterogeneity From Lucia, 2007 Brian Moss & Steve Cannon LPS March

64 Porosity types Logs measure conductivity of reservoir fluid: C=1/R If Sw is 100% then: C b = (1-φ)C m + φc f ; because C m = 0, C b = φc f In more complex reservoirs a cementation factor m is introduced C b = φ m C f How does it change when hydrocarbons are present? From Asquith,1985 Brian Moss & Steve Cannon LPS March

65 What happens with hydrocarbons and water? When Sw<100% then C b = φ m C f x S w n n is the saturation exponent usually estimated as 2. Conductivity decreases as m increases but also when S w decreases a is the tortuosity factor and not always equal to 1 in carbonates In carbonates you must get m right for every point interpretation Rt = aφ m R w S n w Brian Moss & Steve Cannon LPS March

66 Determination of cementation factor Traditionally making electrical measurements on core samples and using x-plot techniques Pickett plots, plot deep resistivity against porosity from logs; m is the slope of the line in the water leg With two porosity measurements sonic (φ s ) and/or density-neutron (φ t ) you have another method Sonic measures matrix porosity only D-N measures total porosity Doesn t work quite so well in oomoldic limestones φ m φ 2(logφs ) m logφ vug = 2 t s ( φ φ ) = φ φ t t vug Brian Moss & Steve Cannon LPS March

67 m is not always constant From Lucia, 2007 From Focke and Munn, 1987 Brian Moss & Steve Cannon LPS March

68 Formation Factor vs Porosity - Core data Intergranular φ m ~ 2 a ~ Brecciated samples m = ~1.25 a > Formation factor (Ro/Rw) ohm Plug porosity % F Brian Moss & Steve Cannon LPS March

69 a and m in resistivity evaluation can be very different from standard values in fractured intervals All data Selected data With > 1% porosity Red dashed line approximates a = 1, m = -2

70 Logging Tool response Fabric-sensitivity is the key. - Pore type and connectivity, rather than simple pore volume, are the controlling characteristics of interest. - The standard suite [SP, GR, density, neutron, resistivity, sonic], responds predominantly to pore volume, and is only weakly fabric-sensitive - Pay attention to log resolution - Need higher resolution logs e.g. images Brian Moss & Steve Cannon LPS March

71 Image magic - I From Lucia, 2007 Brian Moss & Steve Cannon LPS March

72 Image magic - II From Lucia, 2007 Brian Moss & Steve Cannon LPS March

73 Image magic - III From Lucia, 2007 Brian Moss & Steve Cannon LPS March

74 Logging Tool response Log signatures require more than simple mineral/porosity discrimination - Sonic wavetrain (Stoneley) measurements - An elemental/geochemical log - A dielectric log - A magnetic resonance log - Formation testers - Production logs (with well tests) Such a log suite should be run in at least one key reference well, which should also be fully cored Brian Moss & Steve Cannon LPS March

75 Case Study Canada, Bitumen in Vuggy Dolomite Modern solutions LithoScanner* NMR (CMR*) Shortest echo spacing captures fluid porosity in smallest pores. Organic matter decays too fast to be measured Dielectric (ADT*) Responds to water-filled porosity from dielectric permittivity dispersion Deliver accurate saturation estimates in difficult conditions *Mark of Schlumberger Brian Moss & Steve Cannon LPS March 2019 Ref. SPE , 2013, P.R.Craddock et al. 75

76 Workflow schematics Ref. Bust et al., 2011, SPE Brian Moss & Steve Cannon LPS March

77 Conclusions It s mostly about pore geometry - need to distinguish and classify, then find log-definable partitions Mineralogy may be important but the degree of importance needs to be established Pay attention to sampling plans Use modern logging tools including images for at least the key well(s), which should be cored Make no assumptions as to appropriate algorithms or methodology - Use Exploratory Data Analysis to investigate relations/partitions/functions Be open for the most effective partitioning criteria Be open to non-archie behaviour Brian Moss & Steve Cannon LPS March

78 Some key references Asquith, George B., Handbook of log evaluation techniques for carbonate reservoirs, Methods in Exploration 5, AAPG Bust, V.K., Oletu, J.U., Worthington, P.F., 2011, The Challenges for Carbonate Petrophysics in Petroleum Resource Estimation. SPE Reservoir Evaluation and Engineering, February SPE Chilingarian, G.V., Mazullo, S.J., Rieke, H.H., 1992, Carbonate Reservoir Characterisation: a geologic engineering analysis part I. Elsevier. Developments in Petroleum Science 30 Chilingarian, G.V., Mazullo, S.J., Rieke, H.H., 1996, Carbonate Reservoir Characterisation: a geologic engineering analysis part II. Elsevier. Developments in Petroleum Science 44 Clerke, E.A., et al., SPWLA Annual Symposium, 2014 Emery, D., & Meyers, K.J. (eds), Sequence Stratigraphy, Blackwell Science, Oxford Focke, J.W., and Munn, D., 1987, Cementation Exponents in Middle Eastern Carbonate Reservoirs. SPE Formation Evaluation 2 (2); p SPE Lucia, F.J., 1999, Carbonate Reservoir Characterisation. Springer, 226pp. Lucia, F.J., 2007, Carbonate Reservoir Characterisation. An Integrated Approach. Springer, 336pp. 2 nd Edition Moore, C.H., 2001, Carbonate Reservoirs. Porosity Evolution and Diagenesis in a Sequence Stratigraphic Framework. Elsevier. Developments in Sedimentology 55. Skalinski, M., Gottlib-Zeh, S. and Moss, B.P., 2005, Defining and Predicting Rock Types in Carbonates Preliminary Results from an Integrated Approach using Core and Log Data from the Tengiz Field. SPWLA 46 th Annual Logging Symposium, June 19-22, 2005, New Orleans, Louisiana, Paper Z. Skalinski, M., Se, Y., Playton, T., Theologou, P., Narr, W., Sullivan, M. and Mallan, R., 2015, Petrophysical Challenges in Giant Carbonate Tengiz Field, Republic of Kazakhstan. PETROPHYSICS Vol 56, No. 6 (December 2015); Pages Sung, R.R., Clerke, E.A. and Buiting, J.J., 2013, Integrated Geology, Sedimentology and Petrophysics Application Technology for Multimodal Carbonate Reservoirs. Saudi Aramco Journal of Technology, Fall Tucker, M.E. and Wright, V.P., 1991, Carbonate Sedimentology. Blackwell Scientific Publications Wright and Burchette in Reading (1996); Sedimentary Facies Brian Moss & Steve Cannon LPS March

79 Brian Moss & Steve Cannon LPS March

80 Case study Middle East Carbonate Ideally both core and log data are required for a robust HU solution There is a fundamental requirement for accurate depth matching of the two data sets Generally a RRT scheme will be in place against which these results will be calibrated However by starting with preconceived notions of variables such as HAFWL and log derived water saturations, can make this exercise a major challenge! Dominant Lithology MRT FZI > 1 MRT FZI < 1 Mean Phi GMean Perm Grainstone MRT 1 N/A Floatstone MRT 2 N/A Grain Dominated Packstone N/A MRT Packstone MRT 4 MRT / /2.8 Microsparite MRT 6 MRT / /1.0 Muddy Packstone/ Wackestone MRT Cemented Reservoir MRT Dense Layers MRT Brian Moss & Steve Cannon LPS March

81 Spreadsheet analysis Core data should be reviewed and cleaned of poor quality data The HU terms PhiZ, RQI and FZI should be calculated 1.00 RQI vs PhiZ The results can be grouped by rocktype and displayed for trends or predetermined FZI classes used The mean value for each FZI class is used to predict permeability from porosity: RQI 0.10 FZI <0.25 FZI FZI FZI FZI >1.5 k = 1041( F zi ) 2 φ 3 e ( 1 φ ) 2 e PhiZ Brian Moss & Steve Cannon LPS March

82 Quality control Core Permeability vs Predicted Permeability Plot of core permeability against predicted permeability shows excellent 1:1 correspondence except for the best RRT s: r 2 value is Predicted Permeability 1000 y = x R 2 = MRT 1 MRT 2 MRT 3 MRT MRT5 MRT 6 MRT7 MRT All Data Trend Core Permeability Brian Moss & Steve Cannon LPS March

83 Porosity against Predicted Permeability 1000 MRT 1, 2 & 4 are the best reservoir rock types in terms of reservoir quality: grainstones, grain-dominated packstones and floatstones 100 MRT 6 represents a good quality microsparite MRT 3, 5, 7 & 8 are mud dominated rocktypes: muddy packstones and wackestones 10 1 MRT 1 MRT 2 MRT 3 MRT 4 MRT 5 MRT 6 MRT 7 MRT Brian Moss & Steve Cannon LPS March

84 Not-used slides follow Brian Moss & Steve Cannon LPS March

85 Petrophysical Evaluation Two primary sources of data: Wireline Log: in situ, in direct measurement Core: ex situ, direct measurement Challenges: Sampling issues: volume and bias Data quality: acquisition and interpretation Integration: correlation and calibration Petrophysics is not just log analysis Reservoir evaluation is: Petrophysics constrained by geology Brian Moss & Steve Cannon LPS March

86 Describing the rock fabric Spatial geological data integrated with quantitative engineering data leads to a description of the reservoir rock fabric Results in a generic petrophysical classification of the carbonate pore space All inputs are needed to describe the reservoir for 3D reservoir property modelling Bureau of Economic Geology, UTA F. Jerry Lucia Brian Moss & Steve Cannon LPS March

87 Use of bulk volume water (BVW) Bulk volume water is product of water saturation and porosity: BVW = S w x φ BVW indicates whether a reservoir is at irreducible water saturation when all the water is held by capillary forces and therefore produces dry oil X-plots of S wirr and porosity on BVW charts identify where water free production might occur Brian Moss & Steve Cannon LPS March

88 Brian Moss & Steve Cannon LPS March