Fractured Hydrothermal Dolomite Reservoirs in the Devonian Dundee Formation of the Central Michigan Basin

Home , Bell Shale, Michigan Basin, Rogers City Limestone

Fractured hydrothermal AUTHORS John A. Luczaj Department of Natural and Ap- dolomite reservoirs in plied Sciences, University of Wisconsin–Green Bay, Green Bay, Wisconsin 54311; [email protected] the Devonian Dundee John Luczaj is an assistant professor of earth science in the Department of Natural and Applied Sciences at the University of Wisconsin–Green Bay. He earned Formation of the central his B.S. degree in geology from the University of Wisconsin–Oshkosh. This was followed by an M.S. Michigan Basin degree in geology from the University of Kansas. He holds a Ph.D. in geology from Johns Hopkins Uni- versity in Baltimore, Maryland. His recent interests John A. Luczaj, William B. Harrison III, and include the investigation of water-rock interaction in Natalie Smith Williams Paleozoic sedimentary rocks in the Michigan Basin and eastern Wisconsin. Previous research activities involve mapping subsurface uranium distributions, reflux dolomitization, and U-Pb dating of Permian ABSTRACT Chase Group carbonates in southwestern Kansas. The Middle Devonian Dundee Formation is the most prolific oil- William B. Harrison III Michigan Basin Core producing unit in the Michigan Basin, with more than 375 million bbl Research Laboratory, Western Michigan University, of oil produced to date. Reservoir types in the Dundee Formation Kalamazoo, Michigan 49008; [email protected] can be fracture controlled or facies controlled, and each type may William B. Harrison, III, is the director of the Michigan have been diagenetically modified. Although fracture-controlled res- Basin Core Research Laboratory and is professor emeritus in the Department of Geosciences at West- ervoirs produce more oil than facies-controlled reservoirs, little is ern Michigan University. He is also the director of known about the process by which they were formed and diageneti- the Michigan Center of the Midwest Region of the cally modified. Petroleum Technology Transfer Council. He holds In parts of the Dundee, preexisting sedimentary fabrics have a Ph.D. in paleontology and sedimentology from the been strongly overprinted by medium- to coarse-grained dolomite. University of Cincinnati. His interests include pa- Dolomitized intervals contain planar and saddle dolomite, with leontology and stratigraphy of Ordovician and Silurian carbonates in the central United States, minor calcite, anhydrite, pyrite, and uncommon fluorite. Fluid- oil and gas resources of the Michigan Basin, Devo- inclusion analyses of two-phase aqueous inclusions in dolomite nian stratigraphy and depositional facies of the and calcite suggest that some water-rock interaction in these rocks Michigan Basin, and methods of improved oil recov- occurred at temperatures as high as 120–150jC in the presence of ery from depleted or abandoned oil and gas fields. dense Na-Ca-Mg-Cl brines. These data, in conjunction with pub- Natalie Smith Williams Department of lished organic maturity data and burial reconstructions, are not Geosciences, Western Michigan University, Kala- easily explained by a long-term burial model and have important mazoo, Michigan 49008 implications for the thermal history of the Michigan Basin. The Natalie Smith Williams holds an M.S. degree in earth data are best explained by a model involving short-duration trans- science from Western Michigan University and a port of fluids and heat from deeper parts of the basin along major B.A. degree in geology from DePauw University. fault and fracture zones connected to structures in the Precam- brian basement. These data give new insight into the hydrothermal ACKNOWLEDGEMENTS processes responsible for the formation of these reservoirs. We acknowledge a grant from the U.S. Department of Energy (Project Number DE-AC26-00BC15122) awarded to J. R. Wood, T. J. Bornhorst, W. B. Harrison, III, and W. Quinlan that partially supported this project. Additional support was made available from the University of Wisconsin–Green Bay. Drill cores and other materials were available through the Michigan Basin Core Research Laboratory. The authors also Copyright #2006. The American Association of Petroleum Geologists. All rights reserved. thank James Wood, Robert Gillespie, and David Barnes Manuscript received May 5, 2005; provisional acceptance December 1, 2005; revised manuscript for ideas and discussion regarding this research and received June 14, 2006; final acceptance June 27, 2006. James Duggan for reviewing the manuscript. DOI:10.1306/06270605082

AAPG Bulletin, v. 90, no. 11 (November 2006), pp. 1787–1801 1787 INTRODUCTION part of the basin. Regionally extensive dolomite in the Dundee Formation is mainly present in the western The Michigan Basin is the classic example of an intra- parts of the basin, although most oil-producing dolo- cratonic sedimentary basin. It contains as much as mitized reservoirs of the Dundee Formation are lo- 5 km (3.1 mi) of Paleozoic and Mesozoic sediments cated in the central part of the basin (Gardner, 1974), that include carbonate, siliciclastic, and evaporite sedi- where the maximum production appears to be related ments (Sleep et al., 1980). The Devonian Dundee For- to fractured, vug-bearing intervals (Montgomery et al., mation presently lies 3200–4000 ft (975–1200 m) 1998). The fields in the central Michigan Basin are in- below the surface in the study area in the central part terpreted as discrete structures with a similar style of of the Michigan Basin (Figures 1, 2). Although the faulting present among various fields. Montgomery Dundee Formation is formally undifferentiated in the et al. (1998) presented additional geologic, stratigraph- subsurface (Catacosinos et al., 2001), it is correlative ic, and production data on the Dundee Formation in to both the Rogers City and Dundee formations along the Michigan Basin. the outcrop belt. However, the Rogers City and Reed The Middle Devonian Dundee Formation is the City units have typically been used as informal mem- most prolific oil-producing unit in the Michigan Basin, ber names to describe parts of the Dundee Formation with more than 375 million bbl of oil produced to date (Figure 2) for subsurface investigations. It consists of from 137 fields, with about half of that production mudstones through grainstones deposited along a car- coming from dolomite-hosted reservoirs (Gardner, 1974; bonate bank and open-marine environment in the cen- Curran and Hurley, 1992; Montgomery et al., 1998; tral and eastern parts of the basin, with lagoon and Wylie and Wood, 2005). Reservoir types can be frac- sabkha-type environments dominant in the western ture controlled or facies controlled, and each type may

Figure 1. Map showing locations of cores examined in this study (1–9) and the distribution of fields productive from the Devonian Dundee Formation, Michigan Basin. Structure contours are on top of the Dundee interval. Details of drill-core localities are described in Table 1. Modified from Montgomery et al. (1998).

1788 Devonian Fractured Hydrothermal Dolomite Reservoirs have been diagenetically modified. Although fracture- in regard to the conditions of fracturing and mineraliza- controlled reservoirs produce more oil than facies- tion (Prouty, 1983; Montgomery et al., 1998). This study controlled reservoirs, little is known about the process presents the first quantitative data on the temperature by which they were formed and diagenetically modified. of dolomitization and the characteristics of fractures in The earliest known reference regarding Devonian Devonian Dundee reservoirs of the Michigan Basin. fractured dolomite reservoirs in the Michigan Basin was a brief article on the Deep River field by Lundy (1969, p. 62). Lundy described the Rogers City–Dundee PURPOSE oil production as being from ‘‘anomalous secondary dolomites believed developed along a fractured and broken Cores of the Dundee Formation from nine localities in zone in the Rogers City limestone.’’ The dolomitized the central Michigan Basin were examined (Figure 1; Deep River field is considered anomalous because it oc- Table 1) to document the distribution and character- curs in the eastern region of the basin where the Dundee istics of fracturing in the Dundee Formation, the style is mainly limestone. The dolomite occurred along a of fracturing present in these reservoir rocks, and their linear N60jW-trending zone approximately 5.5 mi relationship to rock type, grain size, and epigenetic min- (8.8 km) long but less than 0.5 mi (0.8 km) wide. eralization. Fluid-inclusion microthermometric meth- Cuttings were described as fine-grained brown matrix ods were used to document the temperature of saddle dolomite with medium-size white rhombic dolomite dolomite precipitation in an attempt to place constraints crystals. Prouty (1983) also proposed fracturing in the on the conditions of mineralization and reservoir devel- Dundee as a conduit for dolomitizing fluids in several opment in the Devonian Dundee Formation of the cen- central basin structures. Examples of prolific oil fields tral Michigan Basin. are also present in the limestone-dominated part of the eastern Michigan Basin, for example, the South Buckeye field in Gladwin County, where no evidence METHODS of faulting has been identified. Recent unpublished work (e.g., Wood and Harrison, Parts of the slabbed drill core from nine wells were 2002) using production data and structural data derived examined for the distribution of dolomite and the char- from geophysical logs has suggested that some central acteristics of fractures. Alizarin red stain was used to basin Devonian Dundee reservoirs, such as the Vernon distinguish between calcite and dolomite in the cores. field in Isabella County and the Crystal field in Mont- Doubly polished, epoxy-impregnated thin sections calm County, might be analogous to the hydrothermal were prepared for fluid-inclusion work; extreme care dolomite reservoir facies (HTDRF) of the Albion-Scipio was taken to avoid heating the sample at any time be- and Stony Point fields in the Trenton–Black River fore microthermometric measurements were per- formations of southeastern Michigan (D. Barnes, 2002, formed using a method similar to that of Barker and personal communication). Only one small segment Reynolds (1984). No indication of other heating events (3ft;0.9m)oftheoriginaldrill core remains avail- is available that might suggest that the cores were heated able for the Vernon field, but three cores in and around after washing or other testing by the driller. Ultraviolet- the Crystal field were examined as part of this study. cured epoxy was used as the mounting medium to elimi- Whereas drilling records suggest as many as 200 cores nate heating during this part of thin-section prepara- may have been taken from the Dundee Formation tion. Wet sandpaper and diamond-impregnated plastic throughout the basin, only about 50 are currently known discs over plate glass were used under a constant stream to exist (Montgomery et al., 1998; this study). Many of of cool water during the polishing procedure. This meth- these cores are from the South Buckeye field in Gladwin od was used to improve the final polish of the material County, Michigan, in which the Dundee Formation and to eliminate heating of the thin sections on a quickly has not been dolomitized. This leaves only a few tens rotating lapidary wheel. of cores available for study in which the Dundee For- Standard fluid-inclusion measurement techniques mation has been partly or completely dolomitized. of Goldstein and Reynolds (1994) were performed using Previous studies have focused on Trenton–Black FluidInc.-adaptedU.S.GeologicalSurveydesigngas-flow River production in hydrothermal dolomite reservoirs, heating and freezing stages at Western Michigan Uni- such as the Albion-Scipio field. However, studies on the versity and at the University of Wisconsin–Green Bay.

Devonian carbonates in the region have been qualitative Observations of the homogenization temperatures (Th),

Luczaj et al. 1789 Figure 2. (A) Stratigraphic column, Upper Silurian–Devonian section, Michigan Basin. Modi- fied from Montgomery et al. (1998) and Cataco- sinos et al. (2001). (B) Generalized lithology and facies characteristics for Midwest Thelma Rousseau 1-12 core in Mecosta County, Michigan (locality 1). Conventional porosity (ruled) and permeability (solid black) measured from the core at every foot is graphically displayed on the right. This facies and stratigraphic pattern is typical of all the cores examined in this study from the central Michigan Basin.

the final melting temperature of ice (T mice), the FRACTURE CHARACTERISTICS temperature at which hydrohalite breaks down (T hh), and the eutectic temperature (T e) were made. The T h The facies and stratigraphic patterns in the Dundee are and T micevalues for most inclusions were measured similar in cores examined from the central Michigan to the nearest 0.1–0.5jC, when possible. The T h data Basin. Figure 2B shows the generalized lithology and were collected before the sample was subjected to facies characteristics, along with petrophysical data freezing. Vapor bubbles were typically present at room for the Thelma Rousseau 1–12 core in Mecosta Coun- temperatures prior to the initial heating cycle. All tem- ty, Michigan (locality 1). An additional core log and perature data were recorded using cycling techniques core photographs are presented by Montgomery et al. outlined by Goldstein and Reynolds (1994), and freez- (1998) for the Dundee Formation in the Cronus De- ing run data were collected with the presence of the velopment Tow 1–3 HD-1 well (locality 5). vapor phase in each inclusion. Fractures are present in both the Rogers City and Stable isotopic analyses of dolomite were performed Reed City/Dundee parts of the section and are present at Western Michigan University using the carbonate in both limestone and dolostone lithofacies. However, reaction method of Krishnamurthy et al. (1997). only the dolomite lithology contains high-density swarms

1790 Devonian Fractured Hydrothermal Dolomite Reservoirs Figure 2. Continued. of fractures. Macroscopic fractures are almost exclu- interconnected porosity (vuggy and fenestral) have fewer sively present in finer grained mudstones and wacke- macroscopic fractures. Many fractures are isolated, ver- stones. Facies with grainstones, packstones, or abundant tical, and short in length, although long fractures (up to

Table 1. Locations of Drill Cores Used in This Study*

Core Number Core Name Operator County Permit Number Location

1 Thelma Rousseau 1-12 Midwest Mecosta 35426 Section 12, T16N, R8W 2 Stegman and Anderson 3-33 Newstar Isabella 51656 Section 33, T15N, R6W Michigan Consolidated 3 Shuttleworth 1 Gas Company Gratiot 26779 Section 5, T10N, R4W 4 Bessie and Fernon Lee 1 Leonard Oil Montcalm 24011 Section 8, T11N, R5W 5 Tow 1-3 HD-1 Cronus Development Montcalm 50047 Section 3, T10N, R5W 6 McDonald 1-12 Peninsular Oil Newaygo 38437 Section 12, T11N, R11W Michigan Consolidated Michigan Consolidated 7 Gas Company LR 83-2 Gas Company Osceola 29261 Section 5, T17N, R10W Michigan Consolidated 8 Paul Rieman 1 Gas Company Osceola 27191 Section 4, T17N, R9W 9 Hamming 1-22 Dart Missaukee 31448 Section 22, T21N, R7W

*Refer to Figure 1 for map locations.

Luczaj et al. 1791 1 m [3.3 ft]) and swarms of intersecting fractures are southeast and northeast-southwest orientations occur abundant in several wells. in Cambrian through Devonian age rocks throughout Fractures generally range in width from less than eastern Wisconsin and the western upper peninsula re- 1 mm (0.04 in.) to several millimeters across. Most gion of Michigan. Here, fractured, massively dolomi- fractures are either totally filled with cement or have tized carbonate host rocks reveal distinct hydrothermal a thin coating of crystals that holds the fracture to- signatures and are genetically associated with Missis- gether; only a small percentage of fractures are partially sippi Valley–type mineralization and K-silicate miner- cement filled. Cements are mainly saddle dolomite with alization (Luczaj, 2000, 2006). lesser amounts of planar dolomite, calcite, anhydrite, pyrite, and barite. One well in Oceana County contained minor amounts of fluorite, which was also pre- PETROGRAPHY AND FLUID-INCLUSION viously reported as cavity lining cement in wells from MICROTHERMOMETRY the Vernon field in the central Michigan Basin (Fitz- gerald and Thomas, 1932). Petrography Fractures do not appear to be solution enlarged. They have fitted fabrics and angular edges on the brec- Replacive dolomite and euhedral dolomite cements cia fragments, suggesting a mechanical opening pro- are pervasive in nearly all of the Dundee cores ex- cess (Figure 3). Some, such as those in the Tow core amined in the western and central basin. Only a few at 3200 ft (975 m), show classic brittle fracture char- cores examined still contained areas of unaltered lime- acteristics in three dimensions (Figure 3A, B). These stone, which was restricted to mudstone facies and to fractures exhibit patterns similar to those observed in parts of the Bell Shale that overlies the Dundee For- tectonically active settings and likely were an impor- mation (Figure 2A). tant factor in moving the diagenetic fluids (Nelson, The size and texture of dolomite crystals vary, and 2004). Some fractures emanate from or terminate at many crystals have euhedral surfaces, even at the thin- stylolites. Fractures were considered natural, as op- section scale. Fracture and vug-filling saddle dolomite posed to drilling induced, if they were coated by crystals crystals several millimeters in length are abundant in or if the core sample with an open fracture had just most cores and were observed in all cores. Saddle dolo- enough cement present to hold it together. Most mac- mite fills primary fenestral porosity in some cores. Frac- roscopic fractures that appeared to be drilling induced ture and void-filling dolomite formed before calcite and occurred along bedding planes near stylolites or at anhydrite, which are volumetrically insignificant in lithologic contacts. the reservoirs. One sample of late fracture-filling cal- On a regional scale, few outcrops of Michigan Basin cite examined using epifluorenscence microscopy con- rocks exist because they are covered by thick glacial tained petroleum fluid inclusions, indicating that oil sediments. Large-scale fracturing, faulting, and a re- generation and migration most likely began during or gional joint system are along dominantly northwest- after the final stages of dolomitization, but before pre- southeast trends, which can be mapped in outcrop and cipitation of fracture-filling calcite. inferred from subsurface data (e.g., Prouty, 1983). Con- Fluid-inclusion assemblages (FIAs) are defined petro- jugate northeast-southwest–trending structures are also graphically. Primary inclusions are best identified by evident. The locations and geometry of oil- and gas- their occurrence along growth zones in a crystal, and producing fields in the central part of the Michigan these primary inclusions contain a sample of the fluid Basin appear to be related to deep basement structural present during the precipitation of the diagenetic phase trends related to the Precambrian Mid-Continent rift (Goldstein and Reynolds, 1994). system (Wood and Harrison, 2002). In south-central Typical saddle dolomite crystals analyzed contain Michigan, mineralized zones in the Mississippian Bay- a cloudy (inclusion-rich) dolomite core surrounded by a port Limestone may indicate a possible extension of clear, inclusion-poor dolomite overgrowth (Figure 4A). the northwest-trending Albion-Scipio oil field. Quarries All fluid inclusions inside each cloudy, inclusion-rich at Bellevue, Michigan, contain limestone-hosted Mis- dolomite crystal core are treated as a discrete FIA sissippi Valley–type mineralization such as pyrite, mar- because that is the finest petrographically distinguish- casite, and calcite (Blaske, 2003; this study). able assemblage that could be determined (see Gold- Along the western margin of the basin, mineral- stein and Reynolds, 1994). However, some crystals con- ized faults and fractures with the same northwest- tained two well-defined FIAs that were bounded on each

1792 Devonian Fractured Hydrothermal Dolomite Reservoirs Figure 3. Photographs of drill core illustrating fracture and dolomite characteristics in the Dundee Formation. (A) Locality 5 (Tow1-3at3200ft [975.4 m]). Mechanically produced fractures with open pore spaces between angular breccia fragments (view looking upward along core axis). A very thin coating of crystals is present holding the breccia fragments together. (B) Side view of the same core specimen as in (A) showing brittle fracture characteristics in three dimensions. (C) Lo- cality 4 (Lee 1 at 3466.5 ft [1056.6 m]). Abundant brecciation of fine-grained dolostone matrix accom- panied by fracture and vug-filling white saddle dolomite cements. Vugs (V) are partially filled by saddle dolomite. (D) Locality 1 (Thelma Rousseau 1-12 at 3916 ft [1193.6 m]). Brittle fracturing similar to that shown in (A and B), but with coarse white saddle dolomite cements filling fractures. Breccia fragments in the lower half of the specimen have a fitted fabric of fractures in which the fragments can be pieced back together with adjacent clasts. (E) Local- ity 3 (Shuttleworth 1 core segment 1-10-2 at approximately 3272 ft [997 m]). This locality exhibits abundant fracturing with fitted (f) and random (r) breccia fabrics and saddle dolomite cements in fine- grained dolomitized stro- matoporoid (s)-bearing facies. Vugs (V) are partially filled with saddle dolomite.

Luczaj et al. 1793 mite crystals from two separate localities, the Stegman and Anderson 3-33 core and the Thelma Rousseau 1-12 core (Figure 1). Two-phase fluid inclusions in the sad-

dle dolomites from the two drill cores measured had T h values between 76.5 and 180.6jC, with most in the

range of 120–150jC (Figure 5; Table 2). The T h values averaged 131.4jC for 53 two-phase aqueous fluid inclusions from the Stegman and Anderson 3-33 core. In one crystal from the Thelma Rousseau 1-12 core, two clearly distinguishable FIAs were present. The

cloudy inner core of the crystal (FIA 1) had T h values that averaged 124.6jC for 20 inclusions, whereas an

outer growth band of the crystal (FIA 2) had T h values that averaged 145.3jC for 34 inclusions (Figure 5). Possible evidence for necking down of inclusions after a phase change was observed with only two inclusions in this study from the Rousseau core. Fluid inclusions 32 and 41 in FIA 1 are adjacent to one an- other and fall on opposite ends of a histogram plot for FIA 1. Their values were included in the above

T h averages. If necking after a phase change occurred in this case, then the Th for the original combined inclusion would have been somewhere between the two

measured Th values. No correlation between inclusion size and the presence of a vapor bubble was observed. Inclusion 34 from FIA 1 in the Thelma Rousseau well appeared to have leaked 1 day after T measurement. If this in- Figure 4. Transmitted-light images illustrating typical primary h two-phase aqueous fluid inclusions in saddle dolomite from the clusion is not counted (n = 19), the average Th for FIA 1 study area. (A) Dolomite crystal from locality 1 (Thelma Rousseau in the Thelma Rousseau 1-12 sample is only slightly 1-12 at 3941 ft [1201 m]) containing a distinct inclusion-rich core lower at 123.3jC (Table 2; Figure 5). A second in- and a distinct band of inclusions in the outer clear overgrowth. clusion leaked in this FIA before a Th measurement (B) Enlargement of the upper-right part of the same dolomite could be obtained. crystal shown above. Each inclusion-rich part of the crystal was treated as a separate fluid-inclusion assemblage (FIA). Freezing Data

Observations of the final melting temperature of ice side by clear rim overgrowths (Figure 4B). Fluid inclu- (T mice), the temperature at which hydrohalite breaks sions observed within these isolated primary growth down (Thh), and the eutectic temperature (T e) were zones in the dolomite were treated as different FIAs. made on inclusions from both localities (Figure 6; Most of the fluid inclusions in the cloudy crystal cores Table 3). Fluid inclusions from both localities exhib- and overgrowths are irregularly shaped, range in size ited similar behavior at low temperatures. Halite was from about 1 to more than 20 mm, and have consistently not observed in any inclusions from either locality. low vapor/liquid ratios. For seven inclusions in dolomite from locality 1 (Thelma Rousseau 1-12), ice was the last phase ob-

Homogenization Temperature Data served to break down in all inclusions. The T mice values ranged from À25.3 to À37.5jC and averaged

Homogenization temperatures (T h)fromprimary À34.4jC (Table 3). With the exception of one inclu- fluid inclusions provide an understanding of the pre- sion in the outer growth zone FIA, all measured in- cipitation temperatures of the saddle dolomites. Pri- clusions had similar T micevalues in both inclusions mary fluid inclusions were measured in saddle dolo- (Table 3). Eutectic melting (T e) was observed between

1794 Devonian Fractured Hydrothermal Dolomite Reservoirs Figure 5. Two frequency histograms of T h data from primary aqueous inclusions in single crystals of saddle dolomite from localities 1 and 2. The T h data are similar for both localities and suggest precipitation of dolomite over a moderate range of temperatures. This is well illus- tratedbythedatafromlocality1,in which different crystal growth zones yield T h values indicative of warming during precipitation of saddle dolomite. Data from two inclusions in the interior of the crystal at locality 1 were not used because of leakage.

À57 and À64jC, with melting occurring in most in- STABLE ISOTOPE DATA clusions by À61jC. For five inclusions in dolomite from locality 2 The d18O values of carbonates can be helpful in de- (Stegman and Anderson 3-33), ice was the last phase termining whether the minerals formed at elevated tem- observed to break down in all but one of the inclusions. peratures in burial or hydrothermal settings. Elevated

The Tmicevalues ranged from À27.7 to À34.0jC and temperature drives the isotopic composition of diage- averaged À30.8jC (Table 3). For inclusion 8, a solid netic carbonate to negative values (Hardie, 1987; Allan phase was observed to break down at +1.5jC after all and Wiggins, 1993). Although stable isotopic data have ice melted at À27.7jC. Eutectic melting (Te)inin- been reported for Ordovician dolomites of the Michigan clusions from locality 2 was observed between À57 and Basin (Taylor and Sibley, 1986; Budai and Wilson, 1991; À66jC, with melting occurring in most inclusions by Allan and Wiggins, 1993), data from Devonian carbon- À60jC. ates in the basin are limited.

Luczaj et al. 1795 Table 2. Homogenization Temperature Data for Rousseau 1-12 and Stegman and Anderson 3-33 Saddle Dolomite

Thelma Rousseau 1-12 at 3941 ft (1201 m) Stegman and Anderson 3-33 at 3677.7 ft (1120.9 m)

Fluid Inclusion FIA** T h Fluid Inclusion FIA** T h 1 2 180.6 1 1 160.8 2a 2 150.2 2 1 154.9 2b 2 136.0 3 1 167.4 3 2 149.5 4a 1 111.4 4 2 149.9 4b 1 119.5 5 2 152.1 5 1 131.6 6 2 137.0 6 1 143.6 7 2 150.4 7 1 125.0 8a 2 126.0 8 1 122.8 8b 2 154.9 9a 1 137.5 9 2 150.5 9b 1 149.0 10 2 151.5 10 1 132.1 11a 2 162.5 11 1 136.0 11b 2 124.9 12a 1 76.5 12 2 137.5 12b 1 132.1 13 2 144.5 13 1 139.9 14 2 135.5 14 1 139.9 15 2 153.6 15 1 159.9 16 2 162.5 16 1 134.9 17 2 166.5 17 1 126.0 18 2 137.0 18 1 139.9 19 2 157.0 19 1 174.9 20 2 160.5 20 1 134.9 21 2 154.9 21 1 124.0 22 2 149.9 22 1 144.9 23 2 130.8 23 1 141.9 24 2 152.1 24 1 109.1 25 2 143.0 25 1 126.0 26 2 126.3 26 1 121.5 27 2 123.7 27 1 121.5 28a 1 126.0 28 1 157.1 28b 1 127.0 29 1 128.0 29 1 134.0 30 1 134.9 30 1 119.4 31 1 124.9 31 1 121.9 32 1 126.0 32 1 143.1 33 1 123.2 33 1 119.3 34 1 121.0 34* 1 150.2 35 1 110.8 35 1 136.7 36 1 121.1 36 1 119.2 37 1 123.5 37 1 117.6 38 1 122.3 38 1 114.4 39 1 142.1 39 1 128.7 40 1 126.1 40 1 126.8 41 1 112.5 41 1 109.2 42 1 124.9 42 1 107.5 43 1 124.0

1796 Devonian Fractured Hydrothermal Dolomite Reservoirs Table 2. Continued

Thelma Rousseau 1-12 at 3941 ft (1201 m) Stegman and Anderson 3-33 at 3677.7 ft (1120.9 m)

Fluid Inclusion FIA** T h Fluid Inclusion FIA** T h 43 2 146.2 44 1 139.9 44 1 117.0 45 1 129.9 45 2 132.0 46 1 124.9 46 2 132.0 47 1 124.1 47 2 119.9 48 1 128.6 48 1 118.6 49 1 120.8 49 1 135.4 50 1 135.1 50 1 120.7

FIA 1 (n = 19) Average T h 123.3 FIA 1 (n = 53) Average T h 131.4 FIA 2 (n = 34) Average T h 145.3

*Inclusion 34 leaked after heating and was not used to calculate the average T h. **FIA 1 corresponds to the crystal interior, whereas FIA 2 corresponds to an isolated growth zone in the clear outer rim of the saddle dolomite crystal.

Stable isotopic (d18O) compositions of saddle do- ples of white dolomite fragments from cuttings of sev- lomite were determined for samples from the Devonian eral wells in the Deep River field in Arenac County, Dundee Formation in three different oil fields in central Michigan. The d18O data are consistent with precipita- Michigan. Two samples of saddle dolomite cements tion of saddle dolomite at elevated temperatures (Allan were analyzed from drill core from the Tow 1–3 well and Wiggins, 1993). in the Crystal field of Montcalm County, Michigan (locality 5). The d18O (PDB) composition of saddle dolomites from depths of 3196 and 3231 ft (974 and DISCUSSION 985 m) were reported as À9.03 and À9.19x, respectively. Two saddle dolomite crystals from the Michigan The common occurrence of saddle dolomite along frac- Consolidated Gas Company LR83-2 well in Osceola tures suggests that the fracturing predated or was con- County (locality 7) from depths of 3577 and 3586 ft temporaneous with the precipitation of saddle do- (1091 and 1093 m) yielded d18O values of À8.18 and lomite. Fractures observed at the core scale may be À8.64x, respectively. Additional d18O data ranging related to reservoir-scale features because they com- between À6.68 and À8.93xwere reported for sam- monly are in other fractured reservoirs (Nelson, 2004). Core-scale and reservoir-scale fracturing and faulting in the Dundee Formation are analogous to well- documented dolomitization related to deep basement features in the Ordovician Albion-Scipio trend of the south-central Michigan Basin (Hurley and Budros, 1990). The presence of two-phase fluid inclusions that homogenize in the range of 120 to at least 155jCin- dicates that rocks were exposed to temperatures of that magnitude (Barker and Goldstein, 1990). A few homogenization temperatures approaching 175jC for dolomite suggest that rocks were exposed to tempera-

tures above 155jC. However, because Th values above 155jC are only a small proportion of the data sets, un- Figure 6. Frequency histogram of T micedata from primary aqueous inclusions in saddle dolomite from localities 1 and 2. recognized necking down after a phase change or het- T micedata are similar for both localities and suggest precip- erogeneous entrapment of a relatively small fraction of itation of dolomite in the presence of a dense brine. the vapor phase cannot be ruled out.

Luczaj et al. 1797 Table 3. Fluid-Inclusion Freezing Data for Rousseau 1-12 and Stegman and Anderson 3-33 Saddle Dolomite

Freezing Data

Salinity (wt.% Eutectic

Locality Inclusion Number* FIA Type T mice(jC) NaCl Equivalent) Temperature (jC) Thelma Rousseau 1-12 11a Outer growth zone À34.3 31.5 À62 12 Outer growth zone À36.2 32.9 À65 to À61 16 Outer growth zone À33.3 30.8 – 25 Outer growth zone À37.3 33.8 À60 to À57 26 Outer growth zone À25.3 25.8 À64 28a Crystal interior À37.5 34.0 À64 to À57 29 Crystal interior À37.0 33.6 about À66 to À65 Stegman and Anderson 3-33 1 Crystal interior À29.5 28.3 À66 to À63 3 Crystal interior À29.3 28.2 À63 to À57 8 Crystal interior À27.7 27.2 À65 to À57 17 Crystal interior À33.7 31.1 – 51 Crystal interior À34.0 31.3 about À63

*T h data were not collected for inclusion 51.

Inclusions in the FIAs measured have most of their mite likely precipitated over a range of temperatures

Th values spread over an interval of 25–40jC(Figure5). as temperatures rose to at least 145–150jC. Precipi- This suggests that there was either reequilibration of tation of dolomite over a range of temperatures is existing fluid inclusions or original variability in con- consistent with the idea that dolomite precipitation ditions of entrapment of the fluid inclusions (Goldstein is favored as temperature increases (Carpenter, 1980; and Reynolds, 1994). The Th values are considered min- Land,1985).Therefore,itwouldmakesensethatas imum estimates of the temperature of entrapment, and the temperatures rose when the Dundee rocks were no pressure corrections have been applied because the heated during and shortly after regional fracturing burial pressures at the time of entrapment are un- events, both replacement of preexisting carbonate min- known. Any pressure correction applied to the data erals by dolomite and dolomite precipitation would be would be added to the observed T h values (Goldstein favored. and Reynolds, 1994). Bulk salinity (weight percent, NaCl equivalent) Entrapment of fluid inclusions over a small range was calculated using the equation of Bodnar (1992). For in temperature (10–30jC) is the favored mechanism inclusions in dolomite from locality 1 (Thelma Rous- to explain the range of homogenization temperatures seau 1-12), calculated salinities ranged from 25.8 to measured in the saddle dolomites. No obvious corre- 34.0 wt.% (NaCl equivalent) and averaged 31.6 wt.% lation was observed between the size of an inclusion (NaCl equivalent) for all seven inclusions (Table 3). The and its homogenization temperature, which would be observed eutectics are consistent with a complex sys- expected if reequilibration of existing inclusions caused tem similar to a model NaCl-CaCl2-MgCl2-H2O sys- by stretching had occurred (Goldstein and Reynolds, tem. The model NaCl-CaCl2-MgCl2-H2O system has a 1994). stable T e of À57jC, and the initial melting in the in- Evidenceforentrapmentofinclusionsoverarange clusions that occurs several degrees below the stable of temperatures is especially obvious when separate FIAs eutectic was likely the result of metastable behavior within the same crystal are considered (see Figure 5). In (Goldstein and Reynolds, 1994). the case of the Thelma Rousseau core (locality 1), the For inclusions in dolomite from locality 2 (Steg- outer clear dolomite overgrowth contains inclusions man and Anderson 3-33), calculated salinities range with higher Th values than the cloudy, inclusion-rich from 27.2 to 31.3 wt.% (NaCl equivalent) and averaged core of the crystal. In this crystal, the inclusions can be 29.2 wt.% (NaCl equivalent) for all five inclusions. The segregated petrographically into different FIAs with dif- observed eutectics for inclusions from locality 2 are ferent T h ranges. This also suggests that saddle dolo- also consistent with a complex system similar to the

1798 Devonian Fractured Hydrothermal Dolomite Reservoirs model NaCl-CaCl2-MgCl2-H2O system interpreted lead to considerably lower thicknesses of eroded sedi- for locality 1. ments (see Luczaj, 2000, for details).

Together, the Th, T mice, and T hh data indicate that The term ‘‘hydrothermal dolomite’’ is applied to the fluid present during precipitation of the saddle do- these fractured dolomite reservoirs because of the rela- lomite in the hydrocarbon reservoirs was a very dense tively high temperature of saddle dolomite precipita- Na-Ca-Mg-Cl brine at temperatures as high as 120– tion relative to what is expected from heating by burial 150jC. This is similar to the bulk composition of mod- alone. Assuming a 20jC mean annual surface temper- ern Dundee Formation reservoir fluids in the Michigan ature and a 20–25jC/km geothermal gradient, a mini- Basin (White et al., 1963; Dollar et al., 1991). mum of 3000 m (9800 ft) of sediments would need to In the study area, the Dundee Formation lies 3200– have been deposited and then eroded between the late 4000 ft (975–1200 m) below the surface. The ther- Pennsylvanian and the Late Jurassic to satisfy the tem- mal evolution of the basin is controversial regarding peratures measured in this study, if burial heating alone the thickness of missing strata, timing of thermal matu- were the mechanism responsible. These burial con- ration, and magnitudes of past geothermal gradients straints suggesting 3 km (1.8 mi) or more of missing (e.g., Cercone, 1984; Nunn et al., 1984; Illich and Grizzle, sediments are inconsistent with the known stratigraph- 1985; Velbel and Brandt, 1989; Howell and van der ic history for the Michigan Basin. Therefore, the tem- Pluijm, 1990; Cercone and Pollack, 1991; Crowley, 1991). peratures measured are not representative of heating Results vary widely, but some authors have concluded because of burial alone, but instead are caused by in- that less than about 1200 m (4000 ft) of sediments creased local or regional geothermal gradient related were eroded from the basin before the Jurassic (e.g., to fluid flow. The precipitation of saddle dolomite at Hayba, 2004). temperatures above ambient temperature is consistent We assume that the thickness of the eroded sedi- with moving hot brines from greater depths upward ments across the Michigan Basin was probably less than into the Dundee Formation along faults and fractures. approximately 1000 m (3300 ft) based on the following: Strikingly similar fracture characteristics, mineral- (1) the Pennsylvanian and Devonian sediments in the ogic associations, dolomitization fabric, dolomite for- basin show relatively low thermal maturities (Sleep mation temperatures, and stable isotopic values exist et al., 1980; Landing and Wardlaw, 1981; Moyer, 1982; throughout the Dundee of the central Michigan Basin. Dellapenna, 1991; Martini et al., 1998); (2) the pre- Therefore, some or most of the dolomitized Dundee dicted thickness of the eroded late Paleozoic strati- fields of the central Michigan Basin have the same ori- graphic section in the basin is about 300 m (1000 ft) gin and are genetically related to deep-seated fault and (Beaumont et al., 1987); and (3) relatively noncom- fracture systems. pacted Mississippian shales suggest a maximum additional thickness of 850 m (2800 ft) in the center of the basin (Vugrinovich, 1988). SUMMARY Late Jurassic sediments that overlie Pennsylvanian sediments in the basin also place an important con- Fractured dolomite reservoirs appear to account for a straint on the time available for sediment accumula- substantial part of the oil production from Devonian tion and complete erosion (less than approximately rocks in the Michigan Basin. The formation of fractures 150 m.y.). Deposition and erosion rate calculations and the precipitation of saddle dolomite in the Devo- for this missing interval predict a maximum thick- nian Dundee Formation in the central Michigan Basin ness of eroded sediments of about 1125 m (3690 ft). were integral parts of reservoir development because This estimate was calculated using a period of 75 m.y. they both apparently predate oil migration and reser- of deposition immediately followed by a period of voir filling. 75 m.y. of erosion, with deposition and erosion rates The saddle dolomite was formed during hydro- of 15 m/m.y. (50 ft/m.y.), which is close to the aver- thermal fluid circulation of dense Na-Ca-Mg-Cl brines age net sediment accumulation rate for the whole ba- along faults and fractures over a range of temperatures sin during the early and middle Paleozoic. During the as high as at least 135–145jC. Although it appears that Mississippian through the early Mesozoic, net rates of a wide area of the basin was affected by hot dolomi- deposition were probably much lower, as suggested tizing fluids, heating and water-rock interaction were by significant unconformities present within the Mis- likely focused along localized fractured and faulted zones sissippian and Pennsylvanian sections, which would with which the reservoirs are genetically associated.

Luczaj et al. 1799 Some or most of the dolomitized Dundee fields of the Dellapenna, T. M., 1991, Sedimentological, structural, and organic central Michigan Basin likely have the same hydrother- geochemical controls on natural gas occurrence in the Antrim Formation in Otsego County, Michigan: Master’s thesis, West- mal signature, here interpreted to be related to deep- ern Michigan University, Kalamazoo, Michigan, 147 p. seated fault and fracture systems. Dollar, P. S., S. K. Frape, and R. H. McNutt, 1991, Geochemistry Exploration and production models for dolomi- of formation waters, southwestern Ontario, Canada and south- ern Michigan, U.S.A.: Implications for origin and evolution: tized reservoirs in the Devonian Dundee Formation, as Ontario Geoscience Research Grant Program, Grant No. 249; well as other carbonate units in the basin, need to in- Ontario Geological Survey Open-File Report 5743, 72 p. corporate the concepts of faulted, fractured hydro- Fitzgerald, P. E., and W. A. Thomas, 1932, Occurrence of fluorite in Monroe Formation of Vernon Township pool near Mount thermal dolomite reservoir facies models to achieve a Pleasant, Michigan: AAPG Bulletin, v. 16, no. 1, p. 91–92. complete understanding of reservoir properties and Gardner, W. C., 1974, Middle Devonian stratigraphy and deposi- production potential. tional environments in the Michigan Basin: Michigan Basin Geological Society Special Paper 1, 138 p. Goldstein, R. H., and T. J. Reynolds, 1994, Systematics of fluid inclusions in diagenetic minerals: Society for Sedimentary Ge- REFERENCES CITED ology (SEPM) Short Course 31, 199 p. Hardie, L. A., 1987, Dolomitization: A critical view of some cur- rent views: Journal of Sedimentary Petrology, v. 57, p. 166– Allan, J. R., and W. D. Wiggins, 1993, Dolomite reservoirs, geo- 183. chemical techniques for evaluating origin and distribution: Hayba, D. O., 2004, Stratigraphic and thermal modeling of the Michi- AAPG Continuing Education Course Note Series 36, 129 p. gan Basin, in Petroleum systems of the Michigan Basin— A look Barker, C. E., and R. H. Goldstein, 1990, Fluid inclusion technique at remaining and undiscovered oil and gas resources: Petroleum for determining the maximum temperature and its compari- Technology Transfer Council Workshop, September 23, 2004, son to the vitrinite reflectance geothermometer: Geology, v. 18, p. 243–264. p. 1003–1006. Howell, P. D., and B. A. van der Pluijm, 1990, Early history of the Barker, C. E., and T. J. Reynolds, 1984, Preparing doubly polished Michigan Basin: Subsidence and Appalachian tectonics: Geol- sections of temperature sensitive sedimentary rocks: Journal of ogy, v. 18, p. 1195–1198. Sedimentary Petrology, v. 54, p. 635–636. Hurley, N. F., and R. Budros, 1990, Albion-Scipio and Stoney Point Beaumont, C., G. Quinlan, and J. Hamilton, 1987, The Alleghanian fields–U.S.A., in E. Beaumont and N. Foster, eds., Strati- orogeny and its relationship to the evolution of the Eastern graphic traps I: AAPG Treatise of Petroleum Geology, Atlas of Interior, North America, in C. Beaumont and A. Tankard, eds., Oil and Gas Fields, p. 1–37. Sedimentary basins and basin-forming mechanisms: Canadian Illich, H. A., and P. L. Grizzle, 1985, Thermal subsidence and gen- Society of Petroleum Geologists Memoir 12, p. 425–445. eration of hydrocarbons in Michigan Basin: Discussion: AAPG Blaske, A. R., 2003, Geology of the Mississippi Valley–type min- Bulletin, v. 69, p. 1401–1403. eralization at Bellevue, Michigan: Institute on Lake Superior Krishnamurthy, R. V., E. A. Atekwana, and H. Guha, 1997, A simple, Geology, Proceedings and Abstracts, v. 49, pt. 1, p. 5–6. inexpensive carbonate-phosphoric acid reaction method for Bodnar, R. J., 1992, Revised equation and table for determining the the analysis of carbon and oxygen isotopes of carbonates: An-

freezing point depression of H2O-NaCl solutions: Geochimica alytical Chemistry, v. 69, p. 4256–4258. et Cosmochimica Acta, v. 57, p. 683–684. Land, L. S., 1985, The origin of massive dolomite. Journal of Geo- Budai, J. M., and J. L. Wilson, 1991, Diagenetic history of the logical Education, v. 33, p. 112–125. Trenton and Black River formations in the Michigan Basin, Landing, E., and B. R. Wardlaw, 1981, Atokan conodonts from the in P. Catacosinos and P. Daniels, eds., Early sedimentary evo- Pennsylvanian outlier of the Michigan Basin: Journal of Pale- lution of the Michigan Basin: Geological Society of America ontology, v. 55, p. 1251–1269. Special Paper 256, p. 73 – 87. Luczaj, J. A., 2000, Epigenetic dolomitization and sulfide miner- Carpenter, A. B., 1980, The chemistry of dolomite formation: I. alization in Paleozoic rocks of eastern Wisconsin: Implications The stability of dolomite, in D.Zenger,J.Dunham,andR.Ething- for fluid flow out of the Michigan Basin, U.S.A.: Ph.D. dis- ton, eds., Concepts and models of dolomitization: SEPM Spe- sertation, Johns Hopkins University, Baltimore, Maryland, cial Publication 28, p. 111 – 121. 443 p. Catacosinos, P. A., W. B. Harrison, R. F. Reynolds, D. B. Westjohn, Luczaj, J. A., 2006, Evidence against the Dorag (mixing-zone) and M. S. Wollensak, 2001, Stratigraphic lexicon for Michigan: model for dolomitization along the Wisconsin arch— A case Michigan Department of Environmental Quality, Geological for hydrothermal diagenesis: AAPG Bulletin, v. 90, p. 1719– Survey Division, Bulletin, v. 8, 56 p. 1738. Cercone, K. R., 1984, Thermal history of the Michigan Basin: AAPG Lundy, C. L., 1969, Deep River field, in M. Wollensak, ed., Sym- Bulletin, v. 68, p. 130–136. posium on Michigan Oil and Gas Fields: Michigan Basin Geo- Cercone, K. R., and H. N. Pollack, 1991, Thermal maturity of the logical Society, p. 61–68. Michigan Basin, in P. Catacosinos and P. Daniels, eds., Early Martini, A. M., L. M. Walter, J. M. Budai, T. C. W. Ku, C. J. Kaiser, sedimentary evolution of the Michigan Basin: Geological So- and M. Schoell, 1998, Genetic and temporal relations between ciety of America Special Paper 256, p. 1–11. formation waters and biogenic methane: Upper Devonian An- Crowley, K. D., 1991, Thermal history of Michigan Basin and south- trim Shale, Michigan Basin, U.S.A.: Geochimica et Cosmochi- ern Canadian shield from apatite fission track analysis: Journal of mica Acta, v. 62, p. 1699–1720. Geophysical Research, v. 96, p. 697–711. Montgomery, S. L., J. R. Wood, and W. B. Harrison, III, 1998, Curran, B. C., and N. F. Hurley, 1992, Geology of the Devonian Devonian Dundee Formation, Crystal field, Michigan Basin: Dundee reservoir, West Branch field, Michigan: AAPG Bul- Recovery of bypassed oil through horizontal drilling: AAPG letin, v. 76, p. 1363–1383. Bulletin, v. 82, p. 1445–1462.

1800 Devonian Fractured Hydrothermal Dolomite Reservoirs Moyer, R. B., 1982, Thermal maturity and organic content of se- Velbel, M. A., and D. S. Brandt, 1989, Sedimentology and paleo- lected Paleozoic formations— Michigan Basin: Master’s thesis, geography of the Pennsylvanian strata of Grand Ledge, Michi- Michigan State University, East Lansing, 62 p. gan: Michigan Basin Geological Society Spring Field Trip Guide- Nelson, R. A., 2004, Exploration and evaluation of fractured res- book, 33 p. ervoirs: Petroleum Technology Transfer Council Workshop, Vugrinovich, R., 1988, Shale compaction in the Michigan Basin: AAPG Eastern Section Meeting, October 6, 2004, 78 p. Estimates of former depth of burial and implications for paleo- Nunn, J. A., N. H. Sleep, and W. E. Moore, 1984, Thermal sub- geothermal gradients: Bulletin of Canadian Petroleum Geology, sidence and generation of hydrocarbons in Michigan Basin: v. 36, p. 1–8. Reply: AAPG Bulletin, v. 69, p. 1404. White, D. E., J. D. Hem, and G. A. Waring, 1963, Chemical com- Prouty, C. E., 1983, The tectonic development of the Michigan Basin position of subsurface waters, in M. Fleischer, ed., Data of intrastructures, in R. Kimmel, ed., Tectonics, structure, and geochemistry, 6th ed.: U.S. Geological Survey Professional karst in northern lower Michigan: Michigan Basin Geological Paper 440-F, 67 p. Society 1983 Field Conference, Lansing, Michigan, p. 36–81. Wood, J. R., and W. B. Harrison, 2002, Advanced characterization Sleep, N. H., J. A. Nunn, and L. Chou, 1980, Platform basins: An- of fractured reservoirs in carbonate rocks: The Michigan Basin: nual Reviews of Earth and Planetary Sciences, v. 8, p. 17–34. Semiannual Department of Energy Report for Grant Award Taylor, T. R., and D. F. Sibley, 1986, Petrographic and geochemical Number DE-AC26-98BC15100, 32 p. characteristics of dolomite types and the origin of ferroan Wylie, A. S., and J. R. Wood, 2005, Historical production trends dolomite in the Trenton Formation, Ordovician, Michigan Basin, suggest remaining upside for E&D in Michigan: Oil & Gas U.S.A.: Sedimentology, v. 33, p. 61–86. Journal, v. 103, June 20, 2005, no. 23, p. 38–46.

Luczaj et al. 1801