Magoon, LB, and W. G, Dow, eds, 1994, The petroleum system—from source to trap: AAPG Memoit 60. Chapter 17 5 —== = - ™ Maturity Modeling: Thermal Indicators, Hydrocarbon Generation, and Oil Cracking Douglas W. Waples Consultant Dever, Colorato, USA ‘Abstract ‘Thermal maturity modeling is now a highly developed discipline, with a number of excellent software programs available for working geologists. Maturity models are used to describe the behavior of individual thermal indicators (such as various biomarkers and vitrinite reflectance), as well as 3rbon generation and , Maturity modeling must begin with a valid concep- tual model of local geology, and should then follow a logical sequence. First, present-day heat flow and thermal conductivities are selected to predict: the observe presently temperatares, Next, paleoheat flows and the magnitudes of depositional and erosional events are chosen to agree with the regional geology and of basin evolution and to predict correctly the observed values for any available thermal indicators. Finally, modeling of hydrocarbon generation within a source rock and cracking of oil to gas in source or reservoir rocks is carried out once all input data are internally consistent (optimized). Kinetic models for organic maturation are clearly superior to the time-temperature index (TTD approach and should therefore be used in future applications. Some kinetic models now predict hydrocarbon expulsion as well as tion and thus may provide more accurate estimates of hy composition than generation models alone. Faulting and other tectonic complexities, remain problems. One-dimensional maturity modeling (z spatial dimension plus time) seems adequate for most exploration applications. INTRODUCTION - ‘Thermal maturity modeling has become a standard technology in petroleum exploration. When maturity modeling was first developed, the distinction between hydrocarbon generation and maturation of thermal indi- cators (such as vitrinite reflectance) was not consistently made. Now, however, geochemists routinely distinguish between maturation of thermal indicators and hydro- carbon generation and cracking (Table 17.1). Maturation of thermal indicators refers to thermally induced changes ina variety of indicators that record the thermal histories of rocks in different ways Hydrocarbon generation and cracking, in contrast, refer to two different processes: the formation and the destruction of oil and gas. Maturation of thermal indicators, hydrocarbon generation, and hydrocarbon cracking are collectively referred to as ‘maturation and are now modeled under the umbrella term maturity modeling. It is important to be able to model all three types of phenomena, but equally important to keep them conceptually separate. ‘These three types of maturity modeling are important 285 to investigations of sedimentary basins and their petroleum systems. Thermally induced changes to organic matter and other thermal indicators give important clues about the rate and extent of subsurface heating, This heat affects numerous rock properties, such as induration, porosity, and clay mineralogy. Thus, maturity modeling is important to sedimentary basin analysis. However, the moment hydrocarbons are expelled from a mature source rock, that source rock, the expelled hydrocarbons, the migration conduit, and the traps form a petroleum system. Here, the second type of maturity modeling is critical for understanding the origin and evolution of a particular petroleum system. Finally, cracking of oil to gas can take place as the petroleum system evolves, either during late generation. in the source rock or in the reservoir (Blanc and Connan, Chapter 14, this volume). In either case, maturity modeling is useful for predicting the physical nature of the hydrocarbon fluids. Neither maturity modeling (usually carried out in the + spatial dimension through time) nor its companion, basin modeling (carried out in either two or three spatial286 Waples dimensions, plus time [2-D = x, z, 4D =x, y, 2), is truly a process model (see Tetzlaff and Harbaugh, 1989) since phenomena are not modeled molecule-by- molecule, drop-by-drop, or grain-by-grain. Application of process modeling to exploration is inhibited by computer limitations, lack of cata control, and imperfect knowledge of many of the processes that would have to be included. Models are therefore at best crude approxi- mations of the general characteristics of reality, simu- lating the approximate sequence of events in a semi- quantitative way using equations that range from highly precise to highly empirical We should thus recognize from the outset that our models will always contain intrinsic uncertainties and crrors that limit the accuracy of their predictions. These fundamental uncertainties are further compounded by our imperfect knowlexige of the history and nature of the rocks and hydrocarbon fluids being modeled. Successful application of maturity models depends largely on our ability to deal with these uncertainties. The reader who is unfamiliar with maturity modeling may find it beneficial to consult Waples (1980; 1985, chap. 9) before reading this chapter. MATURITY MODELING History and Present Status The history of maturity modeling has been discussed in detail elsewhere (Waples, 1984; Ungerer, 1990) and is, only summarized here. Although various maturity models had been in use in a few major oil companies since the early to mid-1970s, maturity modeling first became widely popular following Waples’ (1980) appli- cation of Lopatin’s (1971) time-temperature index (TTD to petroleum generation. Many early applications of the ‘TH method were carried out by hand, but by the mid- 1980s, commercial software became available to ease the tedium of hand calculations and to permit the use of kinetic models instead of (or in addition to) the TTT method. A large number of software programs for carrying out maturity modeling are now available to the public. A number of computer codes have also been published for independent workers (Elphick, 1989; Elphick and Reed, 1989; Sweeney, 1989, 1990; Wood, 1990), but they lack the convenient user interfaces of commercial software that greatly facilitate the modeling process Objectives All maturity modeling should begin with a clear conceptual model of geologic history because modeling based on inadequate geologic concepts is likely to show serious weaknesses under close scrutiny. Regardless of the amount of attention one gives to construction of the initial geologic model, however, it will probably contain errors and incorrect assumptions. One important objective of maturity modeling, therefore, is to test the original geologic model and suggest improvements. For ‘example, our present conceptual models are inadequate to predict from first principles such important input parameters as heat flow histories and thermal conductiv- ities of rocks (Waples et al, 1992a,b). The second major objective is to obtain information about the status and history of hydrocarbon generation and thus help to deterthine whether a petroleum system exists. In the early days of maturity modeling, these two objectives were not separated because modeling of thermal indicators and modeling of hydrocarbon genera- tion were usually treated as the same problem. However, as the distinction between maturation of thermal indica- tors and hydrocarbon generation has become clear, the differences between these two objectives have also crys- tallized. Thermal indicators are used to check the correct- ness of the geologic model by comparing the modeled behavior of thermal indicators with measured values. Only after this first objective is successfully completed should an attempt be made to model hydrocarbon gener- ation. Any modern maturity modeling study must fulfill bboth objectives, that is, it must model thermal indicators for basin analysis as well as hydrocarbon generation for system analysis. ‘Types of Thermal Maturity Modeling ‘Time-Temperature Index (TT!) Modeling ‘TTI modeling (Lopatin, 1971; Waples, 1980) assumes, that (1) only time and temperature are important factors in_maturation and (2) time and temperature can be substituted for each other (eg, the lower the tempera~ ture, the longer the time required to reach a given level of thermal maturity). Pressure effects are neglected. The ‘way in which time and temperature are interconverted depends on a gross simplification of basic principles of chemical kinetics. TTI values are calculated using equations such as those of Royden etal. (1980) or Waples (1980), ‘The TTI method was originally designed to predict coal rank through its calibration to vitrinite reflectance (Ro), and thus it directly models only the change in a single thermal indicator. A number of different TTI-Rp calibrations have been developed, both ignoring compaction effects (¢.g,, Lopatin, 1971; Waples, 1980; Goff, 1983) and including them (Dykstra, 1987). Some ‘workers have attempted to overcome inherent weak- nesses in the TT] method by providing sedimentary basin-by-basin calibration (eg, Isler, 1984) Because vitrinite reflectance for many years was directly correlated with hydrocarbon generation, the TT method has also been used to predict hydrocarbon generation and cracking (e.g., Waples, 1980, 1988), although it was inappropriate for that purpose. Although the TTT method does not specify kerogen type, a crude dependence on kerogen type can be included by shifting the vitrinite reflectance value at which hydro- carbon generation begins for different types of kerogens (Waples, 1985).17, Maturity Modeling 287 Table 17.1. Division of Thermal Maturity Modeling into Thermal Indicators and Hydrocarbons Maturation of Thermal indicators Hydrocarbons TimeTemperature Max. Temperature Integrated Time-Temperature Generation Cracking Fission track Fhid inclusions Ry on Oitto gas annealing Trax Gas 2OS(20R + 209) steranes 228(22R + 228) hopanes Sterane aromatization Methiphenanthrene index (MPI) Biphenyl polen translucency Fraction of aromatic protons (FAP or PAP)® ‘C-lactor (IR spectrum) Wire (UR spectrum) Fraction of aromatic protons (NMR)? Net cussed wet 86 Fra (nes for ea Kinetic Modeling Kinetic modeling is used both to predict hydrocarbon. «generation and oil cracking and to model the behavior of some thermal indicators. The theoretical foundation of kinetic modeling is much more solid than that of TTI modeling (Tissot et al., 1987), but is still imperfect. Kinetic models assume that a given process (such as hydrocarbon generation or vitrinite reflectance change) consists of one or several parallel chemical reactions, Kinetic parameters for each individual reaction are derived from laboratory experiments, empirical data from wells, or both. The maturation of different types of organic matter (including different types of kerogen) can ‘be modeled using different kinetic parameters. A number of weaknesses exist in kinetic models, however. Fist, the laboratory measurements from which Kinetic parameters are derived always contain some uncertainties, such as temperature control (Espitalié et al,, 1993). Second, most kinetic models are calibrated using data obtained from high-temperature laboratory experiments, but itis far from certain that the processes occurring under laboratory conditions are good analogs for those occurring in nature. Finally, even if laboratory experiments were perfect analogs for natural reactions, there would be statistical errors in extrapolating from laboratory conditions to natural situations. Arrhenius TTI Modeling, ‘Wood (1988) and Hunt et al. (1991) have shown that a simplified Arrhenius (kinetic) model can be used to calculate what they call TTTge values. Hunt ef a. (1991) have developed nomographs for calculating TTl arg values for a variety of different type II kerogens. This, method has two advantages over standard kinetic ‘modeling: (1) the use of nomographs eliminates the need for computers in kinetic modeling, and (2) if one does vwish to use a computer, the computations are faster than standard kinetic calculations. According to these workers, loss of accuracy due to these simplifications is very small. Although these models contain the phrase “TTY” in their names, they have more in common with kinetic models than withthe traditional TTT method. Modeling of Thermal Maturity Indicators Numerous thermal indicators exist today, and others are being developed almost continually. Curiale et al (1989) have recently reviewed many of them, including some whose behavior can be modeled (Table 17.1) All commercially available software programs include kinetic or TTT models for one or more thermal indicators. By far the most widely modeled thermal indicator has been vitrinite reflectance (Re). A number of molecular transformations, including biomarkers, have become fairly popular. Apatite fission track annealing (sometimes called AFTAR) is also in common use, Some progress has been made in modeling Tmax data obtained from Rock-Eval pyrolysis (eg, Peters, 1986). In rare cases, pollen translucency (Lerche, 1988a,b, 1990), fluid inclusion homogenization temperatures (Roedder, 1984; Goldstein, 1986; Burruss, 1989; Barker and Goldstein, 1990; Leischner and Welte, 1993; Walderhaug and Fjeld- skaar, 1993), and other less common indicators (Ritter et al,, 1993) have been studied, ‘in the past, vitrinite reflectance has usually been modeled using the TTI method (e.g,, Waples, 1980), although some other approaches have been used (see Sweeney and Burnham, 1990, for review). Explanations Of the details of TTI modeling are available elsewhere (eg, Waples, 1980, 1985). However, recent development of kinetic models for vitrinite reflectance (Larter, 1988; Bumham and Sweeney, 1989; Sweeney and Burnham, 1990) now offers an alternative that is almost certainly more accurate than the TTI method (Waples et al., 1992b). Nevertheless, because current kinetic models are unable to distinguish among the different types of vitrinite, they may still predict R, values that have an intrinsic uncertainty of as much as 0.2% Ry (Ungerer, 19883) Biomarkers are increasing in popularity and avail- ability as thermal indicators. The biomarker ratios most commonly used are 20S/(20R + 205) for Coo regular steranes, 225/(22R + 225) for homohopanes, and tr and, ‘monoaromatic steranes. The strengths and weaknesses of these parameters and our current knowledge of the kinetic parameters have been evaluated elsewhere288 Waples (Mackenzie and McKenzie, 1983; Waples and Mackihara, 1991; Ritter et al, 1993), Other molecular indicators for which kinetic descriptions have been developed and used include methylphenanthrenes (Abbott and Maxwell, 1988; Tupper and Burckhardt, 1990), biphenyls (Alexander et al,, 1990), esters (Alexander et al, 1991), and several indicators discussed by Ritter et al. (1993) (Table 17.1). Although all these molecular indicators are Useful their applications are fraught with more difficul- ties than originally anticipated. These problems stem both from uncertainties about kinetic parameters and from unexpected complexities of the chemical systems being modeled (Abbott et al, 1990; Ritter et al, 1993). Consequently, at present their accuracy and degree of application vary widely. “Apatite fission track analysis is based on the annealing, of fission tracks in detrital apatite crystals as a function of time and temperature. In principle, fission track data can yield detailed information on the timing and tempera- ture of thermal events in a rock's history below about 100°-170°C, depending on the effective heating time (eg., Naeser et al., 1989). The technique is thus different from other thermal indicators that simply represent the total effects of heating, In practice, however, some of the ‘mathematical framework crucial to the interpretation of fission track data is in dispute (ee Crowley et al, 1991; Naeser, 1993), and unraveling thermal histories, particu- larly quantitative paleotemperatures, from fission track data can be tricky. Because Rock-Eval Tmax data are abundant, much effort has been expended trying to develop a kinetic ‘model for Tax change. Although such models have now been published (e.g., Sweeney, 1989, 1990), many unsolved problems remain, largely as a result of the dependence of Tmax On kerogen type as well as on ‘maturity. Modeling Hydrocarbon Generation Kinetic modeling of hydrocarbon generation began even earlier than TTI modeling, but the initial models (Tissot, 1969; Tissot and Espitalig, 1975) were much less accurate and popular than modern ones. Today numerous Kinetic models exist, some of which are in the public domain (Tissot et al., 1987; Burnham, 1989; Burnham and Sweeney, 1991). Ungerer (1990) has provided an excellent review of the technical status of Kinetic models. ‘The public models provided by the French Petroleum. Institute (FP) and Lawrence Livermore National Labora- tory (LLNL) are the most widely used. They describe the traditional type I (Green River Shale), type II (Toarcian of the Paris basin), and type Ill (Mahakam delta) kerogens, and they agree well with one another. Examples of kinetic data for standard kerogen types I, IL and IIT are shown in Figure 17.1. Each kerogen is thought to consist of several bond types differing in acti- vation energy by one or two kilocalories per mole. In most models, the pre-exponential factor is considered to be the same forall the bonds in the kerogen, generally on the order of 1012-1016/sec. If pre-exponential factors are 800, A 7007 (GREEN RIVER SH] 600 Tye 7 € 500) a) S300} Ez 2 200} 109 ° 35 40 45 50 55 60 65 70 75 80 85 so, 8 PARIS BASIN + ‘Type = 200 zg z S 100 = 2 "So oO 50 «60 ce MAHAKAM 100: DELTA = Tipe I e 3 80 3 60 z4 2 20 30 «40 «50 60 70 80 (90 ACTIVATION ENERGIES (kcal/mol) Figure 17.1. Histograms of activation energies for various ‘bond strengths in the IFP kinetic model for (A) type |, (B) type Il, and (C) type ll kerogens. The y-axis in each |histogram represents the mass of hydrocarbon formed per ‘gram of organic carbon and thus is directly comparable among all three kerogen types. (From Tissot et al, 1987.)17. Maturity Modeling 289 Table 17.2. Comparison of Kinetic Parameters for Standard Type il and Type ill Kerogens in IFP and LLNL Model ‘Type Il Kerogen EP LLNL® il potential from kerogon (mg/g TOC) 492 360 Gas potential from kerogen (ma'g TOC) 0 65 il to gas cracking efficiency (24) 45 50 ‘A tactor, HC generation (1/sec) 9.078 x 1088 3.010 198 ‘factor, cl racking (1/80c) 3.000 x 10 4.014 x10 Activation energy for ol cracking (kcalimol) 87 54 ‘Activation energy for hydrocarbon generation (kea’mo);fraction® 40 o.014 - 48 9.010 = 49 = 0.05 50 0.087 0.20 51 = 0.50 52 0.555 020 53 = 0.05 54 0.298, - 56 0.045, - 58 oor = 60 0.008 - 62 0.006 = ‘Type lt Kerogen it potential from kerogen (mg/g TOC) 201 50 Gas potential from kerogen (mg'g TOC) 0 110 Oil to gas cracking efficiency (95) 55 50 Afactor, HC generation (1/sec) 5.460 x 1014 1.616 1018 A factor, oil cracking (1/sec) 5.501 x 1014 4.014 x 102 Ea for oil cracking (kcalimo)) 62 54 Ez for hydrocarbon generation (kcal/mol); fraction? 48 - 0.04 50 = 014 52 0015 032 54 0.060 017 56 0373 013 58 0.313 = 60 0.094 0.10 62 0.055 — 64 0.035 0.07 66 0.025 = 68 0.020 0.03 70 0.010 = {FP, Freeh Petcoamirtne rts Tso 1567, UN, Lamenoe Lvemare Nana Labor rts Burhan, 188; rmdfedby Bua, sonal communicate, 128, as dnussodin Wapesett, 1523) ‘Fraaon ol he leregen having tae wh apr cao rey. the same for all bond types, changes in activation energy Table 172 shows a comparison of kinetic parameters as small asa few kilocalories per mole can lead to major differences in the temperature and timing of hydro- carbon generation. (If pre-exponential factors vary among the bond types, it is more difficult to make general comments about sensitivity.) Thus, in comparing, kinetic parameters for different models, one must look at both the pre-exponential factor and the activation energy. Different kinetic models can predict similar ‘generation behavior over a range of geologic conditions only if both their activation energies and pre-exponential factors are similar. for type II and III kerogens from the IFP and LLNL. ‘models, The differences in activation energies between the two models are largely compensated by the differ- ences in pre-exponential factors, resulting in similar predictions about oil generation for most geologic histories. The most significant difference between the two models, in fact, is that the LLNL model allows generation of gas directly from kerogen, whereas this version of the IFP model only forms gas by cracking of oil. (A later version of the [FP model (Espitalié et al., 1988} allows direct generation of gas.) Waples et al290 Waples ‘Table 17.3. Kerogens for Which Published Kinetic Data ‘Are Available Kerogen or ‘Source Rock Name Location _Reeferencess_ Monterey California, U.S.A. A Kirkuk frag A La tuna Venezuela B Phosphoria Montara, U.S.A. Bc Woodtord Oklahoma, U.S.A. BC New Albany Winois, U.S.A, 8 Kentucky, U.S.A. B Posidonia Germany D Talang Akar shales Indonesia E ‘and coals, ‘Alum Shale Sweden F ‘Retorerees: A ist tal (16678, Bumham and Bran (1805, 0, Levan (0985: D,Buthar (1090, Node tal (91 F, Lewin ar Buch 1869), (1992b) have concluded that intrinsic differences between the LLNL and IFP models are considerably smaller than. most of the other uncertainties in maturity modeling, Therefore, the choice of a particular kinetic model (eg, LINL or IFP) for correctly identifying type I, I or lil kerogen would be unlikely to affect exploration decisions. Identification of kerogen type, however, can present problems in modeling hydrocarbon generation. Waples et al. (1992b) have shown that normal errors in esti- ‘mating the proportions of standard kerogen types I and Lin atypical mixed-type kerogen can lead to important differences in modeling results. They also showed that incorrectly identifying a high-sulfur Monterey-type en (type IIS) as a standard type IC kerogen (or vice vend could eto significant errors. The same cautions probably apply to a variety of other nonstandard kkerogens, such as high-wax or resin-tich types. Kinetic parameters have now been published for a number of kerogens from source rocks other than thase representing the three main kerogen types (Table 17:3). Because these data have been obtained over several years, by a mumber of laboratories using different equipment and techniques, some discrepancies have been reported (Gee Bumham and Braun, 1990, for discussion). In most cases, these discrepancies are probably of little practical significance for exploration, but they might occasionally lead to different exploration decisions. One must therefore be cautious about using published kinetic data for individual kerogens. Itis now also possible to obtain “personalized” kinetic data for individual kerogen samples at a cost of about $800-$1000 (US. dollars) per sample. Although the cost is too high to justify routine use of this approach, a company with a strong and ongoing interest in a partic- ular exploration area might find it worthwhile to obtain personalized kinetic data for a few samples of the suspected source rock. However, significantly different Kinetic parameters have been reported for the same rock sample submitted to different laboratories (Hermanrud, 1993), It is thus still an open question whether use of personalized kinetics really improves the accuracy of hydrocarbon generation modeling. Two alternative types of kinetic models exist in addition to the standard IFP and LLNL types of model, which predict generation af oil and gas using a distribu- tion of discrete activation energies (Figure 17.1 and Table 17.2), Quigley and Mackenzie (1988) and Burnham and Braun (1990) have proposed that it would be valid to assume a Gaussian (normal) distribution of activation energies for hydrocarbon formation and to express this range of activation energies using only three parameters: a mean tial factor, a mean activation energy, and a standard deviation for the distribution of activa- tion energies. Quigley and Mackenzie (1988) allow each sample to contain two types of kerogen (called labile and refractory), each with its own set of parameters. However, this approach has yet to be embraced by builders of maturity models. ‘The second alternative shows more promise for wide- spread acceptance. It is called compositional kinetics because it subdivides the products of hydrocarbon «generation into several groups rather than just into oil and gas. Ungerer et al. (1988b) allow formation of four products (C34, Co-Cig, C-Cs, and methane), Forbes et al. (1991) have applied compositional kinetics in a regional analysis. Future research may subdivide the hydrocarbon groups even further. The objective of such detailed analysis is a better description of the effects of phase behavior on hydrocarbon expulsion and migration, The kinetics of gas generation from kerogen has received less attention than that of oil generation. Most models have either ignored gas generation or have included it almost as an afterthought, assigning it the same kinetics as oil generation. The total amount of hydrocarbons generated is then simply partitioned into oil and gas according to a predefined ratio for each kerogen type (eg,, Bumham, 1989; Bumham and Braun, 1990; Burnham and Sweeney, 1991). Use of composi- tional kinetic models should improve our ability to predict gas generation, Modeling Oil Cracking In Source Rocks ‘The action of high temperatures over long times is, responsible for the conversion of kerogen to oil and gas However, if ols retained in the source rock after genera- tion, cracking will eventually convert it to gas plus carbon residue, Maturity modeting of any source rock should therefore predict how much cracking has ‘occurred in the source rock. This problem really has two parts: to determine expulsion efficiency and to determine whether the hydrocarbons retained in the source rock were cracked, In the past, most maturity modeling software ignored expulsion, implicitly assuming that the hydrocarbons simply remained in the source rock. Application of such models therefore often led to overly pessimistic appraisals (eg, “overmature, bumed out, worthless, gaswindow") for rocks that may actually have performed beautifully in the past as effective source rocks. Other models have explicitly allowed some of the hydrocar- bons to be expelled and thus have treated cracking in the source rock more realistically. It is now becoming more common to calculate expulsion explicitly (e.g, Nakayama, 1987; Jabour and Nakayama, 1988; Forbes et al, 1991). Such models can distinguish between the fates of hydrocarbons that are expelled and those that remain in source rocks. Expulsion models are discussed in more detail in Chapter 18 Waples, this volume) In Reservoir Rocks The application of cracking models to reservoir rocks, in contrast, has existed for a long time. To estimate the amount of cracking of oil trapped in a reservoir, we usually simply calculate the total thermal exposure of the reservoir. This method implicitly assumes that most heating of the reservoir rock occurred after the arrival of the oil: In the absence of significant erosion, such an assumption is usually valid. More properly, however, we should first determine the timing of the arrival ofthe oil in the reservoir based on our knowledge of when and where it was generated and the duration of migration, and then compute the thermal exposure of the reservoir rock only after arrival of the oil. Even more properly, we could also consider thermal stress experienced by the oil during migration, although the time spent during migration is apt to be much less than that spent in the reservoir (sce England, Chapter 12, this volume). From. the total amount of thermal exposure, we could in principle compute hydrocarbon composition and volume, ‘Types of Models Oil cracking, like hydrocarbon generation, can be modeled ina nuimber of ways. Cracking in reservoirs has been predicted using both the TTI (eg., Waples, 1980, 1985, 1988) and kinetic methods (e.g,, Ungerer et al., 1988a,b; Behar etal, 1988; Bumsham, 1989; Ungerer, 1990; Bumham and Sweeney, 1991; Andresen et al, 1993). Like hydrocarbon generation, oil cracking can be better predicted using the kinetic method, However, kinetic parameters for oil cracking are more poorly known than those for oil generation. Different oil types appear to have slightly different cracking parameters, although the practical significance of these differences is still conjec- tural. Compositional kinetics have been applied to cracking studies in an effort to obtain more detailed information about the products (Ungerer et a., 1988a,b; Behar etal, 1988). All kinetic models use mass balance considerations to limit the amount of gas that can be formed by cracking of oil. Although the exact proportions vary slightly from ‘model to model and oil to oil, at most about 50% of the cracked oil is converted to gas; the rest becomes carbon residue (Ungerer et al., 1988a,b; Behar et al., 1988; Burnham, 1989; Burnham and Sweeney, 1991), Cracking. in the source rock therefore reduces the total mass of hydrocarbons available for expulsion and migration. 17, Maturity Modeling 291 Overview of Modeling Procedures ‘As mentioned earlier, maturity modeling should begin with the development of a conceptual model of regional geologic history that can serve as a framework for reconstruction of the depositional, erosional, and thermal histories, The burial history of sedimentary rocks, including events that occurred during times repre- sented by unconformities, is then reconstructed, and lithologies are assigned to each rock unit. The necessary input data are often obtained from well samples, but important contributions can come from seismic data, well logs, and outcrops. Heat flows and thermal conduc tivities are then adjusted to give an acceptable fit between measured and calculated present-day tempera- tures. Finally, paleoheat flow and geologic events during, ‘unconformity periods are adjusted to give an acceptable fit between measared and calculated thermal indicators Once this “optimization” process is complete, the input burial, thermal, and petrophysical data should be {as consistent and correct as possible. Hydrocarbon gener- ation and oil cracking are then calculated using the optimized model. The various steps outlined in this section are discussed in more detail in the following section. INPUT GEOLOGIC DATA Burial History Rock Unit Ages Maturity modeling requires that ages and thicknesses be specified for all rock units deposited during the time interval being modeled. Ages for extant sedimentary rocks that have been sampled in outcrop or in wells are normally obtained from micropaleontology and are usually precise. Radiometric dates of volcanic rocks can also be useful. Even unusually large uncertainties in rock unit ages will seldom significantly affect the results of maturity modeling (Waples et al, 1992b). Rock unit ages are depicted relative to depth as in a ‘well bore (Figure 17.2, left), but must be converted to a time history for maturity modeling (Figure 172, right. ‘The mechanics of this conversion are simple. The main problem encountered is deciding on the timing and amount of deposition andl erosion during periods repre sented by unconformities. Water Depth In principle, maturity calculations can be carried out without specifying water depths. However, some software programs require the user to enter water depths, which are useful for two reasons, First, water depth influences temperatures at the sediment-water interface, which should be usde as the surface tempera- ture in the thermal calculations (see later section on Surface Temperatures). Second, inclusion of water depth information allows one to create geohistory plots (van Hinte, 1978), which are discussed later in this chapter. Water depths are usually obtained from micropaleonto-292 Waples ROCK UNIT EVENT (THICKNESS, m) ° lDEPOsIT_F (+300) joePosir F (+300) DEPOSIT E (+1200) Deposit (+300) 1000 100 DEPOSIT ¢ (+800) 2000 ERODE 5 (70) L599 TIME (Ma) DEPTH (m) 300 4000! 400 Figure 17.2. On the lf, rock units ina typical stratigraphic ‘column ere plotied on a depth scale. Ages (in Ma) at rock unit boundaries are shown in parentheses. On the right, the same geologic history i plotid on a time scale. The ‘wo unconformities were assumed to have hed simple histories of deposition units B’ and F) followed by ‘erosion. Numbers in parentheses refer to thickness of interval (+) oF amount of erosion (-). All thicknesses of ‘extant rocks are for present-day, and of eroded rocks, at the start of erosion. Depositiona thicknesses are greater. logic data and are thus typically expressed in broad ranges. The distinction between burial history plots and geohistory plots is the absence versus presence of water depth. Rock Unit Thickness Some programs allow present-day rock unit thick- nesses to be specified directly or to be calculated from rock unit tops, whereas other programs accept only one type of data. Thicknesses are more convenient when using data from outcrops, seismic sections, or geologic cross sections; tops are more useful for modeling wells The thicknesses of existing rock units specified by the modeler are present-day thicknesses after compaction. In most software programs, thicknesses of eroded. rocks represent the thicknesses the units would have had if preserved today since the amount of missing section is often determined by comparison with preserved sections. Some software programs, however, consider the thickness of an eroded section to be its thickness immediately before erosion. There is usually little practical difference between the two methods. Original depositional thicknesses are greater than present-day thicknesses by an amount equal to the amount of porosity reduction due to compaction. Compaction is currently calculated in one of two ways. ‘The simplest approach is to compact each rock layer as a function of maximum burial depth according to a lithology-dependent equation provided by the software program. Table 17.4 shows several types of depth- dependent compaction equations used in maturity ‘modeling, inciuding the most popular Athy-type expo- rrential equation. The pros and cons of various types of compaction equations have been reviewed by Baldwin and Butler (1985). The other popular approach is to calculate porosity reduction as a function of effective stress (Ungerer et al, 1990; Schneider etal, 1993). (Overpressuring,(undercompaction) could be consid- cred to some degree in the depth-dependent equations by merely changing the equations’ constants to yield less compaction. The effective stress approach can directly decrease the rate of porosity reduction during overpres- suring. However, Schneider etal. (1993) have concluded that pressure (and therefore also effective stress) cannot be predicted using one-dimensional modeling (that is, = dimension plus time) since in their opinion most fluid flow is horizontal rather than vertical (Gee England et al, 1987) Alternative approaches to predicting porosity reduction have been developed to take factors other than depth or stress into consideration (Table 17.4), but none have yet been integrated into maturity or basin models. Porosity reduction has been described as a function of time and temperature history (expressed as TTI or Ra) for limestones (Schmoker, 1984) and sandstones (Schmoker and Gautier, 1988, 1989; Schmoker and Hester, 1990; Bloch, 1991). These equations do not contain a direct depth or pressure dependence, although temperature is closely related to depth. Waples and Kamata (1993) have ‘modeled porosity reduction as a series of seven chemical and physical processes. It is not yet clear whether future porosity reduction calculations will continue to be based on depth or effective stress or will adopt some more sophisticated approach. ‘When erosion occurs, release of overburden pressure sometimes leads to expansion of the rock and to an increase in porosity. This uncompaction effect is much smaller than the original compaction, however. Some ‘models ignore uncompaction completely; others allow a small increase in porosity—perhaps 10% of the compaction that occurred during deposition of the eroded rock (Ungerer et al., 1990)17. Maturity Modeling 293 ‘Table 17.4, Common Equations tor Porosity as a Function of Depth or Maturity Equations _ Associated name Relerences Depth-Dependent Compaction Equations $= d0exp(-02) Athy; Sclater-Christie Sclater and Christe, 1980 ito + KZ Falvey-Middieton Falvey and Middleton, 1981 (= @)2 Power function Baldwin and Butler, 1985, Maturity-Dependent Compaction Equations «Te ‘Schmoker, 1984 MRIE ‘Schmoker and Gautier, 1988, 1989; ‘Schmoker and Hester, 1990 8 Bloch, 1991 ‘constants, $15 porosly, an It is sometimes desirable to change rock thicknesses independently of deposition, erosion, or compaction, such as due to igneous intrusive events (although the high temperatures still remain a problem), salt intrusion or withdrawal, or removal of limestone or sandstone during stylolitization. Using such software, the modeler specifies the beginning and ending times for the event and the amount of section added or subtracted, as well as, the lithology of the rock in question. This feature is particularly valuable for describing sections containing, salt since even the ephemeral presence of high-conduc- tivity salt can affect temperature and therefore thermal maturity (O'Brien and Lerche, 1988; Carter and Lerche, in press) Unconformities When a hiatus or erosional unconformity exists, a description of the missing events must be provided. There is no standard formula or method for recon- structing those events. Comparison with adjacent strati- graphic sections may provide ideas about the amounts and ages of rock deposited and later removed. If used. with care, several types of measured data can also yield information about amounts of erosion (eg, Theis et al, 1993). For example, differences in sonic transit time (At values) for shales above and below an unconformity indicate that erosion has occurred (Magara, 1978). When deposition recommences after erosion, the difference in At values between shales above an unconformity and those below depends on the amount of erosion that ‘occurred. However, A values also depend on the burial history subsequent to the erosional event: as the uncon- formity is reburied, the difference in At values gradually decreases and eventually ceases to be of value in esti- ‘mating amounts of erosion (Waples etal, 1992). Vitrinite reflectance data (and other thermal indica- tors) can also be used for the same purpose because, ike At values, they also initially show higher maturities below an erosional unconformity than above. (If little erosion is involved, there is no break in the thermal indi- cators.) However, like At values, breaks in profiles of thermal indicators anneal during reburial and thus must be used cautiously in working out amounts of erosion (Figure 17.3). The graphical approach to estimating is dept removal suggested by Dow (1977) can lead to serious errors where annealing has occurred. A few guides to the use of Ry data in estimating amounts of erosion are available (Katz et al., 1988; cet al, 1989; Waples et al., 1992a,b; Theis etal, 1993). Al such reconstructions should be constrained by geologic concepts as well as by mathematical manipula tion of data. Illustrative examples of reconstructing, thermal histories are given by Deming and Chapman (1985); Feinstein etal, (1989); Hagen and Surdam (1989); and Waples et al. (19924). ‘The increased degree of compaction (lower porosities) of rocks below an erosional unconformity compared to normal (uneroded) rocks at the same depth affects all other petrophysical properties as well as sonic velocities. Errors in amounts of erosion thus lead to errors in thermal conductivity and permeability (often calculated. from porosity; see Chapter 18), and in turn lead to errors in conclusions about maturity and fluid flow. The time spent correctly assessing amounts of erosion is a good. investment. Periods represented by unconformities can consist of complex cycles of burial and erosion, but in most cases, little of this complexity can be included in maturity modeling due to lack of data. Fortunately, the missing, detail is usually of little practical importance, although the way in which one partitions a time gap into deposi- tional and erosional periods can sometimes significantly affect thermal maturity calculations (e.g., Theis et al., 1993). The larger the break in thermal indicators or D, values across an unconformity, the more important it will be to reconstruct the events within that unconfor- mity period accurately (eg., Dahl and Augustson, 1993). From another perspective, detailed knowledge of erosional episodes is most important when the source rocks were at or near their maximum burial depth or temperature during the period in question. Lithologies Lithologies specified for rock units control all petro- physical properties of those units, including compaction rates, thermal conductivities, and heat capacities. Conse- quently, they should be chosen with some care, Litholo- gies of rock units present in outcrop sections or well294 Waples VTRNITE REFLECTANCE (72) WTRNITE REFLECTANCE (72) 02 ost 20 oa ast 2 _ og : a, wee oof aa tows rsd he * 080% 5. I 8000 FT fo ob 4 Bre, e| B s. 02 ose ap o2 oe nen J foe a ah ote Ty ate ee . 7800 FT NT DEPTH ( x1000 FT > sq »4 s{ © 02 os to 20 02 os 2 20 ° . 1 °. Pa 1 OR o= 02% 30 Mo 0 Mo s4 3700 FT 5 fy * 0.04% 5500 FT a a DEPTH (x1000 FT) ote os to 20 50 = PRESENT ms 8 » = 6. 7 Ble Figure 17.3. Development and gradual annealing of a break in a vitrinite reflectance profile during burial, erosion, and reburial. Erosion between the times represented in (A) and (B) leads to a surface Rig (0.75%) that ls 0.5% higher than normal. (C) After 2000 ft of reburia, the gap has shrunk to 0.48%; (D) after 4000 ft, 0.42%: (E) ater 6000 ft, 0.29% (F) after 10,000 ft, 0.08%; and (G) after 12,006 ft, the slope is unbroken. (From Katz eta, 1988.)36 a fn | a6 i 5 | Limestone | & 32 Tchichamanga , Ook Ridge ) ogg Le No of Sompies «21 ° Kmin? 155. & 24 K man? 302 2 20 ae roo tit$ss titel 8 20°22 24 26 28°30 32 34 36 30 40 42 Aa G6 4d 5052 G4 GEES GO EI GA 17, Maturity Modeling 295 Limestones No of Somples = 445 Kmin * 130 Kwon * 626 Lititi bial Laer ‘Thermal Conductivity (W/m K) Figure 17.4. Histogram of total thermal conductivities (matrix plus pore fluid) of limestones measured at room temperature. (From Roy et al, 1981; reprinted with permission from McGraw-Hill Cindus,) bores are normally taken from lithology logs, but can be specified on the basis of any available data, It is more difficult to define the lithologies of eroded rocks. Some software programs simply assign missing, rocks the same lithologies as the extant rocks immedi ately above or below the missing section. This approach has weaknesses, especially when large amounts of section are missing. In other software programs, the user is allowed to specify the lithologies ofthe missing rocks. In such cases, lithologies of eroded rocks are usually estimated by analogy with other spatially or temporally related stratigraphic sections using facies concepts. Petrophysical Properties Petrophysical properties of rocks are specified in one of toes way, alfhough most programs do not allow all three options First ifthe desired thology is the same as one of the lithologies in the program's lithology library, petrophysical properties are simply taken from the library. Second, the user can override the library's values with values of his or her own choosing, or third, the program may allow mixing of two or more of the litholo- ies available in the program’s library and will then calculate the petrophysical properties of the mixed lithology, In most models the lithology library consists of a few standard pure lithologies (such as shale, sandstone, and limestone). The petrophysical properties of any units composed of mite rock types ae then calculated fom the values of each of the components using an averaging, technique (harmonic, geometric, or arithmetic). The averaging technique selected depends on different factors in different pr ‘These factors may include relative amounts of each lithology, nature of the mixing, (interbedded or grain-by-grain), and the particular petro- physical property of interest. Calculating properties of mixed lithology rock units as averages presents a number of problems. First, the true values of the various petrophysical parameters for the pure end-member lithologies are difficult to determine because natural variation is large (Figure 17.4). Although some of the variation is undoubtedly due to different porosities and different pore fluids, considerable scatter Temains that must be due to lithologic composition. Second, the physical nature of the mixing present within the unit is often unclear from available data. Third, the choice of the best averaging technique is not always obvious. Finally, grain-by-grain mixing, in which small grains often simply fill pore spaces between larger grains (Figure 175), cannot be handled by any of the standard averaging equations but rather must be dealt with empirically or by some other mathematical technique (Palciauskas, 1986; Marion, 1990). ‘The other approach to specifying properties of mixed lithologies is to provide a large number of premixed lithologies whose petrophysical properties have already been assigned based on empirical data. The user does not create additonal lithologies by mixing and thus must fit each layer into a predefined lithologic category. If the empirical data base is large and of good quality, this ‘method is excellent because it s grounded in reality.296 Waples Porosity aa ae Clay content (weight traction) Figure 17.5. Porosity of clay-sand mixtures as related to clay content. Dashed and solid lines represent two mathe- ‘matical models for predicting porosities of mixtures. (From ‘Marion, 1990; reprinted by permission of the author.) Faulting Many current thermal maturity models ignore faulting completely, and those that consider itare unable to handle all kinds of faulting. Two difficult problems must be overcome in developing models to describe faulting: the problem of finding general ways to represent faulted sections and the problem of knowing the thermal regimes associated with faulted sections. The representation problems arise from the fact that any ronvertical faut involves lateral movement of material Each fault therefore requires construction of two distinct burial and thermal histories, which must be juxtaposed gradually during the period of faulting to create a single final section. Furthermore, rotation occurring along most faults adds a number of complications. The problems related to thermal regimes have been discussed (but not yet solved) by various authors (eg,, Edman and Surdam, 1984; Furlong and Edman, 1984; Warner et al., 1987; Wygrala et al., 1990) and include both basement heat flow and the role of convective heat transfer. Thermal History Maturity models require that the temperature be specified at all depths at all times in the past. However, in modeling hydrocarbon generation or cracking, or the maturation of thermal indicators, the importance of reconstructing the complete thermal history has probably been overemphasized because most matura- tion occurs near the maximum paleotemperature (Hermanrud, 1993). Thus, for practical purposes, recon- structing thermal history often reduces to the problem of specifying the maximum paleotemperature, when it occurred, and how long it persisted, Surface Temperatures Surface temperatures have generally been given little serious consideration in thermal maturity modeling, Factors that should be considered in estimating the mean annual surface temperature for areas above sea level include latitudinal position through time and global climatic trends (Frakes, 1979). Conceptual models for surface temperature are often unconsciously based on the present, which is probably a poor model for most of the Phanerozoic. For areas below sea level, A. Yukler (personal ‘communication, 1989) suggests estimating the tempera- ture at the sediment-water interface by decreasing the ambient land temperature by 4°C/100 m for the ‘uppermost 200 m of water and by 2°C/100 m thereafter, toa minimum value of about 4°C in the open ocean, or a slightly higher temperature in restricted water. It may also be desirable to consider changes in sea floor temper- ature through time as a function of global circulation. Geothermal Gradient Method Subsurface temperatures can in principle be specified (2) directly at all times using geothermal gradients or corrected bottom hole temperatures, or 2) indirectly using the heat flow-conductivity method. If the geot- hermal gradient method is chosen, gradients can be specified both for the present day and for any desired times in the past and can be made as complex as one wishes. The main weaknesses of the geothermal gradient method are as follows: (1) in most software programs, changes in geothermal gradients are tied to specific depths rather than to rock units, as they should be; 2) modelers often use linear gradients due to lack of data, although actively compacting sedimentary sections normally have nonlinear gradients; and (3) geothermal gradients are effects rather than causes and thus do not provide a sound basis for developing a conceptual model. Heat Flow-Conductivity Method Maturity modeling today increasingly favors the heat flow-conductivity method because it deals with more fundamental physical principles. In fact, some software programs have completely abandoned the geothermal gradient method. Using heat flow, one can easily Construct complex thermal histories that are probably more realistic than those devised using the geothermal gradient method. leat flow across a rock unit represents the product of the thermal conductivity ofthe unit times the geothermal gradient across the unit. The many present-day heat flow Values obtained all over the world from a wide variety of geologic settings vary over a broad range (eg,, Gretener, 1981). However, models for heat flow from the mantle through the crust are still incapable of predicting heat flow values accurately for most cases (see Leadholm et al, 1985; Jensen and Dore, 1993). The intensive studies of the relationship between crustal thickness and heat flow have largely focused on extensional settings (e.g., McKenzie, 1978, 1981; Royden et al, 1980), and there are ongoing controversies about the cor and quanti tative aspects of such models (eg, Tlf et a., 1991). In addition, elevated heat flows associated with crustal stretching are probably of less practical importance than often suggested because source rocks are usually either not yet deposited or only at shallow depths during thetimes of highest heat flow (but for a contrary example, see Cohen’s, 1985, analysis of the Reconcavo basin of Brazil). Moreover, models generally neglect radiogenic sources of heat within the sedimentary section itself. Rybach (1986) contends that contributions from radi- cogenic sources may be variable enough as a function of time and lithology that they should be considered in ‘modeling thermal regimes in basins more than 5 km. deep or more than 10 Ma. Radiogenic contributions can be indirectly and approximately considered, however, by simply adjusting basement heat flow values. ‘leat flow through nonextensional sedimentary basins is even more poorly understood. As a result of these various problems, and also because of local variation, the magnitudes of present or paleoheat flow cannot be predicted precisely from first principles (Blackwell and. Steele, 1989). However, by using the idea that crustal thinning leads to higher heat flows, we can often predict trends qualitatively (e.g,, McDonald et al., 1989). One must therefore test and calibrate the original estimates of heat flow using measured temperatures and thermal indicators (see section on Optimization). Convective Heat Transfer and Thermal Refraction ‘The heat flow-conductivity method applied in one dimensional modeling necessarily assumes that all heat transfer is by vertical conduction and thus will be deficient wherever convective heat transfer or lateral heat conduction (thermal refraction) is important. These two phenomena are discussed elsewhere in Chapter 18 (Waples, this volume) Equilibrium versus Disequilibrium Some software programs assume that sedimentary rocks are always in thermal equilibrium, whereas others allow disequilibrium to occur. Although few sedimen- tary sections are in true thermal equilibrium, disequilib- rium is probably of importance mainly when sedimenta- tion or erosion is rapid (e.g., Deming and Chapman, 1989; Hermanrud et al,, 991; Waples et al, 1992b; Vik and Hermanrud, 1993). Other possible causes for thermal disequilibrium have also been suggested (Powell and ‘Chapman, 1990). ‘Thermal Conductivities ‘Thermal conductivities of all sedimentary rocks in the section being modeled are normally provided by the software program from the lithologies entered by the user. Conductivitics calculated in this way should be considered as rough first estimates, however, for several reasons: (1) errors in specifying lithologies, 2) deviations of lithologic compositions and properties from the gener- alized values provided by the software, (3) overpres- suring (undercompaction), (4) anisotropy, and (5) errors in averaging procedures and natural variation within a given homogeneous set of lithologies (Figure 17.4 and Table 17.5) (See Roy et al., 1981; Blackwell and Steele, 1989). Geophysical well logs and seismic velocities may be useful in estimating thermal conductivities (Goss et 17, Maturity Modeling 297 Table 17.5, Ranges of Thermal Conductivities of Matrix and Pore Fluids ‘Thermal Conductivity thology or Pore Fluid nw?) Shale 10.05-1.45, Sandstone 2.50-4,20, Lithie sand 1.25-2.10 Limestone 2.50-3.10 Dolomite 3.75-6.30 Salt 4,80-60.05 Aniydrite 4.90-5.80, Coat 0.18 Granite 2.50-3.35 Basalt and andesite 1.45-2.10 Water 089 Kerosene 0.15 Methane 00.044 Consus of manic pls po hia 20°C. ‘Seace: may rom Scala Stoo 1665, al,, 1975; Leadholm et al, 1985; Blackwell and Steele, 1989; Brigaud et al, 1990). Most thermal models assume that pore fluids are similar to sea water since most sedi- ‘mentary rocks are marine and hydrocarbons can never represent more than a tiny fraction of total pore fluids in a basin, Where large volumes of hydrocarbons have accumulated locally in highly porous rocks, however, the lower conductivities of oil and gas (Table 17.5) can result in thermal anomalies (e.g,, Eggen, 1983; Meyer and McGee, 1985). If available, measured conductivities are of some value, but they are uncommon and, even when available, may not be representative of entire rock units (Powell and Chapman, 1990). Furthermore, conductivities measured parallel and perpendicular to beciding planes are different, particularly in shales (Blackwell and Steele, 1989), but most measured data do not specify orienta tion. This anisotropy could introduce errors where beds dip steeply or where isotherms have been distorted by faulting or thermal refraction (see Chapter 18, this, volume). Deming and Chapman (1989) offer a rare example of measured conductivites abundant enough to provide a valuable local data base, but those data were obtained during many years of sampling and measure- ment in a well-explored area Because of the problems in accurately predicting thermal conductivities from the information available, conductivities must be checked and calibrated with ‘measured temperature data at the same time heat flow is, optimized. The differences in conductivities among common lithologies can be large enough to cause distinct changes in geothermal gradient across rock units (eg, Blackwell and Steele, 1989). Particularly noteworthy are the high conductivities of quartz sandstone (contrasting sharply with lithic sandstone), salt, and anhydrite, and the low conductivities of shale, organic matter, and pore fluids (Table 175) Thermal conductivities of sedimentary rocks decrease sharply with increasing temperature (Roy et al., 1981;298 Waples Sekiguchi, 1984; Blackwell and Steele, 1989; Brigaud et al, 1993) and increase with decreasing, porosity as low- conductivity pore fluids are expelled. Pressure correc- tions are small and often omitted (Sekiguchi, 1984). Since the in situ conductivity of a rock unit depends strongly ‘on its temperature and state of compaction, the conduc- tivity entered in most software programs is the conduc- tivity of the rock matrix alone at room temperature, Optimization The optimization process begins by comparing present-day subsurface temperatures measured in the well bore with calculated temperatures. Measured subsurface temperatures should always be corrected in some appropriate way for the effects of incomplete thermal equilibration in the well bore. Fertl and Wichman (1977), Gretener (1981), Hagen and Surdam (1989), and Deming (1989) provide some guidelines, techniques, and warnings about standard correction procedures and their uncertainties in the well bore. However, detailed analysis of the precision and accuracy of various temperature correction techniques suggests, several things. First, simple methods consistently under- estimate true rock unit temperatures, with standard devi- ations for data sets on the order of 7°C (Hermanrud, 1988). Second, more advanced methods are fairly accurate but still have about the same standard devia- tions as simple methods (Hermanrud and Shen, 1989; Hermanrud et al., 1990). Finally, the most sophisticated methods (seldom used) can reduce standard deviations, to less than 2°C (Hermanrud et al, in press). Deming, (1989) has made the important point that it is impossible to expect to correct measured temperatures accurately and that better temperature measurements must be obtained if present-day temperatures are to be known. more precisely. Once present-day temperatures in the well bore have been established, modeling is started using (1) a plausible set of thermal conductivities (probably those siven as default values by the software program for the defined lithologies), and (2) an estimate of the present- day heat flow. In subsequent simulations, the present- day heat flow and conductivities of one or more of the rock units can be adjusted to achieve a good fit between ‘measured and calculated temperatures. Thus, heat flow and thermal conductivity are optimized simultaneously rather than separately in an attempt to find a combina- tion of heat flow and conductivity that gives the correct present-day temperature (Waples etal, 1992a). The final choice for a present-day heat flow must also be consis- tent with the geologic model and must use thermal conductivities that are plausible for those particular rocks. This procedure establishes the present-day heat flow and thermal conductivities for the rock matrices at surface conditions Figure 17.6 shows the struggle to fit corrected bottom. hole femperatures to calculated temperatures in the MITT Rumoi well (Waples et al., 1992a), Figure 17.6A shows the results using the initial estimate of thermal conduc tivities (default values for the specified lithologies) and present-day heat flow (38 mW/m2). Figure 17.68 gives the results after forcing a mathematically good fit by adjusting conductivities (increased by 50%) and heat flow (48.5 mW/m2) without regard for geologic and petrophysical constraints. Figure 17.6C show the final best fit taking all geologic and petrophysical evidence into consideration (conductivities of sandstones decreased to reflect lithic content; heat flow = 45 mW /m?). The imperfect fit in the upper part of the section in Figure 17.6C is attributed to errors in measured temperatures. The next step is to find a heat flow history that is consistent with the geologic model for regional tectonic evolution and the data from thermal maturity indicators. ‘The heat flow is optimized by varying it through time in a logical way until measured maturity indicators, calcu- lated thermal maturity indicators, and inferred paleotem- peratures (from fission track data or fluid inclusions) agree as well as possible. The allowable amount of variation in paleoheat flow is highly dependent on the tectonic history within each area, Because of the large uncertainties in heat flow histories, most workers vary heat flow through time freely during optimization. Figure 17.7 shows the testing of three alternative hypotheses about paleoheat flow in the MITI Rumoi well (Waples et al, 1992a). Figure 17.7C was finally chosen to represent the “correct” heat flow model, although the predicted Rg values agree less wel in the upper part of the section than for the heat flow models in Figures 17.7A and B. The thermal and erosional models required to produce the fits in Figures 17.7A and B were judged to be geologically unreasonable (see Waples eta, 1992a, for details). A key point in justifying the acceptance of the scenario in Figure 17.7C was that the Ry data in the ‘upper part of the well were judged to be unreliable. Poor Rg data are probably much more common than is usually appreciated. Use of multiple maturity indicators is encouraged because each one gives a different perspective on the time-temperature history and because the suitability of each indicator varies from case to case. Feinstein et al. (1989), He and Lerche (1989), Ritter et al. (1993), and. ‘Theis et al. (1993) have presented examples using two or ‘more thermal indicators. Much emphasis has been placed in the literature on the importance of thermal maturity indicators in constraining paleoheat flow. Mathematical inversion techniques (also known as thermal indicator tomog- raphy) have been developed to extract the maximum information from thermal indicators (e., Lerche et al, 1984; Lerche, 1988a,b, 1990; He and Lerche, 1989), but few workers have adopted the technology. A major problem with the existing inversion techniques is that they assume that all the geologic uncertainty lies in a single parameter (generally heat flow); every other piece ‘of input geologic data is assumed to be correct. A second problem is that itis tempting to use inversion techniques as a black box replacement for careful geologic analysis. ‘The reality, however, is that inversion technology, like all modeling technology, requires that careful attention be paid to all geologic aspects of the problem, and therefore requires clear thinking. Perhaps the greatest value of thermal indicator tomography is to remind us how17, Maturity Modeling 299 DEPTH (m) TEMPERATURE (°C) poorly our measured data usually constrain paleoheat flows. ‘Thermal conductivities are generally not adjusted during this optimization step. However, if an acceptable fit cannot be obtained using the previously established present-day thermal conductivities, we may adjust the conductivities, recalibrate the present-day heat flow, and then try again to fit the thermal maturity indicators (see Waples et al., 1992a). The final combined heat flow- conductivity scenario should be as consistent as possible ‘with our conceptual model of the geologic history of the area, Figure 17.6. Stepwise fiting of calculated present-day {temperature (line) to corrected bottom hole temperatures (dots) in the MITI Rumoi well, Hokkaido, Japan. See text for discussion of graphs A, B, and C. Abbreviations: Tertiary; K, Cretaceous; k, Kotanbetsu Formation; ¢, Chiku- bbotsu Formation; s, Sankebetsu Formation; h, Haboro Formation; my, Middle Yezo Formation; u, upper; m, ‘middle; |, ower. (From Waples eta, 19924.) Many workers now recognize the importance of using thermal maturity indicators to constrain paleotempera- tures. However, optimization of present-day heat flow and conductivities using present-day temperatures is ‘equally important. Any maturity modeling carried out in the absence of either measured present-day temperatures ‘or measured thermal indicators should be considered speculative (Waples et al, 1992b). While modeling under such conditions is often necessary and valuable, particu- larly in frontier areas (e.g,, Iliffe et al., 1991), it is important to bear in mind the potential errors and uncer- {ainties associated with uncalibrated modeling, During the process of optimizing heat flows and conductivities, it is also necessary to make a final decision about the amount and timing of erosion during each period represented by an unconformity. (Prelimi- rary estimates of erosion should have been made prior to this stage) Comparison of measured and calculated values for thermal indicators can now indicate whether the original estimates are plausible. For example, maturity modeling can help estimate the maximum amount of erosion that could have occurred across an unconformity where the measured thermal indicator profile is continuous. In cases where breaks exist, the best estimate of erosion can be made by trial-and-error comparison of measured and calculated profiles of thermal indicators. If a break in a profile of calculated thermal indicators appears where measured data areDEPTH (m) continuous, the estimated amount of erosion is too large Conversely, if a measured trend shows a break that is absent in the calculated trend, the amount of postulated erosion is too small. In such cases, the modeler simply changes the input data and reruns the simulation. However, natural scatter in thermal indicator data can often make it difficult to determine whether breaks in measured data are actually present (Waples, 1990). Proper use of thermal indicator profiles to reconstruct erosion histories is discussed in Katz et al. (1988), Armagnac etal. (1989), and Wapies etal, 1992 ab). ‘The combination of thermal factors and erosion Figure 17.7. Stepwise fitting of calculated R data (line) to ‘measured Ro data (+) for three proposed geologic ‘scenarios in the MITI Rumoi well, Hokkaido, Japan, using LLLNL kinetic model. See text for discussion of graphs A, ‘and C. Abbrovistions as in Figure 17.6. (From Waples et al. 19928) selected as the final “best” model is seldom a unique solution and probably contains errors as well. However, by constraining both conductivities and heat flows within geologically plausible limits, we can be reason- ably sure thal our solution is reasonably close to the truth (Waples eta, 1992). DIMENSIONS Virtually all maturity modeling is carried out in one spatial dimension (one-dimensional), along the z (depth) axis. As a result, we are unable to include directly the effects of convective heat transfer or thermal refraction, both of which might have a significant lateral component. These limitations are usually not trouble- some, however. In cases where the effects of convective heat transfer or thermal refraction are significant, they can be included indirectly and approximately in the estimate of conductive heat transfer. Some software programs offer a multiwell capability, which produces the same output as two-dimensional (x, 2, plus time) simulations. With this approach, a number of wells in an area are modeled as independent one dimensional simulations, and the results are then plotted as plan-view maps on user-selected horizons or as two- dimensional cross sections along user-selected lines (see examples later). For the purposes of maturity modeling, ‘multiwell one-dimensional is probably nearly as good as true two-dimensional modeling and is certainly faster and simpler.Figure 17.8. Geohistory curve for the MITI Rumoi well, Hokkaido, Japan. Datum is sea level; black areas represent ‘water depth through time. Abbreviations as in Figure 17.6. (From Waples et al, 19922.) It is worth noting that when we discuss modeling, dimensions, we refer only to the number of spatial dimensions. What we call one-dimensional modeling also considers the time dimension in addition to its one spatial dimension and is thus actually two-dimensional. In fact, itis the ability of maturity modeling to include time as a dimension in geologic analysis that in part accounts for its usefulness and popularity APPLICATIONS Of the several ways in which maturity modeling can be applied in petroleum exploration, calculation of timing and extent of oil generation is by far the most common. Other applications are often even more important, however. This section will provide an overview of various ways of extracting useful informa- tion from maturity modeling, Checking the Geologic Model Many less experienced users of maturity models are ‘unaware of the power and importance of using maturity ‘modeling to check and improve their geologic model. If the input data are not properly tested for intemal consis- ‘and reasonableness, any further modeling done with a defective set of input data may be in error. Furthermore, the learning that accompanies develop- ment and checking of the geologic model is often as valuable as the final numerical answer about hydro- carbon generation or oil cracking, For example, if one has to struggle with questions about paleoheat flow and amounts and timing of erosion to achieve a fit between measured and calculated thermal indicators, one may 17, Maturity Modeling 301 aw Figure 17.9. Oil or liquid window (shaded) superimposed ‘on burial history diagram for a well in the Big Horn basin, U.S.A. (From Hagen and Surdam, 1984; reprinted with permission of Rocky Mountain Association of Geologists.) learn much about the geologic evolution of the region, its tectonics, and its depositional patterns (Waples et al., 1992a,b). Previous discussions have shown how the thermal and erosional histories can be resolved with the aid of maturity modeling software (Figures 17.6 and 17,7. ‘Use of geohistory curves (van Hinte, 1978) instead of burial history curves is of value at this stage because the inclusion of water depth gives a more complete picture of subsidence patterns and the interplay between subsi- dence and sedimentation. For example, Figure 17.8 shows that although sedimentation was very rapid during the early Miocene in the vicinity of the MITI Rumoi well, water depth increased as sedimentation failed to keep up with basement subsidence. Hydrocarbon Generation Maturity modeling can predict the degree and timing of hydrocarbon generation and to some degree the composition of those hydrocarbons. The confidence we have in the results depends on the confidence we have in the input data: how well the thermal history is constrained by thermal maturity and paleotemperature indicators and how well the kerogen type and its kinetic parameters are known. The traditional superposition of the maturity window on the burial history (or geohis- tory) curves (Figure 17.9) has become less popular as TTI modeling has been gradually replaced by kinetic modeling and as alternative (and possibly superior) methods of displaying output data have been developed. Among these altematives is plotting hydrocarbon gener- ation for one or more source rocks as a function of time (Figures 17.10) or as a funetion of depth for the entire stratigraphic section at any desired time (Figure 17.11). Knowledge of the timing of hydrocarbon generation is important for two reasons. First, migration pathways can change with time through normal compaction processes, diagenesis, cementation, fracturing and faulting, and structural development, To consider where302 Waples ion (mg HC/g TOC) Figure 17.10. Cumulative generation of ol (slid) and gas (shaded) from middle Miocene time from coals athe base ofthe Haboro Formation, MIT! Rumoi wel, Hokkaido, Japan. (From Weples et al., 19923.) the hydrocarbons went after they left the source rock, we must know approximately when they were moving Second, the later the generation, the less time there has been for biodegradation, cracking, seal rupture, and other destructive processes that destroy or damage hydrocarbon accumulations, The amount and composition of hydrocarbons generated can also be determined to some degree by maturity modeling. Most maturity models predict gas-oil ratios (or give even mote detailed breakdowns if compositional kinetics are used), but the accuracy of such predictions is limited by how well one can characterize the kerogen. When kerogen type and its kinetic parame- ters are known reasonably well, such predictions are fairly reliable. Predicting wax contents, API gravities, and other details is beyond the scope of most models today, however. Accurate prediction of volumes of hydrocarbons generated requires not only that we specify total organic carbon content (TOC) values in addition to kerogen type but also that we extrapolate the results of our model to three spatial dimensions. Such extrapolation requires care (Waple, Chapter 18, this, volume). Oil Cracking Hydrocarbon preservation, like hydrocarbon genera: tion, can easily be shown as a function of time for a ons 4 [olresmuc} 5 DEPTH (md | a 5000 Hi a” "yo "200" 80400500 CUMULATIVE HC GENERATION (mp/a TOC) Figure 17.11. Cumulative generation of oll and gas to present day from coals of the Haboro Formation and from a hypothetical source rock in the Middle Yezo Formation in the MIT! Rumot well, Hokkaido, Japan. Abbreviations as in Figure 176. particular rock unit in the same fashion as in Figure 17.11, or as a function of depth for the whole section (Figure 17.12). Cracking is shown indirectly by the decrease (or slower increase) in total amount of oil and by the increase in both gas and carbon residue. Carbon residue is the most unambiguous indicator since itis only formed by cracking, Such representations are usually used for source rocks, but could also be applied toreservoirs. For source rocks at or beyond peak oil generation, there may be considerable differences between the amount of cracking predicted when expulsion is allowed (Figure 17.12A) and that predicted when expulsion is not allowed (Figure 17.12B). The “no-expulsion” model is considerably more pessimistic in predicting mainly gas and condensate, whereas the “expulsion” model considerable oil. The true composition of hydrocarbons emerging from the source rock is probably better predicted using an expulsion model, in spite of our ‘uncertainties about criteria for expulsion (Chapter 18, this volume). Acknowledgments I thank JNOC for permission to publish this work and for support in preparation of the manuscript. Christian Hermanrud, Michael Zeitlin, and Les Magoon Kindly offered helpful comments on an earlier version, and ‘Nancy Naeser provided perspective on fission track analysis, also am grateful to the many other workers who have gener- ously shared their knowledge in discussion and through publications.‘TOTAL HYDROCARBONS. [ot EJ coNneNsArE £23 WET GAS IJORY GAS DEAD ¢ 17, Maturity Modeling 303 ‘TOTAL HYDROCARBONS low 2 convensare 23 wer cas I} GAS DEAD C ° T T T ad Depth (em) Ls ° T T o ° 70 30 0 Generated HC amount (mg HC/s intial dry rock) ° 0 e % ay Generated HC amount (mg HC/g intl dry rock) Figure 17.12. Comparison of hydrocarbon yields using (A) a new model of generation that allows expulsion and (B) a ‘conventional model that ignores expulsion. (Output provided courtesy of Beicip; calculated using Genex® software.) References Cited Abbott, G. D, and J. R. Maxwell, 1988, Kinetics ofthe aromati- sation of rearranged ring C monoaromatic steroid hydro- carbons nL. Mattavelli and L. Novell eds., Advances in (Organic Geochemistry 1987: Oxford, Pergamon Press, p. 381-885, Abbott, G. D.,G. Y. Wang, T-I. Eglinton, A. K Home, and G. ', Petch, 1990, The kinetics of sterane biological marker release and degradation processes during the hydrous pyrolysis of vitrinite kerogen: Geochimica et ‘Cosmochimica Acta, v.54, p. 2451-2461. Alexander, R, R. Marzi, and R. I Kagi, 1990, A new method, for assessing the thermal history of sediments a case study from the Exmouth Plateau in northwestern Australia: APEA Journal, v.30, pt. 1, p-364-372. Alexander, R, P.G. Kralert, Marz, RL. Kagi, and E. J Evans, 191, A geochemical method for assessment of the thermal histories of sediments: a two-well case study from the Gippsland basin, Australia: APEA Journal, pt. 1, p 325-832, Andresen, P,,N. Mills, H:J.Schenck, and B. Horsfield, 1993, “The importance of kinetic parameters in modelling gener2- tion by cracking of ol to gas—a case study in 1D from well 2/4-14,in A.G. Dore etal, eds, Basin Modelling: ‘Advances and Applications: Amsterdam, Elsevier, p. 563-572, Armagnac, C, J. Bucci, C.G,St.C. Kendall, and I Lerche, 1989, Estimating the thickness of sediment removed at an unconformity using vitrinite reflectance data, in N. D. Naeser and T- H. McCulloh, eds, Thermal History of Sedi- mentary Basins—Methods and Case Histories: New York, Springer Verlag, p. 217-238 Baldwin, B, and C’O. Butler, 1985, Compaction curves: ‘AAPG Bulletin, v.69, p. 622-626. Barker, C. Band R. H. Goldstein, 1990, Fuid-inclusion technique for determining maximum temperature in calcite and its comparisor to the vitrinite reflectance geo- thermometer: Geology, v.18, p. 1003-1006. Behar, F, P. Ungerer, A. Audibert and M. Villba, 1988, Experimental study and kinetic modelling of crude oll pyrolysis in relation with enhanced cil recovery processes, INF. Meyer and J. Wiggins, eds, Fourth UNITAR/UNDP ‘Conference on Heavy Crudes and Tar Sands: Alberta Oil Sands Technology and Research Authority, p. 747-759 Blackwell,D.D, and J LStooe, 1988, Thormal conductivity Of sedimentary rocks: measurement and significance, iN. D. Nacser and T. Hl. McCulloh eds, Thermal History of Sedimentary Basins—Methads and Case Histories: New York Springer Verlag, p. 13-36 Bloch, S, 1991, Empirical prediction of poreity and perme- ability in sandstones: AAPG Ballet, v.75, p. 1145-1160. Brigaud, F, D.S. Chapman, and S. Le Douaran, 1990, Este ‘mating thermal conductivity in sedimentary basins using lithologic data and geophysical wel logs: AAPG Bulletin, v.74, MS9-1474, Brigaud, FG. Caillet and G, Vasseur, 1993, Present thermal State in Sedimentary basins predicted from well data: application tothe Viking Graben, in A.G. Dore etal ed, Basin Modeling: Advances and Applications: Amsterdam, Elsevier. Burnham A. K.,1989, A simple kinetc mode! of petroteun formation and cracking: Lawrence Livermore National Laboratory Report UCID-21665, 11 p. Bumham, A. K, 1990, Pyrolysis kinetics and composition for Posidonia shale: Report UCRL-ID-105871, 25 p. Bumham, AK. and R L. Braun, 190, Development ofa detailed model of petroleum formation, destruction, and ‘expulsion from lacustrine and marine source rocks, in B. Durand and F. Behar, eds, Advances in Organic Geochem- istry 1969: Oxford, Pergamon Press, p. 27-3. Bumham, A. K, and j.J.Sweeney, 1989, A chemical kinetic ‘model of vitrinite maturation and reflectance: Geochimica et Cosmochimica Acta, v.83, p. 2649-2557 Bumham, A.K.andJ.).Sweeney, 1991, Modeling the matt- ration and migration of petroleum, in N. Foster and E. A, Beaumont, eds, AAPG Treatise of Petroleum Geology, p. 55-61. Burrus, 8.C, 1969, Paleotemperatures from fluid inclusions “Advances in theory and technique, x N. D. Naeser and.304 Waples H.McCulloh, eds, Thermal History of Sedimentary Basins—Methods and Case Histories: New York, Springer Verlag, p. 119-231. Carter, and I Lerche, in press, Moving salt sheets and sediment maturity: AAPG Bulletin. Cohen, C.R, 1985, Role of fault rejuvenation in hydrocarbon accumulation and structural evolution of Reconcavo basin, northeastern Brazil: AAPG Bulletin, v.69, p. 65-76 Crowley, K.D, M. Cameron, and R.L. Schacter, 1991, Experi- ‘menial studies of annealing of etched fission tracks in fluo- rapatite: Geochimica et Cosmochimica Acta, v.58, p. 1449-1468. Curiale,J.A./S. R. Larter, RE. Sweeney, and B. W. Bromley, 1999, Molecular thermal maturity indicators in oil and gas source rocks, N.D. Naeser and T. H. McCalloh, eds, ‘Thermal History of Sedimentary Basins—Methods and. Case Histories: New York, Springer Verlag, p. 53-72. Dahl, B. and J. H. Augustson, 1993, The influence of Tertiary ‘and Quaternary sedimentation and erosion on hydro ‘carbon generation in Norwegian offshore basin, in A. G. Dore tal, eds, Basin Modelling: Advances and Applica- tions: Amsterdam, Elsevier, p. 419-432. Deming, D., 1989, Application of bottom-hole temperature corrections in geothermal studies: Geothermics, v.18, p. 775-78 Deming, D, and D. $. Chapman, 1989, Thermal histories and ‘hydrocarbon generation: Example from Utah-Wyoming thrust belt: AAPG Bulletin, v.75, p. 1455-1471 Dow, W. G, 1977, Kerogen studies and geochemical interpre- tations: Journal of Geochemical Exploration, v.7,p. 79-98, Dykstra J, 1987, Compaction correction for burial Fistory curves’ application to Lopatin’s method for source rock maturation determination: Geobyte, v. 2 p. 16-23. Edman J.D.,and R.C. Surdam, 1984, lnfluence of over thrusting on maturation of hydrocarbons in Phosphoria Formation, Wyoming-Idaho-Utah overthrust bel: AAPG. Bulletin, v.68 p. 1803-1817. ager, S, 1983, Modelling of subsidence, hydrocarbon genera ‘Ron are heat wanporin the Norwegan North Sea Durand, ed, Thermal Phenomena in Sedimentary Basins: Paris, Editions Technip, p. 271-284 Exphick, RY. 1989, Lopatin modeling for hydrocarbon ‘maturity, part 1: Geobyte, v. 4, April p. 49-54 Eiphick, RY. and J. Reed, 1989, Lopatin modeling for hydrocarbon maturity, part 2~DOS and Mac programs Geabyte,v.4, june, p.28-37. England, W. A, A.S. Mackenzie, D.M. Mann, and T.M. Quigley, 1987, The movement and entrapment of| petroleum fluids in the subsurface: Journal ofthe Geolog- ical Society, London, v. 144, p. 327-37. Espitali, J, P. Ungerer I Irwin, and F. Marquis, 1988, Primary ‘racking of kerogens, Experimenting and modelling Cy, Cr-Ce Co-Cigand Cys, classes of hydrocarbons formed: Organic Geochemistry, v.13, p. 893-88. Espitalié J, F. Marquis, and $, Drouet, 1993, Critical study of ‘Kinetic modelling parameters, in A.C. Doreet al, eds, Basin Modelling: Advances and Applications: Amsterdam, Elbevier, p. 233-242. Falvey, D.A. and M. F. Middleton, 1981, Passive continental margins: evidence fora prebreakup deep crustal metamor- phic subsidence mechanism: Oceanologica Acta SP, p. 103-114 Feinstein, S, B. P. Kohn, and M. Eyal, 198, Significance of combined vitrinite reflectance and fission-track studies in evaluating thermal history of sedimentary basins: an example from souther Israel, N.D. Naeser and T. H. McCulloh,eds., Thermal History f Sedimentary Basins — Method ane Case Histories: New York, Springer Verlag, p. 197-216 Fee W. Hand P. A, Wichman, 1977, How to determine static BHT from well log data World Oi, 181/1, p 105-106, Forbes 2. P.M Ungerer, A.B Kuhfuss FR and "Eggen, 1991, Compesitional modeling of petroleum gonera- tion and expulsion: tal application toa Yocal mass balance in the Smorbukk Sorfel Haltenbanken ares, Norway. ‘AAPG Bulletin, v.75, p.873-893, Frakes, L.A, 197, Climates Throughout Geologic Time ‘Ameierdam, Ekevier, 310. Furlong. K.P, and] D. Edman, 194, Graphic approach to determination of hydrocarbon maturation in overthnust terrains: AAPG Bulletin v.68 p. 1818-1824 Gof }.C, 1983, Hydrocarbon generation and migration fom Taras source rocks in East Shetland bain and Viking graben ofthe northern North Sea: Journal ofthe Geological Society of London, v.10, p. 445-474. Goldstein, RH, 1986, Reequtibration of uid inclusions in fove-temperature calcum carbonate cement Geology, ¥. 4, p75, Gof R, J. Coumbs, and A. Timur, 1975, Prediction of thermal conductivity in rocks from other physical parameters and from standard geophysical wel legs SPWLA I6th Annual Logging Symposium, p. 1-21 Grotener, P1981, Geothermics using temperature in hydrocarbon exploration: AAPG Eatication Course Notes Series 17,155 p Hagen E.S, and R.C.Surdam, 1984, Maturation history and thermal evolution of Cretaceous source rocks ofthe Bighor basin, Wyoming and Montana in J. Woodward, F. F Meisner, and .L. Clayton, eds, Hydrocarbon source rocks ofthe greater Rocky Mountain region: Denver, Rocky Mountain Association of Geologists, p. 521-33, Hagen, E-5, and RC Surdam, 1989, Thermal evolution of Taramie-style basins constraints from the northem Dighor basin, Wyong and Montana, N. D. Naeset and T. H. MeCalloh eds, Thermal History of Sedimentary Basins Methods and Case Histories: New York, Springer Verlag, p. 277-29. He Z, and Lerche, 198, Inversion of multiple thermal indi ‘ators: quantitative methods of determining palecheat fx and geological parameters IV: case histories sing thermal indicator fomograpy: Mathematical Geology, \ 21 p. 503-541 Hermanrad, C, 1968, Detenmination of formation temperature from downiole measurements PRD. dissertation, Univer- sity of South Carolin, Columbia,SC, 23 p. Hermanrud,C, 1993, Basin modelling fechniques—an ‘overview, A. G, Dore et a, eds, Basin Modeling: ‘Advances and Applications: Amsterdam, Ekevier,p. 1H. Hermanrud,, and P, Y-Shen 1989, Virgin rock temperatures from well logs accuracy analysis for some advanced inversion models Marine and Petroleum Geology, v.6,p. 340-263. Hermanrud,C, 5,Cao,and I Lerche, 190, Estimate of virgin rock temperature derived from BEIT measurements: bias and error Geophysics, v.55, p. 924-931, Hermanrud,C, 5 Eggen, and RM. Larsen, 1991, Investga- tion of the thermal regime ofthe iorda platform by basin modeling: implications for the hydrocarbon potential of the Stord basin, northern North Se, A. M. Spence, ec, Generation, Accumlation and Production of Europes Hydrocarbon: Oxford, Oxford University Pres, p73. Hermanzud,C, I. Lerche, and K. K. Mesingse in press, Determination of virgin rock temperature from dil stem tests SPE Formation Evaluation Socety of Petroleum Engineers.