Skip to Main Content

Abstract

The prospect reserves distribution is really an estimate of the range of ultimate volumes of oil and nat-ural gas that may be recovered if the prospect discovers a producible hydrocarbon accumulation, which may become an oil or gas field. As discussed earlier, this value does not equate to “proved,” “probable,” or “possible” reserves, as formally defined engineering parameters (Capen, 1996; Cronquist, 1997). Those engineering definitions arose out of fiduciary needs and are subject to continual revision throughout the life of the field. They involve considerations of reservoir volume as well as detailed reservoir parameters, flow rates and decline curves, and multiple economic assumptions. Obviously, such details are usually not available for exploration ventures.

Accordingly, many companies employ a simpler set of parameters (Figure 6) that are more consistent with the high degree of uncertainty that attends exploratory prospects: Prospect Reserves = Productive Area (in acres, hectares, or kilometers2) χ Average Net Pay Thickness (in feet or meters) χ Hydrocarbon-Recovery Factor (in bbl or mcf [thousand cubic feet] per net acre-foot, bbl per net m3/hectare-meter, or m3 per net km2-m). The parameters shown in Figure 6 are deterministic; that is, single-value estimates for each parameter, all of which, because of substantial geotechnical uncertainty, are much better forecast as a probabilistic range of possible outcomes. Deterministic predictions are generally unreliable; fortunately, their use in the modern exploration industry is diminishing.

Estimating Prospect Reserves

Constituent Parameters

The prospect reserves distribution is really an estimate of the range of ultimate volumes of oil and nat-ural gas that may be recovered if the prospect discovers a producible hydrocarbon accumulation, which may become an oil or gas field. As discussed earlier, this value does not equate to “proved,” “probable,” or “possible” reserves, as formally defined engineering parameters (Capen, 1996; Cronquist, 1997). Those engineering definitions arose out of fiduciary needs and are subject to continual revision throughout the life of the field. They involve considerations of reservoir volume as well as detailed reservoir parameters, flow rates and decline curves, and multiple economic assumptions. Obviously, such details are usually not available for exploration ventures.

Accordingly, many companies employ a simpler set of parameters (Figure 6) that are more consistent with the high degree of uncertainty that attends exploratory prospects: Prospect Reserves = Productive Area (in acres, hectares, or kilometers2) χ Average Net Pay Thickness (in feet or meters) χ Hydrocarbon-Recovery Factor (in bbl or mcf [thousand cubic feet] per net acre-foot, bbl per net m3/hectare-meter, or m3 per net km2-m). The parameters shown in Figure 6 are deterministic; that is, single-value estimates for each parameter, all of which, because of substantial geotechnical uncertainty, are much better forecast as a probabilistic range of possible outcomes. Deterministic predictions are generally unreliable; fortunately, their use in the modern exploration industry is diminishing.

Productive Area

Utilizing geologic and geophysical data in the prospect area, professional staff are asked to make maps showing a reasonable high-side productive area, given optimistic geologic conditions such as seismic velocities, dip rates, contour configurations, fault extents, trap fill-up, and the like. They also construct maps showing a low-side area, assuming pessimistic geologic conditions, and an intermediate area using best-guess conditions. Values derived from such geologic cases are plotted on a cumulative log probability graph (Figure 7), equating to a reasonable, optimistic case (P10%) and a reasonable, pessimistic case (P90%), and a straight, sloping line is “best-fitted” to the data, forcing the distribution to be lognormal. The P50% case is consequential, derived from the intersection of the P50% horizontal line and the sloping line. Next, the provisional distribution is projected out to the low (P99%) and high (P1%) extremes, which are assessed for their plausibility as being highly unlikely but possible outcomes. The P1% value should be so large that it is barely possible, honoring the data; similarly, the P99% value should be just large enough to be consistent with a very small detectable reservoired accumulation. A “final” distribution should be developed, following several iterations and adjustments, utilizing given plausibility checks and reality checks such as those shown on Table 4. There is no “formula” for deriving the P90% area. It must be prospect-specific and consistent with anticipated structural configuration of the closure and stratigraphy of the reservoir section3. In general, however, the P90% area should be small, corresponding very roughly to an onshore, marginally economic area of drainage (assuming pay thickness and hydrocarbon [HC] recovery are sufficient to warrant completion)—a one- or two-well field. Based on theoretical grounds, empirical observations (Figure 8), plots of actual fields in real basins (Figure 9), and detailed analyses (Squire, 1996), the distribution of area forecasts is lognormal in form.

Figure 6

Reserves parameters for exploration prospects (deterministic).

Figure 6

Reserves parameters for exploration prospects (deterministic).

In practice, geoscientists seem to be able to arrive at high-side estimates for prospect area (P1% and P10%) fairly readily. However, settling on appropriate low-side values seems to be more problematic. The most common error is that estimates of P90% and P99% areas are too large, leading to frequent overestimates of mean area and thus mean reserves. But the determination of the area distribution cannot be “formulaic”—it must be consistent with the data bearing on prospect geometry, and it must be prospect-specific. This is discussed in detail in Appendix C.

Two general estimating approaches seem to dominate:

  • Estimates are made of optimistic (P10%) and pessimistic (P90%) cases, which are then plotted on a log probability graph, and a straight, sloping line is drawn between the two points. This line is then extended out to the P1% and P99% extremes, which are evaluated for plausibility. If any of the derived values are implausible, the line is adjusted until a credible best fit is achieved. This approach is preferred, especially for less experienced prospectors.

  • Assuming all plausible conditions maximizing productive area are operative (velocities as they affect dip rates, contouring, trap fill-up), the resulting maximum outline of that area is mapped and measured; this “max area” is provisionally assigned to P1%. Next, the smallest possible area is assumed (consistent with the geologic attributes of the prospect) that would be consistent with a reservoired small HC-accumu-lation just large enough to sustain flow. This area is assigned a provisional P99% value. Both values are plotted on a log probability graph. From the distribution line, the consequent P90%, P50%, and P10% values are derived and checked for credibility. Again, the line is adjusted as necessary to produce a plausible best-fit distribution.

Figure 7

Area, average net pay, and HC-recovery factor are lognormal.

Figure 7

Area, average net pay, and HC-recovery factor are lognormal.

It should be recognized that, following these methods, any outcome larger than the P1% forecast is treated as impossible, which is not strictly true but pragmatically can be dismissed as such. Thus no value larger than the P1% value can contribute to the mean of the distribution.

Average Net Pay Thickness

Using lithofacies, isopach, porosity and net-to-gross maps, geologic studies of pertinent depositional models, and analog field studies, and considering the interactive effects of the oil/water contact, geologic structure, and reservoir distribution, the exploration team generates a probabilistic distribution of estimated average net pay throughout the area of the accumulation based on P1%, P10%, P50%, P90%, and P99% confidence levels plotted on log probability graphs (Figure 7). Such estimates should be derived in an analogous manner to the area estimates previously described and employing similar reality checks (Table 4). This parameter also typically follows a lognormal distribution (Figure 8), but one ordinarily having less variance than the area distribution. In order to consider porous and tight intervals, reservoir net-to-gross ratio must be considered. Also, a geometric adjustment will need to be made to take into account the geometry of the oil/water contact in relation to the reservoir geometry (Figure 10). For most new-field wildcat (NFW) prospects, a distribution of estimated average net pay thickness (determined as described earlier), throughout the productive area will suffice. It is essential to understand that the “average net pay thickness” estimate integrates internal reservoir distribution, porosity cutoffs, trap fill-up, net-to-gross ratio, and the geometry of the “top-reservoir/oil-water contact” couplet.

Gross Rock Volume

A useful alternative to the Area χ Average Net Pay approach outlined previously is to estimate the probabilistic range in Gross Rock Volume (GRV). There are both advantages and disadvantages to this alternative. The advantages are:

  • It eliminates minor problems caused by the common partial dependency of average net pay thickness on productive area;

  • Calculation of the geometry factor required for correct expressions of average net pay can be dispensed with;

  • Integration of maximum (P1%) area with associated vertical column height and gross reservoir thickness (“area vs. depth plots”) will generate a maximum gross rock volume that represents a reality check on the upper limit for GRV (Figure 11);

  • Consideration of minimum pay zone thickness (taking into account expected net/gross reservoir ratios and required HC column heights), together with the resulting associated area of accumulation that would be required to provide a reservoired accumulation of sufficient volume to support sustained flow, provides a reality check consistent with P99% for GRV; and

  • Plotting these upper and lower values as P1% and P99% on log probability graph paper allows derivation of P10%, P50%, and P90% values for GRV.

Table 4

Table 4 Reality checks (1): characteristics of the endpoints of the reserves distribution.

Figure 8

Distributions of estimated parameters for prospect reserves.

Figure 8

Distributions of estimated parameters for prospect reserves.

The disadvantages, however, are more compelling: GRV is a complex number combining area, thickness, and reservoir/water-level geometry considerations. Accordingly, geoscientists find it quite difficult to relate GRV intuitively to maps or cross sections and, therefore, often don't recognize whether their probabilistic estimates of GRV may be implausible. Moreover, when GRV estimates are checked in post-audit drilling reviews, it is not immediately apparent whether errors were related to area, thickness, or geometric misjudg-ments. The fundamental problem here is that explo-rationists conceptualize prospects in relation to maps (area) and logs and cross sections (thickness), and therefore they have a much better subjective appreciation of the credibility of such estimates than of their combination as GRV, which is a parameter having three variables (area, thickness, and geometry), and thus having a wide range of possible combinations capable of giving the same product.

Figure 9

Productive field areas, East Texas, Capen.

Figure 9

Productive field areas, East Texas, Capen.

Figure 10

Graphs to derive geometry factor adjustment.

Figure 10

Graphs to derive geometry factor adjustment.

Figure 11

Area versus depth plot.

Figure 11

Area versus depth plot.

In fact, utilization of both approaches often provides valuable cross-checks for consistency, and usually leads to better estimates of prospect reserves. That's why use of both procedures is recommended.

In any case, however, GRV, combined probabilistically with estimates of reservoir net/gross and HC-recovery factor, yields a lognormal probabilistic reserves distribution that should be compatible with the reserves distribution derived via the “Area χ Average Net Pay χ HC-recovery Factor” method.

Estimation of both the P1% and P99% cases will require construction of an “Area vs. Depth” plot, as shown in Figure 11. Such plots are readily constructed by measuring the areas enclosed by successive contours. “Depth” is understood to indicate the vertical difference from the crest of the feature to the projected oil/water contact (or gas/water or gas/oil contact where appropriate).

Hydrocarbon-recovery Factor

This parameter expresses “reservoir yield” as barrels (bbl) of oil or mcf (thousand cubic feet) of natural gas per acre-foot of reservoir. It is readily adaptable to the metric system, or to expressing oil in metric tons. Some companies choose to break this parameter down into its four constituent components: porosity, HC-saturation, percent recovery, and formation volume factor. This is not wrong, but for most exploratory ventures, it represents false precision and inefficient use of geotechnical effort: HC-recov-ery factor is entirely adequate given the high level of uncertainty that attends most exploratory prospects. However, many companies choose to estimate the four constituent components as reasonable P10% -P50%-P90% ranges, and combine them, via Monte Carlo simulation, as a reality check to ensure that HC-recovery factor estimates are indeed credible values consistent with known or postulated reservoir parameters. Also, expressing yield as bbls or mcf per acre-foot is compatible with the widespread use of analog field models in modern exploration. Again, a range of rational probabilistic estimates is generated by the exploration team, analogous to the process described previously (Figure 7), and plotted on a cumulative log probability graph, then iterated and adjusted until a best-fit is reached. This parameter also takes a lognormal form (Figure 8). The components of HC-recovery factor and metric equivalencies are shown on Table 5. All estimates should be reality-checked (Table 4).

Table 5

Getting a sense of scale about HC-recovery factor.

Generation of the Prospect-reserves Distribution

When probabilistic estimates of prospect area, average net pay, and HC-recovery have been posted on a log probability graph (Figure 7), reconciled with available reality checks (Table 4), and accepted by a consensus of concerned geotechnical staff, the three distributions should then be combined by multiplication into the prospect-reserves distribution. Ordinarily this combination is accomplished through reiterative computer procedures. The two most widely used procedures are Monte Carlo simulation and Latin Hypercube simulation. Through their AAPG school on risk analysis, Capen, Megill, and Rose developed an analytic method that graphically performs the combination on cumulative log probability paper, assuming the three component distributions are lognormal and have variances that are not drastically different (Appendix C). This method was described by Capen (1992), Megill (1992), and Rose and Thompson (1991). Figure 12 illustrates this graphical procedure.

The resultant prospect-reserves distribution (Figures 7 and 12), because it involves the multiplication of three constituent independent variables, is expected to take a lognormal form. The mean represents the single best expression of the distribution's value. Table 6 provides very useful reality checks of the prospect-reserves distribution, relating its variance to the category of exploratory venture, as well as modifying data, such as three-dimensional (3-D) seismic data, and positive indications of direct hydrocarbon indicators (DHIs).

Finally, it should be emphasized that most prospect-reserves overestimates do not arise because the high-side estimates are too high, but rather because the low-side estimates are too high. Accordingly, when reviewing any prospects, all prospectors (as well as their managers) should unfailingly ask the following question: Is there any chance this prospect could turn out to be a mediocre little one-well field? If the honest answer is yes, then the P99% value must reflect such an unfortunate outcome, and the consequential P90% value must also be compatible. Appendix B is an essay that covers this issue in detail, and all prospectors should read it carefully. Table 6 is also useful in addressing such issues.

Note that multiplication of the three 90% values yields a value corresponding to the P98.7% reserves product, and the three P10% values multiply to give a P1.3% product. This pattern, explained in more detail in Appendix C, is important in considering what low-side geologic values are being risked when estimating the chance of geologic success (see pp. 35-36).

Economic Translation of the Prospect-reserves Distribution

Because the mean is the best single expression of the total prospect-reserves distribution, many companies perform an economic analysis on only the mean reserves case of prospects. This is adequate in many cases. For important projects, however, discounted cash-flow (DCF) analyses should be performed on all four key reserve parameters (P10%, P50%, Mean, P90%), in order to understand the relative profitabilities of a full range of prospect outcomes. This can then lead to a profitability distribution for the key prospect, given success. In other words, what we are really after is not the net present value (NPV) of the mean reserves case, but rather, the mean of the NPVs of all four reserves cases. This is especially important in production-sharing contracts and in expensive offshore projects where step-functions create a nonlinear reserves/PV ratio having multiple inflections in the curve.

If project managers can estimate the approximate reserves required for completion of the exploratory well as a commercial venture, on a cost-forward basis, or reserves required for economic profitability on a full-cycle economic basis, the prospect-reserves distribution can then be used to estimate the chance of commercial success, and the chance of economic success, given that flowing hydrocarbons are discovered. The proportion, as a percent, of the reserves distribution that is larger than the minimum reserve amount needed to justify commercial and economic success, respectively, represents the chance of finding those volumes of reserves (or more), given that reservoired mobile hydrocarbons are found at all. This concept is discussed in further detail on pages 39–1.

Figure 12

Graphical method for analytical solution for combining three lognormal distributions by multiplication.

Figure 12

Graphical method for analytical solution for combining three lognormal distributions by multiplication.

Field-size distributions (FSDs) for the basin or play (see p. 11) provide effective reality checks against which the prospect-reserves distribution should be compared. Also, the variance of the prospect-reserves distribution should be compatible with the exploration well class, as shown in Table 6. But the variance of the prospect-reserves distribution is also a function of the quantity and quality of pertinent information, as well as of the geotechnical skill of the professional staff evaluating it, and all such factors must be considered. For all these reasons, the prospect-reserves distribution need not have the same variance as the parent FSD. In fact, prospect-reserves distributions commonly (but not always) show variances smaller than their parent FSD. Because FSDs tend to get smaller as exploration progresses (Figure 5a), prospect-reserves distributions should be compared with fields that have been recently discovered, using compatible exploration concepts and technologies.

Monitoring and Improving Predictive Performance

Until recently, the systematic preservation of reserve predictions and their comparison with prospect outcomes has not been part of the corporate culture of most oil companies. Management often did not know how inaccurate reserve predictions were, whether there were any repetitive patterns of estimation errors, or what the specific causes for persistent errors might be.

From an organizational perspective, it is easy to understand the reluctance of geotechnical staffs to preserve their forecasts and compare against outcomes. But it is difficult to understand why higher management has not insisted on systematic monitoring of predictive performance by geotechnical staff and continuous efforts to improve such forecasts because of their profound impact on project profitability.

Two methods are currently in use. The first (Rose, 1987) employs simple log-log cross-plots (Figure 13) of prospect parameters such as:

  • prospect area

  • average net pay thickness

  • HC-recovery factor (or its constituents—porosity, oil saturation, percent recovery, and formation volume factor)

  • initial production rate

  • percentage decline

  • prospect reserves (= EUR)

By plotting the median of predicted parameter distributions against the actual outcomes, we can, with a relatively small number of trials, gain insight into the variance (scatter) as well as any existing bias. If staff are unbiased, we should have about the same number of overestimates as underestimates. About 10% of outcomes should be larger than their P10% estimate, and about 10% should fall below their P90% estimates. About 80% of outcomes should fall within their P10%-P90% ranges. Causes of undesirable performance patterns can be investigated by going back to individual prospect cases. This method is especially useful to individuals and smaller organizations who may drill only a few wells per year. The method can also be applied to geotechnical predictions made on competitor wells. Additional information can be gleaned by plotting the predicted P10%-P90% range on the horizontal axis and the actual outcome on the vertical axis.

The second method (Clapp and Stibolt, 1991) for monitoring and improving reserves forecasting (Figure 14) employs continuous “tracking” of an ongoing exploration program, comparing actual geologic reserves-added against chance-weighted mean reserves of all prospects in the program. A cumulative “actual reserves-added” curve is plotted in relation to an “envelope” bounded by the cumulative P10% and P90% predictions as well as the “expected” (= mean) curves. A predictive envelope for numbers of new discoveries can similarly be constructed, which can reveal bias in “chance of success” forecasts. Any time the actual performance goes outside the forecast P10%-P90% envelope, management legitimately can suspect that there are serious problems with reserves or chance predictions by geotechnical staff,4 and corrective measures must be instituted. This second method requires Monte Carlo (or Latin Hypercube) simulation to add (rather than multiply) prospect reserves parameters of serial exploration ventures. The only limitation to this method is that additional analysis, such as cross-plots as described earlier, are necessary to isolate and identify the specific causes of estimation bias because the method reveals only bias, not its root causes.

Figure 13

Comparison of predictions versus outcomes (from Rose, 1987).

Figure 13

Comparison of predictions versus outcomes (from Rose, 1987).

Figure 14

(a, b, c, d) Continuous tracking of organizational predictive performance (Copyright SPE #22038. From Clapp & Stibolt, 1991, reprinted with permission.)

Figure 14

(a, b, c, d) Continuous tracking of organizational predictive performance (Copyright SPE #22038. From Clapp & Stibolt, 1991, reprinted with permission.)

Such routine monitoring of actual performance can be instituted in exploration organizations if (1) management insists on such activity being carried out, (2) results are circulated openly within the organization,(3) geotechnical staff accept unbiased predictive performance as part of their professional obligations, and (4) results are used constructively, to improve staff performance, rather than punitively.

Industry Experience

Persistent Overestimation of Prospect Reserves

Since 1993, most oil companies have acknowledged that their geotechnical staffs persistently overestimate prospect reserves, commonly by about 30% to 80%. Three of the four cross-plots on Figure 2 (b, c, and d) document such overoptimistic bias as expressed by individual companies. But overoptimism is not limited to certain companies—it appears to be a chronic industry shortcoming that has proved to be difficult to correct. Figure 15 records the predictive performance of all companies that made discoveries in the Norwegian North Sea during the 8th to 14th rounds, resulting in only 38% of the new reserves predicted. Moreover, this bias is not improved by modern technology: as an example, BP-Amoco (Harper, 1999) reports persistent overoptimistic bias in their deep-water exploration program, which employs state-of-the-art seismic technology, since the early 1990s (Figure 16). BP-Amoco's successful deep-water ventures found only 45% of the reserves predicted. Before addressing the real causes of reserves overestimation, it is important to emphasize that the problem is not caused by comparison of “before-discov-ery EURs” with “after-discovery booked (or fiduciary) reserves.” Exploration estimates of reserves before drilling and after discovery relate to the same parameter: EUR. At least six real causes are responsible for this bias.

Motivational Bias

Individual prospectors, as well as their managers, appear to allow their enthusiasm for drilling the exploratory well to overcome their objective estimates of prospect-reserve parameters. This can be overcome by emphasizing the correct criterion for company success: adding value, not drilling wells. Sometimes the operative motivation seems to be: “I cannot get promoted unless I find oil, and I cannot find oil unless I get my wells drilled,” so prospect parameters are enhanced in order to encourage management to drill inferior prospects. Also, upper management's desire to find large new fields may tend to encourage the entire staff to expect such large discoveries (Boccia, 1996).

Figure 15

Pre- and postdrill discovery sizes.

Figure 15

Pre- and postdrill discovery sizes.

Figure 16

Volume Accuracy

Figure 16

Volume Accuracy

Low-side Estimates Too Large

Large-company employees who have not worked extensively in mature, onshore provinces are simply unaware of how very many small fields can be present in a given basin or trend. Accordingly, they do not set their P90% estimates for prospect area, pay, and recovery factor small enough, so their resulting P99% reserves estimates are too large. Clear evidence of this problem is that the P10%/P90% ratios of prospect reserves for such prospects are too small (i.e. characteristic of development or extension wells rather than exploratory wells). Here are two useful paradigms that help convey reality: (1) The most common field size in the Permian Basin of west Texas is 10,000-20,000 barrels; (2) The P99% field size in most mature provinces is 1,000 to 10,000 barrels. Accordingly, this question should be addressed to every considered prospect: “Could this prospect turn out to be a mediocre little one-well field?” In most cases, the honest answer must be yes. The P99% reserve value should reflect this.

Overconfidence in Geotechnical Discrimination

Another key insight derived from examination of Figure 2 is that geoscientists are not capable of effectively discriminating between prospects that contain large volumes of oil and prospects that contain small volumes of oil (see p. 6). Petroleum prospectors can identify anomalies (called “prospects”) that have an enhanced likelihood of containing oil and gas accumulations. And they can also distinguish, to some degree, those prospects that are large enough to contain large reserves from those prospects that are not. But they usually cannot identify large prospects that are significantly underfilled, thus containing small-volume reserves. Furthermore, other parameters such as aver-age net pay and HC-recovery have such a large impact on reserves and are so variable that our ability to predict them with precision is limited.

Deterministic Estimates Rather than Probabilistic Ranges

Single-value estimates of uncertain parameters predict an outcome that is possible, usually optimistic, and nearly always wrong.

Using Triangular Distributions Rather than Lognormal Distributions

Triangular distributions are a very poor proxy for the prevailing lognormal distribution and usually lead to substantial overestimation.

Nonrepresentative Analog Field-size Distributions

FSDs are very useful as reality checks, indicating the characteristic sizes of fields that exist in a basin or trend. However, it is important that the FSD employed as analog is in fact a valid example: it must comprise fields that have been discovered in current or recent exploration campaigns, using technology consistent with your prospect. It is incorrect and misleading to include in a “current” (2000) FSD, discoveries that were made 40 years ago, during the “flush” period of exploration.

Failure to Monitor Long-term Field Growth

Another factor may actually work the other way: work by Wood et al. (1990) and Attanasi and Root (1994) has emphasized that long after their discovery, oil and gas fields continue to grow. Such reserve appreciation goes on for 50 years or more in large fields, and it is surprisingly substantial: reserve appreciation of booked reserves is nearly 10 times the initial reserve estimate of U.S. oil and gas fields and more than 8 times the initial estimate for U.S. gas fields. When such subsequent field growth is not monitored and periodically compared against initial prospect reserve predictions, some prospect-reserve predictions may erroneously appear to have been more overopti-mistic than they really were. However, unpublished data by Mobil Oil Corporation demonstrate that no bias exists in forecasts of field EURs made immediately after discovery, compared with field-reserve projections made after extensive drilling and production data were available: there is great uncertainty in such estimates, but little or no apparent bias (Dave Cook, personal communication, 1997). Unpublished data from BP-Amoco show a similar lack of bias.

Remedies

The few oil companies that can report unbiased reserves estimations (McMaster, 1998) are organizations whose staff professionals know that management desires their objective geotechnical estimates, that management is monitoring staff predictive performance, and that unbiased predictive performance will be recognized and rewarded. Such staffs have received and utilized training in estimating proposed parameters, and they know the patterns leading to bias. They typically accept professional accountability for objective estimating. Nevertheless, it must be acknowledged that overestimation of prospect reserves is a widespread industry bias that has proved difficult to eliminate (Johns et al., 1998; Alexander and Lohr, 1998; Harper, 1999). The single most effective remedy is to ensure that the P99% values reasonably represent a very small reservoired accumulation that would be flowable.

Reducing Uncertainty—What Is Possible?

With regard to reducing uncertainty, companies that rigorously employ methods like those described earlier report that they can consistently reduce predictive ranges of reserves from 2 to 3 orders of magnitude at 90% confidence, to 1.0 to 1.5 orders of magnitude at 80% confidence. Obviously, data quality and quantity have substantial influence here. For the components of prospect reserves (area, average net pay, and HC-recovery), capable geotechnical professionals should expect consistent predictions within about 1 order of magnitude at 80% confidence, and they should strive for accuracy of 0.5X to 2X at 80% confidence. However, consistently achieving such levels of predictive performance is, realistically, unlikely, using conventional exploration methods. However, where geologic conditions render application of 3-D seismic data feasible and economic, reserve forecasting at such levels may be achieved, especially if amplitude anomalies and other DHIs reduce the likelihood of very large and very small reserves outcomes.

Table 6

Reality checks (2): characteristic ranges associated with oil and gas ventures having different magnitudes of uncertainty.

Table 7

Expected value examples (coin toss).

Chance of Prospect Success

The Expected Value Concept

Imagine that you have the opportunity to participate in a simple game in which you are asked to correctly call the fair toss of a coin. If your call is correct, you will win $20,000; if it is incorrect, you will win nothing.

If you were able to play such a game free of charge, the Expected Value (EV) of each trial would be (+)$10,000 (Table 7). If you had to pay $10,000 each time you played, the EV would be zero, so that, statistically, you then would be “trading dollars.” If you were willing to invest $4,000 in one trial of this game, the EV would be (+)$6,000. In this example, there are only two possible outcomes, and you are restricted to one trial. The chances of either outcome, as provided by our knowledge about coin-tossing, are essential to calculating the EV of the venture.

In order to calculate the EV for an exploratory well, we will use our knowledge about petroleum geology to estimate the chance (= our confidence) that a reser-voired petroleum accumulation is in fact present and will be encountered by the drill bit. It is important to emphasize that for most companies involved in oil and gas exploration there are many ventures, each with an uncertain outcome. Furthermore, the employment of EV as a decision criterion encourages repeated trials, so that EV is the average profit per decision, assuming repeated trials are made. The power of the EV concept is that it allows (1) a high-risk prospect with large reserves potential to be compared with a low-risk prospect having small reserves potential; and (2) an excessively risky prospect (one with a negative EV) to be identified and avoided.

Faced with choosing among several options, the decision rule is to select the option having the highest EV. Obviously, when operators choose to participate in ventures having negative expected values, they are gambling—“in effect, betting against the house.”

In exploratory ventures, the cost of failure usually includes dry-hole cost, cost for lease bonuses of the condemned leases, and some geological and geophysical (G&G) costs. For development ventures, there may also be substantial additional capital investments plus expense items that will have to be written off— expenditures that were needed in order to determine the viability of the project, such as several confirmation or delineation wells, equipment, materials, and supplies. Newendorp (1975) presents the subject of expected value very thoroughly for the reader who wishes additional background.

Requirements for a Corporate System to Estimate Geologic Chance

In order to calculate a prospect's expected value, we must have a basis for estimating the chance that nature has provided a detectable HC-accumulation in the objective section under the drilling location. Geology and geophysics provide that basis. Moreover, for most companies, many such prospects will be proposed annually, from many different basins, each competing for precious corporate capital. In order to construct a consistent system for evaluating all prospects equitably, 10 requirements must be met:

  • The system must be geologically sound, so that all geologic aspects of oil generation, migration, reservoir emplacement, containment, preservation, and geotechnical detection are considered;

  • The system must be readily usable by many different geotechnical professionals;

  • The system must apply equally well in all geologic provinces;

  • The system must apply equally well to all types of petroleum traps—structural, stratigraphic, combination, basin-centered (generational), hydrodynamic, etc.;

  • The system must apply just as well to exploration plays as it does to the individual prospects that constitute the play;

  • Chance, expressed probabilistically as geologic confidence, must be expressed numerically, not subjectively;

  • The system must relate to reality (i.e., calibration) by periodic comparisons of portfolio outcomes (actual success-ratio) against forecasts (predicted success ratio);

  • The geologic components of chance must be independent of one another, or, if dependency is suspected, its influence must be understood and estimated;

  • The system must be independent of economic requirements such as minimum required reserves and flow rates—that is, the system must work just as well in a mature onshore province, such as the Permian Basin of west Texas, as it does in an economically demanding province such as the North Sea; and

  • All prospectors utilizing the system must be trained so they understand its basis and application.

Figure 17

Exploration failure and exploration success (economic, commercial, and geologic).

Figure 17

Exploration failure and exploration success (economic, commercial, and geologic).

Recorded Success Rates versus Geologic Success Estimates

Geologic success (Pg) is not necessarily the same as commercial success (Pc) or even economic success (Pe). The well-known phrase “geologic success but economic failure” refers to this issue—there are different definitions of success. According to official exploration drilling statistics, such as those reported annually by state and national petroleum agencies, the conventional definition of “success” means simply that the subject well was completed and did produce some hydrocarbons. This does not mean that the venture made a profit! In fact, such standard definitions of “success” contain five possible “successful” outcomes (Figure 17):

  • The well was completed as the discovery well for a field in which average wells will generate sufficient production revenues to recover the cost to drill, complete, and operate them (as well as the sunk costs to find the field), plus a reasonable profit. This is an economic success, on a full-cycle basis, represented as PV > 0 at the firm's discount rate.

    There are three possible outcomes comprising commercial success or completion success, wherein the exploratory well was completed but was not profitable on a full-cycle economic basis:

  • The well was completed because anticipated future production revenues will return a profit on the cost of completing and operating it, but not on the costs of exploratory drilling, leasing, and seismic, which are thus viewed as sunk and not recoverable (Capen, 1991). Such a well is an incremental success. Ordinarily, no more wells would be drilled on such a prospect by the investor, assuming that subsequent events do not provide new encouragement to drill again.

  • The well was completed as either an incremental success or an apparent economic success, but subsequent performance was inadequate even to recover completion and operating costs, resulting in early abandonment; completion of such a well was clearly a mistake.

  • The well was completed only for business reasons; that is, to hold a lease position or to satisfy a con-tractual or regulatory obligation. Some production revenues will be recovered, but perhaps not even enough to cover completion and operating costs.

  • The fifth outcome we call geologic success, meaning that a reservoired accumulation was found that was at least large enough to support a flowing test. For most onshore provinces any well that flows is likely to be completed, but many such small reservoirs encountered offshore are often reported only as shows.

The geologic chance factors (Rose, 1992a, 1995) are defined (pp. 34-36, 38) so as to exclude an onshore well that discovers a petroleum accumulation too small to warrant the expense of completing and operating it (Figure 17). Such very small accumulations are com-monly recognized only as shows. Practically speaking, we have eliminated this class by introducing concepts of minimum dimensionality or volumetrics into the definitions of the geologic chance factors, using the U.S. onshore as an effective minimum standard. This accomplishes three important purposes:

  • It allows the geologic chance factors to yield a product that corresponds effectively with the world's most liberal definition of chance of com-pletion success, thus permitting the comparison of conventional reporting standards of success with independently derived geologic estimates of probability of success;

  • It provides a general and basic standard against which subsequent adjustments can be made for exploration projects having more demanding economic requirements, such as deep overpres-sured tests, offshore prospects, remote frontier ventures, or international contract areas with severe financial obligations; and

  • It ensures approximate compatibility between onshore completion success and minimum detectable reserves volumes, assuming that minimum but finite dimensions are required for area of accumulation, average net pay, and HC-recovery factor. The concept here is that some small but finite volume and minimum reservoir quality must exist for an accumulation even to be detected by an operator. In other words, the lower limit of an accumulation, thus defined, is substantially larger than one barrel of oil! Two independent lines of evidence suggest that appropriate minimum reserves values are generally quite small (consistent with trap geometry): Drilling experience indicates that reserves volumes in the range of perhaps 1,000 to 20,000 BOE are generally capable of being detected and sustaining flow into the borehole, at least for a few hours. Also, P99% values in FSDs in mature onshore trends and basins are commonly around 1,000 to 10,000 BOE (see p. 30 and Appendix C).

Chance of Success Equals “Flowability”

As previously stated, it is important to recognize that the conventional reporting standard for exploratory success in most petroleum-producing nations is not whether the exploratory well discovered a commercial new field—rather, it is simply whether the exploratory well was completed for production. The beauty of such a reporting standard is that it is unequivocal, and results of all wells ordinarily become matters of public record. In the case of expendable exploratory wells, most governmental agencies require the operator, within a reasonable time, to declare whether or not the well encountered commercial hydrocarbons.

Moreover, different companies have different criteria for commercial success, thus rendering profitability to be an inconsistent criterion among exploratory wells drilled by many different operators. So, in order to compare geologic chance of success estimates with actual outcomes by all companies, we have set up the geologic chance system to be consistent with the chance of discovering enough oil or gas to complete the well for production in the case of an onshore, mature petroleum province (Rose, 1992a). A practical proxy for this case is “encountering enough reser-voired oil or gas to sustain flow.” This criterion is widely used by most exploration firms as the basis for exploration success because it is independent of variable economic requirements of different trends or basins.

Geologic Components of Prospect Chance of Success

Requirements for a Hydrocarbon Accumulation

Petroleum geologists generally agree that for a subsurface accumulation of HCs to exist, there must be porous and permeable reservoir rock, HCs that have moved from a petroleum source rock to the reservoir rock, and a sealed closure or trap capable of containing the HCs (Landes, 1951; Dott and Reynolds, 1969). All three of these general requirements must be met for a HC accumulation to form; if any one of these requirements fail, no accumulation will be present. This paradigm becomes the fundamental basis for employing geologic chance factors in estimating probability of geologic success (Table 8). In its simplest form, each of the three geologic chance factors is treated as an independent variable having a probability ranging from 0 to 1.0.

Serial multiplication of all three factors produces a decimal fraction equivalent to the probability that a Hc-accumulation is present, which is the probability of geologic success, or Pg, 0.28 in this example. By subtracting Pg from 1.0, we get its derived counterpart, the probability of geologic failure, or Pf. These expressions, along with estimates of prospect value and failure cost, are needed to calculate the expected value of the venture.

Explorationists commonly ask whether one geologic chance factor is more important than another, thereby deserving more weight. The answer is definitely no. The chance factors should be thought of as links in a chain: if any link breaks, the chain fails (p. 62). By analogy, if any one of the geologic factors is zero, the prospect will be dry! In other words, all of the essential geologic chance factors must coincide in space and time if one or more reservoired HCaccumulations can occur. Coincidence is discussed further on pages 48 and 80.

Many different schemes and combinations of geologic chance factors have been proposed and utilized by petroleum explorationists, involving as many as 14 different geologic chance factors. Most companies today employ four or five critical chance factors, sometimes with subfactors assigned to each main chance factor. In most systems, however, the principle is the same: based on the geologic evidence, geoscientists are asked to estimate their confidence, expressed as decimal fractions or percentages, in the existence of specific geologic conditions in the subsurface under the prospect area. Serial multiplication of the chance factors then yields the geologic chance of success for the prospect.

A few companies have used geotechnical chance factors to focus on the chance of failure, believing that better results may be obtained by focusing on the specific geologic conditions that could cause project failure.

Expressions of geologic confidence are subjective probability estimates, and they depend on three factors:

  1. reliability of evidence (direct, intermediate, indirect);

  2. judgment about adequacy relative to the P99% reserves case, and to the P90% cases for area, average net pay thickness, and HC-recovery factor (more than adequate, adequate, less than adequate, inadequate); and

  3. professional experience in estimating chance factors.

Geologic Chance Factors— Recommended System

For all exploratory prospects, including extensions (= step-outs), deeper-pool or shallower-pool tests, and NFWs, explorationists should independently express their confidence, as probability, in five critical geologic attributes of the prospect. Each of the five geologic chance factors has several subcomponents that must be considered in arriving at a confidence estimate for the chance factor. Because most of the subcomponents are partially dependent, we recommend use of the “weak link” approach, in which the lowest probability assigned to the several subcomponents within one chance factor is used as the probability for the parent chance factor (Rose, 1996a).

Table 8

Example calculation of simplistic probability of geologic success and failure using three geologic chance factors.

Geologic Chance FactorProbability
Reservoir Rock0.7
×Hydrocarbon Charge
0.8×
Sealed Closure0.5
Product = Probability of Geologic Success—Pg0.28
Probability of Geologic Failure—Pf = (1 − 0.28)0.72
Geologic Chance FactorProbability
Reservoir Rock0.7
×Hydrocarbon Charge
0.8×
Sealed Closure0.5
Product = Probability of Geologic Success—Pg0.28
Probability of Geologic Failure—Pf = (1 − 0.28)0.72

Hydrocarbon Source Rocks

First, we assess the probability (= confidence) that thermally mature HC source rocks are present in adequate thickness, extent, organic richness, and type to provide at least a modicum of HC-charge to the prospect area. Components that must be considered are:

  • quantity (thickness, extent, organic richness),

  • HC type (oil, natural gas, mixed), and

  • thermal maturity.

Discussion In many frontier basins, confidence in HC generation may be relatively low, and therefore it is often one of the most important requirements that must be met if exploration is to proceed. In productive basins and established trends, however, confidence in this chance factor tends to be considerably higher. HC volumes generated must be at least large enough to satisfy the P99% reserves value of the prospect. For confirmation wells, step-outs, and development wells, this requirement has ordinarily been met.

Migration

Second, we evaluate the probability (= confidence) that HCs have migrated, utilizing conduits, carrier beds, fractures and/or faults, following migration pathways into the location of the existing closure(s), in volumes of oil and/or natural gas sufficient to charge such clo-sure(s). Components to consider are:

  • conduits (carrier beds or zones, fractures or faults),

  • migration routes from “kitchen” to prospect area,

  • efficiency (concentration during transmission vs. dispersal), and

  • timing (sealed closures existed when migration occurred).

Discussion Again, this chance factor may be an important uncertainty in frontier exploration, but it is of much less concern in known productive basins and trends, and it may be entirely satisfied in cases of confirmation wells, step-outs, and development wells. For NFWs, explorationists should consider whether migration was dispersed, or concentrated by “structural focus,” whether it was primarily vertical or lateral, and the degree to which migration has been impeded by subsurface geologic barriers (Demaison and Huizinga, 1994). Efficiency of migration must be at least adequate to provide the P99% reserves volume required for the prospect. Finally, was timing correct; that is, did sealed closures (= traps) exist at the time migration occurred (“critical moment” of Magoon and Dow, 1994)?

Reservoir Rock

The third geologic chance factor is the probability (= confidence) that reservoir rock is present in the prospect area in sufficient volume, porosity, and deliv-erability to support one or more flowing wells. Reservoir thickness and quality in the exploratory well bore must be consistent with the P90% forecast for prospect average net pay and HC-recovery factor. Essential considerations are:

  • volume (thickness, extent),

  • porosity, and

  • reservoir performance (permeability, drive mechanism).

Discussion Note that the reservoir chance components are set up such that some minimum threshold standards of volume, porosity, and deliverability must be met or exceeded—specifically these reservoir components must be adequate, on an independent basis, to allow detectable, sustained flow of reser-voired HCs into the borehole. Under this approach, encountering a water-bearing, reservoir-quality sandstone would not be a failure in the reservoir category, but rather, a failure in one of the other four categories, such as an unexpected structural low, an absence of HC-charge, or a leaky trap. However, the presence of a 1-ft tight siltstone where a 10-ft porous sandstone objective was forecast would constitute a failure in reservoir prediction. Reservoir thickness must be compatible with at least the P90% average net pay forecast for the prospect, and porosity and reservoir performance with the P90% HC-recovery factor fore-cast (pp. 24-25).

Closure

Fourth, we assess the probability (= confidence) that a structural and/or stratigraphic closure involving the reservoir objective, and of minimally adequate area (consistent with the P90% area forecast) and vertical relief, is present in the prospect area, and can be detected using current geotechnical means. Components to be considered are:

  • Closure exists and is of adequate area and vertical relief to contain a volume of reservoired HCs suf-ficient to support flow (given that reservoir rock is present), and

  • We have confidence in our ability to detect and delineate them using available geotechnological methods.

Discussion Closures can be of any type—structural, stratigraphic, diagenetic, hydrodynamic, or basin-centered. Additionally, the geoscientist is asked to estimate his or her confidence that such target closures can be detected and delineated using whatever technology is being applied. Seismic resolution, velocity conditions, and statics, as well as data quality, den-sity, and reliability, all must be considered, for prospects relying on seismic mapping. Closure confidence must be compatible with at least the P90% area estimate (pp. 24-25).

The closure chance factor is formulated to apply equally well to stratigraphic as well as to structural traps. In combination with the reservoir chance factor, it focuses on the geometry of the envisioned oil or gas accumulation and on the volumes of fluids necessary to sustain a production test or prudent drillstem test. Stated in this way, the scheme can apply to all types of traps.

Containment

The fifth geologic chance factor is the probability (= confidence) that containment has occurred—that effective sealing rocks are present adjacent to the reservoir, and that emplaced hydrocarbons have been preserved. Essential considerations are:

  • seal effectiveness (differential permeability, seal thickness, absence of open fractures),

  • preservation from subsequent spillage (fault leakage, later fracturing, breaching, tilt-and-spill, etc.), and

  • preservation from hydrocarbon degradation (biologic degradation, oxidation, thermal destruction).

Discussion Three issues are addressed here—first, the question of sealing capability between reservoir and top seals, seat seals, and lateral seals, whether formed by stratigraphic contrasts, diagenesis, or fault-gouge, and compatible with at least the P99% reserves forecast (pp. 24-25). The second issue concerns later geologic events that may have resulted in leakage and/or flushing of HCs from the trap. The third issue deals with degradation of reservoired HCs by biologic, chemical, or thermal agents. For most (but not all) confirmation wells and step-outs, most of the “containment” requirements have been met; for development wells, all have been met—otherwise there would not be a field to develop!

Subjective Probability Estimates in Exploration

Expert judgments about the probability of discovery of any drilling prospect are classic examples of subjective probability estimates, which some geoscien-tists resist or even reject (Rose, 1992a); they claim it is just guessing. The record argues otherwise. Given a logical procedure, knowledgeable explorationists can generate such estimates with surprising consistency, agreeing not only on discovery probability but also on the relative certainty or uncertainty of the several geologic chance factors in a given prospect.

Geoscientists have many reasons for their reluctance to estimate the chance of success. One has to do with the traditions of geology as an observational and descriptive natural science, not a predictive and quantified one. A second reason is that many professionals have never been trained in techniques of subjective probability estimating, nor have they been encouraged to examine the accuracy of their prior predictions. Third, a significant chance always exists that such predictions may turn out to be wrong, and our cultural and corporate values often associate scientific error with mediocrity or even moral turpitude, thus generating criticism, guilt, and loss of status. Yet a fourth reason is that geoscientists may not wish to acknowledge the high degree of uncertainty involved in petroleum exploration, still preferring to believe that the secret to exploration success lies almost entirely in geotechnical skill and effort. Finally, we hear the very common excuse: “We don't have enough data to make an estimate.” Unfortunately, explorationists never have “enough” information—this is inherent in the business! Moreover, the more uncertainty that attends a given prospect, the more a systematic expression of subjective probability is needed—not less! All that can be reasonably requested is the best objective estimates possible, given the time, skill, and budget available. This is indeed the professional explorationist's responsibility.

Practical Aspects of Implementation

No substitute exists for actual experience in assessing and estimating confidence (= probability) in the various geologic chance factors, and comparing forecasts with outcomes. Figure 18 relates various subjective phrases used by the writer in relation to a complete probability scale, which may help the novice get started. One point should be emphasized here: the use of probabilities of 1.0 for any geologic chance factor involving a NFW prospect in new trends or basins should be reserved for those cases where the positive evidence is overwhelming. Often the focused question—“What could hurt me regarding this chance factor?”—can illuminate problems that otherwise might be overlooked.

Figure 19 is a matrix widely used among many companies, whose origin is not known to the writer. Assuming that a geological model, or concept, is recognized, it compares (1) quantity and quality of information against (2) what the information is signifying with respect to at least minimal adequacy about the particular geologic chance factor. A partial consequence of Figure 19 is that in order for us to render judgments of high confidence, either encouraging or discouraging, we require considerable data of good quality. Conversely, sparse or poor-quality data frequently allow only intermediate confidence statements. Many exploration teams find this matrix to be quite helpful in consistently assigning confidence values to various geologic components of chance pertaining to their prospects. Above all, geoscientists should recognize that the absence of information does not, by itself, imply a negative outcome—only that there are no data.

Figure 18

Subjective expressions of confidence.

Figure 18

Subjective expressions of confidence.

For development wells and most extension and infield exploratory ventures, the HC source rock, petroleum migration, and trap/seal factors have ordinarily been met. Only reservoir adequacy and the structural aspect remain as serious unknowns.

Rigorous analysis and discussion of real prospects with a peer group are an effective method for acquiring confidence in estimating geologic chance factors and probability of discovery. This also helps explorationists standardize their definitions and procedures.

Naturally, accuracy of predictions on probability of discovery cannot be judged on a single prospect or even on two or three. Only the outcome of a program of exploratory ventures can provide a fair indication as to whether the assignment of discovery probability has been optimistic, objective, or pessimistic. Of course, if the program has involved many high-risk wells, a larger sample may be required.

However, results of even a relatively small program can be instructive with regard to correct identifi-cation of several geologic chance factors. You should review predrilling projections of dry holes to see whether those geologic chance factors identified as the high-risk factors (“critical risks”) did indeed correspond to the geologic reasons why the hole was dry (Rose, 1987). Although dry holes are an inevitable aspect of petroleum exploration, the capable professional will usually—but not always—find that he or she has correctly anticipated the real geologic reasons for failure. If not, this indicates that the geology of prospects is not adequately understood or identified. Again, the writer emphasizes that assessment of geo-technical performance regarding chance requires a completed program of wells, not a single well.

Explorationists can also gain experience by making estimates on wells drilled by competitors operating in the same trend or basin. Commonly, the geoscientist has considerable knowledge about the subsurface conditions attending such wells, which thus provides expanded opportunities for developing a useful experience base for predictive confidence.

Finally, it is difficult to overemphasize the power of independent, multiple judgments in assessing the geologic chance factors and the probability of discovery. Firms are well advised to obtain several different opinions and to combine them into a final estimate of the probability of project success.

Figure 19

Chance adequacy matrix.

Figure 19

Chance adequacy matrix.

Virtues of Geologic Chance Factors

Separation of the various components of geologic chance allows them to be analyzed more thoroughly and objectively, and leads us to better geologic understanding of the prospect. Also, the identification of the geologic chance factor having the lowest probability (= “critical risk”) helps exploration management and staff focus on those items of greatest uncertainty. For example, if closure is the critical risk element of a prospect, additional seismic or reprocessing may be warranted. Conversely, when geoscientists see that additional data acquisition can be curtailed because it is not likely to materially increase the chance associated with a given geologic factor, exploration becomes more cost effective and usually more timely. But most important, explorationists cannot really analyze, and thus improve, their performance in predicting the probability of discovery unless they systematically identify, forecast, and inspect the predictive results that attend the several geologic chance factors for a portfolio of exploratory ventures (Rose, 1987).

Geologic, Commercial, and Economic Chances of Success

The relationship between the prospect-reserves distribution, the geologic chance of success, and differing definitions of success was shown on Figure 17. Remember that the geologic chance factors (especially reservoir and containment) are purposefully defined to be consistent with the presence of a HC accumulation of at least minimum reserve size (or greater)—“at least enough reservoired mobile oil to sustain flow.” Such a definition is equally operative in a mature, inexpensive onshore province such as west Texas, and in an expensive offshore province such as the North Sea.

In a mature, inexpensive onshore province, any exploratory well that flows will probably be completed for production, either as an economic success, recovering at least all investments on a full-cycle basis, at interest (or more), or as a commercial success, which may only return a profitable investment in tubing, tank battery, and completion. In such theaters, geologic success (Pg) and commercial success (Pc) may well coincide. But the same onshore commercial subsurface accumulation encountered offshore North Sea would be described only as a show. A very much larger volume of reservoired HCs would be required there to interest the operator in installing an expensive platform and developing the field discovered by the well—enough reservoired, producible oil to cover all exploratory costs, plus platform construction and full development and operation of the new field, as well as a satisfactory return on the total investment. That amount of discovered oil would qualify as an economic success (Pe). If a lesser volume of recoverable oil is discovered, enough to warrant platform installation and development drilling but not enough to recover all exploration costs, then that smaller discovery would qualify as a commercial success (Pc). Any exploratory well that is completed for production therefore qualifies as a commercial success, “commerce” being the production and sale of oil and/or natural gas. Only some of these are economic successes, on a full-scale basis.

Estimating Chance of Commercial and/or Economic Success

Figure 20 represents the concept and process by which geoscientists may progress from chance of geologic success (Pg), to chance of commercial success (Pc), to chance of economic success (Pe).

Let's assume that your geotechnical staff estimate that Prospect Alpha has a 30% chance of geologic success (i.e., that the well will encounter reservoired flowing HCs). The Prospect Alpha cumulative reserve distribution is represented by the solid black sloping line, which has been truncated above P1%. Reserve parameters are: 

formula

As the manager, you have asked the staff to estimate the chance that Prospect Alpha will discover enough oil (or more) to justify completing it for pro-duction—the chance of a commercial success.

Your engineering and economic departments estimate that, in this location, at least 0.95 million barrels (MM bbl) must be present to warrant a completion; this is the minimum commercial field size (MCFS).

Figure 20

Calculating chance of commercial success and economic success.

Figure 20

Calculating chance of commercial success and economic success.

On the reserves distribution, 0.95 MM bbl occurs at P80%, so 20% of the distribution is smaller than the minimum reserves required, and 80% is larger. Therefore the chance of commercial success (Pc) is: Pc= 0.8 × 0.3 = 0.24.

What is the mean commercial reserve size? Remember that by truncating the original reserves distribution at P80%, your staff deleted the lowest reserves values from the original distribution. That is, a discovered accumulation smaller than 0.95MM bbl will be a dry hole. Furthermore, the surviving distribution is no longer lognormal because the lower part, containing the mode and smaller outcomes, has been removed. What survives has a form that resembles an exponential distribution if plotted in the frequency domain. The mean of the commercial reserves distribution must now be recalculated. The new distribution, from P80% to P1%, now has a new set of reserves parameters; note that in moving from geologic chance to commercial chance, chance of success has decreased (30% -— 24%), whereas the mean of the commercial part of the reserves distribution has increased (20MM ⊒ 24MM): 

formula

Next, the Exploration vice president asks you what the chance is that Prospect Alpha will be an economic discovery—that is, that the project will find enough oil to earn at least the corporate minimum return on investment on a full-cycle basis and pay for all exploration, development, and operating costs. Again your engineers and economists study the problem and advise that an economic discovery will require at least 7.5MM bbl (= MEFS). On the original prospect, reserves distribution, 7.5MM bbl occurs at P40%, so 60% of the distribution is uneconomic, and 40% is economic. Therefore, the chance of an economic success is: 

formula

Again, you must recalculate the mean size of the economic-reserves distribution because you have deleted the smaller, uneconomic accumulations from the parent-reserves distribution. The surviving distribution, from P40% to P1%, now has a new set of reserves parameters: 

formula

So, as the chance of success goes down to reflect decreasing chances of finding larger accumulations of a commercial or economic scale, the commercial and/or economic mean reserves values increase because the noncommercial or uneconomic segments have been deleted through truncation.

Understanding Truncation

Geoscientists and engineers seem to struggle with the “Truncation Problem.” The procedure described previously is useful because it models the actual decision-behavior involved, after an exploratory well has made a discovery, and management selects the logical next steps.

Implicit in this process is the “sunk-cost” concept— that completion decisions must be made on a “point-forward” basis because funds already invested in seismic, leasing, and exploratory drilling are by then gone—sunk. “Can we make a reasonable profit on our investments in pipe, stimulation, tank-battery, and lateral lines in order to bring the well into production?” For offshore projects, such investments may also include platform cost as well, which is usually so large that the “commercial” threshold may well approach the “economic” threshold. However, in mature producing provinces, like the Permian Basin of west Texas, or the Canadian plains, most wells that flow are likely to be completed, so the “commercial” threshold is likely to be at or just above the “geologic chance of success,” and may lie substantially below the “economic” threshold. Obviously no rational investor will continue to drill a series of development wells that are consistently commercial (rather than economic) successes.

Some people are tempted to truncate all projects at the economic threshold, rather than the commercial threshold. This conservative position tends to prevent participation in many projects that could well be profitable. The criterion should be that the average (= mean) of the entire reserves distribution above the commercial cutoff is at least economic. For such a prospect, then, we accept that the outcome may turn out to be a small discovery that is commercial but not economic, but the average of all outcomes is clearly economic; that is, profitable on a full-cycle economic basis. The project's NPV is greater than 0 at the firm's mandated discount rate.

Each firm should determine its own consistent guidelines to be used in predicting the “commercial” cutoff for an exploratory venture. The “commercial” cutoff is not a precise value, and there is no doubt that producing rate must be considered as well as minimum reserves. Many experienced operators correctly believe that the risk associated with the decision to complete marginal commercial wells is the single most underestimated, potentially dangerous risk involved after exploration has made a discovery. Much care is warranted here.

One-step versus Two-step Method for Estimating a Prospect's Chance of Success

Two methods have commonly been used by modern oil companies to estimate a prospect's chance of success.

The One-step Method is favored by White (1993) and assumes that:

  • Modern explorationists can reliably distinguish large reserves prospects from small reserves prospects, so the geotechnical exploration effort can indeed identify and drill only those prospects capable of containing some designated “significant” reserve size or larger;

  • By setting “significant” limits of closure areas, average net pay, and HC-recovery factor, explorers can screen out unprospective parts of mapped trends; and

  • Prospectors can reliably link “adequacy” of source rocks, migration, closure volumes, and seals to the chance of success. Thus, for chance factors such as source rocks or migration to qualify, the geologist must judge whether they were adequate to provide enough oil to a prospect for it to meet or exceed that reserves quantity deemed to be significant.

Advantages of this method are that explorers don't have to worry about the low-reserves ends of FSDs. Estimating chance of success is a one-step process, keyed to finding at least a field of significant (= eco-nomic? commercial?) reserves size or larger.

The Two-step Method is favored by Rose (1995) and is necessitated because all three of White's key assumptions require geologic resolution beyond modern technical capability (i.e., false precision). Instead, in the two-step method Pg is keyed to a very small (“flow-able”) reserves discovery (= P99%) consistent with doc-umented industry performance. Then, in a second step, the part of the prospect-reserves distribution that is considered to be commercial (the percent of the distribution that is greater than the commercial reserves threshold) is determined and then multiplied by the Pg. As a result, estimation of Pc (or Pe) is a two-step process:

  • Use geologic chance factors to find the chance of a small (“flowable”) reservoired accumulation (or larger); then

  • Estimate what percent of the prospect-reserves distribution is of commercial or economic size and multiply the geologic chance by this commercial or economic chance.

Advantages of this method are that the two-step method reflects real industry performance capabilities— there is no false precision. Setting reserves thresholds very low maximizes the likelihood of correctly capturing reality. It allows a company-consistent process to be applied to widely varying economic thresholds in different operating theaters, thus promoting valid ranking and capital allocations. Finally, this method allows reality checks by comparing a consistent geologic chance estimate—usable in all geologic theaters—to a compatible universal public reporting standard. (Most modern companies have now endorsed the two-step method.)

Independent versus Dependent Chance Factors

Many exploratory prospects seek to evaluate a single main reservoir objective on a projected closure. For most such ventures, all the geologic chance factors may be treated as if they are mutually independent. Although there are a few cases where some dependency (either positive or negative) does appear to exist, especially between reservoir presence and structural closure or between seal effectiveness and structural closure, most knowledgeable risk analysts agree that the geologic chance factors, as defined, do not present any serious problems with regard to dependency. However, dependency does become a potential problem in dealing with multiple-objective exploratory prospects or with several prospects contained within one exploration play. Dependency in play analysis is discussed in Chapter 5.

Multiple-objective Prospects

Prospects having multiple objectives tend to be more attractive than one-objective prospects because the combined chance for at least one of the objectives to be productive is higher (see Appendix D). However, a prospect having two objectives will not be twice as likely to be successful as a one-objective prospect—nor does its probability of discovery equal the sum of the probabilities of discovery of the two multiple-objective targets. If all the geologic chance factors for the two objectives are independent of one another, then P(dis-covery) and P(failure) can be derived by binomial expansion as shown in Appendix D.

In most such cases, however, some geologic chance factors are independent and some are dependent (i.e., they are common to both objectives of the prospect). In such cases, Pg and Pf are derived via another two-step process, as shown in Appendix D. Discovery probability (and often EV) for a multiple-objective prospect having independent geologic chance factors is higher than for a multiple-objective prospect having one or more dependent geologic chance factors. Thus, the effect of dependent geologic chance factors is to reduce prospect discovery probability. Dependent geologic chance factors (i.e., common to both objectives of the multiple-objective prospect) most often include the structural aspect and the HC-charge and migration aspects. The reservoir and containment aspects tend to be independent. However, exceptions to this pattern are not uncommon.

An additional complication sometimes occurs when a given geologic chance factor contains several subfac-tors, some independent and others dependent. The task here is to assign relative fractional weights to the subfac-tors so that their product equals the probability of the parent chance factor. Appendix D shows such a case. In general, dependency among the subfactors is common; that is the primary justification for the “weak-link” approach in establishing confidence levels (see p. 34).

Although having legitimate multiple objectives usually makes a prospect somewhat more attractive economically, aggressive explorationists often are tempted to include secondary objectives that are only of marginal value. The improvement to the prospect's value is illusory, however, for several reasons:

  • For most multiple-objective prospects, several of the geologic chance factors are in fact dependent, thus reducing the apparent chance of success significantly from the independent case.

  • The secondary objective(s) commonly represents only a marginal or incremental completion that may pay for pipe, maintenance, and some part of the drilling cost while not adding substantial new reserves. Such ventures do not add much value and therefore should not be actively sought;

  • Dual-objective discoveries are often completed and produced first in the deeper zone until depletion, then completed in the upper zone. Because of the time value of money, this delayed production often reduces the PV of the upper productive zone significantly; and

  • Dual-objective completions are relatively more expensive and require higher maintenance than conventional completions.

For all of these reasons, a multiple-objective prospect having a high chance of success should be viewed with caution. The most attractive multizone prospects are those in which both or most zones clearly stand alone economically, and production from both zones may be commingled.

Nongeologic Aspects of Success and Failure

Technical and Mechanical Effects

Many variables other than geologic chance factors affect exploration success. For example, firms that use state-of-the-art technology seem to have rates of success much greater than firms that drill without benefit of such advanced geotechnical guidance; this should be taken into account.

Also, mechanical chance factors should be considered, such as the chance of not getting the well down to the objective, the chance of incorrectly locating the well, and the chance of geotechnical errors in mapping, logging, or testing. However, if you anticipate that the well will be redrilled if such difficulties occur, you should make probabilistic provision for such trouble costs in the cash-flow schedule for the project rather than as a chance-of-success factor. Generally, such considerations do not make a substantial difference except in economically marginal prospects.

Some authorities suggest including a chance factor that deals with the probability that the exploratory well has been located and evaluated properly (Baker, 1988). It is certainly true that significant fields are occasionally discovered (or recognized) only after several apparently unsuccessful penetrations. However, the writer's experience is that, for most prospects, this aspect can be covered within the closure or reservoir rock categories; that is, if the well turns out to be improperly located, it is commonly perceived as a failure to adequately assess structural or reservoir risk. However, when dealing with frontier basins and plays, it may be advisable to include a separate chance factor assessing the confidence that the well will be located properly and any productive reservoir zones will be identified and evaluated adequately. Careful review of the stratigraphic column and consideration of active petroleum systems will reduce the chance of overlooking a productive zone while drilling. In other words, “serendipity” may be a euphemism for less-than-thorough consideration of all stratigraphic possibilities and thus for incomplete risk analysis.

Expressing Business and Political Risks

In assessing major projects that require large front-end investments or long elapsed time between expenditure and payout, the firm may wish to appraise the likelihood of a severe and extended drop in wellhead prices (or rise in operating costs or taxes). Basically, the procedure here is to identify what sustained low price levels—or elevated costs—would cause termination of the project, then try to obtain estimates from knowledgeable petroleum economists about the probability and timing of such occurrences. The chance of commercial success (Pc) is then multiplied by (1- the chance of such economic failure). Less severe price and cost fluctuations should simply be considered as variant cases within the project cash-flow model.

Political uncertainty can be expressed similarly. Again, knowledgeable, objective political experts should express their opinions about the likelihood of a change of regulation, law, or regime severe enough to cause a project's termination or change its economic status. Commonly, this probability should be directed to the anticipated time of greatest vulnerability—after large capital investments for development of discovered fields but before recovery of those investments via production revenues—which can then be related to the cashflow model. The chance of commercial success (Pc) can then be multiplied by (1- the chance of political failure).

Monitoring and Improving Predictive Performance

If the exploration organization is serious about improving staff performance in estimating the chance of success, management themselves must undertake equally serious procedural changes.

Universal Prospect Risking Scheme

Management must insist on adoption and utilization of a consistent geologic risking system that meets the requirements outlined earlier. Geotechnical staff must be trained in its use; periodic management audits and (where necessary) retraining will ensure its consistent, universal application.

Keeping Records

Management should set up procedures whereby forecasts of geologic chance of success (and individual chance factors) are preserved and routinely compared against actual outcomes of exploratory wells. Individual geoscientists should be encouraged to make geo-logic predictions on competitor wells and to compare them with announced results, as a way to expand the sample size of their predictive experience and reduce the time necessary to begin monitoring, measuring, and improving predictions of chance.

Predicted versus Actual Success Rates

Compare the average predicted chance of success for last year's portfolio with the actual success rate (Rose, 1987). Is there bias? Are some exploration teams biased more than others? Why? Separate all prospects into three groups—high, medium, and low risk. Then see what the average success rate was of each group—are staff distinguishing high-risk prospects from low-risk ones?

Performance Tracking

The methodology of Clapp and Stibolt (1991) (p. 26 and Figure 14) for continuously monitoring reserves additions throughout the annual exploration program may also be applied to chance-of-success forecasts. A P10%-P90% “envelope” of expectations is created, employing Monte Carlo simulation, for the range of numbers of discoveries (Figure 14c and d). Both the mean and P50% trends may be projected within the P10%-P90% envelope. Then as the exploratory wells of the annual portfolio are drilled, the numbers of actual discoveries are plotted as a line rising from the start of the program toward the completion of the portfolio, and are compared with expectations. Any time the actual performance moves out of the P10%-P90% envelope, management is justified in suspecting professional bias affecting the estimates of chance, and in calling for immediate corrective measures. Although performance tracking is an excellent tool for monitoring geotechnical performance, it does not indicate why predictions may be biased. For that, we must turn to dry-hole analysis.

Dry-hole Analysis

Professional staff charged with monitoring and improving geotechnical productive performances should collect and analyze all unsuccessful exploratory efforts. Why were the dry holes dry? Was critical risk (i.e., the geologic chance factor having lowest probability) correctly identified for most dry holes? That is, were most dry holes caused by failure for that chance factor to be satisfied? One U.S. company found that they were correctly identifying reservoir risk but were not recognizing structural risk (Rose, 1987), even though structural errors caused 43% of the company's dry holes—more than any other geologic chance factor (Figure 21). Incorrect predictions of reservoir rock presence were responsible for 40% of the company's exploratory dry holes, but reservoir rock was correctly anticipated as the critical risk in 80% of those dry holes. The same company also found that they were consistently conservative in their chance-of-success estimates (20% predicted vs. 31% actual, over two consecutive years) mostly because their chance estimates of HC-charge were pessimistic (60% predicted vs. 95% actual). In retrospect such conservatism was especially difficult to justify considering that almost all of the company's exploration efforts took place in established petroleum-producing U.S. basins! If this company had properly assessed the chance of HC-charge, its predicted success rates would have matched actual results very closely.

Previously (p. 38) it was emphasized that better estimates of a prospect's chance of success result when the geologic components of chance are analyzed separately. A second advantage of this method is that tech-nology can then be focused on the critical risk—for example, if closure is the greatest risk, additional seismic data may be a cost-effective way to reduce risk. Based on Rose's (1987) experience, a third advantage is apparent: this approach then allows geotechnical staff to improve their predictive performance by identifying and correcting prevalent patterns of error. Amoco (McMaster, 1998), Santos (Johns et al., 1998), and Unocal (Alexander and Lohr, 1998) report analogous learning from their results.

Figure 21

Relative frequency of four geologic chance factors causing dry holes during a company's 1977-1978 exploration programs, plus performance by company geologists in correctly anticipating which geologic chance factor indeed represented greatest technical risk (from Rose, 1987).

Figure 21

Relative frequency of four geologic chance factors causing dry holes during a company's 1977-1978 exploration programs, plus performance by company geologists in correctly anticipating which geologic chance factor indeed represented greatest technical risk (from Rose, 1987).

Figure 22

Global discovery percentages through time (excludes the U.S. and Canada).

Figure 22

Global discovery percentages through time (excludes the U.S. and Canada).

Figure 23

Success rates change with maturity—annual and cumulative success rates compared with annual drilling, Niagaran Reef Trend, Benzie and Manistee Counties, Northern Michigan (Rose, 1992a).

Figure 23

Success rates change with maturity—annual and cumulative success rates compared with annual drilling, Niagaran Reef Trend, Benzie and Manistee Counties, Northern Michigan (Rose, 1992a).

Industry Experience in Estimating Prospect Chance of Success

Actual Industry Performance

In international theaters, NFW success rates since about 1960 have been remarkably consistent at about 25% overall (Figure 22). U.S. exploration success rates during the 1980s (before 3-D seismic) for all exploratory wells were commensurate (20-30%), but showed considerable variation among different classes (Table 9). In particular, annual success rates for U.S. NFWs (onshore and offshore) ranged from 13% to 18% during the 1980s.

Impact of 3-D Seismic Data

In those geologic provinces in which 3-D seismic data collection is feasible, discriminating, and cost-effective, it improves exploration performance in three different ways, especially where DHI technology is also incorporated.

  1. First, by improving their prospect chance-of-suc-cess estimates, explorers can be more selective and thus improve their exploration success rates. This is accomplished through clearer resolution of geologic structure (which leads to improved location of exploratory test wells) and better discrimination of reservoir rock and sealing rock distributions. Also, positive DHI indicators allow enhanced confidence regarding the presence of reservoired HCs in the traps. During the 1990s, provinces such as the North Sea and deep-water Gulf of Mexico report prevailing NFW exploration success rates of 30% or more.

  2. Second, estimates of prospect reserves can be improved through improved resolution of geologic structure and reservoir thickness and extent. Also, DHI signals can reduce uncertainty (variance) on trap volume by indicating approximate position of oil/water or gas/water contacts.

  3. Third, prospect profitability may be improved through optimum location of exploration and development wells, which results in higher initial production rates, larger individual well ultimate recoveries, and fewer required development wells. Nevertheless, 3-D seismic may not be feasible or cost-effective in many onshore provinces, especially in the early stages of exploration.

Table 9

Generalized success rates of various well classes drilled in the U.S. onshore and offshore during the 1980s as reported by the CSD (Rose, 1992a).

Well ClassPercentage Successful
Development wells75-80
All exploratory wells20-30
Extensions (outposts)40-45
In-field wildcats25-35
New-field wildcats13-18
Well ClassPercentage Successful
Development wells75-80
All exploratory wells20-30
Extensions (outposts)40-45
In-field wildcats25-35
New-field wildcats13-18

However, characterizing seismic anomalies as “DHIs” should be rigorous and calibrated against actual experience and should call for the presence of multiple attributes if they are to represent a discriminating basis for elevated Pg estimates, rather than just routine amplitude anomalies.

Characteristic Patterns of Predictive Bias in Estimating Chance of Success

Several international exploration organizations that have adopted companywide geotechnical risk analysis of all prospects report that, during the first year or two, a prevalent performance pattern emerged: for high-risk NFW prospects, geotechnical staff were overly optimistic and overestimated true chances of success (Otis and Schneidermann, 1997; Alexander and Lohr, 1998). This will be addressed further on page 47. For intermediate-risk NFWs, those in the 20-35% range, actual success rates were generally about right and matched predicted success rates fairly closely. For low-risk exploratory ventures, those in the 35-60% range, actual success rates were conser-vative—more of these ventures were successful than predicted. But for high-confidence ventures—those in the 60-90% range—actual results were notably lower. Apparently, prospectors tend to be overopti-mistic when identifying an exploration venture as a sure thing!

However, by maintaining (and circulating) records of predictions vs. outcomes, by making geotechnical professionals aware of the causes and consequences of predictive bias, and by constructively addressing the reasons for error and bias in specific prospects, we can improve forecasts of a prospect's chance of success, and reduce predictive bias if not eliminate it.

Using Trend or Basin Success Rates

In some cases, where geotechnical staff or management mistrust geologically derived prospect success estimates (or lack the geologic skill to reliably derive them), observed success rates (number of discoveries -τ total number of exploratory wells) have been used as a proxy. In some geologic settings, this procedure is acceptable, especially in “statistical plays” where reservoirs are lenticular and beyond predictive geot-echnical resolution. But in most trends, such observed success rates are a poor substitute for prospect-specific chance-of-success determination and tend to give mis-leading results that are either overly pessimistic or optimistic.

There are at least five important unknown aspects of dry-hole ratio trends:

  1. Were the concepts and geotechnical skills of those prior operators commensurate with yours?

  2. Was the quality of their data commensurate with yours? (This particularly applies to vintage and acquisition parameters of seismic data.)

  3. Remember that all wells drilled through the subject zone are counted as valid exploration tests of that horizon, even if they simply passed through the subject zone on the way down to a deeper exploration objective. Therefore, such wells may not have been legitimate exploratory tests of the subject zone.

  4. Why were the completed wells completed? Were the economic parameters of those operators similar to yours? Remember that not all completions are legitimate economic attempts, and a small company may complete a well that a large operator would abandon.

  5. Trend success rates tend to decrease with time (Figure 23); ignoring this pattern may lead to overly optimistic expectations from trend success rates, especially if your venture is in a fairly mature exploration theater.

Accordingly, caution is recommended in substituting observed trend success rates for geologically derived prospect estimates of the chance of success. Certainly they should not be ignored, but their best use is as a reality check after the responsible geoscientist has carefully reviewed and edited such data so that those ventures from which the edited trend success rate was calculated are truly comparable to the prospects in the trend you are exploring.

Historical Changes in Trend Success Rates

In most exploration theaters (basins, trends, and plays), wildcat success rates change through time. Suc-cess rates are characteristically high during the early phases of exploration while larger and more evident fields are being found, then decline as the industry searches for fields that are smaller and/or harder to find. Figure 23 shows actual data from a segment of the Niagaran (Silurian) Pinnacle Reef Trend of northern Michigan. Although the discovery probability should certainly be estimated based on the geotechni-cal characteristics of the prospect itself, the prudent explorationist will also consider the trend's state of exploration maturity in arriving at a final estimate of Pg, Pc, and Pe.

The Trouble with High-risk Exploration

During the 1980s, large, international corporate explorers, such as Shell, Amoco, and Mobil, experi-enced a disturbing result from their high-risk exploration ventures: within the category of all NFWs having a predicted chance of success of 10% or less, less than 1% of those ventures resulted in discoveries. Amoco (McMaster and Carragher, 1996) indicates that, since 1982, such high-risk exploration ventures have destroyed corporate value, not created it.

The same pattern was reported by all three companies independently, and the total number of wells in each company's sample was more than 200. These results cannot be ascribed to vagaries of random sam-pling—Amoco reports that the probability of such a result occurring by chance alone is only about 1%. Clearly, major company geoscientists and managers dealing with high-risk exploration ventures tend to be seriously overoptimistic in predicting chance of success, so such ventures are often overvalued as investment ventures to the detriment of corporate economic performance and the stockholder.

Causes

Some probable reasons for these difficulties have been summarized by Boccia (1996):

  1. Most large companies (the kind that carry out most high-risk, high-potential exploratory ventures) systematically favor such projects because they offer the potential for large, new reserve additions—big new fields that represent opportunities for long-lived, large-profit projects because of applications of advanced technology and efficiencies of scale. Such big companies, having large reserves bases and corresponding cash flows, are—without realizing it—actually exhibiting risk-prone behavior with respect to such exploration ventures. Their managers des-perately want to find such large, new fields, so they are prone to approve such ventures whenever they appear. Ambitious geotechnical staff are therefore prone to overestimate the value of such ventures in their efforts to respond to perceived management needs. This is especially true when exploration organizations have the wrong priorities—when they confuse the need to drill wells with the need to add value.

  2. Risk-analysis tools and methods for evaluating such high-risk, high-potential ventures are oper-ating at the extremes of the risk spectrum— reserves anticipated are very large, whereas chances of success are very small. Both theory and experience suggest that our ability to responsibly assess uncertain events begins to deteriorate at the extremes of either chance or magnitude. For example, most geotechnical professionals are better able to judge whether the chance of occurrence is 50% or 25% than they are at distinguishing between a 5% chance and a 10% chance. Because people are conservative processors of fallible information (Edwards, 1982), we tend to set the ranges of our predictive limits too narrow (Capen, 1976). Also, technologic risk-reduction in certain areas may actually encourage managers to take still higher risks elsewhere. Finally, Boccia suggests that the economics of higher-risk ventures are more sensitive to underrisking than are lower-risk projects. The relative negative impact of overoptimism becomes progressively more severe as the probability of success decreases— “It hurts more, on a per-dollar investment basis, to be wrong by the same degree when the risks are high. Our ability to clearly discern risk is the weakest for exactly those kinds of prospects where it needs to be sharpest.”

  3. In order to have an acceptable chance of making such large discoveries, many trials must be undertaken. Such high-risk, high-potential ventures are not easy to identify and are expensive to carry out. “In effect, high-risk exploration may be failing because companies are underestimating the time and money they need to commit to the high-risk game” (Boccia, 1996). Accordingly, such companies may have difficulty adequately diversifying high-risk ventures within their overall exploration portfolios.

  4. A fourth possible cause, not identified by Boccia, is more fundamental—explorationists and their managers have not come to grips with the reality that the remaining world endowment of undiscovered fields is getting smaller, simply because petroleum exploration has tended to find the giant fields preferentially (see Figure 1). Accordingly, exploration costs must be constrained, consistent with the potential profits of smaller fields. It is not that exploration for large fields must cease—rather, it cannot proceed in the same ways it did in the 1950s through the early 1980s. We must figure out how to be profitable while exploring for (and finding) smaller fields.

  5. Yet another probable cause has to do with inadequate geotechnical verification of coincidence (p. 34). All geologic chance elements must coincide in time and space if reservoired petroleum accumulations are to occur in a basin or trend. Especially in frontier areas, it is essential to map the areas where the various geologic chance elements are present (or probably present) and to restrict exploration to those areas of probable coincidence.

Remedies

What are some remedies to counter these problems?

  1. Geotechnical staff must be vigilant with respect to coincidence—all geologic chance factors must coincide in time and space in the area of the prospect (p. 81). In the case of petroleum generation and migration, their effects must coincide with the presence of reservoirs, closures, and seals in the prospect area.

  2. Actively employ reality checks—analogous experience from other basins, FSDs, and credibility of high-side projections (P1%-P10% range of values). Also, actively consider that all prospects may (unfortunately) result in only a small sub-economic field, and the prospect P99% and P90% forecasts should reflect this!

  3. Solicit independent, multiple estimates, by peer and expert review and/or exploration committee review. Look at such high-risk prospects very carefully, employing Petroleum System Analysis.

  4. Take on geotechnically proficient partners both as a risk-spreading measure and as a way to get independent confirmation of geotechnical and economic merit.

  5. Verify positive expected value and economic feasibility of such projects, assuming a lower chance of success and significantly lower prospect reserves—i.e., what is the economic sensitivity relative to reduced chance and reserves values?

Are some remedies to be avoided? Although geo-technical staff should be admonished to give special technical scrutiny to high-risk prospects, it is probably a mistake for management to select a “chance hurdle”— an arbitrary lower limit for a prospect's chance of success. For example, if a company announced that henceforth, no exploration venture with a chance of success less than 20% would be undertaken, it might only encourage geotechnical staff to find creative ways to elevate what might legitimately be a 15% prospect to one rated 20% or 25%. Also, geotechnical staff must be accountable, and any official prospect review committee should not have power to approve or condemn prospects presented to them. Otherwise, no one is accountable! The expected value concept is a useful screen here. If the prospect-reserves distribution and geologic-chance estimates are well documented and venture EV is strongly positive (even though a scrutinized credible chance of success is only 10%), then the prospect should probably be drilled.

3Elongated trap areas and reservoirs having low net-to-gross ratios will tend to be associated, of necessity, with larger P90% areas than equidimensional closures and high net-to-gross ratios, as described in Appendix C.
4A series of outcomes that collectively fall below the P90% boundary would have less than a 10% chance of being the result of random chance alone.

Figures & Tables

Figure 6

Reserves parameters for exploration prospects (deterministic).

Figure 6

Reserves parameters for exploration prospects (deterministic).

Figure 7

Area, average net pay, and HC-recovery factor are lognormal.

Figure 7

Area, average net pay, and HC-recovery factor are lognormal.

Figure 8

Distributions of estimated parameters for prospect reserves.

Figure 8

Distributions of estimated parameters for prospect reserves.

Figure 9

Productive field areas, East Texas, Capen.

Figure 9

Productive field areas, East Texas, Capen.

Figure 10

Graphs to derive geometry factor adjustment.

Figure 10

Graphs to derive geometry factor adjustment.

Figure 11

Area versus depth plot.

Figure 11

Area versus depth plot.

Figure 12

Graphical method for analytical solution for combining three lognormal distributions by multiplication.

Figure 12

Graphical method for analytical solution for combining three lognormal distributions by multiplication.

Figure 13

Comparison of predictions versus outcomes (from Rose, 1987).

Figure 13

Comparison of predictions versus outcomes (from Rose, 1987).

Figure 14

(a, b, c, d) Continuous tracking of organizational predictive performance (Copyright SPE #22038. From Clapp & Stibolt, 1991, reprinted with permission.)

Figure 14

(a, b, c, d) Continuous tracking of organizational predictive performance (Copyright SPE #22038. From Clapp & Stibolt, 1991, reprinted with permission.)

Figure 15

Pre- and postdrill discovery sizes.

Figure 15

Pre- and postdrill discovery sizes.

Figure 16

Volume Accuracy

Figure 16

Volume Accuracy

Figure 17

Exploration failure and exploration success (economic, commercial, and geologic).

Figure 17

Exploration failure and exploration success (economic, commercial, and geologic).

Figure 18

Subjective expressions of confidence.

Figure 18

Subjective expressions of confidence.

Figure 19

Chance adequacy matrix.

Figure 19

Chance adequacy matrix.

Figure 20

Calculating chance of commercial success and economic success.

Figure 20

Calculating chance of commercial success and economic success.

Figure 21

Relative frequency of four geologic chance factors causing dry holes during a company's 1977-1978 exploration programs, plus performance by company geologists in correctly anticipating which geologic chance factor indeed represented greatest technical risk (from Rose, 1987).

Figure 21

Relative frequency of four geologic chance factors causing dry holes during a company's 1977-1978 exploration programs, plus performance by company geologists in correctly anticipating which geologic chance factor indeed represented greatest technical risk (from Rose, 1987).

Figure 22

Global discovery percentages through time (excludes the U.S. and Canada).

Figure 22

Global discovery percentages through time (excludes the U.S. and Canada).

Figure 23

Success rates change with maturity—annual and cumulative success rates compared with annual drilling, Niagaran Reef Trend, Benzie and Manistee Counties, Northern Michigan (Rose, 1992a).

Figure 23

Success rates change with maturity—annual and cumulative success rates compared with annual drilling, Niagaran Reef Trend, Benzie and Manistee Counties, Northern Michigan (Rose, 1992a).

Table 4

Table 4 Reality checks (1): characteristics of the endpoints of the reserves distribution.

Table 5

Getting a sense of scale about HC-recovery factor.

Table 6

Reality checks (2): characteristic ranges associated with oil and gas ventures having different magnitudes of uncertainty.

Table 7

Expected value examples (coin toss).

Table 8

Example calculation of simplistic probability of geologic success and failure using three geologic chance factors.

Geologic Chance FactorProbability
Reservoir Rock0.7
×Hydrocarbon Charge
0.8×
Sealed Closure0.5
Product = Probability of Geologic Success—Pg0.28
Probability of Geologic Failure—Pf = (1 − 0.28)0.72
Geologic Chance FactorProbability
Reservoir Rock0.7
×Hydrocarbon Charge
0.8×
Sealed Closure0.5
Product = Probability of Geologic Success—Pg0.28
Probability of Geologic Failure—Pf = (1 − 0.28)0.72
Table 9

Generalized success rates of various well classes drilled in the U.S. onshore and offshore during the 1980s as reported by the CSD (Rose, 1992a).

Well ClassPercentage Successful
Development wells75-80
All exploratory wells20-30
Extensions (outposts)40-45
In-field wildcats25-35
New-field wildcats13-18
Well ClassPercentage Successful
Development wells75-80
All exploratory wells20-30
Extensions (outposts)40-45
In-field wildcats25-35
New-field wildcats13-18

Contents

GeoRef

References

Related

Citing Books via

Close Modal
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close Modal
Close Modal