Resource data provide insights into U.S. exploration risk

John D. Grace
Earth Science Associates
Arlington, Tex.

Analysis conducted here of these data show how new field wildcat success ratios have changed over this century and how they vary spatially within oil and gas provinces and among them.

This investigation yielded several surprises. The greatest was the improvement in exploration success since 1980, most dramatically for gas. Second is the difference between the spatial structure of exploration risk for oil versus gas. Finally, the analysis produced a powerful method to aid in the pre-drill prediction of the probability of exploration success for most U.S. oil and gas provinces.

Data and analysis

The USGS, as part of its documentation of the 1995-96 National Oil and Gas Assessment, released data sets summarizing drilling for both the onshore and offshore U.S. These were drawn from Petroleum Information's Well History Control System (WHCS), the most complete data available on U.S. drilling.

A 40-acre grid was established across the U.S. Within each 40-acre grid cell, data for all wells were summarized. This included the number of wells, type of wells, dates of drilling and production, depth, and results of drilling. Separate but related 40-acre grids were developed to differentiate drilling results by the 470 plays analyzed by the USGS onshore and for state waters.

The "province-level" data released by the USGS, which summarize drilling at all depths within a 40-acre cell, cannot be merged directly with the "play-level" data, which report the results of drilling stratigraphically. ESA reprocessed both data sets to merge them for inclusion into its U.S. Oil and Gas Resource System 1.0. The reprocessed data were used in this analysis.

Approximately 1.5 million 40-acre cells contain drilling data.

These data sets formed the foundation for our analysis of the structure of new field exploration risk. The most important results were obtained by investigation of the relationship between exploration risk and the horizontal distance of new exploratory wells from existing production. This relationship was also tracked over time, showing how exploration risk changed in the U.S. during the 20th century.

For each decade between 1900 and 1993, all productive development wells drilled before that decade were posted to a specific theme within a geographic information system. Then, all new field wildcats drilled within each decade, along with their result (oil, gas, or dry) were posted to a separate GIS theme. Finally, circles with radii of one, two, four, eight, and 16 miles were drawn around each of the previously drilled development wells. The number and results of new field wildcats were tabulated within each circle.

Exploration success

It stands to reason that the probability of exploration success declines as distance from established production increases. The clarity of the subsurface picture degrades with greater distance from control, and there is a higher chance that the factors that control accumulations are different the farther away from an existing field.

Conventional wisdom also holds that exploration success in the U.S. has fallen over time, principally because of resource depletion. With over 40,000 oil and gas fields already discovered in the U.S., those that remain are not only smaller but deeper and in environments that are more complex geologically and logistically.

To some degree, both presumptions are correct. Surprises, however, were found within those general truths. Overall exploration success does decline with increasing distance from established production, but there is a big difference between oil and gas. There are also substantial differences between provinces.

While the new field wildcat success ratio has also dropped through the century, it turned around after 1980 for oil and starting after 1960 in the case of gas. Some of the increase is due to better technology and geologic knowledge and some is due to dropping minimum economic field size. The improvement in gas can also be traced to new plays and the fact that gas was increasingly the primary target for drillers.

Fig. 12 [34538 bytes] shows the change in the distribution of new field wildcat wells drilled over this century with respect to distance from existing production at the time of drilling. A proportionately large number of wells was drilled relatively far from production in the early part of the century. Starting around World War II, however, new field exploration became increasingly concentrated close to previously discovered fields.

There are three reasons for this. First is that distant drilling is often associated with the initiation of a new play in a province, and most U.S. plays were established in the first half of the century. Second, as overall density of drilling (and discoveries) increased, there were fewer areas that were far from an existing field-so the new areas in which to look were shrinking. Finally, after most of the largest fields were discovered, explorationists moved closer in to established production to increase the probability of success.

Oil, gas differences

Changes in the composition of new field wildcat drilling were coincident with changes in the outcome of exploratory wells. These results, however, were substantially different for oil and gas.

Fig. 13 [34804 bytes] shows the history of wildcats resulting in oil discoveries by decade over the 20th century. Looking at the century as a whole, success has obviously declined. It is also very clear that risk increases with distance from production. There are, however, two important qualifications to these conclusions.

First, since 1980, success at all distances from established production was higher than it was from 1960-79. When the period of 1990-93 is examined separately, the improvement is even greater. Second, while the reduction in success with distance from production is a persistent phenomenon, for distances greater than two miles success has been quite stable since 1940. The biggest drops are close to existing production.

Given the volume of drilling in the areas within two miles of existing fields, it should be little surprise that the number, size, and quality of targets have diminished.

The picture for gas is different in three ways (Fig. 14 [26002 bytes]). In the first half of the century, gas was found far less often than oil in successful new field wildcat wells. Second, compared to oil, there is a comparatively small influence of distance from production on gas exploration success. Finally, the increase in exploration success for gas since 1980 was higher than it was for oil.

The low gas success rate in the early part of the century, particularly close to existing production, was largely an economic and regulatory phenomenon. After World War II, with changes in the treatment of natural gas in interstate commerce and its greatly increased use in the economy, gas exploration took on its own identity, rather than being the consolation prize in the search for oil. As this occurred, oil and gas success rates converged.

Much more interesting is the relative constancy of gas success as a function of distance from production. While success does decline on average, the rate of decline is less than for oil and the variance is much higher. As there is roughly as much gas as oil in the U.S., it is possible to conclude that these hydrocarbons are much more diffusely distributed in sedimentary rocks. This might be expected in light of its higher relative permeability, making longer-distance migration more likely and trapping efficiency en route lower.

Finally, there is the improvement in exploration success ratio since 1960. Although oil exploration gained, comparing performance in the 1980-93 period to 1960-79, gas improvement was twice as high as oil for exploration within two miles of existing production, about the same in the four to eight mile range, and twice as high again for eight to 16 miles and a major increase at distances greater than 16 miles.

Some of the improvement in gas success was due to the increasing role of "unconventional" sources in both gas exploration and production, particularly since 1980. The distributions of both "tight" gas and coalbed methane are not areally restricted in the same fashion as conventional accumulations, discretely bounded from below by a gas/water contact. Consequently, the distinction of "new fields" is less clear. In these environments of spatially diffuse concentrations of gas, wildcats drilled even many miles from established production may, in fact, be closer to extension exploration than to the establishment of a "new field" in the conventional sense.

Province differences

The decline in exploration success with increasing distance from production is very commonly observed in U.S. oil and gas provinces.

There are, however, strong differences between provinces, both in the success ratios seen very close to production and in the rate at which success deteriorates with distance. In some regions, the relationship does not hold at all.

To empirically investigate the differences among U.S. oil and gas provinces, the results from 48 of 71 onshore and state waters provinces were subjected to regression analysis. The 23 provinces not subjected to the analysis had little or no drilling data.

The curves in Figs. 13 [34804 bytes] and 14 [26002 bytes], particularly the former, suggest that success ratio declines exponentially with distance. Therefore, a log-linear regression model was adopted, where exploration success was posited as a function of the natural logarithm of distance from existing production (Equation 1 below). To make the results practical for today's explorationists, the analysis was restricted to the period of 1970-93. The results presented here combined oil and gas for total hydrocarbon exploration success.

Equation 1 Success ratio = Intercept + [slope coefficient x ln (distance from production in miles)] 

The regressions produce estimates for the intercept and slope.

The intercept is interpreted here as the success ratio of new field wildcat exploration immediately in the vicinity of existing production. As such, it can be viewed as the average upper limit of success for new field wildcat exploration in a province.

For mathematical and logical reasons, the intercept cannot be interpreted as the success ratio at zero miles from production, so it must be thought of as being as close as possible while still being a new field wildcat.

In log-linear form, the slope coefficient is the decline in the percentage of exploration success for each one unit increase in the log of distance from production.

The parameter estimates for 31 U.S. provinces are given in Table 6 [42728 bytes]. The final value in the table, adjusted R2, reflects the fraction of observed variance in exploration success "explained" by distance from existing production. The closer this value is to 1, the greater the explanatory power of the equation.

Inspecting the results in Table 6 [42728 bytes] as a group, there is a significant negative correlation between the intercept and the slope coefficients. That is, the higher the average upper limit of exploration success in a province (the intercept), the faster exploration success diminishes with distance from existing production (the slope coefficient), and vice versa. In some instances, this is because regions with high intercepts are at a low stage of exploration maturity (e.g., in Alaska), while others clearly are not.

The negative correlation between intercept and slope implies, as has been noted elsewhere, that there are geologically concentrative and dispersive environments.7 In the former, processes that created hydrocarbon accumulations formed largely contiguous trends of fields. In the latter, diffusive processes tend to spatially equalize the probabilities of finding oil and gas in a region.

Comparing the provinces individually, there are very significant differences in the results between basins reflecting variations in underlying petroleum geology. Some geographic groupings are apparent. For instance, the Rocky Mountains, the Denver, Big Horn, Wind River, Powder River, and Central Montana provinces all share similar values for both parameters; so do the two divisions of the Gulf Coast basin. However, others are quite different from their neighbors (e.g., Southwestern Wyoming). These differences invite further investigation and disaggregation of results to the play level.

Forecasting success

Before drilling, explorationists must always forecast the probability of success in order to properly assign expected economic value to a prospect. This is typically done by subjective assessment of component risk factors reflecting the expected sufficiency of a prospect's hydrocarbon charge, reservoir quality and trap integrity.

While historical exploration success in a region is often used as a guide, it does not exploit the strong relationship shown here between exploration outcome and distance from existing production. With the data in Table 6 [42728 bytes], however, it is possible to compute the average probability of exploration success for a new field wildcat drilled at an arbitrary distance from existing production.

For example, the average success probability for a new field wildcat drilled in the Wind River basin seven miles from the nearest existing production is 13%. This is found through substituting the Table 6 [42728 bytes] values for the Wind River basin into Equation 1: 12.945 = 17.846 + [-2.519 x ln (7)], or 12.945 = 17.846 + [-2.519 x (1.945)]. This empirically based estimate of exploration success could then be modified based on prospect-specific analysis of geologic factors related to charge, reservoir, and trap.

The explanatory power of the regression equations in Table 6 [42728 bytes] is very high for nearly all provinces reported. For 17 provinces analyzed but not reported in Table 1, the statistical results of indicated that the relationship in Equation 1 was weak or did not hold at all. The regression results for these provinces had low adjusted R2, statistically insignificant estimates of intercept and slope or, in most cases, all three deficiencies. Many of the 17 are small provinces or relatively immature. The most important exception was the Anadarko basin.

This provides confidence in the usefulness of this approach to pre-drill estimation of new field wildcat exploration risk. Although it does not replace analysis of geologic risk factors, by providing a robust starting point, application of this method should increase the accuracy and consistency of pre-drill predictions, which are critical to successful exploration programs.

The same data used in this study of risk at the province level can be analyzed at the play level, where the connection to underlying geologic variables is much clearer and can be directly included. At the play level as well, using geostatistically estimated ellipses, instead of circles, will also improve the precision of success forecasts for individual wells. The relationship of vertically stacked exploration targets on the probability of success is also a factor which requires further examination.

The findings here, as well as the future directions of research they imply, suggest that greater empirical analysis of exploration risk has high leverage in refining pre-drill prediction of the probability of success in new field wildcat exploration. Improvement in risk analysis is a critical determinant in raising exploration profitability, whether applied in the U.S. or abroad.

Conclusions

Several conclusions can be drawn on the structure of exploration success based on the analysis presented here:

1. Generally, the probability of success in new field wildcat drilling has declined in the U.S. over the 20th century. This decline has been greater for oil exploration than for gas. The decline is principally a result of exhaustion of the resource base.

2. Since 1980 for oil and 1960 for gas, exploration success has improved, more for gas than for oil. The increase is due to improving technology and geologic knowledge and a reduction in minimum economic field size. The large relative improvement for gas is related to these same factors and the more deliberate search for it. Some of the more recent improvement may also be related to the increasing role of spatially diffuse "unconventional" gas accumulations in the resource base.

3. Exploration success declines as a function of distance from existing production; the relationship is stronger for oil than for gas. Large differences between provinces are related to underlying petroleum geology. In some large regions, provinces within them show similar results, indicating that the same general geologic controls are operative across the entire region.

4. Statistically estimated regression functions for most U.S. provinces provide a strong empirical foundation for pre-drill forecasts of new field wildcat exploration success as a function of distance from nearest established production. Application of this simple approach should improve the consistency and accuracy of estimating exploration risk.

References

1. 1995 National assessment of oil & gas resources, U.S. Geological Survey Circular 1118, 1995.

2. An assessment of the undiscovered hydrocarbon potential of the nation's outer continental shelf, Minerals Management Service MMS 96-0034, 1996.

3. U.S. crude oil, natural gas and natural gas liquids reserves, annual reports 1980-95, U.S. Energy Information Administration.

4. U.S. crude oil, natural gas and natural gas liquids reserves, annual reports 1947-79, American Petroleum Institute and American Gas Association.

5. Mast, R.F., et al., Estimates of undiscovered conventional oil and gas resources in the U.S.-A part of the nation's energy endowment, U.S. Dept. of Interior, U.S. Geological Survey and Minerals Management Service, 1989.

6. Joint Association Survey on 1994 Drilling Costs, American Petroleum Institute, Washington, D.C., 1996.

7. Klemme, H.D., Field size distributions related to basin characteristics, in Rice, D.D., (ed.), Oil and gas assessment-methods and applications, AAPG Studies in Geology No. 21, 1986, pp. 85-100.

The Author