428
I. Lerche· S. Cao . S. Mallory· K. Petersen· A. Lowrie
of the likelihood of uncertainty. From such measures strategic exploration decisions can be made, tied also to the economic probability of successful exploration for hydrocarbon accumulations (Lerche 1992).
2
Cumulative Probability and Error Assessment
In principle, Monte Carlo procedures should be applied to the study of risk assessment and most probable outcomes of basins analysis models, i.e. the probability distribution of each model output result should be evaluated by considering all possible values of each assumption, each parameter, and each datum
point, within determined ranges, and then an intrinsic probability distribution
of finding each such input control assigned within its given range (e.g. uniform,
Gaussian, triangular, log-normal, etc.), and then, using a multi-Monte Carlo approach, run through a basin modeling calculation to produce a set of outputs.
The probability distributions of the respective outputs can then be recorded and
used in risk assessment (Fig. 2). While the Monte Carlo procedure is completely
general and can be applied to any basin analysis model, once the range and underlying probability distributions for each parameter and for the measurement
errors in data have been provided, the problem is that such a numerical investigation can be extremely computer intensive, perhaps one of the reasons that
such methods are not routinely applied to the results derived from basin analysis
models (Cao and Lerche 1989).
It would, therefore, be helpful if alternative procedures could be used, which
would provide the same probability outcome, but which would not demand such
massive computer time investment as do the Monte Carlo methods. This paper
will demonstrate how to use cumulative probability techniques to achieve such
a goal; the technical mathematical details are given elsewhere (Lerche 1997).
Examples of the procedure are probably more useful illustrations than generic mathematical formulae. Accordingly, we now illustrate use of the cumulative
procedure with two case histories.
3
An Illustration of Cumulative Probability: the Kugrua No.1 Well, Alaska
The Kugrua No.1 well is located in the western section of the National Petroleum
Reserve, Alaska, just to the south of Peard Bay. The well was drilled to a total
depth of 4127 m (12580 ft.). The present-day oil window in the well, based on vitrinite reflectance, encompasses all formations from the Sadlerochit Group
through the Torok Formation as depicted in Fig. 1.
The major problem is that only six vitrinite reflectance measurements are
available for the well, so that it is difficult to provide any reasonable measure of
paleoheat flux based on thermal inversion methods with such sparse data. Fortunately three other wells, Tunalik, Peard and S. Meade share the general regional area, and all three also have vitrinite reflectance values which can be used with
I. Lerche· S. Cao . S. Mallory· K. Petersen· A. Lowrie
of the likelihood of uncertainty. From such measures strategic exploration decisions can be made, tied also to the economic probability of successful exploration for hydrocarbon accumulations (Lerche 1992).
2
Cumulative Probability and Error Assessment
In principle, Monte Carlo procedures should be applied to the study of risk assessment and most probable outcomes of basins analysis models, i.e. the probability distribution of each model output result should be evaluated by considering all possible values of each assumption, each parameter, and each datum
point, within determined ranges, and then an intrinsic probability distribution
of finding each such input control assigned within its given range (e.g. uniform,
Gaussian, triangular, log-normal, etc.), and then, using a multi-Monte Carlo approach, run through a basin modeling calculation to produce a set of outputs.
The probability distributions of the respective outputs can then be recorded and
used in risk assessment (Fig. 2). While the Monte Carlo procedure is completely
general and can be applied to any basin analysis model, once the range and underlying probability distributions for each parameter and for the measurement
errors in data have been provided, the problem is that such a numerical investigation can be extremely computer intensive, perhaps one of the reasons that
such methods are not routinely applied to the results derived from basin analysis
models (Cao and Lerche 1989).
It would, therefore, be helpful if alternative procedures could be used, which
would provide the same probability outcome, but which would not demand such
massive computer time investment as do the Monte Carlo methods. This paper
will demonstrate how to use cumulative probability techniques to achieve such
a goal; the technical mathematical details are given elsewhere (Lerche 1997).
Examples of the procedure are probably more useful illustrations than generic mathematical formulae. Accordingly, we now illustrate use of the cumulative
procedure with two case histories.
3
An Illustration of Cumulative Probability: the Kugrua No.1 Well, Alaska
The Kugrua No.1 well is located in the western section of the National Petroleum
Reserve, Alaska, just to the south of Peard Bay. The well was drilled to a total
depth of 4127 m (12580 ft.). The present-day oil window in the well, based on vitrinite reflectance, encompasses all formations from the Sadlerochit Group
through the Torok Formation as depicted in Fig. 1.
The major problem is that only six vitrinite reflectance measurements are
available for the well, so that it is difficult to provide any reasonable measure of
paleoheat flux based on thermal inversion methods with such sparse data. Fortunately three other wells, Tunalik, Peard and S. Meade share the general regional area, and all three also have vitrinite reflectance values which can be used with