Whoever Took 'Estimating 101?'

BUSINESS SIDE OF GEOLOGY ARCHIVES

Last month we said that modern petroleum explorationists have two main professional responsibilities:

Find opportunities (= prospects).
Measure them objectively, in terms of chance of success; reserves expectations (estimated ultimate recovery); and profitability.

Finding opportunities is the fun part of the exploration business, the value-adding part. It involves new geography, new geology, new tools, secrecy -- and the competitive excitement of the hunt!

However, if we are to stay in business, the measurement part is also important.

Yes, it's often mundane -- even uncomfortable -- especially if our prospects are judged to be uneconomic. And let's face it, a lot of geologists are just naturally less interested in dollars than dolomite or downwarping!

Even so, measuring the economic value of our prospects is a key part of our professional obligations to our clients and investors.

Of course, assessing prospect value necessarily involves estimating many geotechnical parameters (and some economic ones) that affect prospect profitability.

For the geotechnical parameters, estimating is required because:

We cannot crawl down into the subsurface with a measuring tape or microscope.
Mother Earth is a very coarse filter.

A lot of investment capital is riding on our ability to objectively estimate variable factors -- such as productive area, average net pay thickness, hydrocarbon recovery factor (bbl/af or mcf/af), recoverable reserves, etc. -- and existence factors, such as our confidence (= probability) that reservoir, closure, top seal or hydrocarbon charge requirements are satisfied at depth.

This month, we'll discuss the variable factors.

The general industry convention is to estimate the variable factors£-- those uncertainties involving recoverable reserves, or flow rates, or prices and costs -- as probabilistic ranges, using cumulative probability distributions.

In figure 1, for example, the exploration team is 90 percent sure that, given a discovery, the productive area will be at least 40 acres; they think there's only a 10 percent chance it could be larger than 2,000 acres.

What really boggles the mind is that, despite the huge capital investments being laid out yearly by dozens of competing corporations on highly uncertain oil and gas ventures, very few petroleum geoscientists or engineers have ever received any formal training in effective estimating techniques! Lots of math, physics, chemistry, geoscience and computer science, to be sure -- but no training in estimating.

Almost no geologist or engineer ever took a course called "Estimating 101."

We also should learn how to eliminate systematic bias as well -- the tendency for us to consistently overestimate reserves or underestimate costs, to give two examples.

Two biases that commonly bedevil our geotechnical forecasts are overconfidence (setting predictive ranges too narrow, resulting in frequent surprising outcomes) and overoptimism (motivational bias, caused by excessive zeal in "selling" the prospect).

These techniques have proved to be useful in refining our estimates of exploration variables and eliminating bias:

1. Use of multiple working hypotheses and maps.

Prepare several possible interpretations -- optimistic and pessimistic -- to test the envelope of possibilities surrounding all key parameters: MAKE MORE THAN ONE MAP!

(This is one drawback to conventional workstation mapping, which tends to focus on THE answer.)

2. Independent Multiple Estimates.

Get input from exploration team members, exploration committee reviews or other joint-venturers in partnership prospects. For example: Averaging independent estimates by four seasoned experts usually gives a better (and quicker) answer than having one geoscientist work four more weeks on the problem!

3. Honor Nature's Envelopes.

Make predictions employing a parameter's observed (and therefore expectable) distribution (exponential for percent decline-estimates, lognormal for prospect reserves estimates, etc.).

4. Reality Checks.

Is your prospect reserves distribution reasonable compared with a current field-size distribution?

Are your estimates of reservoir parameters credible considering analog fields?

Are the extreme outcomes (P99 percent, P1 percent) plausible?

Is it remotely possible (P1) that the productive area of the prospect in figure 1 could be as large as 10,000 acres? Could it be as small (P99) as eight acres, and still support flow into the borehole?

5. Use Proper Statistical Measures.

For exploration portfolios, the mean
(= average) is the single best representation of the prospect reserves distribution. Use probabilistic ranges, rather than minimum/most likely/maximum.

Avoid the meaningless term "most likely," as well as single-number estimates.

6. Learn From Prior Predictions.

Start preserving all your predictions, and compare them against actual outcomes.

What can you learn about your predictive performance? Do you typically overestimate? Are your ranges too narrow?

How can you improve your estimating ability?

(Next month, a look at existence factors.)