Reserves
Prediction in Exploration - the Trap We Have Got Ourselves Into and How to Deal
with It
Quirk, Dave1, Rick Ruthrauff2
(1) Maersk Oil, Copenhagen K, Denmark (2) Hess
Corporation,
Existing data from wells allow for the
confident prediction of many things in exploration. For example, within mature
plays, geological risk in a new prospect differs little from historical success
in the same play. Reservoir properties such as porosity and net thickness
belong to simple normal or lognormal distributions from which it is
straightforward to calculate cumulative probabilities. Yet exploration programs
rarely deliver what they predict. In an increasingly probabilistic world using
apparently reasonable ranges of parameters, why are prospect reserves usually
biased to the upside? Despite all the technology and data we now have, even
getting consistency in reserves assessments seems unachievable.
The problem lies with the parameter
having the greatest influence on reserves estimates in an undrilled
prospect – trap size. Unfortunately, a realistic distribution for trap size
cannot be built because there are no direct samples of the bounding surfaces.
In other words, cumulative probabilities associated with gross rock volume or
area-net are not predictable, putting into question the way probabilistic
methodologies are currently used in exploration.
Fortunately, there is a solution but it
involves partially sidestepping the trap size issue and locking on to two
things most interpreters can agree on – the culmination and last closing
contour. With this basic information, and a standard way of inputting reservoir
information, we have devised a reliable method of calculating the size and
chance of finding upside reserves (what exploration is really about) and then,
by extrapolation, it becomes easy to fully account for the downside (half of
all discoveries).
AAPG Search and Discover Article #90063©2007 AAPG Annual Convention, Long Beach, California