24-Pet-B3 Petroleum Geology, December 2015, Q9

Question 9

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Tornado diagram. A tornado diagram is a horizontal bar chart used to rank input variables (oil price, reserves, CAPEX, OPEX, royalty rate, etc.) by how much each one, varied independently over its own realistic low–high range while every other variable is held at its base case, swings the output economic metric (typically NPV). Bars are sorted longest-to-shortest from top to bottom, which gives the chart its funnel/tornado silhouette, and immediately shows which one or two variables actually matter to the decision — the variables with the shortest bars can usually be fixed at their base-case value without materially changing the conclusion.

Fig. Q9-1 — tornado diagram: each bar shows the swing in NPV from that variable's low (red) to high (green) case, sorted by swing size.

Spider diagram. A spider diagram is a line chart with the % change in an input variable on the x-axis (centred on the base case at 0%) and the resulting % (or absolute) change in the output metric on the y-axis, with one line drawn per variable, all radiating from the same base-case point. The slope (steepness) of each line shows how sensitive the output is to that variable — a steep line means a small change in that input moves the outcome a lot, while a shallow line means the outcome is relatively insensitive to it — and a non-linear (curved) line flags a variable whose effect on NPV is itself non-linear (e.g., a hurdle-rate or price-floor threshold effect).

Fig. Q9-2 — spider diagram: % change in each input (x-axis) vs. resulting % change in NPV (y-axis), one line per variable.

Both tools are used at the project-screening stage to focus further data-gathering and risking effort on the few variables that actually drive the economic outcome, rather than spreading effort evenly across every input.

Topic: Sensitivity analysis display tools

Key relations: Tornado: one bar per variable, length = |NPV(high) - NPV(low)|, sorted descending. Spider: one line per variable, slope = ΔNPV/Δinput over a common % range.

Why this works: Both isolate ONE-VARIABLE-AT-A-TIME effects on the same base case, which is exactly what is needed to triage where risk/uncertainty analysis effort should go next — a tornado answers “which variable matters most,” a spider adds “and is the relationship linear or does it kink.”

Common pitfall: Reading either chart as if it captured combined/simultaneous variable movement — both are strictly one-at-a-time sensitivities; correlated joint movement of two variables (e.g., price and reserves) needs a separate scenario or Monte Carlo analysis.

Source: Thompson & Wright, Oil Property Evaluation; COGEH Vol. 1, Ch. on risk analysis.

Question 9 of 22

Question 9