Sensitivity (Well Logging / Reservoir)
In well logging, sensitivity is the minimum detectable change in a formation property that a logging tool can distinguish from background noise, defining the tool's resolution floor for parameters such as resistivity contrast, bed thickness, or formation density change; in reservoir engineering, sensitivity analysis quantifies how much the output of a simulation model (recovery factor, plateau rate, NPV) changes in response to systematic variations in uncertain input parameters such as permeability, porosity, and relative permeability endpoints.
Key Takeaways
- Tool sensitivity in wireline logging is determined by detector efficiency, source strength, formation background, and signal processing: a gamma ray tool may detect a 1 API unit change but cannot reliably resolve a 0.1 API unit change in a high-background formation.
- Vertical resolution and sensitivity are related but distinct: vertical resolution is the minimum bed thickness a tool can distinguish, while sensitivity is the minimum property contrast it can detect regardless of bed thickness.
- In reservoir engineering, one-at-a-time (OAT) sensitivity analysis varies each input parameter individually while holding others constant; Monte Carlo sensitivity analysis allows all parameters to vary simultaneously, generating a probabilistic distribution of outcomes.
- Tornado charts (horizontal bar charts ranked by output sensitivity) are the standard visualization tool for ranking uncertain input parameters by their impact on a model output, guiding data acquisition priorities and uncertainty reduction strategies.
- Geomechanical sensitivity analysis evaluates how changes in pore pressure, temperature, and reservoir compaction affect wellbore stability, subsidence, and caprock integrity, informing safe injection pressure limits and reservoir depletion strategies.
Fast Facts
A standard compensated neutron porosity tool has a sensitivity of approximately 0.5 to 1 porosity unit (pu) under typical borehole conditions. Resistivity tools can detect contrasts of less than 0.1 ohm-m in conductive formations but lose sensitivity in highly resistive carbonates. A typical reservoir simulation sensitivity study varies 5 to 15 key parameters over P10-P90 uncertainty ranges, generating 10 to 50 simulation runs to map the output distribution. A well-designed sensitivity study can reduce capital expenditure uncertainty by 30 to 50 percent by identifying which reservoir parameters require additional appraisal data.
Tip: For reservoir simulation sensitivity studies, always test both the high and low end of each parameter's uncertainty range to detect non-linear responses; parameters that appear low-sensitivity in a one-at-a-time study can become critical when combined with other unfavorable values, a behavior that only Monte Carlo or full factorial designs reveal.
What Is Sensitivity in Well Logging and Reservoir Engineering
The word sensitivity describes how a measuring system or model responds to small changes in its input. In well logging, it answers the practical question: how small a change in a formation property (resistivity, density, natural radioactivity, acoustic slowness) must exist before the logging tool can reliably detect it and report it as a real signal rather than noise? This is critical because formation properties often change gradually across lithological transitions or thin pay beds, and whether those gradual changes register on the log determines whether the bed is identified as net pay or dismissed as background.
In reservoir engineering, sensitivity analysis addresses a different but equally important question: of all the uncertain parameters that go into a reservoir simulation model, which ones have the greatest impact on the forecast outcomes that matter to decision-makers? These outcomes might be cumulative recovery, peak production rate, time to water breakthrough, or net present value. By systematically varying each parameter, the engineer identifies where uncertainty reduction investment (more core analysis, more well tests, more seismic) will deliver the greatest reduction in decision risk.
How Sensitivity Works in Logging and Reservoir Analysis
Logging tool sensitivity is characterized during laboratory calibration and field testing by exposing the tool to formations of known properties and measuring the signal response as the property changes in small increments. The minimum detectable signal is defined as the property change that produces a signal equal to twice the standard deviation of the background noise level (a 2-sigma criterion), though some applications use 1-sigma or 3-sigma thresholds. Sources of noise include statistical counting fluctuations (nuclear tools), thermal noise in electronics (resistivity and acoustic tools), borehole rugosity effects, and mud column interference. Environmental corrections applied to raw log data reduce systematic biases but do not improve statistical sensitivity.
Vertical resolution, often confused with sensitivity, refers to the minimum bed thickness that a tool can detect and measure accurately. A thin bed may have a very large property contrast (high sensitivity) but still not be resolved if it is thinner than the tool's vertical resolution. Conversely, a gradual property change over a thick interval may be detectable (sufficient contrast) but not precisely locatable. Deconvolution processing, particularly thin-bed inversion methods applied to resistivity and neutron-density logs, can recover true formation properties in beds below the nominal vertical resolution limit, effectively extending the useful sensitivity of the measurement.
In reservoir engineering sensitivity analysis, the standard approach begins with a base case simulation using the most likely (P50) value for each parameter. Then each parameter is independently varied to its P10 and P90 uncertainty bounds while all others remain at their base case values, and the resulting output is recorded. A tornado chart plots the swing (P90 output minus P10 output) for each parameter as a horizontal bar, ranked from largest to smallest. The parameters with the longest bars are the primary drivers of output uncertainty and the highest priority targets for additional data collection. More sophisticated approaches including Latin Hypercube sampling or full-factorial experimental designs explore interaction effects between parameters that OAT methods miss.
Sensitivity Analysis Across International Jurisdictions
In Canada, the AER requires that reserve submissions comply with the COGE Handbook (Canadian Oil and Gas Evaluation Handbook), which mandates probabilistic reserve estimates (1P/2P/3P) that implicitly require sensitivity and uncertainty analysis. AER Directive 065 sets out specific requirements for volumetric uncertainty reporting. WCSB operators working unconventional Montney and Duvernay plays routinely perform sensitivity studies on key parameters including natural fracture intensity, matrix permeability, and fracture closure pressure, which drive tight gas and liquids-rich condensate production forecasts. The AER's Unconventional Resources Program monitors submitted sensitivity data to track industry understanding of these complex plays.
In the United States, the SEC's reserve reporting rules under Regulation S-X Rule 4-10 and the associated guidance from the Society of Petroleum Engineers require that material uncertainties in reserve estimates be disclosed. Companies with SEC-registered securities typically perform deterministic sensitivity analysis supporting their 1P/2P/3P estimates and may supplement with probabilistic analyses for large, material assets. BSEE does not prescribe specific sensitivity analysis methods for offshore operators, but internal company standards at majors such as ExxonMobil, Chevron, and ConocoPhillips specify detailed Monte Carlo workflows for offshore field development decisions exceeding defined capital thresholds.
In Norway, Sodir's petroleum resources classification framework requires probabilistic reserve estimates for all NCS fields, making formal uncertainty and sensitivity analysis a regulatory compliance requirement. Norwegian operators including Equinor use industry-standard simulation tools (Eclipse, tNavigator, CMG) with integrated uncertainty modules to generate probabilistic production profiles. The Norwegian Oil and Gas Association (NOROG) has published best practice guidelines for uncertainty analysis in development planning that define standardized parameter variation ranges based on field development stage and data availability.
In the Middle East, reservoir simulation and sensitivity analysis support some of the world's largest and most complex field development decisions. Saudi Aramco's reservoir simulation department operates one of the largest in-house simulation groups globally, running full-field compositional models with tens of millions of cells. Sensitivity studies for Arabian supergiant fields such as Ghawar test the impact of aquifer support strength, relative permeability hysteresis, and vertical permeability on 50-year production forecasts. Given the scale of these fields, even a 5 percent sensitivity in recovery factor represents billions of barrels of recoverable oil, making the quality of sensitivity analysis directly proportional to enormous financial outcomes.
Synonyms and Related Terminology
In logging contexts, sensitivity is sometimes called minimum detectable signal, tool resolution, or detection limit. In reservoir engineering, sensitivity analysis is also called parametric study, uncertainty analysis, or perturbation analysis. Related concepts include tornado chart, Monte Carlo simulation, vertical resolution, reservoir simulation, uncertainty analysis, and probabilistic reserves.
Frequently Asked Questions
Q: What is the difference between sensitivity and uncertainty?
A: Sensitivity measures the rate of change of an output with respect to an input, describing how strongly the output responds to a unit change in the input. Uncertainty quantifies the range of plausible input values. Together, they determine the uncertainty in the output: a high-sensitivity parameter with a wide uncertainty range contributes more to output uncertainty than a high-sensitivity parameter with a narrow uncertainty range.
Q: How does logging tool sensitivity affect net pay determination?
A: Net pay cutoffs are applied to log-derived porosity, saturation, and shale volume curves. If the logging tools cannot resolve thin pay intervals (below vertical resolution) or subtle property contrasts near cutoff thresholds (at the sensitivity limit), pay will be missed or overcounted. Sensitivity limitations in resistivity tools in low-contrast pay (slightly hydrocarbon-saturated zones with moderate to high Sw) are a particularly common source of pay undercount in deep or tight reservoirs.
Why Sensitivity Analysis Matters
In well logging, sensitivity defines the limits of what can be known about a formation from a specific measurement, establishing the floor of petrophysical certainty and informing decisions about whether additional or different measurements are needed to resolve critical unknowns. In reservoir engineering, sensitivity analysis is the primary tool for ranking data collection priorities in an appraisal or development program: it answers the question of where to spend the next dollar to reduce the risk of a bad development decision. Fields where the dominant sensitivity is to permeability should prioritize well testing; fields dominated by relative permeability sensitivity should prioritize coreflooding; fields dominated by structural uncertainty should prioritize seismic acquisition. Without sensitivity analysis, data acquisition programs are designed on intuition rather than quantified risk reduction, consistently underperforming their potential to reduce decision uncertainty.