Design Assumptions and Uncertainty

A design assumption is a stated condition or simplification adopted as a basis for analysis. Where it substitutes for missing evidence, its provisional status, rationale and consequences need to remain visible.^[1] In a mine, assumptions connect geological interpretations, extraction arrangements, infrastructure and production expectations. Their value lies in making decisions possible while preserving a clear account of what those decisions depend on.

Uncertainty concerns what remains unresolved about inputs, models or outcomes. Managing it means deciding which unknowns could change an engineering choice, what evidence would be useful, and what commitments can reasonably proceed before that evidence is available.

Separate Evidence, Assumptions and Requirements

These working distinctions help prevent a provisional judgement from becoming an unnoticed design restriction:

The distinction matters when comparing alternatives. In their mining study methodology, Kühn and Visser separate requirements, assumptions and constraints, and recommend revisiting technical solutions as influential inputs become better understood. An assumed operating arrangement can otherwise exclude an alternative before its merits have been examined.^[2]

Distinguish Variability from Incomplete Knowledge

Two useful categories are aleatory uncertainty, representing variability treated as inherent in the analysis, and epistemic uncertainty, arising from incomplete knowledge or model limitations. Additional investigation can reduce the latter; it does not remove the underlying variability represented by the former. NASA's 2024 risk handbook uses this distinction when considering whether further research would justify its cost.^[3]

Applied to mine design, variable ground conditions along a development route and uncertainty about the location of a particular structure pose different questions. Mapping may improve the location estimate while leaving genuine spatial differences to accommodate. The distinction depends on the question and scale: variations treated statistically in a long-range plan may later be resolved individually by closer investigation. A distribution also has uncertain parameters; assigning probabilities does not make knowledge complete.

Make Assumptions Traceable and Reviewable

Documenting the rationale and affected estimate elements provides traceability. Assigning responsibility for checking assumptions, and revisiting them at decision points, connects that record to action.^[1][3]

For the water-services assumption above, a practical record would identify:

• Basis: the investigation, operating records or specialist judgement supporting the statement.
• Applicability: the development area, operating stage and conditions covered.
• Representation: the adopted value, range or qualitative condition, including what its bounds mean.
• Responsibility: the discipline responsible for checking the basis and communicating revisions.
• Dependencies: affected layouts, service designs, schedules and estimates.
• Review trigger: new investigation results, a changed route or an upcoming commitment requiring stronger evidence.

An expiry tied to a decision can be more informative than an arbitrary calendar date. For example, the assumption might support an option comparison but require reassessment before the access arrangement is committed. Retaining earlier versions allows the reason for a changed layout to be reconstructed.

Match Confidence to the Decision

An estimate or study label does not resolve every underlying uncertainty. AACE's 2022 classification guide ties estimate class to the maturity of scope definition and explains that class alone does not determine accuracy. It also warns that uneven definition across interdependent parts of a project can undermine a decision about the whole.^[4]

For mine engineering, the useful question is which unresolved input could reverse the present choice. Materiality also includes consequences for safety, technical feasibility and required service, even when the effect on estimated cost is small. A well-developed excavation arrangement may still depend on uncertain water services. Conversely, further refinement of an input that does not change the preferred arrangement may offer little decision value.

The effort justified by an assumption depends partly on the commitment it supports. Comparing routes can tolerate conditional alternatives; constructing a permanent access creates consequences that are harder to reverse. The evidence sought should answer the decision at hand. A detailed estimate of one component cannot establish the feasibility of an unverified interface.

Test Influential Assumptions and Their Dependencies

One-at-a-time sensitivity analysis varies an input while holding others fixed to expose its influence. GAO cautions that the ranges need a defensible basis and that such tests do not establish the combined uncertainty of the result. Cost contingency likewise needs an explicit risk basis; an arbitrary allowance cannot resolve missing engineering.^[1]

A scenario specifies a coherent set of conditions and may change several related assumptions together. It can test an alternative interpretation without assigning a probability. High and low cases are therefore not automatically confidence intervals. Probabilistic analysis needs justified distributions and relationships between inputs; its conclusions remain conditional on those choices.^[5]

Dependencies deserve explicit attention. Kühn and Visser note that mining study variables may be dependent and that covariance or correlation may need consideration. Physically linked inputs should be represented consistently before interpreting a comparison.^[2] In the water-services example, an investigation result could alter both the expected inflow and the development sequence. Combining the original production timing with the revised water-management arrangement could describe an internally inconsistent case.

Examine the Model as Well as the Numbers

Changing parameter values cannot test every modelling assumption. EPA's 2009 environmental modelling guidance distinguishes uncertainty in inputs from uncertainty in model structure and in applying a model outside its intended setting. It also distinguishes calibration from corroboration against independent evidence: agreement with the observations used to adjust a model does not independently establish predictive performance.^[5]

For a mine water model, the review might therefore ask whether the assumed hydraulic connections, boundaries and time scale suit the proposed excavation sequence. A refined parameter estimate within an unsuitable conceptual model may leave the important question unanswered. Where alternative interpretations remain plausible, their engineering consequences should be compared explicitly. Monitoring can test predictions as access develops, while discrepancies may require revising the interpretation rather than repeatedly adjusting parameters.

Follow a Changed Assumption Through the Mine

Consider this qualitative hypothetical situation: an underground extension is being compared on the assumption that existing water services can support its development. New investigation suggests a connection to a water-bearing structure that the current interpretation did not include.

The immediate question is whether the new interpretation changes the choice of access. A different route might avoid the structure but alter development length and the timing of production access. Retaining the route might require reconsidering water collection, pumping and conveyance, together with their power supply, installation space and maintenance access. Changes in the water-management arrangement could also affect treatment, discharge arrangements and the timing of infrastructure construction.

The next investigation should discriminate between interpretations that lead to different decisions. If the route comparison remains unresolved, more detail on pump procurement alone will not settle it. A provisional alternative can remain in the comparison while evidence is collected, with the associated schedule and cost consequences carried consistently.

When the assumption changes, the affected disciplines need the same revised basis. Updating a water estimate while leaving the access drawing, production schedule and infrastructure scope unchanged would preserve incompatible decisions. The review is complete when each affected output has been checked, the remaining conditional choices are identified, and the next evidence or decision trigger is recorded.