Technical Architecture and Analysis Registers

Technical Architecture and Analysis Registers

A technical architecture describes how a borewell engineering conclusion is built from observations, assumptions, calculations and decisions. The analysis registers make those relationships explicit. They are especially valuable when a project contains many calculations or when the same measured values support pump selection, operating limits, production forecasts and performance comparisons.

The objective is not to produce a long inventory of equations. It is to show which question each calculation answers, whether its assumptions apply and which decision depends on the result. A mathematically correct calculation can still be inappropriate for the site or the intended use. The register must therefore record applicability as well as arithmetic.

Organise the evidence into distinct layers

Use separate layers for asset configuration, raw observations, corrected data, calculated quantities, interpretations and operating decisions. Connect them through stable identifiers. This structure allows a change in one layer to be assessed without rewriting all the others. It also makes it easier to explain which conclusions are measured, derived or judgment-based.

For example, a flow observation belongs to the measurement layer. A period volume calculated by integrating flow belongs to the calculation layer. A forecast based on a future schedule belongs to a planning layer. A permitted operating instruction belongs to a decision layer. Calling all four “measured capacity” obscures their different evidential status.

Create a formula register with enough context

For each material formula, record its identifier, purpose, expression, input definitions, units, output, assumptions and verification method. Identify the source of the equation where relevant. Include the dataset or record that supplies each input. A formula copied into a report without units and variable definitions is not a controlled calculation.

Record known limits and prohibited uses. A volume equation can calculate water discharged over time; it cannot establish a sustainable abstraction rate. A specific-capacity ratio can describe a measured relationship under stated test conditions; it does not independently resolve every aquifer parameter or future operating condition. The register should prevent these category errors.

Verify units before interpreting results

Dimensional checking is an efficient first review. Confirm that rate and time units combine to the intended volume, that level differences use a consistent reference and that electrical energy and water volume share the intended interval. Explicitly record conversions. A hidden conversion factor in a spreadsheet cell is difficult to review and easy to duplicate.

For an illustrative calculation, 2.5 litres per second equals 9 m³/h because 2.5 × 3,600 / 1,000 = 9. Over four hours at that constant rate, volume is 36 m³. These are unit and volume calculations only. The register should state that the example does not establish whether four hours at that rate is an approved duty for any real well.

Keep a groundwater-analysis register

Record each interpretation separately from the equations used within it. Identify the conceptual model, relevant construction and geological information, selected test intervals, treatment of disturbances and limitations. State why the method is considered applicable. A software output should not become a site conclusion merely because the program completed successfully.

Include alternative explanations where they materially affect the decision. If available evidence cannot distinguish between competing interpretations, preserve that uncertainty. The appropriate output may be a constrained decision, further observation or a revised test plan. It is not always a more precise parameter estimate.

Use simple ratios within their limits

Specific capacity is commonly expressed as discharge divided by drawdown under stated conditions. If a fictional test has Q = 12 m³/h and drawdown s = 8 m, the ratio is 1.5 m²/h when written as those units, often described operationally as 1.5 m³/h per metre of drawdown. Preserve the rate, elapsed time and conditions associated with the ratio.

Do not compare ratios from different tests without considering duration, pumping history, well condition and measurement basis. A changed ratio can justify investigation but does not uniquely identify the cause. The analysis register should say whether the comparison is like-for-like, qualified or unsuitable for the proposed conclusion.

Record time-series integration methods

For a variable flow record, specify how volume is calculated across intervals. A piecewise-constant or trapezoidal method may be used when appropriate to the data and purpose, but the selection should be stated. Record treatment of gaps, duplicate timestamps and operating-state changes. Integration should not silently bridge an interval whose behaviour is unknown.

A fictional sequence averaging 8 m³/h over two hours and 10 m³/h over three hours gives 46 m³ when those averages represent the stated intervals. If the rates are only isolated endpoint readings, that same calculation requires an assumption about the intervening behaviour. The register should make the distinction visible.

Separate calibration from validation

Calibration adjusts model parameters to fit selected observations. Validation or other independent assessment examines whether the model is suitable beyond that fitting exercise for the stated purpose. A close fit to the calibration data is not proof that every assumption is correct or that future behaviour is accurately predicted.

Record which observations were used to fit the model and which were used for independent comparison where such a comparison is available. If the available record is too limited for an independent check, say so and constrain the interpretation. Do not create a claim of validation merely by plotting the model against the same data used to tune it.

Make uncertainty actionable

List the uncertainties that could change the decision. Measurement reference, discharge accuracy, timing, boundary conditions and conceptual-model assumptions may have different effects. Where useful, examine sensitivity by changing one defined assumption and observing the consequence. Do not present the resulting range as a statistical confidence interval unless that interpretation is justified.

The register should identify which additional evidence would most improve the decision. This may be a better measurement, a longer relevant observation period, construction information or specialist review. An uncertainty list with no connection to action becomes a disclaimer rather than an engineering tool.

Verify calculations with independent examples

Select test cases that exercise ordinary values and meaningful edge conditions. Check zero or missing denominators, unit conversions, time-boundary handling and invalid inputs. A calculation should fail clearly when required evidence is absent rather than produce a plausible number from an empty cell interpreted as zero.

Use an independent calculation for material results. Independence means more than copying the same formula into a second spreadsheet. A hand calculation, alternate implementation or competent review of the logic can expose errors shared by duplicated formulas. Record the test inputs, expected result and actual outcome.

Control model and formula changes

When an equation, input convention or model configuration changes, identify all dependent outputs. Recalculate the affected results and review whether the operating conclusion changes. Preserve the previous issue and the reason for revision.

Do not revise a model solely to obtain a preferred operating rate. A justified change should arise from evidence, corrected logic or a better-supported conceptual understanding. If the revised interpretation increases uncertainty, the report should reflect that result rather than treating every revision as an improvement in performance.

Fictional register review

A reviewer finds a workbook that reports both monthly volume and specific capacity from a column labelled Q. In one sheet Q is litres per second; in another it is treated as cubic metres per hour. The formula register exposes the mismatch because each calculation must declare its input unit and source.

The correction requires more than changing a heading. The team verifies the original measurements, applies the correct conversion, recalculates affected outputs and checks the decisions that relied on them. The issued correction identifies the affected versions. This example illustrates why explicit definitions can prevent a small data convention error from becoming a large operating error.

Questions about analysis registers

Is a software report a complete analysis record

Not necessarily. Preserve the inputs, configuration, assumptions, processing and interpretation needed to understand the result. The required detail depends on the decision and the software’s own output.

Should every equation be included on the public website

The public article should explain the useful method and its limits. Detailed project registers can remain in the technical handover. Avoid exposing customer data merely to demonstrate that calculations exist.

Does more decimal precision improve confidence

No. Numerical precision should be consistent with the data and method. Extra digits can conceal uncertainty rather than reduce it.

What makes a calculation ready for decision use

Its inputs are traceable, units consistent, method applicable, processing reproducible and result independently checked to the level appropriate to the decision. Material limitations must remain visible to the approver.

Source notes

The benchmark attachment is a directory, not a recovered article body or measurement dataset.

Content-structure method attribution: Eugen Ullrich, eullrich.com, GEO Content Engineering, CC BY 4.0. No ranking or AI citation guarantee is implied.