18.1 Computer Functions, Office Applications & Data Governance

Computer applications support nearly every mining decision, but software does not create reliable information automatically. A system receives input, performs processing, preserves storage, sends output or data movement, and uses control to coordinate operations. A fleet-dispatch system, for example, receives GNSS and payload inputs, calculates assignments, stores events, sends instructions, and applies permissions and logic.

Office Applications by Purpose

Word Processing

Use controlled templates, headings, cross-references, captions, citations, revision marks, comments, and approval metadata for reports and procedures. A report exported to PDF should retain the source, revision, approver, and appendices. Track changes is review evidence, not final content; accept or resolve edits deliberately.

Spreadsheets

Spreadsheets are effective for transparent calculations and exploration but fragile when formulas, units, and overrides are uncontrolled. Structure a workbook:

1. instructions and scope;
2. source inputs with units and references;
3. calculations separated by topic;
4. checks and reconciliation;
5. outputs and sensitivity; and
6. change log and approval.

Avoid hard-coded values inside formulas. Use data validation, named ranges cautiously, protected cells, consistent signs, and error handling that does not hide failures. Inspect formulas across rows and columns for discontinuity. Test limiting cases and reproduce a sample independently.

Presentations

A presentation summarizes a decision; it should not conceal uncertainty with simplified graphics. Label data dates, units, assumptions, and source. Use a clear takeaway, evidence, alternatives, risk, and requested decision. Keep the technical report as the auditable record.

Databases

Relational databases store entities in linked tables with keys, types, constraints, and controlled transactions. A drillhole system separates collar, survey, geology, sample, assay, density, and QA/QC tables linked by identifiers and intervals. Constraints can prevent overlapping samples or assays without a sample record. A spreadsheet copy is not a substitute for the controlled database.

Data Quality Dimensions

• completeness: required records exist;
• validity: type and range conform;
• consistency: related fields agree;
• uniqueness: duplicates are controlled;
• accuracy: values reflect the source or reference;
• timeliness: data are current for the decision; and
• lineage: origin and transformation are traceable.

A zero, blank, not sampled, below detection, and not applicable are different states. Encoding them all as zero corrupts analysis.

Governance and Access

Assign a data owner accountable for meaning and quality and a custodian responsible for technical management. Define authoritative repositories, naming, metadata, permissions, retention, and approval. Use least-privilege access: users receive only permissions needed for their role. Separate development, test, and production systems where practical.

Version control should identify who changed what, when, why, and under which approved request. Hashes or signed releases can confirm that the model reviewed is the model used.

Backup and Recovery

A backup exists only if it can be restored. Use multiple copies, different media, and an off-site or logically isolated copy proportionate to risk. Define recovery-point objective (acceptable data loss) and recovery-time objective (acceptable outage). Test restoration, including application configuration and access—not just raw files.

Cybersecurity and Operational Technology

Phishing, ransomware, weak remote access, shared accounts, removable media, and unpatched systems can disrupt dispatch, ventilation, pumping, processing, or laboratory data. Segment operational technology from business networks, control vendor access, use multi-factor authentication where appropriate, log changes, and maintain manual safe-state procedures. Safety systems should fail predictably when communications are lost.

Spreadsheet Example

A production workbook calculates tonnes as volume times density. One row uses density in kg/m3 while other rows use t/m3, overstating tonnes by 1,000. Dimensional checks and a range rule would flag the result. A chart that looks smooth does not validate the underlying units.

Exam Trap

The most sophisticated software is not necessarily the correct answer. Select the application that matches the data structure and decision, then state validation, access, backup, and review controls.

Formula Assurance

For a high-impact workbook, use a second person to inspect formula consistency, named ranges, external links, circular references, hidden rows, macros, and manual overrides. Reconcile control totals after every import. Lock the approved release as read-only and distribute a checksum or controlled link. If users email editable copies, the organization soon has several plausible sources of truth.

Responsible Automation

Scripts and macros should have an owner, purpose, input schema, tests, error logging, and change control. They must fail visibly on missing or invalid input instead of converting an error to zero. Retain enough documentation for another competent person to reproduce the process.