Build A Practical Compliance Test Harness For New Machinery
A small test harness gives a business a controlled way to check whether a new machine behaves safely, consistently and as specified before it enters everyday production. It combines test scripts, sensors, simulated inputs, expected results and an evidence record, allowing a small manufacturer to find faults without disrupting an entire line.
For an Australian business, the harness should support both engineering assurance and workplace safety obligations. It does not replace a formal risk assessment, commissioning process or advice from a competent person, but it can make those activities more repeatable and easier to defend during an audit, incident review or customer assessment.
| Approach | Best use | Typical cost | Evidence quality | Main limitation |
|---|---|---|---|---|
| Manual checklist | Simple machines and early inspection | Low | Moderate | Inconsistent results |
| Spreadsheet and simulator | Small automated systems | Low to medium | Good | Needs disciplined version control |
| PLC or software test harness | Repeated functional testing | Medium | Strong | Requires technical skills |
| Digital twin or full virtual model | Complex integrated equipment | Medium to high | Very strong | Can be excessive for a small pilot |
Define The Compliance Boundary
Start by describing exactly what the harness will test. Include the machine’s normal operating cycle, start-up and shutdown behaviour, emergency stop response, guarding, interlocks, sensor faults, loss of power, network failure and recovery after an interruption. A narrow boundary is useful because a small business can complete it properly instead of creating a large document that nobody maintains.
Separate legal and safety requirements from performance targets. A guarding interlock may need to prevent hazardous motion, while a production target may require a cycle time below 30 seconds. Both matter, but they should have different acceptance criteria. Record the source of each requirement, such as the supplier specification, internal risk assessment, contract, or relevant Australian and international standard.
For an Australian workplace, check the expectations of the applicable state or territory regulator, such as SafeWork NSW, WorkSafe Victoria or Workplace Health and Safety Queensland. AS/NZS standards may apply, yet the correct standard depends on the machine type and use. A qualified engineer or machinery safety specialist should review the compliance basis before testing begins.
Choose A Simple Test Architecture
A useful harness has four layers: inputs, control logic, observations and evidence. Inputs can come from push buttons, a PLC simulator, software commands or deliberately triggered fault conditions. Observations may include motor status, guard position, speed, temperature, current draw, alarm codes and time stamps. Evidence consists of screenshots, sensor logs, video, signed check sheets and issue records.
Keep the first version modular. A low-cost industrial computer, USB or Ethernet I/O, a spare sensor, a small control panel and a spreadsheet may be enough for a packaging machine or workshop automation cell. The harness should never bypass a real safety device during live operation. Fault simulation should happen in a safe test mode, with physical isolation and a documented method for returning the machine to service.
Use clear naming for every signal and test. For example, GUARD_DOOR_01_OPEN is easier to trace than “input 7”. Store the harness code, wiring diagram, machine software version and test data in a shared repository with read-only released versions. This is especially important when a local provider makes changes remotely or when several people work across Sydney, Newcastle or regional sites.
Build An Evidence Pack
A compliance test is valuable only when another person can understand what happened. Each test record should identify the machine, configuration, operator, date, software version, preconditions, action, expected result, actual result and disposition. Use “pass”, “fail” or “blocked” rather than vague descriptions such as “looked OK”.
Include negative tests as well as normal operating tests. Open a guard during a cycle, disconnect a non-safety sensor, provide an invalid value, interrupt communications and restart the controller. The objective is to demonstrate that the machine moves to a safe state and provides a clear indication of the fault. Never create a hazardous condition simply to produce a test result.
Evidence can also support wider digital-transformation work. A small firm exploring online customer workflows might take ideas from this online visitor quiz, where a simple digital interaction is structured around measurable user responses. The same principle applies here: define the expected interaction, capture the response and preserve the result.
A practical evidence pack usually contains:
- Approved requirements and risk-assessment references
- Test scripts with unique identification numbers
- Raw logs, photographs and relevant video
- Defect records with owner, severity and due date
Account For Australian Workplaces
A test harness must fit the way the machine will actually be used. In many Australian small businesses, the operator may be a tradie, supervisor or production worker who performs several roles in the same shift. Use plain language, visible status indicators and short work instructions rather than assuming the operator will read a long engineering manual.
Consider local conditions such as dust, heat, condensation, vibration and unstable wireless coverage. A machine in a Western Sydney workshop may experience different conditions from one in a clean Melbourne factory or a coastal Queensland site. Test start-up after a long weekend, recovery after a power interruption and operation during the hottest expected shift, subject to the manufacturer’s safe limits.
Training should include contractors and visiting technicians. Record who has been shown the safe operating procedure and who is authorised to alter the harness. If the business is preparing for an end-of-financial-year investment or a grant-funded automation project, retain quotes, invoices and implementation milestones so the technical evidence matches the financial record.
Use Risk-Based Test Cases
Prioritise tests according to the potential consequence of failure. A fault that could expose a person to moving tooling deserves more attention than a dashboard colour that displays incorrectly. Give each case a risk rating, required protective measures and an approval route for any failed or deferred item.
Useful test groups include:
- Safety functions: emergency stops, guards, interlocks and safe restart
- Process control: sequence, timing, limits, recipe and quality checks
- Fault recovery: power loss, sensor failure, network interruption and jams
- Human interaction: alarms, permissions, instructions and manual mode
Do not rely on a single successful run. Repeat important tests under slightly varied conditions and record the number of cycles completed. For a machine that handles cartons, for example, vary carton size within the approved range and test an empty infeed, a misaligned product and a deliberate stop. Define the acceptable result before running the test to avoid shifting the goalposts.
A small manufacturer can pair this approach with practical automation education. A guide to low-cost automation tools may help identify affordable sensors, data collection devices or simple robotic aids, but every addition should pass through the same risk assessment and verification process.
Pilot, Review And Scale
Run the harness first in a controlled commissioning area with a named test lead. Place the machine in a clearly marked test state, restrict access and hold a short pre-start briefing. The test lead should have authority to stop the work immediately if the machine behaves unexpectedly or a control measure is missing.
Review failed tests by separating the symptom from the cause. A machine that does not stop may have a wiring fault, incorrect logic, a misaligned switch or a poorly defined requirement. Assign corrective actions to specific people, retest the affected case and document the final configuration. A closed defect should show the new result, not just a note saying that someone fixed it.
Once the pilot is stable, convert the harness into a repeatable commissioning package. Keep a small regression suite for every software or hardware change, schedule periodic checks, and train at least two people so the process does not depend on one technician. Contact the Shinagawa City business-support program to explore suitable seminars, vetted technology providers and subsidy pathways for a broader automation or digital-transformation rollout. Start with one machine, capture reliable evidence and use the results to build a safer, more productive operation.