Written by Commissioning is the stage when a custom machine moves from an engineered concept to an operating production asset. Mechanical adjustments, control logic, sensors, tooling, and operator procedures must work together under realistic conditions. Data collection gives the commissioning team a structured way to understand what is happening and decide what needs to change.
Without suitable data, teams may rely on isolated observations or memory. With it, they can distinguish a one-time event from a repeatable issue, compare operating conditions, and create a clearer basis for final acceptance and future maintenance.
What data should be collected during commissioning?
The right data depends on the machine and process, but a commissioning plan should normally cover four areas: machine behavior, process performance, quality results, and interventions made by the team.
Machine and control data
Machine data helps confirm whether the equipment is operating as designed. Useful records may include:
- Cycle time and the duration of individual process steps
- Start, stop, reset, and fault events
- Sensor states and actuator commands at critical sequence points
- Servo position, speed, torque, or force values where relevant
- Pneumatic pressure and motion-related observations
- Temperature, vacuum, or other process variables used by the machine
- Alarm messages and the conditions that triggered them
Not every signal needs to be recorded continuously. The aim is to capture information that helps explain sequence behavior, repeatability, or abnormal conditions. Excessive data without a clear purpose can make troubleshooting more difficult rather than easier.
Process and quality data
Commissioning should also show whether the machine is producing acceptable results. Depending on the application, this may include inspection results, test values, reject categories, missing or incorrect assembly conditions, marking verification, leak-test results, or electrical functional-test outcomes.
Each result should be connected to useful context, such as the machine cycle, product variant, station, tooling condition, or recipe. This connection is especially important when several product types or operating modes share the same equipment. A general approach to reliable measurement in test systems can help teams think about how measurement data should be interpreted and controlled.
Define the purpose before choosing signals
A common commissioning mistake is to begin by collecting every available PLC tag. A better approach is to start with questions that the data must answer.
- Is the sequence completing in the intended order?
- Which step is limiting the cycle time?
- Does a fault occur under a repeatable condition?
- Can the machine distinguish a component, product, or recipe variant correctly?
- Are quality results changing with speed, force, pressure, temperature, or tooling position?
- Can maintenance personnel identify the cause of a stop from the available records?
Once these questions are defined, the team can identify the required signals, sampling approach, timestamps, and storage method. This prevents the data collection system from becoming an uncontrolled collection of values that no one uses.
Build a commissioning data plan
A simple data plan creates consistency between mechanical, electrical, controls, production, and quality teams. It can be maintained as a working document throughout commissioning.
1. List the critical process steps
Break the machine cycle into meaningful stages: loading, part presence confirmation, clamping, joining, pressing, testing, marking, unloading, and any other application-specific operations. For each stage, define what must happen, what can go wrong, and which signals can confirm the condition.
2. Assign measurable acceptance criteria
Acceptance criteria should describe the intended result in a way that can be checked. Examples include sequence completion, presence of a required confirmation, a defined test decision, or operation within an approved process window. Criteria should come from the project requirements and process needs rather than being invented after an issue appears.
3. Record changes and interventions
Commissioning often involves adjustments to sensor positions, timing values, motion profiles, tooling, gripper settings, or software logic. Record each significant change with its date, reason, affected station, and observed result. Otherwise, a later improvement may be confused with an earlier setting, making it difficult to identify which change solved or caused a problem.
4. Give every trial a clear identity
Trial records should identify the product or test article, machine state, recipe or program, operator or team where appropriate, and relevant tooling configuration. This does not require complicated software. A consistent naming convention and a controlled worksheet can already prevent confusion between test runs.
Use timestamps to connect events
Data becomes more useful when events can be placed on the same timeline. A fault message, a sensor transition, a motion command, and a quality result should be traceable to the same cycle or test attempt whenever possible.
Consistent timestamps help answer practical questions. Did the sensor fail to detect the part, or did the actuator stop before reaching position? Did a quality deviation occur before or after a recipe change? Was the alarm caused by a process condition or by a communication interruption?
For machines with several stations, timestamps also support comparison between stations. They can reveal waiting time, repeated retries, unnecessary pauses, or a station that regularly extends the cycle. This is more informative than evaluating only the average cycle time.
Turn raw data into commissioning decisions
Data collection has value only when it leads to a decision. During daily reviews, organize observations into a small number of categories:
- Design issue: the machine or tooling does not support the required process reliably.
- Control issue: the sequence, interlock, timing, or fault handling requires revision.
- Material or product issue: incoming variation affects loading, assembly, testing, or handling.
- Method issue: the work instruction, setup, or operator interaction is unclear.
- Measurement issue: the sensor, test method, calibration status, or data interpretation needs review.
This classification encourages the team to address causes rather than repeatedly resetting the machine. It also helps separate a problem that requires engineering work from one that can be resolved through setup or instruction.
Connect commissioning data with traceability
When a machine produces several units or variants during commissioning, linking results to an identifiable cycle is valuable. Traceability does not mean storing every possible signal forever. It means retaining the information needed to understand what happened to a product or process at a useful level of detail.
Depending on the project, a traceability record may include a product identifier, recipe, station result, test decision, fault status, and selected process values. The required scope should be agreed with production and quality teams before implementation. For a broader view of the topic, see this practical guide to production line traceability.
Good traceability also supports troubleshooting after handover. If a quality concern appears later, the available record can help determine whether it is related to a product variant, machine condition, tooling change, or process setting.
Prepare data for FAT, SAT, and handover
Commissioning records should not remain as temporary notes. They can provide useful evidence for factory and site acceptance activities, provided that the test conditions and results are clearly documented.
Before handover, review whether the records show:
- Completion of the agreed functional checks
- Stable operation across the intended product or recipe range
- Known faults, open actions, and their current status
- Final parameter values and approved software versions
- Quality and test results relevant to the process
- Information required by operators, maintenance, and quality personnel
The same discipline supports maintainability. Alarm histories, fault descriptions, adjustment records, and clear recovery steps can reduce the time needed to diagnose future stops. Data should therefore be considered alongside access, documentation, spare parts, and service procedures when designing custom machines for easier maintenance.
Common data collection mistakes
- Collecting values without a question: signals are stored but never linked to a troubleshooting or acceptance need.
- Using unclear names: technicians cannot tell which station, product, or event a value represents.
- Ignoring manual interventions: adjustments and resets are not recorded, so trial results become difficult to compare.
- Recording averages only: short interruptions and repeated retries disappear inside an average value.
- Separating quality from machine data: test results cannot be connected to the cycle conditions that produced them.
- Failing to control revisions: changes to programs, parameters, or tooling are not associated with the recorded result.
Conclusion
Data collection during custom machine commissioning is a practical engineering tool, not an administrative exercise. It helps teams verify sequences, locate delays, investigate faults, assess quality, and document the machine’s final condition.
The most effective approach is selective and purposeful: define the questions, capture the signals that answer them, identify each trial, record interventions, and review the results with the people responsible for production, quality, controls, and maintenance. When commissioning data is organized in this way, it supports a more reliable handover and gives the operating team a stronger foundation for future improvement.