FAT and SAT for Robotic Welding Cells

Acceptance Testing of a Robotic Welding Cell: What FAT and SAT Must Demonstrate Before Serial Production

Posted on

A robotic welding cell can produce an excellent demonstration weld and still not be ready for real production. A single carefully selected component does not prove that the system will maintain the required quality across different batches of workpieces, repeated loading, operator changes, consumable wear and normal production variation.

This is why acceptance should not be reduced to the question, “Does the robot move and weld?” It must demonstrate that the complete system, including the robot, welding power source, fixtures, positioners, sensors, controls, safeguards, documentation and people, meets predefined requirements under conditions representative of future production.

The two main stages are normally FAT (Factory Acceptance Test) and SAT (Site Acceptance Test). They serve different purposes and should not be treated as a formal repetition of the same test.

FAT and SAT Begin with the Technical Specification

IEC 62381:2024 establishes a general framework for FAT, factory integration tests (FIT), SAT and site integration tests (SIT) for automation systems in the process industry. The standard is not specific to robotic welding, but its central principle is fully applicable: acceptance testing should demonstrate conformity with the relevant specification, while the scope, activities and responsibilities should be agreed in advance between the customer and the supplier.

This means that effective acceptance begins when the technical specification is prepared. If the requirement is described only as “robotic welding of product X”, there will be no objective basis at the end of the project for deciding whether the cell can be accepted.

For every key performance indicator, the parties should define:

  • the specific requirement;
  • the test method;
  • the parts, tools and measuring equipment to be used;
  • the conditions under which the test will be conducted;
  • the required number of repetitions;
  • the pass/fail criterion;
  • the responsible party;
  • the record that will serve as evidence;
  • the method for handling deviations.

This turns terms such as “good quality”, “short cycle time” and “easy changeover” into verifiable requirements.

What Is the Difference Between FAT and SAT?

FAT is usually conducted at the integrator’s facility before the cell is dismantled and transported. The aim is to identify as many technical and functional issues as possible while the engineering team, tools and replacement components are readily available.

FAT can cover:

  • completeness of the system against the agreed scope;
  • robot motion, reachability and collision-free operation;
  • operation of fixtures, positioners and sensors;
  • the automatic cycle sequence;
  • signal exchange between the robot, PLC, HMI and welding power source;
  • programs, recipes and product-variant management;
  • weld quality using the agreed test components;
  • preliminary cycle time and productivity;
  • diagnostics, fault messages and recovery procedures;
  • primary safety functions, to the extent that the pre-delivery configuration permits full verification;
  • documentation, backups and training materials.

SAT is conducted after installation at the customer’s plant. It confirms that the cell retains its agreed performance in the final environment and operates correctly with the actual infrastructure and production interfaces.

On-site verification may include:

  • electrical power, compressed air, shielding gas, ventilation and fume extraction;
  • final positioning, levelling and anchoring;
  • integration with upstream and downstream operations;
  • part infeed and outfeed;
  • data exchange with other machines, MES or production systems;
  • operation of the safeguarding system in the final layout;
  • cycle time under the actual work organisation;
  • quality using normal production components;
  • the customer team’s ability to operate, set up, clean and return the system to service.

SAT should not compensate for an incomplete FAT. Changes that can be identified and corrected before transport generally become slower and more expensive once the cell has been installed in production.

One Successful Weld Does Not Prove Production Readiness

The easiest way to make a cell appear ready is to test it with a perfectly prepared component, carefully cleaned surfaces and the most experienced programmer standing at the control panel. This demonstrates that the process is possible, but not that it is robust.

Acceptance testing should include representative production variation:

  • components from different batches;
  • permissible extremes of dimensions and joint gaps;
  • realistic edge and surface conditions;
  • different product variants;
  • repeated unloading and reloading;
  • fixture or recipe changeovers;
  • normal wear or replacement of contact tips and other consumables;
  • operation by trained production staff, not only by the integrator’s team.

A consecutive production run using a predefined number of components is particularly valuable. It exposes problems that a single demonstration may not reveal: spatter accumulation, fixture heating, TCP drift, wire-feeding interruptions, unstable part detection or a gradual increase in manual intervention time. As discussed in our article on robotic welding fixtures, robot repeatability creates value only when the workpiece is also located and clamped predictably.

If statistical process-capability indicators are used, the measured characteristic, sample size and analysis method should be agreed in advance. There is no universal value that applies to every cell and every product.

Quality Must Be Assessed Against the Actual Product Requirements

A visually acceptable weld is not a sufficient criterion. Quality must be verified against the drawing, the applicable welding procedure, the agreed acceptance level and the required inspection methods.

The protocol may include:

  • identification of the product and individual welds;
  • the WPS applied and the associated WPQR/PQR or other qualification basis, where relevant;
  • material, thickness, joint type and welding position;
  • set and actual process parameters;
  • visual inspection results and any required additional testing;
  • critical dimensions and distortion after release from the fixture;
  • deviations, corrective actions and retest results.

ISO 3834-2:2021 places welding within a framework of comprehensive quality requirements, while ISO 17662:2025 addresses the calibration, verification and validation of equipment used to control process variables. The latter also includes guidance on these activities as part of acceptance testing for equipment used in welding and allied processes.

The practical question is not only whether the correct values appear on the display, but also whether the measurement and control of critical parameters are sufficiently reliable for the specific application.

Cycle Time Must Be Measured from Start to Start

When productivity is evaluated, attention is often focused only on the time during which the arc is active. For the manufacturer, however, the complete production cycle is what matters.

It may include:

  • loading and correctly positioning the workpiece;
  • sensor confirmation and clamp closure;
  • locating the weld start or scanning the seam;
  • robot and positioner motion;
  • welding and waiting between passes;
  • torch cleaning and consumable replacement;
  • inspection activities;
  • unclamping and unloading;
  • changeover between variants;
  • foreseeable manual actions and process pauses.

The measurement should therefore begin and end in the same clearly defined state, for example, from readiness to load one component to readiness to load the next. Arc-on time, automatic time, operator time and downtime should be recorded separately. This reveals where the real constraint lies.

A cell with very fast robot motion may still fail to achieve the required takt time if the operator must wait for a shared area to be released, the fixture requires frequent cleaning or a product change requires manual recalibration. In two-station systems, it must also be confirmed that loading can genuinely take place in parallel with welding without conflicting with the safety functions.

Safety Is Not Accepted by Pressing the Emergency Stop Once

ISO 10218-2:2025 addresses safety during the integration, commissioning, operation and maintenance of industrial robot applications and robot cells. Risk assessment in accordance with ISO 12100 and validation of the safety-related parts of the control system are essential to demonstrating that the cell is ready for operation.

Acceptance should cover every defined safety function, not merely the presence of a fence and emergency-stop devices. Depending on the application, this may include:

  • interlocking devices on gates and guards;
  • light curtains and laser scanners;
  • emergency stop;
  • protective stop;
  • selection and control of operating modes;
  • safely limited speed during setup;
  • behaviour when someone enters the loading area;
  • prevention of unexpected restart;
  • response to loss and restoration of electrical, pneumatic or hydraulic power;
  • controlled recovery after a fault;
  • safe access for cleaning, setup and maintenance.

ISO 13849-1:2023 provides a methodology for the design and integration of safety-related parts of control systems. Testing must confirm not only the logical response, but also that each safety function has been implemented and validated against its required Performance Level (PLr) and the intended system architecture.

It is equally important to verify behaviour under realistic error conditions: an open clamp, missing part, incorrect recipe, interrupted sensor signal or an attempted restart from an inappropriate state. A safe cell should not only stop; it should also provide clear diagnostics and a predictable recovery procedure.

FAT and SAT provide evidence within the project, but they do not replace the risk assessment, technical documentation, conformity assessment or any other applicable obligation of the manufacturer or integrator.

Software and Backups Are Also Part of Acceptance

A modern cell contains several interdependent software layers: robot programs, PLC logic, HMI software, welding power-source recipes, sensor configurations, safety parameters and, in some cases, connections to external information systems.

Before acceptance, a validated baseline configuration should be established. It should include:

  • versions of all applicable programs and configurations;
  • complete backups in a usable format;
  • a verified restoration procedure;
  • a list of users, roles and access levels;
  • rules for changing and approving programs;
  • a defined link between product variant, robot program and welding recipe;
  • a record of corrections made during FAT and SAT;
  • licences, handover credentials and any required service tools included in the agreed scope.

A backup that has never been restored is not yet a proven recovery strategy. At least the critical backups should be verified before final handover.

Maintenance and Fault Recovery Must Be Demonstrated

Production readiness does not mean only successful automatic operation. The customer team must be able to return the cell to service following common interruptions, without improvisation and without introducing new risks.

Practical scenarios include:

  • replacing a contact tip or welding wire;
  • cleaning and inspecting the torch;
  • checking or restoring the TCP;
  • removing an incorrectly loaded component;
  • recovering from an interrupted cycle;
  • restarting after a power failure;
  • replacing a worn fixture element;
  • periodically checking sensors and safeguarding devices;
  • reverting to the last validated software version.

For each activity, the required permissions, safe state, tools, sequence and pre-production verification should be clear. These routine interventions often determine the cell’s actual availability more than the robot’s nominal speed does.

Training Must Be Verified Through Practical Action

An attendance sheet does not prove that the team can operate the system. During SAT, operators and technical personnel should demonstrate the key activities associated with their respective roles.

An operator may need to demonstrate correct loading, product selection, start-up, response to common messages and safe cleaning. A weld setter may need to demonstrate program changeover, controlled correction, TCP verification and recovery after an interruption. Maintenance personnel may need to demonstrate diagnostics, component access, backup procedures and work in accordance with the required isolation procedures.

ISO 14732:2025 defines qualification requirements for welding operators and weld setters involved in mechanised and automatic welding of metallic materials. Its applicability depends on the contract and the relevant product standard or regulatory framework. The standard does not apply to personnel who do not control or adjust welding parameters and are not involved in setting up the welding equipment. This distinction is important when roles and competence requirements are defined.

What Documentation Should Remain After Acceptance?

A cell has not been fully handed over if the knowledge required to operate it remains only with the programmer or integrator. The final documentation package should reflect the agreed scope and applicable requirements and may include:

  • final mechanical, electrical and pneumatic drawings;
  • the layout and component list;
  • the risk assessment and description of protective measures;
  • the list of safety functions and their validation records;
  • declarations and conformity documentation corresponding to the applicable role;
  • instructions for operation, setup, cleaning and maintenance;
  • robot programs, PLC/HMI projects, recipes and backups;
  • WPS documents, test results and records for the accepted components;
  • a list of spare and wear parts;
  • a schedule for preventive maintenance, inspection and calibration;
  • training materials and training records;
  • FAT and SAT protocols, a list of outstanding items and evidence that each deviation has been closed.

The documents must describe the configuration actually delivered, not an earlier design revision.

A Practical Plan for Robotic Welding Cell Acceptance

A robust approach can be structured into the following steps:

  1. Freeze the principal requirements. Define the products, variants, quality, cycle time, shifts, interfaces and responsibilities.
  2. Prepare an acceptance matrix. Link every requirement to a test, criterion, item of evidence and FAT or SAT stage.
  3. Provide representative components. Include actual production variation, not only specially prepared samples.
  4. Verify functions without welding. Test motion, signals, interlocks, recipes, diagnostics and recovery.
  5. Validate the welding process. Produce the agreed welds and perform inspection and measurement against the applicable requirements.
  6. Measure the complete cycle. Include operator actions, sensing, positioning, cleaning and changeover.
  7. Conduct a consecutive production run. Verify robustness across a series of components and normal manual interventions.
  8. Validate safety. Test every defined safety function and reasonably foreseeable error condition.
  9. Fix the baseline configuration. Back up the verified programs, parameters and documentation before transport.
  10. Repeat the critical checks during SAT. Confirm installation, infrastructure, integration, quality and cycle time at the actual production site.
  11. Verify team competence. Require operators and maintenance personnel to perform real tasks without continuous assistance from the integrator.
  12. Close deviations with evidence. Every correction should be retested and reflected in the final documentation.

More complex production lines may require separate integration tests between subsystems before overall acceptance. The test structure should follow the real system architecture and the agreed allocation of responsibilities.

Acceptance Is the Bridge Between Engineering and Production

Well-conducted FAT and SAT are not merely the administrative conclusion of a project. They are the point at which design assumptions are tested against real components, real operators and the actual production environment.

When criteria are defined early, acceptance reduces disputes and directs the engineering team towards a measurable outcome. When left until the final day, it can easily become a demonstration under ideal conditions, after which unresolved problems are transferred to production.

A robotic cell is not ready simply because it has produced one good component. It is ready when the system can safely and repeatedly deliver the agreed quality at the required cycle time, with a trained team and a clear technical basis for maintenance and future modifications.

Bullitt Robotics designs, integrates and commissions robotic systems based on the actual products, production variations and acceptance criteria of industrial companies. If you are planning a new welding cell or preparing the acceptance of an automated system, contact our team at +359 89 667 0392 or office@bullitt-engineering.com.

Sources

Similar Posts