Retrofit automation: adding automation to a machine that is already installed and earning

The machine on site is not the machine on the drawing

Surveying belongs before quoting. Electrical drawings will have been amended on the panel door and never in the file, earlier modifications will be undocumented, and the machine will carry parts that no longer match its manual. Spend time and money producing an accurate as-found record before design begins: verified schematics, measured dimensions rather than catalogue ones, an honest account of what the control can offer, and photographs of every enclosure. Integrators quoting from a manual and a brief site visit will price the discovery work as a variation later, by which point the shutdown is already booked.

Mechanical condition sets the ceiling

An operator absorbs a surprising amount of mechanical variation, and automation does not. Worn slides, backlash in a screw, a hydraulic circuit that drifts as it warms, a clamp that no longer repeats: each has been compensated for by somebody adjusting as they went. Once the machine is loaded and started automatically, those turn into scrap. Condition assessment therefore belongs inside the retrofit scope along with whatever mechanical work it identifies, and that work is frequently the largest single line in the project. Where the machine was heading for refurbishment anyway, the case against replacement shifts considerably.

Getting a signal out of a control system you do not own

Older controls range from open and documented through to entirely closed. The best case offers spare inputs and outputs and a documented way to exchange signals. The awkward case is a proprietary control whose supplier sells an interface option, sometimes priced high enough to change the project's economics. The worst case cannot be extended at all, leaving external sensing to infer machine state and external actuation to operate it, which is achievable and always less dependable than a proper handshake. Establish which case applies before committing, because it drives both cost and how well the finished cell will run.

A substantial modification moves responsibility

Adding automation to existing equipment is a modification, and beyond a certain scale it makes the modifying organisation responsible for the resulting machine rather than its original builder. Ask the compliance function what that means in your jurisdiction before work starts. What is unavoidable in engineering terms is that the safety functions were assessed for a machine loaded by a person, and a machine loaded by a robot presents different tasks, different access and different hazards. A fresh risk assessment and validation of the finished installation is required. Support and warranty positions with the original builder need checking rather than assuming.

There is no second machine to develop on

Retrofits are commissioned on production equipment, which removes the safety net a new cell enjoys. No spare machine exists for development and no leisurely ramp is available. Mitigate by building and testing as much as possible away from the machine, simulating its signals on the bench, and staging the work so each shutdown ends with the equipment able to run, manually if necessary. Train operators on the new arrangement ahead of the change rather than during it, and keep manual operation genuinely usable, because a retrofit that removes the ability to run by hand removes your only fallback.

Frequently asked questions

Is retrofitting cheaper than replacing the machine?

Frequently, though not always, and the comparison is often made against the wrong number. A retrofit avoids the machine cost and keeps a proven process, then adds survey work, mechanical refurbishment, control interfacing, a fresh safety assessment and commissioning on live equipment. New equipment arrives with an automation interface designed in, current controls and a warranty. Where the existing machine is mechanically tired or its control is already unsupportable, replacement often wins once both sets of costs are honestly assembled.

What do we do when the original control system is closed?

Ask the builder first, since many sell an interface option far cheaper than engineering around them. Where none exists, external sensing and actuation can substitute: sensors to detect cycle state, an actuator for a door, relays interfaced wherever the control permits. Accept that inferred state is less reliable than a handshake and design the error handling on that basis. Where the machine is old enough that its control is unsupportable anyway, replacing the control within the retrofit is often cleaner.

Does a retrofit need a new risk assessment?

Yes, as an engineering matter whatever the legal position turns out to be. The original assessment described a machine operated a particular way, and automation changes access, tasks, restart behaviour and often the hazards themselves. New interactions appear: recovering a part from a gripper, re-teaching a position, clearing a buffer. Assess the machine as it will be, specify the safety functions that follow, and validate them before production restarts. Handle the question of who becomes legally responsible as parallel work.