Custom Machine Design from Specification to Handover

Custom Machine Design from Specification to Handover

A custom machine starts with a vague sentence and ends with a running machine in a customer plant. Between those two points, most projects lose time not on the hard engineering but on the soft agreements: what the machine must do, who decides when it is done, and what happens when the real parts differ from the drawing. This article maps the path of a custom machine project, from the specification to the handover, and names the points where projects get stuck.

The Specification Is a Contract

The specification is the most important document in the project, and it is usually the worst written. The customer says the machine must be reliable. The builder hears the machine must work. Neither sentence is measurable, and both will be argued about at the handover.

The specification must be written in numbers. The cycle time in seconds, the part size range in millimeters, the tolerance on the critical dimension, the changeover time in minutes, the noise limit in decibels. The maintenance access, the safety standard, the utility requirements, and the spare parts policy all belong in the specification.

The rule: if the requirement is not in the specification, it does not exist. The customer who discovers a missing requirement at the handover pays for the change. The builder who adds the requirement without the specification change eats the cost. The specification protects both sides when it is written down.

The Concept Review

The concept review is where the builder proposes the architecture: the automation level, the motion type, the sensor strategy, the guarding. The review compares two or three concepts against the specification, not against the builder’s favorite. The criteria: cycle time, cost, reliability, maintainability, and the risk.

The concept review is also the last cheap change point. After the concept is fixed, the changes get expensive: the layout changes, the drawings change, the parts change. The team that spends the extra day on the concept review saves the week on the shop floor.

The Design Phase Gates

The design phase runs through gates that keep the project honest. The first gate is the layout approval: the machine envelope, the part flow, the operator position, and the utility connections. The second gate is the critical calculation review: the cycle time math, the motor sizing, the frame stiffness, the safety distances. The third gate is the drawing release: the BOM complete, the tolerances set, the supplier parts ordered.

Each gate has a sign-off, and the sign-off has a name and a date. The gate without an owner is a gate that gets skipped, and the skipped gate is where the problem sneaks into the project.

The Procurement Reality

The custom machine lives and dies on the bought parts. The motor with the eight-week lead time, the cylinder that is out of stock, the controller that needs the license: each one has a lead time, and the lead time belongs in the project schedule from day one.

The procurement rule: order the long-lead items at the concept approval, not at the drawing release. The motor, the controller, the gearbox, and the safety components are ordered on the specification, and the drawings catch up. The builder who orders everything at the drawing release is the builder who waits eight weeks at the end.

The Build and the Assembly

The build phase is where the design meets the shop. The frame is welded, the parts are machined, the subassemblies come together. The assembly is where the small design errors surface: the hole that is two millimeters off, the access that is too tight for the wrench, the cable that does not reach.

The assembly should use the model, not the drawing. The assembler with the tablet and the 3D model finds the interference before the part is bolted. The assembler with the paper drawing and the imagination finds the interference after the part is drilled.

The Debugging Sprint

The machine is wired, the software is loaded, and the debugging begins. The debugging is where the schedule dies or survives. The rule: debug one function at a time, and write down the result. The machine that is debugged in order finds the faults quickly. The machine that is debugged by random adjustment finds the faults after the customer finds them.

The debugging log is the seed of the service manual. Every fault, every cause, and every fix goes into the log, and the log becomes the maintenance guide. The team that writes the log while debugging does not have to write the manual from memory later.

The Acceptance Test

The acceptance test is the formal proof that the machine meets the specification. The test runs the machine at the specified cycle time, with the specified parts, over the specified period. The results are recorded and compared to the numbers in the specification.

The acceptance test is a pass or fail against the numbers, not a negotiation about feelings. The machine that meets the numbers passes, even if the paint has a scratch. The machine that misses the numbers fails, even if the paint is perfect.

The Handover Package

The handover is more than the key to the machine. The package includes the as-built drawings, the model, the BOM, the manuals, the spare parts list, the training session, and the warranty terms. The customer who can run, maintain, and repair the machine from the package is a customer who is happy.

The handover also includes the known limitations, written honestly. The machine that cannot run the very thin parts, the material that was not tested, the speed that was not proven. The honest limitation list prevents the call in three months: why does the machine not do this? Because we said it cannot, here, on this page.

The Project Plan That Holds Up

The custom machine project needs a plan that matches the reality of the work, and the plan starts with the milestones that matter.

The first milestone is the specification sign-off. The customer signs the specification, and the signature means the requirements are frozen. The changes after the sign-off go through the change control, and the change control estimates the time and the cost before the change is approved.

The second milestone is the concept approval. The concept review selects the architecture, and the approval locks the direction. The layout, the calculation, and the procurement all follow the concept.

The third milestone is the design release. The drawings and the BOM are released, and the release triggers the procurement of the remaining parts and the start of the fabrication. The release is the point of no return for the design changes that cost the schedule.

The fourth milestone is the assembly complete. The machine is mechanically complete, the wiring is done, and the commissioning starts. The milestone is measured against the plan, and the slippage at this point is the slippage that hits the delivery.

The fifth milestone is the acceptance. The machine passes the acceptance test, and the delivery starts. The acceptance is the formal transfer of the risk from the builder to the customer.

The plan also contains the buffers. The custom machine project always has the unexpected: the part that arrives late, the fault that takes two days to find, the customer change that slips in. The buffer is placed at the end of the phases, not at the end of the project, because the buffer at the end is the buffer that gets eaten by the early delays.

The Communication Discipline

The custom machine project runs on communication, and the communication needs the discipline of the written record.

The weekly status report lists the completed tasks, the open issues, the risks, and the decisions needed. The report goes to the customer and the internal team, and the report is the shared truth of the project. The conversation that happens on the phone is confirmed in writing the same day, because the written confirmation is the record that protects both sides.

The issue log collects every problem with its status. The interference found in the assembly, the motor that is out of stock, the software fault that appears in the debugging: each one enters the log with the owner and the due date. The review of the log at the weekly meeting keeps the issues moving.

The change request is the formal path for every change. The customer asks for the new feature, the change request estimates the time and the cost, and the approval decides. The change that is implemented without the change request is the change that disappears from the record and the invoice.

The risk register names the risks with the probability and the impact. The long-lead motor is the risk that is retired by the early order. The untested material is the risk that is retired by the sample test. The register is reviewed at every gate, and the retired risks are crossed off.

The communication discipline is not bureaucracy. It is the memory of the project, and the memory is what the project needs when the handover comes.

The Lessons From the Failed Projects

The custom machine projects that fail teach the same lessons, and the lessons are cheaper than the tuition.

The first lesson is the vague specification. The project that started with the phrase make it faster and ended with the dispute about the speed. The specification that was written in numbers would have prevented the dispute.

The second lesson is the late change. The customer request that arrived during the assembly, the change that cascaded through the drawings and the parts and the schedule. The change control that requires the estimate before the approval would have made the cost visible.

The third lesson is the hidden assumption. The builder assumed the customer’s part was within the size range, and the customer assumed the machine handled the range. The assumption that was not written down became the failure at the acceptance.

The fourth lesson is the silent supplier. The motor that was ordered with the optimistic lead time, and the delivery that slipped with no warning. The supplier risk that was not tracked in the risk register became the delay at the end.

The fifth lesson is the skipped validation. The calculation that was not checked, the prototype that was not built, and the failure that was discovered at the customer site. The validation that was skipped at the design stage was paid for at the delivery.

The failed projects are the expensive teachers, and the lessons are written into the process. The process review after the project captures the lesson, and the next project starts with the lesson instead of the mistake.

The Quality System and the Custom Machine

The custom machine quality is not the inspection at the end; it is the checks along the way, and the checks are written into the plan.

The first check is the design review at the gate. The review covers the specification compliance, the calculation check, the risk register, and the open issues. The reviewer is not the designer, because the designer sees what the designer expects.

The second check is the incoming inspection. The critical parts are inspected when they arrive, and the nonconformance is caught before the assembly. The shaft that is out of round and the plate that is the wrong thickness are cheaper to catch at the receiving dock than in the machine.

The third check is the assembly checkpoints. The frame is leveled and checked before the subassemblies mount. The alignment is checked before the coupling is bolted. The safety distances are checked before the guarding is fixed. The checkpoint list is the assembly quality plan.

The fourth check is the functional test. The machine runs the dry cycle, the loaded cycle, and the endurance run. The dry cycle finds the sequence faults. The loaded cycle finds the power and the stiffness faults. The endurance run finds the heating and the wear faults.

The fifth check is the documentation. The test results, the calibration records, and the as-built deviations are collected into the handover file. The documentation is the evidence that the machine meets the specification, and the evidence is what the customer accepts.

The quality system for the custom machine is the discipline of the checks, and the discipline is the difference between the machine that works and the machine that is repaired at the customer site.

Conclusion

Custom machine design is a sequence of agreements, and the agreements are written down. Specify in numbers, review the concept early, gate the design, order the long-lead items, debug in order, test against the spec, and hand over the complete package with the honest limitations. The project that follows the path delivers a machine that runs, and a customer who knows what they bought.