The problem: a shop floor running on stale data
A leading aerospace manufacturer operates precision machining and assembly lines where production schedules are measured in hours and quality tolerances are measured in microns. Yet for years, the people responsible for managing that shop floor — operators, supervisors, plant heads, and planners — were making decisions based on data that was anywhere from two hours to two days old.
The source of that lag was not incompetence. It was architecture. The plant ran SAP as its system of record, and SAP postings happened in batches. Operators would complete work on a job card, set the part aside, and only at the end of a shift — or sometimes the end of a day — would the posting clerk update SAP with actual quantities, scrap, and machine usage. By the time a production manager pulled a WIP report from SAP, the data it contained was a historical record of what had happened, not a live picture of what was happening.
The consequences were predictable. Supervisors would schedule new jobs onto machines that were already occupied. Scrap was discovered at the next operation rather than at the source. Mould maintenance was triggered by failure rather than by condition. Quality checks happened at shift end rather than hourly. Management dashboards were built in Excel from data exported manually every Monday morning, so strategic decisions about capacity and bottlenecks were being made on week-old information.
The ask was straightforward, even if the execution was not: build a system that shows what is actually happening on the shop floor, right now, and keep it integrated with SAP so that the ERP remains the system of record.
The solution: KaiNext MES as the real-time production layer
The deployment of KaiNext MES introduced a dedicated production control layer that sits between the shop floor and SAP. Rather than treating the ERP as a real-time system — which SAP is not designed to be — the MES owns all real-time transactions and synchronises with SAP bidirectionally on a defined cadence and event basis.
At its core, the MES is a work-in-progress tracking engine. Every job card that SAP releases to the shop floor is pulled into the MES. From that point, every move of a part — from raw material to first operation, through successive machining and inspection stages, to the finished goods store — is recorded against the job card in real time. The planned BOM quantity is compared against actual consumption at each operation. Scrap is logged at the point of occurrence, not at the end of the day, and the system immediately recalculates whether the job can still fulfil its order quantity or whether a shortfall needs to be flagged to the planner.
Hourly machine capacity output is captured automatically from the IIoT layer, giving supervisors a live view of how each work centre is performing against its target rate. Operators log downtime events — breakdowns, setup changes, material unavailability — directly from handheld terminals, and the MES classifies these against SMED and maintenance categories. MTTR and MTBF are calculated on live data, not on maintenance logs filled in retrospectively.
Quality checks are embedded in the workflow rather than appended to it. At defined inspection gates, operators record dimensional and attribute checks against the part drawing. The MES holds the inspection plan — which checks are required, what the acceptance criteria are, whether the part can proceed or must wait for a reviewer. Online quality checks replaced the previous model of batching parts for a quality department visit at shift end.
IIoT built in-house: the reliability decision
The most unusual aspect of this deployment was the decision to manufacture the IIoT devices in-house rather than procuring them from an industrial IoT vendor.
The reasoning was grounded in a specific operational concern: an aerospace shop floor is a harsh environment. Coolant spray, metal swarf, vibration, and temperature swings are routine. Commercial IIoT hardware designed for general industrial use tends to be over-specified on connectivity features and under-specified on physical durability for this environment. More practically, when a sensor fails on a machining centre at 2 AM on a Sunday, the question is not whether a replacement is theoretically available — it is whether it is available in the on-site spares cabinet.
By designing the IIoT devices in-house, the team controlled the hardware bill of materials, built the devices to the specific environmental requirements of this shop floor, and maintained a full spares inventory on-site. Firmware updates are managed centrally through the MES. The result is a hardware layer that has achieved the same 99.8% uptime as the software layer — because both layers were designed together for this specific environment.
Barcode printing is integrated throughout the flow. When a job card is released, barcode labels are printed for the traveller cards that accompany each batch through operations. Handheld terminals at each work centre allow operators to scan the traveller, log the operation, record quantities and scrap, and move on — without requiring a workstation at every machine.
The weekly Excel dashboard took half a day to produce and was obsolete before it was distributed. The MES replaced it with a live screen that anyone in the plant can pull up at any time.
SAP integration: bidirectional, not one-way
The integration with SAP was designed from the start to be bidirectional. This distinction matters more than it might appear. Many MES deployments treat the ERP as a data source — pulling work orders, BOMs, and routings — and then write back only summary confirmations at the end. This architecture creates a reconciliation problem: if anything goes wrong during production, the MES knows about it but SAP does not, and the data drifts apart.
In this deployment, every material movement in the MES generates a corresponding posting in SAP as it happens. Goods issues against production orders, scrap bookings, quality inspection results, and process order confirmations all flow back to SAP in real time. The SAP production order is always current. Production managers can run their existing SAP reports and get live data, because the MES is posting to SAP continuously rather than in an overnight batch.
In the other direction, SAP pushes master data changes — engineering change orders, revised BOMs, new routings, updated standard costs — to the MES as soon as they are approved. The shop floor is always working from the current engineering standard, not from a printout that may have been superseded.
Mould maintenance and SMED monitoring
Two operational areas that benefited disproportionately from real-time visibility were mould maintenance and setup time reduction.
Moulds in aerospace manufacturing are expensive, long-lead-time assets. Preventive maintenance intervals are defined by the number of shots or cycles rather than by calendar time. Before the MES, tracking mould usage was a manual counting exercise — a sheet on the press, updated by the operator, reviewed by the toolroom weekly. Moulds regularly exceeded their maintenance interval because the count was never quite accurate.
The MES tracks every cycle on every mould in real time. Maintenance alerts are triggered automatically when the mould approaches its service interval — with enough lead time for the toolroom to schedule the work during a planned break rather than in response to a failure. Since deployment, not a single mould has exceeded its maintenance interval by more than five cycles.
SMED monitoring follows a similar pattern. Setup time — the time from last good part of one job to first good part of the next — is now captured automatically from the machine state data. Supervisors can see setup time distributions by machine, by operator, and by job type. This data fed a focused improvement programme that reduced average setup time by 18% over the first eighteen months of operation.
Results: from weekly reports to real-time dashboards
The shift from weekly Excel reports to real-time MES dashboards is the most visible change, but it is not the most important one. The more significant change is that problems are now visible at the moment they occur, which means they can be addressed before they cascade into larger issues.
- Planned vs actual visibility: Every work centre shows its target output rate alongside its actual output, updated every minute. Supervisors can identify underperforming stations within the hour rather than at shift review.
- Scrap tracking automated: Scrap is logged at the operation where it occurs. BOM vs actual scrap comparison runs continuously. Scrap cost is allocated to the correct cost centre in SAP automatically.
- Quality checks hourly: Online inspection gates ensure that dimensional and attribute checks happen at defined intervals throughout the shift. End-of-batch quality failures — where an entire batch has to be scrapped because an issue was caught too late — have been eliminated.
- OEE and downtime: Shift-wise OEE is calculated and displayed on shop floor screens in real time. Downtime events are categorised by cause, enabling targeted improvement prioritisation.
- Weigh scale automation: Weigh scales are integrated directly into the MES. Material quantities are captured from the scale and posted to the job card without operator transcription, eliminating a significant source of recording errors.
The system has run for more than three years without a single unplanned outage. The 99.8% uptime figure is not a projection — it is the operational record. For a shop floor where production schedules are designed around system availability, this reliability has been as significant as any of the functional improvements.