Connecting manufacturing execution to the machines is not a data collection exercise. It is a set of decisions about where material identity survives and where it does not.
A machine count without an order, a product, a shift and an event definition is not yet a production record. The signal has to carry the business context that makes it usable.
This table is the honest answer to a recall question, and it is the thing to establish before any traceability work is scoped. Most plants carry several of these forms on one line.
The resolution of the whole chain is the resolution of its weakest link, so one hopper in the middle of an otherwise serialised process caps everything downstream of it.
Disconnected MES and shop-floor equipment
MES holds the order and the operator types what happened. The machine already knows, and the typing is a transcription step with an error rate and a time cost.
Manual production data collection
Counts and times written down and keyed later. Accurate to the nearest convenient number, and systematically rounded in the direction that makes the shift look consistent.
Inconsistent machine and production event definitions
Two lines count a stoppage differently, so their downtime is not comparable. This is the second position on this page and it is addressed in section 3.
Limited production order execution visibility
The order is known to be open and nothing more. Which operation it is at, how much is complete and whether it is paused or running are all unavailable until close.
Fragmented PLC, SCADA and historian environments
Several generations and several vendors across one plant, each with its own tag naming and its own idea of what a state means. The connectivity work itself belongs to Industrial IoT & OT Integration.
Legacy equipment connectivity limitations
A share of machines expose nothing useful. That share is a cost and a lead time to be quantified early rather than an obstacle.
Inconsistent production data across lines and plants
The same measure built from different definitions at each site. Comparable reporting requires one event model, which is a governance deliverable rather than a technical one.
Delayed production status updates
Status arriving at the end of the shift is a record. Status arriving during the shift is usable, and the response design for that belongs to Production Monitoring & Optimization.
Complex MES deployment and integration requirements
MES touches orders, materials, quality, equipment and people at once, which is why MES programmes are long. Scoping by product family rather than by module is what makes them finish.
Manufacturing system reliability and downtime concerns
Once the line cannot be confirmed without MES, MES availability is production availability. That changes what the system is and what it needs, and it is covered in section 7.
Machine integration looks like reading a signal. It is actually writing a definition. Whatever the integration decides becomes the plant record, and the controller has no opinion to appeal to.
Step 1
MES and shop-floor architecture assessment
What MES holds today, what the equipment layer exposes, and which transactions currently depend on a person transcribing something.
What MES holds today, what the equipment layer exposes, and which transactions currently depend on a person transcribing something.
Step 2
Manufacturing workflow and equipment mapping
Operations, routings, stations and the equipment at each, mapped from the floor rather than from the drawing, because the drawing is usually a revision behind.
Operations, routings, stations and the equipment at each, mapped from the floor rather than from the drawing, because the drawing is usually a revision behind.
Step 3
Machine data and production event discovery
What each machine can signal: states, counts, cycles, alarms, parameters. Documented per machine type, since one type done once is reusable and a new type is not.
What each machine can signal: states, counts, cycles, alarms, parameters. Documented per machine type, since one type done once is reusable and a new type is not.
Step 4
Shop-floor integration architecture design
Where events are detected, where they are modelled and where they are stored, with the enterprise boundary stated.
Where events are detected, where they are modelled and where they are stored, with the enterprise boundary stated.
Step 5
Equipment connectivity and protocol assessment
Which machines are reachable, by what means, and what share needs added instrumentation. The secure connectivity design is covered on the Industrial IoT & OT Integration page.
Which machines are reachable, by what means, and what share needs added instrumentation. The secure connectivity design is covered on the Industrial IoT & OT Integration page.
Step 6
MES data model and transaction mapping
The event model: every state, every transition and the boundary condition for each. This is the deliverable that makes two lines comparable and it is the stage most often shortened.
The event model: every state, every transition and the boundary condition for each. This is the deliverable that makes two lines comparable and it is the stage most often shortened.
Step 7
MES to equipment and enterprise integration implementation
Downward to the machines, upward to ERP. The ERP side is specified jointly with the business transaction decisions on the ERP & Manufacturing Systems Integration page.
Downward to the machines, upward to ERP. The ERP side is specified jointly with the business transaction decisions on the ERP & Manufacturing Systems Integration page.
Step 8
Production event validation and reconciliation
Captured events checked against observation on the line and against the physical count. A system that is internally consistent and does not match the floor is worse than no system.
Captured events checked against observation on the line and against the physical count. A system that is internally consistent and does not match the floor is worse than no system.
Step 9
Pilot deployment and operational testing
One line, one product family, through a full cycle including a changeover, a stoppage, a rework and a shift handover. Those four are where event models fail.
One line, one product family, through a full cycle including a changeover, a stoppage, a rework and a shift handover. Those four are where event models fail.
Step 10
Multi-line and multi-plant integration rollout
By machine type and by product family, with the event model fixed and the configuration varying. A rollout that renegotiates the event model per line has no event model.
By machine type and by product family, with the event model fixed and the configuration varying. A rollout that renegotiates the event model per line has no event model.
For each one, which side is authoritative is stated. Most are read-only from the plant. The few that are not are the ones that need an engineering and quality review.
MES to PLC integration
Reading machine state, counts and cycle data. Read-only. Any write to a controller is out of scope for what we build and would be an engineering change on your side.
MES to SCADA integration
Reading process state and alarms where SCADA is the aggregation point, which is often the lighter-touch route to the same information.
MES to industrial historian integration
Reading recorded signals for confirmation, parameter capture and investigation. The least invasive integration available and the first one to try.
MES to ERP integration
Orders and master data down, confirmations and consumption up. ERP owns the plan, MES owns the execution record. The transaction decisions behind this are on the ERP page.
Production order dispatch and status integration
Orders reaching the line in the sequence the plant will actually run, with status returning at the chosen confirmation granularity.
Machine status and production event capture
States, transitions, counts and stoppages captured against the agreed event model, with the machine as the authority on what it did.
Material consumption and production confirmation
What was consumed and what was produced, with backflush behaviour stated explicitly because backflush is where plant and ERP inventory diverge.
Shop-floor quality data integration
In-process checks captured at the station and holds raised where they can actually stop movement, rather than recorded for later.
Manufacturing traceability and genealogy
The chain built to the resolution the table in section 2 shows is achievable, with the loss points documented rather than papered over.
Digital work instruction integration
The current revision of the job at the station, for the product actually running. Low glamour and consistently the highest adoption of anything on this page.
Production performance and downtime data integration
Event data made available to reporting and to real-time response, which are two consumers of one event model rather than two collection exercises.
Each of these is routine once the event model exists. The prerequisite named for each is what stalls it when the event model does not.
Automated machine production data collection
Counts and cycles taken from the machine rather than typed.
Production order dispatch to manufacturing lines
Orders arriving at the line in the sequence it will run them.
Production completion and quantity confirmation
Output confirmed from the machine count rather than a declaration.
Machine state and downtime event capture
States and stoppages captured continuously with reason attached.
Material consumption and lot tracking
What was actually consumed, by lot, against the order.
Product and component traceability
The forward and backward chain for a unit or batch.
In-process quality inspection data collection
Checks prompted and captured at the station.
Production routing and operation status updates
Which operation an order is at, live.
Shop-floor operator guidance integration
The right instruction at the station for the product running.
Multi-line MES integration and standardisation
The same execution model across lines and plants.
The ISA-95 levels are a useful shared vocabulary and a poor way to organise a programme, because the boundary that matters in practice is not where the diagram draws a line. It is where authority changes.
ISA-95 aligned enterprise to control integration
Used as the vocabulary for describing which layer owns which transaction, with the authority boundaries above stated explicitly rather than implied by the level number.
MES integration reference architecture
Equipment layer, event layer, execution layer and enterprise layer, with every path marked read-only or read-write. Almost all should be read-only.
PLC, SCADA, DCS and historian connectivity
Preference order for reading the same information: historian first, then SCADA, then the controller directly. Each step increases the load placed on a production system and the approval required.
OPC UA, MQTT, Modbus and industrial protocols
OPC UA where equipment supports it, MQTT for transport, direct protocol reads where neither exists. The secure connectivity design is covered on the Industrial IoT & OT Integration page.
REST APIs and manufacturing web services
Used between MES and enterprise systems, where request and response volumes suit them.
Event-driven production data integration
Production events published once and consumed independently by execution, reporting and real-time response, so adding a consumer never means touching the machine integration again.
Edge gateways and industrial middleware
Where the machines need a translation and aggregation point. The infrastructure and fleet operation of those gateways is covered on the Industrial IoT & Edge Computing page.
Manufacturing data models and event schemas
The event model from section 3, expressed as versioned schemas under change control, with the boundary definitions recorded alongside each state.
Hybrid on-premises and cloud integration
Execution on-premises for most plants, because execution must survive a lost link. Reporting and analysis centrally.
Store and forward and resilient data synchronisation
Local buffering so a link outage does not lose production events, with ordered and deduplicated delivery on reconnection. The downstream consequences of late and out-of-order arrival are set out on the Industrial IoT & Edge Computing page and apply here.
Governance defines who may change an interface, how exceptions get resolved, and how production continuity is protected. Everything below serves one of those three.
That transition is the point at which an execution system stops being an IT system. The controls below are written for that condition rather than for a reporting tool.
IT and OT network segmentation
Zones with controlled conduits, enterprise systems never reaching the control network directly, and the zone model agreed with your OT security function as a condition of the design.
MES and shop-floor access management
Role-based access with named individual accounts, including on the floor. Shift-shared logins at a station are the standard finding and they make every traceability record attributable to nobody.
Production data quality and validation
Captured events validated against physical observation and against the count, with a tolerance agreed rather than assumed. Divergence investigated, not smoothed.
Machine and production event governance
The event model owned by a named role, versioned, and changed through a process. A change to an event boundary changes every historical figure built on it, so the change record is also a reporting notice.
Operational safety and control system boundaries
No write path from MES or the integration layer to a controller, a recipe or a setpoint. Where a transaction has a physical consequence, that consequence is implemented in the control system under your engineering change process.
Integration change management
Changes released through the plant change process, in production windows, with rollback. Software touching production is a production change regardless of which function builds it.
Production traceability and audit trails
The genealogy chain retained with its loss points recorded, so a traceability enquiry returns the achievable resolution and says where it degrades rather than presenting an inference as a fact.
Manufacturing system availability and recovery
An availability target agreed with operations, an on-call route, and a documented way for the line to run and record production while MES is unavailable. That local record then has to load back without creating duplicates.
Shop-floor integration ownership and support
A named owner per interface and an escalation that reaches somebody who can authorise a workaround during a night shift. The IT and OT ownership models are set out on the Smart Factory Solutions page.
Secure industrial protocol and gateway management
Per-device identity, controlled remote access and a patching route that respects production windows. Covered in depth on the Industrial IoT & OT Integration page.
A fast integration that misses events is worse than a slow one that does not, because the gaps are invisible and every figure built on the data is quietly low.
Shop-floor data capture completeness
The primary measure. Events captured against events that occurred, established by observing a line for a period and comparing. The only way to know, and rarely done.
MES integration availability
Per interface, measured against the hours production actually runs rather than as a calendar average.
Production event processing latency
From the event to its availability in MES, at the percentile that matters for the fastest consumer of it.
Machine to MES synchronisation accuracy
Machine count against MES count, per shift, with any difference investigated rather than reconciled by adjustment.
Production transaction failure rates
By failure type, since a rejected confirmation and a timeout need different responses and are usually reported as one number.
Data reconciliation and exception resolution
Open exceptions by age, with a named owner. The age distribution matters more than the count because a stable count can hide a growing tail.
Equipment connectivity health
Per machine, distinguishing a machine that is off from a machine that is unreachable. They need different responses and are routinely confused.
Production traceability completeness
What share of production has a complete chain at the resolution claimed, and where the chain degrades. Reported honestly, including the loss points from section 2.
Multi-line integration scalability
Effort per line in rollout order, by machine type. A flat line by the third installation of the same type means there is no reusable pattern.
Continuous integration reliability improvement
Interfaces simplified, retired or consolidated, tracked deliberately, since an integration estate otherwise only grows.
Connected shop-floor execution, where the machine tells the system what it did rather than a person transcribing it
Improved production data accuracy, which comes from removing the transcription step and from one event model rather than from the integration alone
Reduced manual production reporting, with the remaining manual entry kept where a human judgement is genuinely required
Better production order visibility, at the resolution the chosen confirmation point supports and no finer
Improved manufacturing traceability, to the resolution the table in section 2 shows is achievable for your materials. Where that is a time window rather than a lot, we will say so
Scalable MES integration, evidenced by effort per line falling for the second and third installation of the same machine type
If you have been asked to guarantee lot-level traceability, the first thing worth knowing is whether your materials can support it. That assessment takes days and it changes what gets promised.
CaliberFocus connects manufacturing execution to the shop floor by settling the definitions before capturing the data. We establish where material identity is lost and the traceability each material can honestly support, agree the event boundaries in writing, state the direction of authority per transaction, and never write a setpoint, recipe or program to a controller.
- Traceability resolution stated honestly for each material form
- Event boundaries agreed and signed before capture
- Authority stated per transaction, read-only from the plant
- Capture completeness measured against observation on the line
Security & Compliance
