MES software that records what the line did, not what someone typed.
VZIAM MES takes run, stop and count signals from the controller, asks the operator only for what a machine cannot know (why it stopped, what was rejected) and computes OEE on the server. The figure in Monday’s review is the one the machines produced.
A manufacturing execution system, between your machines and your ERP.
VZIAM MES is a manufacturing execution system: software that records what actually happens on the shop floor (machine states, stops, quantities, rejects, checks and material lots) against the work order that was running, and returns the result to the ERP. The ERP plans the order; the MES records how it was made.
PLCs and sensors
States and counters, over OPC UA, MQTT or a gateway
Touch cockpit at the machine
Stop reasons · rejects · checks · material lots
What the controller cannot know
VZIAM MES
Machine · work order · article · lot
- Timeline, OEE and losses
- Quality, genealogy and the order’s record
- Forms, documents and maintenance
ERP
Orders, articles, BOMs and routings in; quantities, rejects and times back
TV displays and reports
Line states · live OEE · Pareto of losses
Five rules the numbers cannot get around.
The server enforces them, not a procedure someone is asked to follow. This is what makes an OEE figure from VZIAM MES one a plant manager can defend.
- 01
Machine counters are never reset.
Production is the difference between two readings of the controller’s own counter. A rollover or a replaced sensor is detected and absorbed; nothing already counted is subtracted.
Machine data acquisition - 02
OEE runs only while a work order runs.
Running time, output and rejects accrue to the order in production. Counts that arrive with no order running go to a separate counter, so they cannot inflate the order before or after.
Real-time OEE - 03
Nominal speed belongs to the article on the machine.
Performance is measured against the rate set for that article on that machine (in parts per hour or seconds per part), never a catalogue figure for the article alone. Slow cycles and short stops show up as what they are.
Micro-stops and slow cycles - 04
One work order, one finished lot.
Starting an order creates its lot. The material lots it consumes, its rejects and its checks attach to it as production runs, so genealogy is recorded, not rebuilt from paper after a complaint.
Traceability - 05
Every correction leaves a trace.
Operators can reclassify a stop, split it or correct a count: the floor is allowed to be right. Each change keeps the old value, the new one, who made it and when, and the figures are recalculated.
Downtime tracking
One stop, split in two, with the correction on file.
At 09:12 the controller on Line 02 stops counting. Twenty-nine minutes later it starts again, and the operator splits the stop at the cockpit: the line waited for material, then changed over. The original stays in the audit trail.
29 minutes, no reason yet
Two causes, two durations
The correction, kept
- Before
- Unexplained · 29 min
- After
- Material 18 · setup 11
- By
- Line 02 operator
- At
- 09:44
Recalculated: WO-0248 · morning shift
Seven modules write to one record. Start with one.
Each module reads and writes the same machines, orders and lots, so the second one you switch on finds the first one’s data already there.
Production & OEE
Orders started at the cockpit and run from the controller’s signals. Running, stopped, standby and offline on one continuous timeline per machine, split across shifts automatically, with OEE computed on the server to AFNOR NF E60-182.
No gap and no overlap in a machine’s timeline
Quality
Control plans triggered at an order’s start, every so many minutes or every so many parts, scoped to a machine, an article or a workshop. A non-conforming point schedules a recheck, and a critical one alerts the supervisor at once.
A reject without a reason is refused
Traceability
Material lots recorded against the order that consumed them, consumption compared with the bill of materials, and genealogy traced upstream and downstream from any lot.
One work order, one finished lot
Batch records
Each order’s record in one place: the routing as applied, materials and their lots, checks and their rechecks, stops and the corrections made to them, with a PDF summary when the order closes.
Forcing a blocked release needs a written reason
Forms
Start-up checklists, cleaning sheets and readings built from typed fields with tolerances. A value outside its range is flagged as it is entered, and the operator signs on submission.
A submitted form is read-only
Maintenance
Requests raised from the cockpit with the machine and its severity attached; preventive, corrective and improvement interventions; plans triggered by time or usage; procedures with mandatory steps; spare parts; MTBF.
Parts used are booked to the intervention
Documents
Work instructions and one-point lessons attached to machines, articles and orders, opened from the cockpit at the station they apply to.
A new file is a new version; the old one stays
Five roles and a TV wall, all reading the same record.
Each role gets its own screens and its own permissions, scoped by site. Nobody exports a spreadsheet to tell someone else what happened.
A cockpit built for gloves, not for a mouse.
Start an order or a quick run, qualify a stop in three taps (family, component, reason), declare rejects on a numpad, run the checks that are due and call maintenance. Targets are at least 44 pixels, and the screen keeps working when the Wi-Fi does not.
The operator terminal- Stop longer than 60 s
- Orange, asks for a reason
- Qualified in
- Three taps
- Wi-Fi lost
- Declarations queued, replayed in order
The shift, the losses and the arbitration.
The timeline of every machine by shift, Pareto charts of stops and rejects, period comparisons, and the corrections only a supervisor may make: reclassifying after the shift closes, merging intervals, validating large count adjustments.
Downtime tracking- Pareto
- Stops and rejects
- Reports
- Week, month, custom
- Prolonged stop alert
- Threshold you set
Control plans, non-conformities and release.
Define what is checked, when and where; follow non-conformities and their rechecks as a chain; release a lot, or keep it blocked while a releasing point is open. Every forced release carries its justification.
Quality management- Recheck
- Failed points or the full check
- Critical point
- Supervisor alerted at once
- Lot
- Blocked until conforming
Requests from the line, with the context attached.
A request arrives with the machine, the component and the severity the operator felt. Accept or refuse it, plan the intervention, follow its procedure step by step, book the parts used and close it with a report.
Maintenance in VZIAM MES- Severity
- Low · normal · high · blocking
- Preventive plan
- By time or usage
- Spare parts
- Booked to the intervention
Your plant, modelled once.
Sites, workshops, lines and machines; articles and their nominal speed on each machine; the stop and reject reason trees; users and their roles by site; and the datasets and tags that map each controller signal.
The architecture- Structure
- Site › workshop › line › machine
- Nominal speed
- Per article and machine
- Signals
- Datasets, tags and formulas
The state of the line, on the wall above it.
Any browser on a smart TV or a small box: pick the workshop, go full screen, and each line’s card rotates with its state, OEE and cadence against nominal, in figures readable across the hall. Updates are pushed as they happen.
Andon displays- States
- Running · stopped · standby · offline
- Rotation
- One line every 30 s
- On a stop
- Alert banner on the line’s card
OPC UA and MQTT directly. Everything else through a gateway.
VZIAM MES reads OPC UA servers on a polling cycle you set, subscribes to MQTT topics, and accepts HTTP push from edge devices. A Modbus, Siemens S7 or proprietary controller is connected through a gateway (Node-RED or a DAS box) that translates its signals to MQTT. Who supplies and maintains that gateway is settled during the assessment, not at go-live.
- Controller with an OPC UA serverNodes read on the polling cycle you setOPC UA
- Device or broker publishing MQTTTopics mapped to tagsMQTT
- Edge device or scriptPosts readings as JSONHTTP push
- Modbus, Siemens S7 or proprietary PLCRead by the gateway on the lineNode-RED or DAS boxMQTT
VZIAM MES server
- Counts as deltas, never reset
- States with a signal-quality flag
- Formulas on tags, evaluated live
- Machine protocols
- OPC UA · MQTT · HTTP push
- Through a gateway
- Modbus · Siemens S7 · proprietary
- ERP exchange
- REST API · JSON or CSV files, mapped fields
- Runs
- Docker, on a server on your network
- Exports
- CSV for BI tools · PDF reports
Start on one line, with real signals.
Nothing goes plant-wide until one line has run on real signals and you have checked its numbers.
- 01
Assessment
We list every machine and how it can be read (OPC UA, MQTT or a gateway) and agree your stop and reject reasons, articles and nominal speeds before anything is installed.
- 02
Pilot
One line you choose, live on the cockpit, with OEE computed from its controllers. You check the numbers against your own before committing to more.
- 03
Plant rollout
The rest of the plant, connected from what the survey found, and the ERP exchange scoped with your ERP team in the room.
Where VZIAM MES fits next to what you already run.
Most plants asking about an MES already have an ERP, a spreadsheet and a maintenance tool. These pages say what each one keeps doing.
Compare
- MES vs ERP
Where the boundary between planning and execution sits.
- MES vs Excel
Where the numbers come from, and what they can prove.
- MES vs CMMS
Maintenance inside the MES, or a CMMS of its own.
- VZIAM CMMS
A dedicated CMMS, built for maintenance teams.
Guides
- OEE Calculation
The three factors, and what measures each one.
- Identify Bottlenecks
Telling cause from consequence on a coupled line.
Industries
- Automotive
Each finished lot traced to its material and its machine.
- Food Processing
The batch record, written while the batch runs.
- Plastics
Cycle time read from the press, not from the shift report.
- Packaging
Finding the one machine that actually stopped the line.
They can correct them, not hide them. An operator can reclassify a stop, split it or correct a count after a physical recount, and a supervisor can do more after the shift closes. Every correction keeps the old value, the new one, who made it and when, and the figures are recalculated on the server. The cockpit displays OEE; it never computes it.
Controllers with an OPC UA server and devices that publish MQTT connect directly; edge devices can also post readings over HTTP. A Modbus, Siemens S7 or proprietary PLC connects through a gateway (Node-RED or a DAS box) that translates its signals to MQTT.
From the controller. When no signal arrives for 15 seconds the cockpit shows a pre-alert without creating anything; after 60 seconds a stop is created and the operator is asked for the reason. Both delays are settings. A stop nobody qualifies stays visible as unexplained, and an alert is raised once it passes the threshold you set.
The cockpit shows an offline banner and keeps working. Declarations are stored on the device and sent in the order they were entered once the network is back, so the machine’s timeline stays continuous.
Work orders, articles, bills of materials and routings come in through a REST API or JSON and CSV files, with the fields mapped to yours. Good quantities, rejects and real cycle times go back when an order closes. Once an order is in production, a change from the ERP to its quantity or routing needs a supervisor’s approval.
Yes. The timeline exports as CSV for BI tools, with its segments, a KPI summary and the counter diagnostics. OEE reports export as CSV that opens in Excel, with a PDF summary.
It includes maintenance: requests raised from the line, preventive, corrective and improvement interventions, plans by time or usage, spare parts and MTBF. A maintenance team that wants a dedicated application of its own can use VZIAM CMMS instead; the comparison page sets out which fits.
On a server on your network, as Docker containers: the database, the application and an MQTT broker for the machine signals. The database is backed up daily, with 30 days kept.
Plan the change before it reaches the line.
Request a demo and an engineer looks at how your production runs today, shows how the software would fit around it, and where a pilot on one line would start.
01Pick a time
Choose a slot in the calendar, and note what you make and what reads your machines today.
02Screen shared with an engineer
A video call with someone who builds the software, in English or French.
03What would connect
For each machine, its path: native (OPC UA, MQTT or HTTP push), through a gateway, or nothing to build.