Machine monitoring system: what every machine is doing, and what it costs.

What a machine monitoring system does

In most plants, a stopped machine gets noticed when someone walks past it. A machine monitoring system collects signals from your machines all day, on every shift, and answers four questions:

• Is it working? Running, idle, stopped and overload time, by shift.
• What did it make? Part counts, cycle time and OEE.
• What did it cost? kWh and rupees per machine, per shift and per unit made.
• Is it healthy? Vibration, temperature and load, trending over weeks.

How a machine monitoring system works

1. A signal at the machine. An energy meter on the machine’s supply, and where needed a sensor or a part counter. Nothing has to be wired into the machine’s controller.
2. One wired network. Meters, sensors and counters share an RS-485 cable and answer over Modbus RTU, each at its own address.
3. A gateway that reads and records. It asks each device for its readings every few seconds and stores them with the time, inside the plant.
4. Software that turns readings into states. Power draw becomes running, idle, stopped or overload. Each state is split by shift, priced on your tariff and checked against your shift calendar.
5. Alerts and action. The right person hears about an unplanned stop or a machine idling after hours. Where it is part of the project, an I/O card switches the load off.

What you can monitor, and where each signal comes from

The first three rows come from the energy meter alone. The rest are added and configured for your machines as part of your project, on the same wired RS-485 network. Any sensor or output with Modbus RTU connects, and a 4-20 mA sensor connects through a Modbus input card.

Start with the energy meter

A machine’s power draw tells you what it is doing. Above a set level it is running, just above zero it is idle, and at zero it is stopped. That needs no sensor inside the machine and no change to its controller, so it works the same on a 30-year-old press and a new CNC.

Two numbers set up each machine: its idle level and its rated level in kW. Machine running and idle tracking shows how, and what idle time costs on your bill.

Which machines to monitor first

You do not need every machine on day one. Start where a lost hour or a wasted kW costs the most:

• The bottleneck. The machine the whole line waits on. An hour lost there is an hour lost for the plant.
• The biggest loads. Furnaces, large presses, injection moulding machines and compressors. Their idle draw is large enough to see on the bill.
• Machines that run unattended. Night shifts, Sundays and lunch breaks are when machines idle with nobody watching.
• Utilities. Compressors, chillers, pumps and blowers serve every machine, and often keep running when production stops.

Five to ten machines are enough for a first month. The pattern they show tells you where to add the rest. See sub-metering for how to arrange the meters so the totals add up.

Machine downtime monitoring

Every stop is compared with your shift timings, so a stop at 2 a.m. on a one-shift plant counts as planned and a stop at 11 a.m. does not. The morning meeting then looks at unplanned stops with their time and length, instead of what people remember.

The six big losses

OEE work groups lost production into six losses. Three come straight from the energy meter. The other three need a part count or a quality entry.

The first three need nothing but the meter, so they are the place to start.

CNC machine monitoring

CNC controllers differ by make and age, so reading each one is a separate job. Power draw is the same on all of them. A meter on the machine’s supply shows when it is cutting, when it is waiting and when it is off, whichever controller it has. Part counts then come from a Modbus counter on the machine, or from the PLC where it has a Modbus RTU port.

Old machines and mixed fleets

Most plants have machines from many makers and many decades. Reading each controller is a separate job for each make, and many older machines have no data port at all. These signals work on all of them:

• Power draw. A meter with CTs on the machine’s supply gives running, idle, stopped and overload on any electric machine.
• A part count. A proximity sensor or a cam switch on a moving part feeds a Modbus counter.
• A status contact. A spare relay contact, such as cycle running or fault, connects through a Modbus input card.
• An analogue sensor. A 4-20 mA temperature or pressure sensor connects through a Modbus input card.

Newer CNC controls may also offer MTConnect or OPC UA over Ethernet, which helps on the machines that have them. The meter works on every machine. Counters, contacts and sensors are configured for your machines as part of your project.

OEE from the same system

OEE (Overall Equipment Effectiveness) multiplies three figures: availability, performance and quality. Running time comes from the meter; part counts come from a Modbus counter or PLC, and are configured as part of your project. Here is one illustrative shift:

Machine monitoring KPIs

Utilisation, availability, idle share and stop counts come from the meter. MTBF and MTTR need each stop marked as a breakdown or not. kWh per part, OEE and TEEP need part counts.

Machine health

Running time says what a machine did. Health says how it is holding up. Rising vibration, a bearing running hotter, or more kW for the same work all show up well before a breakdown. See machine health monitoring.

Machine monitoring, MES, SCADA and condition monitoring

These systems overlap, and vendors use the names loosely. This is what each one is usually for:

Machine monitoring is a good first step, because it needs no change to the machines. Running hours, stops and energy per part can then go to MES, ERP or costing through an Excel or CSV export. See EMS vs SCADA vs BMS.

From watching to acting

Modbus I/O cards let the system act on what it sees: switch off idle machines outside shift hours, run shift and holiday schedules, and hold back chosen loads when Maximum Demand nears the limit. This is engineered for your plant as part of your project, with the machines and limits agreed with your team. Interlocks, recipes and machine sequences stay with your PLC.

Why on-premises matters for machine data

Machine data shows your output, shifts and downtime, some of the most sensitive numbers a plant has. An on-premises system keeps them inside the plant, and recording carries on when the Internet is down. Owners can still open it from anywhere over an encrypted private network. See on-premises energy monitoring.

Machine monitoring by industry

The pattern is the same everywhere: power drawn while nothing is being made. For where to look in each sector, see manufacturing and food and cold chain.

A first month, step by step

1. Week 1. Fit meters on the five to ten machines from the list above, and set each one’s idle and rated kW.
2. Week 2. Enter your shift calendar and holidays, so planned and unplanned stops separate.
3. Week 3. Review idle hours and unplanned stops in the morning meeting, and pick the two biggest.
4. Week 4. Fix those two, then compare the weeks before and after in kWh and running hours.

At the end of the month you have your own numbers for what idle time and stops cost, and you know where to add meters, sensors or counters next.

Common mistakes

• One threshold for every machine. A 2 kW idle level suits a small lathe and means nothing on a 150 kW press. Each machine needs its own levels.
• No shift calendar. Without one, every Sunday counts as downtime and the real stops get lost.
• Watching running time only. A machine can be on all shift and still waste power idling between parts. Track idle as its own state.
• Sensors everywhere before the basics. Start with meters, then add vibration and temperature where a breakdown costs the most.
• Alerts with no owner. Each alert needs a named person and a limit, or it is soon ignored.
• Data nobody reviews. Pick one meeting, such as the morning production meeting, where the stop list is read every day.

Questions to ask before you buy

• Does it work on our old machines, or only on new controllers?
• Can it show energy and cost per machine, not only running time?
• Can it separate planned stops from unplanned ones by shift?
• Which sensors and PLCs connect, and over what network?
• Can it act on its own, for example switching off an idle machine?
• Where is our production data stored?

For the full checklist on meters, data, the Internet and support, see the buying guide.

How KONDUKTOR® EMS monitors machines

KONDUKTOR® EMS is an energy monitoring system that reads a meter on each machine or panel over Modbus RTU, up to 10 meters per gateway, with more gateways added as the plant grows. The history and dashboard stay on site.

• Running, idle, stopped and overload time per machine, split by shift.
• Energy and cost per machine, and energy per unit made from the production you enter.
• Phase unbalance and voltage, where your meter reports them.
• Vibration, temperature, pressure, flow and part counts over Modbus, configured for your machines as part of your project.
• Automatic idle switch-off and schedules through Modbus I/O cards, engineered for your plant as part of your project.

One system covers the machines and the electricity bill, so you do not need one product for production and another for energy.