Production as a stack of layers
A modern factory is not run by one big computer. It is run by many devices arranged in layers, each with a narrow job and a different time scale. The lowest layers react in thousandths of a second; the highest plan in hours or days. Seeing the layers separately makes it easier to understand where a problem lies and how upgrades fit in. The general theory of control loops is in the overview of industrial control systems and the cross-sector view is in the article on control in modern industry; this one stays on the manufacturing floor.
Layer 0: The field devices
At the bottom is the physical equipment: proximity sensors that detect parts, photoelectric sensors that count them, limit switches, pressure and temperature transmitters, encoders that report position, motors, valves, cylinders and indicator lights. These devices touch the real world. They convert physical events into signals and signals into actions.
The details here matter more than newcomers expect. A sensor that is slightly misaligned or a connector that has worked loose can cause a puzzling stoppage far up the chain. Many faults begin at this level.
Layer 1: Machine and cell control
Above the devices sits the controller that runs a machine or a small cell, usually a programmable controller, sometimes with dedicated motion controllers and drives. It reads inputs, executes the program and updates outputs, repeating in a fast cycle. This is where sequences happen: part detected, clamp closes, drill advances, retract, release, move to the next step. Interlocks and basic fault detection live here too.
Drives and servo systems follow commands from the controller and regulate motor speed or position with their own fast internal loops. Robots usually bring their own controllers that exchange signals with the cell controller. Safety circuits for emergency stops and guarding are typically implemented separately with safety-rated devices.
The benefits of using this kind of controller, and how it replaced relay panels, are described in the article on programmable logic controllers.
Layer 2: Operator interface and local supervision
Operators interact with the machine through operator interface screens or panels. They select products or recipes, start and stop the machine, view alarms and adjust permitted settings. The screens show what the controller reports; they do not replace it. Local supervisory functions may also appear here, such as showing the state of several machines on a line.
Layer 3: Line and area supervision
Where a product moves through many machines, something has to coordinate them: keep conveyors balanced, handle buffers between machines, manage product changeovers and collect line-wide data. This may be done by a line controller, a supervisory system (see SCADA systems), or software that tracks status and counts. Alarms from different machines are collected so supervisors can see the whole line. Data about stoppages, rejects and output is recorded here.
Layer 4: Production management
Higher still, software organizes work: which orders to run, on which line, in what sequence, with what materials. Production management systems may issue work to the line, record what was produced, track materials and quality results, and report performance. This layer ties the plant to business systems for orders and inventory. Its tools overlap with the scheduling and monitoring software discussed in industrial software tools. Time scales are much longer here, typically minutes to days.
How the layers coordinate production
Information flows both ways.
- Downward: orders, recipes and settings. A production system tells a line which product to make; the line controller tells each machine which program to use; the machine controller sets speeds and positions.
- Upward: status, counts, alarms and measurements. Sensors report to controllers, controllers to supervisory systems, and supervisory systems to production management.
Fast, safety-related and time-critical functions stay low in the stack, close to the equipment, so they do not depend on higher layers or network delays. A machine should be able to stop safely, and often to continue or fail safe, even if the layers above are unavailable. Higher layers deal with planning and analysis, where delays are tolerable.
Networks between layers
Each boundary involves a network or interface. Field devices connect to controllers via direct wiring or digital buses; controllers connect to screens and supervisory systems over industrial networks; plant systems connect to business systems through carefully controlled links. Good segmentation between layers supports both reliability and security; see industrial networking and industrial cybersecurity.
Why the layered view helps
Troubleshooting. When output drops, you can ask which layer shows the first sign of the problem. A sensor fault, a program issue, a communication loss and a scheduling mistake all look different at different layers.
Upgrades. A new data collection system can be added at the supervisory layer without touching machine logic. A new robot adds a machine-level controller and a few interface signals to the line.
Standardization. Using consistent naming, signal conventions and program structure across machines makes maintenance far easier, which is why many plants set internal standards.
Skills. Different people often own different layers: electricians and technicians at the device level, controls engineers for programs, IT or operations technology staff for servers and networks. Clear handoffs between them prevent gaps.
Practical cautions
Layers can become tangled when quick fixes bypass the structure, for example by reaching directly into a controller from the office network. Keep documentation current, back up programs at every level and treat changes in a controlled way. Safety functions should be designed and validated by qualified people to the applicable standards and local regulations.
This article is general information, not engineering advice.