Why the classic ISA-95 automation pyramid no longer works
For decades, industrial automation was built on the ISA-95 pyramid model with five distinct levels: from physical sensors at level 0 to enterprise ERP systems at level 4. Each level exchanged data only with its immediate neighbour, and decision-making flowed top-down through the entire hierarchy. This architecture handled the demands of the last century well enough, when equipment operated in isolation and data volumes were measured in megabytes.
But modern manufacturing requirements have changed drastically. Customers want customised products, time-to-market keeps shrinking, and equipment generates terabytes of data daily. Pushing every byte through five intermediate layers is like sending an urgent telegram through a chain of middlemen. Delays accumulate, context gets lost, and decisions are made too slowly for the pace of today's production.
What compact and flat architecture actually means
Compact (or flat) architecture is an approach where the traditional five automation levels are compressed to two or three. Instead of a rigid hierarchy where data passes sequentially from sensors through programmable logic controllers (PLCs), SCADA, MES, and all the way up to ERP, the new model allows shop-floor equipment to communicate directly with cloud and enterprise systems.
The core idea is straightforward: intelligence moves closer to the equipment. Edge controllers process data locally, make autonomous decisions, and pass only aggregated information upstream. Think of how a modern smartphone works: it handles most computation locally and only taps the cloud when it actually needs to.
Key components of a distributed system
Edge controllers and distributed I/O
The heart of compact architecture is the edge controller with built-in analytics capabilities. Unlike classic PLCs that simply run a cyclic program, edge controllers can simultaneously execute predictive analytics algorithms, handle wireless industrial network protocols, and support modern data exchange standards such as OPC UA and MQTT.
Distributed I/O means that input/output modules are placed directly next to the equipment rather than in a central control cabinet. This shortens cable runs, reduces interference, and allows the system to scale without stopping production. The modular design lets you add new data collection points on a plug-and-play basis.
Next-generation communication protocols
The transition to flat architecture is impossible without modern protocols. OPC UA provides semantically rich data exchange — it transmits not just raw numeric values, but self-describing data structures with metadata and context. MQTT is a lightweight protocol ideally suited for resource-constrained devices and narrow communication channels. The most effective solutions combine both: OPC UA inside the plant floor, MQTT for cloud connectivity.
Operator panels and visualisation
In a flat architecture, operator panels (HMIs) are no longer mere display devices. Modern HMIs with HTML5 and web technology support become full-fledged network nodes capable of aggregating data from multiple sources, running local analytics, and providing access to production data from any device via a web browser.
Architecture comparison: centralised vs distributed
| Parameter | Centralised (ISA-95) | Distributed (compact) |
| Number of levels | 5 levels (0-4) | 2-3 levels |
| Decision-making latency | Seconds to minutes | Milliseconds |
| Scalability | Difficult, requires downtime | Modular, no downtime |
| Cabling infrastructure cost | High (long runs) | Low (short connections) |
| Fault tolerance | Single point of failure | Node autonomy |
| Data exchange | Vertical, sequential | Horizontal and vertical |
| Cloud integration | Through gateways and intermediate layers | Direct via OPC UA/MQTT |
| Cybersecurity | Perimeter-based (air gap) | Built-in (zero trust) |
| Maintenance costs | High (many intermediate systems) | Lower (fewer components) |
| Reconfiguration flexibility | Weeks to months | Hours to days |
Practical implementation scenarios
Retrofitting existing plants
Switching to compact architecture does not mean scrapping all legacy equipment. The typical modernisation path starts with adding an edge gateway to the existing system. This gateway connects to the current PLC via standard protocols (Modbus, Profinet, EtherCAT) and begins collecting data for cloud analytics without interfering with the running control system.
In the second phase, distributed I/O modules are installed directly at production stations. This enables higher-frequency data collection with lower latency. In parallel, predictive analytics is introduced — the system begins forecasting equipment failures and optimising energy consumption.
Greenfield projects
For new plants, compact architecture offers fundamental advantages. Instead of designing a central server room packed with dozens of cabinets, engineers place compact edge controllers directly on the shop floor. Variable frequency drives (VFDs) with built-in smart functions become more than just motor speed regulators — they turn into full data collection and processing nodes. They transmit motor health data, load parameters, and energy consumption figures straight to cloud analytics.
Hybrid systems
In practice, most deployments take a hybrid approach. Safety-critical control loops (such as equipment protection) stay on local controllers with guaranteed response times. Less critical tasks — monitoring, reporting, optimisation — move to the edge layer and the cloud. This approach combines the reliability of traditional systems with the flexibility of new technologies.
The role of variable frequency drives in flat architecture
Modern variable frequency drives are far more than power electronics for motor speed control. The latest generation models come equipped with a built-in PLC, Ethernet ports supporting industrial protocols, and the ability to connect expansion boards for additional interfaces. Such a VFD independently handles local control logic, motor condition diagnostics, and upstream data transmission — without a separate PLC acting as a middleman.
This fits the compact architecture concept perfectly. A single device performs the functions that previously required three or four separate components: a frequency converter, a controller, a communication gateway, and a data acquisition module. Fewer components means not just cost savings, but also higher system reliability.
Cybersecurity in distributed systems
Moving from isolated to networked systems creates new cybersecurity challenges. In the classic pyramid, protection was built on a perimeter principle: the production network was physically separated from the corporate one. In flat architecture, this approach fails because shop-floor devices communicate directly with cloud services.
Instead, a zero trust model is applied: every device is authenticated, all traffic is encrypted, and access is granted on a least-privilege basis. OPC UA has built-in security mechanisms — certificates, encryption, and message signing. MQTT is secured through TLS and broker-level authentication mechanisms.
What comes next: near-term trends
Flat architecture is not the endpoint but a platform for further development. Artificial intelligence already running on edge devices will become smarter and more autonomous. Digital twins of production lines will allow changes to be simulated before implementation. The OPC UA Part 14 (PubSub) standard will make data exchange even more efficient, and Time-Sensitive Networking (TSN) technology will deliver deterministic data transmission over Ethernet networks.
For manufacturers modernising their operations, compact architecture opens up the opportunity to leapfrog generations of technology — jumping from outdated centralised systems straight to distributed solutions without intermediate steps. And the equipment needed to build such systems — from frequency converters with built-in intelligence to modular controllers and operator panels — is already available on the market.
