Industrial Data Acquisition and MES Implementation: From PLC to Enterprise Systems

Building a Scalable MES Architecture: From PLC Data to Enterprise Analytics

Manufacturing facilities today generate massive amounts of operational data—production rates, equipment performance, quality metrics, and energy consumption. Yet many plants struggle to capture, centralize, and act on this data effectively. The gap between your shop floor PLCs and enterprise decision-making systems is where operational intelligence gets lost.

Implementing a Manufacturing Execution System (MES) with centralized data acquisition isn't just about installing software. It requires thoughtful architecture design that bridges legacy control systems with modern analytics platforms, ensures reliability under continuous operation, and scales as your facility grows. This guide walks you through the critical decisions you'll face when designing a data collection infrastructure for large process plants.

Understanding the Data Flow: From PLC to Enterprise

Before selecting specific technologies, you need to understand how data moves through your facility. In a typical large process plant, you have:

• Distributed PLCs at various production lines and process areas
• Local networks connecting these devices
• A centralized data collection point (your MES server or historian)
• Enterprise systems consuming that data for analytics, reporting, and optimization

The challenge isn't moving data once—it's moving it continuously, reliably, and at scale. A single production line might generate thousands of tag updates per second. Multiply that across 10, 20, or 50 production areas, and you're managing millions of data points daily.

The Three-Tier Architecture Model

Most successful large-facility implementations follow a three-tier model:

Tier 1: Edge/Control Layer — Your existing PLCs and controllers remain unchanged. They continue running your production processes. The key is ensuring they can expose their data through standard protocols (Modbus, Profibus, Ethernet/IP, OPC UA).

Tier 2: Data Collection & Aggregation Layer — This is where your MES server or data historian lives. It actively polls or subscribes to PLC data, validates it, stores it, and makes it available to upper layers. This tier handles the heavy lifting of data persistence and real-time buffering.

Tier 3: Analytics & Reporting Layer — Enterprise systems, dashboards, and analytics platforms consume data from Tier 2. This separation prevents production systems from being impacted by reporting queries or analytics workloads.

Designing Your MES Server Infrastructure

Your MES server infrastructure must balance three competing demands: reliability (it cannot go down without halting data collection), performance (it must handle your peak data ingestion rate), and scalability (it must grow with your facility).

Hardware Considerations

For a large process plant, avoid single-server deployments. Instead, consider:

Redundant Architecture — Deploy your MES server on a pair of servers in active-passive or active-active configuration. If one fails, the other seamlessly takes over. This is non-negotiable for continuous manufacturing environments.

Network Connectivity — Use redundant network paths to your PLC network. A single network switch failure shouldn't isolate your data collection infrastructure. Westermo industrial switches are purpose-built for this—they provide managed switching with redundancy features like ring topology support and RSTP (Rapid Spanning Tree Protocol) that automatically reroute traffic if a link fails.

Storage Subsystem — Your database will grow rapidly. A 24/7 facility collecting data from 50 PLCs at 1-second intervals generates roughly 1.5 billion data points per month. Plan for RAID storage (RAID 10 preferred for databases) with hot-spare drives. Don't underestimate storage needs.

Compute Resources — Modern MES platforms are resource-intensive. Allocate at least 8 CPU cores and 32GB RAM as a baseline for a mid-sized facility. Larger installations may need 16+ cores and 64GB+ RAM. Monitor CPU and memory utilization—if you're consistently above 70% utilization, you're approaching a scaling limit.

Network Infrastructure for Data Collection

The network connecting your PLCs to your MES server is critical. Industrial networks are noisy—variable frequency drives (VFDs) from manufacturers like ABB and WEG generate electromagnetic interference, and wireless networks can drop packets.

Wired Ethernet is mandatory for reliable data collection. Use industrial-grade switches with managed features. Westermo's product line includes switches designed for harsh industrial environments with wide temperature ranges and vibration tolerance.

Redundancy at the network level matters. If you have multiple production areas, connect each to your MES server through separate network paths. This prevents a single cable cut or switch failure from isolating an entire production line's data.

Bandwidth planning is often overlooked. A single Ethernet connection can handle thousands of PLC tags, but if you're also running video surveillance, VoIP, or large file transfers on the same network, data collection suffers. Segment your network—production data on one VLAN, administrative traffic on another.

Selecting a Data Historian and Database Solution

Your historian is the system that actually stores your PLC data. This is where architecture decisions have the longest-lasting impact.

Time-Series Databases vs. Relational Databases

Traditional relational databases (SQL Server, PostgreSQL, Oracle) can store industrial data, but they're not optimized for it. Industrial data is inherently time-series—you care about values at specific timestamps, trends over time, and aggregations (averages, minimums, maximums) across time windows.

Time-series databases are purpose-built for this workload. They compress data more efficiently, query historical trends faster, and scale better as your data volume grows. However, they require different query patterns and skills to manage effectively.

Hybrid Approach — Many successful implementations use both. A time-series database (or historian) stores high-frequency PLC data for the last 30-90 days. Older data is rolled up (aggregated to hourly or daily values) and moved to a relational database for long-term archival and reporting. This balances performance with cost.

ABB's Approach to Data Management

ABB offers comprehensive solutions for industrial data collection and SCADA. Their ABB 800xA system is a full-featured DCS (Distributed Control System) that includes integrated data historian capabilities. If you're implementing a new control system or modernizing an existing one, 800xA provides unified data collection, control, and analytics in a single platform.

For facilities with existing PLCs that you're not replacing, ABB's Freelance and Compact Control systems can serve as your data collection and control layer, interfacing with legacy equipment while providing modern data management.

WEG and Danfoss Integration

If your facility uses WEG motors and drives, or Danfoss frequency converters, ensure your MES architecture can natively communicate with these devices. Many modern industrial devices support OPC UA, a standardized protocol for industrial data exchange. This eliminates the need for custom drivers or gateways.

Integration Strategies for Legacy PLCs

Most large facilities have a mix of old and new equipment. Your MES must bridge this gap.

OPC UA as the Universal Language

OPC UA (OPC Unified Architecture) is the industrial standard for device communication. If your legacy PLCs support it, use it. It's secure, standardized, and widely supported by MES platforms.

If your legacy PLCs don't support OPC UA natively, you have options:

Gateway Devices — Install an industrial gateway that translates between your PLC's native protocol (Modbus, Profibus, Ethernet/IP) and OPC UA. This is often cheaper than replacing the PLC.

Middleware — Some MES platforms include built-in drivers for common legacy protocols. Verify your chosen platform supports your specific PLC models before committing.

Phased Replacement — If you have many legacy PLCs, plan a multi-year replacement strategy. Replace the oldest or most problematic units first, gradually modernizing your control layer while maintaining production continuity.

Data Validation and Quality

Raw PLC data is often dirty. Sensors fail, communication glitches occur, and PLCs occasionally report invalid values. Your MES must include data validation logic:

• Range checking — Flag values outside expected ranges
• Rate-of-change limits — Detect impossible jumps (temperature jumping 50°C in one second)
• Redundant measurement — If you have multiple sensors measuring the same parameter, compare them
• Timestamp validation — Ensure data is in chronological order

This validation happens in Tier 2 (your data collection layer) before data is stored. Invalid data should be flagged but not discarded—you need to know when data quality issues occur.

Scaling Your Architecture as You Grow

A well-designed MES architecture should handle 3-5 years of growth without major redesign.

Horizontal Scaling

If a single MES server becomes a bottleneck, add more servers. Distribute the data collection load across multiple servers, each responsible for a subset of PLCs. A central database aggregates data from all collection servers.

This requires careful planning—you need load balancing, distributed caching, and conflict resolution. It's more complex than a single-server setup, but it's the path to handling hundreds of PLCs.

Database Scaling

As your historian grows, you'll eventually hit storage limits. Plan for:

• Data archival — Move old data to cheaper storage (cloud object storage, tape archives)
• Data retention policies — Decide how long to keep high-frequency data (30 days?) vs. aggregated data (years)
• Partitioning — Divide your database by time or production line to improve query performance

Practical Implementation Checklist

Before you start your MES implementation:

1. Audit your PLCs — Document every PLC, its model, protocol, and data points you need to collect
2. Design your network — Map out how data will flow from each PLC to your MES server
3. Select your platform — Choose your MES software, historian, and database
4. Plan for redundancy — Design for single-point-of-failure elimination
5. Allocate storage — Calculate your data volume growth and provision accordingly
6. Train your team — Ensure your operations and IT teams understand the new system
7. Plan for maintenance — Schedule regular backups, database maintenance, and system updates

Getting Started with Gross Automation

Implementing a scalable MES architecture is complex, but you don't have to navigate it alone. Gross Automation partners with industry-leading manufacturers—ABB, WEG, Danfoss, Beijer, Westermo, and others—to provide integrated solutions for industrial data collection and control.

Whether you're modernizing your control layer with ABB's 800xA system, integrating legacy PLCs with OPC UA gateways, or designing a redundant network infrastructure with Westermo industrial switches, we have the expertise and products to support your implementation.

Ready to build your MES architecture? Contact Gross Automation to discuss your specific facility requirements. Our technical team can help you design a scalable, reliable data collection infrastructure that grows with your business.