In short: ERP and MES answer two different questions in the same factory. ERP plans what will be produced, when and at what cost; MES monitors and manages how the planned work actually runs on the shop floor, at the level of minutes and shifts. Trying to make one stand in for the other either swamps ERP with shop-floor detail or cuts MES off from commercial decisions. This article explains the division of roles between the two systems, the layer in which each piece of data should live and where to start an MES project.
What is the difference between ERP and MES?
An enterprise resource planning (ERP) system is the company’s commercial and financial backbone: sales orders, purchasing, inventory valuation, the production plan, costing and accounting come together here. ERP’s time scale is days and weeks; the question it asks is «what will we produce, what will it cost and when will we deliver?»
A manufacturing execution system (MES), by contrast, is the system of the shop floor. The dispatching of work orders to machines and operators, the quantities actually produced, downtime and its causes, quality inspection records, lot and serial traceability and operator activities are kept in MES. MES’s time scale is minutes and shifts; the question it asks is «what is happening on the line right now, and where are we against the plan?»
How does ISA-95 describe this division?
The most widely used framework for this division of labour is the ISA-95 series of standards, also known as IEC 62264. The standard separates the activities between business systems and manufacturing control systems into layers and defines how these layers communicate with each other; it bases the equipment hierarchy on the Purdue reference model.
In the common reading, the top layer is business planning and logistics; this is where ERP sits. The layer below it is manufacturing operations management; this is where MES sits. The layers further down are the control systems, sensors and machines on the shop floor. The practical value of this division is that every piece of data gets an «owner», and which information flows between the systems, in which direction and how often, is defined from the outset.
Which data should live where?
The right question is not «which system is better?» but «which decision does this data feed, and who is responsible for it?» A general placement can be thought of as follows:
- In ERP: customer and supplier master records, sales and purchase orders, the master version of the bill of materials and the routing, standard costs, inventory valuation, the production plan and financial records.
- In MES: the status of the work order on the shop floor, actual output and scrap quantities, downtime durations and causes, process parameters, quality measurements, operator and shift records, lot and serial traces.
- Flowing between the two systems: from ERP to MES, work orders, bill of materials and routing information; from MES to ERP, completed quantities, materials consumed, scrap and actual times. This flow keeps cost and inventory in line with reality.
Our assessment is that the most frequent mistake is failing to draw this boundary: a setup that tries to write machine data directly into ERP bloats ERP with unnecessary records, while an MES cut off from commercial information cannot connect shop-floor performance to cost and to delivery commitments.
Where do IIoT and machine data fit into this picture?
The industrial internet of things (IIoT) is the collection of data from machines and sensors and its transfer to other systems. This layer is often MES’s eye on the shop floor: machine status, cycle time, counter and alarm information are captured directly instead of being entered by hand. Open standards such as OPC UA are widely used for this communication between machines and systems; among OPC UA’s own information models there is also a common object model aligned with ISA-95.
The principle is the same here: the raw signal stays on the shop floor, the meaningful event goes to MES and the aggregated result goes to ERP. There is no need to carry every second-by-second sensor reading into ERP; what ERP needs is how long a work order took to complete, with how much scrap and with which materials.
Where should an MES project start?
- Define the decision: which decision will MES help you make better? Starting with one or two concrete questions, such as delivery-date commitments, capacity utilisation, quality losses or traceability, delivers results faster than trying to measure everything.
- Draw the boundary: before the project starts, put in writing which data will live in ERP and which in MES, and how often and in which direction it will flow between the two systems.
- Fix the master data: if the bill of materials, routing and work centre definitions are inconsistent, the data coming from the shop floor is attached to the wrong place as well. The accuracy of MES depends on the accuracy of the master data in ERP.
- Start with one line: starting with a single line or a single bottleneck machine tests machine connectivity, operator use and the transfer to ERP under real conditions; the scope is widened on the strength of that evidence.
- Assign ownership: who is responsible for the accuracy of the shop-floor record? If this responsibility is not defined from the outset between the shift supervisor, quality and planning, the system soon loses its credibility.
The MindDX approach
MindDX starts MES and IIoT projects with measurement: which decision needs which data, where that data is held today and the gap between ERP and the shop floor are established with evidence in the Assess phase; the scope is written into the Validated Scope document and the implementation is planned to that scope. To see how the measure-first, decide-second approach works, explore the MindDX Platform.
Sources
- ISA — ISA-95 / IEC 62264: Enterprise-Control System Integration
- OPC Foundation — OPC UA Online Reference (including OPC UA for ISA-95)
- IEC — IEC 62264-1:2013 Enterprise-control system integration – Part 1: Models and terminology
- OPC Foundation — OPC 10030: OPC Unified Architecture – Common Object Model: ISA-95
- OPC Foundation — OPC Unified Architecture (OPC UA)
- NIST — Current Standards Landscape for Smart Manufacturing Systems (NISTIR 8107)