The previous chapter focused on cost, delivery, quality, energy, and labor challenges. This chapter moves to a more practical question: if ERP already has orders and work orders, why is MES still needed? And if an injection molding machine already contains many parameters, why is the data still difficult to use?

The key is that the systems operate at different levels. ERP knows how many parts a work order requires. The machine knows that the current cycle took 31.8 seconds. Without a layer that connects those two facts, the factory still does not know which physical cycles belong to that work order.

Author: Jush Li

2.1 Four Practical Areas of Smart Manufacturing

Smart manufacturing can be understood as the ability of equipment, production data, and management systems to exchange information continuously so that the factory can see what is happening and respond based on actual operating conditions.

AI, IIoT, MES, and cloud platforms are tools. The fundamentals are whether the data can be acquired correctly, whether the factory knows which machine and work order the data belongs to, and whether sufficient history is retained.

For injection molding, cycle time, cycle count, machine status, barrel temperatures, pressure, and other process variables can all become useful sources. However, the available data varies by machine brand, controller generation, and options. A factory should not assume that every machine can provide the same process curve or update rate.

1. Data as an Asset

Historically, many machine values existed only on the controller screen. If today’s cycle time is 32 seconds but no record is stored, the value cannot be compared a week later.

After connectivity is established, cycle time, cycle count, operating state, and other available machine data can be stored continuously and linked to the work order and product.

The value is not the amount of data. The value is that when a problem occurs, the factory can go back and see what actually happened.

Figure 2.1-1: Digitizing injection molding production data

2. Cyber-Physical Integration (CPS / Digital Twin)

Mold-flow simulation, CAE, and Digital Twin are often discussed together, but they are not the same thing.

CAE is mainly used during product and mold development to predict filling, packing, cooling, shrinkage, and warpage. It helps engineers compare designs before mold trials.

A Digital Twin goes further. It requires a continuing relationship between a physical asset or process, a digital model, and actual operating data. A model used only for offline simulation is better described as simulation or CAE rather than a complete Digital Twin.

Figure 2.1-2: Process curves can be one source for real production analysis

3. Anomaly Detection and Model Analysis

AI can be used for visual inspection, anomaly classification, process analysis, and equipment alerts, but not every AI system should directly change machine settings.

For example, a vision system can classify short shots, flash, or other defects after suitable samples and inspection standards are established. Process data can be used to compare normal and abnormal cycles. Predictive maintenance can be evaluated after enough operating and failure history has been collected.

Whether a system is allowed to write process parameters back to a machine depends on machine capability, validation, permissions, and factory governance. In many plants, “detect and alert first, engineer confirms next” is the more practical sequence.

Figure 2.1-3: Vision systems can assist defect classification after inspection standards are defined

4. Flexible Production

Flexibility in injection molding is not only about changing the product on the machine. Mold changes, material preparation, drying, mold preheating, staffing, and scheduling all have to work together.

MES, scheduling tools, mold management, and SMED can reduce unnecessary waiting. For example, the next mold and material preparation can begin before the current work order is completed.

AGVs, AMRs, robots, and automated handling are another layer. Whether they are justified depends on volume, travel distance, process design, and investment return.

2.2 What ERP Does — and Why MES Is Still Needed

ERP remains essential for manufacturing. Orders, purchasing, inventory, cost, work orders, shipment, and finance are natural ERP responsibilities.

But ERP and injection molding machines operate at different levels of detail.

ERP may know that 10,000 parts are due today. On the floor, stable production may start at 8:20 a.m., stop at 9:35 because of a material issue, resume at 10:02, and later change cycle time after a shift change.

Those minute- or second-level production events are not the main design purpose of a traditional ERP system.

Figure 2.2-1: ERP manages enterprise resources and work orders; machine data requires a different layer

ERP Is Strong at Enterprise-Level Data

• Customer orders and due dates
• Purchasing and suppliers
• Raw-material and finished-goods inventory
• Work orders and required quantities
• Cost and financial information
• Shipment and accounting

The Shop Floor Needs Finer-Grained Data

• When a machine actually started and stopped production
• Actual cycle time during each production interval
• Cycle count and production quantity
• Which mold was used
• Machine downtime and abnormal states
• Operator and shift
• Inspection and quality results

The issue is not that ERP is obsolete or incapable. The issue is that ERP should not be forced to perform every machine-data collection function by itself.

MES / IIoT Fills the Shop-Floor Data Gap

Injection molding machines can provide data through OPC UA, a gateway, or existing machine interfaces. MES then links that equipment data to the work order.

A single value such as Cycle Time = 31.8 seconds has limited meaning. If the system knows it belongs to Machine A, Work Order X, Product Y, Mold Z, at 10:32 a.m., the value becomes traceable production information.

Figure 2.2-2: IIoT acquires machine data for upper-level systems

Modern ERP Can Integrate, but the Roles Are Still Different

Modern ERP systems commonly provide APIs and integration mechanisms. ERP, MES, WMS, and equipment platforms can exchange data without forcing everything into one application.

The more important design decision is to define which system owns which data, and how common identifiers such as work order, product, machine, and time are shared.

A Data Platform Must Solve Relationships, Not Only Storage

Machine time-series data, work orders, quality records, and inventory can live in different databases. The important part is that they can still be joined when a production question needs to be answered.

A data lake with millions of sensor records has limited value if the factory cannot determine which records belong to a specific work order.

Figure 2.2-3: The key is a shared production index across data sources

2.3 What MES Does in an Injection Molding Factory

MES (Manufacturing Execution System) sits between ERP and the production floor. It turns production instructions into work that can be executed, recorded, and traced.

For an injection molder, the most important MES function is not a large dashboard. It is the ability to link the work order to the machine, product, mold, operator, and production period.

Figure 2.3-1: MES functions should be selected according to actual factory needs

MES products do not all include the same modules. The table below reorganizes common manufacturing-operations functions into eight practical categories for injection molding rather than treating them as a fixed universal “eight-function MES standard.”

Injection Molding Needs Additional Domain-Specific Functions

General MES functions are useful, but injection molding has several production objects that require special attention: molds, material lots, recipes, molding conditions, and changeovers.

Figure 2.3-2: Common MES extensions for injection molding

Mold History and Maintenance

The same product may use more than one mold, and a mold may undergo multiple repairs. Recording only the product number is not enough for quality traceability.

Mold ID, machine used, accumulated shot count, repair reason, and repair time are useful parts of the production history.

Recipe and Parameter Management

Molding conditions can be retained as a historical baseline so engineers know approximately what cycle time and process conditions were used previously for the same product and machine.

Parameter management should not automatically be equated with writing settings back to the machine. Automated recipe download requires access control, equipment compatibility, and version management.

SMED and Changeover Support

The system can remind the team to prepare the next mold, material drying, colorant, and related items before the current order is complete.

SMED still depends on process improvement: separating internal and external work and moving preparation outside the machine-stop window. MES can support task and time management but does not replace the method itself.

Raw Material and Lot Traceability

Injection molding often involves resin grades, colorants, recycled material, and lot control. When color, strength, or appearance issues occur, the ability to trace a material lot back to the work order can be important.

2.4 Injection Molding IIoT: EUROMAP 77 / OPC UA and the Value of Connectivity

The goal of machine connectivity is not simply to mirror the controller screen on a PC. It is to make machine data usable by MES, production reporting, quality systems, and other applications.

If cycle time exists only on the machine screen, an engineer can see it at that moment, but the factory cannot automatically compare it with historical production. Once connected, the data can be stored and related to the work order.

Figure 2.4-1: Connected machine data can support monitoring, performance, maintenance, and energy analysis

Real-Time Monitoring

The simplest use is to monitor machine state, cycle time, cycle count, and other available process data.

If a product historically runs at about 30 seconds on a certain machine and current cycles gradually increase toward 35 seconds, the system can flag the difference.

A longer cycle does not mean the part is automatically defective. Cooling, robot waiting, operator actions, material, or other process changes may explain the shift. The system should identify the deviation first and let the production team investigate.

OEE Analysis

OEE requires more than knowing whether a machine is on or off. Planned production time, downtime reasons, performance speed, and good-count definitions must be consistent.

Connectivity can acquire machine state automatically, but mold changes, waiting for material, maintenance, setup, and other reasons still require downtime categories. Otherwise, the result is only a large amount of undifferentiated “stop” time.

Predictive Maintenance

Predictive maintenance requires more data than basic machine networking.

If equipment provides torque, pressure, temperature, vibration, or other health signals, and the factory also retains failure and maintenance history, it becomes possible to evaluate whether certain changes correlate with actual failures.

A few weeks of data without labeled failure events is usually not enough. For many small and mid-sized molding factories, digitizing preventive maintenance, shot counts, and repair history is the more practical first step.

Energy Analysis

Energy analysis also requires actual measurement. A machine controller may not provide complete electrical data, so a factory may need smart meters or an existing energy-management system.

Once energy data is available, the factory can compare different products, machines, and idle periods. A useful comparison is energy per cycle or per kilogram of output rather than only the monthly plant bill.

What Does EUROMAP 77 / OPC 40077 Solve?

Factories often operate injection molding machines from different brands and generations. If every machine exposes a different data structure, the MES layer needs brand-specific interfaces.

EUROMAP 77 / OPC 40077 defines an OPC UA information model for communication between injection molding machines and MES so that machine information, status, molds, jobs, and related production data can be exchanged in a more consistent way.

It is not a specification for MES to remotely control every machine movement. Safety-related control and direct machine-motion control are outside its scope.

EUROMAP 82 series specifications extend OPC UA information models to selected peripheral equipment, such as temperature-control and hot-runner devices.

2.5 What Benefits Should Be Measured?

Smart manufacturing should not be evaluated by claims that “implementation always improves X by Y percent.” Every factory starts from a different baseline. The better approach is to define the baseline first and then measure what changed.

Figure 2.5-1: Smart manufacturing benefits should be verified through production KPIs

ERP, MES, IIoT, and AI do not replace one another. ERP manages enterprise resources and orders. MES manages execution and traceability. IIoT brings equipment data into the system. AI is applied later to problems that are suitable for model-based analysis.