Agility and Variability – Part 2

Part 1 of this series provided an overview of how the ability (or lack thereof) of production and manufacturing processes to handle variability plays a key role in making the overall supply chain more agile and responsive to the business environment.  

Supply chain systems work with the factory through Manufacturing Operations Management (MOM) – the glue and the middleware. For manufacturing to be responsive to supply chain agility needs, MOM systems need to be highly responsive. But they are not. 

These systems operate in the “Operational Time Horizon” of batch duration, shifts, hours, minutes, and seconds, and their decisions are based on information about the entire production system, not one function or one machine.  

The role of MOM systems is to ensure that the “demands” of the enterprise systems (typically ERP) are applied to the production process (typically through automation systems). MOMs systems should inform both the production floor and enterprise systems on how to respond and adjust based on the varying reality of materials, equipment, and quality on the production floor. 

[MOM systems as they are often considered synonymous with Manufacturing Execution Systems (MES). MES is a sub-set of MOM. For a more detailed discussion on this topic, refer to this blog by Bryan Pope.]  

For the MOM systems to be responsive, they need real-time decision-making capability.    

For these MOM systems to inform and respond in the face of variability, they must have the capability to analyze (in real or near real-time) the entire production process and produce a decision. 

However, in most current systems the decision making in the Operations time horizon using MOM data relies on manual analysis and is mostly retrospective. This is the case across process manufacturing, and especially in batch manufacturing. And while investments in ERP systems have added sophistication to planning systems and investments in OT systems have added automation to the manufacturing process, neither of these can provide automated decision-making across the production process. This capability can only be added at the MOM level. This is the biggest challenge and the biggest opportunity to make manufacturing continuously responsive to variability. 

In the context of a brownfield batch process manufacturing business, let’s look at the key data management requirements for moving to automated, real-time, proactive decision making: 

1. Given the scope of the analysis and decision – the entire production process – a “zoomed-out” view of the entire production process is required – both in terms of scope and in terms of the time horizon (time horizon of shifts, hours, minutes, and seconds).  

The view must include – “what is happening”, and what “has happened”. 

This view must therefore be constructed from a real-time temporal representation of the key variables and states of material, equipment, and process.  

2. To perform meaningful analysis and reach a decision, the information about these production components must be viewed in different contexts depending on the perspective of the question being asked to solve the problem. For example, a variability in yield may require a batch context before performing a unit operation analysis that also requires the equipment context. 
3. While performing analysis for real-time usage, historical data and context often need to be combined with real-time data.  

Existing systems for MOM  

To create a responsive MOM system with real-time decision capability, do we need entirely new systems?  

Let’s look at the current state of systems and pseudo systems used to perform MOM functions: 

• BES (Batch Execution Systems) are often non-existent in batch process manufacturing. 
• Similarly, LIMS systems are also non-existent.  
• Despite the heavy investment in MES systems, the key elements of quality and batch orchestration are missing in most implementations. 

Getting usable data from these legacy systems for any decision automation implementation is nearly impossible. Modernizing them to serve varied and complex queries for data required for analysis and event-driven data service is cost and resource prohibitive. 

 The biggest reason is that the legacy systems for MOM were implemented to be systems of execution and record – not information systems 

The stop-gap PIMS (process information management system) systems are inadequate.  PIMS , aka data historians, were brought in to bridge the gap. But these systems have focused primarily on handling continuous time series data, serving as systems of record. In some cases, they act as a substitute for BES systems for the purpose of providing batch context to the data. However, they lack the event and change driven data streams ready for complex queries needed for decision intelligence. 

Efforts to “retrofit” these PIMS and data historian solutions to be “analytics ready” with patch work of “cloud connectors” and data hubs are frustrating to say the least.

A path to Decision Intelligence  

The good news is that it is possible to build decision intelligence capability with existing systems with incremental investment rather than a complete rip-and-replace.  

The fundamental purpose of the existing systems was transactional processing of data to complete production tasks – a capability referred to as OLTP (On-line transaction processing). If we add on-line analytical processing capability (OLAP) as middleware, we create the foundation for decision intelligence with machine learning and statistical methods. 

It is possible to create an intelligent MOM from these legacy systems, with two core components: 

• Modern Industrial DataOps: Event and change driven data integration, contextualization and processing– in the Operational Time Horizon – of real-time and historical data of different types and granularities. Such a system can serve this data to multiple data products (consumers) in real-time or on-demand. 
• Manufacturing Intelligence: Advanced Analytics, SPC, Industrial AI, Optimization and agentic AI acting in synchronization with the MOM’s DataOps.   

The role of manufacturing in the agility of a supply chain is becoming very evident to manufacturing leaders. You can no longer deal with the variability caused by supply chain disruptions and demand side competition without addressing the manufacturing functions. Manufacturing Operations Management (MOM) systems are the glue between supply chain systems and the production floor. Legacy MOM systems (multiple) are becoming the biggest roadblock to making this “glue” flexible for agility. But there is a way to build and intelligent MOM systems to overcome this challenge.