Episode Details

REPLACING THE EXCEL BOTTLENECK: BUILDING A REAL-TIME DECISION PIPELINE

A modern factory is a hive of real-time data, with sensors, controllers, and supervisory systems tracking every machine state, fault code, and temperature fluctuation. Yet, when a critical resource like "Line 3" stops unexpectedly, the immediate response often grinds to a halt. The physical event is instantaneous, but the operational decision that should follow—the choice of how to reroute work, resequence orders, or escalate maintenance—is too often buried under a mountain of manual work. Teams are left scrambling, piecing together information from disparate systems like ERP, MES, and technical logs, often resorting to outdated spreadsheets to find a path forward. In this scenario, real-time data exists, but real-time decisions do not. The missing element is a deterministic, connected pipeline that turns a physical signal into a validated, feasible operational choice with clear authority.

Key Concepts:

• A Machine Event is a Fact, Not a Decision: A technical log can tell you when a machine stopped and what fault code was raised, but it cannot decide how that stop affects delivery promises, whether an alternate machine is feasible, or which priority order should be moved. Value is created only when that fact is connected to a production decision.
  
• ERP, MES, and the Shop Floor: Different Truths: Each system owns a different aspect of production. ERP owns demand and the business record. MES owns production execution and genealogy. OT owns machine signals and local control. The decision pipeline must respect these different truths and connect them only where an operational choice needs to cut across boundaries.
  
• Replacing the Excel Bottleneck: The spreadsheet remains a powerful tool on the shop floor, often acting as the unofficial integration layer because formal systems lack the agility to exchange context. The goal is not to "ban Excel" but to build a visible decision path that brings together source facts and records assumptions, changing the role of the spreadsheet from the primary decision repository to a true tool for personal analysis.
  

Understanding the Factory Scenario: Line 3 StopsA critical bottleneck, Line 3 performs a unique process with only one available alternative, which is also tightly scheduled. A fault code appears, a drive stops, and a sensor flags an abnormality. The plant must now answer four key questions, each sitting in a different domain:

1. Operator: "Can I safely restart?"
   
2. Maintenance: "Is this a quick reset or an inspection requiring downtime?"
   
3. Supervisor: "How long will downstream stock cover the pause?"
   
4. Planner: "What order is at risk, and can it move without creating a bigger problem elsewhere?"
   

Without a decision pipeline, these systems remain isolated, leading to manual investigation and a delayed, risky operational choice.

The Capture Model for DecisionsTo sitting effective operational decisions, an event should pass through a clear deterministic framework:

Capture Detects and produces a traceable event at the source, close to the equipment. This layer must filter out high-speed noise that is only relevant to real-time control logic and pass upstream only the events with distinct ownership and operational consequence.

Interpret Normalizes and validates the raw signal, creating a meaningful, shared definition. A simple "stopped state" tag change from OT becomes "Line 3 Unplanned Stop Active," complete with canonical asset ID, source timestamps, correlation identifiers, and stated conclusions only where supported.

Context Join Connects the event to life production facts from ERP and MES. The pipeline finds the active order, product variant, process ro