Data Rich but Information Poor - Unlocking the "Black Box" in Your Plant

Why does variance exist between expectations and results in a production plant? It boils down to this - our knowledge about the interactions between the affecting variables in a manufacturing process is limited. We have some knowledge about the inputs and constraints that are supposed to produce the desired outputs, but that knowedge is limited - especially as it relates to the interactions between the input variables and constraints. In other words, we have a "Black Box" that we don't fully understand, and until we do, variance, waste and loss will continue.

Fortunately, most manufacturing plants have piles of data related to the various inputs and contrainsts and the actual results. But until the interactions that occur between the the inputs and constraints are known, we continue to to have variance attributed to the "Black Box." Interactions occur in the relationships between raw materials, ambient conditions, human factors, production equipment settings and conditions, etc. They produce waste in the form of unavaialbility, speed or yield reductions, quality defects, excessive energy costs, raw materials waste, wasted labor, etc. (for a visual overview, see the attached figure "Data Mining in a Nutshell").

In a nutshell - you can't be Lean if you don't foigure out the "Black Boxes" that create variance in your processes - Variance = Waste.


Data mining utilizes a range of quantiative techniques to:

  1. Organize seemingly random variables in to factors
  2. Structure factors into linear or non-linear relationships
  3. Define structural relationships as dynamic algorithms.
  4. Convert algorithms into predictive models.
  5. Employ predictive models to gain control and reduce variance, waste and lost.

Algorithms define the "Black Box."

There are two forms of Data Mining Exploratory and Goal Driven

  1. Exploratory - Most plants are aware of some of the "Black Boxes" in their process. However, many of the Black Boxes are hidden. We employ Exploratory Data Mining to uncover the hidden Black Boxes. For example, there are many Black Boxes that reside at the interface points between functional groups in the organization (e.g. marketing, design, supply chain, production, maintenance, etc.). Because each functional group views the business in different ways, communications and hand-offs between the groups are often complicated and unreliable. These are Black Boxes.
  2. Goal-Driven - Once you've identified the black boxes that you want to control, we employ Data Mining to create predictive models to optimize, minimize or maximize whatever we hope to control - cost, profit, emissions, safety, etc. Predictive models and algorithms can be targeted to enable more effective control over a single process, an entire plant, a functional group or the entire value stream.

These powerful Data Mining Techniques can change your business and enable your Lean Business practices.

Remember - Data is the Difference Between Deciding and Guessing. But data is only useful when it's transfomred into information and actions.

I look forward to your comments and questions about the deployment of data mining to enable Lean, Reliable and Safe manufacturing!


Drew D. Troyer, CRE, CMRP