Wednesday, January 20, 2016

Optimizing Manufacturing Operations with Predictive Analytics

For many manufacturers, collecting data has become an important practice in the manufacturing process. However, fewer have found ways to make that data actionable and convert it to business intelligence. Harnessing data to empower predictive analytics can increase efficiency and maximize returns on investments. 

What is predictive analytics?
Predictive analytics is the process of analyzing historical and current data and applying advanced statistical methods and analytical tools to make reliable predictions about the future. In manufacturing, this involves several steps:

Step 1: Identify a Pain Point
Whether a recurring fault, or wasted resources, first the problem must be identified.

Step 2: Collect the Right Data
Look at all of the factors that could be leading to the problem and identify what kind of data is needed to help solve the issue. Examples include temperature, vibration, acoustic forces, deflections and similar technical inputs. Collect the data using sensors and software that can connect to your equipment and systems.



Step 3: Turn Data into Intelligence
Once the data is prepared, a variety of data mining, machine learning and statistical techniques can be used to uncover numerical patterns and relationships between variables. Looking at these results, manufacturers can not only identify what is causing a problem, but also can gain a better understanding of the conditions under which quality and yields are highest.

Step 4: Take Action
This information is then used to create a model that can make reliable predictions. Rule sets and algorithms based on the model can be used to automatically detect errors earlier in the manufacturing process, cutting down on errors, scrap and other wasted resources.


Whether you’re looking to solve a single pain point, or increase efficiencies at a plant or enterprise level, leveraging data is a great place to start. While predictive analytics is not a standalone decision-making tool, it can supplement and support human experience and intuition and improve the quality and accuracy of decisions.


Tuesday, January 5, 2016

The Missing Link in Global MES/MOM Implementations


There is an old MES saying, “MES is not a project event; it is a process. It is 80% cultural and 20% technical.”  Many manufacturers’ corporate IT departments are incorrectly viewing MES as another extension of ERP and thereby incorrectly believing that each plant, each line and each work cell operate under a common set of business and operations models. Individual plants often reject MES implementations because the global application does not have functionality to address their specific set of operations’ pains and needs. Instead, individual plants end up developing their “shadow” IT in custom Excel applications, paper forms, or database applications to help the plant characterize and fix the current problem set.

Two barriers that many manufacturers face when considering an update to their system architecture are:
    The lack of vertical industry instance or templates of standards
    The cost of replacing established legacy systems using disparate data models and metrics supported by point-to-point interfaces



To overcome these issues, automate processes and optimize operations, Global MES Implementation teams must understand that each plant has its own manufacturing form of work processes and personnel culture based on “What it Makes, How it Makes it, Where it Makes it, and Who it is Made for.” Operations process and data standards must be engineered to enable effective communication between systems in plants and their supply network.

Like most technology evolutions, this change in IT architecture usually requires a substantial investment of time and dollars. However, implementing the use of best practices from The Open Group Architecture Framework (TOGAF), Software Engineering Institute’s model, Zackman’s Framework, ISA-88/95 or other similar models can help manufacturers cut down on those costs. If properly implemented, a return on investment can be realized quickly through increased efficiencies and decreased waste, the result of better analysis of a more complete data set.