Viewpoints: Machine Tool Analytics
By Dean Bartles, PhD
Vice President and General Manager
Large Caliber Weapons & Ammunition
General Dynamics-Ordinance and Tactical Systems
St. Petersburg, FL
Isn’t it interesting how a term catches on and spreads like wildfire? Surely the word "analytics" has been around a long time. Recently, however, it seems I can’t get through a magazine article, conference, or blog site without the term coming up. So what does "analytics" refer to, as the term is being used today? Wikipedia defines it as "the application of computer technology, operations research, and statistics to solve problems in business and industry."
Just how ubiquitous is the term? When I did a simple Google search on the word "analytics" about a week before writing this article, there were 196,000,000 hits. A week later, as I sat down to write, I repeated the search and discovered 294,000,000 hits. Can you say "exploding"?
I refined my Google search, went to the "advanced search" feature, and restricted my search to "business analytics" (the exact term) and found 1,480,000 hits. Although less widely used than just the word analytics, it’s nevertheless extremely common. You may have noticed a couple of recent books on the subject of business analytics by Thomas H. Davenport, distinguished professor in information technology and management at Babson College (Wellesley, MA), and Jeanne G. Harris, executive research fellow and director of research at the Accenture Institute for High Performance Business (Chicago). According to these authors, companies are emerging to use business analytics as a source of competitive differentiation.
What about "manufacturing analytics"? An advanced Google search showed only 2730 hits at the time I wrote this piece. Almost everyone involved in manufacturing is familiar with many so-called "analytics" such as percentage of scrap, cycle time, hours per unit, overall equipment effectiveness, SPC data, and efficiency (actual hours per unit compared to a computed standard). And there is the plethora of six-sigma terms we’ve been bombarded with, such as RPN (Risk Probability Number) and Cpk (Capability of the process).
This leads me to the term "machine tool analytics," which I have not seen or heard anywhere. In fact, when I did an advanced search on Google for the term "machine tool analytics", there were zero hits. Nada! Here is my prediction: one year from now, I will perform another advanced Google search, and find thousands of hits, if not tens of thousands.
Machine tool analytics will be the next big thing on the manufacturing shop floor. Imagine every lathe, milling machine, drill press, saw, forge press, heat-treat furnace, and chemical bath streaming data about every single element of its operation. Every single feed, speed, depth of cut, coolant viscosity, coolant temperature, tool-tip temperature, energy consumption, tool-tip vibration, motor vibration, coolant flow, oil viscosity, and acoustic emission on every individual machine in every plant in the country. A virtual tsunami of machine-tool data will flow from manufacturing operations everywhere, everyday.
But won’t data streaming cripple a company’s network? Welcome to cloud computing! Because the type of data I’m referring to are, let’s say, "generic" in nature, and will not reveal the product being machined, nor the dimensions of the product, companies shouldn’t fear transmitting data to the cloud, where third-party service providers will be able to study data, trend data, analyze data, and send extremely useful information back to the manufacturing plant.
Let’s say your experienced CNC programmer writes a program for that new, never-been-machined-before part. As you begin machining, data stream to the cloud for analysis and, based on the database of a million-plus similar machines that have performed similar operations over long periods of time, immediate feedback is provided on how to optimize the operation by increasing the feed rate or using a different insert. Based on historical data, you will have a very high degree of confidence that you will not exceed the limits of the machine and/or the tool. As you continue to run, data streaming from sensors on the machine indicate that the spindle motor is beginning to show a vibration pattern that is typically followed by failure within the next 1000 machining hours. Based on this intelligence generated from the third-party service provider, using the machine tool data that you are sending to the cloud, you schedule the repair during a time when production is not planned.
All of this capability is becoming much easier to obtain due to the proliferation of MTConnect (if you have never heard of it, Google it), affordable sensors, and cloud computing. Machine Tool Analytics is coming. ME
This article was first published in the March 2011 edition of Manufacturing Engineering magazine. Click here for PDF.