fault detection analytics

Manufacturing companies are familiar with the concept of Lean Management, but what about analytics for manufacturing? Consultants have long believed that one way to improve efficiencies is for a manufacturer to have a lean supply chain. However Toyota’s 2010 “Sticky Gas Pedal” recalls led many analysts to conclude that the 19 deaths and nearly 300 accidents could be potential side-effect of such practices. Product gurus came forward to say there was “too much cost-cutting”; while other pundits implicated rapid growth which was the result of “too much ambition”. Even company top management admitted having stretched more than [they] should. But nobody could actually quantify “too much”, i.e., how much is too much?

When you are running a global conglomerate which manufactures dozens of highly complex engineered products, and sells them profitably in as many countries, how can you know if you are overstretching? This may seem to require an almost mystical level of “self-awareness” on a company’s behalf. Would it be even possible to measure something like this?

A product’s profitability hinges on two things: how much it costs to produce and how much premium it can fetch because of its appeal. One can argue that a successful car company had solved the latter problem with their reputation for quality and reliability. It may seem natural then to go after reducing costs. A standard tactic in the industry to reduce costs is by sharing parts across many car models and sourcing as many parts as you can locally. In other words, improving supply chain efficiency.

The argument from analysts was that by commonizing too rapidly, car companies’ end up amplifying the effect of a problem. The sticky gas pedal problem for example, affected as many as 8 models. The basic question now becomes the following: is manufacturing efficiency increased or decreased by pursuing commonality? 

Single mindedly pursuing one narrow objective without regard to the larger picture can lead to such disastrous consequences. The same is true with analytics for manufacturing.

The good news is that with today’s sophisticated analytics tools and processes, acquiring the big picture is very easy. Companies are sitting on what some call the “most compelling asset“: the data that resides in their supply chain information systems. Currently most companies use this asset to monitor only a handful of metrics or Key Performance Indicators (KPIs). But simply obtaining a KPI is not all that this data is good for. To completely leverage this asset, manufacturing companies must develop models and processes that can simulate different scenarios, by integrating data from several different functions. By building and testing such holitic models, manufacturers can be better prepared for dynamic market conditions. 

Today all of this is possible for even the smallest of manufacturers. High performance computing cheaply available on the cloud, combined with highly capable open source analytics software makes further delay in implementing analytics a crying shame. Especially in an industry that is in desperate need of rejuvenation – analytics for manufacturing is no longer a nice-to-have capability. 

Originally posted on Mon, Jan 23, 2012 @ 09:15 AM

No responses yet

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

simafore.ai