Plugging a lean, mean big-data-analyzing machine into manufacturing
I grew up in the Detroit area in the 1960s and 1970s, so I'm quite familiar with the wrenching dislocations caused by a manufacturing-based sector that was too stodgy and overbuilt to adapt in a pinch. Coming of age in an ancient industrial zone, you bear witness to how rapidly a smokestack economy can implode when more agile competitors undercut your reason for being.
Being a white-collar kid in a blue-collar town, you may be looked down on as some sort of effete snob in training. However, that feeling fades as you grow up and begin to realize that a manufacturing-based company's success increasingly pivots on the sorts of analytical skills you honed in long years of schooling. Math, science, logic, language, critical thinking, conceptual agility, creativity and curiosity are where the value is added in new-age manufacturing. Consequently, high-skilled knowledge-intensive positions have crowded the low-skilled out of the manufacturing process in most industrialized societies.
Big data, advanced analytics, and statistical analysis are as pivotal to the manufacturing process now as they are to traditionally white-collar jobs such as finance, marketing, human resources and sales. In fact, the data scientists of the world play a pivotal role in companies' lean manufacturing initiatives.
In that regard, I came across an excellent recent article in McKinsey Quarterly that shows how pervasive data science and computation-driven "lean" process-optimization initiatives are in the manufacturing sector. The article points to success stories in the chemical, electronics, mining, metals and pharmaceutical industries. The case studies range across a spectrum of statistical-modeling approaches, including machine learning, neural networks, Monte Carlo simulation and value-in-use modeling. The range of objectives is equally diverse, including cost reduction, efficiency analysis, root-cause analysis, risk assessment, preventive maintenance, bottleneck analysis, throughput maximization, capacity optimization, inventory analysis, production planning and supply-chain optimization.
The authors (Rajat Dhawan, Kunwar Singh, and Ashish Tuteja) nicely summarize the pivotal role of computational approaches in lean initiatives. "Application of larger data sets, faster computational power,...more advanced analytic techniques [and] sophisticated modeling can help to identify waste. [This can] empower...workers and open...up new frontiers where lean problem solving can support continuous improvement. Powerful data-driven analytics also can help to solve previously unsolvable (and even unknown) problems that undermine efficiency in complex manufacturing environments: hidden bottlenecks, operational rigidities and areas of excessive variability."
Continuous process improvement is the key to manufacturing competitiveness over the long haul. And it's the heart of the "lean" ethos that's been reshaping manufacturing everywhere since the 1970s. The olden days of back-office, top-down operations research in manufacturing—a la Robert McNamara and his post-war Ford Motor "Whiz Kids" are dead and gone.
In this new era, "lean" needs to be encoded in the analytics that drive the machinery and the entire value chain.