Understanding the Impacts of the Fourth Industrial Revolution

by | Dec 8, 2017

On December 6th, we continued our panel series, this time focusing on the Manufacturing Industry, specifically Industry 4.0.  Joined by Phil Naglieri (Director of Technology at Shockoe), Mason Brown (UX LEad at Shockoe), Mike Upchurch (Chief Strategy Officer at Notch), and Will Middleton (CEO at Cloudy Data Systems) we explored the results of the Fourth Industrial Revolution – including predictive maintenance, improved decision-making in real time, anticipating inventory based on production, and improved coordination of jobs and people.

We are currently in the Intelligence Age or Age of Information – The Fourth Industrial Revolution, revolution measured around manufacturing, but really on the impact of everything around us.  The one we are currently living is characterized by data, analytics, and the merger of technology with the physical, digital and biological worlds.

To get the full interview, check out our YouTube Channel

A Brief History of the Previous Industrial Revolutions
#1 – Agrarian to Industrial (Textile and Industry) & Steam:  This revolution took place in during the 18th and 19th century.  It was a period marked by the creation of the iron and textile industries as well as the Steam Engine.

#2 – Electricity, Assembly Line, Light Bulb, and Telephone:  The second revolution started in the late 18 hundreds and lasted through the beginning of WWI, this revolution was marked by the creation of the combustion engine, the light bulb, the telephone, and the assembly line.

#3 – Nuclear Power, Computers, Computing, and Internet:  About half a century later, and really the only revolution any of us lived through is the third revolution, which appeared with the emergence of a new type of energy whose potential surpassed its predecessors: nuclear energy. This revolution also brought with it electronics and the internet—such as the transistor, microprocessor, and computers and the world wide web.

Question and Answer with the Panelists:

Can you give me some insight into the following statement:

“Manufacturing workers are retiring in droves, with an estimated 2.7 million jobs being vacated by 2025. At the same time, the growth and advancement of the industry is expected to create an additional 700,000 jobs for skilled manufacturing employees over the next decade. As a result, manufacturers are scrambling to fill this knowledge and skills deficit with the next generation of workers — millennials.”

Answer (Phil):  Mobile is essential to Manufacturing Companies, who are not naive to the process of hiring new generations to fill the roles left by aging workforces as we’ve seen happen in the recent past, Mobile brings a sense of familiarity and allows for quicker adoption.  Not only will employers onboard new employees quicker, it will allow them to streamline their processes better.

Question:  AI and machine learning are getting a lot of press so can you tell us a little about it?  Specifically, as it relates to Manufacturing, where possible

Response (Mike):  At a high level, think of AI and machine learning as that you are using machines to do what a human would do if they had unlimited time and capacity.  So for example, let’s say you are the production manager at a plant and you are trying to improve yield.  You’d look at a bunch of process and machine data in reports, make changes and then evaluate the impact of your change.  Machine learning will do the same thing except use math to look at thousands of variables and their interrelationships.  The other thing to know is that 80% of the work to use machine learning with be data acquisition and management.  To be good a machine learning, you have to be good at data.  Finally, the process is iterative and exploratory.  For example, a project might start with a question like, “of the 3,000 points of data we capture, what combination of factors cause defects?”.  To solve it the data scientist will try different sets and types of data as well as different models to figure out what combination yields the optimal insight.

Question:  With the recent addition of bitcoin to the Chicago Board Options Exchange and the Chicago Mercantile Exchange, bitcoin and blockchain are in the news today more than ever. How are these different and is the hype really worth it?

Response (Will):  I have been heavily involved in the bitcoin and blockchain space for over a year now and recently became the chapter leader for the Government Blockchain Association Richmond Chapter. Bitcoin and blockchain have been a wild ride this year and the hype is most definitely worth it. The technology provided by the blockchain and the decentralized value exchange of bitcoin are both extremely transformational and disruptive.

Blockchain is one of the underlying technologies behind bitcoin. Though public and private key cryptography, it allows different entities (i.e. an organization, a person, or even a machine) to prove that it is the owner of some data stored on a public database. In Bitcoin, the user can prove they are the owner of the account that sits on the public ledger.

Question:   Earlier you talked about machine learning can you tell us how it’s being used in manufacturing?

Response (Mike):  There are three common categories of use – predictive maintenance, improvements in things like yield, capacity, and quality, and systems optimization.

Using machine learning for predictive maintenance has proven to reduce downtime, in some cases as much as 50% and increase machine life by 20% to 40%.  An example would be predicting part failure and fixing the issue before it happens.
Improving yield/throughput/capacity and quality – studies have shown a wide range of results.  In some industries, the increase can be in the 10% range, but in others, like semiconductor manufacturing, where they are already good at yield, the results can 1%.  However, in that business, a  1% improvement in yield is worth $100M.  This is accomplished by looking at the end-to-end process and understanding not only how each machine contributes to defects, but also the interdependencies of each machine to each other as well as environmental factors; such a factory temperature fluctuations.
As for systems optimization, think of looking at the entire process from sales through manufacturing and delivery. An example would be using 1,000 variables and 10,000 constraints to figure out how to optimize system performance.  The company that did it raised earnings 50%.  Quite dramatic.  A lot of simulation is used here.  An example would be that large oil and gas companies have simulations of their entire plant so they can do things like see how changes in maintenance schedules affect the entire system using software and then optimize the schedule before doing any physical changes.

What are the challenges?

  • Data
  • Security – fog/edge/local, not cloud
  • Operational – ready to adopt?
  • Finding talent

To get the full interview, check out our YouTube Channel