Process Mining – Big Data Working for Manufacturers

In the article Why Manufacturers Need Process Mining – A New Type of Data Analytics, the author is extolling the benefits of what he calls a new type of data analytics – process mining.  Process mining can be used to reduce inventory costs, identify production bottlenecks, improve on-time delivery, and optimize logistics between production sites, distribution centers, and end clients.  It’s hard to justify  that process mining is actually a new type of data analytics that can be used for manufacturers, but process mining does follow the a few of the key rules for successful big data usage.  First, as the name implies, it concentrates on one process.  Implementation of successful big data projects normally requires concentration on one area for improvement.  The specificity of big data projects usually allows them to be successful, and the article looks at why process mining, and big data analytics on the whole, can be successful in saving manufacturers money.

Process mining specifically looks at a very important factor in manufacturer’s processes, KPIs or key performance indicators.  KPIs are exactly what they sound like, the main factors that measure performance and the overall successfulness of process or project.  Process mining’s value lies in the fact that it initially makes one look at KPIs of a process.  It also challenges the manufacturer to validate if they are measuring the right KPIs and understand if the data they are gathering can be related back to KPIs.  Process mining, as all big data does, uses software do the difficult work of visualizing the processes and highlighting specific variances impacting KPIs.  One example used in the article is the examination of throughput times and the ability to identify that specific vendors are not meeting their lead-time commitments.  These types of analyses and results is exactly what big data projects are meant to do and achieve.

Another point that the article points out is that process mining encourages manufacturers to identify inefficiencies and problems within the process.  It encourages companies to embrace their issues.  This ideology is absolutely necessary for continuous improvement, and it is a key to any big data project.  Embracing issues can be difficult, especially for older, engrained processes, but it is the only way to eliminate them.  It certainly is not the easiest thing to do, and requires a humble mindset entering an improvement project.  Process mining has all the key factors of a successful big data software, and it could be very useful for manufacturers that want to embrace the big data revolution.


Do you believe addressing KPIs first is the best way to approach a big data project?

Do you think process mining is actually a new or different type of big data analytics or just a rebranding of basic big data?

Do you think that some manufacturers are reluctant to implement big data projects because they do not want to know their inefficiencies?

Caterpillar is Saving Big Money using Big Data and the IoT

In the article IoT And Big Data At Caterpillar: How Predictive Maintenance Saves Millions Of Dollars, the author examines an interesting case of the company Caterpillar saving significant amounts of money using big data and the IoT.  The best part of this case study is that Caterpillar is seeing a very quick ROI on their big data investment, which is not something that can be said for most companies.  As a Caterpillar manager put it, you don’t have to look for a “grand slam” with big data; sometimes you just need multiple smaller applications of big data to experience significant savings.  In this instance, gathering as much data as possible seems to be the best approach, and utilizing experts in the processes and in the data to analyze and understand the insights gleaned helps realize real value.

Caterpillar utilized big data on in its Marine Division, mainly to analyze fuel consumption for its customers as it most affects the bottom line.  Sensors on the ship monitored everything from generators, engines, GPS, refrigeration, and fuel meters, and Caterpillar utilized Pentaho’s data and analytics platform.  Insights gained have been a correlation between fuel meters and amount of power used by the refrigeration containers, and also that running more generators at lower power instead of maxing out a few was more efficient.  The cost savings here added up to $650,000+/year.  Another insight was to the optimization of a ship’s hull cleaning schedule.  Through the collection of data of cleaned and uncleaned ships, the data showed that cleanings should be performed once every 6 months instead of once every two years.  The savings associated with this optimization was $400,000/ship.

In the grand scheme of big data, predictive maintenance analytics seems to be the most powerful tool consistently being used.  With data being generated just about anywhere you could imagine via the IIoT, understanding trends becomes easier and easier.  Interestingly and contrary to previous articles, Caterpillar believes that you can’t collect too much data.  They point out how data storage is very cheap.  In the words of a Caterpillar manager, you can’t see “relationships about relationships” in the data if you don’t collect it.  Although a more is better approach is definitely not what some companies have ascribed to, it seems to be working well for Caterpillar’s marine division as they continue to pull value of out of big data and analytics.

Quick returns on big data investments seems to be rare, so do you think that companies just aren’t utilizing the big data correctly?

Do you believe in the more is better approach with regard to collecting data?

Do you believe Caterpillar is more likely to invest in big data projects in other parts of their company due to the success in the marine division?

Looking for the Signal in the Noise

In the article A Signal in the Noise: How to Best Manage Big Data, the author advocates some ways to improve the use of data for manufacturers as well as his idea for the best way to approach the use of big data analytics and data management.  The author equivocates big data as sometimes “a needle in the haystack” when it comes to finding and using it in the manufacturing environment.  With so much data produced by even small firms, the management, organization, and analyzation of the data can become overwhelming if not impossible.  Industrial Internet of Things-based systems are estimated to create $4 trillion to $11 trillion in new economic value for manufacturers by 2025, according to McKinsey Global Institute.  With such large value to be had, ignoring or underutilizing data could be a catastrophic mistake for manufacturers.

In a study of manufacturers conducted by the MPI Group, 76% of respondents reported plans to increase their use of smart devices in the next two years while 66% plan to increase their investment in IIoT-enabled products over the same time period.  With many firms taking the necessary steps to collect and use data, a strategic approach is necessary.  The author suggests a number of steps to jump-start the transformation to using data analytics.  Recognizing human limits and the burden of isolation is the first suggested step.  Here the author is advocating that firms understand that human teams are just not capable getting the insights that technology is capable of.  The next step is forgetting the traditional supply chain cycle, and embrace the complexity of modern supply chains.  With the IIoT, data can be captured all along the supply chain which can offer useful insights not previously possible.

Finally, the author advocates understanding four different measures before taking on new data management and analytics initiatives.  Find the actual problem, the “signal in the noise” is the first and foremost issue at hand.  Unless there is direction to the project, the project will almost certainly fail.  Next is understanding the business case.  One has to understand the strategic advantage behind any data management implementation.  Lastly, finding where the optimization is most desperately needed and identifying the experts that are best suited to handle the data being generated is key.  Overall, the biggest pitfall in bringing on new technology is the belief that good things will just happen.  An understanding of the problem, the right people, and a tailored process will allow the technology to do the work it is supposed to.

Do you believe that understanding big data is really about finding the proverbial needle in a haystack?

Do you believe the increase in smart device usage automatically translates to more useful big data, or just more data in general?

As with most strategic IT plays, projects start off to gain a strategic advantage but quickly become part of the IT infrastructure.  Do you think we are in the middle stage between big data being a strategic advantage and a necessary IT infrastructure requirement for manufacturers?

McKinsey Study Shows Big Data is Affecting Numerous Industries Including Manufacturing

In the article How McKinsey’s 2016 Analytics Study Defines the Future of Machine Learning, the investigates a McKinsey study released in January of this year that shows how far big data and analytics have come in the previous 5 years as well as what to expect in the very near future.  The main insights gleaned show that the use of big data is growing in numerous industries, and that the key realizing the value of big data is integration and adopting from the top down.  For manufacturing specifically, machine learning leading to preventative maintenance and predictive analytics is one of the most powerful advantages that has emerged.  Unfortunately, just about every industry examined in the paper has still has not captured more than 50% of the potential value from data and analytics.

The original paper released by McKinsey released in 2011 speculated that manufacturing would find the greatest potential value in data and analytics from possible drops of 50% in product development cost, 25% lower operating cost, and 30% gross margin increases.  As of 2017, the new McKinsey paper shows that manufacturing has only captured 20-30% of this total value with the main issues being siloed data in legacy IT systems along with skeptical and sometimes unsupportive leadership.  With regard to one of big data’s most powerful uses, machine learning, manufacturing stands to realize significant value from machine learning.  In the areas of real-time optimization, predictive maintenance, and forecasting, there is definitely some value to be had in regard to machine learning.  In the areas of strategic optimization, predictive analytics, and understanding data trends, there is substantial value to be had.

Design-to-value, supply chain management and after-sales support are three areas where analytics are making a financial contribution in manufacturing.  Unfortunately, manufacturing as a whole has lagged in adoption with a few advanced players capturing a significant amount of the value.  Digging deeper to see exactly where the money has been saved, analytics has benefited the supply chain and production aspects of manufacturers, but only the small groups that have adopted the technology.  McKinsey estimates that only 5-10% of manufacturers have used analytics for financial savings in the supply chain and production aspects.  In the supply chain, the cost savings has come in the form of 10-15% reduction in product cost, and in production side, the cost savings has come in the form of 10-15% reduction in operating costs.  Overall, the main hindrances to more big data implementation is senior management involvement and designing and implementing the appropriate organization structure to support data and analytics.

Do you believe big data analytics adoption accelerate over the coming years?

Do you believe the results of the study are completely accurate?

How far away are we from a time where if manufacturers are not using analytics, they will not be able to compete in the market?

How Analytic is transforming Supply Chain

The article (Forbes, Bernard Marr, April 22, 2016, talks about how analytic is transforming supply chain management.

There are many applications can be developed for supply chain utilizing analytic. To manage inventory effectively unstructured data can be analyzed with technology such as digital camera analytic to monitor stock level. Forecasting can be benefited by machine learning algorithm with predictive modeling. Distribution center and warehouse can be operated with minimum human intervention.

How visual analytic will change supply chain across manufacturing? How analytic will change supply chain in the transportation industry? How analytic will change supply chain in the retail industry?

Looking Forward at Smart Manufacturing and Manufacturing Industry Changes

Manufacturing has navigated a roller coaster of a landscape over the past 10 years, and the future appears to be ever-changing.  Whether it be the election of a new president in the US who promises to increase manufacturing jobs or the increasingly important technologies such as the IIoTs or big data analytics that make up the Fourth Industrial Revolution, manufacturing is transforming.  This transformation is from legacy, antiquated manufacturing techniques to what is called smart manufacturing.  Smart manufacturing is making data more and more readily available and thus the entire manufacturing process is benefiting.  And as many other articles have stated, any manufacturers that do not invest in this upcoming technological revolution risk being overtaken by competitors that do.

Though the switch to the digital data transformation isn’t quite do or die just yet, it certainly will be soon.  A report from the Conference Board of Canada from earlier this year found that some manufacturers are working with 20-year-old software.  It would appear that some in the industry do not feel that making that adapting to the newest technology is necessary which could be disastrous.  There is a golden opportunity for small to medium size manufacturers to experiment with smart manufacturing solutions as firms of these size have nimbleness and ability to make digital changes quickly.  Furthermore, applying digital technology can be as simple as smart tags for better real-time tracking of inventory or cloud technology to aggregate and transfer information across the supply chain.

Some other simpler but advantageous smart technologies that small to medium sized firms can implement are mobile technologies such as smart phones, smart tablets, and smart watches as well as IIoTs.  Mobile technologies are relatively easy to implement and now affordable, and they allow managers to record and share real-time data easily.  The IIoTs is also catching on quickly with 40% of firms surveyed saying they already have some form of the IoTs in place.  The IIoTs can be used to optimize processes as well the key benefit of preventative maintenance.  Despite these advantages, nearly one-third of declining manufacturers expect to decrease their IT expenditures in the next year, which means firms in this category will be hitting serious roadblocks in the very near future.


Do you think the lack of investment in digital technologies could be due to management’s fear that these technologies might actually lead to a decrease in the need for actual human workers?

Does an decrease in IT spending absolutely mean that a company will not be investing in any smart technologies?

Do you think that another holdup with regard to implementing even simple digital technologies is the lack of ability to quantify the cost savings?

Penn State Beginning $1.5 Million Smart Manufacturing Project


In the article Researchers to Begin $1.5 Million Project for Manufacturing Industry, the author explains how a teams from Penn State, Case Western Reserve University, GE Global Research Center, and Microsoft are collaborating on a smart manufacturing research project.  The project aims to help machines internally recognize signs and wear and fault conditions through the use of cloud-based wireless sensing and prognostic system for monitoring machinery health conditions.  The overall goal is for the machines themselves to make intelligent decisions about fault detection and proactive maintenance scheduling, hence smart machines.  The use of the IoTs, cloud computing big data analytics, and machine learning allows this smart ecosystem to exist.

The project brings experts from across academia and industry together.  On the academic side there are experts from the fields of industrial and manufacturing engineering and mechanical and aerospace engineering.  On the industry side there are experts from GE’s industrial internet of things global research center and senior managers from Microsoft.  Penn State will be hosting the project, but the project is 50% funded by Digital Manufacturing and Design Innovation Institute (DMDII).  DMDII, which was founded in 2014, is funded by the US department of Defense along with number of other entities with the purpose of digitally transforming American manufacturing.

The plan is to begin in February of 2017 and run the project for 18 months.  The actual workings of the project will be done in FAME (Factory for Advanced Manufacturing Education) which is a 10,000-square- foot integrated high-bay laboratory for teaching and research.  Senior Research Associate Wu stated the overall purpose of the project as “The overall goal of this research is to establish a generic framework for real-time process monitoring, diagnosis and prognosis for smart manufacturing using cloud computing and big data analytics.”  It’s easy to see that with lofty goals as described above, this smart manufacturing project could help transform the industry.  Only time will tell how successful the project becomes, and possibly even more important, how the results are viewed and accepted by the greater manufacturing industry.


Do you believe the team will be successful at creating a generic framework that can be used by a majority of the manufacturing industry?

Do you think the project will stay within budget and finish on time?

What are your thoughts on the fact the Department of Defense is funding part of this project?