How AI is making Preventive Maintenance a reality for the Manufacturing Industry

Blog / How AI is making Preventive Maintenance a reality for the Manufacturing Industry

“Innovation distinguishes between a leader and a follower” – STEVE JOBS

Only 60 companies from the fortune 500 list from 1955 could maintain that same status in 2017. (Source: Link) . The secret sauce of their success? “Consistent innovation”. Is the manufacturing industry taking progressive steps to avoid such scenarios? A walk through the history of the manufacturing industry can answer that question.

What mainstream manufacturing is currently and why it needs to get an up-grade: 

The advent of the industrial revolution during the 18th century marked the golden milestone for the manufacturing industry, but the maintenance of those state of the art machines was done only when they broke down. Post WW2, when the entire world was recovering from the destructive war, Japanese engineers started a new trend, called “Planned maintenance”, where consistent audits along with following the manufacturer’s instructions would avoid machine breakdown. This trend is still largely followed by the mainstream manufacturing industry today. Planned maintenance improved efficiency for more than five decades now, but industries today have lakes of data coming from different silos, which are difficult to analyze, and is a signal that planned maintenance needs to be replaced by “Preventive Maintenance”.  A data-driven AI approach to analyzing myriads of data silos is essential to make sure that machines are delivering at the optimum efficiency with minimum downtimes. With AI and machine learning, we have the ability to process massive amounts of sensor data faster than ever before. This gives companies an unprecedented opportunity to improve upon existing maintenance operations.

AI transformation is making machines happier and you more profitable: 

Machines rust if not used and also drill a big hole in your pocket when they break down. Breakdown scenarios not only show up on your variable cost of production but also makes reduce revenue. According to the International Society of Automation, the global downtime cost per year is $647 Billion. That is a huge amount which could be used for producing and delivering better customer experience at a better price. So how can downtime costs be reduced? The inevitability of downtime costs is a thing of the past now. Artificial Intelligence is the answer. This beacon of hope can bring unprecedented cost-cutting for the manufacturing industry. Look at these early adopters who are already experiencing the competitive edge with AI/ML integration.

4 Success  Stories for Preventive Maintenance: 

AbinBev is using data-driven AI & ML solutions to increase the filter run length by 40%-50%:

Anheuser-Busch InBev (AB InBev) makes some of the world’s oldest and most popular beer brands — including Budweiser, Corona, and Stella Artois. Before their process transformation, they used meters to monitor and react to adverse conditions in the filtering process, such as a change in pressure. A top priority at AB InBev was optimizing the K Filter that kicks in at the end of beer brewing, right before packaging. The challenge they faced was regulating pressure in and across the filter alongside other complicated processes that involved many unpredictable variables. All these tasks were manually checked which brought in human errors and influenced the overall taste of the beer. Preventive Maintenance, the AI & ML solution built by Pluto7 Consulting Inc helped AbinBev to increase in the filter run length by 40%-50%. This solution not only helped AbinBev to cut their manufacturing costs but also brought consistency in the taste of the beer produced. Here is the entire case study if you want to learn more. 

Lockheed Martin uses big data for F-35 maintenance: 

Lockheed Martin Corporation is an American global aerospace, defense, security, and advanced technologies company with worldwide interests. They wanted to move away from their traditional maintenance methods. Maintenance technicians manually assessed and tracked damaged areas by placing a transparent film over the areas and traced the reference points. This was a cumbersome and time-consuming process leaving more room for potential maintenance errors. It goes without saying that the accurate assessment of the mechanical structure of an aircraft after combat or environmental hazard is crucial. Even small factors — such as the depth of a scratch or the distance of a hole from supporting structures — can impact flight-worthiness. Lockheed Martin not only designed and built the F-35, but it also delivers F-35 sustainment support. To increase the life of their parts, they needed to prevent any damage which might reduce its flying hours. Keeping that in mind they turned to 3D technology and big data to streamline the diagnostics and maintenance processes for its F-35 and F-22 fighter planes according to a case study provided by the Industrial Internet Consortium. This resulted in more accurate data capturing and increased operational availability of equipment. When an aircraft lands, maintainers on the flight-line can connect to the database and immediately determine if the aircraft is flight-worthy.

Failure alert for wind turbines with 95% accuracy: 

Minimizing operations and maintenance  (O&M) costs is of critical importance for wind farms. On average, O&M accounts for up to 30% of the cost per kWh produced over the life of each wind turbine.  Any failure could results in long waiting periods due to wind farms usually being in remote locations and spare parts that are difficult to find. Not to mention the entire logistics cost involved in delivering these expensive and sometimes massive parts. In total, approximately 5% of revenue was lost due to unplanned shutdowns. It looked inevitable to the leadership until preventive maintenance knocked on their doors. With one year of historical data analysis, the deployed solution was able to alert the maintenance team on an average of 52 hours before downtime with an accuracy of 95%. This resulted in reduced maintenance budget, longer uptime, and increased revenue.

South American steel manufacturer predicting failure with 93% accuracy, reducing operational cost by 15%: 

The steel industry has been known for innovation ever since the golden period when Andrew Carnegie walked this planet. Supporting a growing economy and infrastructural expansion, steel manufacturers had to produce at scale at an optimum price. To achieve this they took the support of advanced machinery.  And this continues even today!  The only thing that has remained constant is the demand for maintenance of these machines. A major South American steel manufacturer was getting hamstrung due to unscheduled downtime and high O&M cost due to labor overtime expenditures. Both were negatively impacting production and revenues. With the AI-powered Preventive Maintenance solution, the alerts were triggered 8 days prior to failure, with an accuracy of 93%. This lead the manufacturer to a 30% reduction in downtime and 15 %reduction in operational cost. 

Where to go from here: 

The answer to the above question is limited by your imagination. Though the discussion was centralized around manufacturing, predictive maintenance can be extremely useful to other industries as well. The global economy is on the first step of a big technology revolution. If you wish to dive deep into and discover how we are using AI and ML to transform organizations in multiple industries, follow our official page Pluto7 – ML and Smart Analytics Experts on Linkedin.