How Predictive Maintenance Solutions Boost the Manufacturing Process

#Machine Learning
March 27, 2020 13 min read

Predictive maintenance is often discussed by McKinsey, an American management consulting company, in its reports. They announced a $1-4T value creation in 2025 because of IoT (Internet of Things) in factory settings globally. And a huge part of this value is the result of industrial analytics and predictive maintenance implementation.

Applying predictive maintenance tools isn’t necessarily smooth—most companies discard 98% of the data collected. They’re unable to integrate collected data fully into the operational workflow due to the lack of predictive capabilities. In such cases, predictive maintenance software is one of the ways to adapt collected data into learning necessities to carry maintenance out. To implement the described tool, a variety of PdM solutions are introduced to the market.

So, what does predictive maintenance mean and why does it matter? Let’s talk about this and try to identify the basic industry challenges, enumerate PdM benefits, and outline strategies for the successful application of IoT predictive maintenance programs.

What Is Predictive Maintenance?

PdM is a tool for assets’ (equipment) condition learning to detect possible failures in operation via sensor devices. This preventive technology provides real-time data to perform maintenance beforehand.

Four steps can be distinguished in PdM implementation:

  • Baseline and benchmark settings in the form of acceptable condition limits, which have sensors
  • IoT devises installation (e.g., usage of temperature sensor on the boiler to prevent overheating or vibration meter on gears)
  • Comprising devices and software (the IoT device is connected with CMMS (computerized maintenance management system)
  • Schedule technical checks (inspections are automatically launched by a CMMS)

PdM is currently introduced in a number of industries and perceived as a smart solution to tackle problems such as high repair costs, long equipment downtime after unexpected breaks or quick machine depreciation. So, the role of PdM solutions is growing, developing, and providing smarter products.

Glimpses On PdM Evolution

Predictive maintenance is an advanced technical solution for equipment care. There are technical checks based on time benchmark and reactive maintenance. The drawback of the time-based one is a risk to monitor too much or fail to control when it’s critical. Under reactive maintenance, equipment is fixed on demand after something is broken, which isn’t cost-effective. PdM is able to perform both reactive and time-based care as a part of the business strategy.

Predictive maintenance tools, such as vibration monitoring, for example, appeared in the beginning of the 21st century. Their active development was caused by the low efficiency of offline or periodic monitoring.

In this study, three PdM cases are discussed: “They identify vibration measurements from time to time, and compare the data collected.” The quote describes one of the basic principles: establishing a benchmark, performing periodical measurements, and launching maintenance under exceeding set baselines.

In the field of service industry a lot of efficiencies can be performed. Tiny adjustments to variables such as time necessary to perform repairs can lead to huge consequences, and IoT technology, sensors and real-time monitoring, give key starting points to grow.
Bernard Marr,
Strategic performance consultant

PdM tools have been developed and improved since 2000, and now, a lot of programs and applications for condition monitoring have been presented to the market. We can name the most popular fields of monitoring: vibration, ultrasonic, infrared, oil, motor-circuit analyses, laser-shaft alignment, and the newest one—video analyses. All of them address a variety of issues to smooth manufacturing workflow and reduce equipment depreciation.

Predictive Maintenance 1

How PdM Works

As we mentioned above, the first step is establishing baselines. This stage is arranged before sensor installation: you monitor assets’ conditional baselines and learn about abnormalities. Then, after sensor tuning, every time any equipment part shows abnormal parameters, the sensor is triggering a special protocol and CMMS assigns to the technicians required repair.

Predictive maintenance implementation starts with the research of baselines and continues with mounting condition monitoring devices and collecting conditional data. Then, another stage begins: identifying baseline breaches, creating work order (workflow protocols should be designed beforehand), and performing demanded repair or audit.

It’s worth remembering that unexpected problems can occur because, on the first stage, no relevant data has been captured, collected, and analyzed. So, from time to time, conditional monitoring should be repeated and protocols undergo some changes and updates to anticipate unexpected situations.

Forbes, in one of its business articles, presented quite an insight, treating predictive maintenance not as a “silver bullet,” but it’s still worth implementing because it’s reasonable to monitor the basics automatically, and be well-prepared for the unexpected accidents.

A predictive system utilizes information about quality data sets, performance reports, machine learning, and AI algorithms to provide a number of guidelines to address problems as they occur. Using predictive system insights, manufacturers can improve quality in real-time, nullifying issues such as machine downtime, scrapping, or recalls. A/B testing (comparing and deciding on which protocol is more efficient as an action plan) is available as well.

Predictive maintenance is a good start to launch smart manufacturing. When data blocks depicting quality control are available, businesses can quickly gain benefits from AI. Let’s look at the pros in the field of PdM usage.

PdM Benefits for Manufacturers

The objectives of predictive maintenance are to improve production and maintenance efficiency. Standard process improvements (reactive or time-based equipment care) cost much, take a lot of time, and do not solve the whole problem usually without any contribution to overall maintenance strategy (to foresee possible failures).

Modern machine learning techniques allow business to leverage collected data to the fullest and increase KPI, streamline repair workflow, and gain more possibilities to be sustainable (e.g., energy efficiency due to sensor devices to regulate energy consumption).

Production efficiency improvements gained via PdM can increase machine uptime and reduce the costs for machine service. Predictions are made on the basis of the data collected from the devices. It enables technical care to be performed not in case of urgent necessity or under standard benchmarks (e.g., miles driven by car), but being assigned in case of predictive algorithm notices abnormal alarms.

With customized PdM software, it’s even possible to predict the condition of the equipment when considering defects it created in its outputs. Predictive maintenance leads to customizing technical care for each machine as well as some separate machine parts.

For example, we can think of a wind farm with sophisticated machines that have sensors and computer chips that control the wind turbine. Sensors not only control, but they also send data constantly with the state of each wind turbine component.

It gives the possibility to establish personalized care for all the facilities without missing any anomaly. In case there are no devices, and technical maintenance is performed on time or after a break, the business owner will spend more money to provide scheduled monitoring of all well-working machines or to repair those, down-time that can be predicted.

So, let’s follow the list of PdM advantages:

  • Increase in assets uptime
  • Reduce unexpected failure
  • Reduced downtime as a period when machines are not working at all
  • Possibilities to predict serious breaks via predictive analyses

Keith Mobley in “Plant Engineer’s Handbook” presents the following figures regarding PdM software usage:

  • Costs for technical care 50% less
  • Accidental and urgent breaks are cut down by 55%
  • Repair and down-time is reduced by 60
  • Inventory of spare parts can be reduced by 30%>
  • 30% increase in MTDF (machinery meantime between failures)
  • 30% increase in working without a break

Even if we treat these figures skeptically, it’s worth noting that in accordance with Hitachi calculations, even a 10% reduction in maintenance costs will lead to about a 40% increase in sales. This company’s specialists say that with properly tuned PdM software, it’s possible to replace about 30% of your PMs (predictive maintenance tasks). You are also welcome to become acquainted with a real-life case of PdM’s huge pros.

Example of Predictive Maintenance

A coal plant operates using a centrifugal pump motor, which is the key equipment critical for operations. The owners of one plant decided to use predictive technology to focus on unscheduled downtime preventions. Considering that a centrifugal pump is a huge piece of equipment performing rotations, it’s obvious that a vibration meter was chosen.

The coal plant team learned normal baseline measurements, visualized them in the form of a wave graph, and attached a vibration meter close to the inner bearing of the pump. Few operational months later, and the meter identified an acceleration spike. New data was reviewed remotely by the team, and they scheduled an inspection. Loose ball-bearings were found and fixed in time.

Sometime later, the coal plant owners outsourced CMMS design and connected it to the meter. This PdM software implementation resulted in constant pump monitoring and predicting ball-bearing with triggering inspection protocol. No staff is involved in the process except the technicians making repair work.

This example proves that PdM consists of several stages (establishing baselines, connect sensor devices and monitor the benchmarks, develop CMMS and arrange distant equipment auditing, and design action protocol in case of machine failures) that deliver substantial benefits.

The definition of predictive maintenance as a powerful tool to improve technical care is true, and it has huge potential to perfect your equipment service and operate smarter. Here are some success stories about PdM implementation to illustrate the practical efficiency of this modern industry boosting tool.

Success Story

Speaking about powerful examples of success in PdM, we’d like to mention the market leader in FSM (field service management), ServiceMax. They assist the clients to manage problem machine cases offsite dealing with mobile machinery and equipment.

In 2019, they partnered with PTC (an industrial IoT leader) to perform a cloud solution based on the Internet of Things components. With this Connected Field Service platform, companies can conduct predictive maintenance if they do not want to outsource the development of a customized PdM product.

This platform was used by the medical equipment manufacturer, Medivators, and after one year of usage, they reported a 78% increase in repair events performed distantly without the necessity of the technician on-site and predicted beforehand with equipment down-time as short as possible.

Company founder and CSO Athani Krishna says: “Company has ‘window’ into what may happen, and they can manage the case remotely.” That is the very sense of PdM software performance.

Innovecs is another innovative software development company that follows the trends, ready to provide you with high-tech solutions or support your manufacturing business with machine learning consulting services.

Predictive Operation: A Faster Route Into Industry 4.0

The participants of the success story described above are usually striving for more efficient production, increased revenue, and a faster route into Industry 4.0 (smart factories via IoT and Internet of Systems).

Predictive quality systems are used to reach this objective. They utilize data, analytics, machine learning algorithms, and AI power. Quality issues are tackled because enough information about machine failures are identified systematically and easily.

Within the Industry 4.0 concept, it’s much easier to monitor machine conditions, apply captured data, and launch service protocols. It’s possible to start with simple data sets processing, then scale up, explore predictive operations, and move further toward an IoT-driven factory.

PdM software is a critical part of the Industry 4.0 generation plant. It enables revenue to be boosted via smart reporting as well. This can be proved by the African gold mine case, in which they identified a problem with oxygen levels, and started saving $20 million annually.

Predictive operations are also performed widely by Amazon to maintain their robots delivering goods around the warehouse. And the main challenge is not only to cut the staff but to employ unskilled workers able to deal with PdM software.

PdM: Business Path to Success

If it’s decided to implement predictive analyses and quality control, manufacturers gain a variety of business perks via a PdM launch, such as equipment downtime reduction, increase in productivity, staff number decrease, repair costs optimization, coming closer to new generation smart production concept (Industry 4.0).

To introduce predictive maintenance is not a single step. It involves the following stages: establishing machinery baselines, mounting sensors, designing acting protocols in case of machine failure, monitoring data collected by sensors, and performing assigned repairing works.

After an enterprise has made a decisive move to introduce PdM into its operations, it’s worth considering a reliable partner to develop a customized solution to monitor the equipment state automatically. Innovecs can become such a collaborator on your path to optimize manufacturing processes via digital solutions.

March 20, 2020
Supply Chain Risk Management Solutions To Run Your Business Globally
In this article Rocky Osborn, Senior Vice President, Global Business Development & Client Success at Innovecs, a global software development company, provides insights in supply chain risk management, discusses main challenges and solutions, and describes real cases. In “Supply Chain Risk Management: Advanced Tools, Models, and Developments,” Dr. Yakob Khojasteh assumed that the statement, “all the goods produced can be sold,” is no longer true. Today, production businesses should consider optimal output, supply levels, and apply risk pooling (as a form of risk management) to operate efficiently—because factors such as consumer needs diversification and business competition globalization have decreased product life cycles. SCRM solutions are demanded on the market since they help companies to evaluate, forecast, and mitigate evolving risks in their developing supply chain networks. The world’s largest logistics companies, such as DHL, provide tools for business continuity planning, logistics control towers, supplier risk management, trade compliance, supply chain insurance, and sustainable procurement. Software development companies, including Innovecs, can customize your transportation management, assist in collecting data from multiple sources, establish tracking processes, or optimize warehousing. The company follows best practices in SCRM to enhance business opportunities and decrease risks. So, let’s learn more about modern possibilities in supply chain risk management software implementation.
Article, Logistics, Big Data & Highload, Software Development