As industries progress, the perception of the maintenance sector follows suit. Instead of waiting for problems to arise, we see value in creating plans and executing them before problems happen.
This shift also brings about a transformation within the maintenance landscape itself. The focus is no longer solely on repairing machines; it’s about maintaining a steady pace and ensuring constant operations.
Even though maintenance teams are still responsible for fixing significant failures, ideal maintenance plans need us to go beyond corrective measures.
That’s where predictive maintenance (PdM) comes in: to improve maintenance routines and anticipate failures.
What Is Predictive Maintenance?
Predictive maintenance combines data analysis and predictive analytics techniques to anticipate and prevent equipment failures or breakdowns.
It constantly monitors the condition and performance of assets to detect any anomalies or early signs of potential issues.
PdM uses data sources – like sensor data and historical data – to create models that predict the future health of assets. Maintainers can use this collected data to increase asset reliability and availability.
Differences Between Predictive and Preventive Maintenance
Predictive maintenance and preventive maintenance are two distinct approaches to maintenance management. While both aim to prevent equipment failures and optimize maintenance practices, there are key differences between them.
Preventive maintenance involves conducting routine maintenance activities at scheduled intervals, regardless of the actual condition of the equipment. It follows a predetermined maintenance plan to replace components or perform maintenance tasks based on time or usage.
On the other hand, predictive maintenance relies on real-time data analysis and predictive analytics techniques to predict failures. It involves monitoring equipment condition and performance indicators to detect anomalies or early signs of potential issues.
Predictive maintenance is a more data-driven approach to maintenance than traditional preventive maintenance. It uses sensors and other data-gathering tools to monitor the condition of assets and identify potential problems before they occur. This can help to maximize uptime, reduce maintenance costs, and extend the lifespan of assets.
Techniques Used in Predictive Maintenance
Data analysis in predictive maintenance is crucial – it’s how we can anticipate faults and failures.
There are multiple ways to collect and analyze data, one being MDSA. Maintenance Data Systems Analysis is the process of collecting, monitoring, auditing, evaluating, and analyzing data from your CMMS.
Aside from MDSA, here are the 5 main techniques used to collect data:
a. Vibration Analysis
Vibration analysis is vital for detecting mechanical failures and understanding the underlying causes of various defects that can halt production.
Analyzing vibrations shows us changes that impact production, such as mass unbalance, shaft misalignment, bearing wear, structural issues, and backlash.
b. Thermography
Thermography is a graph or image representation of the infrared radiation emitted by a piece of equipment.
This image helps us identify and monitor changes in the temperature of components to avoid overheating.
c. Ultrasound
Ultrasound analysis increases the frequency of sound waves, allowing human auditory systems to detect them.
This frequency increase identifies leaks in gas transportation systems as well as electrical current leaks and mechanical defects.
d. Crack analysis
Conducted using magnetic particle testing, crack analysis is an industrial technique utilized to identify both surface and subsurface anomalies.
This analysis enables the detection of defects like cracks, cold joints, double lamination, lack of penetration, folds, segregations, and more.
e. Oil analysis
The primary goal of oil analysis is to offer insights into both the lubricant’s condition and the machine’s overall state.
By monitoring oil contaminants, analyzing wear, and assessing additive levels, we can determine the best time for an oil replacement.
Oil analysis enables the identification of various issues, including pollutants, gear failures, oxidation, misalignment, additive degradation, and mechanical component wear.
How to Implement Predictive Maintenance
Implementing a predictive maintenance program with cost-effective asset sensing and monitoring solutions is a highly worthwhile investment for any company. It’s advisable to begin with techniques that align with the specific equipment types within your organization.
To facilitate efficient monitoring, we’ve compiled a predictive maintenance checklist – a step-by-step guide to implementing this technique.
a. Start With Most Critical Equipment
Critical equipment refers to those whose failure can lead to significant costs and issues for industries. The level of criticality corresponds to the fundamental role the asset plays in production, thus influencing the priority of maintenance implementation.
ABC analysis is a widely-used method for assessing criticality and risk levels of assets. It typically considers three criteria: failure frequency, difficulty in detection, and impact on overall operations caused by the failure.
There are three classes of criticality:
- Class A: high criticality
- Class B: medium criticality
- Class C: low criticality
Ideally, facilities should start predictive maintenance data analytics on equipment categorized as high and medium criticality (A and B). This will considerably reduce the downtime of the machines, and ensure excellent results.
b. Track Failure-related Information
After listing the critical assets, we analyze the root causes of failures and determine the variables to use for monitoring them.
This is when you’ll start implementing the aforementioned techniques – as you shift focus over to sensing and condition monitoring. You must identify reliable tools, sensors, and software that can accurately collect the necessary data and implement techniques to predict failures.
Predictive maintenance has two ways that it monitors assets: offline and online.
Offline PdM involves manual data collection through sensors, relying on maintainers to perform the data collection process. In contrast, online monitoring utilizes Artificial Intelligence (AI) and Internet of Things (IoT) sensors. These use machine learning to collect data constantly, so maintenance professionals can spend less time physically visiting assets.
c. Notify the maintenance team for action
With the right information, maintenance teams can take action to prevent equipment failures. So, instead of waiting for equipment to break down, work to identify and address problems early on. This will save you both time and money.
AI-based predictive maintenance software can be a valuable tool for managers who want to analyze equipment data. These platforms can deliver information and diagnostics in real time, so managers don’t have to waste time interpreting the data themselves.
To show how this works in practice, let’s take a look at one of TRACTIAN’s customers.
The image below illustrates how TRACTIAN’s maintenance management software has helped Embraer move from corrective and preventive maintenance to predictive maintenance.
With more assertive maintenance management, they began to avoid problems. For example, they stopped a potential unbalance and saturation deviations in a motor-pump filter.
With a more proactive approach to maintenance, they were able to prevent problems before they occurred. For example, potential imbalances and saturation deviations in a motor-pump filter were found, and addressed before they caused a breakdown.
TRACTIAN’s platform generated an insight (as shown in the image above), which allowed the Embraer maintenance team to quickly inspect the asset. In most cases, they were able to find the defect before it caused serious consequences for production.
By anticipating failures, which can be as serious as a total stop or equipment breakdown, the industry has seen a significant increase in the availability and reliability of assets that have received sensors.
Additional benefits of predictive maintenance include:
Technology in Predictive Maintenance
For preventive or reactive maintenance, managers often use spreadsheets to record and analyze data. Paper work orders are also common for documenting a team’s work, but this can get tedious.
Predictive maintenance uses technology to replace manual records, which are time-consuming and inefficient. IoT sensors and artificial intelligence (AI) in software can make the work of maintenance technicians more efficient.
Although AI or big data in predictive maintenance may seem out of reach, it’s becoming more affordable and feasible. The application of these technologies can be a great investment for businesses of all sizes.
TRACTIAN is a predictive maintenance system that incorporates USPTO-patented fault-detection technology to help companies transform their maintenance routines. It consists of two parts:
- IoT sensors that monitor equipment
- An integrated and intelligent platform that manages and monitors assets across an entire industrial plant
TRACTIAN’s suite of hardware and software tools help maintenance and reliability teams in their day-to-day operations. Our solutions are mobile, accurate, and easy to use.
To learn more, schedule a demonstration with one of our experts.
