Many organizations report the cost of downtime at $100,000 to $300,000 per hour or more. At that rate, taking measures to predict and prevent breakdowns is an important investment. And with the power of machine learning, big data, and condition monitoring at your fingertips, maintenance teams can use predictive maintenance analytics to increase equipment uptime by up to 20%.
We’ll share all the details of leveraging predictive maintenance analytics, and show you how it’s done.
How data is collected
In a predictive maintenance program, sensors are used to collect data on assets. There are different types of sensors that can measure temperature, vibration, pressure, and more. Some of the sensors commonly used include:
- Infrared analysis sensors are used specifically to compare the difference in temperature of a component over time. These sensors can be used to monitor bearing temperatures in large motors, measure piping conditions, calculate process temperatures, estimate plumbing conditions, verify solar panel conditions, evaluate the condition of electrical components, and more.
- Vibration analysis sensors measure parts vibration to detect faults related to misalignment, mechanical looseness, gear defects, lack of lubrication, resonance, rubbing, cavitation, corrosion, and more.
- Ultrasonic analysis microphones use a sensitive microphone to pick up high-frequency sounds and detect the need for electrical inspection, steam trap maintenance, improved lubrication, leak maintenance, mechanical inspection, electric arc flash detection, and valve testing.
All sensors must be connected to CMMS software. Through the power of the Internet of Things (IoT), the software uses real-time sensor data and predictive models to generate maintenance notifications.
This way, maintenance staff can monitor the health of a machine and act on the root cause of an issue before it becomes a catastrophic breakdown.
Big data and predictive maintenance analytics
Big data refers to large and complex data sets that grow at ever-increasing rates. Within the context of predictive maintenance, it means that as sensors keep sending real-time information on asset condition, the amount of data on hand grows at a constant pace.
In the field of maintenance, the term big data often refers directly to the use of predictive analytics. And that’s because big data is the foundation of predictive maintenance solutions.
Two vital components are necessary to perform big data analytics for maintenance: building a database, and applying predictive analytics models to the data.
Building a database or “data ecosystem”
Before any predictive algorithms or models can be used, all data must be gathered, structured, aggregated, and stored so that it is accurate and usable. This initial stage constitutes around 70% of most data science projects.
These datasets have traditionally been stored and analyzed independently of one another. However, with the power of the Internet of Things (IoT) technology and other advances, it is now possible to connect the systems that measure and collect data, and those that perform an analysis. In the end, this offers a more complete picture of the asset lifecycle.
Data Requirements for Predictive Analytics
Predictive modeling applies statistics to a machine’s historical data to predict outcomes such as potential machine breakdown. However, the accuracy of any prediction depends on three things: relevance, quantity, and quality of data.
When generating “training” data — that is, historical data used to develop predictive models— the data must relate directly to the predictions you want the model to make. This means that you must have an idea of the types of predictions and equipment failures you want the algorithm to detect.
For instance, if the goal is to predict the failure of a traction system, the training data has to encompass all the different components for the traction system.
Designing prediction systems around specific components and their failures rather than larger systems will help provide more specific and actionable predictions.
Another important aspect to consider when building the “training” data is the amount of information needed to make accurate predictions.
Since every piece of equipment is different, there is no one definitive answer to the question of how much data is enough data. However, the more data about specific equipment you have, the more accurate your predictions will be.
Keep in mind that there are failures that are rare in some types of equipment, which means that the “training” data won’t gather this event unless you make sure to include it.
Quality is perhaps the most critical element of building historical data. Data is probably your most valuable intangible asset, so making sure it is accurate will be worth the effort. Data scientists regularly verify and “cleanse” data before using it to train or develop a predictive model.
Make sure to incorporate data quality fundamentals into your PdM project plan such as maintaining an audit trail as “training” data is being prepared, having an assigned individual responsible for data quality, and obtaining independent quality assurance.
Sources for condition-based machine data
There are multiple sources for historical machine data. The main sources include failure history, maintenance history, condition monitoring data, and equipment metadata. Learn how to collect and use machine data
As you know, most equipment doesn’t have unplanned downtime often. Therefore, it becomes challenging to obtain data directly from failure events.
The data used to predict failures tends to come from a component’s normal operations. When building predictive models, algorithms use maintenance records and parts replacement history to find failure events. For this reason, data that includes detection of anomalies can also help predict potential failures.
Maintenance data history is a vital source of useful data. This data includes details about spare parts used, components replaced, repair activities performed, equipment condition at the time of the repair, and more.
Every maintenance activity performed on an asset can contribute to the algorithm’s development of degradation patterns. Failing to include this vital information in the training data can lead to misleading model results.
Machine operating conditions
Real-time data from condition-monitoring sensors indicate an asset’s performance and operation condition. A key assumption in PdM is that a machine’s health status degrades overtime during its routine operation, and real-time condition data helps identify those patterns..
With historical data, artificial intelligence can determine characteristics that capture patterns of aging as well as any anomalies that indicate deterioration.
Long-term data is required for the algorithm to learn the failure and non-failure patterns over time. Based on this information, the algorithm learns to predict how many more units of time a machine can continue to work before it needs to be replaced.
Static feature data
Different from time-based data, static data is information about specific features and characteristics of each asset. Some examples include:
- Manufactured date
- Start date of service
- Control system location,
- Other technical specifications
Adding this type of information to the data used to make predictive models helps the algorithm make more exact predictions.
Leverage PdM Analytics to Streamline Decision-Making
Predictive analytics can be a game-changer for your maintenance team. But the implementation can be challenging if you don’t have the right resources to do it.
The first step is determining whether your organization is suited to this type of investment. And if unplanned breakdowns represent large losses for your business, the high ROI will be worth it.
When problems can be predicted in advance, they become solutions. Predictive maintenance strategies allow teams to lay out a plan to action with minimal effect on production and very little cost to the operation.
Limble CMMS connects a complete Predictive Maintenance module with other crucial modules such as Work Order management, Preventive Maintenance management, Enterprise Asset Management, Spare Parts Inventory, and more.