How to create an effective Predictive Maintenance Program

In order to establish a Preventive Maintenance Program, it is critical to establish a solid foundation, consisting of clearly defined methodology and desired outcomes. Organizations must adopt a robust strategy to successfully embrace predictive maintenance programs. Some basic steps towards achieving a successful implementation are:

1. Establish a plan and start small
2. Choose right assets for PdM
3. Choose the right condition monitoring methods & equipments
4. Establish Data Collection & Analysis mechanism
5. Pilot Testing
6. Decide on a response procedure
7. Build a data analysis strategy
8. Establish a continuous implementation and improvement process

Let’s dive a little deeper into each of the steps, to understand how-to create an effective Predictive Maintenance Program:

1. Establish a plan and start small

Switching to predictive maintenance is a process and not a single event. It involves changes in hardware, software, and manufacturing operations. Success in predictive maintenance begins with careful, strategic planning especially while applying predictive maintenance to multiple assets. This can avoid data deluge and data loss.

The ideal approach is to start small, focus on a single pilot program at a time and establish a robust and reliable process. Once the pilot becomes a success, other processes will fall in place.  

2. Choose right assets for PdM

The success of the pilot program depends upon the right selection of assets for PdM. There are many classes of assets that may be the right candidate for predictive maintenance.

• Assets with a history of failure can be excellent candidates for the pilot program.
• Critical assets that need constant condition monitoring.
• Troubled assets that can stop anytime leading to shutting down the line or unscheduled downtime.
• Assets that are difficult to replace either financially or logistically.
• Hard to reach, remote assets that may either be in an inaccessible place or a hazardous location.

3. Choose the right condition monitoring methods & equipments

Once you choose the assets for predictive maintenance, select the right condition monitoring procedure. This may include the parameters to be monitored viz. sound, vibration, RPM, temperature etc. and the instruments to monitor them. It is essential to choose the correct parameters and instruments as it can make or break the maintenance prediction.

4. Establish Data Collection & Analysis mechanism

The next step is to monitor and collect data from the assets for establishing a baseline. This will help the algorithms to evaluate the health of the machines and predict any anomalies. With enough data, the algorithms can learn the machine statistics to predict any future maintenance requirements and the time-till-failure for these machines / components. Cloud-based storage and advanced artificial intelligence systems that display results on a simplified dashboard, can help analyze vast amounts of machine-health data, and provide faster conclusions, thereby making it a simple and easy-to-use mechanism.

5. Pilot Testing

Choosing a pilot program depends upon whether you are looking for short-term results or long-term benefits. For instance, vibration analysis is the most effective monitoring system for rotating assets as it acts as a continuous monitoring system which will have long-term impact on machine health

6. Decide on a response procedure

The next step in predictive maintenance is establishing a procedure to respond to anomalies. You may need a full-time technician who can monitor the statistics on the dashboard and communicate any anomalies / discrepancies in the health of the machines. Based on the results, they may give instructions to either shut down the system, run for a limited time or run indefinitely with constant monitoring.

7. Build a data analysis strategy

Before extending the predictive maintenance system to more assets, a proper data analysis strategy must be established, especially for continuous monitoring.  It is a crucial component of any PdM program. Data analysis can be performed either by third-party service providers or by in-house experts. Organizations must determine if the staff need to be trained or an expert with the required skills need to be hired. Companies with limited resources can avail the services of third-party individuals for online condition monitoring. An online condition monitoring sensor can enable them to access data remotely, track machine health and troubleshoot any issues.   

8. Establish a continuous implementation and improvement process

Once you gain some success in the pilot venture, you can begin a cycle of implementation of predictive maintenance for all other assets. Run your data against current KPIs to measure the success and ROI. To ensure the growth of the program, the PdM system must regularly monitor the data from critical assets. This needs proper understanding of the processes. From the experiences of past implementations, you can improve the predictive maintenance program to reduce maintenance costs, repair spending, unplanned downtime, and breakdowns.  

Conclusion

In an ever-changing business environment, with competition increasing every minute, predictive maintenance can make all the difference in gaining a competitive advantage. It can change the landscape of the manufacturing sector and help organizations reach unprecedented levels, by focusing on reducing downtimes and improving productivity.  

Taking the right steps in a methodical fashion is the key to establishing a robust predictive maintenance regime.