Hello everyone! I would like to share some insights with those in this forum who are engaged in measuring maintenance performance over time. If you are in this situation, understanding the trend of a specific indicator can be more beneficial than just looking at its previous behavior. By looking into the future rather than reflecting on the past, you can focus on what truly matters and take preventive actions when necessary. There are two approaches you can use for this purpose: one that treats all past observations equally important and another that assigns decreasing importance based on the age of the observations. The latter method is demonstrated in the attached Excel file "Trend Signal", which utilizes an exponential smoothing algorithm with two adjustable parameters: alpha and beta. Alpha and beta can range from 0 to 1, where higher values of alpha prioritize recent observations while higher values of beta emphasize the trend of recent observations. Both parameters can be fine-tuned using SOLVER. I hope you find this information helpful for optimizing your maintenance performance measurements. Best regards, Rui.
When calculating Mean Time Between Failures (MTBF) over a period, failures are typically given equal importance. However, the question arises as to why it may be crucial to adjust the significance of failures over time in practice.
Apologies for the delay in responding to your question, Josh. It is essential to consider the current circumstances when evaluating any event, as circumstances are subject to change. Therefore, it is more prudent to prioritize recent signals that reflect the recent past and current situation, rather than relying on indicators from the distant past that are unlikely to occur again. This approach is standard practice in performance management. Best regards, Rui.
Discover the significance of prioritizing recent events in maintenance and reliability. Uncover examples of situations where giving priority to recent events can significantly impact maintenance and reliability outcomes.
When analyzing past events, it makes sense to give more weight to recent data compared to older data, especially in the context of organizational changes like autonomous maintenance. This approach allows for a more accurate representation of trends over time. For example, consider a scenario where improved teamwork among operators and maintainers of a production line resulted in observed time between failures of 81, 93, 88, 110, 107, and 115 hours during a certain period. If a simple average is used, the Mean Time To Failure (MTTF) comes out to be 99 hours, which may not reflect the most recent data and fails to indicate a consistent increase in performance. However, by assigning decreasing weights to older data points, a more accurate average can be calculated. For instance, using weights of 1/1, 1/2, 1/3, 1/4, 1/5, and 1/6, the MTTF is calculated to be 106 hours. This weighted average better captures the current trend of improvement. In cases where no ongoing changes are happening, a simple average is sufficient as observations are random around the mean with no discernible trend. However, for forecasting purposes, smoothing random and trend analysis using tools like spreadsheets can be highly advantageous. By optimizing parameters like Alfa and Beta, forecasts for future periods can be generated. In the attached spreadsheet, forecasts for Period 1, Period 2, and Period 3 are calculated to be 131 hours, 139 hours, and 148 hours respectively. These forecasts provide a reasonable estimate of future performance, assuming the current trend continues. Adjustments may be necessary if unexpected changes occur.
I understand your question. Can you provide some guidance on how to accurately assign weightage during the analysis process? Thank you.
Hey Rui, this is a brilliant take! You've hit the nail on the head about how focusing on trends rather than individual data points can provide a more accurate picture, performed proactively instead of reactively. I definitely appreciate the distinction you've drawn between the two approaches to assigning importance to past observations. And, your clarification about the use of alpha and beta parameters in the exponential smoothing algorithm is invaluable. I'm about to dig into the "Trend Signal" Excel file to see how I can fine-tune these parameters. Thanks for shedding light on this crucial aspect of maintenance performance measurement.
✅ Work Order Management
✅ Asset Tracking
✅ Preventive Maintenance
✅ Inspection Report
We have received your information. We will share Schedule Demo details on your Mail Id.
Answer: - Understanding the trend of a specific indicator can be more beneficial than just looking at its previous behavior because it allows you to focus on the future and take preventive actions when necessary.
Answer: - The two approaches are treating all past observations equally important and assigning decreasing importance based on the age of the observations.
Answer: - The Excel file utilizes an exponential smoothing algorithm with two adjustable parameters: alpha and beta. These parameters can range from 0 to 1, where higher values of alpha prioritize recent observations and higher values of beta emphasize the trend of recent observations.
Answer: - The parameters alpha and beta can be fine-tuned using SOLVER, a tool in Excel that can automatically adjust the values of these parameters to optimize the trend signal analysis for maintenance performance measurements.
Join hundreds of satisfied customers who have transformed their maintenance processes.
Sign up today and start optimizing your workflow.