Use predictive maintenance to monitor and manage your compressed air system

Incorporating predictive technologies into day-to-day operations has the potential to improve reliability, increase uptime, and reduce maintenance costs.

As Kaeser Compressors’ engineering manager, Neil Mehltretter has conducted and overseen thousands of industrial compressed-air studies and helped users achieve energy savings and operational improvements. Timothy Hitzges is the product engineer for master and local controllers at Kaeser Compressors Inc.

During the live Q&A portion of the webinar "Using Predictive Technologies with Compressed Air," Mehltretter and Hitzges tackled several attendee questions on how to proactively improve your compressed air system.

PS: What is the best interval for taking a compressor oil analysis sample?

NM: It depends. Usually, you want to take an oil sample right away after the first run or after a few weeks to get a good baseline. Then you want to take another sample at approximately 2,000 hours, depending on your service intervals and the type of oil/fluid used. Some fluids are designed for longer service life. Best practice is to do it at every scheduled fluid change. Installation site factors like humidity, dusty environments, etc may warrant more frequent fluid sampling.

PS: What are some tips for collecting energy or other operating data between or before doing a trend analysis?

TH: First you will want to decide on the time period. Typically, when you're performing a trend analysis you are looking at data over a several weeks or months.  However, you can perform a trend analysis based on two weeks of data, if your plant operations allow for it. Some plants have very stable demand profiles and two weeks of tracking is probably a decent sample.

Many plants, however, have significant changes in production during the course of a year. If production changes seasonally you many want to sample during slow periods as well as high production periods. From a trend-analysis, you can look at flow profiles over time and figure out how your leaks have changed and what your baseline flows are. The only way to really do this is if you have more data. If your plant is seasonal, then December might be your highest month and March might be your lowest month. If you have a three-month interval, that might be a great opportunity for you to figure out how your temperatures run, how your flow performs, your leak load during federal holidays, and more. If you are conducting a temperature analysis on your machines, the compressor room might be very cold in the winter, and your temperatures are going to be relatively low compared to summer months. From that standpoint, you might need to have seasonal information or trends over a year.

For predictive technologies to be truly beneficial, a trend analysis should be ongoing to help the system continually improve.

PS: Do you recommend investing in leak detection equipment, or should you outsource the task?

NM: That's a tough question to answer. Some people want to do everything in house and some people want to outsource some of these smaller tasks. It depends on your company. If you have limited resources and a very small maintenance staff that is always busy, then I would definitely contract that out. If you have a large facility and maintenance staff and you're doing quarterly assessments on your leak detection, then it makes more sense to invest in some equipment and training. You can conduct leak detection in stages rather than doing the whole plant in one session. If you know that some equipment is more prone to leaks, it makes to have the tools to stay on top of it.

PS: What is the most important data point?

TH: I just have to smile a bit because there's so single most  important data point or factor. There are two general areas to consider: compressed air application and compressor environment. Pressure and temperature are two important  values equipment function and compressor health. If you really want to narrow it down and you have to sacrifice data points, then that's certainly something you want to look at. It’s not necessarily that there's one important data point; it's always a combination of the data that you need.

Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments