MRC Intern Uses Meridium Weibull Analysis to Eliminate Pipe Erosion
Results in Documented Savings and Efficiency and Safety Improvements

By Michael T. Saale, UT Graduate Student, Mechanical Engineering

Working with Eddie Bozman of DuPont and Eric Watkins (another UT intern) we analyzed the rate at which a high temperature mix of aqueous titanium dioxide was eating through the inside of the tubing of its heat exchanger. The pipe thicknesses are being collected on a monthly basis using a program called Ultra Pipe. The old procedure for deciding when to replace a section of pipe was to schedule the section for replacement when it reached 1/4". If the thickness dropped to 1/8", then the section became a priority and might even merit shut down.

Previous attempts to determine the rate at which the pipes were eroding yielded inconclusive results. This is where Meridium came into play. By analyzing the data using Meridium’s Weibull Analysis, we were able to separate out what we initially thought were outliers.

When the outliers were looked at by themselves, it was clear that most of them were also displaying a trend. The first erosion trend was caused by normal operation. The second erosion trend was later determined to have been caused by loss of process control resulting in a decrease in the pH of the process fluid.

Despite the clarity of the erosion patterns displayed, the beta value for each trend wasn't as high as would be preferable. Eddie looked at what Eric and I had done and filled in an important gap in our knowledge. The formation of precipitate in the process fluid was sinking to the bottom of the pipe and causing that section to erode faster than the rest of the pipe. He then sent me out to observe the data collection procedure so that I could see which reading corresponded to which locations in the pipe section.

In order to extend the life span of each section, during each shut down, various sections of pipe were being rotated 45 degrees. If a section made it seven rotations without a loss of process control event reducing a section of the wall thickness to 1/8", then it was scheduled for replacement at the next opportunity.

When I separated the data points with respect to their position in the pipe and accounted for the temporary jump in erosion rate immediately after the section was rotated, the erosion rate at each position became clear.

Using this research, a planned maintenance event was developed allowing the facility to conservatively save $120,000 this year. It will also generate additional savings as more is learned about the effects of a loss of process control. This data could lead to a better formula which would allow for a tighter margin of safety.