Jake was in a quandary; his Six Sigma project was due and he hadn’t even selected a project let alone worked on it. He had taken the classes but was so busy with other projects that he had neglected his own. What to do? Then, he noticed a moderately low brine cooling refrigeration machine due for a tune-up. Maybe there would be some savings there he could claim.
[pullquote]
Jake liked some aspects of the training, especially the statistics and the ability to cross-reference variables to determine if he had a measurement or metering problem. So, using all of his measurement expertise, he began setting up his project.
Measurement would be critical in this -25°C system. He checked the metering runs to see if he had the proper length upstream and downstream. He asked operations to pull the orifices to make sure they weren’t fouled or damaged. He logged the results. Next, he looked at the temperature sensors and their locations. The temperature elements were calibrated and this also was logged into his project files.
[callToAction ]
In a previous project, Jake and a Black Belt had discovered an instrument problem that caused the interplay between various levels of cooling across the plant to be out of balance; that inefficiency led to excessive energy losses. Jake suspected this system also was out of balance, so he wrote his project statement accordingly.
Jake began his project with a baseline performance test extended over three months. He checked the plant’s online data historian and soon realized that the historian truncated data — if conditions didn’t change, the historian only would store the beginning and ending data point for that period. Jake understood that could be a problem. He analyzed his data at the end of that, and as one of the statisticians said, “you have to torture the data till it gives you what you want.”
To cross-check the data, Jake set up an energy- and mass-flow balance around the machine. As one of his coworkers, Ray, always said, what goes in should equal what goes out. From the baseline, Jake formulated a plan for changes to the system to reduce energy consumption. One of the issues with this machine is that it had been converted to an alternative refrigerant and was operating at near the lowest temperature possible for the machine. Jake proposed raising the operating temperature slightly and measuring the effects on process conditions.
Operations agreed to the change and Jake set out to collect another three months of data using the historian. He then analyzed the data and shared the results. Jake cross-checked the instrumentation using the heat and mass balances. The data appeared to be within the predefined tolerances. Then, Jake worked on “torturing the data.” He discovered that the change in temperature had little effect on the process; there might be additional room for further increases. The annualized savings from the temperature increase during the test worked out to approximately $150,000 per year. There also was potential for another $75,000/year.
Jake completed his project using online monitoring and a data historian. So, how do you go about doing the same at your plant? You’re probably not planning to complete a Six Sigma project; however, keep two key things in mind before doing online testing.
The first is measurement accuracy. Are your temperature elements accurate and in the proper location? If your temperature difference is 7°C, a 0.5-°C inaccuracy represents about 9% error. If you don’t correct this error, how do you ensure the test is accurate and the results are genuine? A similar case can be made for flow and pressure measurements. Perform a cross-check using an energy and flow balance to see how far off you are and then you must determine where the inaccuracy exists.
The second is the data historian. Do you know how the data are collected? What is the time period of this data collection? Is it an average for the time period or is it for the time it was collected? Is the data truncated? Do you know how to extract that information? What time period are you going to do your testing? Can the monitoring and data collection system handle your requirements? Can you add calculation blocks to the data historian to cross-check the heat and mass balances to insure data accuracy?
Good luck and happy energy hunting!
