Ethanol plant learns why pipe cracks

Emerson reports progress on predicting control valve-related vibration failures.

By Paul Studebaker

Absolute Energy has been producing ethanol since 2008 at a plant near Lyle, Minnesota. Now it’s trying to expand plant capacity through optimization of existing equipment and systems. “We’re pushing our plants, and we’re finding the weak points, such as vibration failures,” says Travis Rosenberg, maintenance manager, Absolute Energy. “We don’t need help finding the failures, they find us. We’re looking for solutions to help us prevent them.” 

Rosenberg presented the session “Control valve vibration problems—monitoring, predicting and avoiding them” with Shawn Anderson, senior research specialist, and Adin Mann, simulation technology lead, both with Emerson Automation Solutions, at the 2017 Emerson Global Users Exchange. 

“At the Emerson Innovation Center in Marshalltown, Iowa, we are looking for places where we can monitor valves and vibration, and see where we can predict failure,” says Anderson. “Valves never fail in the field the same way they fail in the lab, and we need field data to support our lab data and help us tie all this data together.” 

Absolute Energy was experiencing repeated fatigue crack failures on a steam let-down system for an evaporator feed (Figure 1). The system let the steam pressure down from 125 psig to 12-14.5 psia prior to a steam diffuser. “We had cracking taking place in an area around the inlet to the diffuser,” Rosenberg says. “Diffuser pressure varies, and it’s often under vacuum. We had repeated weld failures even though we changed the diffuser, tried different materials, and had a system vibration analysis done, which did not lead to a solution. When it failed, we lost revenue as well as the repair cost. 

“One consultant suggested we move the components around to change the dynamics, which was not realistic.” They did try moving the valve further from the diffuser. 

Consultants often recommend trying to overcome a vibration problem by adding mass or stiffening the structure. “We wanted to know if we could address the problem at the source instead of trying to work around it,” says Anderson.  

Emerson and Absolute collaborated on a system to monitor vibration levels before and after the valve was moved. “We have sensors we can put on the valve, log in and see the data,” says Anderson. “We also collected sound level measurements and process pressures—we merged in some historian data.” 

As a side note, Anderson says there’s a lot of mature vibration monitoring technology out there, but it’s not for control valves. Existing technologies are looking at lower frequency vibration, less than about 1,000 Hz. The lab is now looking at vibration levels of 3-5 kHz and even higher, which are acoustic-range vibrations. “The approach is not a product yet, but it may lead to a product in the form of a monitoring system or service, as an add-on or option on new valves,” Anderson says. 

At Absolute, “We put magnetic sensors upstream, downstream, on the actuator body, low on the actuator, and on the valve itself (Figure 2). We also collected the sound pressure level,” Anderson says. A plot of acceleration over time for the sensors (Figure 3) showed that while “some were mild and some were wild,” they were generally below 2-3 G. Then they would suddenly pop up. “We learned the vibration was related to process changes,” Anderson says. 

Meanwhile, Rosenberg realized “We have a lot of similar pieces of equipment in other plants and I wondered, why don’t we see more failures on those? I tried working on the steam saturation, and realized that I had pushed the desuperheat adjustment to raise efficiency, which led to the problem. 

“When we get it fixed, we’ll push it again, but for now, we’re going to give up some efficiency to prevent equipment failure.” 


Fixing the problem 

Mann showed details of how piping configurations and valve characteristics affect acoustic vibration (Figure 4). “Closely coupled systems, where valves are closer to elbows and equipment, have more problems due to flow-induced turbulence,” Mann says. The turbulence from the valve can interact with the turbulence from an elbow and “add up.” The system at Absolute Energy also has a flex connector. “Noise radiates from the connector, and it wobbles,” he says. “Some of the bolts were loosening and coming out.” 

In general, it’s best to observe the guideline to keep elbows at least six diameters before and 20 diameters after a valve. Mann says, “Fisher valve sizing software also provides a warning when a valve application is expected to exceed 110 dBa. This is because there is danger to strain on downstream piping and in the valve, it’s not just because of noise.” 

White paper: How to specify valves and  positioners that don't compromise control

Outlet noise is higher on smaller as compared to line-size valves, and ValveLink software provides noise measurements for both the trim and the outlet, so you can see the effects. 

The valve at Absolute Energy is not a Fisher valve, but “We want to investigate all kinds of valves so we can build generic scenarios,” Anderson says. There’s also a time interval effect—valves that are noisy on startup may be quiet in operation, and it’s hard to be sure how much damage occurs when, so “We are trying to take the data and evaluate the damage as it happens, not overall over time, to help us make predictions,” he says. 

Work in the lab has shown how valve vibration correlates with actuator tubing damage, and how acoustic-induced vibration can cause failure of small-bore connections upstream and downstream of a valve. “There, the problem was frequency coincidence,” Mann says. An 1800-Hz tone coincided with the pipe resonance. “The solution was to quiet it down, to use a quieter valve,” Mann says. “It was as simple as an accelerometer measurement of the frequency and a calculation by someone who understands the basic physics.” 

Back at Absolute Energy, “We didn’t fix the problem yet,” Anderson says. “We’ve looked at the valve sizing, the system, and the diffuser sizing, and we determined the diffuser sizing is too big. The vibration is being caused by different pressure drops due to different operation at different times. The answer may be a smaller valve, or two valves. On the Emerson side, we learned a ton. We want these chances to monitor valves that are having or causing a problem, so we can learn from them with semi-continuous or continuous monitoring.” 

On the Absolute Energy side, it’s not clear to Rosenberg whether they’ll be buying one valve or two, but he says, “With all that these guys have done, I’m pretty sure they’ll be green.”