SUBMITTED BY: Jamie Duart
TITLE: Level III Vibration Analyst
EQUIPMENT: Carbon Furnace Draft Fan
TECHNOOGIES APPLIED: Vibration
PROBLEM DESCRIPTION:
The Carbon Furnace Induced Draft Fan is an overhung, radial blade fan mounted on top of the carbon furnace. It is a direct-drive, variable speed application. Vibration amplitudes vary greatly from month to month and overall vibration is over 1 in/sec on some surveys. A long history of work orders exist in our CMMS system ranging from physical removal of fan for cleaning to bearing replacement to adding mass and stiffness.
INVESTIGATION:
Analysis: With many variable speed applications vibration data goes in and out of alarm from survey to survey. In many cases this is a result of resonance. Resonance is when a forcing frequency such as imbalance, misalignment, looseness, bearing defects, gear defects, etc., coincides with a natural frequency. The natural frequency may be that of the rotor itself or may be that of the support frame or foundation or even drive belts. It was suspected that the Carbon Furnace Induced Draft Fan at our facility was experiencing resonance and different testing methods were employed and much research was done to prove our theory.
Comparative Analysis: On the following page is a horizontal, vertical and axial view of the coupling side fan bearing on the Carbon Furnace Induced Draft Fan. The spectrums show that the readings in the horizontal direction are significantly higher than in the other two directions. One of the characteristics of resonance is that it causes much higher vibration in one direction as compared to the other tri-axial directions. When resonant, it is common for the resonant direction to be 5 to 15 times higher than in the two other directions.
Natural Frequency and Resonance Detection: Component natural frequencies can be detected utilizing several different methods including Impact/Impulse Natural Frequency Testing (also referred to as a “Bump Test”) and Run-up and Coast-down Testing. Impact Naturally Frequency Testing should be performed with the piece of equipment shutdown and could not be performed due to production schedules. A Run-up Coast-down Test was the best option in this case and we employed a technique of this test called a Bode’ Plot. This Plot consists of two different plots – one of amplitude versus RPM and the other of phase versus RPM. Past vibration history showed that the Coupling side fan bearing, horizontal position showed the most response and was the point chosen for our testing. The results of this test can be found on the next page. Data shows that the characteristic 90-degree shift and rise in vibration amplitude seen at resonance occurs at approximately 64% of these units operating speed or 1360 RPM.
Questions arose as to whether this resonance may have been caused by the fan being out-of-balance or misaligned. These conditions could have been the forcing frequencies that coincided with the structures natural frequency and resulted in resonance excitation. A laser alignment had been performed immediately before conducting the Run-up/Coast-down test and the assumption was made that this was not the forcing frequency responsible for the resonant condition. The phase shift that had occurred at 1360 RPM and the steady decrease in vibration as RPM was increased to 2081 RPM was uncharacteristic of an out-of-balance condition. Attempts to balance this fan would likely have been unsuccessful without changing the natural frequency of the structure. This natural frequency may be lowered by adding mass or raised by adding stiffness.
Physical and Analytical Research: Closer examination of the structure confirms that support and stiffness is inadequate. Conversations with Robinson Fans (the manufacturer of this fan) stated that they recommend the foundation supporting the assembly weigh a minimum of five (5) times the weight of the assembly (motor, fan, bearings, shaft and base). In this case the assembly weighs approximately 3700 lbs. total. Five times this amount would be an approximate value of 18,500 lbs. Since this fan sits on two I-beams mounted to the top of a furnace it is doubtful that the base approaches this value. The tank-like construction of the furnace may not be able to support a massive foundation like what is recommended.
The fan company representative also addressed our concerns about the A-frame base that the motor and fan bearings are mounted to not being adequate to dampen frequencies excited by the fan. He explained that the weight of the base assembly (including the motor and fan bearings) provided by Robinson was roughly ten (10) times that of the fan and should provide adequate damping (fan and shaft weight is approximately 320 lbs. and the motor, base and fan bearing weight totals approximately 3700 lbs.).
Another concern we had was that this fan may not have been designed for a variable speed application and that we were exceeding the fans operating speed or operating at a critical frequency of the fan. Robinson explained that they typically design their fans with a critical speed at least 40% above the maximum speed rating. Maximum speed listed on the fan curve is 2183 RPM and currently the fan does not exceed 2081 RPM.
RECOMMENDATION:
After careful review of the data including vibration analysis, fan company recommendations and work order history in our CMMS system it has been determined that the best course of action would be to relocate this fan to the level above the furnace. This would allow installation of a more substantial base to change the natural frequency of the unit and eliminate the resonant condition that is occurring.
CONCLUSION:
In April of 2006 the Carbon Furnace ID Fan was moved from the top of the Carbon Furnace to a newly constructed platform on the level above. Structural I-beams were added and a concrete pad was poured. Fan vibration was checked upon startup and levels were found to have risen at 2081 RPM from approximately .25 in/sec to nearly .5 in/sec. Although vibration amplitudes had risen there was little change in phase. Phase only fluctuated about 30 degrees from 40 to 100%. Vibration amplitude did see a slight rise at about 80% but did not appear to be a resonance (characteristic phase shift was not present). Time constraints prevented completion of a precision field balance but running speed vibration was lowered to approximately .25 in/sec. Balancing was accomplished the following month with final readings of .04 in/sec at 2081 CPM.
Below are pictures of the new structure as well as pictures of the old location on top of the furnace.
