F e a t u r e A r t i c l e Vibration Reduction with a Dynamic Absorber This problem has to do with two Worthington ERPN 80 centrifugal pumps driven by 125 HP motors at 3,562 RPM. The pumps are used to transfer liquid ammonia during production; at least one pump operates 24 hours a day. Each pump contains two mechanical seals that cost $3,200 per set. The pumps have common suction and discharge headers and are spaced about four feet apart. They are located outside and thus are subjected to ambient temperature variations from -40C to +40C. The pumps had been modified in several ways. The suction piping supports were altered because shifting was causing misalignment so severe that the coupling shim packs broke apart. Mechanical seal failures had to be repaired. Bearings wore prematurely due to lubrication problems; the interior of the fabricated component of the pump base had to be grouted. Modifications have been made to the discharge piping and hanger supports on the minimum flow line. Pump A always ran more smoothly than pump B; vibration amplitudes were close to 0.2 IPS on A and 0.3 IPS on B. The check valve and piping on the B pump were worked on during a major shutdown. Upon start-up the vibration level increased to almost 0.6 IPS, sufficiently high to cause early mechanical seal failure. The spectrum was dominated by vibration at operating speed. One possible cause of the increase in vibration level was that something had fallen into the impeller while the discharge piping was open during the shutdown. However, no damage was found on the impeller or inducer when the pump was removed from service and inspected. Runout, impeller unbalance, and signs of rubs were checked. The pump was carefully re-installed with attention to coupling, Gary M. Einarson Saskferco Products Belle Plaine, Saskatchewan, Canada Summary. A dynamic absorber was installed on a centrifugal pump to solve a resonance problem. Vibration levels were reduced by more than an order of magnitude. alignment, and soft foot. Vibration levels remained high, sometimes to 0.8 IPS. The pump was reserved as a spare for short-term use whenever maintenance was to be performed on the A pump until a solution could be found. When the pump was again operating, it was discovered that loosening poorly installed pipe hangers on the minimum flow line decreased the vibration level between 0.3 IPS and Figure 1. Design of Dynamic Absorber. 109847 Vibration mag (11-03) r4 12/3/03 9:16 AM Page 3 4 Vibrations Vol 19 No 4 December 2003 0.4 IPS. The decrease was tolerable for short-term operation but was higher than it should have been. The vibration levels along the piping were measured to determine the locations of nodes and antinodes. Rigid hangers were installed at appropriate locations to achieve an acceptable decrease in pump vibration. However, the vibration levels again increased to almost 1.0 IPS overall. One day when pump A was running, it was discovered that the vibration levels on pump B, which was not operating, were higher than those on pump A. Resonance could have been confirmed with a bump test or a test with an FRP hammer, but it was concluded that resonance was the cause of the high vibration levels. The mass or stiffness of a system can be changed to correct a resonance problem; the forcing frequency is moved away from the natural frequency. Another more economical solution is to add a dynamic absorber to the pump. A dynamic absorber was designed, built, and installed to reduce the vibration amplitude of the operating speed frequency of 3,562 CPM. The absorber was simple so that it could be attached easily to the pump using two of the bolts that hold the end cover onto the drive end of the pump. The design involved a relatively simple calculation and some minor guesswork. The equations are shown in Figure 1. The frequency of vibration that had to be controlled was calculated using spectral analysis and was the operating speed, or 3,562 CPM. Steel was chosen as the spring material. The cross section of the spring (b and h in Figure 1) could have been obtained by clamping a piece of bar stock in a vise, finding the resonant frequency with a bump test, and adjusting the length (L) and required weight (W2). Instead a cross section (b and h), length (L), and weight location (a) were assumed to calculate the required weight in pounds (W2). The weight was found by calculating and weighing flat bars and hardware that add up to the required weight, W2. The weight on the spring (a) had to be sufficiently far from the end of the spring so that the absorber could be tuned in the field. The final test was to mount the absorber onto the pump and run the pump up to speed. A good rule of thumb is that the end of the absorber should displace a maximum of 0.5 inch peak to peak. A higher displacement could cause the absorber to fatigue. The dynamic absorber worked, and the location (a) of the weight (W2) was fine tuned. A vibration transducer was mounted to the pump at the point of highest amplitude; the location of W2 was adjusted until the amplitude dropped to its lowest point. The top of the absorber displaced almost 0.5 inch peak to peak and appeared to blur as the absorber Figure 2. Spectrum Before Dynamic Absorber. Figure 3. Spectrum After Dynamic Absorber. Figure 4. Pump After Installation of Dynamic Absorber. was adjusted closer to the resonant frequency of the pump. Spectral analysis verified the success of the solution; the operating speed component was reduced from 0.92 IPS (Figure 2) to 0.05 IPS (Figure 3). The pump after the dynamic absorber had been installed is shown in Figure 4. 109847 Vibration mag (11-03) r4 12/3/03 9:16 AM Page 4