The Rumford, Maine area is, for all practical purposes, the home of paper making in the United States.
The New Page Corporation mill at Rumford has been in operation since 1901 and continues today as a significant contributor to US paper production. The mill currently has one kraft and one ground-wood pulp mill, four coated paper machines plus supporting equipment and one pulp dryer. Each year, the facility produces an average of 640,000 tons of paper and approximately 500,000 tons per year of hardwood, softwood and ground-wood pulp.
Full Spectrum Diagnostics was contacted by Mr. Mark Brown, Reliability Engineer, to investigate recent elevated noise levels near the Super Calendar 22 (SC- 22) stack assembly (Figure 1).
The noise levels were noted to be approaching the OSHA specified limits for this area. Normal Accelerometer route measurements have failed to identify the frequency source. In addition, the hard-wall surfaces behind SC-22 create significant reverberation, adding to the elusive nature of the problem. The only in-plant sound measurements were collected with an overall Sound Pressure Level (SPL) microphone. Also note that Sound Power Levels tend to double when reflected off a close hard surface, making amplitude and source detection even more difficult.
Machine Designation: SC-22 Calendar System (see schematic in Figure 2)
Queen Roll (1)
Diameter: 28.188 inch
RPM @ 2800 FPM: 379 RPM
Steel Roll (3)
Diameter: 16 - 20.50 inches
RPM @ 2800 FPM: 534 RPM
Cotton Denim Roll (4)
MRF Blue Denim
Hardness: Shore "D" 88-90
RPM @ 2800 FPM: 509 RPM
King Roll (1)
Diameter: 32.023 inches
RPM @ 2800 FPM: 334 RPM
Bearings: TOR A2670 / TOR A2665
Bottom Nip: 1800 PLI
Top Nip: 700 PLI (new rolls) /
1000 PLI (old rolls)
New Page Corporation Rumford Paper Mill contacted Full Spectrum Diagnostics to investigate recent elevated noise levels in their Super Calendar (SC-22) stack. This excessive noise level was close to exceeding OSHA standards requiring double ear protection. To reduce noise levels to acceptable levels the SC-22 line speeds were reduced, limiting production.
An initial visit to the site involved natural frequency testing and Operating Deflection Shape analysis of SC-22. This test series focused on the barring phenomena in the Cotton (Denim) rolls that required frequent replacement. The barring fault was thought to be an underlying contributor to the audible acoustic tone. Some structural problems were found in the calendar rolls that were likely contributors to the noise problem, but no single silver bullet solution was found.
A second site visit was planned as an acoustic investigation. The test plan involved the use of sets of acoustic microphones. A single microphone is sufficient for measuring Sound Pressure Levels (SPL) and determining sound frequency response. However, a single microphone will pick up sound energy from all directions. Locating a source in an "open" area (free-field) is straight-forward, but detecting a directional source in a near or reverberant field is problematic. The addition of a second microphone allows the measurement of Acoustic Intensity. The Intensity comes from the cross-power spectrum measurement between the two closely spaced microphones. This method allows the detection of directional sound pressure vs. oblique sound pressure waves. The intensity measurement is ideal for reverberant sound fields such as the SC-22 calendar system.
Structural measurements with accelerometers are usually limited to rotating components. Measurements with microphones can be collected anywhere around the system or source.
It was suspected that the audible tone was emanating from one of the roll bearings or from the roll nip interfaces. The SC-22 calendar system included eight (8) nip interfaces. Both SPL and Intensity microphone measurements were collected along each nip line in seven locations from bearing to bearing. Forty- nine measurements were collected on each of the front and back sides of the calendar.
A 1/12 Octave SPL spectral (overlay plot) is presented in Figure 3. The dominant tone was defined at 550 Hz. The secondary peak is present at 625 Hz. Additional clusters of peaks at integer multiples (2x and 3x) of the 550 Hz tone were also present. The timebased responses showed a "Beat Frequency" pulsation between the 550 Hz and 625 Hz tones. Looseness and/or alignment problems were possibly indicated by the 550 Hz harmonics.
A 1/12 Octave frequency plot in Figure 4 shows an overlay of the Acoustic Intensity spectra calculated for the pair of microphones. The signatures are similar to the SPL plots with dominant 550 Hz and 625 Hz peaks.
The measurement data collected near the nip lines throughout the calendar stack were helpful in defining the dominant tones emanating from the system. The initial analysis focusing on mechanical faults was obviously not the main source of the noise problem(s).
In themselves, the plots of Figures 3 and 4 only define the spectral content of the sound response. When this same data is presented visually using animation software, a color contour plot of the microphone energy provides an indication of the problem source. ME'scope software was used to construct an acoustic surface around the calendar structures. The SC-22 computer model is provided in Figures 5 - 7.
The Sound Pressure Level (SPL) response from a single microphone includes the direct and reflected sound fields. Figure 8 shows acoustic surface profiles measured approximately 2 feet from the nip lines. The dominant amplitude response was located on the back side of the calendar and noted by the yellow-white colors in the center of the roll stack. The front side profile in Figure 9 indicated lower amplitudes with no single dominant apparent source.
The acoustic intensity response on the back side of the calendar shown in Figure 10 indicates that the S2 Steel Roll appears to be at the center of the sound problems. The intensity plots at 550 Hz are presented in Figures 10 and 11 for the back side and front side of the calendar, respectively.
During the analysis, some accelerometer responses were collected on the King Roll bearings. The 550 response was detected at low levels; however the peak did not match known bearing defect frequencies for this roll. The SPL and Intensity plots indicated that a bearing fault was not a likely source of the audible tones.
When calculating the Acoustic Intensity, the cross-phase response from the microphone pair is considered. The measurement indicates the average rate of flow of sound energy through a unit area normal to the propagation direction. Reflection energy is attenuated, thus a more distinct and local source profile is represented.
At this point, the source of the tone appeared to be the nip lines adjacent to the S2 Steel Roll and mating D2 and D3 Denim Rolls. A mechanism for creating this tone was yet to be determined. To gain further insight, a series of Synchronous Time Averaging (STA) measurements were collected based on triggered microphone measurements from the King Roll, the D3 Roll and the S2 Roll. The measurements will slowly "average-out" nonsynchronously related peaks in the frequency spectrum. If the 550 Hz tone was related to the rotational speed of the trigger; as would likely be the case for looseness, misalignment, or barring type faults, the peaks and their harmonics should be retained. The STA measurements for 5, 40, and 100 averages showed that this was not the case, and the dominant peaks were attenuated (removed) for each set of measurements.
A final set of Natural Frequency measurements were collected on the spare Steel Rolls that were slated as replacements for this calendar. King Roll, the D3 Roll and the S2 Roll. The measurements will slowly "average-out" nonsynchronously related peaks in the frequency spectrum. If the 550 Hz tone was related to the rotational speed of the trigger; as would likely be the case for looseness, misalignment, or barring type faults, the peaks and their harmonics should be retained. The STA measure- The rolls were mounted on their bearings and rested on wood timbers near the calendar.
Impact Natural Frequency testing indicated a 485 Hz response on each roll which was defined as the roll first bending mode. This frequency is approximately 12% below the 550 Hz tone, however when considering the stiffness induced by the nip line pressure during operation, it was possible that this mode could be pushed very close to the 550 Hz response.
The concentration of the tone energy near the center of the roll span suggested that the acoustic measurements may be detecting the first bending mode of the roll. It was suspected that the deflected response during calendar operation would likely produce a repetitive opening and closing of the nip on alternate sides of the roll. This "gapping" was suspected to be inducing the tone and tone harmonics noted in the sound field data. The secondary peak at 625 Hz may be the out-of plane bending mode of the roll due to nonsymmetric stiffness (nip pressure)..
Nice theory, but additional testing was required for proof. It was recommended that additional acoustic measurements under varying nip pressures be performed. Slight variations in the nip line pressure may produce enough of a stiffening (or softening) effect to alter the steel roll's critical speed (first bending mode). The condition of the current rolls did not permit additional testing during this site visit; however this testing was performed in the following days.
A series of High Speed Trials were performed on Super Calendar 22. Included in these trials were alterations to the nip pressure in the roll stack. It was found that changes to the pressure induced a significant audible tone change in the machine. The trials also involved modifications to the new roll "break-in" procedure and to the time periods that the rolls were idle and left in nip under pressure between calendar operations. Addressing each of these variables appeared to be beneficial to both the calendar noise levels and the onset of barring issues in the denim rolls.
Current operation of Super Calendar 22 includes slower line speeds that will meet current demands. However when additional volume is needed, a path to smoother and quieter high speed operation is now available.
Dan Ambre, P.E. is a Mechanical Engineer and founder of Full Spectrum Diagnostics, PLLC, a Full Service Predictive Maintenance Consulting company. Dan specializes in Resonance detection, Experimental Modal Analysis, and Operating Deflection Shape machinery diagnostics. Full Spectrum Diagnostics provides Vibration Analysis level I, II, and III training and certification, as well as training in advanced diagnostic techniques. Dan is a certified software representative for Vibrant Technology, Inc., the creators of ME'scope VES software tools. He also provides ME'scope VES Software Training targeting the needs of the In-Plant Vibration Analyst. Please visit his web site at www.fullspec.net
Mark Brown from New Page Corporation, Rumford Operations, is a Reliability professional with over 13 years experience in vibration analysis. He has developed predictive maintenance programs in the Pulp and Paper industry and Nuclear Power industry. Mark holds a B.S. in Industrial Technology from the University of Southern Maine. In addition to holding an ASNT Level III Certification in Vibration Analysis, Mark is also a Certified Oil and Lubrication Specialist.