All measurements were taken on a Conn-Weld 2.44-m × 4.88-m (8-ft × 16-ft) horizontal vibrating screen with dual G-Master 1000 vibration mechanisms, as shown in Fig. 1. The screen was placed in the center of the OMSHR National Voluntary Laboratory Accreditation Program (NVLAP)-accredited reverberation chamber⁴⁷ and rested on the floor with wooden wedges driven under the frame rails to prevent the screen from rocking during operation. The belt guard was removed to avoid measurement inconsistencies from rattling observed in previous testing³⁴. The screen was run without coal or water feed, since previous studies had shown that coal and water flow noise was not a significant contributor to sound levels in the plant environment³⁰. Sound power tests were performed per ISO 3743-2⁴⁶. Data were simultaneously recorded from 15 microphones arranged in a random distribution⁴⁷. Three 30-second measurements were taken for each test configuration and logarithmically averaged.

Temperature measurements were made for those noise controls installed on the vibrating screen that had an effect on mechanism temperatures. To avoid mechanical failure of the bearings or other components, the screen manufacturer, Conn-Weld, recommends that the mechanisms run at a maximum temperature of 82.2°C (180°F) for standard oil and an absolute maximum temperature of 90.6°C (195°F) for high-temperature oil when measured at the surface of the housings. A variety of cooling methods were attempted to keep the operating temperature within acceptable limits after the noise controls were installed.

Five type J thermocouples were connected to the mechanism housings with thermally conductive cement (Fig. 2). A sixth thermocouple measured ambient air temperature. The thermocouples were connected to a DATAQ DI-715B data acquisition device, which recorded the data to a laptop. Temperatures were recorded until steady state was reached, until it was apparent that a cooling method was unsuccessful, or until the temperature reached 82.2–90.6°C (180–195°F).

Spring inserts were designed to eliminate spring chatter noise by preventing the last full coils and the cut and ground coils on each of the spring ends from coming into contact. An earlier prototype of the inserts, tested in the lab, was a curved slab of 70 durometer natural rubber with tapered grooves to seat the spring coils. The final inserts developed for field testing were composed of 70 durometer natural rubber bonded to a 6.4-mm-thick (1/4-inch-thick) curved steel shell along the outer perimeter. The steel shell was intended to provide a striking surface for hammering the spring inserts into place between the spring coils. The rubber portions of the inserts were tapered to match the contours of the springs for better contact, and had a lip to prevent them from sliding out of place during screen operation, as shown in Fig. 4. Spring inserts were designed for both the inner and outer springs of the vibrating screen, so that all springs on the screen were prevented from chattering.

Constrained layer damping (CLD) is commonly used to treat a vibrating surface that is generating noise. It consists of a layer of flexible damping material applied to the vibrating surface, with the other side of the damping layer constrained by a more rigid material such as steel. Vibration energy from the base layer is transformed into a shear deformation within the damping layer, which reduces the noise radiated by the system while generating a small amount of heat.

For this application, a thin sheet of 80 durometer, 0.64-mm-thick (0.025-inch-thick) elastomeric damping material was bonded on one side to the flat faces on the front, top, and back of each mechanism housing using epoxy. These layers of damping material were then constrained by being bonded to 6.4-mm-thick (1/4-inch-thick) steel plates. Figure 5 shows constrained layer damping treatments on the top and front face of the mechanism housings. To ensure a good bond between the housings, damping material, and constraining plates, the paint from the housings was removed using a grinder and the constraining plates were sandblasted and cleaned with isopropyl alcohol prior to applying the treatments.

A tuned mechanism suspension (TUMS) was developed to selectively transmit forces at the mechanism rotation speed of 900 RPM, while attenuating forces at higher frequencies that resulted from the gears and bearings. The mechanism suspension is “tuned” by adjusting the spring rates of the mechanism suspension to shift the rigid body resonant frequencies of the screen with the added mechanism suspension to either a higher or a lower frequency. In a previous publication⁴⁵, a model of the screen and TUMS was examined within ANSYS to determine the approximate spring rates that would be needed. The goal was to have the rigid body frequencies fall above the operating speed of the screen and below a target value for isolation at 100 Hz⁴⁵.

The TUMS consisted of a mounting plate, or “raft,” for the mechanisms, a modified H-beam, a series of gussets to support the added material to the H-beam, and a series of two-piece vibration isolators between the raft and the H-beam, as shown in Fig. 6. In combination, these elements decouple the mechanism housings and the H-beam. The vibration mechanisms were mounted on top of a large steel plate, or raft, which was in turn mounted to the modified H-beam through the vibration isolators. The goal of adding a mechanism suspension is to separate the primary source of mechanical energy from the noise-radiating structure, but without degrading the performance of the screen.

The H-beam was modified by adding a “cap,” as illustrated in Fig. 7, over top of the H-beam to create a wider platform for mounting the tuned mechanism suspension in order to avoid problems with the pitching modes involving the mechanism housings being close to the operating frequency of the screen⁴⁵. The resulting modified H-beam was 7.3 cm (2–7/8 inch) thick and 64.1 cm (25.25 inch) wide. Eight 1.9-cm-thick (3/4-inch-thick) gussets supported the H-beam cap and attached it to the H-beam. Twenty-four natural rubber two-piece vibration isolators, as shown in Fig. 8, were used to separate the raft from the modified H-beam. The isolators used an external steel structure to prevent the rubber from being cut by the H-beam cap or mounting hardware once the mount was compressed.

An acoustic enclosure was designed that surrounds both mechanism housings and is attached to the H-beam^34,35. The motor and drive belts were not enclosed due to space constraints in coal preparation plants. Additionally, they were not found to be significant contributors to noise³². We created a modular panel-on-frame design with components that weigh less than 222 N (50 lb.) and can be hand-carried and installed on any Conn-Weld 2.44-m × 4.88-m (8-ft × 16-ft) horizontal vibrating screen. The modular nature of this design accommodates different numbers and spacing of vibration mechanism housings and the design can be modified to adapt to different sizes and types of vibrating screens.

As shown in Fig. 9, the enclosure consists of a steel frame, which provides a stiff structure, and panels that enclose the noise source. The frame is made up of a series of sub-frames that can be bolted together to make a larger or smaller enclosure as needed. The panels act as a barrier to contain the noise, while a layer of acoustic foam on the inside of the enclosure prevents the reverberant build-up of noise. The bolt-on panels can be easily removed to reach a fill port, drain plug, or bearing cover, as shown in Fig. 9. An entire frame section with the panels attached can be removed to change a mechanism. Cooling ducts bolt onto the side panels separately and can be reconfigured as necessary, as shown in Fig. 10.

Enclosure construction involved a variety of materials and screen modifications. The frame was composed of steel angle stock and U-channel. It was isolated from the H-beam using strips of 57 durometer natural rubber. Panels were fabricated with 3.2-mm-thick (1/8-inch-thick) Dynalam^™ damped steel (a CLD steel). The insides of the panels were lined with 25.4-mm-thick (1-inch-thick) Polydamp^® acoustic foam. A boss made of 57 durometer natural rubber isolated the right panel from the bearing cover plate.

Because the enclosure consists of damped steel panels and is lined with acoustic foam, it makes a good thermal insulator. As a consequence, temperatures at the mechanism housings increase with the enclosure installed. A variety of methods to cool the enclosure were explored, including cooling the entire enclosure with compressed air, adding an 80.2-L/s (170-CFM) push–pull fan system at the inlet and exhaust ducts of the enclosure, exhausting the enclosure air with a 330.3-L/s (700-CFM) fan attached to a modified enclosure duct, and running this same fan at 75% airflow.

Early in the project, we observed spring chatter both in the lab and in the field. Later, through a beamforming study, it was shown that the suspension springs were one of the three main noise sources of the screen³¹. It was also shown that the spring chatter resulted from impacts between the cut and ground coils on the spring ends and the coils that they contact⁴⁸. When separating the contact between these coils with wooden wedges or pieces of closed-cell foam, the spring chatter was eliminated and noise was reduced in the 1/3-octave-bands from 3150 Hz and above⁴⁹. Because spring chatter is intermittent and variable, the maximum potential reduction from the spring inserts is not known. However, spring inserts have been used consistently throughout all testing to eliminate a known noise source, which also eliminates a source of variability when testing the other noise controls.

After eliminating the spring chatter with the spring inserts, the next controls installed were the constrained layer damping treatments. As shown in Fig. 11, the greatest reductions were achieved above 1 kHz, which is the frequency range that the spring inserts and CLD plates were designed to address, and where the mechanism housings are the most significant noise source. Overall, the combination of both noise controls reduced the sound power level by 2.7 dB.

The TUMS system was installed on the screen next, in addition to the spring inserts and CLD plates. With the original 67 durometer vibration isolators, the bolt heads impacted against the modified H-beam, causing the noise to increase during operation. The sound power level increase was primarily due to a lack of clearance between the mechanism bolt heads and the modified H-beam.

In addition, the first set of vibration isolators was not stiff enough to hold all of the components together during operation. Examining the operation of the screen with a stroboscope revealed that the tops of the isolators were losing contact with the top of the modified H-beam. When the two-piece mounts are installed, the mounts are pre-compressed so that both the top and bottom of the mounts undergo static deflection. Because the top half of the isolator is slightly stiffer than the bottom half of the isolator, the static deflection of the top half of the isolator is slightly less than the static deflection of the bottom half of the isolator. The relative motion between the raft and the modified H-beam exceeded the static deflection of the top half of the isolator. This caused the top half of the isolator to become unloaded and allowed separation of the isolator from the top of the H-beam.

Several changes were made to address these problems. The height of the isolator tops was increased to 3.2 cm (1.25 inch) to reduce the spring rate of the top half of the isolator relative to the bottom half. This ensured that the top half would have enough static deflection to prevent the components from separating under operation. The durometer of the mounts was also increased to increase the mount stiffness. Finally, various spacers were used to increase the preload on the vibration isolators and the clearance of the bolt heads from the modified H-beam, as shown in Fig. 12. These changes ultimately resulted in the “tuning” of the system that was desired.

Sound power level reductions were achieved with these design changes to increase preload, clearance, and mount stiffness. Table 1 is a summary of the successful iterations of TUMS development. The greatest reductions were accomplished by both the 80 durometer mounts with a 6.4-mm (1/4-inch) spacer on top and none on the bottom and the 90 durometer mounts with a 3.2-mm (1/8-inch) spacer on top and none on the bottom—each of which yielded a 1.4-dB improvement versus baseline.

Performance for individual 1/3-octave bands varied with durometer and preload, as shown in Fig. 13. The 80 durometer mounts had better performance than the 90 durometer version in the 125–315 Hz frequency range. All mounts increased the sound power level compared to the baseline in the 2–10 kHz range, but the combination of 80 durometer mounts with the 6.4-mm (1/4-inch) top spacers and the 3.2-mm (1/8-inch) bottom spacers showed the highest increase in this range.

Another concern in mount selection was heating of the mounts under operation, which caused the sound power level to increase over time. Previous measurements of failed mount configurations showed a trend of increasing sound power as the testing progressed. Thermocouples were installed on one of the earlier mount prototypes to observe how they were heating up. The results showed that the mount temperature increased as the vibrating screen continued to run. When the mount temperature reached 93.3°C (200°F), the test was aborted per the isolator manufacturer’s recommendation that the mount temperature should not exceed 93.3°C (200°F). Due to internal damping, the isolators heated up during operation as the rubber stretched and compressed. As the isolators increased in temperature, their stiffness was reduced. The design of the TUMS is such that the vibration mode characterized by relative motion between the raft and the modified H-beam occurs at a frequency much higher than the 15-Hz operating frequency of the screen. As the mount temperatures increase and the mount spring rates decrease, the frequency of this vibration mode shifts closer to the operating speed of the screen. This results in an increase in motion of the mechanisms, which in turn causes the impact forces within the gears and bearings to increase. The temperature increase is a probable cause for the increase in noise from the early prototypes.

While the 80 durometer mounts with spacers on top and bottom showed the greatest stability in terms of sound power levels over time, their overall poor performance in sound power level reductions eliminated them as a candidate. Removing the bottom spacers from the 80 durometer mounts caused the mounts to increase in temperature and exhibit the same trend of increasing sound power levels over time. However, the 90 durometer mounts showed a trend of decreasing sound power levels with time, suggesting that the ideal mount stiffness is between 80 and 90 durometer if the stiffness could remain constant with temperature. For durability concerns, the 90 durometer mount configuration with minimum preload was selected. This configuration yields a sound power level reduction from the CLD and spring insert combination of 1.4 dB (Table 1).

Finally, an acoustic enclosure was added to the noise control package. Originally, the enclosure had been installed directly onto the H-beam in previous testing^34,35, but due to installation of the TUMS the enclosure frame had to be modified so it could be welded directly onto the TUMS “raft.” Figure 14 shows that the enclosure reduces overall A-weighted sound power levels by 1.5 dB from the configuration with CLD, spring inserts, and the TUMS. The reductions are seen from the 315 Hz 1/3-octave-band and higher, with some degradation in the 1/3-octave-bands below 315 Hz. This decrease in performance at these frequencies was not observed when the enclosure was originally tested and compared to baseline^34,35, and Fig. 14 shows that the enclosure is an improvement over baseline even in this lower frequency range. The extra surface area due to the addition of the enclosure could radiate more noise than the TUMS configuration alone, and may be an attribute to investigate further.

As shown in Fig. 14, the entire noise control package reduced the overall A-weighted sound power level by 5.6 dB, which is nearly a 75% reduction in sound power. All 1/3-octave-band levels are reduced with the exception of a small increase in the 160 Hz 1/3-octave band. The greatest reductions are seen at the 250 Hz 1/3-octave band and above.

As a consequence of adding the acoustic enclosure, temperature data were taken as discussed previously to determine if the enclosure would cause premature damage to the bearings due to excessive heat. For all tests the thermocouple located at position 4 (see Fig. 2) recorded the highest temperature. Since maximum temperature is the main concern, all plots show the temperature data recorded from position 4. Initial measurements indicated that the system reached the critical temperature of 90.6°C (195°F) in less than five hours with the enclosure installed. This result compares unfavorably to the baseline, where a maximum temperature of 79.4°C (175°F) was reached after 12 hours.

Next, several cooling methods for the enclosure were evaluated. Because the free convection provided by the enclosure ducts was not sufficient for cooling, the enclosure was cooled using shop air, because that may be more convenient for the end user. An air line was run to the enclosure duct and an adapter was made for the duct-to-air line connection. However, the cooling performance was only slightly better than free convection, so this option was not pursued further. To create a push–pull system of airflow, 80.2-L/s (170-CFM) fans were then installed at the inlet and exhaust ducts. This showed moderate success (10 degrees cooler than the enclosure-only configuration after 4 hours), but because it was clear that the cooling was not sufficient, the test was aborted.

The final cooling configuration utilized a 330.3-L/s (700-CFM) exhaust fan. Additionally, the enclosure duct was modified to connect the top of the duct to a remotely located exhaust fan. As shown in Fig. 15, after 10.5 hours, this configuration reached 84.4°C (184°F) running at full flow with an initial ambient temperature of 23.3°C (74°F), which is sufficient cooling for a production day in a prep plant. Running at 75% of flow, the critical temperature of 90.6°C (195°F) is reached in 10 hours and 50 minutes with an initial ambient temperature of 26.7°C (80°F).

The 330.3-L/s (700-CFM) exhaust fan for the enclosure provides adequate cooling. However, further efficiencies could be gained by optimizing the cooling system to improve interior air flow. This may result in reduced airflow requirements for the fan. For example, the current duct configuration could be modified to reduce turbulence from bends and sharp corners. This has the potential to eliminate airflow losses in the duct. Further, during the testing described above for the 80.2-L/s (170-CFM) fans, the coolest mechanism housing temperature location at point 3 (Fig. 2) was predicted to reach the critical temperature after 9 hours, while the hottest housing location (point 4) was predicted to reach the critical temperature after 5.5 hours. It may be possible to reduce airflow requirements of the fan by better mixing of the interior air in the enclosure, thus improving cooling efficiency.

For each test, the average A-weighted 1/3-octave-band sound levels were calculated. Figure 20 shows a comparison with and without the barrier material with all of the screens running except for screen #169. The results show that the barrier material reduced the background noise by 9 dB. With the reduction in background noise, the net result is that the noise floor is 7 dB below the level of the screen and the background noise combined, yielding a background noise correction factor of 0.8 dB.

When the noise control package was installed, spring chatter noise had not been observed for the baseline. However, the spring inserts still needed to be evaluated for durability and ease of installation. Although the final prototypes of the spring inserts were originally designed to have an outer steel shell to aid in installation, the steel shell was found to be both problematic and unnecessary. First of all, prior to the field testing, it was determined that the prototype inserts for the inner spring did not fit within the outer spring, so all of the inner spring inserts were modified by removing the outer steel shell (Fig. 4). Later, during installation at the preparation plant, we found that hammering on the steel shell to install the spring inserts was not as effective as prying the end coils apart and placing the inserts in the proper position. Figure 21 shows the spring inserts in place with the springs installed on the vibrating screen.

To determine the effect of the noise controls on overall sound levels, it is necessary to compare the data from all of the screens running with the barrier material in place versus the same conditions with the noise controls installed. This makes it possible to determine the reduction in sound level due to the noise controls. A comparison of these conditions in Fig. 22 shows that the overall sound level was reduced by 1 dB, with reductions across most of the frequency range.

A prior study conducted in the lab showed that the CLD plates alone reduced the overall sound power levels by 1 dB³⁴. We were expecting at least a 1-dB reduction with some additional improvement from the spring inserts. However, given that no spring chatter was observed on screen #169 before the installation of the noise controls, the spring inserts are not likely to have affected the noise generated on this particular screen. Therefore, the results achieved in the plant are in line with expectations when the test conditions and past laboratory results are considered.