Final Design Review
1. Introduction
This design review will mostly be used to review the small scale testing that was conducted at Kansas State University (KSU) with representatives from both SPX and the KSU design team present. We will review the data that was accumulated and compare this to updated calculations to verify our calculations. This report will also include our group’s justification for the results that were gathered during testing.
This small scale data will be used to help design the most effective sound attenuation possible. We will also use this data to help predict what will happen when testing is done on our full scale models. This knowledge will help us determine which full scale configurations are worth looking into.
This report will also finalize and confirm the plans for full scale testing that will be conducted in Kansas City at SPX headquarters. Once this testing is complete, our team will gather all the information we acquired and use said information to present final designs to SPX for consideration.
This small scale data will be used to help design the most effective sound attenuation possible. We will also use this data to help predict what will happen when testing is done on our full scale models. This knowledge will help us determine which full scale configurations are worth looking into.
This report will also finalize and confirm the plans for full scale testing that will be conducted in Kansas City at SPX headquarters. Once this testing is complete, our team will gather all the information we acquired and use said information to present final designs to SPX for consideration.
2. Small Scale Insertion Loss Results
The following data will show the decibel readings and pressure drop data for each of the configurations that were tested. The paragraphs following the tables will attempt to explain or justify the numbers that were found through experimentation.
Small scale testing was done on 8 different configurations that were decided on beforehand. All 8 configurations had various amounts of success in terms of lowering the noise created by the fan. We will now look at each configuration individually and attempt to explain the results that were obtained.
The first configuration we attempted was 5 small rectangular baffles aligned in parallel. This configuration had the smallest amount of exposed surface area and this was shown in the decibel readings. The overall dBA across all frequencies was found to be 69.6 dB. This showed basically a 3 dB drop from measuring the sound with no baffles. This configuration lowered the noise level the least due to the fact that the baffles were lined up parallel to each other and only extended half the length of the box. This allowed many of the sound waves to move through the box uninterrupted and therefore the noise level was not changed as much as other configurations.
Small scale testing was done on 8 different configurations that were decided on beforehand. All 8 configurations had various amounts of success in terms of lowering the noise created by the fan. We will now look at each configuration individually and attempt to explain the results that were obtained.
The first configuration we attempted was 5 small rectangular baffles aligned in parallel. This configuration had the smallest amount of exposed surface area and this was shown in the decibel readings. The overall dBA across all frequencies was found to be 69.6 dB. This showed basically a 3 dB drop from measuring the sound with no baffles. This configuration lowered the noise level the least due to the fact that the baffles were lined up parallel to each other and only extended half the length of the box. This allowed many of the sound waves to move through the box uninterrupted and therefore the noise level was not changed as much as other configurations.
Small scale testing was done on 8 different configurations that were decided on beforehand. All 8 configurations had various amounts of success in terms of lowering the noise created by the fan. We will now look at each configuration individually and attempt to explain the results that were obtained.
The first configuration we attempted was 5 small rectangular baffles aligned in parallel. This configuration had the smallest amount of exposed surface area and this was shown in the decibel readings. The overall dBA across all frequencies was found to be 69.6 dB. This showed basically a 3 dB drop from measuring the sound with no baffles. This configuration lowered the noise level the least due to the fact that the baffles were lined up parallel to each other and only extended half the length of the box. This allowed many of the sound waves to move through the box uninterrupted and therefore the noise level was not changed as much as other configurations.
The second configuration used the same size baffles as the previous set-up but with the baffles set up at an angle instead of parallel with the flow. The thinking behind this was that it would force more of the sound waves to hit the attenuator and therefore lead to a large insertion loss. The results we saw seemed to prove this assumption to be correct. The overall dBA was found to be 68.8 dBA which was almost a full decibel drop from the previous configuration. This showed that there is a possible value to using angled baffles instead of baffles parallel to flow.
The third configuration was very similar to configuration number 1 except each baffle extended the entire length of the box. We expected this to increase insertion loss due to the increased exposed surface area. The results we observed confirmed this thought. The 67.9 dBA was lower than what was seen in configuration 1 and therefore proved the increased surface area did in fact make a difference. More research would have to be done to see whether the increased cost is worth the improved insertion loss.
The fourth configuration yielded some interesting results. Based on the first three results, it would not have surprised us if the angled large baffles were to improve upon the large baffles that were parallel to airflow. However, the results we recorded show that there is no difference between angled and parallel alignment with the larger rectangular baffles. The 67.9 recorded dBA was exactly the same as the previous configuration. A possible explanation for this is that the longer baffle causes the sound waves to bounce off the baffle multiple times and therefore the sound still reaches the other side just like in the parallel configuration. This result shows that angling the baffles only creates an advantage up to a certain length. A longer baffle reduces the need for an angled configuration.
The next two configurations were tested to see if the number of baffles played a big role in the amount of insertion loss seen across the attenuator. Another advantage of the fewer amount of baffles was the ability to use a bigger angle in the angled baffle set up. The first one we tried was the angled configuration with 4 baffles. The 68.9 dBA result was expected and proved that more baffles leads to better insertion loss which is what was expected. Once we tried this configuration, we wanted to test the four baffles when they are parallel to the flow to see if the large angle had any effect on the sound waves.
The sixth configuration featured the 4 large baffles aligned in parallel with the air flow. The results showed that the overall dBA was essentially the same as the angled configuration. This gives us another data point that shows the longer baffles have almost no difference between angled and parallel configuration. The angled set up only seems to make a difference when using baffles with a shorter depth. This was the last test we conducted with rectangular baffles.
The next two configurations were of great interest to our team. We have detailed in previous reports why we believed round tube baffles would be beneficial to use for sound attenuation purposes. The first configuration we tried with round tubes was two rows of 4 tubes in line. We believed that this set up would stop more of the sound from reaching the other side of the attenuator. We were pleased to see that the results we observed backed this up. The overall dBA of 66.5 was the lowest decibel reading we recorded for any configuration. The round tubes thus proved to be a worthy design to consider.
The last configuration we tried was 2 rows of 4 baffles but staggered so that the line of sight was obscured. We thought that this might also increase insertion loss because it would be harder for sound waves to pass through the attenuator. The results we found seemed to prove this theory wrong. The decibel reading actually went up compared to the in line configuration. A possible explanation for this is that the circular design of the tubes caused the sound waves to bounce off at inconsistent angles and thus many of the waves went through the second set of tubes without actually hitting the baffles. This is a possible reason why the decibel reading increased between these two configurations.
The first configuration we attempted was 5 small rectangular baffles aligned in parallel. This configuration had the smallest amount of exposed surface area and this was shown in the decibel readings. The overall dBA across all frequencies was found to be 69.6 dB. This showed basically a 3 dB drop from measuring the sound with no baffles. This configuration lowered the noise level the least due to the fact that the baffles were lined up parallel to each other and only extended half the length of the box. This allowed many of the sound waves to move through the box uninterrupted and therefore the noise level was not changed as much as other configurations.
The second configuration used the same size baffles as the previous set-up but with the baffles set up at an angle instead of parallel with the flow. The thinking behind this was that it would force more of the sound waves to hit the attenuator and therefore lead to a large insertion loss. The results we saw seemed to prove this assumption to be correct. The overall dBA was found to be 68.8 dBA which was almost a full decibel drop from the previous configuration. This showed that there is a possible value to using angled baffles instead of baffles parallel to flow.
The third configuration was very similar to configuration number 1 except each baffle extended the entire length of the box. We expected this to increase insertion loss due to the increased exposed surface area. The results we observed confirmed this thought. The 67.9 dBA was lower than what was seen in configuration 1 and therefore proved the increased surface area did in fact make a difference. More research would have to be done to see whether the increased cost is worth the improved insertion loss.
The fourth configuration yielded some interesting results. Based on the first three results, it would not have surprised us if the angled large baffles were to improve upon the large baffles that were parallel to airflow. However, the results we recorded show that there is no difference between angled and parallel alignment with the larger rectangular baffles. The 67.9 recorded dBA was exactly the same as the previous configuration. A possible explanation for this is that the longer baffle causes the sound waves to bounce off the baffle multiple times and therefore the sound still reaches the other side just like in the parallel configuration. This result shows that angling the baffles only creates an advantage up to a certain length. A longer baffle reduces the need for an angled configuration.
The next two configurations were tested to see if the number of baffles played a big role in the amount of insertion loss seen across the attenuator. Another advantage of the fewer amount of baffles was the ability to use a bigger angle in the angled baffle set up. The first one we tried was the angled configuration with 4 baffles. The 68.9 dBA result was expected and proved that more baffles leads to better insertion loss which is what was expected. Once we tried this configuration, we wanted to test the four baffles when they are parallel to the flow to see if the large angle had any effect on the sound waves.
The sixth configuration featured the 4 large baffles aligned in parallel with the air flow. The results showed that the overall dBA was essentially the same as the angled configuration. This gives us another data point that shows the longer baffles have almost no difference between angled and parallel configuration. The angled set up only seems to make a difference when using baffles with a shorter depth. This was the last test we conducted with rectangular baffles.
The next two configurations were of great interest to our team. We have detailed in previous reports why we believed round tube baffles would be beneficial to use for sound attenuation purposes. The first configuration we tried with round tubes was two rows of 4 tubes in line. We believed that this set up would stop more of the sound from reaching the other side of the attenuator. We were pleased to see that the results we observed backed this up. The overall dBA of 66.5 was the lowest decibel reading we recorded for any configuration. The round tubes thus proved to be a worthy design to consider.
The last configuration we tried was 2 rows of 4 baffles but staggered so that the line of sight was obscured. We thought that this might also increase insertion loss because it would be harder for sound waves to pass through the attenuator. The results we found seemed to prove this theory wrong. The decibel reading actually went up compared to the in line configuration. A possible explanation for this is that the circular design of the tubes caused the sound waves to bounce off at inconsistent angles and thus many of the waves went through the second set of tubes without actually hitting the baffles. This is a possible reason why the decibel reading increased between these two configurations.
3. Pressure Drop Results
Now that we have looked at the decibel readings for the different configurations, we must consider the pressure drop across the attenuation set ups. The following table shows the pressure drop that was observed over the attenuation in small scale testing.
This table shows a very small pressure drop reading which is to be expected on testing on baffles this small. Notable difference that should be noted are that angled baffles cause more pressure drop than non-angled alignments. This is also what was expected intuitively. Now we want to compare these values to the tube baffle pressure drops that we had calculated earlier for accuracy.
After looking at this summary chart, we can see that our model actually overestimates the pressure drop but that they are the same order of magnitude. This allows us to use our pressure drop calculations for larger scale applications, with the knowledge that the model may give us a larger value than what will be seen in experimentation.
4. Cost Analysis
This portion of the report will take a look at a cost comparison between tubed and rectangular baffles. This will not be an exact analysis because the cost of manufacturing the tube baffles (rolling, etc) is something of an unknown. However, we will use known facts and try to get an estimate on cost of the two baffles.
![Picture](/uploads/2/5/3/2/25322222/5826829.png?571)
The graph breaks down how much money was spent on each baffle per ft^2 (surface area) and ft^3 (volume). SPX spent $302.09 per baffle for a rectangular baffle and $302.08 per tubed baffle. The tubed baffles would probably be cheaper if a large amount were ordered. They were more expensive due to the fact that only 12 were needed for full scale testing.
The numbers shown above show that the surface area and volume of the tubed baffles is well below the amount of metal required for the rectangular baffles. With current costs, we are spending about 4x more per feet^3 for the tubed baffles. Knowing this, it might be worth looking into negotiating a better deal. However, we are not sure how much more it would cost to manufacture tubed baffles as compared to the simple rectangular baffle.
5. Summary
These results give us a good starting point to evaluate our attenuation design. Full Scale testing will be done the week of April 6th and that will give us an even more accurate representation. Small scale testing did provide us with useful information that can be applied in our design models. We got to see in practice how effective angled baffles and tube baffles when it comes to insertion loss.
Angled baffles gave us interesting results as the success of these baffles was dependent on the length of the baffle. The shorter baffles had a noticeably better insertion loss when they were angled. However, as the baffles increased in length, the effectiveness of the angled baffles decreased. The longer baffles had the same insertion loss regardless of if they were angled or not. Since the pressure drop was greater for the angled baffles, it is clear that for longer baffles it makes more sense to keep them parallel to the airflow.
Our round tube testing provided us with some unexpected yet useful knowledge. The aligned rows of tubes was actually our most successful configuration with regards to insertion loss. The staggered tube configuration was not as successful as we were hoping with noise reduction. Based on these results, the aligned configuration is actually more effective. This is nice because it also happens to be easier to manufacture than a staggered alignment would be.
The full scale testing should give us a more accurate representation of what kind of results rolled tube baffles will give us. We also have to take into account the cost factor of both set ups. We are hopeful that the tubed baffles will be a more effective attenuation option by both cost-effectiveness and success in lowering decibel levels.
The next step in our design process will be to evaluate full scale testing and to develop full scale designs to submit to SPX. We will be in contact with SPX throughout the final stages of the process to ensure good communication and proper completion of the project.
Angled baffles gave us interesting results as the success of these baffles was dependent on the length of the baffle. The shorter baffles had a noticeably better insertion loss when they were angled. However, as the baffles increased in length, the effectiveness of the angled baffles decreased. The longer baffles had the same insertion loss regardless of if they were angled or not. Since the pressure drop was greater for the angled baffles, it is clear that for longer baffles it makes more sense to keep them parallel to the airflow.
Our round tube testing provided us with some unexpected yet useful knowledge. The aligned rows of tubes was actually our most successful configuration with regards to insertion loss. The staggered tube configuration was not as successful as we were hoping with noise reduction. Based on these results, the aligned configuration is actually more effective. This is nice because it also happens to be easier to manufacture than a staggered alignment would be.
The full scale testing should give us a more accurate representation of what kind of results rolled tube baffles will give us. We also have to take into account the cost factor of both set ups. We are hopeful that the tubed baffles will be a more effective attenuation option by both cost-effectiveness and success in lowering decibel levels.
The next step in our design process will be to evaluate full scale testing and to develop full scale designs to submit to SPX. We will be in contact with SPX throughout the final stages of the process to ensure good communication and proper completion of the project.