UNIVERSITY OF IDAHO GAUSS ENGINEERING Comparing Flow and Pressure Drop in Mufflers A Statistical Analysis Jeremy Cuddihy, Chris Ohlinger, Steven Slippy, and Brian Lockner 10/24/2012
Table Of Contents Topic Page(s) Introduction 1 Methodology...2-3 Costs...3-4 Reporting..4 Preliminary Results...5-6 References 7
Comparing Flow and Pressure Drop in Mufflers By Jeremy Cuddihy, Chris Ohlinger, Steven Slippy and Brian Lockner I. INTRODUCTION In last year s competition, the Formula Hybrid team had issues with meeting the proper decibel levels in their exhaust system. In order to meet the proper decibel requirements, the team trimmed the overall exhaust length, thus affecting the engine s tune. In altering the engine s tune, large amounts of heat were induced in the exhaust system, and several mufflers were destroyed in the process. This year, a low decibel, high output muffler was obtained from FMF, and research on this muffler is desired to determine whether the muffler will significantly restrict horsepower and torque. By conducting an experiment on the mass flow rates and pressure drops through the new muffler, we can determine how much horsepower and torque will be restricted compared to the muffler used in last year s car. The Formula Hybrid Car is an ongoing senior design project. Research and experimental results lead to improvements every year. By conducting this research, we can determine what, if any, exhaust improvements need to be made. If we determine that the exhaust doesn t drastically restrict flow, we can present these results to the team, and engine tuning can begin. There are several advantages to performing this research other than the obvious tuning and performance aspects. Last year s exhaust tuning was basically trial and error, which led to inaccuracies and a shot-gun effect. Since we know that the new muffler being tested is rated at low decibels, the time in modifying the exhaust to meet decibel requirements can be reduced significantly. We can also save quite a bit of money in not destroying mufflers due to excessive exhaust heat which burns up the insulation used for sound dampening. If we determine that the exhaust won t restrict flow, proper tuning on the engine can begin immediately as opposed to experimenting with mufflers for several months in order to reduce sound. In conducting this experiment, we will be working with multiple dependent and independent variables. The independent variables will include the mufflers (last year s model and the new muffler) and the vacuum speed levels. We will be testing three vacuum speeds based on the number of vacuums turned on during each test (2, 4, and 6 motors will be turned on at a time). In performing an ANOVA test, we will break the vacuum speeds down into high, medium, and low categories to represent the corresponding number of vacuums. The dependent variables will include the pressure drop across each muffler and the mass flow rate through each muffler. The accuracy of our experiment will be limited by the accuracy of the digital sensors being used. The digital pressure sensor being used in the experiment reads two decimal places, so the sensor is accurate to ±.01 psi. The mass flow rate will be measured using a digital multi-meter, so our accuracy is restricted to ±.01 cfm.
II. METHODOLOGY Figure 1. Diagram of flow bench being used for testing. We will perform all experimentation on the flow bench shown in the diagram above. The flow bench consists of 6 electric vacuum motors, a mass flow sensor, a pressure sensor, and a system of tubing. Each muffler will be placed on the air inlet, and the pressure drop as well as the mass flow rate will be measured using the two sensors. The number of electric motors used at one time can be controlled by a bank of switches on top of the flow bench. A main switch provides power to six individual switches (1 controlling each motor), and one motor is controlled by a variable switch (allowing one motor s speed to be changed variably). When a motor is turned off, air is restricted from flowing though the motor by a rubber flap outside of the tubing connected to the motor. The electric motors are held inside of a sealed box that contains wiring and the exit port of each motor on the outlet side and a vacuum chamber on the inlet side. Each vacuum motor spans the distance between the inlet and outlet sides which allows air to be pulled through the system. The sealed container is joined to a network of pipes that eventually lead to the air inlet where the muffler, or other object being tested, is placed. The tubing used is 3-inch pvc, and is joint using pvc connections with silicon sealant at each joint. The smooth tubing restricts friction and pressure losses throughout the system. In between the tubing and the sealed container, a mass flow sensor is used.
Figure 2. Mass flow sensor being used (from Toyota Supra). The mass flow sensor is from a Toyota Supra and must be powered by an external power source. The mass flow sensor has three main wires that include an input, ground, and output. We will provide power to the sensor using a 12 volt power battery and will read the output voltage using a voltmeter. The output voltage and corresponding mass flow rate will be compared to the manufacturer s specifications, and we will use our results to determine whether or not the sensor needs calibration. We will test each muffler three times by comparing the flow rate and pressure drop at three motor settings. Instead of using the variable motor, we will use full power on two, four, and six motors respectively. In order to make sure we are achieving consistent results, we will need to carefully monitor the sealing of each muffler against the air inlet flange on top of the flow bench. Our initial experimentation revealed inconsistent results when mufflers lost their seal. We will need to obtain muffler tape to seal the old muffler because high heat melted a few holes in it. We will also need to obtain muffler tape and carefully monitor the old muffler because the high heat burned several holes in the muffler. Failure to seal these holes would produce poor, skewed results. Since we will be testing two mufflers with similar sizes and characteristics, we expect similar results (a low decibel, high performance muffler is desired in this application). We hope to compare the two mufflers and find that there is very little if any difference between them. A low flow rate and high pressure drop on the new muffler would reveal poor flow characteristics in muffler and is what we hope to avoid! III. COSTS Since the flow bench is already built and mostly functioning, there will be very few costs associated with this experiment. We are currently going through the wiring and fuses to determine what components will need to be updated or fixed. Our initial experimentation showed a problem with several of the electric motors ( a few of the motors didn t turn on). We found that several wires
were loose, switches broken, and we also found issues with some of the flaps covering the motors. Further experimentation will hopefully reveal the cause of our problem. If we find that several of the fuses need to be replaced, the cost of each 20 amp bus fuse is $12.50. Fuses will be relatively easy to obtain, and we found this price from McMaster Carr. If we find that several of the fuses need to be replaced, a deeper analysis of the system will need to be conducted in order to determine the cause of fuse failure. Some of the wiring may need to be replaced or upgraded, and we will speak to someone that knows more about electrical wiring if this turns out to be the case. We found that 300 volt wiring ranges from 9-20$/ft. If we need funding for wiring, this won t be an issue because NIATT can cover much if not all of the funding. In our initial experimentation, we noticed an air leak around the muffler. We can prevent this leak by using caulk or glue on the air inlet at the top of the flow bench. We may need to use bike tire tubing or some other means of sealing the air inlet, but the flow bench has several feet of tubing already with it. Our overall costs in sealing the system will be around 3$ for a tube of caulk. Lastly, we will need to replace the cover on the flow bench. In tearing everything apart, we broke the particle-board cover. Since two team members are carpenters, this will be an easy fix and will cost around 10$ (for the board and a tube of caulk). It is important that the vacuum side of the flow bench is adequately sealed; otherwise, we will obtain undesired results. IV. REPORTING We will work together on team as far as drafting and editing the reports. We are currently subdividing the written portions of the report based on knowledge of the portion of the experiment being conducted. Steven Slippy will be in charge of the final editing of the reports and putting everything together in a presentable way. Steven will also put together presentations and will determine what needs to be presented in the final class presentation. Jeremy Cuddihy will be troubleshooting the flow bench and making sure that the vacuum motors are operating correctly. Jeremy will be involved in the testing of the mufflers and making sure that leaks in the system are avoided in order to obtain proper results with each muffler. Jeremy will also assist Chris in putting together the log book as well as monitoring the results for consistency. Since the flow bench is owned and used by NIATT, Jeremy will send his results to Dan Cordon as well as keeping track of the results for Senior Lab. Chris Ohlinger will be performing the tests on each muffler and analyzing the results. We have manufacturer s specs that will need to be placed in MATLAB, and Chris will take care of this. We will be placing our final results in MATLAB, and Chris will be in charge of this as well as creating plots, graphs, etc. Brian Lockner will be assisting in experimentation as well as printing off final, color reports. It s important that our flow diagrams and wiring schematics be printed off in color because the wiring schemes and overall system is relatively complex. Brian will also be in charge of creating a presentation for the class.
Mean and Confidence Interval (psi) V. PRELIMINARY RESULTS We conducted an experiment on two random mufflers to determine whether or not the flow bench and our method of experimentation would work. Since we wanted to test two mufflers that would have different results, we picked two mufflers with different diameters and performance levels (the diameter of one muffler was 1.5 inches, and the other was 1.75 inches). After selecting two mufflers that we thought would have different results, we began collecting data. We collected 15 data points showing the pressure drop across each muffler at a set speed (only one motor was working on the flow bench at this time and we set the motor at maximum power). It took two team members to collect data and make sure that the mufflers were sealing correctly. With one team member watching the pressure sensor, we were able to verify that the mufflers were getting an adequate seal against the air inlet. Between each data point, we allowed ample time for the system to de-pressurize so as to avoid inaccurate results (we knew that the mufflers should hit a peak pressure drop and level off, but we did this as a precaution). Since we wanted to compare the average pressure drops of the two mufflers, we decided to use confidence intervals to analyze our results. Confidence intervals were an excellent way to show and compare our results side-byside. 0-0.05 CI On Difference of Means 0 2 4 6 8 10 12 14 16-0.1-0.15-0.2-0.25 Mean Upper Limit Lower Limit -0.3-0.35 Samples Figure 3. Difference of means between two mufflers tested. An analysis of the difference of means of the two data sets showed a clear difference between the mufflers. The smaller-diameter muffler showed higher pressure drops across the entire data set, which we expected. Based on the results of this experiment, we were able to determine that the two mufflers measured were clearly different (95% confidence interval), and we have no reason to believe that our results were skewed or incorrect.
Although we received expected results in experimenting with the flow bench, we found some mechanical flaws in the design of the flow bench. We had initially planned on testing the pressure drop across each muffler using all six motors to make sure that everything was working correctly, but we found that only two of the motors were working correctly. Upon tearing the flow bench apart, we found that several of the leak-preventing flaps outside of each motor were falling apart and not working correctly. We also found that two of the switches controlling motors were broken, which prevented 4 motors from working. We are working on correcting these issues for future experimentation.
VI. REFERENCES 1.) Compressible Flow Bench. (n.d.). Yeditepe University Department of Mechanical Engineering. Retrieved from http://me.yeditepe.edu.tr/courses/me401/me401_compressibleflow.pdf 2.) Society of Automotive Engineers. (2001). Three Dimensional Numerical Study on the Pulsating Flow Inside Mufflers with Complicated Flow. Publication number 03-05-2001. 3.) Erkman, Ned. (n.d.). Mathematical Formulas for Superior Performance. Retrieved from http://www.rollingthunderz.com/rfs_formulas_intake.shtml Note: We didn t need to include in-text citations for this portion of the write-up. When we finish developing our mathematical correlations, several of the above references will be included (in our final lab report).