AC : EXPERIMENTAL STUDY OF WASTE EGRESS FROM COLLECTION VEHICLE

Transcription

1 AC : EXPERIMENTAL STUDY OF WASTE EGRESS FROM COLLECTION VEHICLE Richard Cuprak, Arizona State University, Polytechnic campus John Rajadas, Arizona State University Polytechnic Scott Danielson, Arizona State University American Society for Engineering Education, 2008 Page

2 Experimental Study of Waste Egress from Collection Vehicle Engineering Technology programs focus on delivering a hands-on based engineering education. The students get introduced to the theoretical development of engineering concepts first. Then they apply the concepts to solve practical problems and test the concepts in carefully designed experiments carried out in appropriate facilities. One of the key areas of focus in the Mechanical Engineering Technology program at Arizona State University Polytechnic (ASU Poly) is Thermofluids where therodynamic and fluid dynamic concepts are addressed. The Graduate Degree (M.S.) program in the Mechanical and Manufacturing Engineering Technology (MMET) department at ASU Poly has a variety of activities ongoing in this important area. The graduate student typically works on applied research projects designed with educational and research objectives. Most of these projects involve theoretical and experimental elements. The present paper describes one such project underway in the MMET Department. The project addresses engineering design issues associated with a dry waste collection truck to reduce the potential for load egress during transit. One of the major problems associated with waste collection process, especially the light weight material collected for recycling, is that in the low speed transit segment of the operation, in which the vehicle moves around residential and business neighborhoods collecting the material with the collection bin uncovered, the aerodynamic forces cause the material to become airborne and leave the bin littering the streets. This has a negative impact on several factors associated with the operation not the least of which is public discontent. The project reported here undertakes to address this problem in an experimental investigation using a low speed wind tunnel and appropriately scaled model(s). Flow variables such as velocity and pressure are measured and the dynamics of the problem are analyzed in a systematic manner. Data is generated by employing a two-level factorial experimentation approach. A key requirement for this process to be successful is the availability of a wind tunnel facility that is capable of addressing the engineering tasks outlined for the project. An existing low speed wind tunnel facility at ASU Poly was modified for the purpose of conducting the experimental investigation required. The tunnel modifications included major changes to the inlet section to ensure that the flow entering the test section was well conditioned, a pressure survey setup involving several pressure transducers along with the attendant measurement systems such as a data logger coupled to a desktop computer and a myriad of smaller but essential changes in and around the test section of the tunnel. The modifications were designed, analyzed, fabricated and tested by the graduate student (First Author of the present paper) working on the project. The modification was based on both the short term goal of getting the facility ready for the project at hand and the long term goal of arriving at a configuration, which will enhance the measurements and testing aspects of the engineering technology curriculum being offered in the department. This redesign process will have an immediate impact on the course AET420: Applied Aerodynamics & Wind Tunnel Testing being offered in spring semesters in the department. A wind tunnel design and fabrication process requires the coupling of a few key disciplines in engineering technology. They are aeronautical, mechanical and manufacturing engineering technologies, which form the core focus areas of MMET dept. Some of the changes to the wind tunnel included design of a new inlet convergence, fabrication of this convergence, and subsequent installation. The redesign process was modular which allows additional changes to Page

3 be made with minimum effort in the future. On the instrumentation front, pressure transducers and a data-logger were fully integrated with the wind tunnel measurement system. Variable inputs were geometric design parameters related to the truck cab and the collection bin sections, wind speed, and yaw angle (cab orientation). Output was dynamic pressure from fifty-four test points in the truck. Statistical analysis of normalized data using Minitab included analysis of variance, linear regression analysis to establish significant input variables, and contour plots of the pressure fields generated in Excel. The main effects account for half of the measured response, there are interactions between some main effects, and other probable variables exist. Brief descriptions of the changes to the wind tunnel, design and fabrication of the truck models, factorial design, and the experimental process are given below. As mentioned above, the experimental study described in the present paper supported a M.S. Thesis work in the department. The investigation was comprehensive, rigorous and led to useful results and information that will be of immediate use to the industry at large. Additionally the tasks associated with the study such as the use of rapid prototyping equipment for making the test models will serve as the template for future such endeavors. Even though the redesigned tunnel was used to address the specific problem faced by the industry right now, the students gain valuable experience in solving practical problems of interest to present day industry as they work on a variety of applied projects using the tunnel. Introduction Dry waste collection vehicles are faced with certain problems specific to the curbside collection process on residential streets. Specifically, this process involves the dumping of material from a container into the open section of the vehicle and moving to the next collection point. During the transit time, the truck cavity is open and the potential exists for the material to be forced to egress from the vehicle due to aerodynamic effects. The project described here was accepted based on industry request to study this issue and attempt to determine what critical elements are involved in exacerbating this condition. To accomplish this, an experimental investigation was deemed to be the most logical approach. This required a wind tunnel of sufficient size and speed range as well as appropriately constructed models of the truck design being studied. Additionally, an experimental matrix with test parameters and appropriate output measures was necessary. The first step in this process was to establish key variables that could be studied and the measures indicating the response. With this task completed, it was important to assure that the tunnel was capable of achieving these conditions. The factors established for the study were wind speed, truck cab design (exterior configuration only), collection cavity design, and orientation of the front part of the truck to the wind. A two level fractional factorial design was established as a way to screen these variables in terms of their impacts as main effects to the response. The response chosen was the pressure distribution within the vehicle collection cavity (the open part of the bin). These factors could all be simulated in the tunnel through appropriate design of the tunnel and vehicle model, along with model modifications. With the factors set and the levels of test established, the wind tunnel was reviewed relative to test capability. While important for the test at hand, this tunnel is also used extensively in the Page

4 Technology program for aeronautical, thermal science, and fluid mechanics studies and demonstrations associated with classroom instruction. Therefore, long term factors such as flexibility of design and upgrade were considered. Review of literature for tunnel design quickly revealed that the existing inlet convergence profile, while appropriate, was not optimal and that the convergence ratio was too low to assure good flow quality within the test section 1,5,8,10. While other parameters such as flow conditioning screens could be incorporated or the length of the test entry section changed, the current performance parameters were found to be sufficient for the task at hand. Modifications A new convergence profile was designed incorporating the recommended features of a 6:1 to 12:1 area ratio (inlet-to-test section area ratio), as well as a fifth order polynomial contraction profile. This was done initially using Excel to establish the profile, then constructing a 3-D solid model using Solid Works. After confirming the fit in the solid model software, the features were flattened and the corresponding IGES files were generated. This allowed the fabrication of the components on automated manufacturing equipment available in the MMET Department. Since rolling equipment of sufficient size was unavailable, the design incorporated bends, and the cut metal was formed on a standard sheet metal break. An angle gauge was used to assure the proper degree of bend based on the design drawing. Altogether, the process involved the fabrication of eight component pieces that would be assembled into an inlet section. With the flanges and general contour established, the entire convergence was taken into the welding facility. Starting at one corner, and moving around the arranged pieces, tack welds were applied at approximately two inch intervals, working upward and around in a spiral manner. The intent was to create a rigid assembly that could then be moved or oriented to allow for continuous seam welds. This allowed fabrication of the component without any defects, reducing the likelihood of distortion during the final welding process. Once assembled with the tack welds, work began on the supporting frame. The supporting frame was constructed of angle iron, and the construction again proceeded quickly using the pre-cut sections. Continuous seam welding was used to join the frame and sheet metal and the frame was further stiffened, wheels to allow ease of movement were added, and the internal seams of the square convergence were smoothed. The entire assembly was painted and finally attached to the main tunnel structure. A picture of the final assembly is shown in Fig.1. The entry is 60 inches square, while the test section is 18 inches square, generating a contraction ratio of 11:1. Ports for static pressure measurement were added to three locations in the convergence, and measurements were taken at points along the entire tunnel using an inclined U-tube manometer. Analysis of the data in this table indicates consistent static pressure at the wall in the regions of constant cross-section for the new convergence, indicating stable velocity. Additionally, the pressure increases with increasing fan speed across the range, indicating increasing tunnel velocity with fan RPM. Page

5 Fig.1 Wind Tunnel Inlet The next step in the validation and characterization process was to measure the actual test section speed from the edge to the center and past using a one-quarter inch diameter pitot static probe. The particular tube used was a Dwyer # and the measurements performed were intended to assure that the core flow was stable and that a relatively uniform velocity core existed in the section of the tunnel where the model was placed. A comparison of speeds at one and a quarter inches and in the center of the test section for the old and new convergence was conducted and indicated that improvement in core flow profile consistent with the findings in static pressure measurements were present. It was quickly realized that the measurement process using a U-tube manometer was slow due to the nature of repeated pressure tap connections and manual recording from a single measurement device. Even though a multitube manometer was available for use, the resolution of measurements was not acceptable for the magnitude of the expected measurements from the truck model. The decision was made to incorporate a data-logger and pressure transducers. Four electronic zero to ten volt transducers were purchased, set up in a common fixture as shown below (Fig. 2), and wired into the data-logger also shown. The electronic transducers were Dwyer Model 311 Magnesense and the data-logger was a Hewlett-Packard Model 34970A. Fig.2 Pressure Transducer Measurement System The pressure transducers were compared to each other in an undisturbed flow test to assure consistency between the specific devices and inclined manometer readings. The transducers Page

6 yielded readings which were within one one-hundredth of a volt device to device, and within four one-hundredths of the expected value based on the inclined manometer. This difference most likely resulted from the nature of method used (visual) in obtaining the inclined manometer readings. Readings on the inclined manometer were all estimated within five-thousandths of an inch, which is difficult to justify with any accuracy due to parallax and similar errors. Once completed, testing continued without regard to the specific transducer used for any test location. Readings were taken in groups of four using the data-logger, and subsequently downloaded to a spreadsheet for analysis. This single change allowed for a streamlined testing process for all subsequent measurements and for the actual factorial experiment. Design and fabrication of the truck model in appropriate scale was a concurrent activity with some of the actions already presented. The first step was to compile measurements from the actual full-scale truck. This had previously occurred, and the researcher simply converted these into a solid model using Solid Works. The model was scaled to one eighteenth actual size and subsequently partitioned for fabrication on a rapid prototype machine. The model was constructed as a shell in order to provide an open cavity so the pressure measurement tubes could be placed and routed without significantly disturbing the external truck profile. The particular prototyping equipment used was a machine with maximum dimensions of eight inches by eight inches by twelve inches. This required three pieces to be formed, and then assembled for testing. A section of three-quarter inch aluminum channel measuring one-half inch leg by one-inch web was chosen as the chassis support, and the individual pieces were bolted to this and attached to the wind tunnel base. The assembled truck prior to modification for pressure taps is shown below (Fig.3). Fig.3 Side view of the truck model The model was then modified by adding pressure taps and tubing to connect the points to the pressure transducers in preparation for the experiment. An unblocked resolution four factorial design with replication was the study method. Replication provided an estimate of standard error for the experiment, and increased the study validity and precision. The experiment was established with all input at two levels (high and low). The truck with final modifications in place is shown below. Page

7 Fig.4 Pressure ports distribution in bin Initial levels of the variables were wind speed of 10 and 30 ft/sec, orientation to the wind (yaw) of 0 and 25 degrees, as-manufactured truck profile and a modification, and finally asmanufactured cavity and a modification. Selection of wind speed at these levels was consistent with the potential speeds of operation in residential areas and representative wind gust speeds. In many cases, the only impinging wind was due to the forward motion of the truck. However, the possibility of prevailing high winds during the operation of a full scale truck that may create side or angularly impinging forces to the moving vehicle was considered and changes in yaw was deemed an important parameter to investigate. For this case, the angle chosen allowed the open side of the vehicle to face the wind vector. This angle was the composite vector of a fifteen mile per hour forward speed and a thirty mile per hour direct side wind. Many options are available relative to the angle and speed of crosswind and this is an arbitrary selection for analysis while still allowing the wind speed limits of the factorial experiment to be maintained. Finally, the two design criteria are quite subjective and two somewhat arbitrary choices were made relative to the modifications made to the truck cavity and truck profile. It is common for many blunt cab trucks to have some type of parabolic deflector mounted to the roof of the cab to ensure smooth airflow over the truck top surface including the area above the open collection bin cavity. The other change was to the tapered exit wall of the collection area. In this case, it was thought that the taper might encourage pressure drop and therefore generate favorable gradients. An insert was fabricated to form a completely blunt wall. The test matrix follows indicating the test sequence, randomization order, and factors. Minitab was used as the generating and evaluating program for the factorial design. The experimental output variable, pressure, was recorded from the pressure transducers via the data-logger. The four transducers made this a rather simple process, and the data-logger cycled to collect at least three distinct readings for each position. This output was then directly downloaded into Excel software and finally into the Minitab software for analysis. The data was comprised of fifty-two or fifty-four pressure measurements at each of the test conditions, dependent on the use of the cavity insert. Page

8 StdOrder RunOrder Wind Speed Yaw Cab Cavity Modified Original Original Original Modified Original Original Original Modified Original Modified Modified Original Modified Original Modified Original Modified Modified Modified Original Modified Modified Original Original Original Modified Modified Modified Modified Original Original Table 1. Test matrix for the experimental study Changes to the position and configuration of the truck model were easily made due to previously integrated design features. The model was mounted on a threaded shaft, which allowed easy rotation to change yaw angle. Wind speed was adjusted by changes to fan speed, set by a variable speed controller. Finally, the cab and cavity modifications were quickly changed by simply attaching polystyrene features that helped to alter the profile geometry of the truck airframe. The model had steel plates and magnets integrated in the area where the features attached, allowing quick attachment with good durability and consistent orientation. After each change, sufficient time was allowed for the flow profiles to recover. This procedure persisted for the sixteen tests, following the randomized run order established by Minitab. On average, each run condition required about a half-hour to collect data from all pressure port, with additional time required for model manipulation and data-logger file record archiving. Results After obtaining the pressure measurements, the individual pressure point data was converted to a pressure contour plot in Excel for each run condition. An example of this is shown below for high and low speed tunnel testing, with a scaling factor of 10X applied. Note the similarity of contour, although the absolute magnitude of the response is quite different for the two conditions. Run 2 is at 30 ft/sec wind speed, 25 yaw, with an unmodified cab and cavity. Run 5 is at 10 ft/sec wind speed, 25 yaw, and a modified cab but unmodified cavity. Page

9 Run 2 Run 5 Inches H20 (x10) R 1 R 3 R 5 R 7 R 9 R 11 C11 C10 C9 C8 C7 C6 C5 C4 C3 C2 R 13C Inches H2O (x10) R 1 R 3 R 5 R 7 R 9 R 11 C11 C10 C9 C8 C7 C6 C5 C4 C3 C2 R 13C Fig.5 Pressure mapping inside collection area Visual assessment of the plots indicates an impact of wind speed on pressure field, and indications of possible relationship to yaw angle and cab profile. However, due to the nature of fractionated experimentation, at least two factors are changed during each run. This creates a situation where it impossible to clearly establish key causes through basic one factor variation analysis methods. Minitab allows for the statistical analysis of these effects due to the orthogonal test array originally generated. The only condition necessary is the normalization of data sets to prevent bias in the analysis. That is, due to the nature of the high and low speed testing it is necessary to look at the range of response rather than the absolute response. The normalization involved taking the average of all high speed tests and subtracting this from the individual value for each point measurement. The same occurred for low speed tests, and a new data table was generated and grouped into subsets, which were subsequently averaged for all runs and replications. These groups turned out to be roughly in lines arranged from the front of the truck to the back as defined below. These decisions on measurement point groupings were held constant for all runs and throughout the analysis. The data was then entered into Minitab as a reduced data set of fifteen point clusters for each run, based on the normalized average for the established grouping. A linear regression analysis generated results accounting for only half of the measured response difference based on main effects. Although certain factors are significant, based on the Pareto charts for standardized effect, there was an unexplained element still affecting the pressure response. An attempt was made to reduce the model, as well as apply non-linear regression methods and both resulted in the same condition; half of the pressure response pattern is unaccounted for based on this set of experimental conditions. However, several important conditions do present. First, yaw angle is the most important general factor in affecting pressure conditions within the truck cavity. At 25 angles, the pressure is always higher than when the truck is oriented directly into the wind. Second, wind speed is an important factor in generating pressure increases; higher speed means higher pressure. Finally, use of the cab deflector has the result of reducing the magnitude of pressure throughout the cavity area, independent of speed or orientation. Page

10 Fig. 6 Map of pressure ports locations Conclusions This study was successful in demonstrating several important issues. Technology, and technology based instruction, encompasses a multi-disciplinary set of variables and considerations. The student enrolled in this program should be able to integrate these factors into a solutions based approach to the problem presented. In this study, knowledge of design and tolerancing issues, fabrication, thermofluids principles, and rapid prototyping along with use and application of various software packages and experimentation methods were required. This is similar to the conditions for which we are teaching our students for successful performance in industry. The use and application of fractionated designed experimentation can lead to results or provide a direction more quickly than one-factor experimentation techniques. These design of experiment methods are widely applied in industry due to the reduced cost of experimentation, and should be applied as part of the integrated instruction within our technology programs. While not presented in this paper, it was shown through single factor testing that similar results were obtained to those predicted by the fractionated model. This would require an additional 16 runs, if replication is desired, doubling the time and cost involved. Relative to this study, equipment and methods are now in place to support ongoing testing and demonstrations in the Technology program at this University. Additionally, the instrumentation and data-logging equipment will facilitate more precise measurements and measurement analysis will be improved through automated data transfer. Second, due to the modular design of the changes made, future improvements will allow for conversion to higher velocity within the test section while maintaining reasonable levels of turbulence necessary for a low speed open circuit wind tunnel. Finally, future studies relative to this particular model and problem will involve Page

11 alternative cavity designs to alleviate the high pressure regions along with continued investigation of the factors already part of this study. 1. Bell, J., & Mehta, R. (1988). Contraction design for small low speed wind tunnels (NASA CR ). Washington, DC: National Aeronautics and Space Administration. 2. Dejaegher, B., Smeyers-Verbeke, J., & Heyden, Y. V. (2005). The variance of screening and supersaturated design results as a measure for method robustness. Analytica Chimica Acta, 544(1-2), Fessler, H., et. al., (1995) Wind loading on articulated trailers during tipping. Proceedings of the Institution of Mechanical Engineers. Part D, Journal of Automobile Engineering, 209 (D2), Mehta R.D. (1985). Turbulent boundary layer perturbed by a screen. American Institute of Aeronautics and Astronautics Journal. 23(9), Mehta R.D., Bradshaw P. (1979). Design rules for small low speed wind tunnels. Aeronautical Journal, 83, Natalini, B.; Marighetti, J. O.; Natalini, M. B. (2002). Wind tunnel modeling of mean pressures on planar canopy roof. Journal of Wind Engineering & Industrial Aerodynamics, 90(4/5), Padgett, C. B. (2001) Trash & tribulation? Waste Age, 32(8), Ramaseshan, S., & Ramaswamy, M. (2002). A rational method to choose optimum design for two-dimensional contractions. Journal of Fluids Engineering, 124(2), Roy, Christopher J. et. al. (2006). RANS Simulations of a Simplified Tractor/Trailer Geometry. Journal of Fluids Engineering, 128(5), Sargison, J.E. Walker, G.J. Rossi, R. (2004). Design and calibration of a wind tunnel with a two dimensional contraction. 15th Australasian Fluid Mechanics Conference. The University of Sydney, Sydney, Australia, Thompson, B. (2002). Visibility improvements with overplow deflectors during high-speed snowplowing. Journal of Cold Regions Engineering, 16(3), Page