EVALUATING SPINNER-DISC CONTROL TECHNOLOGY FOR THE DISTRIBUTION OF POULTRY LITTER Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. _________________________ Clayton Marshall Campbell Certificate of Approval: ________________________ ________________________ John P. Fulton, Co-Chair Wesley C. Zech, Co-Chair Assistant Professor Assistant Professor Biosystems Engineering Civil Engineering ________________________ _______________________ Timothy P. McDonald Larry G. Crowley Associate Professor Associate Professor Biosystems Engineering Civil Engineering _______________________ ________________________ Charles W. Wood George T. Flowers Professor Dean Agronomy and Soils Graduate School EVALUATING SPINNER-DISC CONTROL TECHNOLOGY FOR THE DISTRIBUTION OF POULTRY LITTER Clayton Marshall Campbell A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn, Alabama May 9, 2009 iii EVALUATING SPINNER-DISC CONTROL TECHNOLOGY FOR THE DISTRIBUTION OF POULTRY LITTER Clayton Marshall Campbell Permission is granted to Auburn University to make copies of this thesis at its discretion, upon requests of individuals or institutions and at their expense. The author reserves all publication rights. _________________________ Signature of Author _________________________ Date of Graduation iv VITA Clayton Marshall Campbell, son of Steve and Sherre Campbell, was born August 4, 1984, in Gadsden, Alabama. He graduated from West End High School in 2002, and then attended Snead State Community College, in Boaz, Alabama, for two years. In 2004, he enrolled at Auburn University where he graduated with a Bachelors of Biosystems Engineering degree in August of 2006. Directly after graduation, he entered the Auburn University Graduate School to pursue a Master?s of Science degree in Civil Engineering. During graduate school he worked as a graduate research assistant in the Biosystems Engineering Department. He married Kimberly Jackson, daughter of Kem Jackson and Deborah Cunningham, on August 9, 2008. v THESIS ABSTRACT EVALUATING SPINNER-DISC CONTROL TECHNOLOGY FOR THE DISTRIBUTION OF POULTRY LITTER Clayton Marshall Campbell Master of Science, May 9, 2009 (B.B.S.E., Auburn University, 2006) 150 Typed Pages Directed by John P. Fulton and Wesley C. Zech Technological advancements, such as variable-rate technology (VRT), in agricultural application equipment have led to the belief that application accuracy of crop inputs have improved. However, minimal research has been conducted to thoroughly validate this assumption; especially for organic fertilizers such as poultry litter which is inherently variable making it difficult to uniformly apply. Therefore, research was conducted to characterize and compare poultry litter mass and nutrient distribution patterns for a closed-loop system (CLS; spinner-disc control) and an open-loop system (OLS) determining: 1) whether spinner disc-control improves the distribution of litter, 2) vi the association of nutrient and mass patterns, and 3) if spread variability exists along the direction of travel. A typical litter spreader equipped with an electronically adjustable hydraulic flow control (proportional) valve was used to test the CLS and compare these results to the OLS, using a manual valve. Three application rates of 2242, 4483, 6725 kg/ha were selected for applying broiler litter using a two-dimensional pan matrix to assess spread distribution. The results indicated that the CLS was able to maintain more consistent spinner-disc speeds thereby producing less variable distribution patterns over the rates tested. The CLS also produced smaller coefficients of variation, 22% to 34%, for the majority of the mass and nutrient treatments improving spread uniformity by up to 17% over the OLS. Mass (p = 0.0524) and nutrient (p = 0.0657) pattern comparisons revealed that overall differences existed between the two systems. The nutrient patterns were highly correlated (r > 0.98) with their respective mass patterns indicating that even though particle size variability exists across the width of spread, the distribution of mass reflects nutrient distribution. The longitudinal results determined that variability along the direction of travel does exist when litter is applied; however, it was considered random. Overall, the CLS is recommended over the OLS especially if variable-rate application (VRA) is utilized or if application rates are changed frequently. vii ACKNOWLEDGEMENTS First and foremost, I thank God for guiding me through the good times and strengthening me though the bad, without Him I could do nothing. I would like to express my gratitude to all of those in the Biosystems Engineering Department that helped, in any part, toward the finishing point of my research. If it were not for those who helped in the field, Jonathan Griffith, Christian Brodbeck, Corey Kichler, Daniel Mullenix, Ajay Sharda, Tye Harbuck, Russell Davis, etc., I would not have made it to the completion of this thesis. My sincerest appreciation goes out to those who served on my committee, Dr. Tim McDonald, Dr. Wes Wood, and Dr. Larry Crowley, and to Dr. Wesley Zech for serving as the co-chair of this project. I want to especially thank Dr. John Fulton for advising me for the past two years. Thanks for all the leadership, patience, knowledge, and wisdom that you shared with me. Lastly, I would like to thank my wife (Kimberly), my parents (Steve and Sherre), and the rest of my family and friends for all the love and support that they have shown me throughout my academic endeavors. viii Style manual used: Auburn University Graduate School Guide to Preparation and Submission of Theses and Dissertations Computer software used: Microsoft Office 2002 - Microsoft Word 2002, Microsoft Excel 2002; SAS v. 9.1.3 ix TABLE OF CONTENTS LIST OF TABLES ........................................................................................................xv LIST OF FIGURES.....................................................................................................xvii CHAPTER ONE INTRODUCTION..............................................................................1 1.1 PREFACE.................................................................................................................1 1.2 JUSTIFICATION ........................................................................................................3 1.3 OBJECTIVES ............................................................................................................6 1.4 ORGANIZATION OF THESIS.......................................................................................7 CHAPTER TWO LITERATURE REVIEW ..................................................................8 2.1 POULTRY LITTER .....................................................................................................8 2.1.1 PHYSICAL AND CHEMICAL PROPERTIES.............................................................9 2.1.2 NUTRIENT MANAGEMENT...............................................................................11 2.1.3 ENVIRONMENTAL ISSUES................................................................................15 2.1.4 PHOSPHORUS INDEX .......................................................................................16 2.1.5 LITTER STORAGE............................................................................................17 2.2 CALIBRATION........................................................................................................18 2.3 VARIABLE-RATE TECHNOLOGY .............................................................................20 x 2.4 UNIFORMITY .........................................................................................................23 2.5 DISTRIBUTION PATTERNS ......................................................................................25 2.5.1 GRANULAR FERTILIZER TESTING ....................................................................28 2.5.2 POULTRY LITTER TESTING..............................................................................29 CHAPTER THREE SPINNER-DISC TECHNOLOGY TO ENHANCE POULTRY LITTER APPLICATION............................................................................................31 3.1 ABSTRACT ............................................................................................................31 3.2 INTRODUCTION......................................................................................................32 3.3 SUB-OBJECTIVES...................................................................................................33 3.4 MATERIALS AND METHODS ...................................................................................34 3.4.1 DATA COLLECTION ........................................................................................36 3.4.2 DATA ANALYSIS ............................................................................................38 3.5 RESULTS AND DISCUSSION ....................................................................................39 3.5.1 SPINNER SPEED ANALYSIS..............................................................................40 3.5.2 SINGLE-PASS ANALYSIS .................................................................................43 3.5.3 OVERLAP-ANALYSIS ......................................................................................46 3.5.4 DISTRIBUTION PATTERN COMPARISON BETWEEN SYSTEMS.............................50 3.6 SUMMARY.............................................................................................................52 xi CHAPTER FOUR EVALUATING THE APPLICATION OF POULTRY LITTER ON A NUTRIENT BASIS ................................................................................................54 4.1 ABSTRACT ............................................................................................................54 4.2 INTRODUCTION......................................................................................................55 4.3 SUB-OBJECTIVES...................................................................................................57 4.4 METHODOLOGY ....................................................................................................57 4.4.1 SAMPLE PREPARATION FOR NUTRIENT ANALYSIS............................................58 4.4.2 LECO AND ICAP PROCEDURES......................................................................59 4.4.3 DATA ANALYSIS ............................................................................................60 4.5 RESULTS AND DISCUSSION ....................................................................................62 4.5.1 SINGLE-PASS ANALYSIS .................................................................................62 4.5.2 OVERLAP ANALYSIS.......................................................................................63 4.5.3 COMPARING MASS AND NUTRIENT PATTERNS.................................................68 4.5.4 PARTICLE SIZE ANALYSIS...............................................................................68 4.6 SUMMARY.............................................................................................................72 CHAPTER FIVE APPLICATION UNIFORMITY ALONG THE DIRECTION OF TRAVEL....................................................................................................................74 5.1 ABSTRACT ............................................................................................................74 5.2 INTRODUCTION......................................................................................................75 xii 5.3 SUB-OBJECTIVE ....................................................................................................76 5.4 METHODOLOGY ....................................................................................................76 5.4.1 DATA ANALYSIS ............................................................................................77 5.5 RESULTS AND DISCUSSION ....................................................................................78 5.6 SUMMARY.............................................................................................................83 CHAPTER SIX CONCLUSION..................................................................................84 6.1 CONCLUSIONS.......................................................................................................84 6.2 PRACTICAL CRITERIA FOR CONTROL SYSTEM SELECTION.......................................86 6.3 OPPORTUNITIES FOR FUTURE RESEARCH................................................................87 REFERENCES..............................................................................................................89 APPENDIX A TRACTOR AND SPREADER SPECIFICATIONS ............................95 A.1 JOHN DEERE MODEL 6420 TRACTOR.....................................................................96 A.2 CHANDLER EQUIPMENT COMPANY C/L LITTER AND SHAVINGS SPREADER ............97 APPENDIX B SPREADER PARTS............................................................................98 B.1 HYDRAULIC SPINNER MOTORS..............................................................................99 B.2 PRINCE HYDRAULIC PTO PUMP ............................................................................99 B.3 CROSS HYDRAULIC RELIEF VALVE .....................................................................100 APPENDIX C HYDRAULIC FLOW CONTROL VALVES.....................................101 xiii C.1 BRAND HYDRAULICS: ELECTRONICALLY ADJUSTABLE PROPORTIONAL PRESSURE COMPENSATED FLOW CONTROL VALVE ....................................................................102 C.2 BRAND HYDRAULICS: MANUAL FLOW CONTROL VALVE ....................................102 APPENDIX D ZYNX X15 AND X20 ......................................................................103 D.1 KEE TECHNOLOGIES ZYNX X20 CONSOLE...........................................................104 D.2 KEE TECHNOLOGIES ZYNX X15 CONSOLE...........................................................105 APPENDIX E SENSORS AND ELECTRONIC SPECIFICATIONS .......................106 E.1 INDUCTIVE PROXIMITY SENSORS.........................................................................107 E.2 DICKEY JOHN ENCODER......................................................................................107 E.3 PRESSURE TRANSDUCERS....................................................................................108 E.4 MEASUREMENT COMPUTING USB 1608FS ..........................................................108 APPENDIX F VISUAL BASIC PROGRAMS .........................................................110 F.1 PROGRAM TO COLLECT SPEEDS ...........................................................................111 F.2 PROGRAM TO COLLECT SPEEDS AND PRESSURES..................................................113 APPENDIX G SPINNER SPEED DATA..................................................................119 G.1 SPINNER SPEED DATA FOR INDIVIDUAL REPLICATIONS........................................120 APPENDIX H MASS DISTRIBUTION DATA ........................................................121 H.1 MASS OVERLAP DISTRIBUTION PATTERNS ..........................................................122 APPENDIX I NUTRIENT SIMULATED OVERLAP DATA...................................124 xiv I.1 SUMMARIZED NUTRIENT SIMULATED OVERLAP DATA ..........................................125 APPENDIX J LONGITUDINAL ANALYSIS REPORTING THE SUMMARIZED RESIDUAL DATA ..................................................................................................127 APPENDIX K PRESSURE DATA............................................................................130 xv LIST OF TABLES TABLE 3.1. WIND SPEEDS OBSERVED DURING PATTERN TESTING......................................38 TABLE 3.2. POULTRY LITTER MC AND BD FOR EACH REPLICATION..................................40 TABLE 3.3. SUMMARY OF PRE-TEST SPINNER-DISC SPEED DATA........................................41 TABLE 3.4. SUMMARY OF SPINNER-DISC SPEED DATA COMPUTED FOR ALL REPLICATIONS. 42 TABLE 3.5. SUMMARY STATISTICS FOR OVERALL SIMULATED OVERLAP PATTERNS............47 TABLE 3.6. ANOVA RESULTS FOR THE OVERLAP PATTERN DATA.....................................48 TABLE 3.7. PEARSON CORRELATION COEFFICIENTS COMPARING THE MEAN SINGLE-PASS DISTRIBUTION PATTERNS OF EACH SYSTEM. .............................................................52 TABLE 4.1. SUMMARY STATISTICS FOR OVERALL NUTRIENT SIMULATED OVERLAP PATTERNS...............................................................................................................66 TABLE 4.2. ANOVA RESULTS FOR THE NUTRIENT SIMULATED OVERLAP PATTERN DATA. .67 TABLE 4.3. OVERALL MEASURED POULTRY LITTER NUTRIENT CONCENTRATIONS..............70 TABLE 5.1. EXAMPLE RESIDUAL CALCULATIONS AT SELECTED PAN LOCATIONS FOR THE CLS 2242 KG/HA TESTS SHOWING THE 4 ROWS, LABELED A, B, C, AND D WITH A REPRESENTING THE FIRST ROW RECEIVING LITTER BY THE SPREADER........................81 TABLE 6.1. EVALUATION CRITERIA FOR SELECTING A CONTROL SYSTEM. .........................87 TABLE G.1. REPLICATION 1 SPINNER SPEED SUMMARY. .................................................120 TABLE G.2. REPLICATION 2 SPINNER SPEED SUMMARY. .................................................120 TABLE G.3. REPLICATION 3 SPINNER SPEED SUMMARY. .................................................120 xvi TABLE H.1. REP 1 SIMULATED MASS OVERLAP PATTERN SUMMARY STATISTICS..............123 TABLE H.2. REP 2 SIMULATED MASS OVERLAP PATTERN SUMMARY STATISTICS..............123 TABLE H.3. REP 3 SIMULATED MASS OVERLAP PATTERN SUMMARY STATISTICS..............123 TABLE I.1. REP 1 SIMULATED NUTRIENT OVERLAP PATTERN SUMMARY STATISTICS. .......125 TABLE I.2. REP 2 SIMULATED NUTRIENT OVERLAP PATTERN SUMMARY STATISTICS. .......126 TABLE I.3. REP 3 SIMULATED NUTRIENT OVERLAP PATTERN SUMMARY STATISTICS. .......126 TABLE J.1. SUMMARIZED RESIDUAL DATA FROM THE CLS 2242 KG/HA APPLICATION RATE. ............................................................................................................................128 TABLE J.2. SUMMARIZED RESIDUAL DATA FROM THE CLS 4483 KG/HA APPLICATION RATE. ............................................................................................................................128 TABLE J.3. SUMMARIZED RESIDUAL DATA FROM THE CLS 6725 KG/HA APPLICATION RATE. ............................................................................................................................128 TABLE J.4. SUMMARIZED RESIDUAL DATA FROM THE OLS 2242 KG/HA APPLICATION RATE. ............................................................................................................................129 TABLE J.5. SUMMARIZED RESIDUAL DATA FROM THE OLS 4483 KG/HA APPLICATION RATE. ............................................................................................................................129 TABLE J.6. SUMMARIZED RESIDUAL DATA FROM THE OLS 4483 KG/HA APPLICATION RATE. ............................................................................................................................129 TABLE K.1. SUMMARIZED SPINNER-DISC SPEED AND PRESSURE DATA FOR THE CLS. ......131 TABLE K.2. SUMMARIZED SPINNER-DISC SPEED AND PRESSURE DATA FOR THE OLS. ......131 xvii LIST OF FIGURES FIGURE 1.1. U.S. FERTILIZER PRICING FROM 1960 TO 2007 (ERS-USDA, 2007).................1 FIGURE 1.2. AMOUNT OF LITTER PRODUCED DURING 2006 ALABAMA (BROWN DOTS). ........2 FIGURE 3.1. TRACTOR, LITTER SPREADER, AND ILLUSTRATION OF COLLECTION PAN MATRIX UTILIZED DURING TESTING (A) AND REARVIEW IMAGE OF SPREADER SHOWING SPINNER-DISCS (B). .................................................................................................34 FIGURE 3.2. ILLUSTRATION OF THE TWO DIFFERENT FLOW CONTROL VALVES USED DURING TESTING WITH THE PROPORTIONAL VALVE SHOWN ON THE LEFT AND MANUAL VALVE ON THE RIGHT. ........................................................................................................35 FIGURE 3.3. PAN LAYOUT FOR SINGLE-PASS TEST (A) AND EQUIPMENT TRAVERSING PANS (B)..........................................................................................................................36 FIGURE 3.4. OVERALL MEAN SINGLE-PASS DISTRIBUTION PATTERNS FOR THE TWO CONTROL SYSTEMS.................................................................................................................44 FIGURE 3.5. MEAN SINGLE-PASS DISTRIBUTION PATTERNS AND 95% CI. ..........................45 FIGURE 3.6. ILLUSTRATION OF SYSTEM BY RATE INTERACTION........................................48 FIGURE 3.7. OVERALL SIMULATED OVERLAP DISTRIBUTION PATTERNS. ............................49 FIGURE 3.8. OVERALL STANDARDIZED DISTRIBUTION PATTERNS......................................51 FIGURE 4.1. THE RED RECTANGLE ILLUSTRATES HOW LONGITUDINAL PANS WERE COMBINED PRIOR TO THE NUTRIENT ANALYSIS PROCEDURES. ...................................58 xviii FIGURE 4.2. OVERALL MEAN NUTRIENT SINGLE-PASS DISTRIBUTION PATTERNS FOR THE TWO CONTROL SYSTEMS..........................................................................................63 FIGURE 4.3. OVERALL NUTRIENT SIMULATED OVERLAP PATTERNS FOR EACH CONTROL SYSTEM BY APPLICATION RATE TREATMENT.............................................................64 FIGURE 4.4. COMPARISON OF THE MASS AND NUTRIENT STANDARDIZED DISTRIBUTION PATTERNS FOR BOTH CONTROL SYSTEMS AT EACH TEST RATE...................................69 FIGURE 4.5. TRANSVERSE PARTICLE SIZE DISTRIBUTION CATEGORIZED BY SIEVE SIZE.......71 FIGURE 5.1. PAN LAYOUT AND SELECTED POSITIONS FOR REPLICATION 1 CLS 2242 KG/HA TEST. ......................................................................................................................77 FIGURE 5.2. SINGLE-PASS MASS PATTERNS FOR THE 2242 KG/HA APPLICATION RATE SEPARATED BY ROW TO ANALYZE VARIABILITY ALONG THE DIRECTION OF TRAVEL...79 FIGURE 5.3. APPLICATION SURFACES FOR THE 2242 KG/HA APPLICATION RATE TO VISUALIZE THE VARIABILITY ALONG THE DIRECTION OF TRAVEL IN KG/HA. ...............80 FIGURE 5.4. RESIDUAL PLOTS FOR EACH CONTROL SYSTEM AND TEST RATE......................82 FIGURE A.1. JOHN DEERE 6420.......................................................................................96 FIGURE A.2. CHANDLER EQUIPMENT CO. LITTER SPREADER. ...........................................97 FIGURE A.3. LITTER SPREADER REAR GATE, CONVEYOR CHAIN, FLOW DIVIDER AND SPINNERS................................................................................................................97 FIGURE B.1. PARKER HYDRAULIC SPINNER MOTOR. .........................................................99 FIGURE B.2. PRINCE HYDRAULIC PTO PUMP....................................................................99 FIGURE B.3. HYDRAULIC PRESSURE RELIEF ON THE INPUT TO THE CONVEYOR VALVE. ....100 FIGURE C.1. BRAND PROPORTIONAL VALVE USED FOR SPINNER AND CONVEYOR CONTROL. ............................................................................................................................102 xix FIGURE C.2. BRAND MANUAL VALVE USED FOR SPINNER CONTROL. ...............................102 FIGURE D.1. X20 CONSOLE WITH SPREADER CONTROL SOFTWARE.................................104 FIGURE D.2. X15 CONSOLE USED FOR COLLECTING SPEED AND PRESSURE DATA. ............105 FIGURE E.1. SENSORS TO MONITOR SPINNER SPEEDS......................................................107 FIGURE E.2. ENCODER TO MONITOR CONVEYOR SPEED. .................................................107 FIGURE E.3. HYDRAULIC PRESSURE SENSOR. .................................................................108 FIGURE E.4. USB DATA LOGGER FOR PRESSURE DATA. ..................................................108 FIGURE H.1. MASS OVERLAP PATTERNS FOR EACH REPLICATION BY TYPE OF CONTROL SYSTEM AND APPLICATION RATE. ..........................................................................122 1 CHAPTER ONE INTRODUCTION 1.1 PREFACE In Alabama and across the United States, use of organic fertilizer has increased considerably over inorganic fertilizers due to recent escalating prices of manufactured fertilizers. Figure 1.1 illustrates fertilizer pricing in the U.S., for three of the most common types of fertilizer, from 1960 to 2007. From 2002 to 2007 alone, ammonium nitrate price increased by 49%, super phosphate by 47%, and potassium chloride by 41% (ERS-USDA, 2007; figure 1.1) leading producers to consider using organic fertilizers since they are typically cheaper while providing similar fertilizer value for producing crops. 0 50 100 150 200 250 300 350 400 450 1960 1970 1980 1990 2000 2010 Year Pr ice ($ /to n) Ammonium Nitrate Super Phosphate 46% Potassium Chloride 60% Figure 1.1. U.S. fertilizer pricing from 1960 to 2007 (ERS-USDA, 2007). 2 Similarly, poultry production, especially in Alabama, has also increased over the last decade. In Alabama, there was approximately 1.7 million tons of poultry litter produced during 2006 (figure 1.2; Mitchell et al., 2006). The quantity of litter produced in the state has become a problem due to the fact that the majority of poultry farms are located in the northern half of the state with 28% of poultry (broiler) production occurring in four neighboring counties: Blount, Cullman, Marshall, and Dekalb (NASS- USDA, 2007). Dense poultry production in this area promotes over-application of litter which is attributable to the high cost of transporting the litter to areas of low soil fertility. Figure 1.2. Amount of litter produced during 2006 Alabama (brown dots). Several past and current research efforts have been conducted to offset rising fertilizer costs and more efficiently manage organic fertilizer production. However, limited research, if any, has addressed using variable-rate technology (VRT) to more efficiently apply organic fertilizers, such as poultry litter. VRT is growing among the 3 agricultural community and in recent years has been recognized as a method to increase input use efficiency for applying inputs (seed, fertilizer, lime, pesticides, etc.) to cropland while improving environmental stewardship. The concept of applying poultry litter using VRT is a new method which has not been thoroughly investigated. This research makes a step toward the concept of variable-rate application (VRA) of poultry litter by testing the hypothesis that controlling spinner disc speeds when varying application rate on a standard spinner disc spreader will improve the distribution of poultry litter. 1.2 JUSTIFICATION Limited research has been conducted to thoroughly investigate the application of poultry litter. Most studies have focused on synthetic fertilizers, such as urea, potash, ammonium nitrate, etc., with little attention toward organic fertilizers. With the increase in poultry litter production in Alabama and the potential negative environmental effects that are related with the over-application of litter, it is believed that measures need to be taken to improve distribution of poultry litter during application. Recent precision agriculture (PA) technology could have the potential to improve over-application issues associated with poultry litter. Poultry litter is often used as a fertilizer and soil amendment even though it is inherently variable in its physical characteristics, making it difficult to maintain the desired uniform distribution using standard spinner-disc spreaders; the most common equipment used to land apply litter. In previous research, numerous tests have been conducted to determine the effects of certain spreading variables on the application of poultry litter but no attempt has been made to control the speed of the spinner discs of a 4 standard litter spreader on-the-go. This type of technology is capable of providing a more desirable spread pattern, improving uniformity, and ultimately lowering the potential for environmental degradation especially if application rates are being changed. Maintaining an acceptable distribution of litter is essential to reduce over- application of litter in environmentally sensitive areas. The foremost environmental concern with litter is its phosphorus (P) content. On average, litter has an average fertilizer rating of 3-3-2 (N-P2O5-K2O; Wood, 1992), meaning that it contains as much P2O5 as it does nitrogen (N). This becomes a problem when farmers apply litter to meet N requirements which is typically much higher than the P requirement. This type of action leads to a buildup of P in the soil potentially causing harmful amounts of P to be deposited into surface waters via runoff. This is particularly an issue in Alabama where approximately 90% of litter generated is used to fertilize crop and pasture lands. Thus, more recently states are basing litter application on P to meet environmental regulations. If too much P reaches the surface water, the process of eutrophication can take place. Eutrophication is simply an increase in chemical nutrients. In this case, P initializes rapid growth of algae in water, often known as an algae bloom. Microorganisms in the water then feed on the algae, taking in large amounts of oxygen, along with the decomposition of the algae, depleting the water of the required amount of oxygen for aquatic life to survive. This creates a harmful environment for aquatic life. The Phosphorus index (P index), a tool to assess P movement across the landscape, is used to help eliminate some of the hazards associated with P application. 5 When applied appropriately, poultry litter is a good fertilizer source providing all the major nutrients. Wilhoit et al. (1993) measured N and carbon (C) concentrations but not P and K. To effectively manage litter nutrients it is important to study the effects that all the macronutrients have on the spread pattern. Also, particle size can impact distribution and thus needs to be considered when taking the variability of litter into account. In most cases, particle size tends to affect the nutrient management in poultry litter. Koon et al. (1992) determined that the nutrient fraction for each of the macronutrients (N, P, and K), increased as the particle size fraction decreased. However, Wilhoit et al. (1993) reported that carbon (C) concentrations increased with increased particle size and nitrogen (N) content within each size fraction varied randomly. It is common for higher concentrations of nutrients to reside in the smaller particles when dealing with poultry litter. With this being the case and the fact that when using a spinner spreader, large and small particles are distributed differently across the swath, it is important to determine how nutrient and mass patterns interact with one another. Traditionally, only the amount of material applied (mass basis) is considered with no thought about how the nutrients may be distributed. If mass and nutrient distribution varies differently then maybe nutrient content rather than material mass should be considered when applying litter. As previously stated, variability of poultry litter has been studied for mass, nitrogen, and carbon patterns. These studies however were only conducted in one location with one row of collection devices. More focus needs to be put on how to obtain a better understanding of uniformity as the applicator traverses a certain area. Multiple rows of collection devices would potentially inform the operator about the overall 6 applicator performance both longitudinally (along the direction of travel) and transversely. This 2-D testing can better help understand variability of the spreader. A key goal when applying fertilizer is to apply the desired amount. Utilizing VRT to vary application rate across a field can reduce over-application of litter by spatially applying the proper amount to meet local fertility needs. Due to the natural variability of litter, concerns exist that the use of VRT with spinner-disc control may actually have a negative impact on litter application. In essence, VRT might increase application errors associated with litter. However, if it does improve litter application, then it is assumed that it can improve distribution of synthetic fertilizers when VRT is implemented as long as the proper spreader and control settings are utilized. A new technology being considered to improve the distribution of granular or dry products by spinner spreaders is spinner-disc control. The idea behind this technology is that a closed-loop system is used to maintain the set spinner-disc speed no matter the mass flow of material conveyed onto the discs. Traditional spinner disc speed control uses an open- loop system which is unable to compensate for varying material flow onto the discs resulting in speed fluctuations. Therefore, maintaining the set speed could improve application uniformity. 1.3 OBJECTIVES The overall goal of this research was to determine if maintaining a constant spinner disc speed via a closed-loop system (CLS) can improve the distribution of poultry litter compared to a traditional open-loop system (OLS) since speed variations can exist with the open-loop system. The objectives of this research were to: 7 1. Evaluate a traditional open-loop system (OLS) for spinner-disc speed control on a poultry litter spreader versus a closed-loop system (CLS) over a range of application rates. 2. Compare and contrast characterized litter mass and nutrient patterns to determine if nutrients are spread differently than mass along with assessing the difference between an OLS and CLS for spinner-disc speed control. 3. Determine if longitudinal variability exists when applying poultry litter with the open- and closed-loop systems over a range of application rates. 1.4 ORGANIZATION OF THESIS This thesis is presented in manuscript format. Chapter 1 provides introductory statements justifying the emphasis that was put toward this research followed by the main objectives. Chapter 2 is an extensive literature review supplying information on the characteristics of poultry litter and poultry litter application. Each of the Chapters 3 through 5 represents an individual manuscript that focuses on different portions of this research. The results within these chapters illustrate the manner in which poultry litter is distributed using two spinner-disc speed control systems. Chapter 3 characterizes and assesses the mass distribution patterns provided by these systems; where Chapter 4 does the same for the nutrient distribution patterns and forms a comparison with the mass and nutrient patterns. Chapter 5 covers the variability that an applicator provides as it traverses longitudinally across an area. Chapter 6 summarizes the project, presents the overall conclusions, and includes suggestions for opportunities of future research. At the end of the thesis, a single Reference section and Appendices was developed. 8 CHAPTER TWO LITERATURE REVIEW With escalating prices of inorganic fertilizers, organic fertilizers, such as poultry litter, are being heavily utilized as a major source of crop nutrients. In recent years, this increasing trend has initiated research projects focusing on the different variables associated with using organic fertilizers. Several publications were reviewed to gain knowledge on the characteristics of organic and inorganic fertilizers, mainly focusing on poultry litter. These articles included physical and chemical properties, storage, and environmental impacts of poultry litter as well as its ability to be uniformly distributed during field application to cropland. However, limited research has been conducted to fully understand the fertilizer value of poultry litter and our ability to apply it accurately based on site-specific crops needs or fertility levels. Therefore, several manuscripts were reviewed to understand the use of variable-rate technology (VRT) to more efficiently spread granular fertilizers and lime. Other publications related to application distribution and uniformity were also reviewed to quantify the potential of using similar ideas when applying poultry litter. 2.1 POULTRY LITTER In Alabama, 1.5 to 2 million tons of poultry litter are produced annually (Mitchell and Tyson, 2007). Approximately 90% of this litter is utilized as fertilizer by applying it to crop and pasture lands. Litter particles are variable in size and nutrient concentration 9 making it difficult to uniformly apply based on crop and soil requirements. The majority of Alabama poultry production is located in the northern half of the state with production facilities densely located within a few counties with limited land around these facilities to apply litter. Therefore, since litter is not a dense material, it is not economical to transport over large distances resulting in litter being applied near these facilities and leading to multiple applications within the same field or pasture over the years. Transporting the material to locations of low soil fertility is an option but is generally a route that is not taken due to high fuel prices (Wood et al., 1992). Decades of over- application have led to environmental issues with high phosphorus (P) levels in surface water which initiated the creation and use of the P index to manage the application of litter. 2.1.1 PHYSICAL AND CHEMICAL PROPERTIES When trying to attain the most efficient spread pattern, physical and chemical properties of the material being utilized are important since they can impact uniformity. Koon et al. (1992) conducted a study to determine if physical and chemical characteristics of pine shavings, used as a bedding material in a poultry house, changed over a four grow-out period. Samples were taken after each grow-out for a period of four grow-outs and analyzed for particle and chemical analyses. After each grow-out, fresh pine shavings were placed on top of the old litter, after samples were taken. Each sample was sieved, and then each size fraction was averaged and blended together for chemical analyses. Three samples of fresh pine shavings were also sieved and analyzed in the same manner. Results indicated there was little variation in particle size of the pine shavings poultry litter over the four grow-out period. As for chemical analyses, the 10 nutrient concentration for each of the macronutrients (nitrogen (N), phosphorus (P), and potassium (K)) increased as the particle size fraction decreased However, the majority of nutrients were retained on the larger sieves resulting from more mass being retained on the larger sieves. In contrast, Wilhoit et al. (1993) reported that carbon (C) concentrations increased as particle size increased and the N content within each size fraction varied randomly. Pezzi and Rondelli (2002) noted that particle size of poultry litter seemed to decrease with longer storage times. Glancey and Hoffman (1996) investigated physical properties of poultry manure and compost to determine bulk mechanical properties and their effect on material handling systems, such as spreaders. The properties investigated were bulk density, moisture content (MC), angle of repose, maximum lump size, and static frictional characteristics. Tests were conducted on fresh poultry manure clean-out and crusted material, crusted and clean-out poultry manure stored at 5 weeks and 14 weeks, and fresh compost material under three conditions: poultry manure composted with dead chickens, municipal solid waste (MSW) composted with dewatered sludge, and MSW composted with poultry manure. An analysis of variance determined that outside storage and an exposure to rainfall of poultry manure significantly increased the MC, static coefficient of friction, and wet bulk density (majority of the increase was within the first 5 weeks of outside storage). Outside storage did not affect the angle of repose or lump size of poultry manure. A dependence of wet bulk density on MC across all of the solid wastes evaluated was also determined. Results from this finding illustrated that MC is more important than knowing the source of waste material. There was little practical difference of the static coefficients of friction for poultry manure with a high or low MC 11 with regard to designing material handling systems. However, it should be noted that unscreened waste has larger lump sizes and should be considered when designing these systems. Wilhoit et al. (1993) also determined the importance that litter MC and number of flocks raised on the litter can play when trying to attain uniform application when considering mass as well as nutrient content. 2.1.2 NUTRIENT MANAGEMENT Many researchers over the years have studied ways to more efficiently manage broiler litter as a fertilizer (Coloma et al., 2004; Wood, 1992; Mitchell et al., 2007). Coloma et al. (2004) and Wood (1992) have both concluded that litter should be combined with an inorganic N fertilizer then applied. This blending will meet the soil P requirements as well as the crop nutrient requirements while minimizing environmental impacts. This blend would also cut down on the hauling expense of fertilizer to the field as well as the over application of P (Coloma et al., 2004). Wood (1992) also stated the reason poultry litter is often preferred over other manures is because of its high nutrient content and the fact that it can produce relatively equivalent yields as synthetic fertilizers but at lower costs. Coloma et al. (2004) conducted tests on untreated broiler litter as well as treated (screened) to determine the available nitrogen (AN):P2O5 and C:N ratios. Results illustrated that screening the litter significantly lowered the bulk density of the retained fraction indicating a higher porosity in the retained fraction. However, the screened fraction contained a higher portion of nutrients. Results of the C:N tests indicated that the retained fraction was significantly higher than the untreated litter, however, for its use in composting there would still need to be some carbon material added to reach the 12 desired C:N ratio. The results of the AN:P2O5 test showed the AN:P2O5 ratio for the screened litter was not significantly increased as compared to the untreated litter. In conclusion, they determined that the fraction of nutrients passing through the screen followed a similar pattern as the fraction of raw litter mass and by blending an inorganic fertilizer with the screened fraction could reduce the material application rate by 72% when compared to untreated broiler litter. This indicates that only a portion of the smaller fraction of litter after sieving would need to be used to supply the nutrient requirements of specific crops. This is based on the finding that nutrient concentration is higher in smaller particle sizes. This method would require added N fertilizer for crops or pastures but could help minimize environmental concerns associated with P runoff by reducing the amount of applied litter. The NRCS Code 590 (USDA-NRCS, 2002) was created to manage all aspects of nutrient application to the soil by setting regulations on timing, amount, source, and placement of nutrients. Many laws were generated to attempt to reduce environmental pollution related to applying animal waste to the soil. The regulations that pertain to the application of poultry litter in Alabama are: application shall be 15.24 m from surface waters of the state, 30.48 meters from the nearest occupied dwelling, church, school, hospital, park, or non- potable water wells, 61 meters from Outstanding National Resources Water, Outstanding Alabama Water, potable water wells, or public water supply, and it is not to be applied across property boundaries unless the adjoining property owner consents in writing. All precautions should be taken to eliminate or minimize nonpoint source pollution to the ground and surface waters. Each site, farm, or field shall be evaluated using the P index and the Leaching Index to assess the movement 13 of applied nutrients in the soil to protect the quality of the water resources in the state. For those fields that are located in environmentally high risk areas, erosion, runoff, and water management controls shall be installed. To determine the allowable amount of nutrients that can be applied, a soil test must be conducted using either the Auburn University Soils Testing Laboratory or an acceptable laboratory. Soil tests older than three years shall not be used for nutrient planning. It is recommended that soil amendments, such as lime, should be used to adjust soil pH prior to nutrient application. When it comes to nutrient application, it states that the application of nutrients needs to be based on current soil test reports and that the application shall not exceed 10% of the intended rates of the field. When applying organic by-products, such as poultry litter, the acceptable rate is generally based on the amount of P that can be applied to the soil due to the P index rating of the field. When the vulnerability rating (P index rating) is very low/low, litter can be applied to meet the N requirement even if it means the P rating exceeds 10% of the established application rate. However, when a rating above medium is determined, the litter should be applied to meet the P intended rate and in this case an additional source of N can be used to meet the N requirement. Organic by-products can not be applied in Alabama during the fall and winter seasons unless it is on actively growing crops. In north Alabama, no application can occur between November 15 and February 15 due to crop inactivity. Mitchell et al. (2007) addressed issues of nutrient management when dealing with broiler operations to protect water quality. The authors discussed the need for a Comprehensive Nutrient Management Plan (CNMP) on every farm and the requirements needed to apply the CNMP. The nutrients of primary concern were N and P due to their 14 leaching and runoff characteristics, respectively. There were five steps to this CNMP. The first being, estimate broiler litter amount (pounds of meat produced per year * ? pound of litter per pound of meat), compost production, and storage facilities. Next, estimate the nutrient value of the litter and compost. Poultry broiler litter is generally a 3-3-2 fertilizer rating (Mitchell et al., 2007 and Wood, 1992). Then, map and calculate land area for spreading using an aerial photo or topo map. Next, determine the crop and nutrient needs for each field using recent soil tests. The P index also needs to be utilized to determine the amount of P that can be applied. Finally, determine uses for excess litter and compost. Armstrong et al. (2006) examined irregular soil sampling on a field of long-term litter application to predict the areas of accumulation and loss of nutrients in a field. Soil samples from plots with different topography on an irregular grid were collected and analyzed for nutrient content. Nutrient accumulations for both fields were identified easily with the irregular soil sampling method. For field A, the elevation increased as the N and P concentrations decreased, however, field B had a positive relationship between elevation and N and P. This research determined that by using irregular soil sampling points and focusing on topography and landscape of a field that nutrient accumulation after long-term litter application can be determined. This was accomplished by examining topography and landscape positions to determine water movement via hydrological pathways. This type of sampling could help create more accurate nutrient management plans that in the long term would reduce surface and groundwater pollution in areas of long-term poultry litter application. 15 A study was conducted in north Alabama on pastures of long term litter application and on pastures with no litter application to determine the severity of litter application to the landscape (Wood, 1992). It was found that in the litter applied fields more nitrate (NO3-N) was found below 50 inches indicating that excessive NO3-N leaching occurs on litter applied fields. Also, in the long term litter applied fields the extractable P concentrations averaged 530% higher than the other fields in the upper 0.61 m of the soil. Wood (1992) concluded that long term litter application at the disposal rate degraded the environment and one way to minimize this degradation was to apply litter based on the soil P test which also decreased the NO3-N leaching. 2.1.3 ENVIRONMENTAL ISSUES Even though poultry litter is a good source of fertilizer, it can potentially impact the environment (Wood, 1992; Coloma et al., 2004; and Armstrong et al., 2006). The majority of environmental contamination occurs in dense poultry producing regions such as the mountainous regions of Alabama and Arkansas. This problem originates because there is not enough land to safely spread all the litter that is produced, leading to over application in areas of high slope and shallow soils to the bedrock. Other hazards that are associated with broiler litter application are: poor timing of disposal, low efficiency of nutrient recovery, and lack of knowledge concerning nutrient, heavy metal, and soluble salt release (Wood, 1992). Nitrate leaching and P runoff are the two major environmental concerns when discussing poultry litter application. When NO3 reaches the groundwater it can have harmful effects on humans as well as livestock if too much of it is consumed (Armstrong et al., 2006 and Wood, 1992). Farmers often apply litter to meet the N requirement of 16 their crop; however, in doing so they over apply P as well as N leading to the previously stated issue. This over application often leads to a buildup of P in the top layer of the soil and through runoff and erosion makes it to the surface water in terms of a pollutant, diminishing the water quality and putting the aquatic life into a hazardous situation (Coloma et al., 2004; Armstrong et al., 2006; and Wood, 1992). In certain regions, the environmental hazard related to litter could be controlled if the cost of the transportation was low enough to haul the material to areas of low soil fertility or high yielding crops. 2.1.4 PHOSPHORUS INDEX The Phosphorus Index, commonly referred to as the P index, is nothing more than a tool utilized to assess the risk of P movement into surface waters. The P index is used widely across the agricultural community as well as many other environmental agencies. It is important for farmers, agronomist, engineers, etc. to understand the P index and its parameters. Without this knowledge it is easy to over apply and apply excess P that can be discharged into surface water. The main purpose of the P index is to identify sites that are of potential hazard to the environment. These hazards are associated with the potential risk of P movement to water bodies. The movement of P can be categorized into three main factors: transport, P management, and P source (USDA NRCS, 1994). Over the past many years the P index has become an exceptional tool when it comes to protecting and preserving the environment. Many versions of the P index exist due to differing regional and geographic conditions. Alabama?s P index determination method is somewhat different than the method proposed above by the NRCS. It incorporates other factors that are specific in nature to Alabama (USDA-NRCS, 2001). Best management practices (BMP) are being put into affect to reduce site vulnerability to P 17 applications ultimately reducing the P index. Some of these BMPs are grassed waterways, setbacks from streams, filter strips, limited animal access to surface waters, and lower P applications (Mitchell and Tyson, 2007). 2.1.5 LITTER STORAGE Each state has its own set of rules and suggestions when it comes to poultry litter storage. Regulations do seem to very from state to state; however, each state always seems to have one main point, environmental quality. Literature summarized below was selected from two states and the issue of environmental quality is addressed in each. According to the Virginia Department of Environmental Quality (VDEQ), if poultry litter is not going to be used immediately it must be stored properly. The storage facility must be of adequate size and located where it will not cause environmental risk to water quality. The site must be 30.48-m from surface water, intermittent drainage, wells, sinkholes, and rock outcrops with a slope no greater than 7%. If litter is to be stored outside longer than 14 days it must be covered with an impermeable layer that will not allow storm water to run onto it or under it and it must resist wind. If it is to be stored where the water table is less than 0.61-m, then an impermeable layer should be placed underneath the litter. Acceptable layers are: 30.48-cm of compacted clay, 10.16-cm of concrete, or other impermeable layers with a minimum permeability rating of 0.0036- cm/hr. No one is permitted to store litter where the water table is less than 0.030-m. One must remove all litter residues from the storage area when storage is no longer needed. The Alabama Cooperative Extension System (ACES) identified BMPs to help minimize litter storage in hopes to reduce possible environmental risks. Determining 18 storage requirements and sizing storage structures is generally determined by estimating broiler production and the density of the litter (on average 500-kg/m3). Managing litter can reduce the need for litter storage. BMPs for reducing litter storage include: schedule cleanouts so they can be land applied and reduce wet spots in the house by using more efficient drinker lines. There are many ways to store litter, such as, open stockpile (must be compacted), covered stockpile, covered stockpile with temporary ground liner, covered stockpile with permanent ground liner, and a roofed storage structure. No matter the storage method, the litter must be protected from rainfall, leaching, and runoff. Effective storage of litter retains nutrients in the manure as well as protects the environment (Donald et al., 1996). 2.2 CALIBRATION Calibration is important to determine the rate and uniformity that the spreader is operating at and is a key component in maintaining a target rate. It also helps setup the hardware and software when using VRT. Fulton et al. (2005b) found that the simulated overlap plots displayed that pattern adjustments could be made to produce better distribution patterns for all applicators and also that overlap patterns should be generated at calibration to more efficiently quantify application uniformity. If improper calibration of an applicator is conducted then the applicator could be off target with the desired application rate and distribute material incorrectly. Proper calibration can also reduce environmental risk associated with applying poultry litter (Mitchell and Tyson, 2001). Marsh et al. (2003) and the Virginia Cooperative Extension stated that it is important to apply manure at the desired rate to meet, however not exceed, the nutrient requirements of a specific crop. 19 The American Society of Agricultural and Biological Engineers (ASABE) Standard, S341.3, Procedure for Measuring Distribution Uniformity and Calibrating Granular Broadcast Spreaders, provides a uniform method to test, analyze, and report performance data on spinner spreaders (most common type of fertilizer applicator). This standard establishes guidelines for test setup, collection devices, test procedures, determination of application rates, and effective swath width. Examples of a few of the standard test setup variables include ground slope (<2%), wind velocity (<8-km/h), and hopper fill level (at least 40% to 50% capacity) (ASABE Standards, 2004). The ACES published an article identifying a procedure for calibrating poultry litter spreaders considering the large amount of litter produced in Alabama each year (Mitchell and Tyson, 2001). Many factors affect and should be monitored during calibration including: ground speed, power take off (PTO) speed, discharge opening, and swath width. Mitchell and Tyson (2001) discussed three methods of calibration. The first method was just to apply the litter uniformly over a field of known size and can only be accomplished if the litter load weight is known. The next method utilized a tarp to cover a known portion or the ground and then making three equally spaced passes (equal to swath width) over the tarp. Finally, the material on the tarp was weighed, divided by the tarp area, and converted to an application rate. The last method included setting pans out in the field and again making three passes over the pans. Then the material in the pans were weighed and plotted to determine the material distribution and uniformity. The Virginia Cooperative Extension proposes very similar calibrating procedures as the ACES. One of the major differences is they propose using the tarp method to determine 20 uniformity and swath width rather than the pan method. To ensure the correct amount of litter is applied, do not change the spreader settings after calibration (Marsh et al. (2003)). Parish (2000) used three commercial fertilizer spreaders and two products to compare delivery rates calculated from using collection trays with delivery rates from spreader calibration tests. Pattern tests were conducted on all the spreaders and pans were set out to conform to the ASABE S341.3 standard. The spreaders were passed over the pans three times and then the application rate was determined by converting the mass in the collections pans to kg/ha. Then calibration of the delivery system was conducted for each spreader. The distribution mechanism was removed from each of the spreaders to allow the material to be caught in a bucket and the application rate was determined. Results indicated that half of the comparisons between the rates determined by pattern data and rates determined by calibration were statistically significantly different. In most cases, the rates from pattern data were higher. This is assumed to be due to the fact that the tests were conducted on a hard surface causing the particles to bounce into the collection pans. In conclusion, the study confirmed that significant spreader delivery rate errors can be generated from pattern tests when conducted on a smooth surface; however, errors may or may not occur on a rough surface. Parish (2000) suggests that rate calibration be conducted after an effective swath width is determined by pattern testing. 2.3 VARIABLE-RATE TECHNOLOGY Utilizing variable-rate technology (VRT) to more uniformly apply fertilizers can reduce over-application of nutrients by spatially applying the proper amount to meet local fertility needs. Studies have been performed to determine the affect of VRT on fertilizer 21 application (Fulton et al., 2001; Lawrence and Yule, 2005; Molin et al., 2002). However, no literature was found on using VRT for applying poultry litter. It is assumed that if VRT can improve the application of inorganic fertilizers then it can be utilized to improve litter application. Lawrence and Yule (2005) evaluated the different spreader testing protocols used throughout the world and the potential for the machine to perform VRT was assessed. A spreader truck with dual spinners operating at 750 rpm and urea application rates of 80, 100, and 150 kg/ha was tested. The protocols tested included two ISO standards, ASABE S341.3, European Standard, the ACCU-Spread, and the Spreadmark standard. A pan matrix of 1400 pans which represented 18 simultaneous tests was laid out to conduct the tests. The coefficient of variation (CV) was used to compare the different testing methods with a CV of 15% deeming an acceptable pattern. The conclusion made between the different testing methods was the only significant difference in calculating the maximum swath width of all the methods was with the ISO(i) and the Spreadmark methods. Also, single transverse tests did not fully represent the actual spread pattern, for this multiple tests need to be conducted. As for the potential for VRT, it was concluded that for variable-rate technology to be effective for spreading in the farming industry a greater understanding of the current spreading equipment performance would be required. Fulton et al. (2001) used a spreader truck with dual rear spinner discs equipped with VRT to test variable-rate (VR) distribution and uniformity. The ASABE standard 341.2 was followed for all aspects of the testing. Multiple tests were conducted to determine fixed-rate application as well as variable-rate application of potash at high and 22 low application rates. Analysis found that CVs over 20% were calculated for the average transverse patterns at high and low application rates. The uniform application at high and low rates was modeled from the average transversal spread patterns. It was determined that modeled application at the uniform rates and both rate changes predicted the actual application well. It was also noted that there was good uniformity at the low application rate but pattern changes occurred at the high application rate suggesting that modifications need to be made to the spreader to gain uniformity. A study was conducted to determine the effect of VRT when using a three-point hitch mounted fertilizer spinner spreader (Molin et al., 2002). Tests using urea were conducted in the transversal and longitudinal directions where pans were set out to meet the ISO 5960 standard. The CV was chosen to compare results of the transversal tests, 15% being acceptable. For the transversal distribution tests, a swath width of 24-m resulted in the best uniformity at all application rates (50, 150, and 250 kg/ha). For the longitudinal tests, two treatments were run changing the application rate on-the-go to look at the effect of the change in rate using the VRT. These tests concluded that the response time for the VRT was 3.1 seconds to an increasing step and 5.6 seconds for a decreasing step. Finally, the flow rates obtained during the tests were found to be lower than the desired flow rate. Lark and Wheeler (2003) tested two technologies, VRT of an input following a treatment map and yield monitoring to measure the crops response, to investigate the response of a combinable crop to an input. These technologies were used during fertilization and harvesting, respectively. The findings confirmed that local response functions can be estimated from designed fertilizer experiments harvested using 23 commercial yield mapping. This technology can help farmers decide if they need to apply or not apply fertilizers or other inputs to maximize economic returns for each field. Schueller and Wang (1994) described some of the different methods available for applying fertilizers. Typical methods used for these types of applications are Automatic Control (sensors sending feedback to a controller) and Temporally Separate Control (use of a prescription map). Global Positioning Systems (GPS) is often used for each of these methods and the accuracy of variable-rate application (VRA) depends on the accuracy of the GPS data utilized to map the fields (Chan et al., 2002). VR applicators are desired for this form of work and generally use some type of feed-forward control which allows the appropriate rate of fertilizer to be applied. Note the applicators tested were liquid applicators; however, with some modifications the same principles apply to dry applicators. The accuracy of these applicators depends on the immediate response to a command change which relates back to the time constant of the system. A simulation test was conducted on a desired spatially variable field and results indicated that the feed- forward control could reduce the error in application considerably. Without this pre- command technology, the use of spatially variable control will not reduce application errors (Schueller and Wang, 1994). 2.4 UNIFORMITY Various research attempts have been made to improve uniformity of fertilizer application using standard spinner disc spreaders (Smith et al., 2004; Kweon and Grift, 2006; and Hofstee, 1995). Most of the studies used manufactured fertilizers not poultry litter, but knowledge can be gained by evaluating the investigations on manufactured 24 fertilizers. Effects of wind speed and direction, along with other climatic parameters, on the uniformity of fertilizers have been examined. Parish (2002) determined that material flow increased while uniformity decreased. Also, many new technologies, such as feedback control, use of optical sensors, and Doppler velocity meters, have been tested to determine the parameters that most affect the spread pattern. Smith et al. (2004) researched the effect of wind speed and wind direction on lateral and transversal spread uniformity of ammonium nitrate and potash, as well as the overlap pattern uniformities of triple 13 fertilizer using a spinner spreader truck operating at a spinner speed of 640 rpm. Pans were set out for all tests and results are presented using the progressive method of distribution. For ammonium nitrate, the CVs generally increased as the swath width increased for all wind speeds and wind directions while the potash CVs remained constant. The average CVs for the potash and ammonium nitrate tests were 19.2% and 21.1%, respectively at their respective effective swath widths. For the 13-13-13 tests, uniformity under cross wind applications were more uniform than head wind applications. Finally it was determined that the amounts of N, phosphate, and potash, in the triple 13 blend, collected in the pans were proportionately distributed in relation to the total deposit. Researchers have worked toward providing improved distribution of granular materials over varying application rates without recalibration using automatic control systems; however, uniformity is not guaranteed (Kweon and Grift, 2006). Kweon and Grift (2006) utilized optical sensors, measuring velocity and particle size exiting a spinner spreader, to predict particle landing position and then send feedback to an algorithm which controls the feed gates of the material onto the spinners. Results 25 indicated that without the feed gate adaptation method unacceptable patterns were produced at the high application rates with acceptable patterns at the low rates. When the feed gate adaptation (VRA) method (use of optical sensor and control algorithm) was used to simulate spread patterns, acceptable patterns were produced for both spreaders at any application rate. The authors stated that feed gate control needs to be field tested with the optical sensor and control algorithm before applying to every day use. Hofstee (1995) studied the effect that physical properties of fertilizer have on uniformity when using spinner disc spreaders. An important factor that affects the behavior of the spread pattern is the motion of the particles on the vane surface with the most important physical property of the particles being the coefficient of friction on the vane surface. A simulation model and spreading tests were used to determine the influence of the particle parameters on the spread pattern. A pair of Doppler Velocity Meters was used to determine the velocity and direction of the particles that passed the measuring zone. Results indicated the effect of the coefficient of friction on motion of the fertilizer particles developed by the simulation model is not easily expressed through spreading tests. In conclusion, mass flow played the main role of the motion of the particles on the disc; however, it could not be modeled. After this finding it was suggested that when applying fertilizers on a site-specific basis, it would be better to vary ground speed rather than mass flow to maintain an optimum application rate. 2.5 DISTRIBUTION PATTERNS The distribution patterns of fertilizer products are dependent upon the type of spreader and hardware settings being utilized. These patterns also vary depending on the 26 type of fertilizer being spread. Researchers since the mid 1900?s have studied distribution patterns of various fertilizers, including poultry litter, based on the variables previously mentioned to determine the most efficient way to distribute fertilizer based on mass as well as nutrients. Parish (1999b) and (2003a) conducted tests to determine if ASABE S341.3 standard provided the best method to evaluate spinner spreader performance. Parish (2003a) compared pattern tests laid out by the ASABE standard to alternate tests where the applicator and collection devices were stationary, proposed by some spreader manufacturers. After analyzing both testing procedures, it was concluded that the stationary method was more difficult and yielded data inconsistent with the accepted ASABE S341.3 protocol. Parish (1999b) used multiple spreaders to test the theory laid out in the ASABE S341.3 that all tests should be conducted with the hopper filled and leveled at 40 to 50% capacity. When the hopper fill level was between 50 to 100%, there were no rate changes observed; however, when the fill level dropped below 50% highly significant changes occurred. The author concluded that hopper fill level can be a concern especially when the fill level drops to 10% capacity and that ASABE?s requirement of 40 to 50% is acceptable. In recent studies, some focus has been put toward the design of spinner discs on a spreader. Parish (2003b) theorized that impeller angle affected the distribution pattern of fertilizer by changing the material trajectory and drop location, ultimately changing the point where the particle leaves the impeller. In all cases as the impeller was angled upward patterns were skewed more to the right and as it was angled downward the patterns were skewed more to the left. The author suggests mounting a bubble level on 27 the hopper of the spreader in view of the operator to ensure that the spreader stays level during operation. Yildirim (2006) focused on determining if vane number on a single disc rotary spreader affected the distribution patterns, of multiple fertilizers, over a range of application rates. Six different vane numbers (2, 4, 6, 8, 10, and 12) were used with orifice diameters of 30, 40, and 50-mm to allow for three flow rates. The author stated that single disc spreaders typically have six vanes where traditional twin disc spreaders have two vanes per spinner disc. Vane height was pre-selected for each fertilizer based on the finding that vane height did significantly affect uniformity. Tests showed that while vane number increased the CV also increased for each flow rate. This study also determined that for each vane number the CVs increased with increasing flow rate. Results illustrated that the best distribution patterns for all fertilizers tested were generated with two vanes at the lowest flow rate. As it is stated previously the ASABE S341.3 outlines the acceptable method to test spinner spreaders; however, in some cases researchers believe there could be a better way to tests such applicators. Reed and Wacker (1970) studied the effect of indoor broadcast testing, justifying the process by eliminating environmental factors, such as wind. The tractor and spreader remained stationary while the collection pans were moved to simulate field operation. The authors concluded that indoor testing provided acceptable data and allowed for testing of many of the spreader variables. Kaplan and Chaplin (1997) proposed a new method to compare patterns generated by different sized collection devices concluding that it was acceptable compared to other methods. Parish (1999a) evaluated the effect of multiple passes over a single row of pans compared to a single pass believing that run-to-run variations could be averaged out with multiple 28 passes. Results indicated that data will be consistent when using either a single test or multiple tests for spreader pattern testing. 2.5.1 GRANULAR FERTILIZER TESTING Research on granule fertilizers is more common than organic fertilizer considering its usage from an agronomic standpoint. Many technologies have been proposed and utilized to assess their effect on the distribution patterns of these fertilizers. Fulton et al. (2005b) used VRA of muriate of potash with four applicators to differentiate the distribution patterns over different application rates. Two spinner spreaders and two pneumatic spreaders were utilized. Results indicated that the pneumatic applicators provided consistent patterns; however, the optimal swath width with the lowest CVs was produced by one of the spinner spreaders. Diallo et al. (2004) installed and tested an electronically actuated gate on a ground driven pull-type (buggy) spreader using granulated fertilizer. Results concluded the relationship between mass flow and gate opening were linear for all materials tested and CVs fewer than 15% were generated for swath widths 3-m or less. The author noted that a ground driven spreader cannot guarantee a stable spinner speed but still allowed for low CVs. Diallo et al. (2004) also compared truck spreader patterns to buggy spreader patterns and determined that they were similar when operated under comparable conditions. New technology always seems to be on the rise to improve distribution. Grift and Hofstee (2002) developed a sensor with the capabilities of collecting data from all angles around the rear of the spreader to allow for real-time prediction of the spread pattern. This sensor mounts on and rotates around the rear of the spreader. Results showed the sensor produced an excellent indication of fertilizer dispersion behind the 29 spreader and provided similar results on particle size diameter when compared to the hand measurements. It also demonstrated that the spread pattern was skewed to one side and found that the optimal swath width was lower than that of the manufacturer?s recommendations. The author?s stated that the sensor needs to be validated using the ASABE S341.3 testing standard. 2.5.2 POULTRY LITTER TESTING Pezzi and Rondelli (2002) conducted a study on a prototype manure spreader to more efficiently distribute poultry litter by increasing the swath width and making it easier to fertilize orchards. The litter was composed of different degrees of composting and MC?s. The spreader was pull type with an auger conveying system and tubular mixing mechanism in the hopper. Hydraulics were utilized to power the rear gate as well as the distribution system to allow for the drop location of material on the spinners to be adjusted. Vanes on the spinners were modified to extend approximately two inches past the spinner and a rear adjustable shield was put in place to allow for spreading in broadcast method or in bands. During testing, the spinner speeds, drop location, and rear shield were adjusted to determine the appropriate spreader settings. The CV was utilized to analyze the results with a CV of less than 30% considered acceptable. Results for broadcast distribution indicated high spinner speeds and drop locations away from center of the spinners were desired. This is ironic considering most manufacturers? recommend drop locations near the center of the spinners (Chandler Equipment Company, 2008). Based on the results the prototype spreader performed better when the percentage of large particles was the lowest (Pezzi and Rondelli, 2002). 30 Wilhoit et al. (1993) used a centrifugal-type broadcasting spreader to evaluate the distribution pattern of poultry litter based on weight along with studying the effect that particle size has on nutrient distribution across the swath. The spreader used had a ground driven chain with two spinners that ran off of the tractors hydraulic system and operated at 600 rpm. Litter samples within each pan were weighed and sieved, and then the samples were recombined and analyzed for nutrient content (N and C) at each location. A CV was used to assess uniformity. Bulk litter analysis illustrated that C concentrations increased with increased particle size and the N content within each size fraction varied randomly from 3.2 to 4.16%. The pan analysis found that both C and N concentrations were uniform across the swath indicating that even though variations occurred between particle size and nutrient content these variations did not affect nutrient uniformity. Other results illustrated that the smaller particles tended to land more directly behind the spreader (3.7-m to either side) with the larger particles being distributed farther out (6.1-m to either side). The manufacturers recommended swath width was not found to be the optimum width and results suggested that swath widths lower than the manufacturers recommendation may need to be utilized when applying poultry litter. 31 CHAPTER THREE SPINNER-DISC TECHNOLOGY TO ENHANCE POULTRY LITTER APPLICATION 3.1 ABSTRACT As technology advances for applicators, it is assumed that control and distribution of material should improve. A study was conducted to evaluate if spinner disc-control improves the distribution of poultry litter. A typical litter spreader equipped with an electronically adjustable hydraulic flow control (proportional) valve was used to test a closed-loop system (CLS), spinner-disc control, and compare these results to a traditional open-loop system (OLS), using a manual valve. Three application rates of 2242, 4483, 6725 kg/ha were selected for applying broiler litter. Litter was collected based on ASABE S341.3 testing protocol but modified to assess pattern uniformity using a two- dimensional pan matrix. Analyses included assessing variability and consistency of distribution patterns and spinner speeds between the two systems. The CLS was able to maintain more consistent spinner-disc speeds only allowing 1-6 rpm differences between the spinner-discs where the OLS allowed 1-12 rpm differences. The CLS also consistently provided smaller CVs (23% to 28%) over the range of application rates. However, pattern comparisons revealed no overall differences existed between the two systems. Based on the results, significant differences were found between the systems (p=0.0524) recommending the CLS when performing variable-rate application (VRA). 32 3.2 INTRODUCTION The use of poultry litter as a fertilizer and soil amendment has increased recently due to the increasing prices of inorganic fertilizers. However, poultry litter is inherently variable in its physical characteristics making it hard to maintain the desired uniform distribution using standard spinner-disc spreaders; the most common equipment used to land apply litter. Recent precision agriculture (PA) technology could have the potential to improve the distribution of poultry litter. Over the past several years, various research attempts have been made to gain a better understanding of variables that affect distribution of granulated fertilizer products using various types of spreaders including spinner-disc spreaders. Most of these studies focused on manufactured fertilizers, such as urea, potash, ammonium nitrate, etc., with few focusing on organic fertilizers, such as poultry litter. Numerous tests have been conducted to determine how the spread pattern and uniformity is affected by spreading variables such as material chemical and physical properties, weather conditions, machine parameters, and the use of variable-rate technology (VRT). Further, environmental concerns associated with applying poultry litter must also be considered. Therefore, the ability to maintain acceptable litter distribution is needed to minimize the over-application of litter in environmentally sensitive areas potentially leading to runoff issues. To maintain an acceptable distribution with poultry litter, it is believed that some control of the spinner-discs needs to be obtained during application. Many other spreading variables have been tested when using poultry litter, such as with the prototype spreader by Pezzi and Rondelli (2002), on-the-go feed gate control (Kweon and Grift, 2006), and variable ground speed rather than varying mass flow (Hofstee, 1995). 33 However, no research has been conducted on spinner speed control. By utilizing spinner speed control via a hydraulic flow control valve, the speeds of the spinners will remain constant rather than fluctuating as mass flow varies onto the discs allowing a more uniform and desirable application. In Alabama, approximately 90% of the litter generated annually is used for fertilizing crop and pasture lands. Due to this heavy usage over the years, there has been an accumulation of phosphorus (P) in the soil potentially causing harmful amounts of P to be deposited into surface waters via runoff. Utilizing VRT to more efficiently apply fertilizers can reduce over-application of litter by spatially applying the proper amount of litter to meet local fertility needs. Due to the natural variability of litter, concerns exist that the use of VRT with spinner-disc control may actually have a negative impact on litter application by increasing application errors. However, the use of VRT for litter application could improve the overall distribution of litter across a field providing an alternative to reduce environmental concerns while maximizing yields. 3.3 SUB-OBJECTIVES The objectives of this investigation were to: (1) characterize the distribution patterns of poultry litter and compare a closed-loop system (CLS) to an open-loop system (OLS) for spinner disc speed control over a range of application rates and (2) evaluate the accuracy of litter application under simulated field operation to determine if spinner speed control (CLS) improves the distribution of poultry litter. 34 3.4 MATERIALS AND METHODS A standard litter/shavings spreader manufactured by Chandler Equipment Company was used for this investigation (figure 3.1a). This pull-type spreader utilized hydraulically controlled dual rear spinner-discs and apron chain (figure 3.1b). The spreader utilizes 0.6-m diameter spinner discs with four uniformly spaced vanes. A John Deere 6420 tractor was used during testing to pull the spreader and was equipped with a John Deere GreenStarTM AutoTracTM system using real-time kinematic (RTK) correction (Appendix A). Topcon?s Zynx X20 computer/controller loaded with Topcon?s Spreader Control software program (Appendix D) was used and provided both VRA and spinner- disc speed control capabilities. The X20 uses inputs such as spreader variables (ex. gate height), product density, and swath width along with ground speed to maintain the desired application rate. Sensors were mounted under both spinner-discs (52 pulses/rev) and on the rear shaft of the apron chain (360 pulses/rev) to control and monitor the speeds for the Spreader Control program. A Visual Basic (VB) code was developed to log these speeds during testing (Appendix F) and was loaded onto a Zynx X15 computer (Appendix D). The VB program wrote all data to a .TXT file. (a) (b) Figure 3.1. Tractor, litter spreader, and illustration of collection pan matrix utilized during testing (a) and rearview image of spreader showing spinner-discs (b). 35 Hydraulic flow control for the apron chain and spinner-discs was maintained using Brand proportional valves with the Spreader Control program using speed sensor feedback (CLS) to maintain the desired speeds. The apron chain speed was controlled by a 57-L/min valve (Model No. EFC 12-15-12) while the spinner-disc speed was controlled by a 76-L/min proportional valve (Model No. EFC 12-20-12; figure 3.2). A standard Brand, manual valve (open-loop with no flow compensation) was used as the traditional hydraulic control system for the litter spreader (figure 3.2; Appendix C). The manual valve does not allow for feedback flow adjustment; therefore, the spinner disc speeds can fluctuate as mass flow onto them varies. Figure 3.2. Illustration of the two different flow control valves used during testing with the proportional valve shown on the left and manual valve on the right. Three application rates were selected: 2242, 4483, and 6725 kg/ha based on the recommended application range for poultry litter in Alabama. Therefore, a randomized complete block design was used with blocking based on the CLS and OLS with the three rates randomized within each block for a total of six tests. A block experimental design was needed since hydraulic hoses had to be disconnected between valves thereby minimizing potential oil spill. All tests were replicated three times with replications performed on different days. 36 Poultry litter was used for all tests. All tests were conducted to meet the requirements of the ASABE S431.3 testing standard (ASABE Standards, 2004). Prior to testing, both hardware and software calibration procedures were performed based on manufacturers? published literature. A single row of 35 pans, uniformly spaced at 0.8-m, was used during calibration. The pans were 50.8-cm long, 40.6-cm wide, and 10.2-cm tall. The pan on either side of the center pan was removed to allow the tractor and spreader to pass unobstructed. Calibration was conducted at the median application rate of 4483 kg/ha. Adjustments were made to the spreader until the most uniform application was accomplished. A swath width of 9.2-m was found to be the optimal, effective swath width for this spreader. 3.4.1 DATA COLLECTION Four rows of pans were used for each of the six tests using a two-dimensional collection pan matrix (figure 3.3). A flag was placed at the front center of each pan to ensure the pan was placed in the same location for every test. A 3.1-m longitudinal and 0.8-m transversal pan spacing was utilized requiring a total of 140 pans. (a) (b) Figure 3.3. Pan layout for single-pass test (a) and equipment traversing pans (b). 37 Prior to testing, the spreader was loaded to meet the ASABE Standards (2004). Bulk samples of litter were collected randomly out of each hopper load to determine the moisture content (MC) and bulk density (BD). Only one load was required to run the six tests of an individual replication. All bulk samples were sealed in a plastic bag and labeled. Also prior to testing, three samples were taken and an average BD was measured, by a known volume container and digital scale, for input into the Spreader Control program. A gate height of 35.6-cm and a desired spinner speed of 600 rpm were used for all tests based on the calibration results. The spinner speed was programmed into the Spreader Control program for the CLS tests. For the OLS tests, the manual valve was hand-set to a flow that corresponded to 600 rpm when the spinners were loaded at the median application rate. Once the manual valve was adjusted to the desired setting, it was not adjusted again to ensure consistency for all replications. However, prior to each OLS test, observations were made to ensure that the proper spinner speed was being maintained. This procedure was also used for the CLS tests. For each test, the target application rate was entered into the Spreader Control program and a nominal ground speed of 6.4 kph was maintained. The wind speed was measured for each test and it never exceeded the 2.2 m/s maximum allowable wind speed for any test (table 3.1; ASABE Standards, 2004). The applicator was turned ON far enough in advance to allow the spreader to be at normal operating conditions prior to traversing the pans. The John Deere AutoTracTM guidance system was used during all tests to ensure the same path, over the center pans, was driven to minimize operator error. 38 Table 3.1. Wind Speeds observed during pattern testing. System Rate Rep 1 Rep 2 Rep 3 (kg/ha) (m/s) (m/s) (m/s) 2242 1.61 0.27 0.00 Closed-Loop 4483 0.89 1.21 0.00 6725 1.34 0.45 0.00 2242 0.67 1.25 0.00 Open-Loop 4483 0.89 0.89 1.12 6725 0.80 1.12 0.89 3.4.2 DATA ANALYSIS The material collected in each pan for each test was placed in a plastic bag, sealed, and labeled accordingly for laboratory analyses. A digital scale that measured to the nearest 0.01 g was used to weigh each sample. A mean mass measurement along with the standard deviation of the four rows at each transverse position was calculated, converted to an application rate, and used to generate the single-pass distribution pattern for each test. The 95% confidence interval (CI) was calculated for each transversal position and plotted to show the confidence around the mean. Microsoft Excel was used to summarize this data. Simulated overlap patterns, based on the progressive method (ASABE Standard, 2004), were also created from the single-pass patterns using the effective swath width. The mean, standard deviation, and coefficient of variation (CV) was computed and used to assess application uniformity. The mean and standard deviation was calculated by using every point in the overlap pattern. The CV was used to analyze the uniformity of the overlap patterns and was calculated by dividing the standard deviation by the mean. Perfect uniformity is defined by the overlap pattern resulting in a straight line when plotted or a CV = 0; indicating the same amount of 39 material was applied across the swath. The single-pass patterns were then standardized based on the mean application rate computed using the overlap data to evaluate pattern similarities or shifts. The standardized patterns were determined by dividing each position in the spread pattern by the mean calculated rate from the simulated overlap patterns. For example, each position for the CLS low rate was divided by 2419 kg/ha (table 3.5) To assess the true nature of the spread pattern, the mean patterns for the three replications were averaged to generate overall mean patterns for the three different rates (2242, 4483, 6725 kg/ha). In the end there were three patterns for the CLS and the OLS. Using the Statistical Analysis System (SAS Inst., NC), an analysis of variance (ANOVA) was conducted using the General Linear Model (GLM) to determine if statistical differences existed between the two different control systems based on the mean overlap patterns for each rate. The ANOVA compared the means between the systems, days (reps), and the rates as well as determined if an interaction occurred between the systems and rates (system by rate) and between the systems and days (system by day) (table 3.6). The Least Squares Means (LSM) procedure in SAS was used in conjunction with the ANOVA to confirm the possible significance of the overlap means. Pearson?s Correlation Coefficients were calculated using SAS?s CORR procedure to evaluate similarities or shifts between single-pass distribution patterns for each control system. An alpha value of 0.10 was used for statistical comparisons. 3.5 RESULTS AND DISCUSSION The average MC and BD for the litter applied in each test was 25.8% and 542.5 kg/m3, respectively (table 3.2). The litter used for calibration and replications 1 and 2 40 came from the same location even though there were slight differences in the litter from reps 1 and 2 according to the MC and BD data. These differences illustrate the natural variability of litter and how its physical characteristics can differ even when from the same farm. A different load of litter with similar physical characteristics was used for replication 3. When comparing the litter between replications, one can see that the BD from rep 3 was similar to that of rep 1 and rep 2 with only the MC being slightly less. Table 3.2. Poultry litter MC and BD for each replication. Replication Moisture Content Bulk Density (%) (kg/m3) 1 26.7 538.9 2 27.1 550.6 3 23.7 538.2 Average 25.8 542.5 3.5.1 SPINNER SPEED ANALYSIS During reps 1 and 2, an unusual characteristic was noticed when dealing with the spinner speeds. The hydraulic motors (Appendix B) that power the spinner discs were connected in series meaning the hydraulic fluid flows from the tank to the left motor (spinner 1), then to the right motor (spinner 2), and back to the tank. With the spinner discs operating with no load, they worked as expected with spinner 1 running slightly faster that spinner 2. However, once litter was conveyed onto the discs, spinner 2 was rotating faster than spinner 1 (Appendix G). In theory, this scenario is impossible considering the motors have equivalent displacement since they are matching gear motors. Unloaded spinner disc speeds were collected prior to each test for rep 3 (table 3.3). In all cases, spinner 1 ran faster than spinner 2 verifying the spinners were operating properly before being loaded, but this did not provide any explanation of what 41 was observed during the tests. Another item to note is that the measured OLS pre-test speeds were approximately 50 rpm faster than the desired speed of 600 rpm. This higher speed was because the manual valve was set to run at the desired speed when the spinners were loaded at the middle application rate (2242 kg/ha). For the CLS, the spinner discs should operate at the desired speed since the proportional valve adjusts the hydraulic flow to maintain the desired speed whether loaded or unloaded. Table 3.3. Summary of pre-test spinner-disc speed data. Spinner 1 Spinner 2 Target Mean Std. Dev. Diff. 1 1 Mean Std. Dev. Diff. 1 1 Diff. 2 2 System (kg/ha) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) 2242 601.1 3.9 1.1 595.5 4.2 -4.5 -5.5 4483 603.2 1.8 3.2 598.7 1.7 -1.3 -4.4 Closed-loop 6725 601.3 0.8 1.3 596.1 0.6 -3.9 -5.3 2242 652.9 1.0 52.9 648.9 0.8 48.9 -4.0 4483 655.7 4.2 55.7 654.5 1.1 54.5 -1.3 Open-loop 6725 654.6 0.9 54.6 649.4 0.9 49.4 -5.3 1) Diff. 1 is the difference of the spinners from the desired speed of 600 rpm; positive and negative indicating when the actual speed is greater than or less than the desired speed, respectively. 2) Diff. 2 is the difference between spinners 1 and 2; negative indicating spinner 1 is faster. Table 3.4 summarizes the speed results for the two different control systems over the three rates. The CLS was able to maintain spinner disc speeds near the desired 600 rpm (Diff. 1 column in table 3.2). Considering the magnitude difference between the three application rates, the CLS performed consistently with spinner 1 operating between 598 and 599 rpm and spinner 2 between 600 and 605 rpm. These are small differences when compared to the OLS results which varied more as shown by the higher standard deviations and speed differences (Diff. 2) between spinner 1 and 2. The CLS system also had smaller speed differences between the two spinner discs; 1 to 6 rpm. Of note, these 42 values were the mean speed data over all replications. The speed differences between the spinners seemed to vary between replications, for both systems. However, the CLS provided more consistent spinner disc speeds over all the tests, the only exception coming during rep 2 where the OLS provided differences that were less than the CLS, from 1 to 5 rpm smaller (Appendix G). The reason behind this difference is unknown. For rep 2, spinner 2 operated uncharacteristically high, anywhere from 7 to 15 rpm, for all tests; however, this was not the case for reps 1 and 3. Table 3.4. Summary of spinner-disc speed data computed for all replications. Spinner 1 Spinner 2 Target Mean Std. Dev. Diff. 1 1 Mean Std. Dev. Diff. 1 1 Diff. 2 2 System (kg/ha) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) 2242 599.5 6.4 -0.5 600.2 8.4 0.2 0.7 4483 597.8 11.5 -2.2 604.1 13.5 4.1 6.4 Closed-loop 6725 598.7 9.6 -1.3 604.9 15.3 4.9 6.2 2242 630.4 15.2 30.4 631.5 14.6 31.5 1.1 4483 602.1 11.8 2.1 613.6 13.7 13.6 11.5 Open-loop 6725 598.3 25.5 -1.7 610.0 25.9 10.0 11.6 1) Diff. 1 is the difference of the spinners from the desired speed of 600 rpm; positive and negative indicating when the actual speed is greater than or less than the desired speed, respectively. 2) Diff. 2 is the difference between spinners 1 and 2; negative indicating spinner 1 is faster. The main point about the spinner speed data is that the CLS was able to maintain the desired speed and a smaller differential speed between the spinner-discs when compared to the OLS. Initially, it was assumed the OLS would not be able to maintain the desired speed over a range of application rates without needing to reset the valve for each rate. The question then became, could the OLS maintain a constant spinner speed between the two spinner-discs better than the CLS, and it did not. These results indicated how an OLS was unable to maintain the desired and constant spinner speed allowing the 43 speeds to vary as the load on the spinner-discs changed. However, no inferences can be made about if a CLS can improve the distribution of litter based solely on this data. 3.5.2 SINGLE-PASS ANALYSIS The single-pass patterns for the closed- and open-loop systems are illustrated in figures 3.4a and 3.4b, respectively. A symmetrical pattern is desirable for spinner-disc spreaders since they rely on overlap from the adjacent passes. Thus, a symmetrical pattern will provide a more uniform distribution of material. When comparing the patterns of the two systems, the CLS patterns were more symmetrical and exhibited less variation along the center portion of the spread width (between -5 and 5 m) than those for the OLS. The patterns produced by both systems were ?W? shaped. This type of pattern is usually undesirable due to its tendency to negatively affect the overlap pattern by generating non-uniform material distribution. This undesired shape is a prominent characteristic to note when trying to gain uniformity considering the center portion of the swath is where a majority of the material is applied, significantly impacting overlap patterns. Typically, hardware adjustments can be made to minimize or eliminate the ?W? shape. Through extensive calibration of this spreader, no additional hardware adjustments were possible to minimize the final pattern shapes. However, Chandler has made adjustments to the hardware on the newer spreaders to minimize these errors. The single-pass patterns generated by the CLS were more desirable than the OLS patterns. Nevertheless, the CLS minimized the magnitude of the ?W? shape for the two higher application rates. In comparison, the intensity of the ?W? shape pattern increased with application rate for both systems which is typical when applying granular material. Another point of interest is the center peaks of the higher rates. The amount of material 44 applied at these positions was much higher (approximately 2200 kg/ha) than the intended rate. The exact explanation behind this result is unknown. However, considering the over-application did not occur at the lower rates, it is possible that the spinners were overloaded with material bypassing the spinner disc and being deposited directly to the ground. Over-application is not desirable since it increases environmental risks following field application. 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 6725 kg/ha 4483 kg/ha 2242 kg/ha 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 6725 kg/ha 4483 kg/ha 2242 kg/ha (a) Closed-loop system (b) Open-loop system Figure 3.4. Overall mean single-pass distribution patterns for the two control systems. The 95% confidence intervals (CI) were computed for each mean pattern to illustrate their variation. For the 2242 kg/ha rate, the CLS had a wider CI (figure 3.5e). This variation was also the case with the high application rate (6725 kg/ha; figure 3.5a); however, with the middle rate (4483 kg/ha) the OLS displayed the larger CI (figure 3.5d). In all cases variability decreased as the patterns extend to their outer limits and increased toward the center of the swath between ? 4.6 m. The greater variations resulted from larger amounts of material being applied at these locations as well as the uncharacteristic distribution of particle sizes (figures 3.5a and 3.5d). Normally, when applying poultry litter the larger particles are deposited on the pattern outer boundaries with the smaller 45 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 6725 kg/ha 95% CI 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 6725 kg/ha 95% CI (a) (b) 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 4483 kg/ha 95% CI 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 4483 kg/ha 95% CI (c) (d) 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 2242 kg/ha 95% CI 0 2000 4000 6000 8000 10000 12000 -15 -10 -5 0 5 10 15 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 2242 kg/ha 95% CI (e) (f) Closed-loop system Open-loop system Figure 3.5. Mean single-pass distribution patterns and 95% CI. 46 particles toward the center. With the CLS 6725 kg/ha pattern (figure 3.5a) and the OLS 4483 kg/ha pattern (figure 3.5d), large clumps of litter were deposited at the 1.5 m location in one of the tests resulting in a large CI around that point. This anomaly helps clarify that litter variability is difficult to control when spreading. In summary, the 95% confidence intervals revealed there was little difference in terms of variability about the mean pattern between the two systems. 3.5.3 OVERLAP-ANALYSIS Table 3.5 presents the overall summary data for the simulated overlap patterns. Figure 3.3a illustrates the multiple passes needed to model the overlap patterns. In previous research conducted with inorganic fertilizers, a CV of 20% or less is considered acceptable. There was no literature found specifying an acceptable CV when applying poultry litter. All CVs were greater than 20%. Considering the inherent variability of litter compared to inorganic fertilizers, one would expect measured CVs to be higher thereby CVs up to 25% may be acceptable. Five out of the six tests produced CVs less than 30% with the middle rate for the CLS at 23%. This rate by far provided the best CV with the possible reason that the applicator was calibrated at this level. One would gather that this lower CV would also be the case for the OLS at the middle rate due to the same reason but it was not. For the OLS, the CVs increased going from the low to high rate. None of the CLS CVs averaged higher than 28% and none of the OLS CVs averaged lower than 28% (table 3.5). While the low rate treatment CVs differed by only 1.6% (26.6% vs. 28.2%), this difference increased with the application rate to 7.0% for the middle rate and 10.8% for the high rate. Further, the OLS CVs ranged from 28% to 38% with the lowest CV (28.2%) occurring at the low rate and the higher two rates being 47 equal to or greater than 30%. In contrast, the CVs for the CLS ranged between 26.6% and 27.6% being more consistent over the application rates tested. Therefore, the CLS provided improved uniformity over the OLS. Comparing how close the mean rates were to the target rates showed that the CLS deviated more than the OLS for every rate. The CLS at the 6725 kg/ha rate over-applied by 739 kg/ha while the OLS under-applied for every rate (negative value in the Rate Diff. column). The rate differences tended to follow the same trend as the CVs with the lowest CV occurring at the CLS middle rate, the rate used for calibration, indicating improved spreader performance at the lower rates. Table 3.5. Summary statistics for overall simulated overlap patterns. Target Mean 1 Std. Dev. CV Rate Diff. 2 System (kg/ha) (kg/ha) (kg/ha) (%) (kg/ha) 2242 2419d 644 26.6 177 4483 4383c 1009 23.0 -100 Closed-loop 6725 7463a 2060 27.6 739 2242 2080d 587 28.2 -162 4483 4417c 1326 30.0 -66 Open-loop 6725 6500b 2494 38.4 -224 1) Mean rates with similar letters indicate they are not statistically different at the 95% confidence level. 2) Positive and negative rate differences indicate over- and under-application, respectively. Based on the LSM and ANOVA results, the mean application rates within each system were significantly different from one another as expected since a range of very different rates were chosen for the experiment (table 3.5 and 3.6). The letters next to the mean values in table 3.5 indicate whether there is a significant difference or not between the means. When comparing between systems, a significant difference was found only for the high rate. Differences between these control systems at the high rate are 48 illustrated in the overlap patterns (figure 3.6). A p-value of 0.0524 was attained from the ANOVA for the systems illustrating there is a significant difference in the two control systems. Overall there was no statistical interaction between the systems and the rates denoted by a p-value of 0.1478 (P>0.10). However, figure 3.6 displays that a slight interaction did occur but still demonstrates that the systems responded in a similar fashion when applying the three different rates. The interaction could have been caused by the fact that the systems were calibrated at the middle rate. There was a significant difference in the Day (replication) response (p = 0.0617) indicating the need to block the replications into different days. Even though a difference was found for Day there was no ?System by Day? interaction identifying that the systems performed similar in nature. Table 3.6. ANOVA results for the overlap pattern data. Source DF Sum of Squares Mean Square P-value System 1 804,347 804,347 0.0524 Day (Rep) 2 1,250,557 625,278 0.0617 Rate 2 67,373,417 33,686,709 <.0001 System by Rate 2 761,620 380,810 0.1478 System by Day 2 53,669 26,834 0.8444 0 1000 2000 3000 4000 5000 6000 7000 8000 0 2000 4000 6000 8000 Desired Rate (kg/ha) Ap pli ed R ate (k g/h a) CLS OLS Figure 3.6. Illustration of System by Rate interaction. 49 Figure 3.7a and 3.7b depict the overlap patterns of the CLS and OLS, respectively. One can observe that the CLS produced more uniform spread patterns and that the overlap patterns followed the same ?W? shaped pattern as the single pass patterns. For the CLS, the middle application rate again showed the least variation around the actual mean with both sets of patterns increasing in magnitude as the application rate increased. This proved that with the swath width of 9.2 m the single pass patterns directly affected the uniformity when using the progressive method of application. For the overall average patterns, the most uniform pattern illustrated by the lowest CV (23%) was the 4483 kg/ha rate for the CLS. The pattern with the worst uniformity was the 6725 kg/ha rate for the OLS (figure 3.7b). 0 2000 4000 6000 8000 10000 12000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 6725 kg/ha 4483 kg/ha 2242 kg/ha Desired 0 2000 4000 6000 8000 10000 12000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) 6725 kg/ha 4483 kg/ha 2242 kg/ha Desired (a) Closed-loop system (b) Open-loop system Figure 3.7. Overall simulated overlap distribution patterns. A horizontal line represents perfect uniformity (CV = 0) and is desired when analyzing overlap patterns. To obtain this type of scenario with this data the high rate values need to be shaved off to fill in the low areas for each pattern, considering the mean application rates are, for the most part, close to the desired rates. This relocation of material across the swath could require hardware adjustments on the spreader; such as, gate height, spinner speed, and divider location or a change in effective swath width. It is 50 important to note that results of this investigation highlight the need for spinner speed control when implementing variable-rate application (VRA) of litter. As application rates change under a VRA scenario, spinner speed control can maintain the desired spinner speed as flow rate onto the disc varies, resulting in a more uniform distribution of material. Overall the CLS provided an improved distribution over the OLS by generating smaller CVs for every application rate. This uniformity improvement was directly related to the CLS ability to maintain constant spinner disc speeds, via the proportional valve, of approximately 600 rpm rather than the speed variation allowed by the OLS. The OLS did provide less variation in actual rate compared to the desired rate and overall a significant difference was found between the two systems. If VRA is to be utilized then the CLS is recommended. 3.5.4 DISTRIBUTION PATTERN COMPARISON BETWEEN SYSTEMS To be able to effectively compare the distribution patterns of each system at the different application rates, standardized patterns were analyzed. Figures 3.8a and 3.8b illustrate the standardized patterns plotted for the CLS and OLS, respectively. Both systems generated consistent patterns over the rates tested with slight pattern shifts taking place in both systems. The standardized patterns of the CLS showed some similarities and differences between the different rates. The 4483 and 2242 kg/ha patterns were consistent with one another with only a slight difference on the left peak. The 6725 kg/ha pattern illustrates a few more differences the most noticeable occurring with the center peak. This issue was brought about earlier with the single-pass analysis with the possible cause being that the spinner discs were being loaded to heavy too maintain speed. The 51 OLS standardized patterns also demonstrated similarities and differences. The left and right peaks for each rate seem to be fairly consistent with one another however some minor shifts occurred toward the center of the spread patterns. Again the most prominent difference with these patterns occurs at the center peaks. This is a trend with both systems and is directly related to the large amounts of material being deposited at the center location in return creating greater variability. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 -15 -10 -5 0 5 10 15 Transverse Distance (m) Sta nd ar diz ed D ist rib uti on Pa tte rn 6725 kg/ha 4483 kg/ha 2242 kg/ha 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 -15 -10 -5 0 5 10 15 Transverse Distance (m) Sta nd ar diz ed D ist rib uti on Pa tte rn 6725 kg/ha 4483 kg/ha 2242 kg/ha (a) Closed-loop system (b) Open-loop system Figure 3.8. Overall standardized distribution patterns. It was difficult to determine which system provided the most consistent patterns, so a correlation was conducted in SAS using the Pearson correlation procedure. Table 3.7 displays the results of the Pearson correlation. The low, mid, and high labeling represent the 2242, 4483, and 6725 kg/ha rates, respectively while the P and M are representative of the proportional and manual valves which are utilized in the CLS and OLS, respectively. The values under the correlation values (<.0001) in table 3.7 are the probability levels associated with each correlation. The results indicated the two control systems performed similarly over the range of tested application rates. The 2242 kg/ha rates of the two systems were highly correlated with a correlation value of 0.9851 with the 4483 kg/ha rates also being highly correlated (r = 0.9746). Even though the LSM 52 results indicated that the means for the 6725 kg/ha rates were significantly different, the Pearson correlation showed the patterns were correlated (r = 0.9578). Table 3.7. Pearson Correlation coefficients comparing the mean single-pass distribution patterns of each system. Low P Mid P High P Low M Mid M High M Low P 1.0000 0.9906 0.9466 0.9851 0.9714 0.9461 <.0001 <.0001 <.0001 <.0001 <.0001 Mid P 1.0000 0.9448 0.9938 0.9746 0.9548 <.0001 <.0001 <.0001 <.0001 High P 1.0000 0.9223 0.9879 0.9578 <.0001 <.0001 <.0001 Low M 1.0000 0.9620 0.9411 <.0001 <.0001 Mid M 1.0000 0.9790 <.0001 High M 1.0000 3.6 SUMMARY The distribution patterns of two spinner-disc speed control systems, CLS and OLS, were characterized and statistically analyzed to determine if the CLS provided an improved distribution over the OLS. The patterns from both systems were ?W? shaped with the OLS patterns having a more defined shape exhibiting more variation. Spinner- disc speed analysis concluded that the CLS allowed for less spinner speed variation between the two spinners, generated consistent speeds, and maintained the desired speed of 600 rpm better than the OLS. Pattern comparison results determined that no overall interaction was found between the systems and the rates indicating the systems responded in a similar fashion over the range of rates; however, differences were found between the 53 6725 kg/ha rates. Statistically, the systems performed different based upon the low p- value of the ANOVA results (p = 0.0524). Pattern correlations concluded that the CLS and OLS patterns were all highly correlated. Simulated overlap analysis was carried out to evaluate the uniformity of each system. The 4483 kg/ha application rate for the CLS provided the most uniform pattern with a CV of 23% with the worst CV of 38% coming from the 6725 kg/ha pattern of the OLS. The overall CVs for the CLS patterns did not exceed 28% and the OLS CVs never made it below 28% concluding that the closed-loop system provided an improved distribution over the open-loop system. The speed variations negatively affected the distribution of litter highlighting the importance of maintaining constant spinner speeds when making rate changes or as flow rate changes due to litter physical variability. The CLS is recommended over the OLS based on practical and statistical significance and due to the improved CVs for the CLS. Spinner speed control with the CLS is suggested for VRA use and standard use. This type of control would be beneficial even if VRT is not available especially if farmers are applying litter on multiple fields that require different application rates. 54 CHAPTER FOUR EVALUATING THE APPLICATION OF POULTRY LITTER ON A NUTRIENT BASIS 4.1 ABSTRACT Poultry litter is typically difficult to uniformly apply because of its natural variability in moisture content, particle size, and density. Nutrient concentrations tend to increase with decreasing particle size initializing a concern regarding spreading litter based on its nutrient content rather than mass. A study was conducted to determine nutrient distribution (N, P2O5, and K2O) in relation to the traditional mass distribution while comparing a closed-loop system (CLS), spinner-disc control, to an open-loop system (OLS). Three application rates of 2242, 4483, 6725 kg/ha were selected for applying broiler litter. Litter was collected based on ASABE S341.3 testing protocol but modified to assess pattern uniformity using a two-dimensional pan matrix. The CLS provided more uniform nutrient patterns with coefficient of variations (CVs) ranging from 22% to 34% compared to 26% to 39% generated by the OLS. Based on pattern comparisons the nutrient patterns were highly correlated with their respective mass pattern indicating that even though particle size variability exists across the width of spread, the distribution of nutrients reflects mass distribution. Practical and statistical differences were found between the two systems (p = 0.0657) concluding the CLS provided CV improvements up to 17% and it is thereby recommended over the OLS. 55 4.2 INTRODUCTION Poultry litter has become a reliable source of fertilizer in many parts of the country due to its nutrient value. On average, poultry (broiler) litter has a 3-3-2 (nitrogen (N)-phosphorus (P)-potassium (K)) fertilizer rating (Wood, 1992). The litter utilized during this research had a fertilizer rating or 3-6-5. One of the key goals when applying fertilizer is to apply the proper amount to meet crop or forage requirements. In certain parts of the Southeast where poultry production is dense, litter is used as the sole fertilizer to meet all nutrient requirements. This use of litter as fertilizer has produced environmental issues in some areas due to the over-application of P to the soil. In the past, poultry litter has been applied to meet the N requirement leading to the over- application of P considering the similar concentrations of N and P in the litter and the minimal P requirement for crops. Armstrong et al. (2006) stated that application of litter to meet N requirements causes an over-application of P, causing water quality issues ultimately creating a potential to harm aquatic life. Thus, many states are now basing litter application on P to meet environmental regulations. The environmental issue associated with over-application of P is called eutrophication. When excessive P is applied to the landscape, the soil is not able to retain all of it; therefore, in some cases harmful amounts of P is deposited into nearby surface water via runoff. The P initializes the rapid growth of algae in the water, often known as an algae bloom. Microorganisms in the water feed on the dead algae, taking in large amounts of oxygen depleting the water of the required amount of oxygen for aquatic life to survive. The Phosphorus index (P index), a tool to assess P movement across the landscape, is used to help eliminate some of the hazards associated with P application. 56 Particle size tends to affect nutrient management in poultry litter. Koon et al. (1992) determined the nutrient fraction for each of the macronutrients (N, P, and K), increased as the particle size fraction decreased. However, Wilhoit et al. (1993) reported that carbon (C) concentrations increased with increased particle size and nitrogen (N) content within each size fraction varied randomly. They also determined the variation in the litter moisture content (MC) and the number of flocks raised on the litter can play an important role in attaining a uniform application when it comes to mass as well as nutrient content. Only a few research attempts have been made to determine the spread pattern of nutrients in conjunction with the mass pattern of poultry litter. Wilhoit et al. (1993) measured N and C concentrations but not P and K. To simultaneously manage litter nutrients and protect the environment, it is of importance to know how macronutrients are being distributed. Many researchers have studied the effect of mass distribution with such granular fertilizers as ammonium nitrate, potash, urea, etc (Fulton et al., 2005b; Diallo et al., 2004; Grift and Hofstee, 2002). For many granulated fertilizers, nutrients are equally distributed with mass; however, poultry litter nutrient uniformity can be affected by variation in particle size. When applying litter with a spinner disc spreader it is common that the larger particles are distributed more towards the outer boundaries of the swath with the smaller particles applied at the center. Therefore, it is important to know how the nutrient patterns correlate with the mass patterns to determine if adjustments need to be made to more uniformly apply litter based on nutrient content. 57 4.3 SUB-OBJECTIVES The objectives of this investigation were to: (1) compare and contrast characterized nutrient and mass distribution patterns of poultry litter for both an open- loop system (OLS) and closed-loop system (CLS), (2) evaluate the uniformity of nutrient application to determine if spinner disc speed control (CLS) improves the nutrient distribution of poultry litter, and (3) determine particle size distribution across the swath width along with its impact on nutrient application. 4.4 METHODOLOGY A Chandler litter/shavings spreader equipped with hydraulically controlled dual rear spinner discs and apron chain was utilized for this investigation. Two spinner disc speed control systems, CLS and OLS, were used to conduct three tests each for three replications. Topcon?s Zynx X20 computer/controller loaded with its spreader control program provided spinner speed control for the CLS. Four rows of 35 pans (140 total pans) were used to collect litter samples at three application rates (2242, 4483, and 6725 kg/ha) for the two systems. Rather than analyzing all 140 samples for each test, the samples were combined at each transversal location after weighing was completed (figure 4.1). All samples were analyzed for N, P, and K. One load of litter was required to complete each replication. Three bulk samples were collected from each load of litter and analyzed for moisture content (MC), bulk density (BD), and nutrients. Twenty sub-samples were taken from the bulk samples of each replication totaling sixty sub-samples. For each replication, ten samples were analyzed for nutrients (N, P, and K). The other ten samples were sieved with each 58 particle size class analyzed for nutrients. The sieving process was conducted to determine the nutrient concentration of each particle size class and used to determine the particle size distribution across the swath width. Figure 4.1. The red rectangle illustrates how longitudinal pans were combined prior to the nutrient analysis procedures. 4.4.1 SAMPLE PREPARATION FOR NUTRIENT ANALYSIS All litter samples were refrigerated prior to preparing them for nutrient analysis. After the samples were combined at each transversal location, they were oven dried for 48 hours. Following the drying process, all samples were ground pass a 1-mm sieve for performing the nutrient analysis with the exception of all the 4483 kg/ha treatment samples (6 or 1/3 of all tests conducted) which were initially analyzed for particle size and then nutrients. Samples were segregated into four particle size classes using a standard mechanical sieving method. The sieve sizes utilized were a No. 4 (4.75-mm), No. 10 (2- mm), and a No. 60 (0.25-mm). The particle size ranges produced were: x>4.75 (retained on the No. 4 sieve), 2.00 5 BUFFER_ARRAY = BUFFER_LEFTOVER & MSComm1.Input 'ADD NEW MESSAGE TO BUFFER START_POS = InStr(BUFFER_ARRAY, "%") 'DEFINE START OF MESSAGE STRING AT CHANNEL 1 END_POS = START_POS + 7 'InStr(BUFFER_ARRAY, "!") ' DEFINE END OF MESSAGE STRING BUFFER_LENGTH = END_POS - START_POS 'Len(BUFFER_ARRAY) 'BUFFERLENGTH SET EQUAL TO LENGTH OF MESSAGE If START_POS = 0 Then Exit Sub If END_POS = 0 Then Exit Sub 112 'DATA_STRING = BUFFER_ARRAY DATA_STRING = Mid(BUFFER_ARRAY, START_POS + 1, END_POS - START_POS + 1) 'SPLIT DATA STRING DATA_ARRAY = Split(DATA_STRING, ",") CounterID = DATA_ARRAY(0) Frequency = DATA_ARRAY(1) COUNTER(CounterID).Text = Frequency Spinner1.Text = COUNTER(1) * 1.2 Spinner2.Text = COUNTER(2) * 1.2 Chain.Text = COUNTER(3) / 6 'PLACE CHAIN SPEED CONVERSION VALUE HERE GSRSpeed.Text = COUNTER(4) * 0.0177 'PLACE GSR VALUE HERE FOR CONVERSION 'If (LogData.Value = 1) And (OPENFILE = True) Then ' Write #1, count1.Text, count2.Text, count3.Text, count4.Text, count5.Text, count6.Text 'End If BUFFER_LEFTOVER = Right(BUFFER_ARRAY, END_POS + 1) 'COLLECT UNUSED BUFFER If (LogData.Value = 1) And (OPENFILE = True) Then Write #1, Spinner1.Text, Spinner2.Text, Chain.Text, GSRSpeed.Text End If Wend End Sub Private Sub ChooseFilename_Click() On Error Resume Next CommonDialog1.DialogTitle = "CHOOSE DATA FILENAME" CommonDialog1.ShowOpen FilenameDisplay.Text = CommonDialog1.FileName Open CommonDialog1.FileName For Append As #1 Print #1, " SPINNER1, SPINNER2, CHAIN, GSRSPEED" 'Prints labels at the top of text file OPENFILE = True LogData.Enabled = True End Sub Private Sub OpenPort_Click() MSComm1.PortOpen = True End Sub 113 F.2 PROGRAM TO COLLECT SPEEDS AND PRESSURES Public OPENFILE As Boolean Dim Ulstat As Long Dim BoardName As String Dim BoardNum As Integer Public org_time As Long Dim TempBoard As String Dim TempNum As Integer Dim filelocation As String Private Declare Function GetTickCount Lib "Kernel32" () As Long Dim Con As Integer Private Sub Command1_Click() End End Sub Private Sub Command2_Click() On Error Resume Next 'Student = InputBox("Please enter your name.") 'des = InputBox("Please enter the testdescription.") CommonDialog1.DialogTitle = "Choose Data FILENAME" CommonDialog1.ShowOpen FilenameDisplay.Text = CommonDialog1.FileName Open CommonDialog1.FileName & ".txt" For Append As #1 Print #1, "Student Name.", Student Print #1, "Test Description.", des Print #1, "Date", Date Print #1, "Time, SPINNER1, SPINNER2, CHAIN, GSRSPEED, Pressure, Pressure1, Pressure2, Pressure3 " 'Print #2, "Student Name.", Student 'Print #2, "Test Description.", des 'Print #2, "Date", Date 'Print #2, "Time, Average" OPENFILE = True LogData.Enabled = True End Sub Private Sub cmdStartConvert_Click() cmdStartConvert.Visible = False cmdStopConvert.Visible = True cmdStopConvert.Default = True tmrConvert.Enabled = True End Sub Private Sub cmdStopConvert_Click() tmrConvert.Enabled = False End End Sub Private Sub Form_Load() MSComm1.PortOpen = True CommonDialog1.InitDir = 114 "C:\DocumentsandSettings\Christian\Desktop\Ajay\Data" CommonDialog1.DefaultExt = ".txt" LogData.Enabled = True org_time = GetTickCount BoardNum = 0 '<======this is the default board number 'change it to what IstaCal has assigned for your USB/PMD-1608FS BoardName = " " Ulstat = cbGetBoardName(BoardNum, BoardName) Myboard = BoardName Myboard = Trim$(Myboard) bdlen = Len(Myboard) - 1 Myboard = Left(Myboard, bdlen) If (Myboard <> "PMD-1608FS") And (Myboard <> "USB-1608FS") Then MyMessage = "A USB/PMD-1608FS was not assigned to Board " & BoardNum & " in InstaCal." & Chr$(13) _ & "Please run InstaCal to verify the board number" & Chr$(13) _ & "and/or change BoardNum = " & BoardNum & " in the Form_Load event" & Chr$(13) _ & " to the correct board number. Then re-run this program." r = MsgBox(MyMessage, vbExclamation, "USB/PMD-1608FS not detected.") End End If Ulstat = cbErrHandling(PRINTALL, DONTSTOP) If Ulstat <> 0 Then Stop ' If cbErrHandling is set for STOPALL or STOPFATAL during the program ' design stage, Visual Basic will be unloaded when an error is encountered. ' We suggest trapping errors locally until the program is ready for compiling ' to avoid losing unsaved data during program design. This can be done by ' setting cbErrHandling options as above and checking the value of ULStat ' after a call to the library. If it is not equal to 0, an error has occurred. End Sub Private Sub tmrConvert_Timer() ' Collect the data with cbAIn() ' Parameters: ' BoardNum% :the number used by CB.CFG to describe this board ' Chan% :the input channel number ' Gain :the gain for the board. 115 ' DataValue% :the name for the value collected Chan% = 1 'Set input channel. In Single Ended you can set 'Chan%' between 0 and 7 'In Differential Input mode, you can only set 'Chan%' between 0 and 3. 'You can set 'Chan%' to a different value to suit you needs Range = BIP5VOLTS 'Set the input range for the PMD-1608FS. 'When in Singled Ended you MUST use this range. 'cbAIn returns a value in counts (a value between 0 and 4095 for a 12 bit converter). Ulstat = cbAIn(BoardNum%, Chan%, Range, DataValue%) If Ulstat <> 0 Then Stop '------------------------------------------------------------------ Range = BIP5VOLTS 'DataValue% comes from the cbAIn function above 'EngUnits! is the value calculated from the DataValue% 'Use the cbToEngUnits function to convert the raw counts value to volts (engineering units) Ulstat = cbToEngUnits(BoardNum%, Range, DataValue%, EngUnits!) If Ulstat <> 0 Then Stop ShowData.Text = DataValue% ' print the counts ShowVolts.Text = Round(EngUnits!, 2) & " Volts" ' print the voltage ShowPressure.Text = Round(EngUnits! * 750, 2) 'Image1.Visible = False ShowTime.Text = Time ShowDate.Text = Date Chan% = 2 'Set input channel. In Single Ended you can set 'Chan%' between 0 and 7 'In Differential Input mode, you can only set 'Chan%' between 0 and 3. 'You can set 'Chan%' to a different value to suit you needs Range = BIP5VOLTS 'Set the input range for the PMD-1608FS. 'When in Singled Ended you MUST use this range. 'cbAIn returns a value in counts (a value between 0 and 4095 for a 12 bit converter). Ulstat = cbAIn(BoardNum%, Chan%, Range, DataValue%) If Ulstat <> 0 Then Stop '------------------------------------------------------------------ Range = BIP5VOLTS 'DataValue% comes from the cbAIn function above 'EngUnits! is the value calculated from the DataValue% 'Use the cbToEngUnits function to convert the raw counts value to volts (engineering units) 116 Ulstat = cbToEngUnits(BoardNum%, Range, DataValue%, EngUnits!) If Ulstat <> 0 Then Stop ShowData1.Text = DataValue% ' print the counts ShowVolts1.Text = Round(EngUnits!, 2) & " Volts" ' print the voltage ShowPressure1.Text = Round(EngUnits! * 750, 2) Chan% = 3 'Set input channel. In Single Ended you can set 'Chan%' between 0 and 7 'In Differential Input mode, you can only set 'Chan%' between 0 and 3. 'You can set 'Chan%' to a different value to suit you needs Range = BIP5VOLTS 'Set the input range for the PMD-1608FS. 'When in Singled Ended you MUST use this range. 'cbAIn returns a value in counts (a value between 0 and 4095 for a 12 bit converter). Ulstat = cbAIn(BoardNum%, Chan%, Range, DataValue%) If Ulstat <> 0 Then Stop '------------------------------------------------------------------ Range = BIP5VOLTS 'DataValue% comes from the cbAIn function above 'EngUnits! is the value calculated from the DataValue% 'Use the cbToEngUnits function to convert the raw counts value to volts (engineering units) Ulstat = cbToEngUnits(BoardNum%, Range, DataValue%, EngUnits!) If Ulstat <> 0 Then Stop ShowData2.Text = DataValue% ' print the counts ShowVolts2.Text = Round(EngUnits!, 2) & " Volts" ' print the voltage ShowPressure2.Text = Round(EngUnits! * 750, 2) Chan% = 4 'Set input channel. In Single Ended you can set 'Chan%' between 0 and 7 'In Differential Input mode, you can only set 'Chan%' between 0 and 3. 'You can set 'Chan%' to a different value to suit you needs Range = BIP5VOLTS 'Set the input range for the PMD-1608FS. 'When in Singled Ended you MUST use this range. 'cbAIn returns a value in counts (a value between 0 and 4095 for a 12 bit converter). Ulstat = cbAIn(BoardNum%, Chan%, Range, DataValue%) If Ulstat <> 0 Then Stop '------------------------------------------------------------------ Range = BIP5VOLTS 'DataValue% comes from the cbAIn function above 'EngUnits! is the value calculated from the DataValue% 117 'Use the cbToEngUnits function to convert the raw counts value to volts (engineering units) Ulstat = cbToEngUnits(BoardNum%, Range, DataValue%, EngUnits!) If Ulstat <> 0 Then Stop ShowData3.Text = DataValue% ' print the counts ShowVolts3.Text = Round(EngUnits!, 2) & " Volts" ' print the voltage ShowPressure3.Text = Round(EngUnits! * 750, 2) 'If (LogData.Value = 1) And (OPENFILE = True) Then ' Write #1, ShowTime.Text, Spinner1.Text, Spinner2.Text, Chain.Text, GSRSpeed.Text, ShowPressure.Text, ShowPressure1.Text, ShowPressure2.Text, ShowPressure3.Text 'End If End Sub Private Sub MSComm1_OnComm() 'MSComm2 ROUTINE DEFINES OPERATIONS ON NEW SERIAL MESSAGE FOR CNTR BOARD On Error Resume Next If MSComm1.CommEvent = comEvReceive Or (MSComm1.CommEvent = comEvRing) Then 'CHECK FOR NEW MESSAGE RECEIVED BUFFER_LENGTH = MSComm1.InBufferCount Else BUFFER_LENGTH = 0 End If While BUFFER_LENGTH > 5 BUFFER_ARRAY = BUFFER_LEFTOVER & MSComm1.Input 'ADD NEW MESSAGE TO BUFFER START_POS = InStr(BUFFER_ARRAY, "%") 'DEFINE START OF MESSAGE STRING AT CHANNEL 1 END_POS = START_POS + 7 'InStr(BUFFER_ARRAY, "!") ' DEFINE END OF MESSAGE STRING BUFFER_LENGTH = END_POS - START_POS 'Len(BUFFER_ARRAY) 'BUFFERLENGTH SET EQUAL TO LENGTH OF MESSAGE If START_POS = 0 Then Exit Sub If END_POS = 0 Then Exit Sub 'DATA_STRING = BUFFER_ARRAY DATA_STRING = Mid(BUFFER_ARRAY, START_POS + 1, END_POS - START_POS + 1) 'SPLIT DATA STRING DATA_ARRAY = Split(DATA_STRING, ",") CounterID = DATA_ARRAY(0) Frequency = DATA_ARRAY(1) 118 COUNTER(CounterID).Text = Frequency Spinner1.Text = COUNTER(1) * 1.2 Spinner2.Text = COUNTER(2) * 1.2 Chain.Text = COUNTER(3) / 6 'PLACE CHAIN SPEED CONVERSION VALUE HERE GSRSpeed.Text = COUNTER(4) * 0.0177 'PLACE GSR VALUE HERE FOR CONVERSION 'If (LogData.Value = 1) And (OPENFILE = True) Then ' Write #1, count1.Text, count2.Text, count3.Text, count4.Text, count5.Text, count6.Text 'End If BUFFER_LEFTOVER = Right(BUFFER_ARRAY, END_POS + 1) 'COLLECT UNUSED BUFFER If (LogData.Value = 1) And (OPENFILE = True) Then Write #1, ShowTime.Text, Spinner1.Text, Spinner2.Text, Chain.Text, GSRSpeed.Text, ShowPressure.Text, ShowPressure1.Text, ShowPressure2.Text, ShowPressure3.Text End If Wend End Sub Private Sub OpenPort_Click() If MSComm1.PortOpen = True Then MsgBox "The Port Is Already In Use" Exit Sub End If MSComm1.PortOpen = True End Sub 119 APPENDIX G SPINNER SPEED DATA 120 G.1 SPINNER SPEED DATA FOR INDIVIDUAL REPLICATIONS Table G.1. Replication 1 spinner speed summary. Target Mean Std. Dev. Diff. 1 1 Mean Std. Dev. Diff. 1 1 Diff. 2 2 (kg/ha) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) 2242 599.8 5.9 -0.2 597.7 6.6 -2.3 -2.1 4483 597.2 12.1 -2.8 598.0 11.3 -2.0 0.8 6725 599.4 5.4 -0.6 599.8 5.9 -0.2 0.3 2242 624.7 5.1 24.7 636.0 5.0 36.0 11.3 4483 597.7 4.5 -2.3 606.6 5.7 6.6 8.9 6725 609.6 10.5 9.6 623.3 6.5 23.3 13.8 System Closed-loop Open-loop Spinner 1 Spinner 2 Table G.2. Replication 2 spinner speed summary. Target Mean Std. Dev. Diff. 1 1 Mean Std. Dev. Diff. 1 1 Diff. 2 2 (kg/ha) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) 2242 599.3 6.2 -0.7 607.5 6.6 7.5 8.2 4483 597.7 11.2 -2.3 614.9 13.1 14.9 17.2 6725 598.6 13.9 -1.4 614.7 23.5 14.7 16.1 2242 648.7 4.3 48.7 643.7 4.3 43.7 -5.1 4483 613.0 4.5 13.0 629.2 6.8 29.2 16.2 6725 618.2 7.9 18.2 629.2 5.9 29.2 11.0 System Closed-loop Open-loop Spinner 1 Spinner 2 Table G.3. Replication 3 spinner speed summary. Target Mean Std. Dev. Diff. 1 1 Mean Std. Dev. Diff. 1 1 Diff. 2 2 (kg/ha) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) (rpm) 2242 599.2 7.5 -0.8 596.1 7.5 -3.9 -3.1 4483 598.8 10.5 -1.2 602.7 9.3 2.7 3.8 6725 598.0 8.8 -2.0 602.8 9.8 2.8 4.8 2242 614.4 6.5 14.4 610.3 5.5 10.3 -4.1 4483 594.9 15.0 -5.1 604.3 11.1 4.3 9.4 6725 565.3 14.4 -34.7 575.4 15.0 -24.6 10.1 1) Diff. 1 is the difference of the spinners from the desired speed of 600 rpm; positive and negative indicating when the actual speed is greater than or less than the desired speed, respectively. 2) Diff. 2 is the difference between spinners 1 and 2; negative indicating spinner 1 is faster. Spinner 1 Spinner 2 Closed-loop Open-loop System 121 APPENDIX H MASS DISTRIBUTION DATA 122 H.1 MASS OVERLAP DISTRIBUTION PATTERNS The simulated overlaps patterns in figure H.1 illustrate the differences occurring between replications. The results in Chapter 3 are based upon the mean of these patterns for the three replications. 0 500 1000 1500 2000 2500 3000 3500 4000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap plc ati on R ate (k g/h a) Rep 1 Rep 2 Rep 3 Desired Avg P - 2242 kg/ha 0 500 1000 1500 2000 2500 3000 3500 4000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) Rep 1 Rep 2 Rep 3 Desired Avg M - 2242 kg/ha 0 1000 2000 3000 4000 5000 6000 7000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) Rep 1 Rep 2 Rep 3 Desired Avg P - 4483 kg/ha 0 1000 2000 3000 4000 5000 6000 7000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) Rep 1 Rep 2 Rep 3 Desired Avg M - 4483 kg/ha 0 2000 4000 6000 8000 10000 12000 14000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) Rep 1 Rep 2 Rep 3 Desired Avg P - 6725 kg/ha 0 2000 4000 6000 8000 10000 12000 14000 -5 -4 -3 -2 -1 0 1 2 3 4 5 Transverse Distance (m) Ap pli ca tio n R ate (k g/h a) Rep 1 Rep 2 Rep 3 Desired Avg M - 6725 kg/ha a) Closed-loop control system b) Open-loop control system Figure H.1. Mass overlap patterns for each replication by type of control system and application rate. 123 The data in the following tables corresponds to the patterns illustrated in figure H.1 with data presented for each replication. Table H.1. Rep 1 simulated mass overlap pattern summary statistics. Target Mean Std. Dev. CV Rate Diff. 1 (kg/ha) (kg/ha) (kg/ha) (%) (kg/ha) 2242 2297 646 28.1 55 4483 4214 674 16.0 -269 6725 8081 1078 13.3 1356 2242 2027 582 28.7 -215 4483 4072 1289 31.7 -411 6725 6778 2001 29.5 53 System Closed-loop Open-loop Table H.2. Rep 2 simulated mass overlap pattern summary statistics. Target Mean Std. Dev. CV Rate Diff. 1 (kg/ha) (kg/ha) (kg/ha) (%) (kg/ha) 2242 2185 581 26.6 -57 4483 4141 1131 27.3 -342 6725 6758 1565 23.1 33 2242 1903 591 31.0 -339 4483 4523 1124 24.9 40 6725 5723 2241 39.2 -1002 System Closed-loop Open-loop Table H.3. Rep 3 simulated mass overlap pattern summary statistics. Target Mean Std. Dev. CV Rate Diff. 1 (kg/ha) (kg/ha) (kg/ha) (%) (kg/ha) 2242 2776 598 21.5 534 4483 4795 1115 23.2 312 6725 7553 3000 39.7 828 2242 2311 564 24.4 69 4483 4656 1579 33.9 173 6725 7001 3138 44.8 276 1) Positive and negative rate differences indicate over- and under-application, respectively. Open-loop System Closed-loop 124 APPENDIX I NUTRIENT SIMULATED OVERLAP DATA 125 I.1 SUMMARIZED NUTRIENT SIMULATED OVERLAP DATA Tables I.1, I.2, and I.3 summarize the simulated nutrient overlap data for the three replications. The average of the data in these three tables was used to compute the results reported in Chapter 4. Note the CV differences between replications which highlight the variability of litter and the difficulty of uniformly applying litter. Table I.1. Rep 1 simulated nutrient overlap pattern summary statistics. Rate Nutrient Mean Std. Dev. CV (kg/ha) (kg/ha) (kg/ha) (%) N 57.0 16.2 28.5 P2O5 110.9 29.7 26.8 K2O 84.2 24.6 29.2 N 102.5 16.6 16.2 P2O5 203.2 29.7 14.6 K2O 143.6 22.1 15.4 N 132.5 25.0 18.9 P2O5 357.5 67.3 18.8 K2O 309.4 62.6 20.2 N 51.1 13.8 26.9 P2O5 91.7 25.0 27.2 K2O 72.5 20.0 27.5 N 102.3 32.9 32.1 P2O5 179.8 60.8 33.8 K2O 141.9 47.9 33.7 N 174.3 50.2 28.8 P2O5 315.6 87.8 27.8 K2O 251.8 70.4 28.0 Open-loop 2242 4483 6725 System Closed-loop 2242 4483 6725 126 Table I.2. Rep 2 simulated nutrient overlap pattern summary statistics. Rate Nutrient Mean Std. Dev. CV (kg/ha) (kg/ha) (kg/ha) (%) N 56.8 15.2 26.8 P2O5 95.9 25.5 26.6 K2O 77.3 19.8 25.6 N 106.1 29.4 27.7 P2O5 202.2 60.1 29.7 K2O 151.9 43.0 28.3 N 140.9 40.1 28.5 P2O5 335.2 76.6 22.8 K2O 257.6 61.6 23.9 N 47.8 13.6 28.4 P2O5 80.7 23.5 29.1 K2O 63.0 18.7 29.7 N 117.9 29.4 25.0 P2O5 222.3 95.2 42.8 K2O 161.5 40.4 25.0 N 144.6 56.6 39.2 P2O5 253.9 98.2 38.7 K2O 202.6 79.4 39.2 Open-loop 2242 4483 6725 System Closed-loop 2242 4483 6725 Table I.3. Rep 3 simulated nutrient overlap pattern summary statistics. Rate Nutrient Mean Std. Dev. CV (kg/ha) (kg/ha) (kg/ha) (%) N 68.5 14.7 21.4 P2O5 119.2 25.2 21.1 K2O 102.9 21.7 21.1 N 119.0 28.0 23.5 P2O5 221.5 44.5 20.1 K2O 186.6 37.1 19.9 N 184.6 69.9 37.8 P2O5 353.3 139.1 39.4 K2O 288.2 117.1 40.6 N 57.6 13.5 23.5 P2O5 105.0 21.4 20.4 K2O 88.0 19.2 21.8 N 112.1 40.5 36.1 P2O5 209.8 78.2 37.3 K2O 173.4 61.1 35.2 N 173.5 76.0 43.8 P2O5 316.6 142.9 45.1 K2O 263.7 117.6 44.6 4483 6725 2242 4483 6725 2242 Open-loop System Closed-loop 127 APPENDIX J LONGITUDINAL ANALYSIS REPORTING THE SUMMARIZED RESIDUAL DATA 128 Table J.1. Summarized residual data from the CLS 2242 kg/ha application rate. Rep Pan Location Overall Mass Mean Row A Rate Row B Rate Row C Rate Row D Rate Row A Residual Row B Residual Row C Residual Row D Residual (m) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) 1 -3.81 1805.8 885.4 1616.8 1520.0 1384.3 920.3 188.9 285.8 421.4 1 4.57 1404.9 1117.9 1398.9 1490.9 1539.3 287.0 6.1 -86.0 -134.4 2 -6.10 485.8 575.4 735.3 410.8 657.8 -89.6 -249.5 75.1 -172.0 2 3.05 2104.0 1684.7 1272.9 1597.5 1510.3 419.4 831.1 506.6 593.8 3 -2.29 1630.6 2173.9 1863.9 1704.0 1713.7 -543.3 -233.3 -73.5 -83.2 3 6.10 555.3 439.8 609.3 715.9 425.3 115.4 -54.1 -160.7 130.0 Table J.2. Summarized residual data from the CLS 4483 kg/ha application rate. Rep Pan Location Overall Mass Mean Row A Rate Row B Rate Row C Rate Row D Rate Row A Residual Row B Residual Row C Residual Row D Residual (m) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) 1 -3.05 3771.9 3602.8 2987.6 3263.7 3728.7 169.1 784.3 508.2 43.2 1 3.05 3524.9 2852.0 2362.8 1999.5 4111.4 672.9 1162.1 1525.4 -586.5 2 -5.34 1554.3 2290.1 1239.0 1117.9 2101.2 -735.8 315.2 436.3 -546.9 2 2.29 2605.0 1989.8 1713.7 2745.4 2411.2 615.2 891.3 -140.5 193.8 3 -6.10 927.4 958.1 861.2 1161.5 1151.8 -30.7 66.2 -234.1 -224.4 3 4.57 2191.2 1883.3 2953.7 2769.7 3675.4 308.0 -762.5 -578.4 -1484.2 Table J.3. Summarized residual data from the CLS 6725 kg/ha application rate. Rep Pan Location Overall Mass Mean Row A Rate Row B Rate Row C Rate Row D Rate Row A Residual Row B Residual Row C Residual Row D Residual (m) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) 1 -6.86 1159.9 2435.4 1093.7 1088.9 1350.4 -1275.5 66.2 71.0 -190.5 1 7.62 1112.3 1379.5 2299.8 1989.8 1147.0 -267.2 -1187.5 -877.5 -34.7 2 -3.05 5679.9 5734.0 4426.2 3917.6 4983.3 -54.1 1253.7 1762.3 696.7 2 5.34 2752.3 2227.2 2609.8 1917.2 2454.8 525.1 142.5 835.1 297.5 3 -4.57 3899.9 4280.9 3612.5 5874.5 4508.6 -381.0 287.4 -1974.6 -608.7 3 2.29 4201.8 4440.8 3258.9 4731.4 4571.5 -239.0 942.9 -529.6 -369.7 129 Table J.4. Summarized residual data from the OLS 2242 kg/ha application rate. Rep Pan Location Overall Mass Mean Row A Rate Row B Rate Row C Rate Row D Rate Row A Residual Row B Residual Row C Residual Row D Residual (m) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) 1 -5.34 768.8 735.3 740.1 783.7 938.7 33.5 28.7 -14.9 -169.9 1 1.52 883.0 677.2 861.2 788.6 745.0 205.9 21.8 94.5 138.0 2 -3.05 1848.5 1045.3 1214.8 1834.8 1471.5 803.3 633.7 13.7 377.0 2 7.62 278.0 231.5 236.4 357.5 154.0 46.4 41.6 -79.5 123.9 3 -3.05 1848.5 2159.3 2227.2 2522.6 2440.3 -310.8 -378.6 -674.1 -591.7 3 5.34 708.6 725.6 672.3 706.2 788.6 -17.0 36.3 2.4 -79.9 Table J.5. Summarized residual data from the OLS 4483 kg/ha application rate. Rep Pan Location Overall Mass Mean Row A Rate Row B Rate Row C Rate Row D Rate Row A Residual Row B Residual Row C Residual Row D Residual (m) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) 1 -4.57 2618.7 1384.3 1292.3 1394.0 2687.3 1234.3 1326.4 1224.7 -68.6 1 4.57 2764.0 2605.0 2140.0 3593.1 3123.3 159.0 624.0 -829.1 -359.2 2 -4.57 2618.7 2377.3 2890.8 2963.4 3331.5 241.4 -272.1 -344.7 -712.8 2 3.81 3635.1 3588.3 2793.9 4038.7 3641.5 46.8 841.2 -403.6 -6.5 3 -3.81 3716.2 3922.5 3476.8 4411.7 4087.2 -206.3 239.4 -695.5 -371.0 3 1.52 2905.2 1631.4 1558.7 1805.8 1980.1 1273.8 1346.5 1099.5 925.1 Table J.6. Summarized residual data from the OLS 4483 kg/ha application rate. Rep Pan Location Overall Mass Mean Row A Rate Row B Rate Row C Rate Row D Rate Row A Residual Row B Residual Row C Residual Row D Residual (m) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) (kg/ha) 1 -6.10 1235.8 1423.1 1064.7 1582.9 1161.5 -187.3 74.3 -347.1 74.3 1 5.34 2031.4 1660.4 1389.2 1888.1 1612.0 371.0 419.4 143.3 419.4 2 -6.86 803.5 924.2 861.2 745.0 735.3 -120.7 68.2 58.5 68.2 2 1.52 2786.6 2367.6 2435.4 2304.7 3462.3 419.0 -675.7 482.0 -675.7 3 -6.86 803.5 715.9 851.5 599.7 774.0 87.6 29.5 203.8 29.5 3 7.62 715.9 987.2 648.1 1050.1 376.8 -271.3 339.1 -334.2 339.1 130 APPENDIX K PRESSURE DATA 131 To be able to gain a better understanding of the hydraulic system on the spreader and determine why spinner 2 was operating at a higher speed than spinner 1, pressure transducers were installed to determine if the pressure drops across each of the motors were equal. The pressure tests were conducted after the data collection tests. Below is the data for the pressure tests. The CLS speed data (table k.1) is not consistent with the data collected during data collection. Something occurred with either the tractors hydraulic system, the CLS flow control valve, or the control algorithm not allowing the controller to maintain a consistent spinner speed when material was being applied. The OLS data (table k.2) was similar to the data collected during the actual field tests; however, there is still no definite evidence to conclude why spinner 2 runs faster than spinner 1. Table K.1. Summarized spinner-disc speed and pressure data for the CLS. Load Spinner 1 Spinner 2 Diff.1 Chain GSR Input Press. Spinner 1 Input Spinner 2 Input Output Press. Spinner 1 Press. Drop Spinner 2 Press. Drop (kg/ha) (rpm) (rpm) (rpm) (rpm) (mph) (psi) (psi) (psi) (psi) (psi) (psi) None 598.3 599.2 1.0 0.0 0.0 281.2 241.4 150.9 73.7 90.4 77.3 2242 607.6 633.8 26.2 4.8 3.5 777.3 733.7 416.2 87.8 317.5 328.4 4483 596.3 674.5 78.1 11.3 4.0 1426.1 1368.6 707.0 105.7 661.5 601.4 6725 530.1 642.5 112.4 17.2 4.1 1899.9 1848.3 1027.6 103.7 820.7 923.9 1) Diff. 1 is the difference between spinners 1 and 2; negative indicating spinner 1 is faster. Table K.2. Summarized spinner-disc speed and pressure data for the OLS. Load Spinner 1 Spinner 2 Diff.1 Chain GSR Input Press. Spinner 1 Input Spinner 2 Input Output Press. Spinner 1 Press. Drop Spinner 2 Press. Drop (kg/ha) (rpm) (rpm) (rpm) (rpm) (mph) (psi) (psi) (psi) (psi) (psi) (psi) None 658.0 657.8 -0.2 0.0 0.0 2611.2 275.2 178.6 82.4 96.7 96.2 2242 638.4 637.4 -1.0 5.7 4.1 2612.9 447.3 274.1 81.0 173.2 193.1 4483 584.5 606.2 21.7 11.3 4.0 2605.7 710.0 373.5 80.3 336.5 293.3 6725 569.4 585.5 16.0 17.2 4.0 2602.2 830.3 429.4 79.9 401.0 349.5 1) Diff. 1 is the difference between spinners 1 and 2; negative indicating spinner 1 is faster.