Back to Contents6.3.2 Investigation of Machinery and Controls Limitations on Input Management Resolution
Investigators: Scott Shearer, Sam McNeill, Tom Mueller, Richard Barnhisel, Larry Wells, and Steve Higgins.
Site-specific management of crop production is limited by the cost of obtaining and characterizing the variability of parameters such as soil fertility. Often overlooked are the capabilities and limitations of variable-rate application equipment. Specification of the control criteria and development of crop response functions provide two of the three building blocks for successful and profitable variable-rate application. Both the variability characterization and equipment capabilities combine to place limitations on the management resolution. This portion of the project is intended to assess and quantify machinery limitations. Equipment concerns can be divided into three major areas: 1) accuracy and precision of positioning information; 2) accuracy, precision and distribution of inputs using variable-rate control; and 3) capital cost and life of the equipment. This portion of the project is intended to address the first two equipment concerns.
Differential GPS (DGPS) Receivers - As GPS technology continues to evolve there is a corresponding reduction in cost and improvement in performance. The burden for correcting positioning information available to civilian users via the coarse acquisition code (C/A code) of the standard positioning service rests with both the public and private sectors through differential correction. Kentucky producers must choose between wide-area and local correction services, and once this decision is made, must live with the accuracy and errors of the system they have purchased. Unfortunately the most appropriate purchase considerations vary from location to location, and with the intended GPS application. Therefore it is important to supply producers with unbiased DGPS performance criteria, and an indication of which systems are available to them.
Field tests will be conducted for three primary methods of differential correction that include: 1) local differential correction via AM-band U.S. Coast Guard radio-beacon, 2) wide-area differential correction via C-band carrier, and 3) wide-area differential correction via L-band carrier. Efforts will be made to insure that the GPS engine used for each correction approach is the same. Kentucky will be divided into between seven and ten grain producing regions. Radio-beacon, C-band and L-band DGPS receivers will be taken to each location to assess carrier acquisition times and circular errors of probability (95% CEP). This portion of field-testing will be constrained to horizontal parameters (latitude and longitude). USGS benchmarks will be selected as reference points, and data will be collected with each receiver for a time period of 20 to 30 minutes, or of sufficient duration to capture in excess of 1000 data points from each receiver. These data will be used as the basis for reporting DGPS accuracy. All DGPS data collected in the field will be transformed from WGS 84 to state plane coordinates using the NAD 83 datum. Horizontal positions will be described using Easting and Northing directions and units of meters.
Static tests, as indicated above, are a valuable tool for assessing the accuracy of DGPS receivers. However of interest to the producer is the ability to return year after year to the same position for the purpose of re-sampling to track fertility, to map and delineate field boundaries, and for real-time control of inputs to the production unit. These applications are considered to be dynamic, and are influenced to some extent by the software and hardware used to acquire and display positioning information from the receiver. In view of these considerations, field tests will be conducted to determine the accuracy of the three previously defined DGPS receiver configurations. Dynamic field- testing will be accomplished by mapping the boundary of a single production unit or field on fifteen different occasions. DGPS receivers will be mounted on an ATV equipped with laptop computer for logging data. This test series will be conducted at three locations.
Field boundary mapping results will be compared for accuracy by constructing a base polygon defining the field. This base polygon will be constructed using the intersection of rays originating from the weighted-average centroid of all fifteen field boundary polygons, and arc segments defining the field boundary for each of the fifteen test cases. The number of equally spaced rays used for this part of the analysis will equal the average number of vertices collected during boundary mapping. RMS area errors will be determined for each polygon by calculating the area between the base and observation polygons for between pairs of rays, squaring these values, and summing these around the centroid. Additionally, polygon areas will be determined using Green's Theroem and line-integrals.
Students' T-tests will be used to compare the population means of the various absolute test parameters for the three DGPS receiver configurations. The variances will be compared for both static and dynamic testing using an F-test to establish if there are significant differences in receiver performance and repeatability.
Variable-Rate Granular and Liquid Applicators and Seeding Equipment - Application accuracy and distribution are important properties to assess on multiple-bin, variable-rate granular fertilizer and liquid application systems. Steady-state control system errors and lag times contribute significantly to application error. To assess the current level of application errors associated with variable-rate application both field and bench tests will be conducted to evaluate two commercially available control systems.
Field tests will consist of evaluating the deposition from a two-bin variable-rate spinner disc spreader applying granular potash and phosphate fertilizer, and agricultural lime. Application rate and distribution tests will be conducted in-situ. ASAE standards (S341.2 Procedure for Measuring Distribution Uniformity and Calibrating Granular Broadcast Spreaders in ASAE, 1997) describes a uniform method of determining performance data of broadcast or spinner spreaders for granular materials, and provides a basic test procedure to compare spreader distribution patterns. Test procedures that include well-defined setup, collection device specifications, test procedures, distribution specification, effective swath width specification, and application rate determination. The test procedure will be modified to include a two-dimensional array of collection pans to evaluate the effect of rate changes via GPS control. The size of the array normal to the direction of travel will include 13 collection pans, spaced evenly across the anticipated distribution pass. Along the axis of travel will be 13 evenly spaced row of collection trays. The spacing will be determined as a function of the speed of response of the controller and the ground speed of the applicator. The spacing will be adjusted as parameters warrant.
Test cases to be investigated for the application of granular P and K fertilizers will include: 1) fixed-rate application of single nutrient at a low rate; 2) fixed-rate application of single nutrient at a high rate; 3) fixed-rate application of two nutrients at low rates; 4) fixed-rate application of two nutrients at high rates; 5) variable-rate application of a single nutrient from low to high rate; 6) variable-rate application of a single nutrient from high to low rate; 7) variable-rate application of one nutrient from low to high with low fixed rate of the second nutrient; 8) variable-rate application of one nutrient from high to low with low fixed rate of the second nutrient; 9) variable-rate application of one nutrient from low to high with high fixed rate of the second nutrient; 10) variable-rate application of one nutrient from high to low with high fixed rate of the second nutrient, 11) variable-rate application of two nutrients, both from low to high; 12) variable-rate application of two nutrients, both from high to low; and 13) variable-rate application of two nutrients, one from low to high and the second from high to low. Test cases for variable-rate application of lime will include: 1) fixed-rate application at a low rate; 2) fixed-rate application at a high rate; 3) variable-rate application from low to high rate; and 4) variable-rate application from high to low rate. Selected tests will be rerun with multiple parallel passes to evaluate the effect of overlap on uniformity. Data will be collected at this time to assess how accurately the operator can judge the distance between parallel passes.
The Biosystems and Agricultural Engineering program currently maintains a custom built two-bin variable-rate fertilizer truck. The truck employs spinner disc technology to apply granular products. A Midwest Technologies, Inc. TASC 6200 Controller opens and closes motorized flow control valves to vary the speed of metering augers and chains. This truck will be used for all of the granular application field studies.
Particle samples will be collected, bagged and labeled for each of the field tests. All samples will be weighed to the nearest 0.01 grams. Particle sizes will be determined by sieving all samples. For fertilizer samples where more than one nutrient has been applied, chemistry lab analysis will be required to determine the relative quantities of P and K. Distribution plots will be generated for all test cases. Application errors will be summed over the test area and reported as a percentage of the total mass of applied products.
Results from field testing of granular fertilizer application will be used to develop a simulation model for predicting application accuracy given the recorded DGPS trace of the application track. Monte Carlo simulation will be used to project application errors associated with pattern overlap, particle dynamics, and spinner disc dynamics. This methodology will also enable the prediction of fertilizer applied beyond the field boundary.
Laboratory bench tests will be conducted to evaluate the response and steady-state control errors associated with two models of controllers, Midwest Technologies, Inc. TASC 6200 Controller, and the Rawson, Inc. Accu-Plant drive. Hydraulic power will be supplied via a test bench power station located in the Agricultural Engineering Building. This station will provide constant and consistent hydraulic power for controller evaluation. The drive motors will be loaded in a manner similar to what might be expected under field conditions (e.g., back pressure for fluid metering systems, or dry friction for granular fertilizers). Tests to be conducted for either dry granular or liquid products include: 1) fixed-rate application at a low rate; 2) fixed-rate application at a high rate; 3) variable-rate application from low to high rate; and 4) variable-rate application from high to low rate. Controller errors will be monitored by tracking the speed of motor shafts via Hall Effect or capacitance speed sensors. Test results will be plotted to illustrate the difference between input commands and actual drive speeds. These plots will be characterized by defining time constants, or damping ratios and natural frequencies, depending on the form of the response.
Managing inputs to crop production systems using precision agriculture or site-specific practices is currently predicated on manual soil sampling, and laboratory analyses. The per sample cost ranges from $7.00 to $8.00, and therefore is the primary constraint on the economics associated with this technology. Productivity gains are often offset by the cost of obtaining soil fertility data. Further, it is this cost that fixes the management resolution (e.g., one-acre grids, two and one-half acre grids, etc.), either by balancing soil sampling cost against returns, or more often by what the market will support. On the analytical side, laboratory capacity continues to increase, while the cost of analysis remains the same, $3.50 to $4.00 per sample analyzed. Approximately one-half the cost of soil sampling is attributed to boundary mapping, gridding, field collection and composting of soil cores, and data manipulation. Experience indicates that manual extraction and compositing of multiple soil cores, limits worker productivity to 18 samples per hour. The objective of this portion of the work is to increase sampling productivity by three-fold (to 60 samples per hour) by mechanizing soil core collection, and to reduce the potential for errors associated with data management via use of bar codes to identify soil samples.
The envisioned high volume soil sampler will consist of a mechanical device capable of extracting five soil cores at a predetermined radius about a grid point or other pre-specified location, composite the sample, and deposit it in a single, pre-labeled sample bag. Components of each of the five coring devices will include an auger, soil tube, hydraulic motor to power the auger, and hydraulic cylinder and appropriate hydraulic valves to control the movement of the auger and soil tube into the ground. As the soil is augered from the ground and transported via a vacuum system, a cyclone separator will be used to remove the soil from the air-stream, composite cores from all five probes, and drop the soil into a plastic bag at the base of the separator. The goal is to complete the coring process in less than 20 seconds. Initial testing of a single coring device prototype has confirmed the potential to reach sampling times of less that 20 seconds. A 30 kW utility tractor will be used as the system platform with hydraulic power to be provided through a power-take-off driven hydraulic pump. The soil coring devices will be located radially around the tractor, with the geometric center to coincide with the center of gravity of the tractor. The antenna of the GPS receiver will be located vertically from the geometric center of the coring devices. The tractor referenced above is currently on-hand in Biosystems and Agricultural Engineering. The Department also maintains a machine shop and personnel capable of fabricating the sampling system.
Data tracking and management within the field will be performed using bar codes. The hardened personal computer will be equipped to obtain GPS location information, provide field-mapping capabilities, and print sample bag labels complete with bar codes. A CODE 39 format will be used to identify soil samples. CODE 39 bar code format is gaining acceptance for application not suited for UPC (Universal Product Code) formats. Bar codes can be scanned at the laboratory to track sample origins and to minimize errors in transcribing sample analyses values. And consistent with the proposed soil fertility data dictionary (Agricultural Electronics Association, 1998), position information will be organized and stored accordingly.
The successful completion of this phase of the project will result in a soil sampling system that can be used at field days and in support of the research efforts as defined in other areas of this project. System capacity is targeted at collection of 300 plus composited soil samples per day, which significantly exceeds current manual core extraction practices.
The intent of the field and laboratory investigations is to characterize and model variable-rate application of granular and liquid materials. Using these findings, and the results from the DGPS receiver tests, accuracy of variable-rate application can be determined for a variety of grid resolutions. Acceptable management grid resolution ranges will be specified in accordance with equipment limitations. This will help to refine soil-sampling strategies for optimal returns to the producer. Findings of the soil sampler develop portion of this project will be shared with equipment manufacturers, service providers and soil testing laboratories with the intent of reducing the cost to producers, and eliminating the potential for errors associated with managing soil fertility data. The results of all field and laboratory investigations will be published in refereed journals as appropriate, and in circulars and extension bulletins for dissemination to producers. This work will also be used in support of the handbook development as noted in Section 8.0 of this proposal. Information will also be provided to researchers under Objective 4 to aid in the assessment of whole farm economics.