University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln
Biological Systems Engineering: Papers and
Publications
Biological Systems Engineering
2016
In-field fuel use and load states of agricultural field
machinery
Santosh Pitla
University of Nebraska-Lincoln, spitla2@unl.edu
Joe D. Luck
University of Nebraska-Lincoln, jluck2@unl.edu
Jared Werner
University of Nebraska-Lincoln
Nannan Lin
Ohio State University
Scott A. Shearer
Ohio State University, shearer.95@osu.edu
Follow this and additional works at: http://digitalcommons.unl.edu/biosysengfacpub
Part of the Applied Mechanics Commons, Bioresource and Agricultural Engineering Commons,
Computational Engineering Commons, Computer and Systems Architecture Commons, Data
Storage Systems Commons, Environmental Engineering Commons, Other Civil and Environmental
Engineering Commons, Other Computer Engineering Commons, and the Other Mechanical
Engineering Commons
Pitla, Santosh; Luck, Joe D.; Werner, Jared; Lin, Nannan; and Shearer, Scott A., "In-field fuel use and load states of agricultural field
machinery" (2016). Biological Systems Engineering: Papers and Publications. 469.
http://digitalcommons.unl.edu/biosysengfacpub/469
This Article is brought to you for free and open access by the Biological Systems Engineering at DigitalCommons@University of Nebraska - Lincoln. It
has been accepted for inclusion in Biological Systems Engineering: Papers and Publications by an authorized administrator of
DigitalCommons@University of Nebraska - Lincoln.
Published in Computers and Electronics in Agriculture 121 (2016) 290–300. doi:10.1016/j.compag.2015.12.023
Copyright © 2016 Elsevier B.V. Used by permission.
Submitted 10 August 2015; revised 13 November 2015; accepted 31 December 2015; published online 12 January 2016.
digitalcommons.unl.edu
In-field fuel use and load states of agricultural field machinery
Santosh K. Pitla,1 Joe D. Luck,1 Jared Werner,1 Nannan Lin,2 Scott A. Shearer2
1 Department of Biological Systems Engineering, University of Nebraska-Lincoln, Lincoln, NE, USA
2 Department of Food Agricultural and Biological Engineering, Ohio State University, Columbus, OH, USA
Corresponding author — S. Pitla, 207 L.W. Chase Hall, Department of Biological Systems Engineering, University of Nebraska-Lincoln,
Lincoln, NE, USA. Tel.: +1 402 472 1466. E-mail address: spitla2@unl.edu (S.K. Pitla).
Abstract
The ability to define in-field tractor load states offers the potential to better specify and characterize fuel consumption
rate for various field operations. For the same field operation, the tractor experiences diverse load demands and corresponding fuel use rates as it maneuvers through straight passes, turns, suspended operation for adjustments, repair
and maintenance, and biomass or other material transfer operations. It is challenging to determine the actual fuel rate
and load states of agricultural machinery using force prediction models, and hence, some form of in-field data acquisition capability is required. Controller Area Networks (CAN) available on the current model tractors provide engine performance data which can be used to determine tractor load states in field conditions. In this study, CAN message data
containing fuel rate, engine speed and percent torque were logged from the tractor’s diagnostic port during anhydrous
NH3 application, field cultivation and planting operations. Time series and frequency plots of fuel rate and percent torque
were generated to evaluate tractor load states. Based on the percent torque, engine speed and rated engine power, actual load on the tractor was calculated in each tractor load state. Anhydrous NH3 application and field cultivation were
characterized by three distinct tractor load states (TS-I, TS-II and TS-III) corresponding to idle states, parallel and headland passes, and turns, whereas corn planting was characterized by two load states (TS-I and TS-II): idle, and a combined
state with parallel, headland passes and turns. For anhydrous NH3 application and field cultivation at ground speeds of
7.64 km h–1 and 8.68 km h–1, average tractor load per tool and fuel use rate per tool of the implement were found to be
7.21 kW tool–1, 3.28 L h–1 tool–1, and 1.31 kW tool–1, 0.64 L h–1 tool–1, respectively. For planting, average tractor load per
row and fuel use rate per row were found to be 4.65 kW row–1 and 1.70 L h–1 row–1 at a ground speed of 7.04 km h–1 .
Keywords: Controller, CAN, Tractor, Machinery, Load, Performance
tractor loading states such as working periods (e.g., parallel and
headland passes) and non-working periods (e.g. field adjustments
and repairs) is required. Understanding actual load profiles of the
tractor in different working states has the potential to yield true average load conditions. Improved fuel consumption estimation, and
better tractor and implement matching are some of the benefits of
in-field tractor load state determination.
Tractor performance is currently evaluated using OECD 2 test
code (OECD, 2012) where tractors are operated under steady-state
conditions, selected engine speeds and torques which are a subset
of several field operating conditions. Power take-off (PTO) power,
drawbar power, and specific fuel consumption are reported to assess
the performance of a tractor under controlled conditions. However,
measuring the performance of the tractor under field conditions is
central to a more thorough understanding of the actual power consumed by implements for various working phases of field operations.
1. Introduction
Tractors are used for multiple field operations during the entire
working season and hence are subjected to varying load demands.
Further, for a specific operation, the load demands on the tractor
change as a result of ground speed variations, effective implement
working widths and depth of operations, field conditions (e.g., soil
variability and terrain slope), and machine handling by the operator. When selecting and matching equipment complements, data
is readily available for projecting engine load demands of various
field operations (ASABE, 2011a). The reference data provides required draft forces at typical working speeds for specific operations
(chisel plowing, seeding, etc.), however these power requirements
(draft and rotary) of the implements vary within a maximum range of
±50% based on the type of operation (ASABE, 2011a). A more accurate estimation of power drawn by the implements during different
290
In-field fuel use and load states of agricultural field machinery
Engine speed and load also effect emissions, and hence, accurate
load estimation of the tractor will indirectly lead to improved emission calculations and fuel consumption measurements. Thus, determination of in-field tractor load states is essential for improved fuel
efficiency, better matching of implements to tractors, and accurate
estimation of emissions.
Tractor load state estimation and performance testing has been
the subject of many engine development and emission control investigations. More recently, manufacturers have focused their attention on off-highway engine emissions. Specifically, ISO 8178
(ISO, 2006) suggests engine test cycles (e.g. type C1, C2, and D1)
for various classes of engines and equipment. These cycles include
a sequence of steady-state modes for evaluating engine emission
performance. Unfortunately, the test cycle conditions deviate from
engine operating conditions experienced in actual field applications. ASABE (2011b) provides practices to follow when estimating fuel use rate and draft power requirements for hitched and
other types of equipment loads. However, recommendations are
not made for fuel consumption during non-working periods including when the tractor is stopped for field adjustment or repair
and maintenance, when the tractor is making end-of-row turns,
or when the tractor is operated at reduced speeds to accomplish
field border passes.
Efforts are underway to predict off-road equipment emissions.
An emission inventory model known as NONROAD was developed
to predict emissions based on the equipment use (Harvey, 2003). The
model estimates an emission factor which is a function of transient
adjustment factors (TAFs). The TAFs are based on engine speeds
and loads (both transient and steady-state) of off-road equipment.
A load factor of 0.78 was considered for agricultural tractors in predicting the emission factors (Harvey, 2003). This load factor is an approximate indicator of the true load factors of the agricultural machinery, and depending on the type of operation, could have either
overestimated or underestimated the engine load factor.
In-field machine performance data acquisition could be of significant value for determining actual load factors and states of tractors. Burgun et al. (2013) conducted a long term data acquisition
campaign for evaluating mechanical energy needs of the plowing
operation, and suggested dual alternating profile of loads. Further
they used steady-state bench test results to predict operational efficiency and field load conditions. Two indicators, time efficiency (h
ha–1) and area specific fuel consumption (L ha–1) made these predictions possible. Yahya et al. (2009) developed a data acquisition
system for use with an agricultural tractor for mapping tractor-implement performance while disk plowing a field. In a similar effort,
Al-Suhaibani et al. (2010) instrumented a tractor for measuring performance parameters and the draft forces of various implements at
different depths and speeds. The authors found good correlations
between measured and predicted values of draft force, which validated the instrumentation methodology. The availability of the Controller Area Network (CAN) bus on the tractors is allowing researchers to obtain tractor performance data (Lin, 2014; Pitla et al., 2013;
Darr, 2012). Pitla et al. (2014) obtained tractor fuel use rate messages from the CAN bus to determine field efficiencies of row crop
operations based on a threshold fuel use rate methodology. Further, researchers have compared CAN bus fuel use rates of tractors
to physical tractor fuel measurements to understand the accuracy
of CAN fuel rate data (Cupera and Sedlak, 2011; Marx, 2015; Marx
et al., 2015). The study conducted by Marx et al. (2015) concluded
that a maximum error of 6.22% between the physical fuel rate measurement and the CAN bus fuel rate measurement is possible. Fuel
rate errors were found to be higher at lower fuel rates, whereas for
higher engine fuel use rates within the torque curve the errors were
found to be closer to ±1% (Marx et al., 2015). Thus, given the utility and availability of CAN bus data on current day machinery, this
291
source of data provides an attractive alternative for tractor performance evaluation. As part of this research, CAN bus data were recorded for estimating true load states of the tractors performing
typical row crop production operations.
2. Objectives
The specific objectives of this investigation were to:
(1) Obtain CAN messages related to tractor performance from the
communication diagnostic ports of four wheel drive (4WD) and
mechanical front wheel drive (MFWD) tractors during row crop
production field operations (e.g., anhydrous ammonia (NH3) application, field cultivation and planting).
(2) Determine actual fuel use rates and power consumption in different load states of the tractors performing NH3 application,
cultivation and planting.
3. Materials and methods
CAN bus data were logged from a 245 kW rated 4WD tractor (JD
9410R, Deere & Co., Moline, IL) and a 127 kW rated MFWD tractor
(JD 7200R, Deere & Co., Moline, IL) during field operations. The 4WD
tractor (see Figure 1a) was used to pull an NH3 applicator (DW 6032,
Dalton Ag Products, Lenox, IA) and a field cultivator (JD 2210, Deere
& Co., Moline, IL) shown in Figure 1b.The MFWD tractor (see Figure
1c) was used for planting corn with a 16 row central-fill planter (JD
1770 NT, Deere & Co., Moline, IL).
CAN data were logged with a VectorTM CAN data logger (CANcase XL log, Vector, Stuttgart, Germany) and CANalyzer software installed on a laptop computer (see Figure 2). Data were logged from
both the implement and tractor channels of the CAN bus. Tractor
data from a total of six unique fields were collected. Machinery used
for the study, field names, specifications of the implements, and the
CAN data bus loads (%) of the tractors are summarized in Table 1.
CAN messages logs were imported into Excel for sorting and extraction of machine operating parameters. A screenshot of the CANalyzer interface with CAN messages can be seen in Figure 3. While
all messages were logged, only the SAE J1939 messages were considered for the study as the identifiers and data formats were readily available through the SAE J1939 database (SAE, 2013). The primary messages used in this investigation were the Electronic Engine
Controller 1 (EEC1 – CF00400hex – PGN 61444) and Liquid Fuel Economy (LFE – 18FEF200hex – PGN 65266), both highlighted in Figure
3. Data relevant to this investigation included in the EEC1 message
were actual engine torque in percent, and engine speed in rpm. The
LFE message provided the engine fuel use rate in L h–1. From Figure 3, it can be observed that the data in the CAN messages were in
hexadecimal format. Contents of the EEC1 and LFE message frame
in hexadecimal format and their respective message identifiers is
presented in Figure 4. The hexadecimal data in the messages were
converted to engineering units based on the conversion factors and
procedures available in the SAE J1939 database (SAE J1939, 2013).
The LFE message provided the engine fuel use rate in L h–1 with a
resolution of 0.05 L h–1 bit–1, whereas EEC1 message provided actual
percent engine torque with a resolution of 1.0% bit–1, and engine
speed in rpm with a resolution of 0.125 rpm bit–1. As an example, to
convert the hexadecimal data of the LFE message, D0 and D1 data
bytes of the LFE message (see Figure 4a) which corresponded to the
fuel use rate were converted into decimal numbers and combined
to yield bits. These combined bits of D0 and D1 (1123) were multiplied by 0.05 L h–1 bit–1 conversion factor to obtain fuel use rate in
L h–1. D0 and D1 data byte values of the LFE message shown in Figure 4a, yielded a fuel use rate of 56.15 L h–1. Similar procedure was
followed to decode data bytes D2, D3 and D4 of the EEC1 message
292
Pitla et al. in Computers and Electronics in Agriculture 121 (2016)
Figure 1. (a) 4WD tractor pulling the anhydrous applicator, (b) field cultivator pulled by the 4WD tractor and (c) MFWD tractor pulling the corn planter.
Figure 2. CAN bus hardware used for CAN data collection from tractor and implement bus.
Table 1. Summary of the equipment used, field names, and the CAN bus loads of the tractors.
Implement
Anhydrous applicator
Anhydrous applicator
Field cultivator
Field cultivator
Field cultivator
Central fill planter
Central fill planter
Central fill planter
Field name Implement
width (m)
Tractor used
3MSID
4A
1C
2D
2C
2D
1C
12D
4WD
4WD
4WD
4WD
4WD
MFWD
MFWD
MFWD
10
10
13.7
13.7
13.7
12.2
12.2
12.2
which yielded a percent torque of 49% and an engine speed of
1781.25 rpm (see Figure 4b). It was required to apply an offset of
–125% to the converted percent torque data as a last step to obtain
correct percent torque values. For example, the percent torque data
byte D2 of the EEC1 message (see Figure 4b) when converted into
Channel I (tractor bus)
Channel II (implement bus)
Baud rate (kbps) Bus load (%)
Baud rate (kbps)
Bus load (%)
500
500
500
500
500
500
500
500
250
250
250
250
250
250
250
250
18.8
18.4
19.2
19.2
19.2
22.2
22.5
23.6
32.7
32.1
32.3
32.4
32.3
35.5
35.0
35.4
percent torque with a conversion factor of 1.0% bit–1 yielded a percent torque value of 174%. An offset of –125% was applied to yield
the correct percent torque of 49%. Based on the time stamps of the
messages, it was observed that LFE messages were generated at a
rate of 10 Hz while EEC1 messages were generated at a rate of 100
In-field fuel use and load states of agricultural field machinery
293
Figure 3. Screenshot of the CAN messages captured using the Vector CANalyzer software.
Hz. Fuel use rate, engine speed and % torque messages were subsampled and registered to create a data file with a time base of 10
Hz for the duration of each field operation. In addition to CAN data,
National Marine Electronics Association (NMEA) global positioning
(GPS) data was obtained from Trimble’s FMX display (Trimble Navigation, Ltd., Sunnyvale, California) installed in the tractors. Trimble’s
AutopilotTM (Trimble Navigation, Ltd., Sunnyvale, California) was used
for automated guidance during field operations which provided Real
Time Kinetic (RTK) GPS. Ground speed, latitude and longitude information obtained from Trimble hardware was combined with CAN
message data for tractor path analysis.
4. Results and discussion
The fuel rate profile for NH3 application presented in Figure 5 illustrates the periods of high and low fuel consumption rates. The
peaks corresponded to the parallel and headland passes when
the implement was engaged in the soil, whereas the valleys represented end-of-row tractor turns. High fuel rate regions were separated by low fuel rate regions in a sequence which correlated to
parallel and headland passes, and end-of-row turns. In addition to
the peaks and valleys, between times 3500 s and 3700 s there was a
period of time when the fuel rate dropped to less than 10 L h–1. This
Figure 4. (a) Decoded Liquid Fuel Economy (LFE) message data and (b) Decoded Electronic Engine Controller (EEC1) message data.
Figure 5. Fuel rate profile of the tractor for NH3 application in Field 3MSID.
294
Pitla et al. in Computers and Electronics in Agriculture 121 (2016)
corresponded to the idling of the tractor indicating that the application was stopped by the operator while waiting to exchange NH3
tanks. Thus, fuel rate profile revealed the working and idle states
during field operations. It was observed that the valleys and peaks
of the CAN fuel rate profile correlated well with the tractor working states in turns and straight passes, respectively (see Figure 6).
The corresponding fuel rate frequency distribution of the tractor
is shown in Figure 7. There were three distinct distribution regions
within this frequency plot. The first tractor state (TS-I) was an idle
state where the tractor did not perform any useful work. The tractor
experienced low fuel rates (approx. less than 10 L h–1) during TS-I as
the engine was lightly loaded. The second tractor state (TS-II) corresponded to the end-of-row turns, whereas the third state (TS-III)
represented parallel and headland passes of the tractor where the
NH3 applicator was engaged with the soil. Differentiation between
the tractor states was performed graphically followed by confirmation with the threshold fuel rates of implements. Threshold fuel
rates correspond to the fuel use rate of the tractor when the implement is engaged in the soil and is a function of the draft force, implement width and the speed of operation (Pitla et al., 2014). TS-III
of the NH3 applicator covered significant portion of the frequency
distribution indicating that the applicator was in a working state for
a significant portion of the field operation. Further, fuel use rate values corresponding to TS-III were higher compared to other states
(TS-I and TS-II) indicating that the tractor was loaded heavily when
contrasted with the other two states (see Figure 7). In addition to
fuel use rate, percent torque is another good indicator of engine
load and hence the percent engine torque frequency distribution
was plotted (see Figure 8). The percent torque distribution profile
trends similar to the fuel use rate distribution as anticipated exhibiting three distinct regions corresponding to idle state (TS-I), end-ofrow turns (TS-II), and working state in parallel and headland passes
(TS-III) for the NH3 applicator.
Fuel use rate was plotted against the power (kW) to confirm the
validity of the CAN bus data (see Figure 9). A coefficient of determination of 0.96 was obtained indicating a good fit between the power
requirements of the drawn implement and the tractor fuel use rate.
Weighted averages of the fuel use rate, percent torque and engine
speed were calculated based on their occurrences for each of the
three states in accordance with Eq. (1)
Xwavg =
∑ ni=1 xi fi
∑ ni=1 fi
(1)
where
Xwavg is the weighted average of the parameter under consideration: fuel use rate, percent torque and engine speed.
i
is the index of the data point.
n
is the number of data points in each tractor state.
xi
fi
is the current data point.
is the frequency of the current data point.
As an example, to obtain weighted average fuel use rate in idle
state (TS-I), the frequency of fuel use rate value was multiplied by
the corresponding fuel use rate and a summation was performed
for the distribution falling within TS-I. The resultant value was divided by the sum of frequencies in the TS-I (idle state) to yield a
weighted average value of 6.28 L h–1 (see Figure 10). A similar procedure was followed to obtain weighted averages of percent torque
and engine speed (rpm). The percentage of field time the machine
operated in each tractor state was calculated based on the duration
of each state relative to the overall field time (all three states combined). Percent weighted torque, weighted fuel use rate, total fuel
consumption and the percent of time expended by the tractor in
each of the states (TS-I, TS-II and TS-III) during NH3 application in
field 3MSID are presented in Figure 10.
From Figure 10 it can be seen that for 10 min, which corresponded to 7% of the total work time, the NH3 applicator was using 8% of the rated torque, consuming 6.28 L h–1 with a total fuel
consumption of 1.06 L during the idle state. Total fuel consumption
in liters for each working state of the NH3 applicator was calculated
based on the data collection rate of 10 Hz (data point every 0.1 s)
and the fuel use rate in L h–1 at each data point. During TS-II, which
lasted for 33 min (24% of total time), the percent weighted torque
was 23%, fuel rate was 20.39 L h–1 and the total fuel consumption
was 10.73 L. For TS-III, which corresponded to parallel and headland passes, the NH3 applicator used 52% of rated torque at a fuel
rate of 52.71 L h–1. TS-III took 96 min (69% of the total working time)
and used 91.10 L of fuel during this state indicating that TS-III was
the high energy demand state of the NH3 applicator. Perhaps the
best opportunities for energy savings may be realized in this work
state as it corresponded to the high energy consumption state of
the tractor operation. The procedure discussed above was repeated
to obtain weighted percent torque, weighted fuel rate and percent
of total field time of each tractor state during field cultivation and
planting operations. NH3 application and field cultivation operations
were represented by three distinct working states, whereas planting
had only two distinct tractor states. For planting operation, working
state in parallel and headland passes was not distinguishable from
the end-of-row turns. Percent torque and fuel use rate distribution
plots for corn planting in field 12D can be seen in Figures 16a and
11b, respectively.
TS-I is the idle state of the tractor, whereas TS-II corresponded to
combined working state in turns, headland and parallel passes. Unlike NH3 application and field cultivation, where only drawbar power
of the tractor was used, planting operation used drawbar, PTO and
hydraulic power of the tractor, simultaneously. The central fill planter
Figure 6. Turns and straight passes compared to the valleys and peaks of fuel rate profile.
In-field fuel use and load states of agricultural field machinery
295
Figure 7. Fuel rate distribution of the tractor for anhydrous NH3 application in Field 3MSID.
Figure 8. Percent torque distribution for NH3 application in field 3MSID.
Figure 9. Correlation between fuel use rate and the power (kW) of the
4WD tractor in field 3MSID.
Figure 10. Weighted average percent torque, weighted average fuel use
rate and fuel consumption of NH3 application in Field 3MSID for tractor states (TS-I, TS-II and TS-III).
used the hydraulic-drive fans to transfer seed from the central tank
to the individual row units and to provide airflow at the seed meters for singulation. The planter used power from multiple sources
of the tractor there by placing high demand on the engine. Even
when the planter was not engaged in the soil the tractor was still
under load due to power drawn from the PTO and hydraulic selective control valves. This could be one of the reasons why there was
no distinct separation in the percent torque frequency plot between
working state in turns (planter row units disengaged) and working
state in parallel and headland passes (planter row units engaged).
For planting in field 1C, the tractor was in idle state for a significant portion of time. This was evident from high frequencies of percent torque values between approximately 25% and 45% torque values in Figure 12a. The idle state of the tractor is also represented in
percent torque profile of the tractor in Figure 12b.
The highlighted areas in Figure 12b correspond to idle state of
the tractor where the percent torque values were low for extended
periods of times indicating that the tractor was stopped frequently.
Based on the field notes taken it was confirmed that the operator
stopped the tractor frequently to check the seed spacing and depth
within the furrow. As field 1C was the first field to be planted, the
operator evaluated the quality of seed placement on multiple occasions during which time the tractor stopped while the operator was
making adjustments.
Weighted average percent torque, weighted average fuel rate,
and percent total field time of the 4WD tractor operating the NH3
applicator and field cultivator in each tractor state (TS-I, TS-II and
TS-III) are summarized in Table 2. On an average the 4WD tractor
worked approximately 70% of the field time in TS-III for both NH3
application and field cultivation operations indicating that the tractor
was in high load state for majority of the field time (see Table 2). For
NH3 application, the weighted average percent torque in TS-III was
296
Pitla et al. in Computers and Electronics in Agriculture 121 (2016)
Figure 11. (a) Percent torque distribution and (b) fuel use rate distribution for planting in field 2D.
Figure 12. (a) Percent torque distribution and (b) percent torque profile for planting in field 1C.
Table 2. Weighted average percent torque and weighted average fuel use rate of 4WD tractor during anhydrous NH3 application and field cultivation in TS-I, TS-II and TS-III.
Field operation
NH3 Application
Field Cultivation
Field
name
Avg. ground
Weighted average torque (%)
speed (km h–1) TS-I
TS-II TS-III
Weighted average fuel rate (L h–1)
% of total field time
TS-I
TS-II
TS-III
TS-I
TS-II
TS-III
3MSID
4A
1C
2C
2D
7.45
7.91
8.08
9.17
8.80
6.28
6.46
13.20
11.66
7.11
20.39
23.14
31.69
33.18
30.91
52.71
54.00
66.57
74.55
65.61
8
8
4
3
3
22
21
25
27
24
70
71
71
70
73
8
9
13
13
13
23
24
33
35
33
52
50
66
68
63
In-field fuel use and load states of agricultural field machinery
approximately double the percent torque values of TS-II whereas, the
fuel use rate values in TS-III were approximately two and half times
the fuel use rate values occurring in TS-II (see Table 2). For field cultivation, in all the fields, weighted percent torque and the fuel use rate
in TS-III were found to be approximately twice that of the weighted
torque and fuel use rate values of TS-II. As expected the weighted
average percent torque and fuel use rates were relatively low in TS-I
states where the tractor was in idle state. These results clearly indicate that the load and fuel demands on the tractor change significantly within the field depending on whether the implement is
engaged in the soil or not. For field 2C during field cultivation, the
weighted fuel use rate of 74.55 L h–1 in TS-III was much higher than
the fuel use rate in field 1C and field 2D as the tractor was travelling at a higher ground speed of 9.2 km h–1 compared to the ground
speeds of 8.1 and 8.8 km h–1 in fields 1C and 2D, respectively.
Rated power (4WD: 245 kW and MFWD: 127 kW) and rated engine speeds (4WD and MFWD: 2100 rpm) obtained from the official Nebraska tractor test reports of the tractor models were used
to determine the actual rated torque values of the tractors. The
weighted percent torque obtained from CAN data in each state was
converted to torque in N m based on the rated torque value. Percent weighted torque, as summarized in Table 2, was converted to
weighted torque (N m) (see Table 3) based on the rated power and
engine speed of the 4WD tractor. The torque obtained was converted to power (kW) using the weighted engine speed in rpm of
each tractor state (see Table 3).
Highest load demand of 155.91 kW on the tractor was found to
be in TS-III of the field cultivation operation in field 2C, whereas the
lowest demand of 8.26 kW on the tractor was found to be in TS-I of
the NH3 application in field 3MSID. Weighted average fuel rate and
power were weighted averaged across the fields for NH3 application
and field cultivation based on the percent time of the tractor states
separately to yield overall tractor load states of each field operation.
For NH3 application, the weighted average loads across the fields in
TS-I, TS-II and TS-III were found to be 9.20 kW, 46.96 kW and 118.10
kW with tractor operating at an average ground speed of 7.64 km
h–1 (see Figure 13). These load values were divided by the rated engine power of the tractor.
These load values were divided by the rated engine power of
the tractor to yield the percent loads of the tractors in each state.
The weighted average fuel use rates in each of the three states were
found to be 6.38 L h–1, 21.67 L h–1 and 53.36 L h–1 (see Figure 13).
The load states of NH3 application indicated that high demand of
118.10 kW occurred for 70% of the field time, whereas medium load
and low load demands of 46.96 kW and 9.20 kW occurred for 22%
and 8% of the field times (see Figure 13). Similar procedure was followed to determine the weighted average fuel use rate and loads of
the field cultivation operation across three fields considered in the
study. Weighted average power and fuel use rates of the tractor in
three states for the field cultivation across the fields are summarized
in Figure 14. The 4WD tractor was under higher load during field cultivation relative to NH3 application based on the weighted average
297
Figure 13. Weighted average power and fuel use rate of the 4WD tractor
in TS-I, TS-II and TS-III across the fields (3MSID, 4A) for NH3 application.
Figure 14. Weighted average power and fuel use rate of the 4WD tractor in TS-I, TS-II and TS-III across fields (1C, 2C, 2D) for field cultivation.
power and fuel use rates in TS-I, TS-II and TS-III. For 71% of the field
time, the cultivator was exerting a load demand of 142.06 kW on the
tractor in TS-III, 63.24 kW for 25% of the field time and 20.33 kW
for 4% of the time in TS-II and TS-I, respectively. Thus, the actual infield load states of the 4WD tractor while pulling both the NH3 applicator and field cultivator in three tractor states were quantified.
The weighted percent torque and fuel use rate values of the
MFWD tractor during planting operation in three fields are summarized in Table 4. Comparing the percent of total field times of tractor states, it is evident that in fields 2D and 12D for 96% and 98%
of the field time, respectively, the tractors were in TS-II which corresponded to a combined working state of parallel passes, headland passes and turns (see Table 4). However, for field 1C, the tractor spent 41% of the time in TS-I where the tractor was in an idle
state, and the remaining 59% of the time the tractor was working in
parallel passes, headland passes and turns.
As discussed earlier, the tractor was stopped on multiple occasions by the operator in field 1C to evaluate seed depth and spacing which resulted in multiple idle time periods. The fuel use rate
and percent torque in TS-II were more than double compared to
the fuel rate and percent torque in TS-I (see Table 5) confirming the
fact that the tractor went through distinct load states in the field.
The weighted torque (N m) and power (kW) required to operate
the planter in TS-I and TS-II are summarized in Table 5. The tractor in field 1C consumed the least power in both tractor states as it
was not loaded for significant amount of time because of extended
non-working periods. Similar to NH3 application and field cultivation,
Table 3. Weighted average torque and weighted average power of 4WD tractor during anhydrous NH3 application and field cultivation in TS-I, TSII and TS-III.
Field operation
NH3 application
Field cultivation
Field name
3MSID
4A
1C
2C
2D
Weighted average torque (N m)
Weighted average engine speed (rpm)
Weighted average power (kW)
TS-I
TS-II
TS-III
TS-I
TS-II
TS-III
TS-I
TS-II
TS-III
89
100
145
145
145
257
268
368
390
368
580
558
736
758
703
884
955
1455
1543
905
1711
1718
1568
1696
1541
2009
1959
1767
1964
1827
8.26
10.03
22.08
23.42
13.71
45.95
48.13
60.40
69.30
59.36
121.95
114.34
136.14
155.91
134.37
298
Pitla et al. in Computers and Electronics in Agriculture 121 (2016)
Table 4. Weighted average torque and weighted average fuel use rate of MFWD tractor for planting operation in TS-I and TS-II.
Field operation
Corn Planting
Field name
1C
2D
12D
Avg. ground
speed (km h–1)
6.14
7.16
7.83
Weighted % torque
Weighted fuel rate (L h–1)
% of total field time
TS-I
TS-II
TS-I
TS-II
TS-I
TS-II
37
36
35
76
81
82
10.15
12.37
11.80
24.70
27.50
27.74
41
4
2
59
96
98
Table 5. Variation of load on the MFWD tractor for planting operation in TS-I and TS-II.
Field operation
Corn planting
Field name
1C
2D
12D
Weighted engine torque (N m)
Weighted engine speed (rpm)
Weighted engine power (kW)
TS-I
TS-II
TS-I
TS-II
TS-I
TS-II
213
208
202
438
467
473
904
1433
1492
1514
1532
1545
20.19
31.13
31.51
69.44
74.89
76.45
Figure 15. Weighted average power and fuel use rate of the MFWD tractor in TS-I and TS-II for planting operation.
weighted averages of power and fuel use rated were performed
across fields 2D and 12D based on the percent field time in each
tractor state to estimate representative load states for planting operation. Field 1C was not considered in the weighted average calculation to avoid inaccuracies in the weighted average values (see
Table 5). For fields 2D and 12D, the planting operation used approximately 75.68 kW in TS-II and 31.36 kW in TS-I, whereas an average
fuel use rate of 27.62 L h–1 and 12.18 L h–1 resulted in TS-II and TSI, respectively (see Figure 15).
Weighted power and fuel use rates in TS-I, TS-II and TS-III were
weighted averaged based on the percent field times of each state
to determine a composite power and fuel use rates of the NH3 application and the field cultivation operations. A comparison of the
power consumed by the implements, load percentages and the fuel
use rates of the tractors pulling the NH3 applicator, field cultivator
and the planter is presented in Figure 16. It should be noted that
the field cultivation and NH3 application were performed by a 4WD
tractor with 245 kW rated power and a MFWD tractor with 127 kW
rated power was used for planting. From comparison it can be seen
that the field cultivator had the highest demand on the tractor in
terms of power consumed and the fuel use rate. From Figure 16
it can be seen that the average load on the 4WD for NH3 application operating at a ground speed of approximately 7.64 km h–1 was
found to be 93.74 kW with a fuel use rate of 42.63 L h–1. The 4WD
tractor was loaded on an average by 38% of the rated power during this operation. Also, the tractor was loaded more optimally for
planting operation at 59% engine load compared to 38% for NH3
application and 48% for field cultivation indicating that the tractors
were oversized in the latter two cases. The power and fuel use rate
were divided by the number of rows, 16 in the case of planting operation and number of tools, 13 in the case of NH3 applicator, and
90 for the field cultivator, to yield specific power used by the implements. The specific power and fuel rate values of implements presented in Figure 17 provides a comparison of the load demands of
the three field operations. On a per tool basis, NH3 application demanded the highest power of 7.21 kW tool–1, whereas the field cultivator demanded the lowest (1.31 kW tool–1). One of the reasons
for high power requirement for the NH3 applicator is the working
depth of 22 cm compared to field cultivator which was operated at
a depth of 7 cm (secondary tillage). The specific power requirements
and fuel use rates of field cultivation and planting obtained from the
CAN data were compared to the specific power requirements and
fuel use rates suggested by the ASABE standards (ASABE, 2011a).
No suggestions were made for NH3 application in the ASABE
standards and hence specific power requirements obtained in this
study for NH3 application were compared to the NH3 application
power requirements suggested by Godwin and O’Dogherty (2007).
Figure 16. Comparison of weighted average power and fuel use rates of NH3 application, field cultivation and planting operations.
In-field fuel use and load states of agricultural field machinery
299
Figure 17. Comparison of power and fuel use rates per row/tool for NH3 application, field cultivation and planting.
Table 6. CAN specific power and fuel use rate comparison to specific power and fuel use rate values obtained from equations of ASABE standards
and other research studies.
Anhydrous application
Specific power (kW tool–1 or kW row–1)
Specific fuel use rate (L
h–1
tool–1
or L
h–1
row–1)
Field cultivation
Corn planting
CAN data Godwin & O’Dogherty (2007) CAN data ASABE (2011a)
CAN data ASABE (2011a)
7.21
5.96
1.31
1.34
4.65
4.37
3.28
3.12
0.64
0.56
1.70
1.72
From Table 6 it can be seen that the specific power and specific fuel
rate of NH3 application determined from CAN data were slightly over
estimated compared to the specific power and fuel rate values suggested by Godwin and O’Dogherty (2007). For cultivation and planting, specific power and fuel rate values obtained from the CAN data
were closer to the specific power and fuel rate values suggested by
the ASABE standards (2011a), but no particular trend was observed.
Extensive data have to be collected from a large sample of tractors
and fields to further investigate the comparison of specific power
and fuel rate values obtained from machine CAN data to the power
requirements of the implements suggested by the ASABE standards
and other research studies.
5. Conclusions
In-field tractor data was successfully collected from tractors performing NH3 application, field cultivation and planting operations.
It was evident from the load and fuel use rate data that multiple
tractor load states with different fuel use and load demand magnitudes existed within a given field for the same operation. Based on
the fuel use rate and percent torque frequency distributions, NH3
application and field cultivation were found to have three distinct
work states: TS-I (idle state), TS-II (working state in turns) and TS-III
(working state in parallel and headland passes). Planting operations
were divided into two tractor states TS-I (idle state) and TS-II (working state in parallel passes, headland passes and turns) as it was not
possible to separate turns from parallel and headland passes in the
frequency distribution. Identification of different tractor load states
provided an insight into the actual power needs of the implements
during working and non-working periods. The 4WD tractor, on average, was loaded 38% for NH3 application and 48% for field cultivation indicating that a less rated power tractor model could have
been used. NH3 application and field cultivation, on average, required a specific power of 7.21 and 1.31 kW tool–1, and fuel use rates
of 3.28 and 0.64 L h–1 tool–1at 7.64 km h–1 and 8.68 km h–1 grounds
speeds, respectively. The planter was better matched to the tractor
when contrasted with the NH3 applicator and field cultivator noting
the tractor load of 59% of the rated power during the planting operation. The planter required an average specific power of 4.65 kW
h–1 row–1 and a fuel use rate of 1.70 L h–1 row–1 at a ground speed
of 7.04 km h–1. The specific power needs and the fuel use rates determined in this study can be used to update power sizing, and fuel
rate estimation procedures of the ASABE machinery management
standards. Data corresponding to typical field times at a certain
tractor load, power and fuel rate requirements of the implements
can be used by manufacturers and producers for estimating actual
power and fuel requirements of production operations, and for improving the field and fuel efficiencies of the machinery. Off-road vehicle emission studies will benefit from the actual load state values
(power and fuel rate) of the tractors. Further, standardized tractor
tests, which currently use steady state test conditions only, could use
the tractor load state information presented for developing testing
procedures reflecting actual field conditions.
Acknowledgments — The authors would like to thank Nathan Douridas, Chuck Gamble, and Matt Sullivan at the Farm Science Review, The
Ohio State University, for their support during in-field CAN data collection from agricultural field operations. Help from students Andrew Klopfenstein, Dustin Wolters, Brittany Schroeder, and Karl Klopfenstein at the
Department of Food, Agricultural and Biological Engineering at The Ohio
State University is appreciated.
References
Al-Suhaibani, S.A., Al-Janobi, A., Al-Majhadi, Y.N., 2010. Development and evaluation of tractors and tillage implements instrumentation system. Am. J. Eng.
Appl. Sci. 3 (2), 363–371.
ASABE, 2011a. ASAE D497.7 MAR2011 Agricultural Machinery Management Data.
ASABE Standards 2011. St. Joseph, MI.
ASABE, 2011b. ASAE EP496.3 FEB2006 Agricultural Machinery Management. ASABE Standards 2011. St. Joseph, MI.
Burgun, C., Lacour, S., Delacroix, A., Descombes, G., Doyen, V., 2013. Computing
time and fuel requirements to assess efficiency of a field work from conventional laboratory tests: application to a plowing operation. Agric. Food Sci. 22
(2), 247–261.
Cupera, J., Sedlak, P., 2011. The use of CAN-bus messages of an agricultural tractor for monitoring its operation. Res. Agric. Eng. 57 (4), 117–127. Darr, M.,
2012. CAN bus technology enables advanced machinery management. Resour. Mag. 19 (5), 10–11.
Godwin, R.J., O’Dogherty, M.J., 2007. Integrated soil tillage force prediction models. J. Terramech. 44 (1), 3–14. doi:10.1016/j.jterra.2006.01.001
300
Pitla et al. in Computers and Electronics in Agriculture 121 (2016)
Harvey, C., 2003. EPA’s newest draft nonroad emission inventory model. In: 12th
International Emission Inventory Conference ‘‘Emission Inventories-Applying
New Technologies”, San Diego, California. April 28–May 1, 2003.
ISO, 2006. ISO 8178-1 Reciprocating Internal Combustion Engines – Exhaust Emission Measurement – Part 1: Test-bed Measurement of Gaseous and Particulate
Exhaust Emissions. ISO: CH-1211 Geneva 20 Switzerland.
Lin, N., 2014. Prediction of Wheeled tractor Fuel Use Rate Distributions from CANbus Data for Agricultural Field Operation. Master’s Thesis. The Ohio State
University.
Marx, S.E., 2015. Controller Area Network (CAN) Bus J1939 Data Acquisition Methods and Parameter Accuracy Assessment Using Nebraska Tractor Test Laboratory Data (Master’s thesis). University of Nebraska-Lincoln, Lincoln, NE.
Marx, S.E., Luck, J.D., Hoy, R.M., Pitla, S.K., Darr, M.J., Blankenship, E., 2015. Validation of machine CAN Bus J1939 fuel rate accuracy using Nebraska Tractor Test
Laboratory fuel rate data. Comput. Electron. Agric. 118, 179–185.
OECD, 2012. Standard Code for the Official Testing of Agricultural and Forestry
Tractor Performance. OECD. http://www.oecd.org/tad/code/Code%202%20
-%20Final.pdf
Pitla, S.K., Lin, N., Shearer, S.A., Luck, J.D., 2014. Use of controller area network (CAN)
data to determine field efficiencies of agricultural machinery. Appl. Eng. Agric. 30 (6), 829–839.
Pitla, S.K., Luck, J.D., Shearer, S.A., Lin, N.N., Schroeder, B.A., Klopfenstein, A.A., 2013.
Work and load performance profiles for agricultural field machinery. In: Conference Agricultural Engineering 13, VDI-Meg, Land-Technik, 29-36 VDIBerichte Nr 2193-2013.
SAE, 2013. Standard J1939: Serial Control and Communications Heavy Duty Vehicle Network – Top Level Document. http://standards.sae.org/j1939_201308/
Yahya, A., Zohadie, M., Kheiralla, A.F., Giew, S.K., Boon, N.E., 2009. Mapping system for tractor-implement performance. Comput. Electron. Agric. 69, 2–11.