Performance Evaluation of Electrically Driven Agricultural Implements Powered by an External Generator

: In the last decade, many studies have been conducted on tractor and agricultural machinery electriﬁcation. In particular, the electriﬁcation of power take-off (PTO)-powered implements could support many beneﬁts, such as improved comfort and safety during implement connection, less noisiness, accurate control of the implement rotational speed, and fuel consumption reduction. However, commercially available tractors do not generate sufﬁcient electric power to run electriﬁed implements. A solution to this issue is powering eventual electriﬁed implements with an external electric generator powered by the PTO and mounted with the front three-point linkage. This study aimed to evaluate the potential beneﬁts of using this combination with respect to PTO powered implements. The types of implements analyzed in detail in this study were a sprayer and a mulcher. Field tests were performed acquiring performance, operational, and environmental parameters. Results show that on the electriﬁed implements, the absence of the cardan shaft and hydraulic remotes shortened the time required for the hitching phase and reduced the in-work noisiness. Field tests demonstrated that the electriﬁed implements permitted an improvement of the fuel consumption per hectare, up to 33.3% and 29.8% lower than their PTO-powered homologue for the sprayer and the mulcher, respectively.


Introduction
The agricultural sector is responsible for about 21% of the world s greenhouse gas emissions, mainly due to the use of fossil-based fertilizers, the combustion of biomass, and the polluting gas emissions of agricultural machinery [1]. Regarding the latter, the vast majority of self-propelled agricultural machinery is powered by endothermic engines run with fossil fuels, which represent 95% of the energy used for their propulsion [2]. European governments have demanded a reduction in fossil fuel consumption and manufacturers of agricultural machines are concentrating a lot of efforts on increasing the efficiencies of their machines [3,4]. Over the last two decades, agricultural machinery manufacturers have been focused on engines and drivelines [5], but recently they have been concentrating on other limiting components, such as accessories [6] and tractive performance [7]. The power take-off (PTO) is a component that can lead to operational inefficiencies since it must operate at 540 and 1000 rpm in function of the attached implement [8]. These speeds are reached at specific engine speeds in function of PTO operating mode: standard or economy. Generally, the standard and economy PTOs are located at the speeds where the engine delivers the peak power and torque, respectively. If this is reasonable when the engine operates at full load, in partial loads, the engine may not operate at the most efficient point since it operates to greater speeds than strictly necessary, which can lead to accessory overdrive [6]. The best option would be to design a PTO drive able to run the engine at the lowest speed, where it can deliver enough power for the operation. That is achievable with constant speed PTOs, which are able to maintain the PTO speed at

e-Source Specification
The front three-point linkage (TPL) of the tractor carried an electric generator, called e-Source in the following, developed by CNH Industrial (CNH Industrial N.V., Amsterdam, NL, USA) ( Figure 1).

e-Source Specification
The front three-point linkage (TPL) of the tractor carried an electric generato e-Source in the following, developed by CNH Industrial (CNH Industrial N.V., dam, NL, USA) ( Figure 1). The e-Source was run by the front PTO of the tractor, then the developed energy was modulated by a high voltage inverter controlled by the generator e control unit (GECU). The connection to the implements was granted by a stand high-voltage power interface (AEF HVPI in the following) [15]. The operator cou itor and control the operation of the device through the virtual terminal of the tra voltage output ( ) was monitored by an embedded sensor, the output signal quired by a CAN-Bus datalogger described in the following. Further e-Source s tions are reported in Table 2. The e-Source was run by the front PTO of the tractor, then the developed electric energy was modulated by a high voltage inverter controlled by the generator electronic control unit (GECU). The connection to the implements was granted by a standard AEF high-voltage power interface (AEF HVPI in the following) [15]. The operator could monitor and control the operation of the device through the virtual terminal of the tractor. The voltage output (V eS ) was monitored by an embedded sensor, the output signal was acquired by a CAN-Bus datalogger described in the following. Further e-Source specifications are reported in Table 2.

Implement Specifications
Mulchers and sprayers specifically designed to work in orchards and vineyards were used in this study. For each implement type, the PTO and electric-powered versions were tested, both were equal in terms of external dimensions and machine capacity. All the implements were developed and produced by Nobili S.p.a. (Molinella, Bologna, Italy).

Sprayers
The PTO-driven and electrically driven sprayers are, in the following, denoted mSprayer and eSprayer, respectively. These are shown in Figure 2 and their main specifications are reported in Table 3.

Implement Specifications
Mulchers and sprayers specifically designed to work in orchards and vineyards used in this study. For each implement type, the PTO and electric-powered versions tested, both were equal in terms of external dimensions and machine capacity. A implements were developed and produced by Nobili S.p.a. (Molinella (BO), Italy).

Sprayers
The PTO-driven and electrically driven sprayers are, in the following, de mSprayer and eSprayer, respectively. These are shown in Figure 2 and their main sp cations are reported in Table 3.   One can note that the main differences between the two sprayers are the power drive and a slight difference of mass caused by the high-voltage electric components (e.g., the electric motor) installed on the eSprayer. Indeed, on the mSprayer the power required by the implement rotor is provided by the cardan shaft attached to the rear PTO of the tractor, while on the eSprayer the rotor is driven by an electric motor powered by the e-Source through the AEF HVPI.
The transmission ratio between the fan and the PTO speed on the mSprayer could be set on two different gears, these are selectable through a mechanical lever positioned on the implement. This does not permit the mSprayer to change the gear on the go. The e-Sprayer does not have any gearbox, the fan speed (n f ) is independent of the engine speed ( Figure 3) and can be changed on the go with a remote control installed inside the tractor cab. One can note that the main differences between the two sprayers are the pow and a slight difference of mass caused by the high-voltage electric components electric motor) installed on the eSprayer. Indeed, on the mSprayer the power req the implement rotor is provided by the cardan shaft attached to the rear PTO of th while on the eSprayer the rotor is driven by an electric motor powered by the through the AEF HVPI.
The transmission ratio between the fan and the PTO speed on the mSprayer set on two different gears, these are selectable through a mechanical lever posit the implement. This does not permit the mSprayer to change the gear on the g Sprayer does not have any gearbox, the fan speed ( ) is independent of the engi ( Figure 3) and can be changed on the go with a remote control installed inside th cab. , * data can be used as surrogate data for the fan speed. One can note th eSprayer, , * was nearly constant and consequently fan speed was independent of The two implements are equipped with the same hydraulic pump, howeve eSprayer it is electrically driven, while on the mSprayer it is PTO driven. The de different fan drive required a complete redesign of the fan of the eSprayer in contain the fan drive, which is bulkier for the eSprayer than that of the mSpray Figure 3. Power required by both sprayers (P impl, * ) with respect to the engine speed (n eng ) during engine ramping. For both sprayers, the tests were carried out with the same fan nominal speed (i.e., 1800 rpm). P impl, * data can be used as surrogate data for the fan speed. One can note that for the eSprayer, P impl, * was nearly constant and consequently fan speed was independent of n eng . The two implements are equipped with the same hydraulic pump, however, on the eSprayer it is electrically driven, while on the mSprayer it is PTO driven. The design of a different fan drive required a complete redesign of the fan of the eSprayer in order to contain the fan drive, which is bulkier for the eSprayer than that of the mSprayer. However, this bulkiness is counterbalanced by the high efficiency granted by the installed electric motor, indeed the efficiency coefficient of this motor typology varies from 92% to 97% [23]. A standard double cardan shaft could reach even higher values of mechanical efficiency, though these values could heavily drop when there is a remarkable height difference between the tractor PTO and the implement splined shaft. Indeed, in the latter case, the joints of the cardan shaft would form huge angles with the input and output shafts, reducing the mechanical efficiency [24]. Moreover, if the angles that the cardan shaft form with the PTO of the tractor and the splined shaft of the implement are different, a perfectly constant transmission ratio is not reachable [25].
A remarkable difference between the two implement versions is the fan design, which is the same in diameter but with different fan blades. Especially, the eSprayer is equipped with serrated trailing edge blades, which are typically adopted for reducing the noise [26] ( Figure 4).
Agronomy 2021, 11, x FOR PEER REVIEW ever, this bulkiness is counterbalanced by the high efficiency granted by the installed tric motor, indeed the efficiency coefficient of this motor typology varies from 92% to [23]. A standard double cardan shaft could reach even higher values of mechanica ciency, though these values could heavily drop when there is a remarkable height d ence between the tractor PTO and the implement splined shaft. Indeed, in the latter the joints of the cardan shaft would form huge angles with the input and output s reducing the mechanical efficiency [24]. Moreover, if the angles that the cardan shaft with the PTO of the tractor and the splined shaft of the implement are different, a per constant transmission ratio is not reachable [25].
A remarkable difference between the two implement versions is the fan de which is the same in diameter but with different fan blades. Especially, the eSpray equipped with serrated trailing edge blades, which are typically adopted for reducin noise [26] (Figure 4).

Mulchers
The PTO-driven and electrically driven mulchers are, in the following, den mMulcher and eMulcher, respectively. Both were used for the test activities shown in ure 5, and their main specifications are reported in Table 4.

Mulchers
The PTO-driven and electrically driven mulchers are, in the following, denoted mMulcher and eMulcher, respectively. Both were used for the test activities shown in Figure 5, and their main specifications are reported in Table 4.
The differences in terms of power drive and implement mass already observed for the sprayers are also present for the mulchers.
Consequently, on the mMulcher, the rotor speed is dependent on the PTO speed, controlled by a fixed ratio through a belt transmission, although on the eMulcher, the rotor speed is controlled by a remote control installed inside the tractor cab.

Sensors and Acquisition System
Tractor CAN-Bus data were recorded through a stand-alone CAN-Bus data-logger optimized by CNH Industrial according to the approach introduced by Molari et al. [27].
In particular, the CAN-Bus signals with the Suspect Parameter Numbers (SPNs) and parameter numbers group (PNGs) [28,29] reported in Table 5 were used for the analysis.

Mulchers
The PTO-driven and electrically driven mulchers are, in the following, d mMulcher and eMulcher, respectively. Both were used for the test activities shown ure 5, and their main specifications are reported in Table 4.   Engine torque as a per cent of reference engine torque; the value includes the torque developed in the cylinders required to overcome friction.

M eng% (−)
Nominal friction percent torque 514 5398 Torque contribution of frictional and thermodynamic losses of the engine itself, pumping torque loss and the losses of fuel, oil, and cooling pumps.
Engine fuel rate 183 65266 Amount of fuel consumed per unit of time.
The hourly fuel consumption ( . f ) is derived by the pulsations of the solenoid valve injector controlled by the vehicle electronic control unit (ECU), so some discrepancy may be found between calculated fuel rate and the actual fuel rate. However, noticeable differences between the CANBUS message and the actual fuel rate values are observable only during engine transient phases (up to 6.22%), while during steady phases or averaged analysis the error is minimal [30,31].
The tractor actual speed (V t ) and its geospatial position were acquired with a Global Navigation Satellite System (GNSS) receiver embedded in the data-logger. Moreover, auxiliary sensors were connected to the data-logger to record information that was not otherwise available in the tractor CAN-Bus network. In particular: Agronomy 2021, 11, x FOR PEER REVIEW during engine transient phases (up to 6.22%), while during steady phases or ave analysis the error is minimal [30,31]. The tractor actual speed ( ) and its geospatial position were acquired with a Globa igation Satellite System (GNSS) receiver embedded in the data-logger. Moreover, auxiliary sensors were connected to the data-logger to record inform that was not otherwise available in the tractor CAN-Bus network. In particular: The following performance parameters were obtained from the previously des signals: • The engine power ( ) was calculated with the Equation (1): • The e-Source input power ( , ) coming from front PTO was calculated wi Equation (2): The power absorbed by the PTO driven implements ( , ) was calculated Equation (3): The following performance parameters were obtained from the previously described signals: • The engine power (P eng ) was calculated with the Equation (1): • The e-Source input power (P eS,in ) coming from front PTO was calculated with the Equation (2): • The power absorbed by the PTO driven implements P impl,m was calculated with Equation (3): • The power absorbed by the electrified implements P impl,e was calculated with Equation (4):

Experimental Tests
In order to perform a comprehensive analysis of the performances of the e-Source and the electric powered implements, the experimental tests presented in the following sections were carried out.

Implement Hitching Time
The time needed by a professional farmer with several years of experience to mount and dismount each implement on the tractor linkages was measured to evaluate any operational benefit brought by the usage of an electrically powered implement. In fact, Agronomy 2021, 11, 1447 9 of 20 the usage of the AEF HVPI on the electric implements could allow a reduction of the hitching time since it replaces both the cardan shaft and the hydraulic remotes. The time necessary to mount and dismount each implement was measured with a stopwatch and each operation was repeated three times. Then, the mean value and the standard deviation of the three repetitions was calculated for each implement and operation.

Noise Tests
Noise tests were performed for each implement with the tractor maintained still and with the rear window open following the UNI EN ISO 11201 standard [32] using a Casella CEL-246 phonometer (IDEAL Corporate, Sycamore, IL, USA) (Figure 7).

Implement Hitching Time
The time needed by a professional farmer with several years of experience t and dismount each implement on the tractor linkages was measured to evaluate erational benefit brought by the usage of an electrically powered implement. In usage of the AEF HVPI on the electric implements could allow a reduction of the time since it replaces both the cardan shaft and the hydraulic remotes. The time n to mount and dismount each implement was measured with a stopwatch and ea ation was repeated three times. Then, the mean value and the standard deviatio three repetitions was calculated for each implement and operation.

Noise Tests
Noise tests were performed for each implement with the tractor maintained with the rear window open following the UNI EN ISO 11201 standard [32] using a CEL-246 phonometer (IDEAL Corporate, Sycamore, IL, USA) (Figure 7). The implements were tested with the same set up adopted for the field test the mMulcher and the mSprayer, both PTO modes were tested, while for the e and the eSprayer, was maintained at 1300 rpm. Moreover, for both spraye (i.e., 1300 and 1800 rpm) values were tested. Each test was performed for 30 s, sented results are the mean values of three repetitions performed in the same con

Field Tests
The sprayer field tests were performed in a 2 ha pear orchard with a row sp 4.1 m located in Cadriano (BO, Italy) at the Experimental Farm of the University gna (UNIBO) (Figure 8). The implements were tested with the same set up adopted for the field tests, so for the mMulcher and the mSprayer, both PTO modes were tested, while for the eMulcher and the eSprayer, n eng was maintained at 1300 rpm. Moreover, for both sprayers, both n f (i.e., 1300 and 1800 rpm) values were tested. Each test was performed for 30 s, the presented results are the mean values of three repetitions performed in the same condition.

Field Tests
The sprayer field tests were performed in a 2 ha pear orchard with a row spacing of 4.1 m located in Cadriano (BO, Italy) at the Experimental Farm of the University of Bologna (UNIBO) (Figure 8).
Both sprayers were used in the same operational settings; in particular, they were tested at the same nominal fan speed (n f ) (i.e., 1300 and 1800 rpm), pump pressure (i.e., 1.5 MPa), number of open nozzles (i.e., 6), and target V t (i.e., 5.5 km h −1 ). Regarding the mSprayer, tests were carried out with PTO operating at 540 rpm in both modes: standard (denoted as 540 in the following) and economy (denoted as 540E in the following). n eng was nearly 1900 rpm and 1500 rpm for 540 and 540E PTO modes, respectively. Since the n f on the eSprayer is independent of the n eng , the latter was set at the lowest value that simultaneously permits the engine to deliver consistently enough power to perform the operation. Preliminary tests performed by the e-Source manufacturer demonstrated that with n eng set at 1300 rpm, the aforementioned conditions were satisfied for the majority of the agricultural operation on plain ground. This set-up, theoretically, allows farmers to obtain the lowest possible specific fuel consumption since it allows for a reduction in the power demands from accessories [6]. Both sprayers were used in the same operational settings; in particular, they tested at the same nominal fan speed ( ) (i.e., 1300 and 1800 rpm), pump pressure 1.5 MPa), number of open nozzles (i.e., 6), and target (i.e., 5.5 km h −1 ). Regardin mSprayer, tests were carried out with PTO operating at 540 rpm in both modes: stan (denoted as 540 in the following) and economy (denoted as 540E in the following). was nearly 1900 rpm and 1500 rpm for 540 and 540E PTO modes, respectively. Sinc on the eSprayer is independent of the , the latter was set at the lowest value simultaneously permits the engine to deliver consistently enough power to perform operation. Preliminary tests performed by the e-Source manufacturer demonstrated with set at 1300 rpm, the aforementioned conditions were satisfied for the maj of the agricultural operation on plain ground. This set-up, theoretically, allows farme obtain the lowest possible specific fuel consumption since it allows for a reduction i power demands from accessories [6].
The mulcher tests were performed on a 3 ha agricultural field with old corn resid and tall grass (SimVin in the following, Figure 9) located in San Pietro Capofiume Italy). That field was chosen because it could reproduce a comparable environme vineyard interrow and because during the period of the tests, no orchards or viney with the desired conditions were available.
during the tests was maintained at around 3.1 km h −1 and the mulcher rotor s ( ) was set at 2220 rpm for both implements. For both mulchers, was set at the s values as those of the sprayers.
All the field tests were performed with the tractor air conditioning system turne to reproduce the most realistic working environment. The mulcher tests were performed on a 3 ha agricultural field with old corn residuals and tall grass (SimVin in the following, Figure 9) located in San Pietro Capofiume (BO, Italy). That field was chosen because it could reproduce a comparable environment of vineyard interrow and because during the period of the tests, no orchards or vineyards with the desired conditions were available.
Agronomy 2021, 11, x FOR PEER REVIEW 11 Figure 9. Field with old corn residuals and tall grass to simulate a vineyard interrow (SimVin during the eMulcher test.

Data Analysis
The data acquired from the field tests were elaborated with MATLAB software MathWorks, Inc, Natick, MA, USA). Firstly, the recorded signals were interpolated a Vt during the tests was maintained at around 3.1 km h −1 and the mulcher rotor speed (n r ) was set at 2220 rpm for both implements. For both mulchers, n eng was set at the same values as those of the sprayers.
All the field tests were performed with the tractor air conditioning system turned on to reproduce the most realistic working environment.

Data Analysis
The data acquired from the field tests were elaborated with MATLAB software (The MathWorks, Inc, Natick, MA, USA). Firstly, the recorded signals were interpolated at 0.1 s through a spline interpolation algorithm. Then, the passes were separated from the headland turns, observing the rate of change of P impl, * calculated with signal differentiation (Figure 10).

Data Analysis
The data acquired from the field tests were elaborated with MATLAB software MathWorks, Inc, Natick, MA, USA). Firstly, the recorded signals were interpolated s through a spline interpolation algorithm. Then, the passes were separated from headland turns, observing the rate of change of , * calculated with signal differ tion ( Figure 10). After that, the mean values and the standard deviations of , , , * , and acquired during the passes were calculated; the average values along p were denoted with the overbar (i.e., represents the raw signal, while represen average value during the passes). The e-Source efficiency ( ) was calculated with E tion (5): The following operational and environmental indexes were calculated, consid the data acquired during the headland turns as well: Figure 10. Portion of tractor speed (V t ), and power absorbed by the mSprayer (P impl,e ) during an acquisition (a). Portion of tractor speed (V t ), and power absorbed by the eSprayer (P impl,e ) during an acquisition (b). One can note the effectiveness of the pass selection. The V t was mostly constant during headland turns for the large headlands that permitted the machinery to turn with no braking or changing of throttle position. A similar methodology was adopted for both mulchers.
After that, the mean values and the standard deviations of n eng , V t , P impl, * , P eS,in , . f , and P eS acquired during the passes were calculated; the average values along passes were denoted with the overbar (i.e., V t represents the raw signal, while V t represents the average value during the passes). The e-Source efficiency (η eS ) was calculated with Equation (5): The following operational and environmental indexes were calculated, considering the data acquired during the headland turns as well:

•
The field capacity (F cap ) was calculated with Equation (6): where b is the implement width for the mulchers or the distance between rows for the sprayers • The fuel consumption per hectare ( f ha ) was calculated with Equation (7): f /F cap (7) • The CO 2 emission per hectare (E CO 2 ) was calculated with Equation (8): where c = 2.65 was based on the calorific value of diesel with a density of 0.835 kg dm −3 [33]. Moreover, the obtained indexes were applied to a UNIBO experimental farm case study, where there are 29 ha devoted to various types of orchards and vineyards. However, orchards and vineyards undergo more than one treatment during the year for spraying and mulching. Using the activity log of the UNIBO experimental farm for the year 2019, the total amount of worked hectares for spraying and mulching (UB ext ) was estimated. The average row spacing of orchards and vineyards of the UNIBO experimental farm was calculated equal to 4.1 m, and this value was used for calculating the number of passes 174 ha. The following indexes were calculated:

•
The yearly time spent for the agricultural operation (t year ) was calculated with Equation (9): here UB ext was 375 ha and 174 ha for the sprayers and mulchers, respectively.
• The diesel liters used during one year for the agricultural operation ( f year ) was calculated with Equation (10): The yearly diesel cost for the agricultural operation (C f ,year ) was calculated with Equation (11): where p = 0.9 € L −1 was the average agricultural diesel price in Italy (March 2021).

•
The CO 2 emitted per year (E CO 2 ,year ) was calculated with the Equation (12):

Results
This section is divided into subheadings. It provides a description of the experimental results and their interpretation.

Implement Hitching Time Results
The implement linkage processes consist of two phases: • Implement linkage to the TPL of the tractor. • Power transmission linkage.
The first phase is identical for electrified and mechanical implements, while the second one is remarkably different, as shown in Figure 11. Indeed, the power transmission linkage on the electrified implements is limited to the connection of the AEF HVPI, while on the mechanical ones it occurs through the connection of the cardan shaft, and, only for the mMulcher, also the connection to the hydraulic remotes.
The average times needed to perform the implement hitching operations are reported in Table 6. The average times for connecting the mulchers were longer than those of the sprayers due to the higher number of couplings: three hitches instead of two (the mulchers are mounted implements, while the sprayers are trailed implements) and the connection of the hydraulic remotes. Table 6. Hitching time of the implements. In the brackets, the standard deviation is reported for each parameter. The first phase is identical for electrified and mechanical implements, while the s ond one is remarkably different, as shown in Figure 11. Indeed, the power transmissi linkage on the electrified implements is limited to the connection of the AEF HVPI, wh on the mechanical ones it occurs through the connection of the cardan shaft, and, only the mMulcher, also the connection to the hydraulic remotes. The average times needed to perform the implement hitching operations are report in Table 6. The average times for connecting the mulchers were longer than those of sprayers due to the higher number of couplings: three hitches instead of two (the mulch are mounted implements, while the sprayers are trailed implements) and the connecti of the hydraulic remotes. Table 6. Hitching time of the implements. In the brackets, the standard deviation is reported for each parameter. One can note that a slight time saving was obtained using the electrically powered implements, mainly for the power transmission linkage. This was mainly due to the faster joining of the AEF HVPI since the time spent to connect the implements to the TPL of the tractor was the same. The measurements were conducted on brand new machinery, so the time spent to mount the cardan shaft could worsen in the long term due to wear and dirt during real in-field activities. In addition to that, the time required for coupling of the hydraulic quick-coupler could also be higher in real-world conditions, since occasionally, high pressure may rise in the pipeline and it must be released to allow the connection of the hydraulic remotes [34]. This issue cannot occur with the AEF HVPI. Moreover, an important improvement could be reached in terms of comfort and safety for the user since the operation of joining the hydraulic remotes and the cardan shaft to the rear PTO is more uncomfortable and potentially more unsafe than joining the AEF HVPI. Indeed, joining mechanical couplings requires greater loads and in uncomfortable positions (i.e., with the back bent), this may be a potential source of musculoskeletal disorders.

Noise Tests Results
The noise test results obtained with the eSprayer and mSprayer with n f set at 1300 rpm and 1800 rpm are reported in Table 7 and in Table 8, respectively. One can note from Tables 7 and 8 that the eSprayer produced conspicuously lower sound pressure levels than the mSprayer, a maximum reduction of 11.3 dBA was measured with n f set at 1800 rpm. However, this remarkable result was caused by the engine downspeed permitted by the driveline electrification, the absence of the cardan shaft rotating at high speed, and the different design of the fan blades of the eSprayer. It is arduous to estimate the contribution of each factor on the noise reduction.
The mulcher results presented in Table 9 show that the eMulcher registered lower sound pressure levels than the mMulcher, reductions of 4.5 dBA and 5.0 dBA reduction were measured with the PTO in 540E and 540 modes, respectively. This improvement was mainly due to the engine downspeed and mulcher drivelines; in particular, the absence of the cardan shaft rotating at high speed on the eMulcher ensures that the latter is less noisy since no mechanical vibrations are transmitted to the tractor and the implement.

Field Performance
The performance parameters obtained with the eSprayer and mSprayer with n f set at 1300 rpm and 1800 rpm are reported in Table 10 and in Table 11, respectively. One can note that the V t during all the tests ranged from 5.37 km h −1 to 5.62 km h −1 , which is consistent with the chosen test procedure, where the chosen target V t was 5.5 km h −1 . When n f was set at 1800 rpm, the differences in terms of P impl,m at 540 and 540E modes were limited, as expected, since the sprayer fan rotates almost at the same speed. With n f set at 1300 rpm, differences of P impl,m between the two PTO operating modes were more evident than those at n f set at 1800 rpm; however, those were still acceptable and caused by the fact that tests were performed in real working conditions and not in a controlled environment. Moreover, with n f set at 1300 rpm, P impl,e was significantly lower than P impl,m . In particular, −40% compared with that of the mSprayer in 540E mode and −33% in 540 mode. With n f set at 1800 rpm, the eSprayer still absorbed less power than the mSprayer, but in this case there were limited differences (around 10-12% lower). These P impl, * differences were mainly due to the different transmission drive and to the different fan blade design. The analysis of the P impl, * standard deviation shows that higher values were obtained on the mSprayer. This was due to the slight variability of the angles that the cardan shaft formed with the PTO of the tractor and the splined shaft on the implement during the operation, causing a non-constant transmission ratio, and this led to a certain variability in P impl,m that was not observed in P impl,e ( Figure 10). The e-Source was demonstrated to have a high η eS in both configurations, indeed values of 0.85 and 0.92 were measured in the two tests with n f set at 1300 rpm and 1800 rpm, respectively. One can note from the previous tables that a remarkable engine downspeed is reachable with the adoption of the eSprayer. Indeed, the P eng measured on the eSprayer was markedly lower than those measured with the mSprayer, therefore even the . f showed a similar behavior. In particular, with the eSprayer with n f at 1300 rpm, . f was 33.3% lower than the one obtained with the mSprayer with the same n f actuated by the PTO in 540 mode. Furthermore, with the n f set at 1800 rpm, . f with the eSprayer was 21.4% lower than the one obtained with the mSprayer with the same n f actuated by the PTO in 540 mode. The analysis of the results obtained with the mulchers (Table 12) shows that the V t during all the tests ranged from 3.06 km h −1 to 3.12 km h −1 , which is consistent with the chosen methodology where the target V t was chosen to 3.1 km h −1 . The η eS measured during the eMulcher test was 0.90, very similar to those obtained during the eSprayer tests. The P impl, * measured during the mulcher tests were lower than those measured on the sprayers and all the values lay in the range between 3.0 kW and 4.7 kW. Theoretically, P impl, * of the eMulcher and the mMulcher should be very similar, since the two implements have the same area capacity, rotor design, and they were used in the same working conditions. Therefore, the differences in P impl, * showed in the results were mainly caused by the different transmission systems installed on the two mulchers, the slight differences in V t and the natural variability of the residuals on the fields. Even if a slight difference in terms of P impl, * between the mMulcher and eMulcher can be observed, the latter registered a reduction of . f of 29.8% with respect to the mMulcher used with the PTO in 540 mode. This was due, once again, to the independence of n r from n eng , allowing the machines to perform the field operation at lower n eng and, consequently, lower P eng and . f .

Operational and Environmental Indexes
Operational and environmental indexes obtained with the eSprayer and mSprayer with n f set at 1300 rpm and 1800 rpm are reported in Table 13 and in Table 14, respectively.  The saving in terms of f ha followed the same trend observed for the . f since the travelling speeds were very similar (and consequently the F cap ) for each test. Regarding the UNIBO farm case study, the yearly money saving due to the saved fuel ranged from 337 € to 506 € for n f set at 1300 rpm and from 270 € to 405 € for n f set at 1800 rpm, depending on the considered PTO mode.
Even for the mulchers (Table 15), the savings in terms of f ha followed the same trend observed for the . f for the same aforementioned reasons. Regarding the UNIBO farm case study on SimVin, the yearly money saving due to the saved fuel ranged from 360 € with the PTO in 540E mode to 673 € with the PTO in 540 mode. Both for the eSprayer and the eMulcher, besides the money-saving due to a fuel consumption reduction, a cost that could be lower compared to their rear PTO homologues is the maintenance cost. In fact, the greasing operation on the cardan shaft and on other moving parts is not necessary. Regarding E CO 2 ,year , the usage of the tested electrified implements could permit a reduction of CO 2 emission of up to 2019 kg year −1 . This value may seem not very significant, but it gains fundamental importance if it is multiplied for the thousands of machines worldwide that could benefit from this technology in the future.

Discussion and Conclusions
The necessity to increase the efficiency of agricultural machinery is pushing manufacturers in searching for novel solutions to increase fuel efficiency. Electrification is probably the solution in which manufacturers are seriously investing currently. In this study, two electrified implements were compared with their equivalent mechanical counterparts. The electric implements were powered with an electric generator powered with the front PTO of the tractor. The electric power was delivered to implements through the AEF HPVI electric plug, which would replace hydraulic remotes and the PTOs. The field test reported a greater fuel efficiency of the electrified implements with respect to the mechanical counterparts and the fuel efficiency benefits were greater, in percentage, with low-demand operations due to the engine downspeed. Moreover, the fuel, money, and CO 2 savings estimated in this study were calculated assuming that all orchards and vineyards have the same row spacing, and this may not always occur. So, the actual savings may be significantly different than those calculated in this study. However, these figures allowed us to evaluate how much farmers may benefit with this solution.
The engine downspeed would be beneficial also in terms of the increased durability of rotating the components of the transmission due to the fact that the number of damaging cycles will be reduced [35].
A more in-depth economic analysis that also includes the generator and implement purchase prices and the repair and maintenance costs is not possible at this stage of their development processes. Indeed, these devices are not currently on the market and a more extensive measurement campaign is needed to obtain additional data to fill these shortcomings.
Electrification of implements will also permit the introduction of additional functionalities, which is very important, especially for sprayers where novel solutions for avoiding overdosing or drift have been studied [36]. The advantages of the proposed solution with respect to the mechanical and the hydraulic counterparts are also enhanced with improvements in the safety of farming, which is still plagued by a high number of accidents worldwide. Indeed, injuries from mechanical PTOs are rather frequent and appear to be around 2000 injuries per year in the U.S [37]. Major causes are the lack of adequate shielding and the presence of any protrusion (i.e., locking pin, bolt, cotter pin, grease fitting) that can catch operator clothing. Even if these accidents are not as frequent as others (i.e., tractor rollovers [38]), the consequences are severe since they may lead to death, amputation, and fractures [39][40][41]. Hydraulic power is much safer than mechanical PTO, but it is not completely free from any danger since injection injuries may occur in certain conditions [37]. Injection injuries occur when a body part is in contact with a high pressure fluid and the impact of the injury is seriously dependent on the toxicity of the fluid [42]. The surplus weight on the tractor front axle due to the presence of the e-Source could increase the soil compaction. However, this issue is counterbalanced by the absence of the ballast, which is typically mounted on the front linkage instead of the e-Source. Finally, the solution of an external electric generator is a solution that permits farmers to obtain full advantage of electrified implements, even if they use old tractors.