- Email: [email protected]
An innovative flexible head for the harvesting of cardoon (Cynara cardunculus L.) in stony lands
Industrial Crops and Products 94 (2016) 471–479
Contents lists available at ScienceDirect
Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop
An innovative flexible head for the harvesting of cardoon (Cynara cardunculus L.) in stony lands Luigi Pari ∗ , Angelo Del Giudice, Daniele Pochi, Francesco Gallucci, Enrico Santangelo Consiglio per la ricerca in agricoltura e l’analisi dell’economia agraria (CREA) – Unità di ricerca per l’ingegneria agraria, Monterotondo, Roma, Italy
a r t i c l e
i n f o
Article history: Received 19 April 2016 Received in revised form 11 August 2016 Accepted 1 September 2016 Keywords: Harvesting Cardoon Head height control
a b s t r a c t Cardoon (Cynara cardunculus L.) is a promising crop for utilizing marginal lands. In these terrains, the lack of adequate tillage or soil levelling and the excessive presence of stones requires a high cutting height during harvesting, with remarkable biomass loss occurring. CREA-ING designed a new flexible bar driven by a system for sensing and signalling the presence of obstacles during the forward of motion of a combine harvester. A test-track was prepared to monitor the activation sequence of three sensor systems placed on the cutter bar as follows: four piston transducers measuring the flexion of the blade and counter-blade, two opposite wire transducers computing the movements of the blade-holder hinged on the left and right side of the frame, and one wire transducer placed between the head and the combine for measuring the lifting of the head operated by the hydraulic system. The combine was driven at an average speed of 0.7 km h−1 on a row of progressively higher obstacles (from 10 to 40 cm) placed in lateral, intermediate and central positions. The tests were carried out in the left half of the cutter bar. The output sent by the different sensors varied as a function of their position and the position of the obstacle, thus highlighting that the presence of an obstacle was correctly perceived by the sensor system. The signals originated from the left and right transducers had opposite trends. At the narrow “bell-shape” showed by the graph of the left transducer corresponded to a “reversed” bell generated by the right transducer, thus representing graphically the flexibility of the bar. The four piston transducers detected the necessary flexion, then lift and return the cutter bar to the starting point in response to each obstacle. The head lifting varied from 1.25 s to 2.51 s, but a threshold value could be observed as follows: below 25 cm the lifting occurred between 1.0 s and 1.5 s, while for higher obstacles the head lifting required 2.0–2.5 s. Such movement was the result of the signals sent by the sensors to the control unit before the head lifting began. The difference between the input sent by the transducers and the head lifting ranged from 1.22 s to 2.54 s in relation to the position of the obstacle. The tests showed that the head elements activated efficiently during the overcoming of an obstacle. However, if increasing speeds are needed, the reduction of Act or Dsig will require a modification of the electrohydraulic components. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Crop harvesting is a crucial stage in the supply chain of biomass production because it affects both product quality and efficiency. Specifically, some energy crops might require innovative solutions for assuring their integral exploitation and the profitability of cultivation. Cardoon (Cynara cardunculus L.) is one of the most interesting non-food crops for biomass production. It is a perennial species
∗ Corresponding author. E-mail address: [email protected] (L. Pari). http://dx.doi.org/10.1016/j.indcrop.2016.09.005 0926-6690/© 2016 Elsevier B.V. All rights reserved.
with an annual development cycle that can be repeated for more than 10 years (Angelini et al., 2009; Gherbin et al., 2001). The plant can reach a height of 3 m and the rooting system can reach a depth up to 7 m (Fernández et al., 2006). Cardoon is well adapted to the xerothermic conditions of southern Europe and Mediterranean climatic-type areas (Tuck et al., 2006; Grammelis et al., 2008), and could be preferred to others crops (fibre sorghum, giant reed, miscanthus, switchgrass) when water is a limiting factor (Solano et al., 2010). Recently, the growth of cardoon in marginal areas of South (Sicily region) or Central (Latium region) Italy has been studied, where the crop improved the soil fertility of degraded lands (Mauromicale et al., 2014; Francaviglia et al., 2016).
472
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
In terms of energy conversion, the main products are the lignocellulosic biomass for energy production (Fernández et al., 2006; Gominho et al., 2009) and the oil extracted from seeds for biodiesel production (Gominho et al., 2011; Sengo et al., 2010), similar to other botanically related oil crops, such as the sunflower or the safflower (Fernández and Manzanares, 1990). New systems for the separation of lignocellulosic biomass and seeds were studied to aid in the development of the crop (Pari et al., 2008, 2009a,b, 2011). Such systems integrate the components of a maize head with those of a classic wheat head (Pari et al., 2011). The upper portion, derived from a maize head, provides for the detachment of the capitula and then transports it to the threshing operation. In the lower portion (developed from a wheat head), the aboveground biomass is mowed, conditioned and windrowed. The capitula residue is released in the windrow, while the seeds are collected from the combine harvester. During previous harvesting trials performed in stony areas of Sardinia (Italy), the traditional head for wheat with a rigid cutter bar showed high susceptibility to stone damage, forcing operators to increase the cutting height to 40 cm, which resulted in excessive harvesting losses. The cutter bar is one of the most important elements affecting the combine performance since it is the first mechanical contact with the crop and the first point where the yield losses occur. The use of a flexible header was reported as a popular method for the harvesting of pulses, with particular reference to soybean harvesting (Hirai et al., 2005; Hummel, 1983; Zyla et al., 2002). The idea of using a flexible cutting system for reducing the harvesting losses of traditional combines dates back to 70’s (Quick and Buchele, 1974). The authors observed that approximately 85 percent of combine losses were borne from the head, mostly due to the action of the reciprocating cutter bar. Floating, flexible and floating flexible cutter bars follow the contour of the ground, thus minimizing the loss of low hanging pods (Siemens, 2006; Glancey, 1997). As a result, the harvesting losses were reduced by 3–10% of the total yield (Hummel, 1983; Nave et al., 1977; Quick and Buchele, 1974). A particular case for using the flexible cutter bar was recently proposed by Mahmoodi et al. (2007) for harvesting of furrow-hill plantings. The bar was composed of multiple sections hinged to each other and regulated by a leveller to the shapes and sizes of furrows and hills. To find a technical solution for reducing losses in stony soil, CREA-ING designed a new flexible bar driven by a system for sensing and signalling the presence of obstacles. The present work reports the results obtained in a test-track prepared for monitoring the behaviour of the flexible bar and the ability of the sensors system to detect obstacles of different heights. The goal of the work was to analyse the timing of the sensor system in detecting the obstacles and to verify their ability to correctly overcome the obstacle while maintaining the height of the cut at the minimum distance from the soil.
2. Materials and methods 2.1. Description of the head The new head for the harvesting of cardoon in stony lands was designed by CREA-ING in collaboration with the company Cressoni Ltd. The head has three different elements which are activated by the presence of obstacles. The first element is a cutting bar fixed on a harmonic steel structure, which allows for an oscillation up to 150 mm high. Such a blade is capable of following the profile of the ground surface, keeping the height of the cut at the minimum distance from the ground, and flexing when a stone is encountered.
Fig. 1. Boot-shaped coulters of the self-leveling head.
The second element is represented by the structure that supports the blade. It is hinged on both sides of the head’s main body and is free to move vertically, following the contours of the ground. The lower part of the structure contains 28 coulters, fitted with 20 mm wide sleds shaped to climb the stones, while avoiding their contact with the blade (Fig. 1). Behind the coulters, a structure consisting of six articulated sections in wear-resistant steel, protects the moving parts from rubbing with stones and rocks. Since each arm can float independently from the other, the blade can flex, adapting itself to the roughness of the ground also in the transverse direction. To this aim, the housing of the lower auger (fixed part) is connected to the mobile part by a lamina of harmonic steel. The task of the lower auger (300 mm diameter) is to convey the cardoon stalks towards the centre of the head and to discharge them to the ground between the wheels forming the windrow. The lower auger moves with the mobile part through a sliding system between metal sheets. For overcoming the obstacles, the three systems act independently and sequentially. When the head encounters an obstacle, the sleds prevent its contact with the blade by lifting it from the ground in response to the contact point. Subsequently, when the blade reaches the highest point allowed by its flexibility, the mobile part hinged on the sides of the head and supporting the blade raises. The oscillations of the mobile portion are measured by angular transducers installed on both oscillating arms and transmitted to the control unit of the combine. When the lifting exceeds a threshold, the control unit commands the hydraulic system to lift the whole head while the signal persists. When the obstacle is overcome, the interruption of the electrical signal causes the lowering of the head. Then, the mobile portion drops as well and the blade straightens (Fig. 2). A corn head is mounted above the flexible cutter bar, which provides for the detachment and separation of the capitula. It is composed of 9 groups spaced 500 mm apart. In each group, a pair of chains transfers the capitula to an upper auger. Below the chains, two detachment blades can be regulated hydraulically according to the diameter of both the capitula and the apical portion of the stem. Underlying these are the counter-rotating harvesting rollers. Equipped with 4 blades, the stem is simultaneously pulled down and comminution occurs until the capitula is detached by the blades. The first prototype for cardoon harvesting was designed as part of the BIOCARD Project (Pari et al., 2008, 2009a,b, 2011). The head used in this study is the second version developed by CREA-ING in collaboration with Cressoni Ltd. in the framework of the BIT3G project for developing operation in the marginal stony fields of the Sardinia Region (Pari et al., 2014).
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
473
Fig. 2. Diagramatic representation of the sensor system.
2.2. Methodological approach A test-track was prepared to monitor the activation sequence of the three systems. Seven trapezoidal wood blocks were fixed to the ground and arranged in a single row on a flat stripe of soil 30 m long. The obstacles had a frontal surface inclined at 45◦ from the ground with progressively increasing heights (10, 15, 20, 25, 30, 35, 40 cm). They were spaced 4 m apart, the distance needed by the head to restore its normal position after overcoming each obstacle. Considering that the bar flexibility is the highest in the centre and the lowest on the sides (Fig. 3), three series of tests were carried out varying the relative position between the bar and the obstacle row, as follows: Pos 1 – lateral (LAT), on the left side of the head (outside face of the wheel) 0.25 m from the head edge; Pos 2 – intermediate (INT), near the inside face of the wheel, 1.65 m from the left head edge; Pos 3 – central (CEN), at the midpoint of the head, between the wheels, 2.25 m from the left head edge. 2.3. Sensor system The combine and the mobile elements of the head were equipped with potentiometric transducers, capable of monitoring the movements by measuring the electric signal variations they induce. The data were stored by an acquisition control unit (Fig. 4). Moreover, the combine ladder was equipped with a “wheel of Peiseler” (Fig. 4), measuring the distance and the actual travel speed. The tests were carried out in the left half of the head, assuming that the behaviour of the head was similar on the opposite side. To monitor the activation sequence of the three systems responsible for avoiding the contact between the blade and obstacle, the following sensors were mounted (Fig. 5): 1. four piston transducers (PT1, PT2, PT3 and PT4) for measuring the flexion endured by the blade and the counter-blade on the obstacles. In each test, the transducers were installed on the cutting bar in relation to the obstacle row, according to the above mentioned LAT, INT, CEN (Fig. 5);
2. two wire transducers for measuring the movements of the bladeholder hinged on the frame, on both left and right side (LT and RT) (Fig. 6); 3. one wire transducer placed between the head and combine (Fig. 6) for measuring the lifting of the head (HL) operated by the hydraulic system.
For each position (LAT, INT, CEN), the test was replicated five times for a total of 15 runs. In each run, the sensors registered the following: the distance travelled and time required, the heights of the piston transducers (PT1, PT2, PT3, and PT4), the height of the blade holder at both left and right sides (LT and RT), and the hydraulic lifting height of the head (HL). The machine started each run at a distance sufficient to reach the desired speed (approximately 1 km h−1 ), before hitting the first obstacle. All five replicates were conducted by the same driver.
2.4. Analysis of the signal During each run, the signals sent by the sensors were detected at a frequency of 10 Hz. Depending on the length of the test-track and the time required to accomplish a run, the number of acquisitions per run ranged from 1100 to 1800. To allow for a comparison of overlapping measurements, for each row position and for all replications, the acquisitions occurring ± 1 m from the peak of the maximum height of the bar relative to the obstacle were considered. For a detailed analysis of the signals, we examined the behaviour of the inputs registered for each sensor prior to the maximum point reached by the HL sensor (Fig. 7). The activation delta (Act ) and the duration of the signal (DSig ) were taken into account. The activation delta (Act ) was the difference (in seconds) between the start of the signal sent by the transducers (LT, RT, PT1, PT2, PT3, PT4) and the start of the command for head lifting (HL). The duration of the signals (DSig ) represented the time taken for the sensors and the head to reach the maximum value. For both measures, the values registered for obstacle 2 were not included in the analysis because signal anomalies were found in all treatments (see below).
474
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
Fig. 3. Arrangement of the wood obstacles. Lateral (A) and top view (B) of the test-track. Above each obstacle it is reported the corresponding height in cm.
Fig. 4. Control unit (left) and “wheel of Peiseler” (right).
Fig. 5. Piston transducers placed on the left side of combine (PT1, PT2, PT3, PT4) when the obstacles were in lateral (left) and in central position (right).
2.5. Statistical analysis All data were analysed with the software PAST. Homoscedasticity and normality were checked prior to testing. If the normality of distributions was verified, the data were analysed using one-way ANOVA. In this case, the Duncan’s multiple range test was used for the separation of the means. When the data distribution deviated from normality, the Kruskal-Wallis non-parametric test was
applied and the differences were tested according to the MannWhitney pairwise test. 3. Results and discussion The combine travelled for a net distance slightly shorter than the whole length of the test-track (Table 1). The variability observed among the three positions of the obstacles is related to the need
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
475
Fig. 6. Wire transducers between head and combine for measuring the movement of the head (HL) (a), wire transducers for measuring the displacement of the mobile part placed on the right (RT) and the left (LT) side of the cutter bar (b).
Table 1 Space, time and speed of the combine registered for the different positions of the obstacles.
Fig. 7. Parameters considered for the analysis of head movement (DSig is represented only for the HL).
of the operator to reach the first obstacle at a steady speed. The row in the lateral position required the driver to pay closer attention to maintaining the correct direction, which caused an increase
Obstacle
Acquisition (n.)
Space (m)
Time (s)
Speed (km h−1 )
Lateral Intermediate Central
1599 ± 158,31 1414 ± 105,74 1412 ± 252,21
27.73 ± 0.4 29.12 ± 0.8 27.47 ± 1.2
159.02 ± 15.8 140.52 ± 10.6 140.30 ± 25.2
0.63 ± 0.06 0.75 ± 0.05 0.72 ± 0.12
Mean
1475 ± 172,09
28.10 ± 1.1
146.60 ± 19.2
0.70 ± 0.09
in time and a reduction in the forward speed for this test. On the other hand, the presence of the obstacles in the intermediate position needed more space before the running began to position the combine correctly. The different types of sensors showed a distinct range of oscillations according to the position in which they were applied and the type of measurement monitored (Fig. 8). Moreover, the displacement of the obstacles row (LAT, INT, CEN) also influenced the signals of the sensors. As a general rule, the magnitude of the oscillation decreased as follows: head lifting (HL) > left and right edges (LT, RT) > blade and counter-blade (PT1, PT2, PT3 and PT4). The behaviour of the wire sensors monitoring the lifting and lowering of the left and right edges of the blade-holder was particularly divergent when obstacles were in the lateral or central position. As described below, the transducers of the left side showed a signal shift towards positive values, while those on the right were negative. The data reflects the flexibility of the bar as follows: during the overcoming of the obstacle, it temporarily assumed an S shape, with positive values in respect to the starting point, in response to the
Fig. 8. Box plot showing the 25–75 percent quartiles, the median (horizontal line) and the minimal and maximal values (whiskers) of head lifting (HL), signals of left and right wire transducers (LT and RT) and the piston transducers (PT1, PT2, PT3 and PT4), in lateral (A), intermediate (B) or central (C) position.
476
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
Fig. 9. Height of the bar (mm) in correspondence of the relative position of the obstacles. For each point of maximum lifting the significance of F value is showed (* = P < 0.05; *** = P < 0.01; ns = not significant).
obstacle, whereas on the opposite side, the blade holder dropped below the reference level. During the run, the lifting of the head was somewhat associated with the height of the obstacle (Fig. 9). The height of the bar showed a tendency to rise in the presence of higher obstacles. The graphics of the bar height were almost overlapping for the lateral and central positions of the rows of obstacles, whereas when the row was in the intermediate position, the head lifting was more pronounced. In general, the signals for the second obstacle were not easily identifiable, therefore the corresponding measurements were maintained in the general analysis of the signal level, but were disregarded in the detailed analysis of the signals (act and DSig ) to avoid biased results. When compared with the other obstacles, the signal observed in the second obstacle would appear anomalous. However, such an anomaly was independent from the position of the obstacle, and the detection of the obstacle did occur, but at a lower intensity. This highlights at least two important aspects of the system. The first is its capacity for detecting a specific combination of adjacent low obstacles (the first and the second). The second concerns the distance. Although not analysed in the present work, such a degree
Fig. 10. Signal of the wire transducers (LT and RT) compared to the height of head recorded by the HL sensor in presence of increasing obstacle in lateral (A), intermediate (B) and central (C) position.
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
477
of sensitivity should allow for the detection of closer obstacles (well below 4 m): the timing of the signal should remain the same but it would be reasonable to expect a variation of its intensity. Such a hypothesis needs to be confirmed, however, either in a test-track or in non-uniform conditions. The signal relating to the displacement of the mobile part was influenced by the point of application of the transducer and by the position of the obstacle row (Fig. 10). The main difference between the LT and RT signal concerned the shape of the signal. Both sensors reached the maximum level of the input immediately prior to the maximum lifting of the head, but, while the LT assumed a shape of narrow “bell”, the RT dropped negatively once the maximum value was reached, thus producing a “reversed bell” diagram. This was clearly visible when the obstacles were in the lateral position (Fig. 10A) because the left side of the mobile part was subject to an upward solicitation while the right side folded downward. Such behaviour represented graphically the effect of the flexibility. While the left half of the cutter bar overcame the obstacles, the right half curved below the “0-position” of the transducers, and the magnitude of folding was driven to some extent by the position of the obstacles. When the obstacles were in the central or intermediate positions (Fig. 10B and C) the RT maintained the same type of signal. Conversely, the LT showed an enlargement of the “bell” shape which was much wider as the LT was far from the obstacle row. As a final result, however, both transducers correctly detected the obstacle by starting the signal before the command of head lifting. Being positioned directly on the cutter bar, the signals sent by the piston transducers were much clearer, uniform and coordinated than the wire transducers’ signals (Fig. 11). The four transducers accurately detected all stages of overcoming (flexion, lifting, and return to the starting point) performed by the cutter bar in response to each obstacle, showing also that the PT signal preceded the lifting of the head. The tops of the curves were reached when the bar began the ascending phase. As previously observed for the LT position, the signal associated with each obstacle had the shape of a narrow “bell” when the row was lateral (Fig. 11A), while such configuration was modified by the presence of a “shoulder” when the combine encountered intermediate or central obstacles. 3.1. Analysis of the signals Once the signals generated by the sensors reached the control unit, the system took a time interval (Dsig ) ranging from 1.25 s (obstacle 4) to 2.51 s (obstacle 7) to lift the head before overcoming the obstacle (Fig. 12). Interestingly, the data revealed a threshold value beyond which the head required a different range of time as follows: below 25 cm, the lifting occurred between 1.0 s and 1.5 s, while for obstacles comprised between 30 cm and 40 cm, the head lifting time ranged from 2.0 s to 2.5 s. Such observations prompt some needed explanation regarding some hidden implications in the study. Cardoon is a “poor” crop suitable for marginal areas where the accurate soil tillage may lag behind. The use of practices such as minimum or no-tillage as well as the absence of soil levelling or the presence of obstacles of various natures (often stones) leads the combine to work on an irregular surface where the difference in height can reach a remarkable value. Thus, the apparently unrealistic sizes of the highest obstacles were not a mere experimental exercise, but were aimed at reproducing field situations typical of highly uneven terrain. Further points of discussion emerged by extending the analysis of DSig to all sensors in response to the positions of the three obstacles (Fig. 13). The RT showed the shortest signal duration in both the lateral (0.89 s) and intermediate positions (1.56 s). In the same test, the corresponding and opposite wire transducer (LT) produced a signal statistically longer than the RT. When the obstacle row
Fig. 11. Signal of the piston transducers (PT1, PT2, PT3 and PT4) compared to the height of head recorded by the HL sensor in presence of increasing obstacle in lateral (A), intermediate (B) and central (C) position.
Fig. 12. Mean (±SE) of the duration of head lifting. Means sharing common letters are not statistically different for P ≤ 0.01, following post-hoc Duncan’s test.
478
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
Fig. 13. Mean (±SE) of the length of DSig of the six transducers and the head lifting with lateral (A), intermediate (B) and central (C) obstacles. Means with different letters are significantly different for P < 0.05 (lowercase letters) or P < 0.01 (capital letters), following post-hoc Mann-Whitney test.
was equidistant between the LT and RT (CEN) the signal length was comparable, thus highlighting that the presence of an obstacle is accurately perceived by both sides of the head. The time elapsed between the beginning of the signal from the piston transducers and the angular transducers input for head lifting was over 2 s and was the longest whenever the obstacles were positioned. It should be noted that the obstacle row in the intermediate position caused an increase in the signal length, suggesting that some specific positions could require a more careful “reading” to precisely identify the obstacle. As expected, although to a different extent, the raising of the head was always preceded by the signalling of the sensors (Fig. 14). The obstacles in the lateral position were detected within an interval of 1.49 s (RT)–2.09 s (PT3) before the head lifting, while when the position was intermediate, the sensors sent the signal 2.08 s (LT) – 2.54 s (PT4) earlier. The detection of a central obstacle was more delayed, requiring between 1.22 s (LT) and 1.82 s (PT3). Thus, it can be speculated that although the system is able to perceive the soil irregularity along the entire head width, some specific positions can be trickier and require an improvement in the method of detection. The wire transducers (LT and RT) showed an opposite behaviour in relation to the obstacle position. The LT anticipated the RT in sending the signal just when the obstacles were in the lateral position (hence, on the same side of the transducer). Shifting the obstacle far from the left side caused an earlier activation of the RT with a difference that became significant for the central
Fig. 14. Mean (±SE) of the activation delta (act ) of the six transducers with lateral (A), intermediate (B) and central (C) obstacles. Data were calculated considering the lifting of the head as point 0. Where reported, means with different letters are significantly different (P < 0.05), following post-hoc Mann-Whitney test.
obstacles. The signals from the piston transducers started a little bit earlier than those from the LT and RT when the obstacles were lateral and central, while in the intermediate position they started simultaneously. The extreme sensibility of the PT is a consequence of their point of application, corresponding to the obstacle row, for detecting the necessary flexion of the blade and counter-blade. Their earlier activation confirms that the signalling started just as the coulters of the bar began to “feel” the obstacle. The experimental data has some practical implications tied to the correct tuning among the components of the combine. The sum of Act and Dsig accounted for the time between the electric input and the complete head lifting (Table 2). In one specific case, travelling with an average speed of 0.7 km h−1 , corresponding to 0.19 m s−1 , the combine needed a space of 0.74 m to overcome the obstacle. Cardoon harvesting can be carried out at 0.80 m s−1 (Pari et al., 2014), a value four-fold higher than the speed maintained in the present study. Assuming the same reaction time is maintained, increasing speeds will require a reduction in the Act or the Dsig . The first concerns the electronic system and hence the frequencies of the pulse of the input signal. The modification of the Dsig would require a change to the hydraulic system, namely the section width of the pipeline controlling the head lifting. Such outcomes agree in part with those of
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479 Table 2 Time and space required for overcoming different obstacles (combine speed 0.19 m s−1 ). Obstacle
Act (s)
Dsig (s)
Total Time (s)
Space required (m)
1 (10 cm) 3 (20 cm) 4 (25 cm) 5 (30 cm) 6 (35 cm) 7 (40 cm)
2.80 1.73 1.55 2.32 1.82 1.91
1.29 1.39 1.25 2.29 2.44 2.51
4.1 3.1 2.8 4.6 4.3 4.4
0.78 0.59 0.53 0.88 0.81 0.84
Mean
2.02
1.86
3.9
0.74
Xie et al. (2010, 2013), which identified the mechanical configuration (combine and head) and the electrohydraulic actuation of the head as the main subsystems affecting the performance limitations in the control of head’s height. The authors also listed the replacing of tires with tracks, the use of active suspension, the redesign of suspension elements and the use of very high performance servo hydraulics as possible means for improving the performance (Xie et al., 2013). 4. Conclusion In difficult terrains, the use of conventional combines leads to a substantial loss of biomass because the bar must be raised off the ground 40–50 cm to avoid breakage. The flexible head studied in the present work showed technical characteristics useful for reducing such losses. The integration of the sensors system combined with the flexibility deriving from the specific building material, allowed for the adjustment of the bar height in a dynamic way, almost point-by-point, as can occur in irregular terrains. The control system was able to quickly and effectively detect obstacles ranging from 10 to 40 cm, a deliberately excessive height tested to evaluate the technical limits of the system. Furthermore, the test showed that the three head elements activated during the overcoming of an obstacle combined efficiently with each other. Consequently, the adoption of the system can be recommended, independent of the species, whenever it is necessary to maintain the cutting system as close as possible to the ground. However, the bottleneck of increasing the combine speed remains. Further in-depth studies are required to address the challenging objective of integrating possible mechanical and/or electronic solutions able to efficiently exploit the potential of the sensor system while maintaining a reasonable cost. Acknowledgements This work was supported by the BIT3G (MIUR) and Suscace (MIPAAF) projects. References Angelini, L.G., Ceccarini, L., o Di Nasso, N.N., Bonari, E., 2009. Long-term evaluation of biomass production and quality of two cardoon (Cynara cardunculus L.) cultivars for energy use. Biomass Bioenergy 33 (5), 810–816, http://dx.doi.org/ 10.1016/j.biombioe.2008.12.004. Fernández, J., Manzanares, P., 1990. C. cardunculus L. a new crop for oil, paper pulp and energy. In: Proceedings 5th European Conference on Biomass for Energy and Industry, London, UK, pp. 1184–1189. Fernández, J., Curt, M.D., Aguado, P.L., 2006. Industrial applications of Cynara cardunculus L. for energy and other uses. Ind. Crops Prod. 24, 222–229, http:// dx.doi.org/10.1016/j.indcrop.2006.06.010. Francaviglia, R., Bruno, A., Falcucci, M., Farina, R., Renzi, G., Russo, D.E., Sepe, L., Neri, U., 2016. Yields and quality of Cynara cardunculus L. wild and cultivated cardoon genotypes. A case study from a marginal land in Central Italy. Eur. J. Agron. 72, 10–19, http://dx.doi.org/10.1016/j.eja.2015.09.014. Gherbin, P., Monteleone, M., Tarantino, E., 2001. Five year evaluation on Cardoon (Cynara cardunculusL. var.altilis) biomass production in a Mediterranean environment. Ital. J. Agron. 5, 11–19.
479
Glancey, J.L., 1997. Analysis of header loss from pod stripper combines in green peas. J. Agric. Eng. Res. 68 (1), 1–10, http://dx.doi.org/10.1006/jaer.1996.0122. Gominho, J., Lourenc¸o, A., Curt, M., Fernández, J., Pereira, H., 2009. Characterization of hairs and pappi from Cynara cardunculus capitula and their suitability for paper production. Ind. Crops Prod. 29 (1), 116–125, http://dx.doi.org/10.1016/j. indcrop.2008.04.022. Gominho, J., Lourenc¸o, A., Palma, P., Lourenc¸o, M.E., Curt, M.D., Fernández, J., Pereira, H., 2011. Large scale cultivation of Cynara cardunculus L. for biomass production–a case study. Ind. Crops Prod. 33 (1), 1–6, http://dx.doi.org/10. 1016/j.indcrop.2010.09.011. Grammelis, P., Malliopoulou, A., Basinas, P., Danalatos, N.G., 2008. Cultivation and characterization of Cynara cardunculus for solid biofuels production in the Mediterranean region. Int. J. Mol. Sci. 9, 1241–1258, http://dx.doi.org/10.3390/ ijms9071241. Hirai, Y., Schrock, M., Maski, D., Taylor, R., 2005. Loss evaluation of soybean harvesting on four types of headers. In: ASAE Agricultural Equipment Technology Conference, Louisville, Kentucky, pp. 1–4. Hummel, J.W., 1983. Impact cutting of soybeans using flexible cutting systems. Trans. ASAE 26 (5), 1315–1319, http://dx.doi.org/10.13031/2013.34123. Mahmoodi, E., Jafari, A., Mobli, H., Salokhe, V.M., Jayasuriya, H.P.W., Soni, P., 2007. A new idea for construction of a flexible cutter bar mower for listed lands. In: Salokhe, V.M., Jayasuriya, H.P.W., Soni, P. (Eds.), Proceedings of the International Agricultural Engineering Conference 2007. Cutting Edge Technologies and Innovations on Sustainable Resources for World Food Sufficiency. Bangkok, Thailand (unpaginated, ISBN 978-974-8257-48-8). Mauromicale, G., Sortino, O., Pesce, G.R., Agnello, M., Mauro, R.P., 2014. Suitability of cultivated and wild cardoon as a sustainable bioenergy crop for low input cultivation in low quality Mediterranean soils. Ind. Crops Prod. 57, 82–89, http://dx.doi.org/10.1016/j.indcrop.2014.03.013. Nave, W.R., Hummel, J.W., Yoerger, R.R., 1977. Air-jet and row-crop headers for soybeans. Trans. ASAE 20 (6), 1037–1041, http://dx.doi.org/10.13031/2013. 35698. Pari, L., Fedrizzi, M., Gallucci, F., 2008. Cynara cardunculus exploitation for energy applications: development of a combine head for threshing and concurrent residues collecting and utilization. In: Proceedings of the 16th European Biomass Conference & Exbition, Valencia, Spain, pp. 504–508. Pari, L., Civitarese, V., Assirelli, A., Del Giudice, A., 2009a. Prototype for Cynara cardunculus capitula threshing and biomass windrowing. In: Proceedings of the 17th European Biomass Conference & Exibition, Hamburg, Germany, pp. 262–267. Pari, L., Civitarese, V., Assirelli, A., Del Giudice, A., 2009. Il prototipo che abbatte i costi della raccolta del cardo Agroenergie dall’impianto alla raccolta. Special issue of L’Informatore Agrario 29, 8–11. Pari, L., Civitarese, V., Assirelli, A., Del Giudice, A., 2011. Sviluppo di un prototipo per la raccolta differenziata dei diverse frazioni di cardo. In: Pari, L. (Ed.), Lo sviluppo delle colture energetiche in Italia. Il contributo dei progetti di ricerca SUSCACE e FAESI. Nuova Cultura, Rome, pp. 671–682 (ISBN 9788861347304). Pari, L., Gallucci, F., Civitarese, V., Acampora, A., Sanzone, E., Alfano, V., Assirelli, A., 2014. Innovazione meccanica per la raccolta del cardo in ambiente arido. In: Valorizzazione delle risorse agricole nell’ottica del concetto di bioeconomia. Progetti di ricerca SUSCACE e FAESI. Special issue of Sherwood 203, 43–46. Quick, G.R., Buchele, W.F., 1974. Reducing combine gathering losses in soybeans. Trans. ASAE 17 (6), 1123–1129, http://dx.doi.org/10.13031/2013.37044. Sengo, I., Gominho, J., d’Orey, L., Martins, M., Duarte, E., Pereira, H., Ferreira-Dias, S., 2010. Response surface modelling and optimization of biodiesel production from Cynara cardunculus oil. Eur. J. Lipid Sci. Technol. 112, 310–320, http://dx. doi.org/10.1002/ejlt.200900135. Siemens, M.C., 2006. Effect of guard spacing, guard attachments and reel type on chickpea harvesting losses. Appl. Eng. Agric. 22 (5), 651–657, http://dx.doi.org/ 10.13031/2013.21997. Solano, M.L., Manzanedo, E., Concheso, R., Curt, M.D., Sanz, M., Fernández, J., 2010. Potassium fertilisation and the thermal behaviour of Cynara cardunculus L. Biomass Bioenergy 34, 1487–1494, http://dx.doi.org/10.1016/j.biombioe.2010. 04.025. Tuck, G., Glendining, M.J., Smith, P., House Jo, I., Wattenbach, M., 2006. The potential distribution of bioenergy crops in Europe under present and future climate. Biomass Bioenergy 30, 183–197, http://dx.doi.org/10.1016/j.biombioe. 2005.11.019. Xie, Y., Alleyne, A., Greer, A., Deneault, D., 2010. Header height control of a combine harvester system. Cambridge, Massachusetts, USA In: ASME 2010 Dynamic Systems and Control Conference, vol. 1, pp. 7–14. Xie, Y., Alleyne, A.G., Greer, A., Deneault, D., 2013. Fundamental limits in combine harvester header height control. J. Dyn. Syst. Meas. Control 135 (3), 1–8, http:// dx.doi.org/10.1115/1.4023209. Zyla, L.E., Kushwaha, R.L., Vandenberg, A., 2002. Development of a new crop lifter for direct cut harvesting dry bean. Can. Biosyst. Eng. 44, 2–9.
Contents lists available at ScienceDirect
Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop
An innovative flexible head for the harvesting of cardoon (Cynara cardunculus L.) in stony lands Luigi Pari ∗ , Angelo Del Giudice, Daniele Pochi, Francesco Gallucci, Enrico Santangelo Consiglio per la ricerca in agricoltura e l’analisi dell’economia agraria (CREA) – Unità di ricerca per l’ingegneria agraria, Monterotondo, Roma, Italy
a r t i c l e
i n f o
Article history: Received 19 April 2016 Received in revised form 11 August 2016 Accepted 1 September 2016 Keywords: Harvesting Cardoon Head height control
a b s t r a c t Cardoon (Cynara cardunculus L.) is a promising crop for utilizing marginal lands. In these terrains, the lack of adequate tillage or soil levelling and the excessive presence of stones requires a high cutting height during harvesting, with remarkable biomass loss occurring. CREA-ING designed a new flexible bar driven by a system for sensing and signalling the presence of obstacles during the forward of motion of a combine harvester. A test-track was prepared to monitor the activation sequence of three sensor systems placed on the cutter bar as follows: four piston transducers measuring the flexion of the blade and counter-blade, two opposite wire transducers computing the movements of the blade-holder hinged on the left and right side of the frame, and one wire transducer placed between the head and the combine for measuring the lifting of the head operated by the hydraulic system. The combine was driven at an average speed of 0.7 km h−1 on a row of progressively higher obstacles (from 10 to 40 cm) placed in lateral, intermediate and central positions. The tests were carried out in the left half of the cutter bar. The output sent by the different sensors varied as a function of their position and the position of the obstacle, thus highlighting that the presence of an obstacle was correctly perceived by the sensor system. The signals originated from the left and right transducers had opposite trends. At the narrow “bell-shape” showed by the graph of the left transducer corresponded to a “reversed” bell generated by the right transducer, thus representing graphically the flexibility of the bar. The four piston transducers detected the necessary flexion, then lift and return the cutter bar to the starting point in response to each obstacle. The head lifting varied from 1.25 s to 2.51 s, but a threshold value could be observed as follows: below 25 cm the lifting occurred between 1.0 s and 1.5 s, while for higher obstacles the head lifting required 2.0–2.5 s. Such movement was the result of the signals sent by the sensors to the control unit before the head lifting began. The difference between the input sent by the transducers and the head lifting ranged from 1.22 s to 2.54 s in relation to the position of the obstacle. The tests showed that the head elements activated efficiently during the overcoming of an obstacle. However, if increasing speeds are needed, the reduction of Act or Dsig will require a modification of the electrohydraulic components. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Crop harvesting is a crucial stage in the supply chain of biomass production because it affects both product quality and efficiency. Specifically, some energy crops might require innovative solutions for assuring their integral exploitation and the profitability of cultivation. Cardoon (Cynara cardunculus L.) is one of the most interesting non-food crops for biomass production. It is a perennial species
∗ Corresponding author. E-mail address: [email protected] (L. Pari). http://dx.doi.org/10.1016/j.indcrop.2016.09.005 0926-6690/© 2016 Elsevier B.V. All rights reserved.
with an annual development cycle that can be repeated for more than 10 years (Angelini et al., 2009; Gherbin et al., 2001). The plant can reach a height of 3 m and the rooting system can reach a depth up to 7 m (Fernández et al., 2006). Cardoon is well adapted to the xerothermic conditions of southern Europe and Mediterranean climatic-type areas (Tuck et al., 2006; Grammelis et al., 2008), and could be preferred to others crops (fibre sorghum, giant reed, miscanthus, switchgrass) when water is a limiting factor (Solano et al., 2010). Recently, the growth of cardoon in marginal areas of South (Sicily region) or Central (Latium region) Italy has been studied, where the crop improved the soil fertility of degraded lands (Mauromicale et al., 2014; Francaviglia et al., 2016).
472
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
In terms of energy conversion, the main products are the lignocellulosic biomass for energy production (Fernández et al., 2006; Gominho et al., 2009) and the oil extracted from seeds for biodiesel production (Gominho et al., 2011; Sengo et al., 2010), similar to other botanically related oil crops, such as the sunflower or the safflower (Fernández and Manzanares, 1990). New systems for the separation of lignocellulosic biomass and seeds were studied to aid in the development of the crop (Pari et al., 2008, 2009a,b, 2011). Such systems integrate the components of a maize head with those of a classic wheat head (Pari et al., 2011). The upper portion, derived from a maize head, provides for the detachment of the capitula and then transports it to the threshing operation. In the lower portion (developed from a wheat head), the aboveground biomass is mowed, conditioned and windrowed. The capitula residue is released in the windrow, while the seeds are collected from the combine harvester. During previous harvesting trials performed in stony areas of Sardinia (Italy), the traditional head for wheat with a rigid cutter bar showed high susceptibility to stone damage, forcing operators to increase the cutting height to 40 cm, which resulted in excessive harvesting losses. The cutter bar is one of the most important elements affecting the combine performance since it is the first mechanical contact with the crop and the first point where the yield losses occur. The use of a flexible header was reported as a popular method for the harvesting of pulses, with particular reference to soybean harvesting (Hirai et al., 2005; Hummel, 1983; Zyla et al., 2002). The idea of using a flexible cutting system for reducing the harvesting losses of traditional combines dates back to 70’s (Quick and Buchele, 1974). The authors observed that approximately 85 percent of combine losses were borne from the head, mostly due to the action of the reciprocating cutter bar. Floating, flexible and floating flexible cutter bars follow the contour of the ground, thus minimizing the loss of low hanging pods (Siemens, 2006; Glancey, 1997). As a result, the harvesting losses were reduced by 3–10% of the total yield (Hummel, 1983; Nave et al., 1977; Quick and Buchele, 1974). A particular case for using the flexible cutter bar was recently proposed by Mahmoodi et al. (2007) for harvesting of furrow-hill plantings. The bar was composed of multiple sections hinged to each other and regulated by a leveller to the shapes and sizes of furrows and hills. To find a technical solution for reducing losses in stony soil, CREA-ING designed a new flexible bar driven by a system for sensing and signalling the presence of obstacles. The present work reports the results obtained in a test-track prepared for monitoring the behaviour of the flexible bar and the ability of the sensors system to detect obstacles of different heights. The goal of the work was to analyse the timing of the sensor system in detecting the obstacles and to verify their ability to correctly overcome the obstacle while maintaining the height of the cut at the minimum distance from the soil.
2. Materials and methods 2.1. Description of the head The new head for the harvesting of cardoon in stony lands was designed by CREA-ING in collaboration with the company Cressoni Ltd. The head has three different elements which are activated by the presence of obstacles. The first element is a cutting bar fixed on a harmonic steel structure, which allows for an oscillation up to 150 mm high. Such a blade is capable of following the profile of the ground surface, keeping the height of the cut at the minimum distance from the ground, and flexing when a stone is encountered.
Fig. 1. Boot-shaped coulters of the self-leveling head.
The second element is represented by the structure that supports the blade. It is hinged on both sides of the head’s main body and is free to move vertically, following the contours of the ground. The lower part of the structure contains 28 coulters, fitted with 20 mm wide sleds shaped to climb the stones, while avoiding their contact with the blade (Fig. 1). Behind the coulters, a structure consisting of six articulated sections in wear-resistant steel, protects the moving parts from rubbing with stones and rocks. Since each arm can float independently from the other, the blade can flex, adapting itself to the roughness of the ground also in the transverse direction. To this aim, the housing of the lower auger (fixed part) is connected to the mobile part by a lamina of harmonic steel. The task of the lower auger (300 mm diameter) is to convey the cardoon stalks towards the centre of the head and to discharge them to the ground between the wheels forming the windrow. The lower auger moves with the mobile part through a sliding system between metal sheets. For overcoming the obstacles, the three systems act independently and sequentially. When the head encounters an obstacle, the sleds prevent its contact with the blade by lifting it from the ground in response to the contact point. Subsequently, when the blade reaches the highest point allowed by its flexibility, the mobile part hinged on the sides of the head and supporting the blade raises. The oscillations of the mobile portion are measured by angular transducers installed on both oscillating arms and transmitted to the control unit of the combine. When the lifting exceeds a threshold, the control unit commands the hydraulic system to lift the whole head while the signal persists. When the obstacle is overcome, the interruption of the electrical signal causes the lowering of the head. Then, the mobile portion drops as well and the blade straightens (Fig. 2). A corn head is mounted above the flexible cutter bar, which provides for the detachment and separation of the capitula. It is composed of 9 groups spaced 500 mm apart. In each group, a pair of chains transfers the capitula to an upper auger. Below the chains, two detachment blades can be regulated hydraulically according to the diameter of both the capitula and the apical portion of the stem. Underlying these are the counter-rotating harvesting rollers. Equipped with 4 blades, the stem is simultaneously pulled down and comminution occurs until the capitula is detached by the blades. The first prototype for cardoon harvesting was designed as part of the BIOCARD Project (Pari et al., 2008, 2009a,b, 2011). The head used in this study is the second version developed by CREA-ING in collaboration with Cressoni Ltd. in the framework of the BIT3G project for developing operation in the marginal stony fields of the Sardinia Region (Pari et al., 2014).
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
473
Fig. 2. Diagramatic representation of the sensor system.
2.2. Methodological approach A test-track was prepared to monitor the activation sequence of the three systems. Seven trapezoidal wood blocks were fixed to the ground and arranged in a single row on a flat stripe of soil 30 m long. The obstacles had a frontal surface inclined at 45◦ from the ground with progressively increasing heights (10, 15, 20, 25, 30, 35, 40 cm). They were spaced 4 m apart, the distance needed by the head to restore its normal position after overcoming each obstacle. Considering that the bar flexibility is the highest in the centre and the lowest on the sides (Fig. 3), three series of tests were carried out varying the relative position between the bar and the obstacle row, as follows: Pos 1 – lateral (LAT), on the left side of the head (outside face of the wheel) 0.25 m from the head edge; Pos 2 – intermediate (INT), near the inside face of the wheel, 1.65 m from the left head edge; Pos 3 – central (CEN), at the midpoint of the head, between the wheels, 2.25 m from the left head edge. 2.3. Sensor system The combine and the mobile elements of the head were equipped with potentiometric transducers, capable of monitoring the movements by measuring the electric signal variations they induce. The data were stored by an acquisition control unit (Fig. 4). Moreover, the combine ladder was equipped with a “wheel of Peiseler” (Fig. 4), measuring the distance and the actual travel speed. The tests were carried out in the left half of the head, assuming that the behaviour of the head was similar on the opposite side. To monitor the activation sequence of the three systems responsible for avoiding the contact between the blade and obstacle, the following sensors were mounted (Fig. 5): 1. four piston transducers (PT1, PT2, PT3 and PT4) for measuring the flexion endured by the blade and the counter-blade on the obstacles. In each test, the transducers were installed on the cutting bar in relation to the obstacle row, according to the above mentioned LAT, INT, CEN (Fig. 5);
2. two wire transducers for measuring the movements of the bladeholder hinged on the frame, on both left and right side (LT and RT) (Fig. 6); 3. one wire transducer placed between the head and combine (Fig. 6) for measuring the lifting of the head (HL) operated by the hydraulic system.
For each position (LAT, INT, CEN), the test was replicated five times for a total of 15 runs. In each run, the sensors registered the following: the distance travelled and time required, the heights of the piston transducers (PT1, PT2, PT3, and PT4), the height of the blade holder at both left and right sides (LT and RT), and the hydraulic lifting height of the head (HL). The machine started each run at a distance sufficient to reach the desired speed (approximately 1 km h−1 ), before hitting the first obstacle. All five replicates were conducted by the same driver.
2.4. Analysis of the signal During each run, the signals sent by the sensors were detected at a frequency of 10 Hz. Depending on the length of the test-track and the time required to accomplish a run, the number of acquisitions per run ranged from 1100 to 1800. To allow for a comparison of overlapping measurements, for each row position and for all replications, the acquisitions occurring ± 1 m from the peak of the maximum height of the bar relative to the obstacle were considered. For a detailed analysis of the signals, we examined the behaviour of the inputs registered for each sensor prior to the maximum point reached by the HL sensor (Fig. 7). The activation delta (Act ) and the duration of the signal (DSig ) were taken into account. The activation delta (Act ) was the difference (in seconds) between the start of the signal sent by the transducers (LT, RT, PT1, PT2, PT3, PT4) and the start of the command for head lifting (HL). The duration of the signals (DSig ) represented the time taken for the sensors and the head to reach the maximum value. For both measures, the values registered for obstacle 2 were not included in the analysis because signal anomalies were found in all treatments (see below).
474
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
Fig. 3. Arrangement of the wood obstacles. Lateral (A) and top view (B) of the test-track. Above each obstacle it is reported the corresponding height in cm.
Fig. 4. Control unit (left) and “wheel of Peiseler” (right).
Fig. 5. Piston transducers placed on the left side of combine (PT1, PT2, PT3, PT4) when the obstacles were in lateral (left) and in central position (right).
2.5. Statistical analysis All data were analysed with the software PAST. Homoscedasticity and normality were checked prior to testing. If the normality of distributions was verified, the data were analysed using one-way ANOVA. In this case, the Duncan’s multiple range test was used for the separation of the means. When the data distribution deviated from normality, the Kruskal-Wallis non-parametric test was
applied and the differences were tested according to the MannWhitney pairwise test. 3. Results and discussion The combine travelled for a net distance slightly shorter than the whole length of the test-track (Table 1). The variability observed among the three positions of the obstacles is related to the need
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
475
Fig. 6. Wire transducers between head and combine for measuring the movement of the head (HL) (a), wire transducers for measuring the displacement of the mobile part placed on the right (RT) and the left (LT) side of the cutter bar (b).
Table 1 Space, time and speed of the combine registered for the different positions of the obstacles.
Fig. 7. Parameters considered for the analysis of head movement (DSig is represented only for the HL).
of the operator to reach the first obstacle at a steady speed. The row in the lateral position required the driver to pay closer attention to maintaining the correct direction, which caused an increase
Obstacle
Acquisition (n.)
Space (m)
Time (s)
Speed (km h−1 )
Lateral Intermediate Central
1599 ± 158,31 1414 ± 105,74 1412 ± 252,21
27.73 ± 0.4 29.12 ± 0.8 27.47 ± 1.2
159.02 ± 15.8 140.52 ± 10.6 140.30 ± 25.2
0.63 ± 0.06 0.75 ± 0.05 0.72 ± 0.12
Mean
1475 ± 172,09
28.10 ± 1.1
146.60 ± 19.2
0.70 ± 0.09
in time and a reduction in the forward speed for this test. On the other hand, the presence of the obstacles in the intermediate position needed more space before the running began to position the combine correctly. The different types of sensors showed a distinct range of oscillations according to the position in which they were applied and the type of measurement monitored (Fig. 8). Moreover, the displacement of the obstacles row (LAT, INT, CEN) also influenced the signals of the sensors. As a general rule, the magnitude of the oscillation decreased as follows: head lifting (HL) > left and right edges (LT, RT) > blade and counter-blade (PT1, PT2, PT3 and PT4). The behaviour of the wire sensors monitoring the lifting and lowering of the left and right edges of the blade-holder was particularly divergent when obstacles were in the lateral or central position. As described below, the transducers of the left side showed a signal shift towards positive values, while those on the right were negative. The data reflects the flexibility of the bar as follows: during the overcoming of the obstacle, it temporarily assumed an S shape, with positive values in respect to the starting point, in response to the
Fig. 8. Box plot showing the 25–75 percent quartiles, the median (horizontal line) and the minimal and maximal values (whiskers) of head lifting (HL), signals of left and right wire transducers (LT and RT) and the piston transducers (PT1, PT2, PT3 and PT4), in lateral (A), intermediate (B) or central (C) position.
476
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
Fig. 9. Height of the bar (mm) in correspondence of the relative position of the obstacles. For each point of maximum lifting the significance of F value is showed (* = P < 0.05; *** = P < 0.01; ns = not significant).
obstacle, whereas on the opposite side, the blade holder dropped below the reference level. During the run, the lifting of the head was somewhat associated with the height of the obstacle (Fig. 9). The height of the bar showed a tendency to rise in the presence of higher obstacles. The graphics of the bar height were almost overlapping for the lateral and central positions of the rows of obstacles, whereas when the row was in the intermediate position, the head lifting was more pronounced. In general, the signals for the second obstacle were not easily identifiable, therefore the corresponding measurements were maintained in the general analysis of the signal level, but were disregarded in the detailed analysis of the signals (act and DSig ) to avoid biased results. When compared with the other obstacles, the signal observed in the second obstacle would appear anomalous. However, such an anomaly was independent from the position of the obstacle, and the detection of the obstacle did occur, but at a lower intensity. This highlights at least two important aspects of the system. The first is its capacity for detecting a specific combination of adjacent low obstacles (the first and the second). The second concerns the distance. Although not analysed in the present work, such a degree
Fig. 10. Signal of the wire transducers (LT and RT) compared to the height of head recorded by the HL sensor in presence of increasing obstacle in lateral (A), intermediate (B) and central (C) position.
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
477
of sensitivity should allow for the detection of closer obstacles (well below 4 m): the timing of the signal should remain the same but it would be reasonable to expect a variation of its intensity. Such a hypothesis needs to be confirmed, however, either in a test-track or in non-uniform conditions. The signal relating to the displacement of the mobile part was influenced by the point of application of the transducer and by the position of the obstacle row (Fig. 10). The main difference between the LT and RT signal concerned the shape of the signal. Both sensors reached the maximum level of the input immediately prior to the maximum lifting of the head, but, while the LT assumed a shape of narrow “bell”, the RT dropped negatively once the maximum value was reached, thus producing a “reversed bell” diagram. This was clearly visible when the obstacles were in the lateral position (Fig. 10A) because the left side of the mobile part was subject to an upward solicitation while the right side folded downward. Such behaviour represented graphically the effect of the flexibility. While the left half of the cutter bar overcame the obstacles, the right half curved below the “0-position” of the transducers, and the magnitude of folding was driven to some extent by the position of the obstacles. When the obstacles were in the central or intermediate positions (Fig. 10B and C) the RT maintained the same type of signal. Conversely, the LT showed an enlargement of the “bell” shape which was much wider as the LT was far from the obstacle row. As a final result, however, both transducers correctly detected the obstacle by starting the signal before the command of head lifting. Being positioned directly on the cutter bar, the signals sent by the piston transducers were much clearer, uniform and coordinated than the wire transducers’ signals (Fig. 11). The four transducers accurately detected all stages of overcoming (flexion, lifting, and return to the starting point) performed by the cutter bar in response to each obstacle, showing also that the PT signal preceded the lifting of the head. The tops of the curves were reached when the bar began the ascending phase. As previously observed for the LT position, the signal associated with each obstacle had the shape of a narrow “bell” when the row was lateral (Fig. 11A), while such configuration was modified by the presence of a “shoulder” when the combine encountered intermediate or central obstacles. 3.1. Analysis of the signals Once the signals generated by the sensors reached the control unit, the system took a time interval (Dsig ) ranging from 1.25 s (obstacle 4) to 2.51 s (obstacle 7) to lift the head before overcoming the obstacle (Fig. 12). Interestingly, the data revealed a threshold value beyond which the head required a different range of time as follows: below 25 cm, the lifting occurred between 1.0 s and 1.5 s, while for obstacles comprised between 30 cm and 40 cm, the head lifting time ranged from 2.0 s to 2.5 s. Such observations prompt some needed explanation regarding some hidden implications in the study. Cardoon is a “poor” crop suitable for marginal areas where the accurate soil tillage may lag behind. The use of practices such as minimum or no-tillage as well as the absence of soil levelling or the presence of obstacles of various natures (often stones) leads the combine to work on an irregular surface where the difference in height can reach a remarkable value. Thus, the apparently unrealistic sizes of the highest obstacles were not a mere experimental exercise, but were aimed at reproducing field situations typical of highly uneven terrain. Further points of discussion emerged by extending the analysis of DSig to all sensors in response to the positions of the three obstacles (Fig. 13). The RT showed the shortest signal duration in both the lateral (0.89 s) and intermediate positions (1.56 s). In the same test, the corresponding and opposite wire transducer (LT) produced a signal statistically longer than the RT. When the obstacle row
Fig. 11. Signal of the piston transducers (PT1, PT2, PT3 and PT4) compared to the height of head recorded by the HL sensor in presence of increasing obstacle in lateral (A), intermediate (B) and central (C) position.
Fig. 12. Mean (±SE) of the duration of head lifting. Means sharing common letters are not statistically different for P ≤ 0.01, following post-hoc Duncan’s test.
478
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479
Fig. 13. Mean (±SE) of the length of DSig of the six transducers and the head lifting with lateral (A), intermediate (B) and central (C) obstacles. Means with different letters are significantly different for P < 0.05 (lowercase letters) or P < 0.01 (capital letters), following post-hoc Mann-Whitney test.
was equidistant between the LT and RT (CEN) the signal length was comparable, thus highlighting that the presence of an obstacle is accurately perceived by both sides of the head. The time elapsed between the beginning of the signal from the piston transducers and the angular transducers input for head lifting was over 2 s and was the longest whenever the obstacles were positioned. It should be noted that the obstacle row in the intermediate position caused an increase in the signal length, suggesting that some specific positions could require a more careful “reading” to precisely identify the obstacle. As expected, although to a different extent, the raising of the head was always preceded by the signalling of the sensors (Fig. 14). The obstacles in the lateral position were detected within an interval of 1.49 s (RT)–2.09 s (PT3) before the head lifting, while when the position was intermediate, the sensors sent the signal 2.08 s (LT) – 2.54 s (PT4) earlier. The detection of a central obstacle was more delayed, requiring between 1.22 s (LT) and 1.82 s (PT3). Thus, it can be speculated that although the system is able to perceive the soil irregularity along the entire head width, some specific positions can be trickier and require an improvement in the method of detection. The wire transducers (LT and RT) showed an opposite behaviour in relation to the obstacle position. The LT anticipated the RT in sending the signal just when the obstacles were in the lateral position (hence, on the same side of the transducer). Shifting the obstacle far from the left side caused an earlier activation of the RT with a difference that became significant for the central
Fig. 14. Mean (±SE) of the activation delta (act ) of the six transducers with lateral (A), intermediate (B) and central (C) obstacles. Data were calculated considering the lifting of the head as point 0. Where reported, means with different letters are significantly different (P < 0.05), following post-hoc Mann-Whitney test.
obstacles. The signals from the piston transducers started a little bit earlier than those from the LT and RT when the obstacles were lateral and central, while in the intermediate position they started simultaneously. The extreme sensibility of the PT is a consequence of their point of application, corresponding to the obstacle row, for detecting the necessary flexion of the blade and counter-blade. Their earlier activation confirms that the signalling started just as the coulters of the bar began to “feel” the obstacle. The experimental data has some practical implications tied to the correct tuning among the components of the combine. The sum of Act and Dsig accounted for the time between the electric input and the complete head lifting (Table 2). In one specific case, travelling with an average speed of 0.7 km h−1 , corresponding to 0.19 m s−1 , the combine needed a space of 0.74 m to overcome the obstacle. Cardoon harvesting can be carried out at 0.80 m s−1 (Pari et al., 2014), a value four-fold higher than the speed maintained in the present study. Assuming the same reaction time is maintained, increasing speeds will require a reduction in the Act or the Dsig . The first concerns the electronic system and hence the frequencies of the pulse of the input signal. The modification of the Dsig would require a change to the hydraulic system, namely the section width of the pipeline controlling the head lifting. Such outcomes agree in part with those of
L. Pari et al. / Industrial Crops and Products 94 (2016) 471–479 Table 2 Time and space required for overcoming different obstacles (combine speed 0.19 m s−1 ). Obstacle
Act (s)
Dsig (s)
Total Time (s)
Space required (m)
1 (10 cm) 3 (20 cm) 4 (25 cm) 5 (30 cm) 6 (35 cm) 7 (40 cm)
2.80 1.73 1.55 2.32 1.82 1.91
1.29 1.39 1.25 2.29 2.44 2.51
4.1 3.1 2.8 4.6 4.3 4.4
0.78 0.59 0.53 0.88 0.81 0.84
Mean
2.02
1.86
3.9
0.74
Xie et al. (2010, 2013), which identified the mechanical configuration (combine and head) and the electrohydraulic actuation of the head as the main subsystems affecting the performance limitations in the control of head’s height. The authors also listed the replacing of tires with tracks, the use of active suspension, the redesign of suspension elements and the use of very high performance servo hydraulics as possible means for improving the performance (Xie et al., 2013). 4. Conclusion In difficult terrains, the use of conventional combines leads to a substantial loss of biomass because the bar must be raised off the ground 40–50 cm to avoid breakage. The flexible head studied in the present work showed technical characteristics useful for reducing such losses. The integration of the sensors system combined with the flexibility deriving from the specific building material, allowed for the adjustment of the bar height in a dynamic way, almost point-by-point, as can occur in irregular terrains. The control system was able to quickly and effectively detect obstacles ranging from 10 to 40 cm, a deliberately excessive height tested to evaluate the technical limits of the system. Furthermore, the test showed that the three head elements activated during the overcoming of an obstacle combined efficiently with each other. Consequently, the adoption of the system can be recommended, independent of the species, whenever it is necessary to maintain the cutting system as close as possible to the ground. However, the bottleneck of increasing the combine speed remains. Further in-depth studies are required to address the challenging objective of integrating possible mechanical and/or electronic solutions able to efficiently exploit the potential of the sensor system while maintaining a reasonable cost. Acknowledgements This work was supported by the BIT3G (MIUR) and Suscace (MIPAAF) projects. References Angelini, L.G., Ceccarini, L., o Di Nasso, N.N., Bonari, E., 2009. Long-term evaluation of biomass production and quality of two cardoon (Cynara cardunculus L.) cultivars for energy use. Biomass Bioenergy 33 (5), 810–816, http://dx.doi.org/ 10.1016/j.biombioe.2008.12.004. Fernández, J., Manzanares, P., 1990. C. cardunculus L. a new crop for oil, paper pulp and energy. In: Proceedings 5th European Conference on Biomass for Energy and Industry, London, UK, pp. 1184–1189. Fernández, J., Curt, M.D., Aguado, P.L., 2006. Industrial applications of Cynara cardunculus L. for energy and other uses. Ind. Crops Prod. 24, 222–229, http:// dx.doi.org/10.1016/j.indcrop.2006.06.010. Francaviglia, R., Bruno, A., Falcucci, M., Farina, R., Renzi, G., Russo, D.E., Sepe, L., Neri, U., 2016. Yields and quality of Cynara cardunculus L. wild and cultivated cardoon genotypes. A case study from a marginal land in Central Italy. Eur. J. Agron. 72, 10–19, http://dx.doi.org/10.1016/j.eja.2015.09.014. Gherbin, P., Monteleone, M., Tarantino, E., 2001. Five year evaluation on Cardoon (Cynara cardunculusL. var.altilis) biomass production in a Mediterranean environment. Ital. J. Agron. 5, 11–19.
479
Glancey, J.L., 1997. Analysis of header loss from pod stripper combines in green peas. J. Agric. Eng. Res. 68 (1), 1–10, http://dx.doi.org/10.1006/jaer.1996.0122. Gominho, J., Lourenc¸o, A., Curt, M., Fernández, J., Pereira, H., 2009. Characterization of hairs and pappi from Cynara cardunculus capitula and their suitability for paper production. Ind. Crops Prod. 29 (1), 116–125, http://dx.doi.org/10.1016/j. indcrop.2008.04.022. Gominho, J., Lourenc¸o, A., Palma, P., Lourenc¸o, M.E., Curt, M.D., Fernández, J., Pereira, H., 2011. Large scale cultivation of Cynara cardunculus L. for biomass production–a case study. Ind. Crops Prod. 33 (1), 1–6, http://dx.doi.org/10. 1016/j.indcrop.2010.09.011. Grammelis, P., Malliopoulou, A., Basinas, P., Danalatos, N.G., 2008. Cultivation and characterization of Cynara cardunculus for solid biofuels production in the Mediterranean region. Int. J. Mol. Sci. 9, 1241–1258, http://dx.doi.org/10.3390/ ijms9071241. Hirai, Y., Schrock, M., Maski, D., Taylor, R., 2005. Loss evaluation of soybean harvesting on four types of headers. In: ASAE Agricultural Equipment Technology Conference, Louisville, Kentucky, pp. 1–4. Hummel, J.W., 1983. Impact cutting of soybeans using flexible cutting systems. Trans. ASAE 26 (5), 1315–1319, http://dx.doi.org/10.13031/2013.34123. Mahmoodi, E., Jafari, A., Mobli, H., Salokhe, V.M., Jayasuriya, H.P.W., Soni, P., 2007. A new idea for construction of a flexible cutter bar mower for listed lands. In: Salokhe, V.M., Jayasuriya, H.P.W., Soni, P. (Eds.), Proceedings of the International Agricultural Engineering Conference 2007. Cutting Edge Technologies and Innovations on Sustainable Resources for World Food Sufficiency. Bangkok, Thailand (unpaginated, ISBN 978-974-8257-48-8). Mauromicale, G., Sortino, O., Pesce, G.R., Agnello, M., Mauro, R.P., 2014. Suitability of cultivated and wild cardoon as a sustainable bioenergy crop for low input cultivation in low quality Mediterranean soils. Ind. Crops Prod. 57, 82–89, http://dx.doi.org/10.1016/j.indcrop.2014.03.013. Nave, W.R., Hummel, J.W., Yoerger, R.R., 1977. Air-jet and row-crop headers for soybeans. Trans. ASAE 20 (6), 1037–1041, http://dx.doi.org/10.13031/2013. 35698. Pari, L., Fedrizzi, M., Gallucci, F., 2008. Cynara cardunculus exploitation for energy applications: development of a combine head for threshing and concurrent residues collecting and utilization. In: Proceedings of the 16th European Biomass Conference & Exbition, Valencia, Spain, pp. 504–508. Pari, L., Civitarese, V., Assirelli, A., Del Giudice, A., 2009a. Prototype for Cynara cardunculus capitula threshing and biomass windrowing. In: Proceedings of the 17th European Biomass Conference & Exibition, Hamburg, Germany, pp. 262–267. Pari, L., Civitarese, V., Assirelli, A., Del Giudice, A., 2009. Il prototipo che abbatte i costi della raccolta del cardo Agroenergie dall’impianto alla raccolta. Special issue of L’Informatore Agrario 29, 8–11. Pari, L., Civitarese, V., Assirelli, A., Del Giudice, A., 2011. Sviluppo di un prototipo per la raccolta differenziata dei diverse frazioni di cardo. In: Pari, L. (Ed.), Lo sviluppo delle colture energetiche in Italia. Il contributo dei progetti di ricerca SUSCACE e FAESI. Nuova Cultura, Rome, pp. 671–682 (ISBN 9788861347304). Pari, L., Gallucci, F., Civitarese, V., Acampora, A., Sanzone, E., Alfano, V., Assirelli, A., 2014. Innovazione meccanica per la raccolta del cardo in ambiente arido. In: Valorizzazione delle risorse agricole nell’ottica del concetto di bioeconomia. Progetti di ricerca SUSCACE e FAESI. Special issue of Sherwood 203, 43–46. Quick, G.R., Buchele, W.F., 1974. Reducing combine gathering losses in soybeans. Trans. ASAE 17 (6), 1123–1129, http://dx.doi.org/10.13031/2013.37044. Sengo, I., Gominho, J., d’Orey, L., Martins, M., Duarte, E., Pereira, H., Ferreira-Dias, S., 2010. Response surface modelling and optimization of biodiesel production from Cynara cardunculus oil. Eur. J. Lipid Sci. Technol. 112, 310–320, http://dx. doi.org/10.1002/ejlt.200900135. Siemens, M.C., 2006. Effect of guard spacing, guard attachments and reel type on chickpea harvesting losses. Appl. Eng. Agric. 22 (5), 651–657, http://dx.doi.org/ 10.13031/2013.21997. Solano, M.L., Manzanedo, E., Concheso, R., Curt, M.D., Sanz, M., Fernández, J., 2010. Potassium fertilisation and the thermal behaviour of Cynara cardunculus L. Biomass Bioenergy 34, 1487–1494, http://dx.doi.org/10.1016/j.biombioe.2010. 04.025. Tuck, G., Glendining, M.J., Smith, P., House Jo, I., Wattenbach, M., 2006. The potential distribution of bioenergy crops in Europe under present and future climate. Biomass Bioenergy 30, 183–197, http://dx.doi.org/10.1016/j.biombioe. 2005.11.019. Xie, Y., Alleyne, A., Greer, A., Deneault, D., 2010. Header height control of a combine harvester system. Cambridge, Massachusetts, USA In: ASME 2010 Dynamic Systems and Control Conference, vol. 1, pp. 7–14. Xie, Y., Alleyne, A.G., Greer, A., Deneault, D., 2013. Fundamental limits in combine harvester header height control. J. Dyn. Syst. Meas. Control 135 (3), 1–8, http:// dx.doi.org/10.1115/1.4023209. Zyla, L.E., Kushwaha, R.L., Vandenberg, A., 2002. Development of a new crop lifter for direct cut harvesting dry bean. Can. Biosyst. Eng. 44, 2–9.