Introduction
Otter boards, used to spread the net wings of most trawl gear, play an important role in the geometry of fishing gear and catch efficiency. However, their bulky size and weight cause problems in fishing operations and can negatively impact seabed ecosystems. Recent studies have suggested substituting flexible, kite-like canvas devices for otter boards (Bae et al., 2006; Balash et al., 2015).
The motion of the otter board (e.g., angle of attack or tilt) can change the coefficient of shearing force or distance between the otter boards, and affects net gear motion in terms of net height or spread (Crewe, 1964; FAO, 1974; Lee et al., 1986). It has been shown to impact shrimp trawl catches (Broadhurst et al., 2012; Priour and Prada, 2015). Crewe (1964) reported that the main motion factors were roll and pitch. The tilt of the otter board, caused by rigging (FAO, 1974), waves, and contact with the bottom (Parketal., 1993a, 1993b; Matsushitaetal., 2005; Salaetal, 2009; Politis et al., 2012), can be affected by stability in action and spreading efficiency (Matuda et al., 1990). Most studies on the hydrodynamics of otter boards have focused on the shearing coefficient by varying tilt (Park et al., 1993b, 1996).
The tilt of otter boards has been measured under field conditions using various methods (Sivadas, 1970; Lee et al., 1986; Lin et al., 1989). Time series of variations in tilt have been regarded that changes are periodic (Lee et al., 1986), which has also been observed in variations in warp tension (Matsushita et al., 2005). The variations in bottom trawl otter board tilt have been analyzed during fishing operations in terms of periodicity or amplitude relative to the tilt variation of the codend (Kim, 2014). Flow visualization methods have shown that the otter boards can also cause a complex turbulence wake over angle of attack (Ko et al., 1990Ko et al., 1990; Park et al., 1994a, b; Fukuda et al., 1997). The overall motion of the otter board (tilt or flow) may be closely related to unbalanced forces caused by warp and net-pendant tension due to the ship’s motion (O’Neill et al., 2003) and through whole-net drag. The tilt variations and turbulence in the codend caused by oscillating motions of the gear affect fishing selectivity and fish escape (Kim, 2013).
However, whether periodic tilt variation produces a similar periodic wake in the otter board during fishing operations has not been investigated. Flow measurements or visualization are generally based on models in flume tanks due to the difficulty associated with field observations. The goal of this study was to compare the relationship between 3-D tilt and flow variations of otter boards under field conditions and analyze their periods and amplitudes.
Seven fishing trials were conducted on the training ship Saebada as preliminary experiments on the bottom-trawl otter boards using Vector, the built-in 3-D tilt and flow sensor. Tilt data (i.e., yaw, pitch, and roll) and 3-D flow velocity were analyzed for period and amplitude using global wavelet and event analysis methods. Then the period phase differences between tilt and flow were compared. These results showed rough, complex periodic variation in tilt and flow and may be useful for otter board design (Takahashi et al., 2015) and wake turbulence control.
Materials and Methods
The experimental fishing gear was a bottom trawl towed by the training vessel Saebada, a 999-GT stern trawler (Gyeongsang National University, LOA 70.57 m, 3,000 PS) (Kim, 2014). The otter pendant, hand rope, and net pendant of the trawl were 13.8 m, 96 m, and 47 m, respectively. The otter board (Fig. 1) was a cambered super-V type (Baekkyung Ind., Pusan, South Korea) and had a length (L) of 3.4 m, chord (C) of 2.0 m (aspect ratio 1.7, 3 panels as 3 colors and 2 slots in Fig. 2), 12% camber ratio, 6° upper and 12° lower sweep angles, and an underwater weight of 1,697 kg at a 24° angle of attack (Kim, 2014).
The tilt and 3-D flow velocity of the otter board were measured by Vector (Nortek, Rud, Norway; Kim, 2013), which was 55 cm long, 7.5 cm in diameter, and weighed 1.5 kg in water. Vector was fixed 40 cm (0.2C) from the end of the trailing edge aligned with the towing direction (24° angle of attack) of the otter board on the port side using an extended stainless steel frame (15-mm diameter) bolted into the otter pendant holes (Fig. 2). The measuring points on the upper trailing edge were 0.2C and 0.17L (trials F1U–F4U in Table 1) and those on the middle trailing edge were 0.2C and 0.5L (trials F5M–F7M in Table 1).
Vector measured 3-D water velocity using a 1.5-cm-diameter water volume 16 cm apart from the end of acoustic probe with an accuracy of ±0.5%. 3-D flow velocity was measured as towing direction U, lateral direction V, and depth direction W at 16 Hz. The resultant flow velocity S was estimated as the root of (U2+V2+W2), which was also used to calculate turbulence rate over 1min (16Hz×60s). Vector also included sensors for measuring temperature (accuracy, 0.1°C), yaw (compass accuracy, 2°; resolution, 0.1°), tilt (pitch and roll, accuracy, 0.2°; resolution, 0.1°), and depth pressure (accuracy, 0.25%; resolution, 0.005%) with a 1-s sampling interval. A positive pitch indicated forward down, and a positive roll indicated tilt to starboard in the range of ±180° when Vector was positioned vertically parallel to the trailing edge of the upper panel of the otter board. Yaw indicated a compass heading of 360°.
Seven fishing operations were conducted, mostly south of Geomun Island in the South Sea of Korea at depths of 90-94 m with a warp length of 500 m, except in trial F4U west of Galdo, which was conducted at a depth of 41-43 m with a warp length of 300 m southwest of Tongyeong (Table 1). During the fishing trials, sea conditions were mostly calm, with wave heights <1 m. The towing speed of the fishing boat was determined using a built-in acoustic Doppler current meter (TS-310, Tokimec, Tokyo, Japan), andover-the-ground speed was measured using a global positioning system.
The periodicities of tilt and flow velocity data from Vector were analyzed using the global wavelet method (GWP) from Torrence and Compo (1998). However, that method estimates period without amplitude. Therefore, peak event analyses (EVP), such as calculation of peak and valley values (Fig. 3), were conducted using custom-made Fortran software (Kim, 2014) following the methods of Narasimha et al. (2007). When the data peaked and then decreased in consecutive time series, the peak values were positive (‘peak event’) and the valley values were negative (‘valley event’). Then, the tilt and flow velocity periods were estimated as the time differences (PR1) between trial TR2 and TR1 (roll) and between TU2 and TU1 (U velocity), respectively (Fig. 3). The amplitude of tilt was estimated as the difference in roll between TR1 and TR3, while flow amplitude was estimated as the difference in velocity between TU1 and TU3. The period time interval between peaks or valleys was limited to <3–11 s for tilt data (1-Hz sampling) and <1–11 s for flow data (16-Hz sampling). The threshold values between the peak and valley had a mean value of ±0.5 standard deviations (SDs). The mean period was estimated from the average of the total intervals between peaks and between valleys, while mean amplitude was estimated from the difference between consecutive peak and valley values.
To calculate the time difference between tilt peaks and flow peaks, the phase difference ratio (Rp) was defined as the time difference (|TU1−TR1|) between flow peak time and tilt period (PR1=TR2−TR1) (Fig. 3) using the following equation:(1)
An Rp value of 0.5 indicates that flow peak occurred at the middle of the tilt peaks, whereas a value of 0 indicates that flow peak occurred simultaneously with tilt peaks. The frequency of Rp was divided into five steps from <0.1 to <0.5 with an interval of 0.1 in each trial.
Results
The tilt of the otter board and codend varied with towing time (Fig. 4). The tilt of the otter board showed periodic variation and the amplitude of the roll was greater than pitch or yaw. The mean values of tilt, resultant flow velocity, and turbulence are shown in Table 2. The mean pitch of the otter board was -3° to -7°, and roll ranged from -5° to -25°, indicating that the orientation of the upper part of the otter board was slightly backward and outward. The SD of roll was greater than that of pitch, indicating that it varied more.
The flow values of U, V, and W from the trailing edge wake also showed periodic variation. V and W varied between positive and negative values, as shown by the transition of the flow direction, whereas U was generally positive. The mean flow velocities were less than or equal to towing speed due to wake turbulence at the trailing edge rather than at the leading edge. The mean turbulence rate (Tr) was 30-50% and kinetic energy (TKE) was 0.1-0.23 m2/s2, except in trial F4U, which occurred in shallow coast water.
Period spectra of flow velocity in the otter board from the global wavelet analysis are shown in Fig. 5. The dominant peak periods of tilts and flow velocities of the otter board are shown in Table 3. The dominant peak periods (2.5-4.9 s) mostly appeared in the 3-D flow velocity of the otter board and there was no difference between upper (trials F1U-F4U) and middle sections (trials F5M-F7M) of the otter board. The dominant period of V, with a mean value of 3.4 s, was significantly shorter than the dominant periods of yaw, pitch, or roll, with mean values of 4.1-4.5 s (t-test, n=7, P<0.05). However, the dominant periods of U or W were not significantly different from any dominant periods of tilt (n=7, P>0.05).
The period and amplitude of tilt and flow velocity from the peak event analysis are shown in Fig. 6 as an example of frequency. Almost all period frequency and amplitude distributions of flow appeared as normal distributions. The mean values of period and amplitude from the peak event analysis are listed by tilt and flow velocity and by trial in Table 4. The product of the number of events (n in Table 2) and the mean period was nearly equal (99-100%) to the total sampling time (Table 2). The mean values of the tilt (yaw, pitch, and roll) periods (5.1 s) were significantly greater than those of the 3-D flow (U, V, and W) periods (3.2-3.4 s) (t-test, p<0.001), respectively. Amplitudes of roll were significantly greater than those of yaw or pitch (p<0.001) while the amplitude of W was significantly lower than those of U or V (p<0.001). Additionally, periods and amplitudes of tilt or flow velocity showed no differences between upper and lower trailing edges, respectively, in only 3-4 fishing operations.
The deviations (SD in Table 2) or amplitudes (Amp in Table 4) of roll were greater than those of yaw or pitch. The relationships between mean deviation or mean roll amplitude and mean turbulence rate of 3-D flow velocity in seven trials are shown in Fig. 7. Although there was no significant relationship between Tr and Amp (r2<0.01, p>0.2), Tr had a significant relationship with SD (p<0.05) represented by the following equation:(2)
The relationship between the mean flow and tilt periods for seven trials using global wavelet (GWP in Table 3) or peak event analyses (EVP in Table 4) is shown in Fig. 8. The relationship between flow period (Pf, s) and tilt period (Pt, s) given by the event analysis (p<0.05) is stronger than that from the global wavelet method (p > 0.05), represented by the following equations:
The peak event time between roll and flow velocity (U, V, W, and S) by the event analysis has some time lag as a peak time phase difference as shown in Fig. 2. As also demonstrated by the mean period difference between tilt and flow velocity (Table 3), flow peak times occurred once or twice between tilt peaks. The Rp values between tilt and flow velocity estimated by eq. 1 are shown in Fig. 9 as a frequency distribution with a step interval of Rp = 0.1. Most Rp values between tilt and U or W in trial F1U and those between tilt and W in trial F5M were significantly different (chi-square test, p<0.05) from random distributions (e.g., frequency ratio 0.2). However, for the other five trials, the distributions of Rp between tilt and flow velocity were not significantly different from random distributions at each step.
Discussion
The periods of tilt components (i.e., yaw, pitch, and roll) determined using the global wavelet method or peak event analysis in this study were similar to an approximated tilt period with short sampling (Lee et al., 1986) and an estimated period of warp tension (Matsushita et al., 2005). The yaw, pitch, and roll of the otter board varied simultaneously as a 3-D deformation (so-called ‘resultant tilt’; Kim, 2014). Similarly, resultant tilts also varied periodically to generate complex turbulent wake around the trailing edge of the otter board (Ko et al., 1990Ko et al., 1990; Park et al., 1994a, 1994b; Fukuda et al., 1997). To the best of the author’s knowledge, this is the first time that these complicated wake flows at the trailing edge of an otter board have been measured and analyzed by Vector under trawl operating conditions. Although Vector can sample at up to 64 Hz, this study used 16 Hz because of memory capacity from total fishing time and longer flow peak variations (>1 s).
The 3-D flow velocities measured at a point on the trailing edge of the otter board during fishing operations showed complex variations with time due to the wake and changes in tilt. Those flow velocities could represent a 1.5-cm water ball vortice at a relatively fixed point on the otter board. The flow data were analyzed using both global wavelet method and peak event analysis (Kim, 2014). The peak event analysis was considered useful for data with highly variable peak point values in long time series to extract period and its amplitude between neighboring peaks (Narasimha et al., 2007; Kim, 2013), whereas the global wavelet method was useful for a more limited time period and provided no amplitudes (Torrence and Compo, 1998). The 3-D flow velocities from the peak event analysis showed little periodicity and had greater deviations in period and amplitude. The period and amplitude between the upper and middle trailing edges were not significantly different. The surface of the measured cross section of the otter board has a relatively rough surface (e.g., from rope holes, slots, and bars) compared with a hydrofoil. Those protruding objects can generate more turbulence until posterior trailing edge rather than even surface hydrofoil (Karbasian and Kim, 2016) or airfoil (Martinat et al., 2008; Prangemeier et al., 2010).
The relationship between 3-D tilt and 3-D flow was represented as the phase difference of period (i.e., time lag between tilt and flow velocity peaks) under the assumption of simultaneous start times, although they had different sampling rates (tilt, 1 Hz; flow, 16 Hz). The period phase differences were random in five of seven trials (irrespective of tilt), whereas those in the other two trials (F1U and F5M) were not random but did not occur in the middle of the tilt period. These two cases may have been affected by side tidal currents in the towing direction near the sea bed (see Table 1). This could be elucidated by further measurements of tidal currents using Vector near the sea bed in fishing grounds (Kim and Hong, 2014).
The observed tilts were showed as pitch up and roll out as general rigging and operation of the bottom trawl. However their variations were considered as higher even under relatively calm weather and resultantly caused higher turbulence at trailing edge. Therefore the device to reduce turbulence could be suggested as wing tip or diffusor adapted from aerofoil. It is also needed more flat surface of the board including bracket, pendant ring etc for boundary layer control. The tilt variation was also affected by irregular contact of the shoe part with angle of attack by pitching mainly. The attachment of separated shoe could be reconsidered as smaller angle of attack ≈0 by changeable hinge to reduce area of the sea bed contact.
Investigations of flow structure and vortices for hydrofoils under pitching conditions have used numerical analyses (Karbasian and Kim, 2016) and wind tunnel experiments (Ducoin et al., 2009), but are difficult to compare due to differences in objects, analytical methods, and water flow conditions. However, computational flow investigations (Jonsson et al., 2013) may be possible and necessary for periodic tilting conditions identical to field operations referring to airfoil cases (Gharali and Johnson, 2013; Luetal., 2013). Additional studies are needed to reduce tilt variation via boundary layer control or to minimize sea bed contact (Ivanovicetal., 2009) by modifying shoe tread of bottom shoe (McHugh et al., 2015).
Conclusion
The overall motion of the otter board (tilt or flow) may be closely related to unbalanced forces caused by the ship’s motion and through whole-net drag. The tilt variations and turbulence in the codend caused by oscillating motions of the gear affect fishing selectivity and fish escape. Seven fishing trials were conducted on the training ship Saebada as preliminary experiments on the bottom-trawl otter boards using Vector, the built-in 3-D tilt and flow sensor. Tilt data of yaw, pitch, and roll and 3-D flow velocity were analyzed for period and amplitude using global wavelet and event analysis methods. The main peak period of 3-D flow was occurred twice within a peak period of the tilt by peak event analysis. The turbulence rate of flow was 30-50% at the trailing edge closely related to roll deviation and possibly reduced spreading efficiency. In order to reduce turbulence boundary control by wing tip or diffusor adapted from aerofoil could be needed in addition to more smooth surface followed model test in flume tank first.
