Otter Trawl Lab: Door Spread, Towing Drag and Codend Mesh Selectivity

Setting up the lab bench...

Presets

Live readouts

Graphs

Codend retention versus fish length (blue: analytic, grey dashed: population, orange dots: live fish, vertical line: minimum landing size)

Yield per recruit versus mesh size at the current fishing mortality (white: current, orange: best mesh)

Door spread versus door angle of attack (vertical line: stall)

Numerical checks

running...

How to use

Drag the 3D view to orbit; use the camera buttons for the vessel, the doors, the net and the codend. Pick a preset, then change the mesh, the doors or the tow.

Compare Cod, 120 mm diamond with Small mesh: watch the retention curve move left and the undersized share of the catch jump.
Raise the door angle past 30° (or load Door stalled): lift collapses and the net spread shrinks.
Increase Codend fill with a diamond mesh: the meshes close and L50 falls; switch to square mesh and it does not.
Watch the live orange dots on the selection graph converge to the analytic curve as fish pass the mesh.

The science

Otter doors: lift sets the spread. Each door is a low-aspect-ratio foil towed at angle of attack α to the tow direction. With dynamic pressure q = ½ρV² and door area A_d it gives an outward force and a drag

L_d = q C_L(α) A_d, D_d = q C_D A_d, C_L = 2πAR/(AR+2) · α (pre-stall), C_D = C_D0 + C_L²/(π e AR)

(AR = 1 here, a deliberate stall model above 30° where lift collapses). The net pulls the door back with half its drag D_n/2 and, through the sweeps (angle γ), inwards with (D_n/2) tan γ. The warp must point along the resultant of all door forces, so in plan view

tan β = [ L_d - (D_n/2) tan γ ] / [ D_d + D_n/2 ], spread = 2 R_h sin β, R_h = √(S² - h²)

with warp length S, depth h and the sweep geometry sin γ = (spread/2 - w/2)/(sweep length). The page solves this self-consistent equilibrium by bisection; the checks compare it with a separate damped fixed-point iteration. The net drag uses D_n = q C_dA (a tuned drag area growing with headline length and with codend fill), warp drag uses a cylinder drag coefficient 1.2. Towing force F_t = 2D_d + D_n + warp drag, engine power P = F_tV/0.45 + 120 kW hotel load, fuel at 0.21 kg/kWh.

Codend selectivity from geometry. A diamond mesh of stretched size M_s has bars of length M_s/2. When open to an angle φ the largest circle that fits the hole has circumference π (M_s/2) sin φ. A fish of length L has girth g L (g about 0.4-0.5), a soft body that can squeeze through with a deformability factor d = 1.15, and both the girth and the real mesh size vary by a few per cent. So the fish escapes if g L ≤ d (π/2) M_s sin φ and the retention curve is a probit (normal CDF) with

L₅₀ = d (π/2) M_s sin φ / g, r(L) = Φ[(L - L₅₀)/σ], σ = L₅₀ √(CV_g² + CV_m²)

A catch-filled diamond codend is pulled shut (φ falls from 72° with load), which lowers L₅₀ and lets small fish be kept; a square-mesh codend stays open (φ = 90°) and gives a higher, sharper L₅₀. The factor d and the opening law are teaching calibrations that give selection factors L₅₀/M_s of about 3-3.6 for roundfish; real selectivity must be measured by cover-net or paired-gear trials and varies with species, season and handling.

What the mesh does to the stock. With von Bertalanffy growth L(t) = L_∞(1 - e^-K(t-t₀)), weight W = aL³, natural mortality M and fishing mortality F S(L), the yield per recruit is the integral of F S N W dt along the cohort (numerical age integration here). For knife-edge selection at age t_c it reduces to the Beverton-Holt closed form, which the checks use as an independent test, as well as the Baranov catch equation for the numbers caught. The spawning-biomass ratio compares spawners per recruit with the unfished value. Yield per recruit ignores recruitment collapse and economics, so the best mesh printed here is not a management recommendation.

Species parameters are illustrative round numbers, not an assessment of any real stock. Fuel and catch rates are order-of-magnitude teaching values. The model has no fish behaviour in front of the net, no bottom contact or sediment clouds, no net shape change under load beyond the codend opening, and no escape through the upper panels. Educational model only.