Hydrodynamics of Paddling
Use science to go faster. Every Spearer is looking for the boat, the stroke, the line, the tide and the conditions that might produce a PB. This article is a reading of the engineering and sports-science literature on why boats are slow — skin friction, wave-making and the drag budget — and what that implies on the Brisbane River. The companion page Kinematics of Paddling is about the motion of the boat and the paddle — how a Garmin Edge or watch turns that motion into stroke rate, impulse and power.
On a steady piece of water, average paddle thrust equals total drag. Cut drag by 5% and you have bought the same gain as 5% more thrust.
Unless noted, the running example is a 65 kg paddler in a 12 kg Nelo Quattro M K1 (5.2 m) at \(v = 2.78\,\mathrm{m/s}\) (10 km/h, 3:00 / 500 m) in 20 °C fresh water — the same mass and speed as Kinematics of Paddling. Numbers are order-of-magnitude passive-drag estimates from a towed K1, not a race predictor.
Contents
Four kinds of drag
A displacement kayak or surfski moving through water is resisted by four physically different mechanisms:
- Skin friction — viscous shear on the wetted hull.
- Pressure or form drag — the net pressure difference from bow to stern, including any separated wake.
- Wave-making — energy radiated in the Kelvin wake, working against gravity.
- Air drag — the paddler and the dry hull pushing air.
Older naval-architecture writing lumps form drag and waves together as residuary resistance (whatever is left after a flat-plate friction estimate). That is a bookkeeping convenience, not a single physical force. Gomes et al. (2017) keep friction, pressure and wave drag separate, then subtract an air allowance from a towed measurement. This page follows that split.
Real paddling adds yaw, roll and a paddle that both thrusts and drags — see propulsion. The towed numbers below are the still-hull baseline.
Skin friction
Water sticks to the hull. Viscosity then drags neighbouring layers along, forming a boundary layer. The hull has to keep feeding momentum into that layer; by Newton’s second and third laws that rate of momentum change is a drag force on the boat.
Reynolds number
The ratio of inertia to viscosity along the hull is the Reynolds number
where \(v\) is speed, \(x\) is distance from the stem, and \(\nu\) is kinematic viscosity. Fresh water at 20 °C has \(\nu \approx 1.00\times 10^{-6}\,\mathrm{m}^2/\mathrm{s}\) and density \(\rho \approx 998\,\mathrm{kg/m}^3\). (This page uses 20 °C throughout; 15 °C water is about 14% more viscous.)
For a 5.2 m K1 at 2.78 m/s (10 km/h, a 3:00 / 500 m split):
Laminar versus turbulent flow
On a smooth flat plate, transition often begins somewhere between \(Re_x \sim 10^5\) and \(5\times 10^5\), and the layer can be fully turbulent by \(Re_x \sim 3\times 10^6\). Those are plate values, sensitive to free-stream turbulence and roughness. A kayak bow is three-dimensional and has a free surface, so they are a sketch, not a measurement on your ski.
At 10 km/h, \(Re_x = 10^5\) is only about 4 cm from the stem, and \(Re_x = 3\times 10^6\) is about 1.1 m. Over almost the whole boat the boundary layer is turbulent.
The 99% thickness of a Blasius (laminar) layer is \(\delta \approx 4.91\,x / \sqrt{Re_x}\); a fully turbulent 1/7-power layer from the leading edge is \(\delta \approx 0.37\,x / Re_x^{1/5}\). Using the turbulent formula all the way to a 5.2 m stern gives \(\delta\) of order 7 cm. That is the edge of the sheared layer, not “the thickness of water that causes drag.” Wall shear lives much closer to the hull (the momentum thickness is a fraction of \(\delta\)).
Estimating the friction force
A mixed laminar–turbulent plate formula used by Gomes (from swimming papers by Mollendorf et al., 2004, and Pendergast et al., 2006) is
where \(A_S\) is wetted area. A ship would more often use the ITTC-1957 friction line plus a form factor \((1+k)\) for three-dimensional flow. Equation \eqref{eq:skin-friction} is an analogue, and Gomes then fitted multipliers to towing data — so the raw plate number is not their measured \(D_F\).
Static wetted area on a Nelo Quattro M, from a 3D scan and on-water waterlines (Gomes et al., 2017):
| Paddler | Wetted surface area (m²) | Frontal submerged area (m²) |
|---|---|---|
| 65 kg | 1.524 | 0.0253 |
| 75 kg | 1.627 | 0.0284 |
| 85 kg | 1.725 | 0.0315 |
Wetted area is not proportional to weight: 65 → 75 kg is +15% mass and only +7% area. At speed the boat squats and the wave profile changes the waterline, so Table 1 understates the running wetted area.
Pressure (form) drag
Even in a fluid with no free surface, the hull carries a high-pressure region near the bow and, if the stern flow does not recover perfectly, a lower-pressure region aft. The axial resultant of that pressure field is form drag (also called pressure drag). It is shear-free — that is the difference from skin friction — and it is not wavemaking, which is the extra cost of gravity waves on the free surface.
The same quadratic law used for any compact body, and the law Gomes used when they wrote \(D_{PR} \propto A_f v^2\), is
where \(A_f\) is the submerged frontal area from Table 1 and \(C_{PR}\) is a pressure-drag coefficient based on that area. A sphere is around 0.4–0.5 on frontal area; a racing K1 is a very slender displacement body (length / \(\sqrt{A_f} \approx 33\)), so \(C_{PR}\) is small. Fitting \eqref{eq:form-drag} to Gomes’ pressure-drag curve for the 65 kg M K1 — about 7 N at 3.3 m/s and about 19 N at 5.56 m/s in their Figure 3 — gives
Our running example is the same 65 kg paddler at \(v = 2.78\,\mathrm{m/s}\) (10 km/h), 20 °C water, \(A_f = 0.0253\,\mathrm{m}^2\):
That is about one third of the 15 N skin-friction estimate at the same speed. Gomes’ three weight curves for pressure drag lie almost on top of each other between 65 and 85 kg, which is what you expect if \(C_{PR}\) is set by hull shape and \(A_f\) only grows slowly. Narrowing the waterline cuts \(A_f\), but the cockpit still has to fit a pelvis, and stability falls.
A naval-architecture cross-check is the form factor: viscous pressure drag is often written as \(k\,D_F\). Here \(5/15 \approx 0.3\), a plausible \(k\) for a slender hull, and it returns essentially the same 5 N at club speed. We keep the quadratic frontal-area law because it is explicit in \(A_f\) and matches Gomes’ \(v^2\) split.
At 15 km/h the same \(C_{PR}\) gives \(D_{PR} \approx 11\,\mathrm{N}\), still about a fifth of the hydrodynamic total at that faster pace.
Wave-making
The hull is a moving pressure disturbance on the free surface. That costs energy because the waves have gravitational potential (and kinetic) energy that is left behind.
Froude number
The dimensionless speed for gravity waves on a hull of waterline length \(L\) is
Use waterline length, not overall length; 5.2 m is LOA for this K1, so the true \(Fr\) is a little higher than the 0.39 you get if you insert 5.2 m at 10 km/h.
Kelvin wake and wavelength
A moving pressure point on deep water makes a Kelvin pattern: diverging arms in a V, and transverse waves trapped inside it.
In deep water the transverse wavelength of waves that travel with the boat is
At 10 km/h, \(\lambda \approx 4.9\,\mathrm{m}\). That is a plausible spacing for the “hump” you sit on when wash-riding; the real wake of a finite hull is a superposition of bow and stern systems, so the first comfortable spot is not guaranteed to sit exactly one wavelength back.
Humps, hollows, and a common mix-up
A real hull is closer to two opposite disturbances: high pressure at the bow, suction at the stern. Their transverse waves add or cancel depending on speed. Constructive interference radiates a larger wake and therefore more resistance, not less. Resistance “humps” are the expensive speeds; “hollows” are the cheaper ones, sitting on top of a general rise of wave drag with Froude number.
Because bow and stern are opposite in sign, the simple rule “humps when \(L = n\lambda\)” (identical sources) is offset by about half a wavelength for a displacement hull. The practical landmark is that wave drag climbs steeply as \(Fr\) goes through ~0.4–0.5. A shorter boat at the same speed has a higher \(Fr\) and dumps more energy into waves; a longer boat has more wetted area and more skin friction. Racing K1 length is capped by ICF rules at 5.20 m, which is already a compromise.
Gomes fitted wave drag as proportional to \(v^4\). Their 65 kg curve is 7.05 N at 5.56 m/s, so
for that paddler and hull \eqref{eq:wave-drag}. At 2.78 m/s this is \(D_W \approx 0.4\,\mathrm{N}\). Gomes’ Figure 3 (our Figure 6) reads under a newton at the same club pace, so the power law is in the right place. Wave drag is the component that grows fastest with speed and with extra displacement: at 15 km/h it is still the smallest of the three hydrodynamic pieces for a 65 kg paddler (~5% of hydro), but it nearly triples from 65 to 85 kg in Gomes’ higher-speed data.
Air resistance
Air drag on the paddler and the dry hull is the same quadratic law as form drag, but in air:
with \(\rho_\mathrm{air} \approx 1.2\,\mathrm{kg/m}^3\). A crouched paddler plus coaming is a bluff body; taking an effective drag area \(C_D A \approx 0.45\,\mathrm{m}^2\) at our 2.78 m/s example (still air, so \(v_\mathrm{rel}=v\)) gives \(D_A \approx 2\,\mathrm{N}\). That is about 9% of the ~23 N passive total below, which is Jackson’s still-air fraction and the allowance Gomes subtract from a towed measurement. It is not an extra 10% on top of the water components.
A headwind raises \(v_\mathrm{rel}\); a tailwind can cancel it. On a windy river the air term is not a fixed 3 N.
Putting the four together
The same 65 kg paddler, 5.2 m K1, 20 °C water, \(v = 2.78\,\mathrm{m/s}\) (10 km/h, 3:00 / 500 m):
| Component | Formula | Force | Share |
|---|---|---|---|
| Skin friction \(D_F\) | Plate formula, \(A_S = 1.524\,\mathrm{m}^2\) | 15.4 N | 67% |
| Form drag \(D_{PR}\) | \(\tfrac12\rho v^2 A_f C_{PR}\), \(C_{PR}=0.05\) | 4.9 N | 21% |
| Wave drag \(D_W\) | \(K_W v^4\) from Gomes 65 kg | 0.4 N | 2% |
| Air drag \(D_A\) | \(\tfrac12\rho_\mathrm{air} v^2 (C_D A)\), \(C_D A=0.45\,\mathrm{m}^2\) | 2.1 N | 9% |
| Total \(D_T\) | \(D_F + D_{PR} + D_W + D_A\) | 22.8 N | 100% |
Club cruise · 10 km/h · 65 kg
The four formulae above, not a remainder from the towed total.
Faster piece · 15 km/h · 65 kg
Same coefficients: form and air follow \(v^2\), waves follow \(v^4\), skin friction a little slower than \(v^2\).
skin friction form waves air
Checking the total with a coast-down
If you stop paddling and sit still, thrust drops out and the boat’s deceleration is the drag:
so \(D_T = m_e |a|\) from \eqref{eq:coast-down}. Here \(m_e = (m_{\mathrm{p}}+m_{\mathrm{b}})(1+\alpha)\) with \(\alpha \approx 0.033\) from the sprint-kayak ellipsoid model in Prétot et al. (2022); 65 kg + 12 kg is then about 79.5 kg. At 10 km/h the 23 N passive budget is a deceleration of about 0.29 m/s². That is a phone-accelerometer or GPS speed-decay measurement: same method as their pure-deceleration trials.
Strategies
- Length is a tradeoff, not a free lunch. Shorter cuts wetted area (friction) but raises \(Fr\) (waves). Below about 5 m a displacement hull at club speeds is already a noisy wavemaker. ICF sprint K1s are maxed at 5.20 m for a reason.
- Pick the hull for the paddler. Table 1 is not linear in weight, and Gomes (2015) found that the lowest-drag size can depend on target speed as well as mass. Dropping 10% body mass does not drop 10% friction or 6% total drag.
- Keep the hull smooth and clean. The flow is already turbulent. Smoothness reduces sand-grain roughness drag. Nicks, tape edges and a fuzzy river-slime layer all count.
- Dimples are a golf-ball trick, not a kayak trick. Dimples trip a sphere’s boundary layer so it separates later and the wake shrinks. A slender displacement hull does not carry a golf-ball wake; extra roughness would mostly buy extra friction.
- Sit the boat flat. Constant edging and poor trim add wetted area and yaw. That is free drag.
- Warm water is slightly faster. Friction coefficient scales roughly as \(\nu^{0.2}\). About 5 °C warmer cuts \(D_F\) by ~2.5–3%, and total drag by a smaller fraction (~1.5–2%) if friction is ~60% of the total. It is real, not a race-winning secret.
- Wind is not 10% on a breezy morning. Air drag follows relative wind squared. A 15 km/h headwind on a 10 km/h boat is a 25 km/h air speed.
- Wash-ride the ~5 m hump. The wavelength formula is the club-useful result: the transverse wave of a slightly faster boat (~5 m at club pace) is a free downhill if you can sit on it without steering yourself into their stern.
- Take the deeper line on a falling tide. At 10 km/h, extra wavemaking starts below about 3.1 m and the critical depth is 0.8 m. At 15 km/h a 2 m channel is already near critical. Shorter on the map is not faster if it sits you on \(Fr_h \approx 1\).
- Boundary-layer additives (polyethylene oxide, guar) can cut pipe friction in the lab. They wash off, and ICF surface rules do not look kindly on them. Not a training tool.
Shallow water
This is the section that actually matters on the Brisbane River. The physics is the same as for any slender displacement hull — racing K1s, canoes and rowing shells — and it is organised by one number: the depth Froude number. Day, Campbell, Clelland, Doctors and Cichowicz (2011) tank-tested that family of hulls in shallow water; Tuck (1966) is the slender-body theory for squat; Faltinsen (2005) is the wave-pattern account they point to.
Deep-water wave speed depends on wavelength. In finite depth \(h\) the dispersion relation is
When \(h/\lambda\) is large (deep), \(\tanh \approx 1\) and you recover \(\lambda = 2\pi v^2/g\). When \(h/\lambda\) is small, \(c \approx \sqrt{gh}\), independent of wavelength: there is a maximum long-wave speed set by depth alone.
The hull’s depth Froude number is
Day’s map for \eqref{eq:depth-froude}, which is the one that applies to a K1, is:
- \(Fr_h \le 0.5\): wave-making still close to deep water.
- As \(Fr_h\) approaches 1: wavelengths, wave heights and wave resistance all increase. The boat sits in a growing transverse wave.
- \(Fr_h = 1\): critical. Long waves cannot outrun the hull. This is the expensive neighbourhood.
- \(Fr_h \gt 1\): supercritical. The transverse system disappears and wave resistance can fall relative to the critical peak; the remaining wash is diverging.
Critical flow is \(h_c = v^2/g\). The deep-water edge of the map is \(Fr_h = 0.5\), i.e. \(h = 4v^2/g\). For the running example (10 km/h, \(v = 2.78\,\mathrm{m/s}\)) those depths are 0.79 m and 3.1 m.
What that does to our K1
Two separate mechanisms, both from slender-body hydrodynamics.
Wave-making. Table 2’s deep-water \(D_W\) is only 0.4 N at 10 km/h — 2% of the 23 N total. Shallow water acts on that term. As \(Fr_h\) climbs, the same hull radiates a larger wake (Day; Faltinsen §4.5). Skin friction and form drag, ~88% of the deep-water budget, are almost insensitive to a 2–4 m river: the K1 draught is ~0.1 m, so \(h/T\) is of order 20, and the blockage \(A_f/(Wh)\) in a wide channel is negligible. The extra force you feel on a falling tide is extra waves, sitting on top of the 23 N deep-water total — not a rescaling of the whole pie.
Squat. Continuity speeds the flow under the hull; Bernoulli drops the pressure; the boat settles. Tuck’s open-water slender-body result, in the form Gourlay uses, is
with \(C_s \approx 1.5\) a conservative shape factor (Gourlay 2011, after Tuck). For 65 kg + 12 kg on 5.2 m, \(\nabla/L^2 \approx 2.9\,\mathrm{mm}\). The K1 is so slender that squat is millimetres until you get close to critical:
| Depth \(h\) | \(Fr_h\) | Day / Faltinsen regime | Tuck squat \(s\) |
|---|---|---|---|
| 5.0 m | 0.40 | still near deep-water wavemaking | 1 mm |
| 3.1 m | 0.50 | onset of finite-depth waves | 1 mm |
| 2.0 m | 0.63 | extra wavemaking growing | 3 mm |
| 1.4 m | 0.75 | intermediate: larger wake (typical Brisbane River) | 6 mm |
| 1.0 m | 0.89 | approaching critical | 16 mm |
| 0.79 m | 1.00 | critical: max transverse wave | — |
| 0.5 m | 1.25 | supercritical: transverse waves gone | trim, not sinkage |
The faster-piece line is the surprise. At 15 km/h, \(h_c = 1.77\,\mathrm{m}\) and the \(Fr_h = 0.5\) edge is 7.1 m — so anything under about 7 m is already in the finite-depth regime, and a 2 m channel sits near critical. Day et al. make the same point for a rowing lake: 3.0 m of water has a critical speed of 5.4 m/s, which elite crews reach inside a stroke. That is why Scragg and Nelson (1993) designed different eight-oar hulls for deep and shallow racing — the optimum shape moves once \(Fr_h\) is no longer small.
Paddling is also unsteady. Day’s oscillating-speed tests on slender hulls in shallow water found a mean wave resistance above the steady towing value, and at stroke-like frequencies the viscous resistance rose as well. Keep the boat tracking; a yawing, surging ski in 2 m of water is paying twice.
Propulsion and active drag
Hull drag is only half of the speed equation. The paddle has to make a wake of its own. Jackson (1995) modelled the blade as producing vortex-ring momentum, then equated mean thrust to hull drag and asked how hull speed changes if you tweak length, weight, blade area or stroke. That sensitivity study is the reason the paper is famous — not the 10% air figure.
Propulsive efficiency is less than one: some of the work you put into the shaft goes into moving water backward rather than moving you forward. A cleaner catch and a blade that does not slip as much raise efficiency; thrashing raises the paddle’s own drag. How a Garmin accelerometer turns each stroke into impulse and power is the subject of Kinematics of Paddling.
Figures 5 and 6, Table 2 and the drag formulae above are passive tow numbers: the boat is pulled with no paddling. Gomes et al. say so explicitly — towing “eliminated changes in the wetted surface area that would occur if there were accelerations and decelerations from the intermittent application of propulsion during actual paddling.” On the water each stroke yaws, rolls and surges the hull, immersion changes, and the paddle itself both thrusts and drags. Day et al. (2011) showed that oscillating speed, especially in shallow water, raises the drag of slender racing hulls above the steady towing value. The resistance that mean paddle thrust has to overcome is therefore active drag, and it is larger than the still-hull curves in Figures 5 and 6. Keeping the boat tracking — less rudder, less correction, less edge — is a drag intervention as much as a style point.
In swimming hydrodynamics, active drag is routinely written as a factor on the same quadratic drag law used for passive drag — enlarging the effective frontal area (or an equivalent multiplier) while keeping the \(v^2\) form. Gatta, Cortesi, Fantozzi and Zamparo (2015) estimate active drag as \(D_a = \tfrac12\rho\,C_D A_p v^2\) with stroke-averaged planimetric area \(A_p\), and find front-crawl active drag about 1.5× the streamlined / passive baseline; a later synthesis puts recent active/passive ratios in the band 1.5–2.5 (Zamparo et al., 2020, citing Formosa, Gatta and Narita). We borrow that bookkeeping for the kayak hull split: apply one active-drag factor \(C_{\mathrm{act}}\) to each hydrodynamic term, leave still-air drag alone, and (below) allow a separate boat-geometry factor \(C_{\mathrm{geom}}\),
That is a desk transfer from passive tow → paddling, not a new tank measurement. With the running example’s passive pieces \(D_F=15.4\,\mathrm{N}\), \(D_{PR}=4.9\,\mathrm{N}\), \(D_W=0.4\,\mathrm{N}\), \(D_A=2.1\,\mathrm{N}\) (\(D_T=22.8\,\mathrm{N}\)), Gatta’s crawl ratio \(C_{\mathrm{act}}=1.5\) and \(C_{\mathrm{geom}}=1\),
One more caveat: Gomes’ formulae and Figures 5–6 are for a Nelo Quattro M K1 — a very low-drag racing hull. A touring, recreational or otherwise beamier boat will sit higher even at the same speed and mass. Try the numbers therefore includes a Boat geometry factor \(C_{\mathrm{geom}}\) (default 1 for the Quattro M) that multiplies the same hydrodynamic terms, so you can scale the desk estimate for a different hull without changing the Gomes baseline.
Try the numbers
Change length, speed, paddler mass, boat mass, temperature, optional depth, the active-drag factor and boat geometry. The base numbers are for a Nelo Quattro M K1 — the hull Gomes towed and the areas in Table 1. With no depth, the four drag terms are the Table 2 deep-water formulae. Enter a depth and Tuck squat raises wetted and frontal area (so \(D_F\) and \(D_{PR}\)), and a finite-depth factor amplifies \(D_W\) above Day’s \(Fr_h = 0.5\) onset — both feed \(D_T\). \(C_{\mathrm{act}}\) and \(C_{\mathrm{geom}}\) then scale skin, form and wave drag per \eqref{eq:active-drag} (air unchanged): default \(C_{\mathrm{act}}=1.5\) (Gatta crawl), \(C_{\mathrm{geom}}=1\) (Quattro M). Set \(C_{\mathrm{act}}=1\) for a passive-tow comparison; raise \(C_{\mathrm{geom}}\) for a higher-drag boat than the Quattro M. Hull power \(D_T' v\) uses the corrected total. Coast-down \(|a| \approx D_T'/m_e\) is shown for the same total — for a pure glide comparison set \(C_{\mathrm{act}}=1\). Same 65 kg and 10 km/h defaults as the kinematics calculator. Neither is a race predictor.
Fitted to a Nelo Quattro M K1 (Gomes et al., 2017). Same coefficients as Table 2: \(C_{PR}=0.05\), \(C_D A=0.45\,\mathrm{m}^2\), \(K_W\) interpolated from Gomes’ 65–85 kg wave-drag points. Skin, form and wave are multiplied by \(C_{\mathrm{act}} C_{\mathrm{geom}}\) (defaults 1.5 and 1); air is unchanged. Coast-down uses \(m_e = (m_{\mathrm{p}}+m_{\mathrm{b}})(1+0.033)\). Forces are fixed-decimal (newtons, matching Table 2); \(m_e\) uses three significant figures like the kinematics desk tool. Depth corrections apply only above Day’s deep-water edge \(Fr_h = 0.5\) (shallower than \(4v^2/g\)); at or below that edge the drag matches infinite depth. Above it: squat \(s\) (Tuck/Gourlay, \(C_s=1.5\)) adds wall-sided immersion \(\Delta A_S \approx 2Ls\) and a matching \(\Delta A_f\) from the Gomes mass table; wave drag is multiplied by \((1-0.5^2)/(1-Fr_h^2)\) (desk estimate, soft-capped at \(Fr_h=0.95\)), then decays as \(1/Fr_h^2\) past the peak. Clear the depth field (or enter a deep value) to see the deep-water drag without a shallow correction.
References
- Gomes, B. B., Machado, L., Ramos, N. V., Conceição, F. A. V., Sanders, R. H., Vaz, M. A. P., Vilas-Boas, J. P., & Pendergast, D. R. (2017). Effect of wetted surface area on friction, pressure, wave and total drag of a kayak. Sports Biomechanics, 17(4), 453–461. https://doi.org/10.1080/14763141.2017.1357748. (Towed K1 drag split into friction, pressure and wave drag; Table 1 areas and Figures 5–6; passive-tow caveat.)
- Gatta, G., Cortesi, M., Fantozzi, S., & Zamparo, P. (2015). Planimetric frontal area in the four swimming strokes: Implications for drag, energetics and speed. Human Movement Science, 39, 41–54. https://doi.org/10.1016/j.humov.2014.06.010. (Active drag as \(D_a=\tfrac12\rho C_D A_p v^2\) with stroke-averaged area; ~1.5× passive in front crawl.)
- Zamparo, P., Cortesi, M., & Gatta, G. (2020). The energy cost of swimming and its determinants. European Journal of Applied Physiology, 120(1), 41–66. https://doi.org/10.1007/s00421-019-04270-y. (Review: recent active/passive drag ratios about 1.5–2.5.)
- Gomes, B. B., Conceição, F. A. V., Pendergast, D. R., Sanders, R. H., Vaz, M. A., & Vilas-Boas, J. P. (2015). Is passive drag dependent on the interaction of kayak design and paddler weight in flat-water kayaking? Sports Biomechanics, 14(4), 394–403. https://doi.org/10.1080/14763141.2015.1090475. (Hull size versus paddler mass and target speed.)
- Jackson, P. S. (1995). Performance prediction for Olympic kayaks. Journal of Sports Sciences, 13(3), 239–245. https://doi.org/10.1080/02640419508732233. (Still-air fraction and blade / hull sensitivity framing.)
- Prétot, C., Carmigniani, R., Hasbroucq, L., Labbé, R., Boucher, J.-P., & Clanet, C. (2022). On the physics of kayaking. Applied Sciences, 12(18), 8925. https://doi.org/10.3390/app12188925. (Coast-down method; longitudinal added-mass coefficient \(\alpha \approx 0.033\).)
- Day, A., Campbell, I., Clelland, D., Doctors, L. J., & Cichowicz, J. (2011). Realistic evaluation of hull performance for rowing shells, canoes, and kayaks in unsteady flow. Journal of Sports Sciences, 29(10), 1059–1069. https://doi.org/10.1080/02640414.2011.576691. (Depth-Froude regimes and unsteady shallow-water resistance.)
- Tuck, E. O. (1966). Shallow-water flows past slender bodies. Journal of Fluid Mechanics, 26(1), 81–95. https://doi.org/10.1017/S0022112066001101. (Slender-body squat theory.)
- Gourlay, T. P. (2011). A brief history of mathematical ship-squat prediction, focussing on the contributions of E. O. Tuck. Journal of Engineering Mathematics, 70(1–3), 5–16. https://doi.org/10.1007/s10665-010-9435-3. (Squat form used here, with \(C_s \approx 1.5\).)
- Faltinsen, O. M. (2005). Hydrodynamics of High-Speed Marine Vehicles. Cambridge University Press. Chapter 4.5, ships in shallow water. (Wave-pattern account behind Day’s map.)
- Scragg, C. A., & Nelson, B. D. (1993). The design of an eight-oared rowing shell. Marine Technology, 30(2), 84–99. https://doi.org/10.5957/mt1.1993.30.2.84. (Deep versus shallow hull optima once \(Fr_h\) is no longer small.)
- Mantha, V. R., Silva, A. J., Marinho, D. A., & Rouboa, A. I. (2013). Numerical simulation of two-phase flow around flatwater competition kayak design-evolution models. Journal of Applied Biomechanics, 29(3), 270–278. https://doi.org/10.1123/jab.29.3.270. (CFD context for racing-kayak free-surface flow.)
- Mollendorf, J. C., Termin, A. C., II, Oppenhein, E., & Pendergast, D. R. (2004). Effect of swim suit design on passive drag. Medicine & Science in Sports & Exercise, 36(6), 1029–1035. (Source of the mixed laminar–turbulent plate friction line Gomes adapts.)
- Pendergast, D. R., Mollendorf, J. C., Cuviello, R., & Termin, A. C., II. (2006). Application of theoretical principles to swimsuit drag reduction. Sports Engineering, 9, 65–76. (Companion swimming paper for the same friction line.)
- Li, D., Yang, Q., Zhai, L., Wang, Z., & He, C. (2021). Numerical investigation on the wave interferences of submerged bodies operating near the free surface. International Journal of Naval Architecture and Ocean Engineering, 13, 65–74. https://doi.org/10.1016/j.ijnaoe.2021.01.002. (Kelvin-wake diagram in Figure 4.)
- Çengel, Y. A., & Cimbala, J. M. Fluid Mechanics: Fundamentals and Applications. (Boundary-layer transition diagram in Figure 2.)
- International Canoe Federation. Racing rules (hull length, minimum mass, surface). https://www.canoeicf.com/rules. (ICF K1 length cap and surface rules.)
- Spearers (2026). Kinematics of Paddling — companion article on stroke motion (accelerometer surge, impulse, power and left/right) and stroke calculator.