The Propeller Drawing: Three Views of One Surface

Three engineers measure the skew of one propeller off one drawing and return three different answers, and none of them makes an arithmetic mistake. The commonest failure is to read the correct view and forget what its horizontal axis actually is. On the reference propeller of this specialisation, the 7.00 m four-bladed unit fitted to our reference Panamax bulker, that one omission over-states the local skew angle at 0.3R by about a quarter.

A propeller plan is not a picture of a propeller. It is three geometric constructions of one surface, each discarding different information on purpose, plus a data block carrying what no view can show.

The surface that comes before the blade

Everything on the plan describes one object: a helicoidal reference surface, swept out by a family of helical lines, one at each radius from hub to tip. The real blade is that surface perturbed, a two-dimensional hydrofoil section dressed onto it at every radius to give thickness and camber. A surface and a stack of foils: every line on the drawing belongs to one or the other.

The helical line at radius r has a pitch P, the axial distance a point on it advances in one revolution, and a pitch angle phi, the angle it makes with the plane of rotation:

tan(phi) = P / (2 pi r)

where P is the pitch in metres, r the radius in metres and phi the pitch angle. Lesson 2 takes this apart properly, because P is not one number and there are three conventions for measuring it. What matters here is that phi falls as r grows: at the nameplate pitch ratio of 0.715, taken for now as if it held all the way in, the ship's blade stands at 37.2 degrees to the plane of rotation at 0.3R and at 18.0 degrees at 0.7R. A blade is twisted because the geometry forces it, not because a designer preferred the look.

Two lines then organise the surface. The generator line is where that surface cuts the plane containing the shaft axis, and on series propellers and most conventional merchant designs it is straight. The blade reference line is the locus of the mid-chord points of the sections. It is a curve in space, and its two projections are the quantities everyone argues about: onto the plane containing the shaft axis it becomes the rake line, onto the plane of the propeller disc the skew line. Rake and skew are one curve seen twice.

A common coordinate system is right-handed, with x along the shaft axis positive towards the bow, y horizontal and z vertical positive downwards. The origin sits where the shaft axis meets the generator line, which sets rake and skew to zero at the root section by construction, so every quoted value is relative to the root. Some offices use the forward face of the hub instead; the one invariant is that x is the shaft axis, so check signs before comparing drawings from two houses.

Three views, three quantities

The transverse view looks along the shaft and shows the blade silhouette on the propeller disc. It carries the projected outline, the skew line, and the total skew angle, which is the maximum of the local skew angle over the radius minus the minimum. That definition earns its keep: a balanced-skew blade swings negative before it swings positive, so the tip value on its own understates it. Lesson 4 works skew properly; here it is simply the quantity this view owns.

The profile view is the elevation on the plane containing the shaft axis. It carries the generator line, the total rake line, and the thickness fairing from which maximum section thickness is read at every radius. Many plans also draw the section shapes across the expanded outline, so a thickness can be read there station by station; the fairing, one curve through every station, is the profile view's. It also shows what the transverse view cannot: rake and skew are not independent. Displacing a section along its own helix by a local skew angle theta_s moves it axially too, by r theta_s tan(phi), so a "zero rake" skewed propeller almost always means zero generator-line rake rather than zero total rake.

The expanded view is where readers go wrong, because it is not a projection. Cut the blade with a cylinder of radius r and unroll that cylinder flat; a cylinder is developable, so nothing stretches or tears. The helical line becomes a straight line of slope phi, and the segment between leading and trailing edge becomes the chord c(r). The diagram stacks every section, each rotated by its own pitch angle so its chord lies horizontal, about its own point on the generator line. Its horizontal lengths are therefore true helical arc lengths, not distances in the propeller disc. It carries c(r) and the offset of each section's mid-chord point from the generator line, measured along its own chord, and the skew angle follows:

theta_s(r) = Delta(r) cos(phi(r)) / r

where Delta is the offset in metres, phi the pitch angle at that radius, r the radius in metres and theta_s the local skew angle in radians. The cosine is the whole point: a displacement along a helix of angle phi has a circumferential component of only Delta cos(phi). Some offices dimension the offset from the leading edge instead, in which case subtract half the chord first.

Distributions, not numbers

Every quantity above is a function of radius, and the drawing tabulates it at standard stations, r/R from 0.2 to 0.9 in steps of 0.1, then 0.95 and often 0.975 and 1.0. The stations crowd near the tip on purpose, because the chord falls to zero there with a steep slope and evenly spaced numerical integration cannot follow that cusp: on 0.1 steps the under-read runs from about one per cent to over four, depending on the rule and on how full the outline is near the tip, and that is exactly the size of error that quietly corrupts a reconstructed geometry. The 0.2R start is not arbitrary either. In the Wageningen B-series the hub is 0.167 of the diameter, and the band out to 0.2 R is the fillet, the transition from boss to the first useful section.

Four numbers are enough to pick a propeller off a systematic series at the preliminary stage: diameter, a nominal pitch, blade number, and the expanded area ratio that lesson 5 defines properly. Everything else on the plan is what separates a preliminary propeller from a good one. A specification reading 7.00 m, four blades, pitch ratio 0.715 at 0.7R and area ratio 0.55 tells you enough to look up an open-water curve and nothing about whether the blade will cavitate, vibrate or crack.

When the plan is incomplete

A twenty-year-old ship, a scanned drawing whose pitch table is illegible, and an owner who now wants a study of an energy-saving device. Recovering the distributions by measurement off the sheet is ordinary work, in a fixed order. Set the scale from two independent known lengths rather than one, because scans stretch more in one direction than the other. Digitise the expanded outline at the standard stations for c(r). Measure the offset along each expanded chord and convert it with the cosine relation. Read the rake ordinates off the profile view and subtract the skew-induced part r theta_s tan(phi); if what remains is a straight line through the origin, you have recovered the rake angle. Read maximum thickness off the fairing. Then check globally, against any area ratio stamped on the sheet and against the stamped weight, using about 7,600 kg per cubic metre for nickel-aluminium bronze.

Take the offset step at 0.3R on our Panamax's propeller, where r is 1.05 m and, still taking the nameplate pitch as if it held there, the pitch angle is 37.2 degrees, so cos(phi) is 0.797. Her skew line swings towards the leading edge at the inner radii, and the offset measured there is 115 mm on the leading-edge side of the generator line: theta_s = 0.115 x 0.797 / 1.05 = 0.0872 rad, which is 5.0 degrees. Divide 115 mm by 1.05 m and stop, and the answer is 6.3 degrees. That is an error of 1.3 degrees from one missing cosine, a quarter of the angle itself, and it grows further inboard, where phi is larger still.

Monday morning

Pull the propeller plan for one ship you look after and check three things while nothing depends on the answer: that all three views are on the sheet, that the data block states the radius and the convention at which pitch is quoted, and that the direction of rotation is written somewhere. Then look at the pitch entry. If it is a single number rather than a table, you have a nominal value and not a distribution, and the next lesson is about exactly why that distinction costs money.