The 3D chine lines were planated approximately. Each 3D chine line was projected to a plane defined by three 3D points of the chine. CAD-Functions: define a plane by 3 points, change temporarily to this plane as XY-coordinates, draw a 2D polyline along the 3D chine. The difference in length was less than 3 mm (.1"), due to the only slight curvature of the chines in Z. These polylines represent (very nearly) the axis of the pipes for the chines. Parallel drawn lines in distance of the pipes' radius are shown on the plotted drawing below, the stern-pipes kneeled up, the bow's pipes down.

This drawing on the work desk is nailed down at the inner edges of the pipes, to control bended pipes. The positions are polyline points, defined at design stage in a distance of about 25 cm (10").

The curvature of a pipe changes steadily. Simply bending by a constant radius along the whole length of the pipe would not achieve the goal. So different radii have to be used for certain sections, approximating a smooth curve by basket arches queued along the desired length. So a bending machine had to be constructed, which allows a partial change of bending radii.

Ballbearing A mounted on eccentric tappet, positioned by lever D, which can be fixed by a tappet snapped into holes along strip E. Also the two counter bearings B were mounted on eccentric tappets, for rough radius adjustment in steps by positioning in 3, 6,9 and 12 o'clock by the lever F and to increase/decrease the radius choice, then screwed down. The triangle C is only used temporarily laid over A and the two Bs, to support against bending pressure. All parts are mounted on a steel plate of 8 mm (1/3").

Consequently along the different hole positions of D in E and lever positions in the Bs, test bendings were rendered to get the resulting radius for each adjustment. The reached radius is derived from the arch rise by a certain ruler length of about 70 cm (30") hold to the bay.

Bending is performed by pulling and pushing the pipes "through" the bearings. The lowest radius needed to be bent was 45 cm (18"), maybe a smaller could be reached. Back bending in case of mis-bending could be performed by turning the pipe, within limited dimensions.

The first attempt to bend pipes by visual judgment, controlled along the nails on the plots intermediatly, was not satisfactory and time consuming. (Bending, pipe removal from the machine, laying down ...). So the bending radii were marked up on the plot in each nail position. The length along the nails measured against a certain pipe end for the income and outgoing of the section to be bent in a certain bending radius. These two lengths very marked on the pipe and the pipe was pushed and pulled within these two marks.

This method was satisfactory for the chine pipes at the bow, due to their low curvature and low change in radius (i.e. long sections). For the stern chines it did not work, due to short ways between radius-change to estimate the bending radii between nail stations on the plot. For that reason the method had to be improved, by dimensioning the drawing in length along the chine line and the corresponding bending radius.

In the bending machine, a certain radius was represented by certain lever positions of the eccentric bearings. The following image maybe a little confusing, but ...

The red spline represents a smoothed function of the queued bending radii for one pipe (scaling in X an Y adopted). It begins with a steep slope down at the straight entry of the pipe (bending section), levels more and more and raises again steeply against the end of the pipe (bending section). Huge Radius means low bend, small radius large bending. (what's not so nice is the bay in the curve left, but for people with high attitude for smooth functions only...)

The blue parallels represent the discrete radii, reached by certain adjustments of the levers at the bending machine. The section between the intersection on both sides with the red line, will give the length to push/pull the pipe through the machine by this radius. Of course this section nearly to the whole pipe in length is a long one with big bending radius (so nearly flat). Toward the pinnacle of the pipe's bending the section gets shorter but with small bending radius.

This all ended up in a spreadsheet list, with the position for the counter bearings B, shown as triangles for clockwise position of levers F, the hole position number for along strip E for lever D and the length along the pipe in centimeters. The lengths were marked upon the pipe. Bending then will be rendered by just to adjust the levers read from the printed list and to push and pull the pipe through the machine.

The marked positions of the levers on the (here) rod can be recognized in the image. The actual bending section is marked by two small tapes, for convenience at the pushing and pulling. A holder at the end of the pipe was mounted, to keep the pipe in flat position to avoid distortion.

Bending now was performed by one push and one pull for the individual machine positions (each radius) along each marked section. For the whole spectrum of radii needed for one pipe, this procedure had to be repeated 15-20 times along a length of 3 m (9'). This method was so exact, that we could ignore control on the plot for each pipe, this was performed at the end of the day of course. The deviations were lower than +/- 1 cm (.4")compared to the nails on the plot and within each pair of pipes.

Although the power needed for push and pull was not very high, teamwork was essential. The second hand will keep the pipe from distorting by holding it flat and as we bent 3 m of 6 m (20') pipes, it would not have been possible to work alone. The machine had also to be mounted rotary on the table, because of the small radii reaching to the end of each procedure and the raising angle between machine and table.

Forty minutes were consumed for one pipe, including preparation of the numbers, but we felt it was a nice job (in spite of this "academic" approach) but with accurate results.

For each alloy type and differing dimensions of pipes/rods/extrusions, bending must be tested along the machine's adjustments to gain the lever positions for each radius. E.g. the 5083 rod did not need such an overbending and lower push/pull power than the 6060 pipes. So the "stations" of the levers were different for the same radius. Dimensions of both were 30 mm (1,1")

Overbending was not explicitly considered here, it is already "included," when figuring out the radii by test bending with machine adjustments.

The slight curvature in Z-direction along the hull's lines, we will reach by inserting the bended pipes into the prepared notches at the frames.

T-Frames for the roof we will bent the same way, but are a little bit easier, because of constant radius along the whole breadth.

The results :

 18 of 24 bent chine pipes (6m) and chine rods (3m)