Vous êtes sur la page 1sur 17

01

Appendix 3.2.1: Drawing Process: Adapting Two-Dimensional Drawings


to Three-Dimensional Plots

Ma
y2

To t.p.a were made available scans of hand-drawn sections of C1-C16, and two sets of
two-dimensional (2D) plots of all frames, i.e., the single frame sections with records of the
various measurements taken (a number of which are found in Fig.3.2.i-1) and a combined
plot with outline measurements only (Fig.3.2.i-2). Judged by the measurements obtainable
from these drawings, Daniel Visnikar, the diver in charge of taking the measurements, in
most cases had drawn his sections and views starting at starboard and subsequently mirrored a technique supplied with the plotting program the necessary elements to portside.
In his drawings he also had assumed that the midship bottom planking had opened up until lying flat on the seabed, and accordingly not noted any initial deadrise of the lower hull.
Aware of the reported list of the hull to starboard but left with no practicable alternative, I
have followed Visnikars plots along the inside measurements of the hull.

Lie
b

ne
r

As the various section drawings were produced in several steps over seven months and
Visnikar in handling the plotting program, Autocad 2004, adopted an approach of learningby-doing, measurements proved not too consistent, and eventually necessitated the various
adjustments of the original plots described below. In his computer plots Visnikar had used
units of metres, thus, e.g., noting 1 m 23 cm as 1.23 units, and not displayed any further digits.
As he had drawn most of his lines by hand without making use of defined line-lengths, on
closer examination the actual drawing objects proved to have any measurement falling into
figures still rounding onto the dimensions he had initially intended, thus in many a case producing measurements that differed by up to 0.0049 units (= 0.49 cm) on a digit display setting
of four digits behind Zero (see, e.g., Fig.3.2.ii-1). As three-dimensional (3D) drawings have to
rely on unambiguously defined planes in 3D space, these ostensibly negligible differences
in many a case obstructed the 3D drafting process. With the aim to increase precision and
ease the plotting process, I used cm and a display setting of four digits behind Zero as the
programs basic drawing unit and made where possible use of pre-set line lengths, snaps

H.

and various other options to generate more precisely fitting lines and sections. However,
attempts at changing all of Visnikars various measurements into consistent and matching
values eventually proved to be fairly futile.

rst

Due to its improved three-dimensional capabilities, I combined Visnikars section plots


into a 3D Autocad 2010/12 file following the spacing of the respective frame stations. The
resulting model then was faired, i.e., cleaned of obvious inconsistencies in the geometry1;
details of this process will be described below. The resulting section profiles were used to
draft each of the recorded constructional members as a single 3D object. These objects

Ho

were generated by lofting, a command that uses cross section profiles as delimiters of a
1

For a discussion of the virtual fairing process see, e.g., http://www.boatbuilding.com/article.php/ DirtySecretsHullDesign.

01
4

virtually solid body2, along the cleaned section measurements. As the lofting function
could not always produce the desired geometry, some objects were constructed along seg-

Ma
y2

ments of the respective profiles and then combined into single objects using the programs
union command. It proved rather impractical to generate splined (curved) 3D models
based on the combined 2D drawings by Daniel Visnikar, and I in most cases decided to reproduce the segments between the cross section profiles as bound by straight lines (in the
program, ruled against smooth fit lofting method). To as far as possible imitate the actual construction process of a ship and its elements, distinct features of the objects (as, e.g., the
holes in tambuku-lugs or the scarfs on plank butts) then were cut into these raw solids
by subtracting other solids representing the respective features. As this process would
produce a problematically complex and overlong file, any finished object was transformed
into an external reference, i.e., saved as an own file containing a record of its placement in
the overall model. Most of the resulting plots illustrate rather technical matter; I hence refrained from applying computer-intensive operations like, e.g., rendering of objects, textured
surfaces or intricate shadowing and lighting3, but instead used those of the programs in-built

ne
r

colours, light settings and visual styles that seemed most appropriate for any particular question at hand.

Lie
b

3D modelling followed the universal procedure of assembling a planked hull, i.e., a virtual laying of the keel; I thus initially evened out measurement values of the keel sections, onto which the succeeding measurements of planks and frames were mapped. I assumed the initial drawing plane an essential requirement for 3D drafting to be parallel to
the keel sections between frames C6-21. Though the distances between the tambuku-lugs
on a number of surfaced plank samples do not entirely concur with Visnikars records of the
distances between the various frame sections, I in lieu of an alternative followed his measurements: conditions on the underwater site evidently were not too supportive for scrupulously precise gauging, and, as far as they can still be reconstructed, observable differences do
not exceed 5-7 cm (see, e.g., sample S573).

rst

H.

Visnikars drawings indicate that the keels top surface broadens noticeably around the
hulls centre: on his overall 2D drawing (Fig.3.2.ii-3), the keel extends to 50 cm between C7C19; in the section drawings, C8-C20 display measurements 50 cm. However, as shown in
Fig.3.2.ii-4, the measurements reported for the various individual keel sections are not consistent and hence do not produce the smooth and realistic curves one would expect. As
access to several sections of the hull was obstructed by concretions, the irregular measurements reported for e.g. C17 and C8-9 very probably are unreliable the 50[.08] and 51[.45] cm
reported for top-surface width on C13 and C14, though, should be dependable enough.
With the aim to reproduce the rather shallow lengthwise curvature of the keel Visnikar
2

Ho

This process is described in some detail in Lin (2003: 13f); for an alternative approach, i.e., combining virtual
surfaces into solids, see Nayling and Jones (2009: 322ff).
3

For a concise overview of the various possibilities in 3D modelling cf., e.g., Wells 2008: 17-25.

01
4

seems to have observed in the necessarily straight 3D drawing, I accordingly fixed the widest top-surface sections onto 51 cm around C14-16, thus close and somewhat aft to the pro-

Ma
y2

posed constructional centre point of the hull. To evenly decrease keel width to fore-ship to
the 20 cm measured at the bow piece, I chose two steps, to bow fixed on C11 and C6, to
stern on C19 and C21, assuming a somewhat beamier aft of the hull: as explained in Chap.
3.2, it has to be supposed that the crown (and thus the less voluminous parts) of the tree
used for the keel was placed to the bow of the hull.
As the hulls bottom was submerged into sediment, keel height could not be measured;
the shapes shown on Visnikars and my drawings thus are assumed (see the discussion of
possible keel-shapes in Chap. 3.2). As described, on the seabed the hulls remains listed to
about 5-10o to starboard, the proposed angle of the initial deadrise of the hull; for the centre
section C14-16 I accordingly reconstructed the lower parts of the keel as a circle segment
defined by two chords at angles of 10o to the lower keel-side edges of the garboard strake
(see Fig.3.2.i-13). Height at C6, C11, C19 and C21 follows Visnikars assumption (and, as

ne
r

noted above, necessity for a stringent 3D reconstruction) that top and bottom surfaces of
the centre parts of the keel were even. The section profiles of these stepping points were
then used as delimiters for lofting the intermediate keel segments.

Lie
b

To the bows, the cross-section at C6 was lofted to gradually comply with the (slightly
revised see below) form and 2D measurements of the bow-wing recorded by Visnikar at
C1, thus resulting in a rise of the keels top surface of about 1o; based on the above supposition that the keel was more beamy to aft ship, this process was repeated for C21 C24,
with, however, assuming a somewhat more shallow aft-wing. Visnikar, though, had envisaged a flat top surface for the whole length of the keel, and measured accordingly; when the
garboard strake had to be (virtually) attached to the keel sections, the elevation of frame
sections C1-C5 and C22-C24 thus had to be changed to match onto the envisaged slight longitudinal rise of the keels top surface. A number of measurements resulting from the above
operations are illustrated in Fig.3.2.ii-5.

rst

H.

Modelling the bow-wing based on Visnikars 2D drawings alone proved impossible:


probably not familiar with the concept, he had not drafted angle and transverse measurements of the wings protruding cheeks visible on the available photographic and video recordings; he had not included the whole length of the actual wings to aft, and not drawn
the slight rake of the bow-wings prow reported by other divers4; as they, possibly, had been
lost when he measured the bows, he had not noted the extant remains of a (first? see
Chpt.3.2) step of the bow-wing that aligned with the forward butt of the starboard garboard hood observed on still and moving pictures taken several months before5 (Fig.3.2.ii-6;

Ho

comp. Fig. 3.2.i-10). The general shape of the bow-wing was discussed and outlined shortly
4
5

E.g., F. Dobberphul, pers.com. and discussions, November 2004, May 2005, July 2012.
The video was not available to Visnikar when drawing his 2d plots, and supplied to t.p.a. only in 2009.

01
4

after the first photographic recordings of the bow-wing became available in May 2005, and a
preliminary 3D model refined in course of the efforts of producing an initial site drawing

Ma
y2

throughout 2005 and 2006. The solution adopted here (Fig.3.2.ii-7) attempts to combine
this preliminary model with Visnikars measurements, the observations reported by other
divers mentioned above and interpretations of the available photo and video material. It
will be noted that the resulting 3D model does not adhere to all measurements reported by
Visnikar: a number of changes resulted from attempts to produce a fair and straking 3D
shape based on his key measuring points; some of the features he had reported are unalike
to or not evident in the available photographic material; it was impossible to align the bowwings measurements with the cross sections he had taken off frame stations C1 and C2, the
following planks and the assumed overall layout of the keel. To produce a consistent overall
3D model it, hence, was unavoidable to adopt a number of compromises, which, too, necessitated changes to the two cross sections C1 and C2 (Figs.3.2.ii-8, -9).

Lie
b

ne
r

Due to the obstruction by keelson and floors, Visnikar in his 2D plot (Fig.3.2.ii-3) had
noted holes for lashing rope in the keels topsides only on C1 and C24; as, however, its edges
display the rabbets replacing the orthodox tambuku-lugs on the hulls floor planking6, such
holes should be present, in an arrangement following the corresponding holes on the garboard strake. Visnikars observation that the V-shaped hull section at C1 would not allow
for enough space between the keels edges and the garboard strake to accommodate a rabbet
of sufficient depth for the horizontal part of lashing rope holes7 was confirmed by the virtual reconstruction of the hulls remains: with the given parameters, the computer-generated
keel/garboard strake profiles (barely) allot the necessary space for such holes from C3 onwards to aft (Fig.3.2.ii-10). On C1-2 I accordingly opted to draw two pairs of holes traversing under the keels top surface; a to a certain extent comparable arrangement can be observed on the bow-wing of the Punjulharjo wreck. On the somewhat wider and less inclined aft section, L-shaped holes could commence at c23 to forward.

rst

H.

Visnikar had allowed about 10 cm for plank thickness, fixed depth of the rabbets between the planks edges and the tambuku (lugs or ridges see Chpt.3.2) on nearly all
planks onto about 5 cm, and their height on about 4 cm. Examination of the available samples, however, showed rabbets depth and tambuku heights varying between respect. 1-5.25
cm and 1.5-4.5 cm, and plank thicknesses of 10-13.25 cm (Fig.3.2.ii-11). To maintain workflow
and eventual discernibility in the 3D models, I changed the rabbets on all planks to 4(H) x 5
cm(D). Around the central keel sections I assumed a thickness of 14 cm for the garboard
strake, thinning down to 11.8 cm at the last strakes. The changes along the planks edges
resulted in a few cases in slight alterations of deadrise on the outer surface of the model,
none of which, however, exceeded 1.5o. The revisions of plank thicknesses were only ap6

Ho

For a discussion of the various shapes of planks in the remaining structure see Chpt. 3.2, espec. Figs.3.2.i-29 31.
7

Pers.com. October 2005; comp. the representation of these holes on his 2d overall drawing (Fig.3.2.ii-3).

01
4

plied to the weatherside surfaces of the hull, thus not changing the original measurements
taken from inside the vessel.

Ma
y2

In both his 2D overall hand-drawing and computer plot Visnikar represented the horizontal lines connecting the various vertical profiles of the plank strakes as fair curves; this
impression is supported by video recordings and photos and complies with the necessities of
naval constructions8. However, when attempting to combine the 2D profiles into a 3D
overall drawing, it became obvious that merely connecting the various points measured
would not result in fair lines, and a number of adjustments had to be made. I largely refrained from applying such adjustments in the lower strakes, and only re-aligned lines when
clearly necessary; the overall drawing thus still retains some slightly non-straking sections
(e.g., C17-18, strakes II-IV). In several cases (e.g., C6, C16, C18, C22) fair and straking plank
lines could be achieved by simply moving plank seams to the next node drawn / recorded
by Visnikar: it would seem, that he during the underwater recording process at several plac-

Lie
b

ne
r

es had mixed up measurement points due to tiefenrausch. Somewhat sizeable modifications


were unavoidable on C5, where, as described in 3.2, the hull had displayed a suspicious
bend (Figs.3.2.ii-12, 3.2.i-24, -27, -28) probably caused by a shift of the cargo to starboard
fore at impact of the hull on the seabed. A number of further adjustments had to be made
around the lateral lines of the strakes scarfs and are probably due to the fairly complicated
geometry at these sections. Frames C17-18 and C22, covered by concretions and thus inaccessible to measuring endeavours, had to be aligned following the lines given at C16 and
C19. In the overall Figs.3.2.ii-13 and -14, it will, however, be observed that most of the
changes are small; except on C5, none exceeds 5 cm in real measurements.
On his overall freehand drawing (Fig.3.2.i-9) Visnikar had sketched only one of the usual
two holes used for attaching frames on several of the foreship drop planks, and a second
hole as placed in the consecutive hood plank. This impression is not supported by the
available photographic and video documentary, and was corrected accordingly in the final
plots (see, e.g., Fig.3.2.i-22). It, too, was impossible to combine the various measured points

H.

along the hooked scarfs at the plank-joints into fair and straking 3D models; to ease the
modelling process I thus adopted a standard model for these scarfs, mainly based on video
recordings, photos and the measured plank scarf on s581 and s583 (Fig.3.2.ii-15).

Ho

rst

It should be noted, that despite their various limitations discussed above Daniel Visnikars measurements and drawings without doubt can be combined into a stringent model of
the ships remains, an achievement even more commendable when taking into account the
intricacies of his task on a site in such an extreme depth.

First of all the necessity of producing a hull that is absolutely symmetrical and fair (having regular curves
and surfaces that are smooth) (Chapelle 1996).

20
14
Ma
y
ne
r
Lie
b
H.
Ho
rst

Fig.3.2.i-1: Example frame section measurements by


D. Visnikar (green) and actual measurements to a
precision of four digits behind 0 (red); all
measurements in m.

ay
20
14
Lie
bn
er
M
H.
rst
Ho

Fig.3.2.i-2: Visnikars overall

frame sections; measurements


in m.

20
14
Ma
y
ne
r
Lie
b
H.
Ho
rst

Fig.3.2.i-3: D. Visnikars overall 2d plot of the ships remains; measurements in m.

ay
20
14

55

covered by concretions:
< 50%
>50%

45

40

35

30

25
0

Lie
bn
er
M

Measurements, cm

50

10

12

14

16

18

20

22

24

Keel Section, C#

55

Measurements, cm

50

45

40

35

H.

30

25

10

15

20

Keel Sections, m

rst

reconstructed measurements, broadest keel sections, C1-C6


reconstructed measurements, top keel
Visnikar's overall plot, top keel
Visnikar's single sections, top keel
Visnikar's single sections, base keel

Ho

Fig.3.2.i-4: Measurements of keel sections. Above, per frame station; below, linear measurements.

20
14
Ma
y
ne
r
Lie
b
H.
Ho
rst

Fig.3.2.i-5: Reconstructed keel-plank; all


measurements in cm. Frame labels in red.

20
14
Ma
y
ne
r
Lie
b
H.
Ho
rst

Fig.3.2.i-6: Bow-wing, D. Visnikars section drawings; measurements in cm.

ay
20
14
Lie
bn
er
M
H.
rst

Fig.3.2.i-7: 3d model of bow-wing (dark), proposed


lost step (transparent), fore end of keel (light

Ho

brown); measurements in cm.

ay
20
14
Lie
bn
er
M

Fig.3.2.i-8: Section cut through bow-wing (dark), remaining planks (light brown), section profile

rst

H.

(grey) and Visnikars frame C1 (black outlined/hatched).

Ho

Fig.3.2.i-9: Bow-wing (dark), remaining bow planks (lighter brown) and Visnikars frame section C2
(black outlined/hatched).

ay
20
14
C1

C2

C3
C4

Fig.3.2.i-10: Lashing rope holes on computer-generated keel and (portside) garboard strake, sections
C4-1, looking forward. Note the undercutting of horizontal holes into the plane surface of the

Sample

Lie
bn
er
M

rabbet on the keels edge at C3, and, to a lesser extent, C4.


Thickness

Depth Plank Edge


Tambuku

Height Planksurface
Tambuku

13

2.5 - 3.5

1.5 - 1.75

10 - 12.5

3.25 - 3.5

3.5 - 4

12.75

2 5.25

4 - 4.25

10.5

n/a

n/a

strake XI, portside, C13-15

11.5 - 12.5

2.5 - 4

3.5

strake VII, portside, C19-20

10.75 - 12.75

1.5 - 2

n/a

strake X, portside, C18

12.25 - 13.25

1.5 - 1.75

3.5 - 5

strake VII, portside, C19-20

10.5 - 13

probably strake X, starboard, C6

11.5 - 11.75

1 - 3.5

4.5

Position

loose, off portside bow; probably part of


plank connections at C3-4
loose, off portside bow; probably part of
plank connections at C3-4
loose, off portside bow; probably part of
plank connections at C7-6
loose, off portside bow; probably part of
plank connections at C7-6

s32
s33
s35
s36
s573
s581
s582
s583
s584

Fig.3.2.i-11: Plank thicknesses and measurements of rabbets on surfaced samples (cm).

Fig.3.2.i-12: C4-7, Visnikars

measurements (red; brown plank solids,


measurements applied in 3D site model
(green; transparent surfaces).

Ho

rst

H.

omitting tambuku lugs/ridges),

ay
20
14
Lie
bn
er
M
H.
rst
Ho

Fig.3.2.i-13: Visnikars original plank profiles (red) and the profiles adopted for the faired
3D model (dark green), single frames.

20
14
Ma
y
ne
r
Lie
b
H.
Ho
rst

Fig.3.2.i-14: Visnikars original plank profiles (red)


and the profiles adopted for the faired 3D model
(dark green), overall model.

ay
20
14
Lie
bn
er
M

Fig.3.2.i-15: Standard plank scarf adopted for 3D modeling. Top: Starboard bow, strakes II-VI;

Ho

rst

H.

below: Strake IV starboard aft. Measurements in cm.

Vous aimerez peut-être aussi