by Peter Yule
Batten was responsible for enforcing contract standards. Batten
recalls that his staff continually checked factories around Aus-
tralia and sampled by physical inspection almost 20 per cent of
the work done. Each month about 160 large reports were made on
the progress of the sub-contractors and the quality of their work.
While the sub-contractors around the world were building
their particular components, ASC’s yard was proceeding steadily
on the construction of the hulls.4 An important milestone was
the ‘keel-laying’ ceremony for the first submarine on 14 Febru-
ary 1990, at which Kim Beazley announced the names of the six
submarines. They were named for six Australian sailors of the Sec-
ond World War – Collins, Farncomb, Waller, Dechaineux, Sheean and Rankin. HMAS Collins, as the first boat, gave its name to the class.5
Each submarine was built in six sections, each of which con-
sisted of several ‘cans’. The sections were numbered 100, 200,
300, 400, 500 and 600, with 600 being the forward section and
100 being the aft section. Each sub-section was numbered; so, for
example, the 200 section is made up of a 210, 220, 230, with 210
made up of two cans, 220 of three cans and 230 of two cans –
seven cans welded together to make the complete section.
The first major work at Osborne was welding three curved
plates of steel to make the cans and then welding the cans together
148
T H E C O L L I N S C L A S S S U B M A R I N E S T O R Y
to make four hull sections for the first submarine. Welding the
cans and sections was among the most critical and risky parts
of the whole project. The high-tensile steel used for submarine
hulls required the highest quality welding, with welding problems
being among the most frequent reasons for delays or even failure
of submarine projects around the world. One of the great suc-
cess stories of the project was that the welding techniques used
by ASC, developed in conjunction with DSTO, proved highly
successful.6
The hull construction shop is 150 metres long and 40 metres
wide and is divided into a series of workstations. At workstation
100 the curved plates from CBI were welded to form the can,
using full penetration, manual metal arc welding. The can of about
20 tonnes then moved to workstation 200. Here the ‘T’ stiffeners
from CBI and bulkheads from Perry Engineering were welded into
the cans. The stiffeners were welded to the inner part of the hull
with an automated submerged arc welding machine. One side of
the can would be fully welded and then the whole can would be
picked up, spun around and put down to allow work on the other
side of the T-stiffener welds. Magnetic particle testing occurred in
the weld prior to welding the second side. After the second side
weld was complete there would be 100 per cent visual, 100 per
cent magnetic particle and 100 per cent ultrasonic inspections;
10 per cent of the welds on the pressure hull were also subjected
to X-ray inspections.
All the material on the hull was micro-alloy steel, designed
to absorb energy and buckle rather than fracture if subjected to
explosions. However, such steel loses its granular structure and
therefore its strength if it is cooled too quickly. Consequently, the
steel structure had to be preheated and postheated for all welding
operations. It was potentially extremely dangerous to use heaters
while welding in confined spaces, so ASC developed a process
of heating the steel using low voltage heaters that lessened the
chances of accidents.
With stiffeners and bulkheads in place, the cans were moved on
to workstation 300, put on rollers and welded using an automated
process developed by ASC. The cans rotated past the fixed welding
machine, joining seven cans to make a complete section.
Four sections were built from the outside in – first putting the
cans together, then adding the frames – but several cans in two
B U I L D I N G S U B M A R I N E S
149
sections were built from the inside out. These were the cans with
flat bulkheads in the front and mid sections. In these cans shear
plates were welded first, followed by the stiffeners, the frames
and the infill plates, and finally the hull plate was welded around
the fabricated section, becoming so strong that the production
workers called it the ‘egg crate’.
Before all the sections were welded together most of the tanks
were welded into the hull. There are over 50 tanks in each sub-
marine, including freshwater tanks, ballast tanks, trim tanks, fuel
tanks, waste water tanks and torpedo compensation tanks.
As the cans were joined into sections, they were put in cradles
so they could be moved around the hull shop by rail, and eventu-
ally the whole submarine could be rolled out to the ship lift.
After the welders had finished as much ‘hot work’ as possible,
sections went to the blast and paint chamber, which was fully
enclosed and designed to take all the submarine sections. It would
take several months to blast and paint each section – a source of
frustration because meanwhile nothing else could be done with
the section.
Blasting in the confined spaces of the submarine sections was
difficult and unpleasant work. At each stage the section had to be
vacuumed and inspected before painting. Blasting and painting the
external hull was comparatively easy, especially compared to work
inside the tanks. It was often impossible to paint in winter because
of cold surfaces, and later ASC installed heating to improve the
schedule.
After several months in the blast and paint shop it was on to the
outfitting shop. G öran Christensson from Kockums was the first
manager of the outfitting shop, and when he returned to Sweden
in 1990 Robert Lemonius from Cockatoo Island took over the
role.
The 200 section of each submarine was always the first to enter
outfitting, where the major equipment received from Australia
and overseas would be installed. For the first boat much of this
came from overseas, but afterwards from Perry Engineering (steel
fabrication) and ASC Engineering (outfitting) in Adelaide. The
machinery and platforms arrived in the outfitting shop with the
sections outfitted sufficiently for platforms to be installed. These
came from Kockums for the first section of the first submarine,
but thereafter were made in Adelaide by Johns Perry.
150
T H E C O L L I N S C L A S S S U B M A R I N E S T O R Y
Some areas of the submarine are almost inaccessible after
the platforms are in place, so everything had to be in perfect
order before major equipment such as the battery compartments
was installed. There was always a temptation to complete plat-
form installation as this was a key milestone in the progress
of each submarine, but it was a serious mistake to do it too
early as it made uncompleted work inaccessible or difficult to
complete.
<
br /> Cabling is always a major challenge in submarine construction
because the cables are installed in continuous lengths wherever
possible, avoiding cable junction boxes and lessening the oppor-
tunities for faults. However, this means that there will be numer-
ous cables rolled up into lengths on sections or on platforms that
can only be connected when the submarine’s sections are joined
up and the cable can be rolled out.
Piping presents similar challenges. Each Collins submarine has
about 23 kilometres of piping, and early in the project ASC
decided to invest in an expensive pipe-bending machine rather
than use straight lengths of pipe and then weld in pre-bent elbows.
The reason for this was that the specification for pipe welding was
extremely demanding, requiring extensive testing and inspection
and making welding expensive.
While the hull construction was proceeding, ASC engineers
worked on the tooling required to install different equipment on
the boats. For example, the propulsion motor weighs about 90
tonnes, and large jigs and fixtures were required to transfer and
load the engines into the submarines at the appropriate stage of
the outfitting process.
Once adjoining sections had been outfitted as completely as
possible, they would be welded together using manual arc welding.
These joints took about four weeks to complete, running teams of
six welders in three shifts around the clock, and had to be perfectly
round.
With joints consolidated and outfitting completed, initial test-
ing of systems and equipment began in preparation for the launch
of the submarine. After the launch the submarine would spend
about nine months alongside the wharf during the ‘set to work’
program, ensuring that all the systems worked to the original
specifications.
B U I L D I N G S U B M A R I N E S
151
On 24 July 1993 the first submarine completed outfitting and
testing and ASC held a rollout ceremony. Simon Ridgway recalls
that:
We had been going for four years in construction and it was
an opportunity to say: ‘Hey, we’ve actually built a finished
product.’ It was a great day to recognise everyone’s
achievement. Everyone was allowed to invite their family and
the submarine was rolled out, using our 40 tonne crane as a
tractor to pull it out of the outfitting shop and onto the
hardstand.
C H A P T E R 14
The automated integrated vision
Part 1: The combat system 1987–93
The success of the weapons update program for the Oberons gave
Australian submariners a vision for a fully-integrated combat sys-
tem and also prompted them to consider combining this with
a highly-automated ship control and management system. The
automated, integrated vision became central to the requirements
for the new submarines and was one of the highlights of Kim
Beazley’s press release on the announcement of the contracts in
May 1987:
The combat system for the new submarines represents about
one third of the construction cost and will be assembled in
Sydney by Rockwell Ship Systems Australia . . . The
computerised combat system will be more advanced than any
yet installed in a diesel-electric submarine . . . All tasks can be
carried out at any of the multi-function common console
work stations in the control centre. There is no central
processor to present a single point of failure and the data
152
T H E A U T O M A T E D I N T E G R A T E D V I S I O N
153
distribution system can sustain significant damage or failure
and still function satisfactorily.
A prime example of the new technology is the ship
management system, which greatly minimises the workload
by providing computer-based control and monitoring though
common multi-function consoles. It is the main reason why
the crew can be reduced to 41 compared with the
Oberons’ 63.1
The ship control and management system is widely regarded as
one of the great successes of the submarine project, but the combat
system never approached the level of performance dreamed of by
those who planned ‘the world’s best combat system’ or demanded
by the contract specifications. Although overshadowed in the pub-
lic mind by the mechanical problems revealed during the trials
period in the late 1990s, the combat system was by far the most
serious and intractable problem of the project. If the combat sys-
tem had worked as planned the history of the project would have
a far rosier glow.
During 1987 and 1988 teams of engineers and designers in
Sweden, California, Australia and France began planning to make
the automated, integrated vision a reality. There was no sense of
foreboding but several critical decisions had already been made
that resonated controversially throughout the project. These deci-
sions have been blamed for most of the difficulties the combat
system project encountered, but there is a total lack of consensus
on how and when these decisions were made and their relative
consequences for the combat system.2
At the most fundamental level, the ‘architecture’ of the com-
bat system is frequently seen as either fundamentally flawed or too
ambitious (or both) and this is usually sheeted home to the orig-
inal requirements drawn up by the project team and the SWSC.
Ian MacDougall as director of submarine policy and warfare was
involved in writing the concept papers for the new submarine in
the mid-1980s, and he believes they made a mistake by speci-
fying distributed architecture as this was too big a step for the
technology available at the time. He still feels bitter towards the
‘big contractors who all signed up and said they could do it, but
then failed to deliver’. Nobody told the navy that ‘this cannot be
achieved’.
154
T H E C O L L I N S C L A S S S U B M A R I N E S T O R Y
Many later observers also see the initial design concept as
overly ambitious. Colin Cooper, who was combat system man-
ager for the new submarine project from 1993 to 2003, argues that
‘the design concept was probably 30 years ahead of the reality of
processor and memory capacity, system hardware and software
able to support that concept’.3 Even Oscar Hughes, the project
director, thinks that in retrospect ‘we were too ambitious with the
combat system’.
It is widely believed that the combat system was designed by
the SWSC and that the design was then imposed on Rockwell.
This is not an entirely accurate view. The centre, together with the
submarine project office, drew up the detailed specifications and
operational requirements for the combat system, but it did not
design the system. Diagrams of the Rockwell and Signaal systems
submitted after the project definition study show that they differed
greatly, and were not simply quoting to build a
system which
was already designed. The SWSC laid down what the combat
system should be able to do and recommended the use of the Ada
language and distributed architecture, but it did not design the
combat system.
Perhaps not surprisingly, those who were involved with the
combat system specifications argue that the problems arose from
poor implementation rather than misguided specifications. They
argue that the Rockwell team made major errors in designing its
system, and these were exacerbated by poor choices of processors,
compilers and other hardware, an overly academic approach to
writing the software and, finally, incompetence at integrating the
various components of the combat system. Mick Millington, for
instance, argues that the star-connected fibre optic data bus could
never guarantee the bandwidth the system required and it was
‘doomed based on that alone’.
The boffins from the SWSC are not alone in their view. Ron
Dicker, with his unique perspective on the issues as a former Dutch
submarine commander, manager of the Signaal bid for the com-
bat system and ASC’s combat system manager, argues that the
main cause of the problems with the combat system was not the
architecture, although the requirement that ‘any console can do
all functions’ and ‘all consoles can do any function’ simultane-
ously created unnecessary difficulties. In Dicker’s view the prob-
lems arose because
T H E A U T O M A T E D I N T E G R A T E D V I S I O N
155
Rockwell in their desire to retain ownership of the system
design had contracted core elements . . . all over the place
[and] that did not allow them in the end to manage the
design evolution and integration and to take advantage of the
rapid developments in computer technology.4
Rick Neilson recalls that he was the person who said when prepar-
ing the requirements, ‘we will do this combat system in Ada’. This
turned out to be a problem, although he argues that Ada as a
language was not the problem; rather it was the newness of the
language, which meant that much of the support for it was uncer-
tain. This showed up particularly in the choice of compilers for
converting the Ada software to machine code compatible with the
68000 processor series. It was not clear which compiler was most
suitable and Rockwell gambled by choosing Verdix, which proved
