Professional Documents
Culture Documents
Quantification of Manufacturing Complexity in Shipbuilding Projects 31-5
Quantification of Manufacturing Complexity in Shipbuilding Projects 31-5
Production
Quantification of Manufacturing
Complexity in Shipbuilding Projects
SDPO.016.015.m
By
L. J. G. Huijgens
Performed at
Company supervisors
Responsible supervisor: Ing. F. M. Verhelst BBA
E-mail: [email protected]
Author details
Student number: 4258819
Author contact e-mail: [email protected]
QUANTIFICATION OF MANUFACTURING COMPLEXITY IN SHIPBUILDING
PROJECTS
SUMMARY
An issue that arises during the pre-contract phase of any large engineering project is correct
estimation of the work associated with execution of the project. In the shipbuilding industry,
estimations in this respect are often based on man hour-per-ton figures, which in turn are
usually based on regression analysis and experience. It is known that such figures vary greatly
in function of vessel type and specifications, and this variance is often attributed to
“complexity”.
Often, an estimation of production man hours required for construction of a ship is made by
looking at previous, comparable projects, and extrapolating the production work by assuming a
constant man hours-per-ton ratio for steel work and outfitting. Although this gives an indication
of the work force costs, it does not provide much insight in the distribution of work content over
different systems. This research aims to provide a method that allows more detailed
estimations to be made without raising the need for exhaustive input data sets.
The method proposed in this report provides a means to compare a shipbuilding project to
previous projects in terms of technological complexity through a set of standardised formulas,
where complexity is expressed as work content connected to the manufacturing process of the
ship. Using estimated constants and exponential factors, based on regression analysis and
engineering judgement, relations between vessel requirements, components dimensions and
man hours can be adapted to a mathematical framework. This way, estimations based on
rough extrapolations of past projects and “gut feeling” can be substituted by a more
mathematical approach, and well-reasoned factors expressing relative complexity per system
already in the earliest phase of the project can be generated. Substantiated numbers of this
kind can be valuable during pre-contract and company-internal negotiations.
In the first chapters, an elaboration on the concept of complexity is given, and it is investigated
how previous research approached the subject. Secondly, a mathematical framework for the
method is set up and linked to the findings from previous publications. Subsequently, a
practical model for demonstration and validation of the proposed method is developed, from
which eventually conclusions with respect to applicability and results can be drawn. The report
ends with recommendations regarding use of derived models and further development of the
estimation method.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 6/153
Quantification of Manufacturing Complexity in Shipbuilding Projects page 7/153
TABLE OF CONTENTS
PAGE
1. INTRODUCTION 11
1.1. Backgrounds 11
1.2. Research Objective and Scope 12
1.2.1. Research Questions 12
1.3. Research Methodology and Structure 13
1.3.1. Preliminary Research 13
1.3.2. Technical Analysis of Newbuilding Specifications 14
1.3.3. Defining Basic Principles of the Quantification Method 14
1.3.4. Solution Design: Method for Quantification of Complexity 15
1.3.5. Results of Demonstrative Calculations 15
1.3.6. Conclusions and Recommendations 15
2. PRELIMINARY RESEARCH 17
2.1. Defining Complexity 17
2.1.1. Types of Complexity 18
2.1.2. Aspects of Complexity 19
2.1.3. Causes for Complexity in Large Engineering Projects 21
2.1.4. Summary and Categorisation of Complexity Aspects and Causes 23
2.1.5. Expressing Complexity 25
2.2. Technological Complexity in Shipbuilding Projects 26
2.2.1. Amount of Systems and Components 27
2.2.2. System Density 27
2.2.3. Casco Geometry, Plate Thickness and Girder Spacing 27
2.2.4. Machinery Spaces 28
2.2.5. Pipe Spools and Ducts 29
2.2.6. Electric Wiring 29
2.2.7. Modular Building Strategies 29
3. METHOD FOR TECHNICAL ANALYSIS OF SHIPBUILDING PROJECTS 31
3.1. Structure of the Technical System Analysis 31
3.1.1. Level 1: Main Systems 32
3.1.2. Level 2: Subsystems 32
3.1.3. Level 3: Primary Components 33
3.1.4. Level 4: Secondary Components 33
3.1.5. Differentiation Groups 34
3.1.6. Application of the Proposed Categorisation 34
Quantification of Manufacturing Complexity in Shipbuilding Projects page 8/153
1. INTRODUCTION
1.1. Backgrounds
Contrary to for instance the automotive industry, where series of cars are built by the thousands
by the use of line work, the shipbuilding industry is much more project based. Although designs
of vessel may often be influenced by each other, even sister ships are often considered as
different projects.
Each ship design has its own characteristics, such as hull dimensions, loading gear, propulsion
systems, ancillary equipment and in many cases, specialised equipment that allows the vessel
to perform its specific task. Because of the near infinite variables in a vessel’s design, the
duration and logistical planning of a building program can seldom be easily derived by looking
at previous projects. This especially goes for the one-of projects executed by the Damen
shipyards such as certain types of naval ships, offshore supply vessels and yachts. These
ships not only possess unique specifications but in some cases also have to be built on varying
locations in collaboration with local yards.
In the shipbuilding industry, figures concerning cost and amount of work hours required for
construction of sections are often based on the weight of the section. Experience however
shows that variations in the planning are inevitable, also between projects which are similar in
terms of weight: system density (the amount of installations per unit of available space) and the
number of different systems result in a varying relation between section weight on one end, and
cost and required construction time on the other end.
This variation in cost is generally attributed to complexity, and is apparent not only during the
outfitting phase of the project, when systems are integrated into the hull, but also during casco
construction, depending on hull geometry. Often, experience allows production engineers to
estimate whether or not a cost figure is realistic, however this usually based on gut feeling
rather than on a detailed, mathematical evaluation of the project. So far, complexity within
Damen Schelde Naval Shipbuilding (DSNS) projects has always been roughly taken into
account based on experience by engineers; predictive models were developed based on
regression analysis of a limited set of in-house projects and these models were never validated
in detail. It is the goal of this research to provide a more exact approach to complexity
anticipation in shipbuilding project planning by identifying relevant systems and analysing their
characteristics and reciprocities.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 12/153
The eventual result of this research project is a substantiated method in the form of a
mathematical framework which allows generating a set of factors giving an indication of what
are the parts which make a shipbuilding project complex in construction, and how complex the
project actually is.
The focus of the research is on the theory behind this method: development of a mathematical
framework and substantiation is the primary objective. Additionally, a model is developed which
serves as an instance of how the quantification method can be applied and is used for
validation of the method.
Two issues are explicitly kept out of scope: logistical uncertainty and technological uncertainty.
Logistical uncertainty is related to external parties such as subcontractors and clients. This
factor is not taken into account, as uncertainty in this respect depends more on the
environment than on the properties of the vessel itself. Expanding the research in this direction
would require considerably more time and data without adding much technical insight in the
actual influence of vessel specifications on construction complexity.
To demarcate the subjects covered in this report, research questions are formulated.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 13/153
To work towards an answer to the main research question in a structured way, and allow
demonstration and validation of results, sub questions are formulated:
In chapter 8, the findings of the research described in this report are summarised such that they
form answers to the abovementioned questions.
To give the report an accessible structure, the contents of the chapters are closely linked to the
chronology of the research.
The first phase of the research is based on review of scientific literature and consultation of the
engineers at the project and production departments.
Complexity: the state of having many parts and being difficult to understand or find an answer
to.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 14/153
However sufficient as a general description, this is too vague to serve as a basis for a research
on engineering projects. The concept of complexity in an engineering environment needs to be
clearly defined by evaluation of publications on this subject, which is done in this chapter. It is
important to note that this research is aimed at developing a mathematical structure, allowing a
substantiated, exact comparison of shipbuilding projects in terms of work content, taking into
account technological complexity, rather than delivering an improved definition of complexity.
Another subject covered during the literature review is the identification of common causes for
complexity in engineering projects. The basic list of complexity aspects resulting from literature
review on this subject is supplemented by input from the project department.
The eventual goal of a technical analysis is to define a set of relevant components which can
be linked to a certain amount of work (usually expressed in man hours). Special attention is
paid to the complexity aspects which have come forward during the literature review and
consultation of the engineering, project and production departments. These aspects (for
instance spatial constraints) are linked to technical properties of components and systems
(such as power requirements) and interference with other systems (for example an increasing
number of components in the same room, leading to difficulties during outfitting).
As an example, a system breakdown in main systems, subsystems and components for the
demonstrative quantification model developed to illustrate the theory in this research project will
be given in Appendix I.
A clear definition of complexity, along with the identified aspects leading to complexity resulting
from the literature review and discussions with project and production engineers, allows for
identification of the principal characteristics which will be used in the quantification method.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 15/153
More specifically, this chapter lists the format and significance of input and output variables. A
straightforward way to express complexity is to calculate a correction factor per complexity
aspect, and then too, it can be argued how these factors are combined (or whether they should
be combined at all). Besides, it might be more interesting not to work with factors but to divide
projects into categories, depending on their overall complexity.
With the aspects of complexity linked to the systems commonly installed on the Damen ships,
and the influence of modular building strategies on complexity identified, a method for
predicting and ranking complexity of projects can be developed, according to the mathematical
principles devised before. The result is a set of factors and formulas which give an estimation of
the relative complexity of each aspect.
The method is demonstrated and validated using a model developed in Microsoft Excel. This
model is an application of the mathematical theory, with values assigned to relevant factors
based on regression analysis and engineering judgement. Validation and adjusting of the
demonstrative model is done by comparing historical project data and by feedback received
from the project, production and engineering departments. The results of the calculations are
given in the final chapters, providing both a quick overview in the kind of results that can be
expected from models based on the quantification method put forward, as well as a validation
that such models can yield useful results with only limited input data.
Several projects for which data is made available by DSNS are processed according to the
proposed method, and results regarding the accuracy of the demonstrative model and, most
importantly, the relevance of the quantification method are discussed. It will be shown that the
method can be of use not only during planning of production planning but also for evaluating of
vessel designs in the earliest engineering phases.
Due to the many facets connected to complexity, and hence the subjective nature of many
assumptions that need to be made, further development of the quantification method may be
appropriate. Furthermore, future projects may offer new insights which allow a more precise
Quantification of Manufacturing Complexity in Shipbuilding Projects page 16/153
estimation of certain aspects. Analysis of other ships and fine-tuning of the method using data
concerning these vessels can make the method applicable to a wider range of ship types.
Remarkable results, limitations and possible future improvements recognised during the
research are listed in this final chapter.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 17/153
2. PRELIMINARY RESEARCH
Firstly, a definition for the term complexity is derived by summarising literature, along with
aspects which are agreed to be determining for the degree of complexity.
The resulting findings are applied on documentation regarding Damen projects in the past, and
supplemented by experience from engineers in the project and production departments.
Finally, the principles of modular shipbuilding are investigated and effects on complexity are
analysed.
Before going into detail concerning definitions of complexity, it must be noted that in this
research, the term complexity is merely used to identify difficulties and bottlenecks in the
construction of ships. Limitations with respect to versatility of the definition used in this research
must be seen in this light; rather than to develop a widely applicable, theoretical definition, this
chapter seeks to combine literature with engineering judgement into a set of relevant aspects
that can be used to quantify relative complexity in ship construction.
From preceding research, a division of the concept of complexity can be deducted. Several
papers propose a division into two major aspects (Baccarini, 1996):
Different proposals regarding definitions of project complexity were put forward in countless
other publications (Bosch-Rekveldt, Jongkind, Mooi, Bakker, & Verbraeck, 2011); however the
elaborated concept of complexity as given in this chapter is considered as most suited for
evaluation of Damen’s newbuilding projects, as this report aims primarily on evaluation of
technical properties of systems and components rather than management issues which are
often the subject of papers on complexity in engineering projects.
A high-level categorisation of complexity into two types can be made: organisational and
technological complexity.
It can be stated that organisational complexity, as defined in the course of this research, covers
both external as internal elements of a project (example given relations with subcontractors and
between internal departments), while technological complexity focuses mostly on the technical
system properties.
2.1.2.1. Differentiation
Differentiation in a system can occur in two ways: horizontally, on a single hierarchic level, and
vertically, referring to the number of hierarchic levels (Hall, 1979).
When considering technological complexity, horizontal differentiation refers to the way in which
tasks are structured; two ways of differentiation can be cited (Beyer & Trice, 1979).
A job could be split up and spread over several workers, each doing a specific part of the job.
This practice is referred to as division of labour, and is observed for example at an assembly
line. Complexity can then be measured by the amount of different job parts, a number directly
proportional to the number of workers.
Alternatively, workers could be skilled in such a way that they can complete a wider range of
technically challenging jobs by themselves; this is called personal specialisation (Hall, 1979). In
this case, complexity can be expressed by the number of specialisations required to finish a job
(Dewar & Hage, 1978).
However often labelled as a technological aspect of differentiation, task assignment is more a
matter of organisation and will not be extensively covered as such. However, in connection to
this, the variety of systems and components can also be considered a measure for
differentiation; this aspect will be explicitly taken into account in the estimation method.
2.1.2.2. Interdependencies
Dependencies can be further broken down into different categories: depending on how
changes in a certain element affect other elements, relations can be categorised as pooled,
sequential, reciprocal or by feedback (Williams, 1999). Same as with the definition of
complexity, multiple ways to categorise interrelation between elements are found in literature.
Like differentiation, interdependency can be horizontal, at the same hierarchic level, or vertical,
between different layers. The levels and hierarchies of technical systems on board will be
covered during the technical analysis later; a description of system and component levels and
the reasoning behind these subdivisions can be found in section 3.1.
As dependencies between organisational layers, when not merely sequential, are the result of
technical interdependencies (for example when iterative corrections are necessary because of
combination of different systems into a single space), and since this research aims at a detailed
evaluation of said technical interdependencies, a further subdivision of organisational
complexity is not made.
Sequential dependency implies that if one element changes, so does another. For instance, a
larger heat exchanger requires a more powerful cooling water pump.
Feedback dependency means that if alterations are made to an element, this has impact on the
element itself, requiring further adjustments. This means that iterative corrections must be
made to the element to ensure correct design and construction. As an example, if a ship with a
specified hull shape and a combined diesel-electric or diesel (CODELOD) propulsion
configuration is to have a design speed higher than usual for that specific class, the maximum
continuous rating (MCR) of the main engines is most likely to be increased, which implies
Quantification of Manufacturing Complexity in Shipbuilding Projects page 21/153
heavier ancillary equipment. This influences design displacement and resistance, which in turn
raises the need for even more powerful engines to maintain the envisioned design speed.
It can be argued whether or not this is also a matter of vertical interdependency; however as
the element itself is ultimately influenced, it is considered a horizontal effect.
Pooled dependency means that changes of several elements together influence one element in
another level. For instance, the specifications of the propulsion system, deck gun installations
and heating, ventilation and air conditioning (HVAC) system all have influence on the
dimensions of the cooling water subsystem.
Reciprocal dependency indicates indirect influences from one element on another. This can be
illustrated using the same example as for feedback dependency: an increased design speed
may require more powerful main engines to be installed, which in turn need larger secondary
components such as fuel pumps, starting air compressors et cetera. It can be stated that
reciprocal dependency is in fact a combination of vertical and/or horizontal sequential
dependencies and as such could be eliminated as an interdependency category.
Preceding research on engineering projects in general has yielded valuable insights in how
experienced project managers judge projects for complexity (Bosch-Rekveldt, Jongkind, Mooi,
Bakker, & Verbraeck, 2011). As a basis for further evaluation on complexity in shipbuilding
projects, the factors of important as identified during such research projects are clarified and
categorised according to the criteria as given in section 2.1.2.
Undoubtedly, there will be overlap in the way the mentioned factors influence complexity. For
the sake of completeness however, all factors deemed relevant will be evaluated.
Dependencies between tasks: refers to how tasks connected to certain system relate
to each other. Finish-to-end relations for example may delay the starting date of certain
tasks, making for peaks in activities.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 22/153
Experiences with partners involved: a subcontractor which has been worked with
during previous projects and is known to be reliable can execute jobs in a more
independent fashion (requiring less guidance and corrections).
Experience with technology: the degree in which a yard’s own work force and
subcontractors are acquainted with installing and commissioning certain technology is a
very important and often cited source for unexpected problems and delays, as was
specifically mentioned in the 2011 paper by Bosch-Rekveldt et al.
Interrelations between technical processes: refers to how systems are related to
each other during operation of the vessel. This is an effective way to estimate which
systems share which components, which fittings (such as piping) are required to ensure
proper operation of these shared components, and how these fittings must be
dimensioned.
Global newness of technology: in close connection with yard experience, the
experience worldwide with certain new technology may be an important factor during
integration of said technology in a newbuilding. For applications of completely new
technology, practical problems may not be easily negotiated, which means more
engineering rework and tailor made solutions.
Number of different nationalities, languages and standards: communication may be
compromised or delayed as a result of language barriers, cultural issues or different
technical standards.
Number of locations: if parts of single ship are constructed on different yards (as is
often the case in modular construction projects), issues such as dimension monitoring
and independence of separate modules must be taken into account.
Number of departments directly involved with construction: this is a clear example
of how vertical differentiation influences organisational complexity. Not only hierarchy
may complicate the transmission of orders and translation of design concepts into
practical solutions; departments operating on the same hierarchic level but handling
different steps in the building process (such as the project and the production
departments) may too experience difficulties while transferring instructions.
Number of systems and tasks: depending on the amount of systems and components
involved, the yard work force will have to be present for a longer time and in greater
numbers. Keeping in mind spatial and resource constraints, this may complicate the
construction process.
Required local content: in some projects, especially those ordered by foreign
governments, requirements are set in respect of the involvement of local industry. This
may influence the project several ways already mentioned in this section. Because this
Quantification of Manufacturing Complexity in Shipbuilding Projects page 23/153
In section 1.2, it was stated that the estimation method will be based on engineering first
principles as these provide a solid basis for extrapolations. This implies that organisational
complexity will not be explicitly taken into account: extrapolations of historic data will be made
based on technical relations, and organisational issues will be accounted for by assuming that
these remain the same for both the historic (or benchmark) project and the newly evaluated
project.
In other words, the research mainly concentrates on technological complexity: it is assumed
that organisational complexity is constant and that learning curves are not applicable. In order
to ensure that this is a valid assumption, it was chosen to perform demonstrative calculations
with two newbuildings that were both first-of-class, and planned to be built on (or have
corrected man hour figures for in case of production on) the DSNS yard in Vlissingen: the
Sigma 10514 (PKR1) and Sigma 6110 surface combatants.
When estimation production work content for ships which are produced in series, it is advisable
to consult publications focussing on organisational complexity and shipyard learning in order to
more accurately assess organisational evolutions such as (Baitello & Martins, 2012) and
(Craggs, Bloor, Tanner, & Bullen, 2003).
To present an overview of how these factors can be investigated, the following categorisation is
proposed:
ORGANISATIONAL COMPLEXITY
Differentiation
Horizontal
Number of different nationalities, languages and standards
Number of locations
Size of the project team
Experience with partners involved
Size in CAPEX
Size in engineering hours
Required local content
Vertical
Number of departments directly involved with construction
Required local content
TECHNOLOGICAL COMPLEXITY
Differentiation
Horizontal
Number of systems and tasks
Variety of tasks
Experience with technology
Experience with partners involved
Global newness of technology
Vertical
Interrelations between technical processes
Dependencies between tasks
Horizontal Sequential
Feedback
Vertical: Pooled
Sequential
Reciprocal
Feedback
First of all, this research assumes that complexity is a function of the amount of work connected
to construction and outfitting activities. Acquisition costs for construction material, equipment
and tools are not considered.
Previous publications on complexity in engineering projects so far always concentrated on
figures based work content due to the intuitive relation with complexity (one would never call a
task complex if it can be done without much effort).
Moreover, cost is only partly driven by the physical properties of the product: supply and
demand mechanisms are always applicable and not necessarily related to complexity.
Additionally, data regarding equipment cost is usually highly classified and dispersed over
different departments of a company (in this case a complete shipyard, which, as a systems
integrator for large projects, can be expected to consist of more departments than an average
engineering company). It is hence highly uncertain whether data regarding equipment cost
could be gathered in time if at all.
Consequently, in this research, it is chosen to unify the concepts of scale and complexity in a
single complexity factor, actually expressing total work content. Although this approach may be
Quantification of Manufacturing Complexity in Shipbuilding Projects page 26/153
semantically disputable, it does deliver a result which is more mathematically correct, allows
direct comparison of projects and can be easily applied in cost estimation and project planning.
From within the DSNS project department, it was coined that a suitable metric for complexity
would be a figure in man hours per ton, as is used currently for yard planning. Such a metric
would compensate for the fact that large projects entail more work, which is considered self-
evident.
Experience has shown however that this way of expressing complexity cannot be consistently
applied: man hour per ton values for outfitting, usually the more complex part of the vessel, can
be very high for ships which can be expected to be not especially complex (as, for example, for
the PSV 5000, as will be discussed in section 7.4.3).
From an academic point of view, this way of expressing complexity is subject to discussion as
tonnage is in fact a variable depending on construction properties such the material used; if, for
instance, a vessel is built out of lighter material with similar construction properties, leading to
similar construction work content, the man hour per ton figure will rise due to the lighter weight,
leading to the misleading conclusion that the project be more complex. For this reason, it is
chosen to express complexity in man hours and ratios between these values for different
projects, rather than by dividing work content by weight of hull and outfitting.
The complexity quantification is primarily based on regression analysis of load balances and
simplified construction models on one hand, and historical production man hour data for the
PKR1 project on the other hand. These analyses and models will be described in Appendix J
and Appendix K.
It can be expected that certain systems and components have a much greater impact on
production man hours than others; it is therefore of interest to assess the main systems in
consultation with the different disciplines within the engineering and project departments. To
make the quantification model valid also if available input data are limited, the key factors to
Quantification of Manufacturing Complexity in Shipbuilding Projects page 27/153
In multiple occasions, it was pointed out that the amount of systems (and hence components)
greatly affected work content. This may not be a surprise; it is however important to keep this in
mind when defining a complexity quantification model based on a limited number of key
systems and components.
According to supervisors in the project and planning department, one can say that as the
system density of a room doubles, the man hours required for installation of major components
quadruples; in other words, a square relation between system density and work content is to be
expected.
As no detailed literature or data allowing thorough analysis of this aspect is available, personal
assessment plays an important role in quantifying relations resulting from differentiation in the
course of this research, rather than regression analysis. Mostly, the square relation mentioned
before will be maintained. Validation of this assumption can only be established in future
research which concentrates on this aspect of complexity.
Often, man hours required for the production of a ship are estimation based on the weight of
the ship. Through experience, this has proven to be an all too crude approximation as the
production man hours per ton very greatly in function of the hull geometry and girder
arrangement (Brown & Barentine, 1996). For example, if a strict weight limitation is imposed on
the vessel, thinner skin plates may need to be used in order to keep the vessel weight below
this limit. This means however that the amount of stiffeners is to be increased in order to
preserve sufficient bending resistance. The additional structural members need to be placed
Quantification of Manufacturing Complexity in Shipbuilding Projects page 28/153
and welded, increasing the man hours involved, while the fact the plate thickness is reduced
does not cause a significant reduction in production hours. Moreover, thinner plates are more
likely to distort as a result of heat absorption and dissipation during welding, which may raise
the need for additional straightening.
Another important issue is the fraction of the hull which consists of straight plates that can be
readily machined. In the 1996 paper by Brown and Barentine, a distinction is made between
“producible” (straight) and “non-producible” (curved) plates, the latter of which require
considerably more man hours to reach their final geometry. In the paper, it is concisely
demonstrated how the hull of an Arleigh Burke-class destroyer can be divided in these two
kinds of plates, and how the design can be altered to increase the fraction of producible plates.
It was also stated that a trade-off between “producibility” and seakeeping and Radar Cross
Section (RCS) could occur if fundamental changes to hull design are made in order to simplify
skin plate production; the underlying considerations for these design aspects are not part of the
scope of this research but are nonetheless closely related to the results of a complexity
estimation model.
Factors expressing the ratio between producible and non-producible skin plates may be useful
for determining a factor for construction complexity of the hull.
In other words, a function is to be defined which allows extrapolation of the production hours
not only based on casco weight, but also on hull form, frame spacing and average plate
thickness.
Installation of machinery accounts for a considerable part of the total work content connected to
ship construction. An obvious reason for this is the fact that component sizes are comparably
large (for instance main diesel engines): moving in a heavy item requires more tools, guidance
and preparations than smaller outfitting parts. Size and weight however cannot be considered
as the determining factors when it comes to work content: the amount of work hours is much
more closely related to the number of main components, as this number is a far better indicator
of the number of fittings and ancillary equipment connected to these components.
For example, a direct diesel drive consisting of two 10 MW engines is much simpler than a
drive making use of four 5 MW engines: hoisting and placing of the engines will not be
fundamentally different, while each engine requires fresh water cooling, lubricating oil and fuel
Quantification of Manufacturing Complexity in Shipbuilding Projects page 29/153
supply connections, along with associated pipe spools, secondary components such as
governors, power take-off connections, fittings and cabling.
A considerable part of man hours in production is spent on mounting of pipes. Usually, these
are divided in pipe spools: loose pieces of pipe which can be prefabricated and subsequently
joined in the ship. As pipe spools are logically secondary components (according to the
definition given in section 3.1.4), their number depends on the amount of primary components
installed on board. There is however also a relation with the number of rooms and the length of
the vessel: more rooms imply more wall and deck penetrations, increasing the amount of
individual pipe spools, while larger ships entail larger distances between some components,
which too may have an increasing effect on the number of pipe spools.
The amount of data and power cables logically depends on the number of systems, the amount
of inputs and outputs per system (which is closely related to the degree of automation), the size
of the ship, the number of separate networks and circuits (also including, for instance,
degaussing coils) and the degree of redundancy of the communication systems. The way in
which the communication lines and sockets of these individual components are designed is an
important factor as well. Modern diesel engines for instance may have many more data outputs
than their older counterparts; due to more advanced electronic components however, these
signals can be bundled in single wires, whereas older engines have separate connections per
output, complicating installation.
For several reasons, modular building of ships can be interesting or even required. A division of
a newbuilding into largely independent modules allows for much more flexible planning and
parallel execution of jobs, assigning of work to local yards specialised in certain jobs, and
meeting requirements regarding local input as is often the case when building for governments.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 30/153
Division of a ship in modules which then are constructed separately decreases organisational
complexity: each module can be considered a project on its own, containing much less systems
than does the entire ship. However, certain additional tasks will have to be performed in order
to successfully integrate all modules into one vessel, especially if the modules are built in yards
far away from each other. This increases technological complexity to some extent.
A prime example of such a strategy is presented by the Sigma class combatants, which were
designed in such a way that main components such as the power plant, navigation and combat
control rooms, aft ship, bow compartment, helicopter deck and funnel can be built separately
and later joined with relative ease. The modules are finished almost completely, also in terms of
outfitting, before being joined; this entails some additional work.
All in all, analysis of historical data shows that the influence on man hours is fairly limited, as
can be concluded from the data given in Appendix M; this subject will not be covered in detail.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 31/153
The breakdown is made from a marine engineering perspective; in this chapter, the first
principle approach of the research becomes apparent. As no literature, describing relations
between shipbuilding strategy and technical specifications in detail, was found, the
categorisation made in this chapter is based on personal judgement and experience in the field
of marine engineering, as well as input from the project and engineering departments at DSNS.
Table 1 Division of systems into categories based on complexity (Hubka & Eder, 1988)
For this research, a categorisation based on the one devised by Hubka and Eder was applied
on shipbuilding projects; the divisions used for the estimation model are given in sections 3.1.1
through 3.1.4.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 32/153
More detailed information on technical system hierarchy in shipbuilding was found in recent
literature and elaborated to provide substantiation for the system breakdown. The 2012 paper
by Gaspar et al. specifically covered marine newbuildings and the way in which they can be
divided into subsystems and components (Gaspar, Rhodes, Ross, & Erikstad, 2012).
An important remark on the following subdivision is that the terms systems and components
must be understood in a broad sense, and the units of these characteristics can be user-
defined as to make the model sufficiently flexible to handle varying degrees of benchmark
detailedness. Components for instance do not need to represent physical items but can also
refer to important characteristics of a system, such as total diesel-electric propulsion power.
The concept of main systems is based on the functions that are to be fulfilled by the vessel, for
example propulsion, electric power generation, navigation, surface warfare and support tasks.
In many aspects, these main tasks depend on each other: navigation requires electric power.
These interdependencies are taken into account by evaluation of the components in lower
levels.
In some cases, it can be argued whether or not different systems are part of the same main
system. For instance, surface-to-surface combat may be covered by both a deck gun and
installations for launching of self-propelled ordnance. If such systems fulfil very similar functions
but are technically independent, they can be considered individual main systems. Note that for
the demonstrative model developed in the context of this research, the categorisation of the
Damen SWBS structure (as given in Appendix M) is maintained; main systems are defined by
this structure to increase applicability of the model for the DSNS project department.
Secondly, a division is made into subsystems: autonomous groups of installations which, often
in combination with other subsystems, are essential for the operation of a main system, or
present the main system with an alternative strategy to fulfil its function (this could be
expressed in AND/OR relations).
On this level, exact requirements are also to be made (e.g. speed at direct-diesel drive,
auxiliary power redundancy factor, complement, armament). These requirements, taking into
Quantification of Manufacturing Complexity in Shipbuilding Projects page 33/153
account relations between (sub) systems, influence the specifications of components at lower
levels.
For instance, subsystems of the propulsion main system may be the diesel-direct drive
subsystem, shaft and screw assembly, and diesel-electric drive subsystem.
In some cases, it can be argued whether a certain system can be classified as a main system
or subsystem. It must be kept in mind that this classification is merely devised to facilitate
component dimensions and numbers and as such is subject to personal judgement.
If a subsystem is out of operation, the parent main system loses at least one of the strategies to
fulfil its function.
The next level consists of primary components. These perform the main tasks necessary for
the subsystem to remain in operation.
Primary components can be made redundant: if one component fails, another takes over to
fulfil its task, so the parent subsystem remains fully operational. Here lies the difference
between primary components and subsystems: the latter always have an impact on
functionality of their parent main system in case of failure.
Finally, the category of secondary components contains the components which are required the
primary components to fulfil their task on short and long term. Secondary components can be
redundant, allowing primary components to function after failure of a secondary component.
Often, these components are smaller and more flexible that the main components.
Definition of the four categories in the foregoing sections allows for identification of
interdependencies. Another function is to be integrated in the mathematical framework in order
to take into account differentiation, which is the result of the number of components in a certain
room rather than relations between these components.
For this purpose, differentiation groups can be defined, which compare the dimensions of a
group of components to the space in which they are grouped. For instance, if a ship is
equipped with sensors for marine research and a control room to monitor these sensors, a
differentiation group for this control room can be defined, containing relevant components. If
more of these components are installed, the ratio component dimensions to control room
volume will increase and hence, the differentiation factor, ultimately influencing project
complexity.
In Appendix I, a system breakdown of the PKR1 project is given. As a result of limited data
concerning work content, the breakdown is limited to just over 100 components connected to
subsystems. The system breakdown given in this appendix will form the basis for
demonstrative calculations in this research.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 35/153
Models for complexity estimation would primarily be of use during the pre-contract phase of a
project, allowing more accurate prediction of man hours connected to a certain project prior to
making a detailed production plan. At this stage, detailed engineering is yet to follow and only
basic data concerning systems and their specifications are available. The estimation model
must thus be developed in a way that it can deliver sound estimates of complexity with only
limited data.
A possible approach is to firstly investigate which data are normally available during the pre-
contract phase and subsequently base the estimation model on these data.
In order to allow for some freedom during the research however, this limitation is not set. The
model will be devised in such a way that it is based on the vessel’s main functions and overall
system requirements, relating these input requirements to expected technical specifications. It
is looked into how detailed the provided input data need to be in order to yield relevant results.
In the past, the issue arose that man hours for DSNS projects were not consistently recorded
due to the multitude of task categories, which ranged up to 15,000 in number. Unofficial
estimates state that only 65% of the hours were correctly booked. For this reason, the past
decennium saw introduction of a new, much simpler man hour registration method which was
based on only about a tenth of the original number of categories. The result was that work
content is now kept in a less detailed but much more accurate way: accuracy is expected to
have risen to above a 90% level.
Two different categories of methods for processing input data into complexity figures are
considered, both of which have their specific benefits and drawbacks.
Absolute estimation methods allow direct calculation of technical complexity of a ship, even if
the specifications are completely different from any other vessel evaluated so far. As estimates
are based on component specifications and man hour factors, and the relations between these
characteristics, it is required that all components of the evaluated project are correctly listed
and man hour factors are known and reliable.
Relative estimation methods on the other hand use data from ships built in the past to deduct
complexity factors based on the different requirements and specifications of the evaluated
vessel.
There are two notable limitations to relative methods; first of all, there is the need for a
predefined benchmark that contains at least basic specifications of the equipment installed on
the evaluated project. If this is not the case, extrapolation of systems not present in the
benchmark would, through mathematical logic, result in factors ranging into infinity which is not
quite a helpful figure in project planning. Consequently, systems present in the evaluated
project but not included in the benchmark must still be defined component per component, as is
the case in the absolute estimation method.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 37/153
Secondly, by applying an extrapolation on the specifications and man hours, it is assumed that
the building strategies and work forces are comparable. In other words, organisational
complexity is not taken into account.
The advantages of relative estimation methods are multiple. First of all, inaccuracies as
mentioned in the previous section are of minor importance since the result of the estimation
model focuses on factorial indicators for complexity, based on physical relations. This means
that, even if an input value is not entirely accurate, the factor resulting from the model is still
valid as the relations still hold.
As the evaluation is linked to a specific benchmark, which is more likely to be based on man
hour data originating from a limited number of projects as compared to a loose set of factors as
used in the absolute estimation method, uniformity of the input data and hence of the resulting
estimation is superior.
Moreover, as a result of the fact that this method is based on extrapolation, definition of all
components to the highest level of detail is not required, as long as all relevant systems are
adequately covered. This makes the relative estimation method very suited for calculations
during the pre-contract phase; an important remark in the context of this particular research.
Mission profiles and comparable parameters are also automatically taken into account.
Additionally, using ratios between requirements to estimate new system and component sizes
eliminates the need for intermediate factors such as density conversion factors, further
simplifying calculations without sacrificing accuracy.
Finally, a relative estimation method would still be based on man hours, and results in
complexity factors per system which allow estimation of absolute figures in terms of man hours,
just as the absolute estimation method would.
In conclusion, it can be said that the relative estimation method is more versatile as it can
predict complexity per main system based on a more limited set of known specifications,
handles inaccuracies in historical data with more resilience, and more directly produces
complexity factors that allow extrapolation of absolute man hour data.
The principles of the prediction model developed in this research will hence be based on the
relative estimation method.
As stated in section 4.1.3, the estimation method is devised in such a way that newbuilding
projects can be evaluated by comparing them to benchmark projects and extrapolating
component dimensions.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 38/153
A consequence of the fact that the complexity estimation method is based on extrapolation and
comparison with benchmarks is that for proper evaluation, a benchmark vessel with similar
functions must be defined, with all possibly applicable functional requirements and the
equipment necessary to fulfil these requirements. It is not necessary that a ship with the exact
specifications actually exists (in fact, it is not likely that a vessel equipped with all possible
machinery for a certain class does exist); it must however be possible to make an accurate
profile of man hours for the said vessel based on historical project data of actual vessels. For
example, complexity of a surface combatant cannot be evaluated using a bulk carrier as a
benchmark.
The benchmark vessels are to contain an as wide as possible range of basic equipment, in
order to cover all possible variations in their respective ship class.
Subsequently, it is evaluated how certain systems are extended or scaled down for individually
evaluated newbuilding projects, based on the functional requirement of these new projects. The
rate with which equipment needs to be altered is based on a technical systems analysis of
previous projects and marine engineering in general, and as such, a substantiated estimation of
the sizes of required systems can be made before detailed engineering of the project is carried
out.
The next step is to investigate the relations between sizes, types and amounts of installed
equipment on one side, and required man hours to install this equipment on the other. These
relations are deducted and quantified by comparison of previous projects and by discussion
with engineers experienced with the subject. After such relations have been determined, man
hours connected to the estimated equipment for a newly evaluated project can be estimated.
As it was chosen to consider work in terms of man hours as the main indicator of complexity in
this research, these steps allow assigning complexity factors to each system of any evaluated
vessel, ranging from 0 (if the system is not present) to a very high number (requiring an
estimated proportional amount of man hours for an extended installation, compared to a
chosen benchmark).
Comparison of projects is not limited to the template vessels put forward during this research.
To a certain extent, a single, previous project can be used as a standard with which a new
project can be compared. It must be kept in mind however that all main systems as well as
most of the components present on the new, evaluated vessel must also be present (at least as
an option) on the project which is taken as a benchmark in order to allow a complete
Quantification of Manufacturing Complexity in Shipbuilding Projects page 39/153
evaluation. If this is not the case, the benchmark or relation factors can be adapted (as will be
demonstrated in section 4.3.1.2), or the systems influenced by the components absent can be
omitted from the evaluation.
4.3.1. Interdependency
For the technical analysis, two main types of relations are considered: vertical and horizontal
interdependency. These dependencies were described in section 2.1.2 and are used to deduct
factors which relate requirements set on the vessel’s functionality to dimensions of the installed
components.
Where:
fi = [-] interdependency factor
da = [kW, kg, t, …] dimensions of component in reference project
db = [kW, kg, t, …] dimensions of component in evaluated project
Quantification of Manufacturing Complexity in Shipbuilding Projects page 40/153
As was stated in 2.1.2, interdependency is divided into horizontal and vertical dependencies.
This means that fi needs to comprise two types of relations, which will be covered in the
following two sections.
Where:
da = [kW, kg, t, …] dimensions of component in reference project
db = [kW, kg, t, …] dimensions of component in evaluated project
l1 = [-] vertical dependency constant
ra = [kn, complement, …] system requirements for reference project
rb = [kn, complement, …] system requirements for evaluated project
e1 = [-] vertical dependency dimension exponent
Units used for dimensions da and db can vary in function of the described component.
Examples of applicable units are [kW], [m3] and [kg]. The same goes for the system
requirements: these can be expressed in knots design speed, complement numbers et cetera,
in function of the requirement. The dependency constant allows assigning a linear factor to the
relation; the dependency exponent is a dimensionless number, determining whether there is a
square, cubic, linear or other relation between these characteristics.
Often, relations involve multiple subsystems, such as the Admiralty power law for propulsion
power prediction which is based on a multiplication of design displacement and design speed.
To take this into account, Equation 2 can be extended:
Quantification of Manufacturing Complexity in Shipbuilding Projects page 41/153
Where:
da = [kW, kg, t, …] dimensions of component in reference project
db = [kW, kg, t, …] dimensions of component in evaluated project
l1 = [-] vertical dependency constant
ra = [kn, complement, …] system requirements for reference project
rb = [kn, complement, …] system requirements for evaluated project
e1 = [-] vertical dependency dimension exponent
g = [-] index for subsystems of influence on vertical dependency
The expression for horizontal dependency is more complicated than that for vertical
dependency as it is devised to also capture virtual components such as frame spacing. The
general format for horizontal dependency is the following:
Where:
da = [kW, kg, t, …] dimensions of influenced components in reference project
db = [kW, kg, t, …] dimensions of influenced components in evaluated project
l2 = [-] horizontal dependency constant
dc = [kW, kg, t, …] dimensions of influencing components in reference project
dd = [kW, kg, t, …] dimensions of influencing components in evaluated project
e2 = [-] horizontal dependency exponent
Quantification of Manufacturing Complexity in Shipbuilding Projects page 42/153
This formula is valid if a component is influenced by only one other component. It can however
be foreseen that this will not always be the case and that the formula for horizontal
interdependency must be adapted to influence by multiple components, each with their own
weight on top of different proportional and exponential factors. The resulting formula to
determine horizontal dependency is given in Equation 5.
Where:
da = [kW, kg, t, …] dimensions of influenced components in reference project
db = [kW, kg, t, …] dimensions of influenced components in evaluated project
s = [-] coefficient of effect (“weight factor”)
l2 = [-] horizontal dependency constant
dc = [kW, kg, t, …] dimensions of influencing components in reference project
dd = [kW, kg, t, …] dimensions of influencing components in evaluated project
p = [-] index for components of influence on horizontal dependency
q = [-] number of components of influence on horizontal dependency
e2 = [-] horizontal dependency exponent
As can be seen, a coefficient sp is added. This factor can be used to assign different weights to
component ratios; for instance, doubling the main engine power has more influence on cooling
water heat exchange requirements than doubling the auxiliary engine power, while these
engines usually are linked to the same (sea water) cooling system. The value for sp for main
engines of 20 MW and auxiliary engines of 4 MW could hence be 5 and 1 respectively, if the
input for the heat exchanger extrapolation is supplied as the total number of engines and
assuming that the required waste heat exchange rate has a linear relation with power.
The combination of components with different dimensions is another situation in which weight
factors are required to yield a mathematically and physically correct equation. In such a case,
the desired weight factors must be multiplied with a value of d-e per individual component in
Quantification of Manufacturing Complexity in Shipbuilding Projects page 43/153
order to maintain equal component dimensions (in other words, all dimensions corrected to 1)
or with a value of d-e+1 in order to maintain the original component dimensions.
As can be concluded from the above elaboration, some feeling with mathematics is required for
correctly defining interdependency factors.
Similar remarks can be made with respect to units for Equation 5 as for Equation 2. Moreover,
units for dimensions of all components in Equation 5 must be the same (for instance [kW] for
summations of prime mover power) in order to obtain a mathematically sound expression. This
requirement can be circumvented by defining separate horizontal dependency factors for
groups of components with different dimension units, and then multiplying these factors to
obtain the final horizontal dependency factor for the evaluated component; such a solution is
applied in the demonstrative complexity quantification model developed for DSNS.
Interdependency constant l2 can be useful when considering components which are not present
in the benchmark. In the relation analysis done in the context of this research, bow thruster
power had to be defined without these components being present on the benchmark, the
Sigma 10514. For these components, l2 was set at 500 (kW) with the corresponding value for
dd indicating the number of bow thrusters, as is shown in Table 13. Note from Equation 5 that
this approach only works for components not included in the benchmark.
To obtain a factor which describes both horizontal and vertical dependency, the ratios
mentioned in sections 4.3.1.1 and 4.3.1.2 must be combined in a single interdependency factor.
The following assumption is made:
Variables and units in Equation 6 are the same as those in Equations 3 and 5.
As a result of the way in which these factors are devised, it can be reasonably assumed that
vertical dependencies, expressed in requirements, will further increase the effect of horizontal
Quantification of Manufacturing Complexity in Shipbuilding Projects page 44/153
Consequently, ratio db to da can be written as follows: (note that in Equation 1, fi was defined as
the relation db to da)
Equation 7 Vertical and horizontal dependency ratios per component, written out
Where:
da = [kW, kg, t, …] dimensions of influenced component in reference project
db = [kW, kg, t, …] dimensions of influenced component in evaluated project
l1 = [-] vertical dependency constant
ra = [kn, complement, …] system requirements for reference project
rb = [kn, complement, …] system requirements for evaluated project
g = [-] index for subsystems of influence on vertical dependency
e1 = [-] vertical dependency dimension exponent
s = [-] coefficient of effect
l2 = [-] horizontal dependency constant
dc = [kW, kg, t, …] dimensions of influencing component in reference project
dd = [kW, kg, t, …] dimensions of influencing component in evaluated project
p = [-] index for components of influence on horizontal dependency
q = [-] number of components of influence on horizontal dependency
e2 = [-] horizontal dependency exponent
4.3.2. Differentiation
density: the amount of systems and their volume per unit available space on board, an aspect
explicitly remarked by DSNS engineers as was mentioned in section 2.2.2.
Per differentiation group (usually referring to a room), a degree of system density will be
calculated and expressed in differentiation factor fd. This factor takes into account ratios
between room volumes and available space for cables and pipes, and ratios between
component dimensions. One differentiation factor is calculated per room and is applicable for
the components installed in that room. The ratio that will be looked for in this respect is the
following:
Where:
fd = [-] differentiation factor
Va = [m3, m] available space in room of reference ship
Vb = [m3, m] available space in room of evaluated ship
Db = [kW, kg, t, …] dimensions of equipment in evaluated ship
Da = [kW, kg, t, …] dimensions of equipment in reference ship
As can be seen, fd is inversely related to the ratio Vb to Va, and proportionally related to the ratio
Db to Da. These ratios are chosen as complexity is expected to increase with decreasing
available space and increasing component dimensions.
Another issue, which is hard to estimate but is mentioned here for the sake of completeness as
it was also listed in section 2.1.3, is experience with the installed technology and
subcontractors involved. As this aspect is very dependent on personal judgement and is not
easy to quantify due to inherent uncertainty (which is not part of the scope of this research), no
factor is assigned to account for this particular aspect.
Where:
fd = [-] differentiation factor for component
da = [kW, kg, t, …] dimensions of influenced component in reference project
db = [kW, kg, t, …] dimensions of influenced component in evaluated project
Va = [m3, m, …] available space for component in reference project
Vb = [m3, m, …] available space for component in evaluated project
i = [-] index for components present in room
p = [-] number of components present in room
Same as in Equation 5, dimensions for all values of da and db must be the same. For variables
Va and Vb, units of length (most notably m3) are used.
The concept of differentiation can be extended to describe any relation where the dimensions
of a group of components, all part of the same structure or system, determine the amount of
man hours that need to be accounted for per component. For example, requirements for a
lighter casco while maintaining the amount of systems on board makes for much more complex
stiffening panels, which in turn increases man hours per ton connected to casco building.
Summarising, it can be said that the interdependency factor quantifies influence on a certain
component dimension by other components and systems, while the differentiation factor
quantifies mainly spatial limitations imposed on said component.
The final step to be made in order to quantify complexity is to connect these specifications per
component to man hours, which were chosen as primary indications for complexity in section
2.1.5.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 47/153
Papers published on the subject of work content and CGT factors indicate that the relation
between differentiation and associated man hours may often be linear, although at a rate which
depends on the ship type (Craggs, Bloor, Tanner, & Bullen, 2003).
As the relation between spatial constraints and man hours is often not linear and can vary in
function of the component, as became apparent during discussions with engineers involved
with planning and production, an additional exponential factor is introduced to account for non-
linear influences.
Hence:
Where:
fc = [-] complexity ratio for component
c1 = [-] component dimensions – man hours constant
fi = [-] interdependency factor for component
c2 = [-] differentiation – man hours constant
fd = [-] differentiation factor for component
e3 = [-] differentiation – man hours exponent
Other variables and their units are the same as in Equation 1 through 9.
The use of power factors based on engineering judgement in Equations 10 and 11 is in line
with earlier proposals in the field of complexity quantification such as those put forward by
Hubka and Eder in their book Practical Studies in Systematic Design (Hubka & Eder, 1988) and
by Valerdi in his Constructive Systems Engineering Cost Model (COSYSMO) (Valerdi, 2005);
although these factors, when not directly resulting from physical properties such as power or
Quantification of Manufacturing Complexity in Shipbuilding Projects page 48/153
weight, are often estimated on an arbitrary basis, no accurate prediction model can be
developed without taking into account non-linear effects.
It can be noted that component dimensions are taken into account twice in the estimation
method described in Equations 10 and 11: both in factors fi and fd, the ratio db to da is taken into
account.
The reason for this is the fact that not only the increase of the size of a component itself is likely
to increase complexity (as a result of, for instance, more fittings and connections), but also the
increase in size and amount of adjacent components and in relation to available space is of
importance (due to increased system density and resulting lack of space during installing of the
systems). It is useful to define these aspects mathematically in an independent way, so their
influence on estimated complexity can be altered as such; it cannot be assumed that for each
component, absolute size and system density influence complexity in the same way.
Complexity ratio fc expresses the ratio of man hours between the two compared projects and
can thus also be written as follows:
Where:
fc = [-] complexity ratio for component
MHb = [hrs] estimated man hours for installation of evaluated component
MHa = [hrs] man hours for installation of reference component
The aforementioned complexity ratio is calculated per component. Such a level of detail is not
particularly suited if the goal is to provide an overview of overall complexity of a newbuilding
project and complexity “hotspots”. Hence, these ratios are combined into complexity ratios per
subsystem, main system and for the entire project; levels that were defined in section 3.1.
These combined complexity factors are calculated by adding and dividing the summations of
estimated and reference man hours.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 49/153
Where:
fcc = [-] combined complexity ratio for higher level
k = [-] index for lower level components, subsystems or main systems
n = [-] numbers of lower level components, subsystems or main systems
MHb = [hrs] estimated man hours for installation of evaluated component
MHa = [hrs] man hours for installation of reference component
These combined complexity ratios are the envisioned result of the evaluation: a mathematical
comparison on different technical levels of shipbuilding projects in terms of construction and
system integration complexity.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 50/153
To substantiate the factors given in Appendix J, and to provide guidance for definition of future
benchmark projects and relations, the following sections give a description of the method used
to determine said factors. As the most extensive derivations were made for the hull structure,
due to the abundance of knowledge on this subject on the DSNS production floor, this is the
subject on which most attention will be paid in this chapter. Additionally, an account is given of
the reasoning behind the extrapolation factors for main and auxiliary machinery and other
outfitting.
The prediction models and methods for determination of the extrapolation factors will not be
described completely, as this would lead to an unnecessary amount of additional, not
particularly valuable sections filled with formulas; an elaboration is made only insofar as
required to provide insight in the method used to determine interdependency exponents and
constants sp, l1, l2, e1 and e2.
The approximation models can be assessed in full detail in the source calculation files, which
are property of DSNS.
5.1. Model Used to Estimate System and Component Relations for Hull Structure
Figure 1 shows the four main parts of the casco estimation model.
4. Helicopter hangar
As the benchmark is defined using PKR1 data, input variables for the casco estimation model
are those for the Sigma 10514 surface combatant.
Dimension Unit
Vessel dimensions
Longitudinal compartments 13 #
Transverse compartments 3 #
Length 105 m
Beam 14 m
Quantification of Manufacturing Complexity in Shipbuilding Projects page 52/153
Draught 3.7 m
Height lower hull 10 m
Height accommodation 6 m
Length accommodation 50 m
Deck height 2.75 m
Heli hangar 1 applicability
Stiffeners
Frame distance longitudinal 0.6 m
Transverse frame depth 0.2 m
Transverse frame t 0.007 m
Frame spacing transverse 2 m
Longitudinal frame depth 0.2 m
Longitudinal frame t 0.007 m
Deck stiffener spacing longitudinal 1.8 m
Deck stiffener spacing transverse 0.5 m
Deck stiffener depth 0.04 m
Deck stiffener t 0.006 m
Plate properties
Average plate thickness 0.006 m
Av. large plate t 0.007 m
Large plate average side 3 m
Large plate average area 9 m2
Av. internal plate t 0.005 m
Internal plate average side 1.5 m
Internal plate average area 2.25 m2
Table 2 Input variables for casco estimation model, based on Sigma 10514
Quantification of Manufacturing Complexity in Shipbuilding Projects page 53/153
To illustrate how this model is used to mathematically define relations, it is described how the
extrapolation factors connected to steel work for the large structural members are deducted.
The number of longitudinal and transverse frames is estimated by the following set of formulas:
Where:
Ft,s = [-] number of transverse frames in superstructure and hull below
AFs = [m2] area of transverse frames in the superstructure and hull below
Ft,h = [-] number of transverse frames in hull on which no superstructure is built
AFth = [m2] area of transverse frames in hull on which no superstructure is built
Fl = [-] number of longitudinal frames
AFl = [m2] area of longitudinal frames
Ftotal = [-] total number of frames
AFtotal = [m2] total frame area
ls = [m] length of superstructure (without helicopter hangar)
ahh = [-] applicability of helicopter hangar
dt = [m] transverse frame spacing
hFt = [m] height of transverse frames
B = [m] ship beam
hs = [m] height of superstructure
hh = [m] height of lower hull
LOA = [m] ship length
dl = [m] longitudinal frame spacing
hFl = [m] height of longitudinal frames
For work content estimation, the assumption is made that steel work for large structural
members is proportional to the frame area (note that steel work excludes welding, which is
estimated by a different set of extrapolation factors). To identify which parameters influence
frame area, input parameters mentioned at the beginning of section 5.1 are varied
independently, and output values are recorded in function of each input parameter. From this, it
becomes apparent that frame area is affected by the following parameters:
Quantification of Manufacturing Complexity in Shipbuilding Projects page 55/153
Frame spacing
Ship length
Ship beam
Number of decks in hull
Number of decks in superstructure
Length of superstructure
Deck height
The recorded values are plotted, which allows curve fits to be derived. As an example, one of
these plots with a curve fit generated by Microsoft Excel is shown in Figure 2.
2,9
Total frame area
1,9
1,4
0,9
Note that, as the mathematical framework for the estimation method is based on requirement
and component ratios, all values are normalised through division by the corresponding values
for the reference project.
This plot and curve fit delivers the extrapolation factors as required by the mathematical
framework.
An issue that occurs is that, at an input parameter ratio of 1, the result of the regression
equations is not 1 due to the regression constant which is slightly more than 1; from Figure 3, it
can be concluded that this situation occurs for most of the curve fits.
These deviations are unacceptable as they imply that an exact copy of the benchmark vessel
would be considered more complex than the benchmark. For this reason, these constants are
corrected to a value of 1, which sometimes required minor adjustments to the exponent to
retain a sufficiently accurate fit with the model results.
In Table 10, it can be seen that frame spacing is considered as a subsystem, due to the many
components which depend on it. Steel work for large structural members is defined as a (non-
physical) component, which makes that the relation between frame spacing and steel work for
large structural members is labelled as vertical dependency. According to Equation 3, the
exponential factors which are to be derived are l1 and e1.
l1 = 1
e1 = -0.931 according to Microsoft Excel curve fit
= -0.951 after correction for l1 (see above)
Equation 22 Vertical dependency: frame spacing and steel work large structural members
Where:
SW e = [hrs] work content steel work large structural members, evaluated project
SW r = [hrs] work content steel work large structural members, reference project
dt,e = [m] transverse frame spacing, evaluated project
Quantification of Manufacturing Complexity in Shipbuilding Projects page 57/153
Note that frame area and hence steel work for large structural members depends on more
factors than transverse frame spacing only. The combined vertical dependencies for this
component, expressed in the same fashion as demonstrated by Equation 22, can be multiplied
with each other, as was also implied in Equation 3.
To assess the validity of the model, a comparison between the generated estimation for steel
weight of the PKR1 project and the DSNS estimation were made. The estimation by the
simplified model is 1007 tons, while the DSNS total estimate is 1134 tons. In other words, the
model yielded a result 12.6% below the Damen estimate. It must be noted though that the
DSNS estimate includes small steel work for outfitting and an additional 5% margin for weld
weight, making that the actual deviation is in fact less than 7.2%.
Based on this observation, the conclusion is drawn that the simplified estimation model is of
sufficient accuracy to allow deduction of extrapolation factors for similar newbuildings.
Another important remark is that the model parameters as used for extrapolation, and hence
the deducted interdependency factors, could be expected to be particularly suited for the
evaluation of surface combatants. From the results in chapter 7 however, it can be concluded
that the model is equally suited to predict dimensions of platform supply vessels, indicating that
extrapolation factors generated by the model described here are applicable for a wider range of
vessels.
The model allows identification of the relations between components listed in Appendix I, based
on physical properties. These relations are expressed in the form of power equations with the
format as proposed in section 4.3, which allows direct implementation in the applied
quantification model.
In Figure 3, it can be seen that, as the component ratio moves further away from value 1, slight
deviations between the estimated interdependency factors and the regression curves become
apparent. In general, it can be said that the regression curves deliver accurate values for
component dimension ratios between ⅓ and 3, which allows comparison of vessels with
fundamentally different dimensions.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 58/153
For a detailed insight in the way the regression analysis was done, reference is made to the
relevant source files; in order to keep the report accessible, it was chosen not to enclose charts
of all analyses as appendices.
2
Steel weight
1,8
Weld seams
1,6 Plates
Total stiffeners
1,2
Weld seams = 1.0296x-0.286
1
Total frames = 1.0033x-0.96
5.2. Estimating System and Component Relations for Machinery and Outfitting
In this section, only the extrapolation factors for main propulsion engines and power grid
cabling are covered. These two items are considered to be representative as they comprise
both vertical and horizontal interdependencies, also making use of weight factors.
For the demonstrative model, relations between machinery too were mostly derived by
approximation models such as for instance load balances, supplemented by technical data
available for the PKR1 project. Certain relations between high-level system requirements and
primary components however can be easily quantified with some basic knowledge of marine
engineering. The relations between design speeds and propulsive power for instance can be
approximated by the Admiralty constant (Klein Woud & Stapersma, 2002):
Where:
Cadm = [t2/3 kn3/kW] Admiralty constant
= [t] displacement
vs = [kn] ship speed
PD = [kW] power delivered to propulsors
In the proposed set of interdependency factors, Cadm is assumed to be the same for the
benchmark and evaluated ship. This assumption is valid if the hull form is similar and design
speed and displacement are in the same order of magnitude. Comparison between, for
instance, a surface combatant such as the PKR1 project and a platform supply vessel with
double the design displacement and half the design speed may yield inaccurate power
predictions; in the demonstrative model however, these predictions can be overruled by manual
input.
Vertical interdependency between subsystem Main propulsion design speed and component
Main engine power:
l1 = 1
e1 = 3
Where:
PPe = [kW] propulsive power, evaluated project
PPr = [kW] propulsive power, reference project
ve = [kn] design speed, evaluated project
vr = [kn] design speed, reference project
s = 1
l2 = 1
e2 = ⅔
Where:
PPe = [kW] propulsive power, evaluated project
PPr = [kW] propulsive power, reference project
∆e = [t] design displacement, evaluated project
∆r = [t] design displacement, reference project
Cable lengths and hence work in general is considered to be dependent on vessel dimensions
comprising but not limited to length and beam, and the amount (or dimensions) of components.
Being substantiated to a much lesser extent than hull structure and power plant components,
as a result of more limited knowledge on technical particulars, this category qualifies for fine-
tuning efforts based on more detailed technical data from past projects, especially considering
the fact that it represents a significant part of the production hours spent on the electric plant.
The distribution of cabling between systems and components is derived from DSNS registered
data insofar as this was possible (for instance, this was possible for generator cabling) and
approximated by a model which delivers a weight factor sp per component (referring to the
cable length associated with the respective component relative to total cable length), as
proposed in Equation 5.
The model on which the weight factors were based took into account vessel dimensions
(length, beam, deck height, number of decks etc.), position of several key components
connected to the power grid (resulting in a fraction of the ship length and height over which the
relevant power cables extend) and redundancy and amount of these components (yielding the
number of passes the cables make through each deck and compartment). The intention was to
estimate the proportion of power grid cabling per category of components relative to total power
grid cabling rather than to predict total power grid cable lengths.
Input data concerning vessel dimensions for the power grid estimation model were based on
the Sigma 10514 as the benchmark too refers to this design; these data can be found in Table
2. Other input data, such as number of decks passed, fraction of the ship’s length to be covered
by power cabling per component and number of passes for each cable, were own estimates
based on consultation of PKR1’s general arrangement plan (GAP) and engineering judgement.
Combining these two datasets, a total power grid length for the considered components was
calculated as well as the fraction of this total for each component. The returned results are
given in Table 3.
A result of this model for instance was the observation that bridge and sensor mast systems
can be expected to make up about 4% of power grid cabling work for PKR1. Consequently, the
horizontal interdependency factors for components Bridge and mast sensor and comm.
installation (expressed as a dimensionless benchmark factor in the demonstrative model) and
Power grid cabling are the following:
Equation 26 Horizontal dependency: sensor and communication systems and power cabling
Where:
PCe = [m] power grid cabling, evaluated project
PCr = [m] power grid cabling, reference project
SCe = [-] mast sensor and communication systems, evaluated project
Quantification of Manufacturing Complexity in Shipbuilding Projects page 63/153
Additional components can be added to the power cabling horizontal interdependency factor
with their own sp factors.
Note that in Table 14, an sp factor of 0.08 is given. This increase with factor 2 in the model is
explained by the fact that component Bridge and mast sensor and comm. installation is in fact
part of Power grid group 1, which in turn makes up 50% of the total power grid group.
Moreover, discussions with the engineering department and evaluation of cable lengths from
other vessels (most notably the Joint Support Ship (JSS) Karel Doorman, as data regarding
cable lengths for the PKR were sparse) led to the conclusion that the physical model
overestimated the effects of redundancy for sensor power cabling and that some adjustments
to the cable ratios were required. Hence, the results listed in Table 3 are not always those
actually used in the demonstrative model.
It must be pointed out that this category is to be among the first to be looked into during future
fine-tuning efforts. The weights used in the demonstrative model are corrected based on data
for the Sigma 6110 and JSS Karel Doorman, without these corrections being implemented in
the physical model as this model was not intended for consideration as a research result. As
such, the relations factors of for these components groups are validated only to a limited extent.
Results of regression and engineering analyses such as described in this chapter make up the
factors given in Appendix J and integrated into to the demonstrative model. In this section, the
results from the analyses given in the previous sections are listed in the way they are in
Appendix J.
Each line describes an interdependency relation. From the columns that are filled out, it can be
deducted whether a horizontal or a vertical dependency is described: in case of vertical
interdependency, an influencing subsystem is given; for horizontal interdependencies, an
influencing component is listed. Note that for vertical interdependencies, weight factors sp are
not applicable (refer to Equation 7).
The figures resulting from the regression analysis in section 5.1 are given in Table 4, which was
taken from Table 12.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 64/153
In Table 4, it can be seen how besides frame spacing, other subsystem requirements are taken
into account for extrapolation of structural member area and hence steel work on large
structural members.
The exponents and weight factors given in sections 5.2.1 (propulsive power) and 5.2.2 (power
cabling) are given in Table 5 and Table 6 respectively, which in turn originate from Table 13
and Table 14.
Table 6 shows how the weight factors for other components, also resulting from the power grid
estimation model, are chosen. Note that these are not always the same as those given in Table
3, also taking into account that Power grid group 1 represents only 50% of the total estimated
power grid.
The way in which these factors can be entered into the demonstrative quantification model is
shown in Figure 8, section 6.1.1.4.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 66/153
The remarks made in sections 6.2 through 6.4 serve to illustrate the model used to validate the
relations and dimensions assumed during this research, as well to provide guidance during
future use of models based on the proposed quantification method.
In order to demonstrate and validate the set of formulas proposed in chapter 4, a model was
developed using Microsoft Excel. In Figure 4, a schematic representation of the structure of the
calculation model is given.
For each calculation, two ships are referred to: a benchmark project, which is used as a
reference, and an evaluated project, which is compared to the benchmark in order to estimate
its complexity. The user input for the benchmark is set only once, when the benchmark is
defined, and is marked darker in Figure 4. The requirements for the evaluated project are
required user input each time a complexity evaluation of a new project is made; this input is
marked in a lighter colour. The white boxes represent intermediate and finals output data of the
model.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 67/153
The final output category, Complexity factors, comprises the dimensionless factors expressing
the complexity of the evaluated project relative to the benchmark. In the demonstrative model,
an function is included which allows to combine estimated man hours per sub and main
system, providing an insight in relative complexity on multiple levels.
Note that the interdependency and differentiation factors and relation exponents are not
explicitly displayed in Figure 4. These factors can be considered as the arrows linking the
different data boxes in the schematic drawing. The exponents actually used for the
demonstrative calculations are given in Appendix J; chapter 5 is dedicated to the method used
to determine these exponents as this makes up a crucial part of defining new reference projects
and fine-tuning of existing benchmarks.
A division is made into four kinds of worksheets: presentation sheets, benchmark sheets,
evaluated project sheets and calculation sheets.
On the presentation sheets (orange in Figure 5), extrapolated component dimensions for the
evaluated vessel are shown, and an overview of the resulting complexity factors is given.
Evaluated project sheets (purple) contain the requirements and component dimensions for the
evaluated projects as supplied by the end-user.
Finally, in the calculation sheets (green), relations between components, system requirements
and work content are quantified by the user. These relations are then used to extrapolate the
benchmark properties.
Complexity_graph: in this sheet, the calculated complexity factors are plotted per main
system: both the relations with the benchmark project as well as the fractions of total
work per main system are given. The layout of the sheet is shown in Figure 6; examples
of charts which form the contents can be found in section 7.4.
Dimension_estimation: here, the benchmark data are displayed, and system
requirements (such as design speed) or component dimensions (such as main engine
power) can be set for the evaluated project. Using the relations defined in the definition
sheets, final dimensions of relevant components are calculated.
Complexity_estimation: using the final dimension resulting from sheet
dimension_estimation, and the work contents defined in the relevant benchmark sheet,
the combined complexity factor fcc per subsystem, main system and total project for the
evaluated ship are determined, resulting in figures ranging from 0 (implying zero work
content) to any number indicating the work content in proportion to the benchmark
project. These combined complexity factors are calculated according to Equation 13,
using results from the sheets mentioned in sections 6.1.1.2 through 6.1.1.4.
Figure 7 Layout of input sheet for benchmark project PKR1 (man hour data were erased)
The patterns and colour schemes of the benchmark sheets are set up in a way that they guide
the user to fill in main and sub systems as well as associated components according to the
hierarchical structure as described in section 3.1.
In these benchmark data sheets, main systems, subsystems, parameters da, dc, ra (refer to
chapter 4) and man hours per dimension unit are defined, as well as the room (or differentiation
group) per component and the associated exponential factor e3 for determination of fd per
component.
In these sheets, users can supply requirements and component dimensions in the same way
as in the benchmark sheets. Layout and input data are the same, except for man hour data and
differentiation groups/rooms which do not need to be given for evaluated projects.
For new projects, new project sheets should be made and referred to in a separate index sheet,
following which this new project can be selected from a dropdown menu in the
Complexity_graph sheet.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 70/153
In these sheets, user can specify the exponents and applicable components and systems for
the relations as given in chapter 4.
Note that the calculations sheets are hidden in the demonstrative model, as end-users are not
expected to require access to the data in these sheets. Developers can unhide these sheets
using Visual Basic for Applications (VBA).
The method devised during this research delivers the most accurate results if man hour data
are available on a component level; such detailed data are however not easily found for Damen
projects. For instance, meetings with electrical engineers learned that cable lengths for specific
systems are not presently available, and the man hours connected to installation of these
cables are not explicitly mentioned either.
A more flexible means of defining system and component sizes must therefore be applied,
within the framework of the categorisation proposed in chapter 4.
As the estimation method is based on proportions and comparisons between data from
different ships, rather than on absolute values, component dimensions that are unknown can
be assigned a size factor instead of an actual measurement. For example, if it is known that the
cabling for a certain communication network took 800 man hours to install, but the total length
of the cabling is unknown, the benchmark component dimension (which would normally be total
length in metres) can be set to value 1. This value can still be extrapolated for other projects by
using the same relations as were defined in the technical analysis.
In some cases, individual component dimensions are known, but no man hour data referring to
these specific components are available; such cases leave several options to extrapolate
production hours.
It can be chosen to consider a specific component as decisive for the number of production
hours connected to a certain system.
Alternatively, the components can be brought back to non-physical dimensions, based on their
production hours in relation to each other rather than on physical characteristics. These
dimensions can subsequently be summated into one final component, representing the entire
system. Man hour estimations for the system in a newly evaluated project can be deducted
from the extrapolation of this single component.
Of course, the abovementioned procedures are partly based on personal insight and hence of
subjective nature; the relations thus identified must be validated by applying the model on past
projects and comparison with previous research.
To ensure relevance of the model results to the Damen Schelde project department, the
Damen SWBS codes are used to categorise main systems in demonstrative calculations.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 72/153
Numbering of subcategories within these main systems is however not necessarily maintained
in order to allow for flexible definition of the model.
To demonstrate the model, the PKR1 project was chosen as a benchmark. This project, a
Sigma 10514 surface combatant built for the Indonesian navy, is documented well enough to
allow a well-reasoned division of construction work content in function of vessel specifications.
The PKR benchmark is then compared to the Sigma 6110, which has different dimensions and
specifications but possesses mostly comparable systems. Additionally, the versatility of the
benchmark is investigated by using the demonstrative model to compare the PSV 5000 and
PSV 3300, platform supply vessels with a layout that differs fundamentally from that of surface
combatants, with the Sigma 10514.
Correct definition of the benchmark, both in terms of technical specifications and in work
content, is of paramount importance to get a good insight in the division and evolution of
complexity as it is defined in this research. A short description of the way in which DSNS data
regarding the PKR1 project were processed as well as the results of these efforts is given in
sections 6.2.2.1 through 6.2.2.3; further elaboration on this subject is given in Appendix L and
Appendix M; this should give an idea of the data generally required to set up a basic
benchmark.
Considering required benchmark data, it can be said that the benchmark needs to be only as
detailed as is required for the evaluated projects, and correct determination of man hours and
relation factors is of more importance than the number of defined components. As a minimum,
key components related to the most important functional requirements (design speed,
accommodation, combat capabilities, sensor systems, cargo capacity) must be given.
Secondly, all systems present on the evaluated project must be also present on the benchmark
vessel as the estimation method is based on extrapolation. For example, the benchmark which
was defined in the course of this research and which was based on the Sigma 10514 surface
combatant, was adapted to include specific types of deck equipment (deck cranes, winches,
fire-fighting installations) in order to arrive at a meaningful estimation for platform supply
vessels.
The first vessel of the PKR programme, a Sigma 10514 surface combatant, is the vessel for
which the most detailed historical and planned man hour data were available, while it also can
be labelled as first-of-class which is important to exclude effects influencing organisational
complexity resulting from, for instance, learning curves, as was described in section 2.1.3.
Secondly, the division of work content for both recorded and planned man hours do not extend
to a sub system level. Data can be combined however to deduct how work is distributed over
various systems as the planned and recorded hours use different categorisations to describe
the same activities.
Thirdly, as mentioned above, armament is included in neither the DSNS estimations nor the
registered data. As a result, these items can only coarsely be taken into account.
As a validation, it was investigated how the recorded production hours for modules III and V
related to the planned hours for the systems integrated in these modules. It was found that the
DSNS planning overestimated the actually required production hours by 3.74%.
Furthermore, subcategories were defined and it was checked how work content within these
categories corresponded. This eventually led to the conclusion that the major categories such
as casco steel work, welding and outfitting were estimated with errors less than 10%.
Based on these observations, the conclusion is drawn that the DSNS estimations for modules I,
II, IV and VI were fit to serve as a basis to define a reference project.
The registered hours for modules III and V were given per SWBS category, per module and per
task (examples being KAMAG transport, pressure testing and steel cutting). The planned hours
were less detailed but still made a division per module and per major category of work
(examples being casco welding, piping and mechanical work).
As the categorisations were not the same, distributions could be deducted from the manner in
which hours were differently grouped for modules III and V. For the other modules, for which
only planned data were available, distributions were based on data supplied by engineering (for
instance, cable lengths) and on estimation models (for example, sensor weights).
This way, a substantiated distribution of work content over 58 components could be made,
which were supplemented by related components to arrive at a PKR1-based benchmark
comprising 26 design requirements and 120 key components.
An example of a system breakdown is given in Appendix I; relations between the systems and
components identified in this analysis are mathematically defined in Appendix J and Appendix
K, according to the method principles given in section 3.1.
In general, the system breakdown given in Appendix I aims to provide a benchmark based on
the PKR1 project. For validation and demonstrative calculations, the Damen Sigma 6110, PSV
5000 and PSV 3300 designs are used as evaluated projects; reference is made to chapter 7 for
descriptions and results.
To enhance flexibility of the estimation model with respect to systems that are not present on
the benchmark vessel, it was chosen to define man hours not in absolute values but in work
content per component unit. This has no major influences on the formulas used but allows end
users to manually define components specific for the newly evaluated vessels without
generating divide-by-zero errors.
Previous publications on man hour estimations were consulted, some of which put forward
basic methods for man hour predictions. Especially the primary factors identified in these
papers as well as the values assigned to these factors are of interest, and to a certain extent,
provided input for the comparative complexity model developed in this research (Chou &
Chang, 2001).
It is important to realise that the extrapolation factors aim to describe the evolution in work
content rather than the evolution of the physical dimensions of the components. In many cases,
these may be directly related (such as the number of pipe spools and man hours involved in
mounting), or there may be a relation which can be assumed to resemble more of a square root
function (such as installation of engines with increased MCR).
The way in which the extrapolation exponents and factors, put forward in chapter 4, can be
correctly determined, was explained in chapter 5.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 76/153
A description of the functional architecture of the demonstrative model was given in section 6.1.
As was described in this section, the mathematical structure of the model is entirely based on
the mathematical framework defined in chapter 4.
For the results listed in this chapter, the Damen SWBS structure will be adhered to for
categorisation of work content. The prediction model makes a much more detailed subdivision
to determine man hours per subsystem and component, the particulars for subsystems and
components will however not be covered in this report considering the amount of data involved
and the limited relevance of this information.
As was stated previously, the demonstrative calculations will be based on project PKR1 (yard
number 414), a vessel of the Damen Sigma 10514 design. Technical specifications of this
vessel are given in Appendix E, the system breakdown structure for this project in Appendix I.
The Sigma 10514 surface combatant offers a solid reference for extrapolation to other projects
as it possesses a wide range of civil and naval appliances and characteristics. For instance, the
CODELOD configuration allows for extrapolation to diesel-electric ships as well as diesel-
mechanical (diesel-direct, separately geared and CODAD configurations) and CODELOD
propulsion systems.
Surface combatants are however not laid out for cargo handling, a task which is the main
function of most sea-going vessels. For this reason, several optional components were added
to the PKR1 benchmark in order to allow it to be used for comparing platform supply vessels as
well. This possibility was deemed desirable as a demonstrative calculation with a platform
supply vessel (PSV), which is fundamentally different from a surface combatant in terms of
layout and function, would give insight in the versatility and limitations of the proposed
extrapolation method.
Three projects were compared with the Sigma 10514 benchmark: the Sigma 6110 surface
combatant and the PSV 5000 and PSV 3300 platform supply vessels. Specifications of these
ships can be found in Appendix E through Appendix H; for the input values used for the
estimations given in the following sections, referral is made to the relevant Excel calculation
files.
As a reference, the division of work content for the benchmark project is given. Complexity
estimations for the Sigma 6110 and PSV vessels in the next sections relate to this project.
7000 ARMAMENT
Figure 9 Distribution of work content for benchmark project PKR1 (Sigma 10514)
Quantification of Manufacturing Complexity in Shipbuilding Projects page 78/153
A notable difference between the Sigma 10514 and the Sigma 6110 is the much smaller size of
the latter. Despite this, the amount of sensor and combat systems is comparable, which
explains the large differences in work content observed for the relevant main systems.
It can be concluded that in general, construction of a Sigma 6110 requires less work than
construction of a Sigma 10514: according to the complexity model, the amount of man hours
required for production of a Sigma 6110 is in the order of 51% of the hours required to produce
Quantification of Manufacturing Complexity in Shipbuilding Projects page 79/153
a Sigma 10514. This is mainly due to the fact that the hull is much smaller; from Figure 9, it can
be deducted that this segment accounts for a considerable part of the production hours.
On the other hand, categories 4000 (command and surveillance) and 7000 (armament) are
relatively complex, compared to other parts of the ship. Armament and sensor systems are
located closer to power generation and auxiliary systems; there is however fewer space per
component compared with the Sigma 10514. This has an increasing effect on installation work
and is taken into account by the differentiation factors. Moreover, a smaller complement raises
the need for a higher degree of automation, which in turns makes the individual components,
and the networks connecting them, considerably more complex.
7000 ARMAMENT
Figure 11 also shows that the work associated with hull erection is comparatively limited. For
this reason, category 9000 (ship assembly and support systems) for Sigma 6110 makes up a
Quantification of Manufacturing Complexity in Shipbuilding Projects page 80/153
larger part of the production process than for the Sigma 10514, despite the fact that the Sigma
6110 is much smaller and is not built in multiple yards. From Figure 10 however, it can be
concluded that this is only a relative difference, and that category 9000 for Sigma 6110 is still
expected to be much reduced in terms of man hours when compared to the same category for
the PKR1 project. Remarkably, category 1000 represents only few more production hours than
category 4000.
In section 2.1.5, it was stated that in this research, complexity is expressed in terms of man
hours. The DSNS project department however often makes use of figures in the format of man
hours per ton; for ease of comparison, a graph with figures in this format is included in Figure
12.
700
600
500
400
300
200
100
0
Hull Outfitting
Figure 12 Man hours per ton: comparison between Sigma 10514 and Sigma 6110
It can be seen that, both for hull construction and outfitting, work content per ton is higher for
the Sigma 6110.
For the hull, this is explained by the fact that weight of the construction decreases with a slightly
higher rate than does the amount of steel plates and the weld distance, due to the use of
Quantification of Manufacturing Complexity in Shipbuilding Projects page 81/153
thinner steel plates and aluminium superstructure parts. This remark is a good illustration of the
reason why in section 2.1.5, it was chosen to relate complexity to man hours rather than man
hours per ton.
For outfitting, this is primarily the result of the higher system density, which is taken into
account through higher differentiation factors, as well as a larger amount of network cabling
which adds virtually no weight.
The available estimate for the Sigma 6110 by the DSNS project department, made in the
context of a possible future project, only comprises work done by the yard, excluding
subcontractor work such as electric cabling and HVAC. The results of the demonstrative
calculations can hence only be partially compared.
The comparison given in Table 7 accounts for 89.43% of the activities listed in the DSNS
estimate. Although this estimate has not been confirmed by actual production data, the
comparison learns that, when it comes to evaluating surface combatants from the Sigma class,
the results of the demonstrative complexity model are in line with those from the models which
have been successfully used by the DSNS project department to predict past projects.
As was stated in section 7.2, the benchmark model was expanded to verify the versatility of the
complexity estimation method. Additionally installed equipment such as tug winches, cargo
cranes and firefighting equipment was added as individual components and from the man hour
estimates made by DSNS, work content for these components was derived.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 82/153
This of course implies that the results for these specific components as well as the auxiliary
components connected to this equipment are merely deducted from other estimations and as
such cannot be considered as validation.
This is the reason why the PSV 3300, a vessel design with the same main functions as the
PSV 5000 but different dimensions, was also evaluated; the results of these calculations can be
found in section 7.4.5.
Whereas differences between the Sigma 10514 and Sigma 6110 were mainly to be found in
dimensions and, hence, system density, the differences between the former and the PSV 5000
are much more extensive.
First of all, the PSV 5000 lacks combat systems. This does not only entail the absence of
armament (as is evident from the zero-value for category 7000), but also implies that much less
sensor systems are installed. As a result, category 4000 (command and surveillance)
represents only a relative small fraction of the total work content.
Secondly, the hull of the PSV is much heavier. Surface combatants are not designed to carry
large amounts of cargo while they do have high speed requirements, leading to designs with
relatively thin plates in order to limit the vessel’s displacement. Platform supply vessels
however are designed for a wide range of tasks, including transport of deck cargo. Hence, hull
rigidity and maximum allowable deck load is of greater importance than speed, resulting in a
greater average plate thickness.
This on its turn greatly increases the work content connected to hull erection. The hull in itself
may not be much more geometrically complex, steel processing (plate forming) however
becomes more intensive as the plate thickness increases. This effect is particularly witnessed
when comparing the Sigma 10514 and the PSV 5000: steel processing is estimated to
consume twice the amount of man hours for the PSV 5000 as for the Sigma 10514, while
activities such as steel work and section welding increase in a more limited degree.
Moreover, the auxiliary systems are different. In Figure 13, it can be seen that category 5000
for the PSV 5000 comprises about the same amount of man hours as for the Sigma 10514. The
systems connected to this category however are not the same: the helicopter hangar and
power supply for armament, sensor systems and electric propulsion are replaced by deck and
firefighting equipment.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 83/153
Finally, category 2000 (propulsion) is reduced more than would be expected, based on the
considerably increased design displacement and a design speed of 14 knots. There reason
behind this is that only electric propulsion motors are installed; the diesel generators delivering
propulsion power are considered as part of category 3000 (electric plant).
From Figure 14, it becomes apparent that for the PSV 5000, work content mainly consists of
hull construction. A second large category is made up by auxiliary systems which comprise,
amongst others, engine room auxiliaries (pumps, heat exchangers and such), piping, HVAC
and deck equipment.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 84/153
7000 ARMAMENT
Highly remarkable: the demonstrative estimation model predicts that outfitting in general is, per
ton of installed equipment, considerably more labour intensive for the PSV 5000 than for the
Sigma 10514.
Closer examination of the results shows that most of the outfitting for PSV 5000 is related to
items with a relatively low physical weight: small electrics and lighting, deck covering,
conservation and (to a certain extent) piping. Although in total and absolute terms, these
relatively labour intensive components (in terms of man hours per equipment piece and per ton)
represent less work than their counterparts in the Sigma 10514, they do form a larger fraction
of the total outfitting portion. As such, they substantially increase the average work per ton
equipment installed.
When the complete vessel is considered however, the number of man hours per ton for the
PSV 5000 is only a fraction of that for the Sigma 10514, as outfitting for only a small fraction of
Quantification of Manufacturing Complexity in Shipbuilding Projects page 85/153
the light ship weight. In this sense, too, figures in man hours per ton may give a distorted view
on work content distribution and complexity.
Figure 15 Man hours per ton: comparison between Sigma 10514 and PSV 5000
As the categories giving in the DSNS preliminary estimate are not always explicit in terms of
the equipment they comprise, only selected parts of the estimate are used for a comparison.
The hours used for the comparison given in Table 8 represent 71.45% of the complete DSNS
estimate. Here, the same conclusion can be drawn as in section 7.4.2.2: the results of the
quantification model (except for quality control, which is hard to relate to physical properties
and not critical in terms of man hours) are in line with the DSNS project department estimates.
7.4.4. Comparison between Complexity Factors for Sigma 6110 and PSV 5000
A brief recapitulation on both the Sigma 6110 and PSV 5000 allows demonstrating the practical
use of the complexity factors resulting from complexity estimations as performed by the
calculation model: complexity factors for both projects are compared in Figure 16.
As the PSV 5000 is larger both in terms of lengths and weights, it can be expected that
category 1000 (hull structure) represents more work. Also categories 4000 (command and
surveillance) and 7000 (armament) hold no surprises: no armament is installed on the platform
supply vessel, and much fewer sensor systems are installed.
Category 0000 (guidance and administration) merely consists of input variables and is hence
constant at value 1, while category 8000 (integration engineering general) refers to engineering
which is kept out of scope, also resulting in a constant complexity factor 1.
The propulsion systems are fundamentally different: the PSV is propelled by two azimuthal
thrusters, supplemented by two bow thrusters for DP operations; all fed by the electric grid. In
short, the PSV 5000 is laid out as an all-electric ship (Klein Woud & Stapersma, 2002).
Quantification of Manufacturing Complexity in Shipbuilding Projects page 87/153
The Sigma 6110 on the other hand has a CODAD propulsion system with four diesel engines,
driving two propellers through two gearboxes; contrary to the Sigma 10514, the vessel is not
equipped for electric propulsion.
The results of these differences in configurations is that, according to the Damen SWBS
structure, category 2000 (propulsion) is somewhat more complex for the Sigma 6110, while
category 3000 (electric plant) for the PSV 5000 is predicted to be more complex due to the
vastly increased electric power requirements.
Another conspicuous difference is category 5000 (auxiliary systems): as the PSV is equipped
with a large amount of deck equipment, requiring much more auxiliaries (such as, for instance,
piping for the firefighting system), said ship requires more work in this respect. The same goes
for category 6000 (outfit and furnishing): increased numbers of crew and passengers as well as
more generous living and messing quarters result in more work on the accommodation.
The PSV 3300 was evaluated to investigate to what extent the extrapolation factors for auxiliary
systems, tuned in such a way that they correctly estimate work content for the PSV 5000, allow
accurate evaluation of platform supply vessels with different dimensions.
Compared to the PSV 5000, production of the PSV 3300 involves fewer man hours, which is
not surprising considering the different dimensions. The distribution of man hours over the
various main systems is also comparable, with some minor yet remarkable differences.
Category 3000 (electric plant) represents a smaller fraction of the total work content than for the
PSV 5000; this is explained by the reduced engine power requirement (lighter ship, slightly
lower design speed) and the absence of external firefighting installations.
Moreover, work connected to category 6000 makes up a smaller fraction due to the
considerable reduction in accommodation for passengers (6 passengers instead of 28 on the
PSV 5000, while the crew remains constant at 16 persons).
7000 ARMAMENT
The work content distribution shown in Figure 18 illustrates how, for smaller vessels with similar
functional requirements and hence auxiliary systems, outfitting forms an increasing portion of
the total production work content. In the case of the PSV 5000 and PSV 3300, the size of the
hull decreases with a greater rate than does the amount of systems and components; this not
Quantification of Manufacturing Complexity in Shipbuilding Projects page 90/153
only reduces relative construction work for the hull but also decreases the space available for
each system installed, further increasing the share of total work content for propulsion, power
generation and auxiliary systems.
The graph displaying man hours per ton for hull construction and outfitting for PSV 3300,
included as Figure 19, shows the same, relatively high labour intensity in terms of man hours
per ton for outfitting as for the PSV 5000, be it to a lesser extent.
The absence of external firefighting equipment as well as a more limited passenger capacity
reduces the amount of small electrical items and pipe spools; items that add limited weight but
entail considerable installation work, when compared to deck equipment such as cranes and
tug winches. As a result, the number of man hours per ton for outfitting is somewhat reduced
compared to the PSV 5000.
Still, this number is significantly higher than for the Sigma 6110, giving the impression that the
outfitting process of the PSV 3300 is more complex while it in fact only indicates that a
considerable share of the components installed is of lower physical weight.
700
600
500
400
300
200
100
0
Hull Outfitting
Figure 19 Man hours per ton: comparison between Sigma 10514 and PSV 3300
Quantification of Manufacturing Complexity in Shipbuilding Projects page 91/153
In the comparison given in Table 9, auxiliary systems such as deck equipment are not taken
into accounted as it cannot be deducted with certainty how these systems are categorised. Of
the total work content resulting from the DSNS estimate, 73.37% is accounted for in this table.
In the case of the PSV 3300, this is an important note as auxiliary systems are estimated to be
considerably more complex in the model resulting from this research than in the DSNS
estimate; this is mainly due to the increase system density. Considering total production work
content, the ratio between the quantification model and the DSNS estimate becomes 1.20.
Whether or not this result is closer to reality than the DSNS prediction is subject to analysis of
actual production data as soon as such data can be obtained; the estimate is however based
on insights from the Damen Schelde production and engineering departments and at least
partly reflects experience with past projects.
The complexity calculations described in section 7.4 were all based on the following 43 input
parameters:
Results of complexity estimations for the Sigma 6110 and PSV 5000 are evaluated in function
of these input categories in order to provide insight how available data influences validity of the
results.
From Figure 20, it can be concluded that the number of available data is not critical for
evaluation of the Sigma 6110: as long as the basic dimensions and functional requirements
(input category I) are known, a good indication of the work distribution can be obtained. In
relevant SWBS categories, differences can be spotted as input categories are added; for
instance, 7000 (armament) increases as soon as input category III is added.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 94/153
2
Category I Category I+II Category I+II+III
1,8
Category I+II+III+IV Category I+II+III+IV+V
1,6
1,4
1,2
0,8
0,6
0,4
0,2
Figure 20 Evolution of estimated work distribution for Sigma 6110, relative to PKR1
Figure 21 does not allow clear conclusions to be drawn with respect to accuracy of the
complexity estimation: although the results show a peak for the case where only input category
I is supplied, it is not known which value can be considered to be correct as no validated
production data are available for the PSV 5000. Moreover, the results do not converge to a
specific value; no trend can be deducted in general, rendering the chart in Figure 21 not
especially useful when it comes to assessing accuracy of the results in function of input
variables.
The general conclusion for the calculations for complexity of the Sigma 6110 is that, for
accurate results, more data are obviously desirable; a quick overview of the differences with
Quantification of Manufacturing Complexity in Shipbuilding Projects page 95/153
other surface combatants in terms of work content and hence complexity can however be
achieved with a set of only 14 parameters.
0,5
Total relative project complexity
0,49
0,48
0,47
0,46
0,45
0,44
0,43
0,42
0,41
0,4
1 2 3 4 5
Figure 21 Total relative project complexity for Sigma 6110 in function of input categories
For the PSV 5000, the evolution of complexity factors per SWBS category appears to be
erratic, as can be witnessed in Figure 22. This is explained by the fact that certain key
categories, most notably 4000 (command and surveillance), 5000 (auxiliary systems) and 7000
(armament) are subject to sudden change when their relevant components are specified.
To obtain sufficiently accurate results for the PSV 5000, all five input categories must be
supplied, which means that 43 input parameters must be known. Differences with the results for
the Sigma 6110 (Figure 20), which evolve much less as the amount of input variables
increases, are clearly visible.
This observation leads to an important conclusion, which in fact should not come as a surprise
considering the mathematical principles of the proposed complexity estimation method.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 96/153
In general, as the differences in functional requirements and hence, component and system
types (note that this is not the same as component and system dimensions) between the
compared projects increase, models based on extrapolation of projects require more detailed
input variables in order to reach the same degree of accuracy.
The statistical sensitivity of the demonstrative model is not investigated in this research:
considering the lack of historical data on the production of the Sigma 6110, possibilities for
validation are very limited and results from a sensitivity analyses would ultimately be
speculative.
2
Category I Category I+II Category I+II+III
1,8
Category I+II+III+IV Category I+II+III+IV+V
1,6
1,4
1,2
0,8
0,6
0,4
0,2
Figure 22 Evolution of estimated work distribution for PSV 5000, relative to PKR1
Quantification of Manufacturing Complexity in Shipbuilding Projects page 97/153
During demonstration and discussing of the demonstrative model with peer researchers, the
suggestion was made that, apart from for comparing of different projects in order to quantify
complexity and substantiate production planning, the quantification model could be of use
during evaluation of production costs of a design as early as in the design stage. Prerequisite to
this is that the key components of this new ship are identified and dimensioned, and that it is
approximately known how many production hours are connected to these components (in other
words, that the evaluated project is defined as a benchmark).
Using interdependency and differentiation relations identified for earlier projects, it can then be
evaluated how production hours evolve in function of changing functional requirements and
component dimensions.
Such an application could aid a yard in optimising their designs in terms of production costs,
especially if there already exists a practice of engineering products out of sets of standardised
components for which dimensions, components and production and installation hours are
known, and as such also have an influence on strategic design choices.
The results of these evaluations could for instance be used to provide input concerning
production costs for product family profit prediction models such as the ES3P-method proposed
by Nieuwenhuis in his doctoral dissertation published in 2012. In this dissertation, profit of the i-
th instance of a series-built product in the j-th year after production of the first ship is assumed
to be dependent on the following costs:
Where:
Pi,j = profit
SPi,j = selling price
DECi,j = design and engineering costs
PCi,j = production costs
ACi,j = costs for material, components, external labour and knowledge
i = index for number of product in the series
j = index for year of production
Quantification of Manufacturing Complexity in Shipbuilding Projects page 98/153
The effect on profit by parameter PCi,j (excluding organisational factors) can be approximated
with relative ease using the complexity quantification method proposed in this research.
To illustrate how evolution of production cost of a design can be evaluated, frame spacing and
plate thickness of the Sigma 10514 benchmark are varied. Thinner steel plates result in an
overall thinner hull steel weight, which in turn reduces the required propulsive power. To retain
sufficient hull stiffness however, more frames must be placed; in other words, frame spacing
decreases. Consultation with structural specialists from the DSNS engineering department
learned that in general, when designing a surface combatant with a displacement hull and
transverse framing system, the following ratio is kept in mind:
Where:
dt = [m, mm] transverse frame spacing
tp = [m, mm] skin plate thickness
0,9
0,8
Figure 23 Work content distribution for Sigma 10514 in function of varying frame spacing
Quantification of Manufacturing Complexity in Shipbuilding Projects page 99/153
Important to note: frame spacing and plate thickness were the only input variables that were
supplied to the model as user input. This implies that all other components could be freely
extrapolated by the estimation model, leading to for instance an increase in installed propulsive
power as plate thickness (and hence light ship weight) increased, as can be witnessed by the
peak in category 2000 (Propulsion) in Figure 23. This mitigates the effect of decreased plate
thickness and frame spacing on total work content as plotted in Figure 24, as the resulting light
weight leads to somewhat smaller propulsion and auxiliary systems.
Figure 24 shows that, while light ship weight evolves linearly with plate thickness, work content
shows a non-linear, rapid increase as plate thickness is reduced. If plate thickness is reduced
to 4 mm for instance, light ship weight is estimated to drop by 17% while work content
increases with more than 10%.
1,2
Estimated total work content
1,15
Estimated light ship weight
1,1
Relative to benchmark
1,05
0,95
0,9
0,85
0,8
4 mm 5 mm 6 mm 7 mm 8 mm
Average plate thickness (equals frame spacing / 100)
Figure 24 Work content and LSW in function of average plate thickness, Sigma 10514
Quantification of Manufacturing Complexity in Shipbuilding Projects page 100/153
8.1. Conclusions
Which systems are typically installed in projects executed by Damen Schelde Naval
Shipbuilding, and what are their relevant characteristics and possible interferences with other
systems?
Quantification of Manufacturing Complexity in Shipbuilding Projects page 101/153
After analysis of the available man hour data and pre-contract specifications of the PKR1
project (Sigma 10514 design), the components listed in Appendix I were taken as a basis for
development of a benchmark project.
Which input data concerning specifications of these systems are required to generate
complexity estimations?
An in-depth answer to this question is found in section 7.5. As a general conclusion in this field,
it can be said that for complexity estimation methods based on extrapolation such as put
forward in this report, the number and variety of required input parameters increases as the
compared projects differ more from each other in terms of functional specifications.
How can these input data of the method be related to figures expressing complexity?
The answer to this question can be found in chapter 4, where the way in which relations can be
mathematically defined is discussed in detail. Linear and exponential factors between system
requirements and component dimensions can be specified; in general, it is assumed that
relation can be expressed by a power factor which can be any rational number.
As a general conclusion, it can be stated that, even with limited historical data being available,
projects can be defined as benchmarks and extrapolated to estimate work content and
complexity for new projects by a mathematical structure as put forward during this research.
Although these factors cannot be inconsiderately applied to predict future projects, they are an
objective measure to quantify complexity of a project as a whole as well as the individual
systems installed, to a degree which is limited by the elaborateness of the benchmark and
Quantification of Manufacturing Complexity in Shipbuilding Projects page 102/153
extrapolation factors on one side, and by the detailedness of the supplied input data on the
other.
Furthermore, it is concluded that the practice of estimating construction work based on steel
and outfitting weight (or, man hours per ton) delivers very rough results. Even when vessels
with relatively similar specifications such as the Sigma 10514 and Sigma 6110 are compared,
results may be distorted: in 7.4.2.1, it was shown how the use of thinner plates and, to a limited
degree, aluminium led to higher hours per ton figures.
Extrapolation methods as proposed in this research are mostly based on changes in hull shape
and size: estimations for casco work are related to weights only for specific activities such as
steel processing. This allows for more advanced estimations without raising the need for
extensive input data sets.
Another conclusion, resulting from the analysis of evolution of estimated complexity factors in
function of the number of supplied input data, is that the validity of the results, for a constant set
of input data, decreases as the differences in functional requirements and specifications
between compared projects increases. Especially for project and planning departments, which
aim to predict production processes (and hence, complexity) in an as early phase as possible
with a very limited set of known variables, this effect is not to be lost sight of.
A final remark is that, when the method proposed in this report is used for cost prediction and
yard planning, it must be kept in mind that influences resulting from the structure of the
organisation as well as factors such as work force learning curves and efficiencies are not
taken into account in the calculations, as was stated in section 2.1.3. This limitation made that,
in order to arrive at correct results for regression analyses and work content estimations, only
first-of-class vessels were used for the demonstrative calculations: the PKR1 is the first of two
vessels, and the figures for the Sigma 6110 are valid for the first vessel in a series of eight
planned for production by DSNS.
8.2. Recommendations
In the course of the research, it has proven to be a challenge to gather data that are detailed
enough to allow extrapolations based on first principles, as not much is known about the work
connected to individual components. Though this may be regarded as of not much use for other
purposes than detailed complexity and man hour analyses, estimation models such as that
developed in this research would benefit greatly from such detailed data collections. It is hence
recommended that, as much as is practically achievable, man hour data connected to key
Quantification of Manufacturing Complexity in Shipbuilding Projects page 103/153
components be registered for future projects, not only for the yard’s own work force but also for
work executed by subcontractors. This would eventually lead to more detailed and accurate
benchmarks.
Another point can could be improved upon is the accuracy of the interdependency and
differentiation exponents. This research mainly focuses on the mathematical method on which
quantification estimation models can be based; the demonstrative model used to evaluate the
Sigma 10514, Sigma 6110, PSV 3300 and PSV 5000 projects was based on extrapolation
models that were not always developed in full detail.
Although validation shows that the estimation model used for demonstration yields useful
results, the accuracy of these exponents could be improved by development of more detailed
system and component dimensioning models (for instance for prediction of piping, which was
already subject to research by DSNS) and subsequent regression analysis.
A specifically interesting topic in this respect is analysis of the influence of plate thickness (and
hence hull steel weight) on the man hours connected to steelbuilding, a remark that was made
in section 7.4.3.2.
Thirdly, the method put forward in this research allows for a standardised yet flexible definition
of benchmark projects. The projects evaluated in this research are limited to surface
combatants and platform supply vessels; through future research on other ship types,
benchmarks could be developed and expanded to allow evaluation of any vessel type,
providing DSNS with a standardised tool which allows a quick overview of any shipbuilding
project within certain margins of accuracy.
This recommendation is further substantiated by the observation made in section 7.5: the larger
the differences between compared projects, the larger the amount of input variables required to
achieve the desired level of accuracy. In order to end up with a standardised tool, allowing a
wide variety of projects to be evaluated with sufficient accuracy with data sets limited to only
basic pre-contract specifications, it is recommended that multiple benchmarks be developed,
covering an as wide as possible range of Damen designs.
Finally, figures in man hours per ton have shown to be not the most reliable way to express
complexity. Although this was excluded from the scope of the research, it was briefly
considered how complexity and scale could be adequately separated, and a proposed metric
was expressed in man hours per unit Gross Tonnage (GT), as it can be reasoned that volume
is a more correct description for scale than is weight. This observation is however subject to
future research.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 104/153
It can be expected that factors giving an indication of complexity per system as early as in the
pre-contract phase are of great value when approaching customers, subcontractors and other
departments within the company. Knowledge on which systems are most contributing to the
overall profit, and which systems are comparatively more labour intensive than previous
projects will, especially in the earliest stage of the project, put the party which possesses this
knowledge in an advantageous position during negotiations. It needs no further explanation
how companies such as DSNS could benefit from the application of early-phase predictive
models.
Predictive methods for manufacturing complexity can however also find application in the
design process, as was demonstrated in section 7.6. If, parallel with the design process, a
hierarchic structure is set up with interdependency and differentiation factors of key
components, the complexity of the design can be evaluated in function of varying system
requirements, allowing fine-tuning of the design while it is still under development.
Moreover, section 7.6 shows how derivatives of the proposed quantification method can find
application on a strategic level when the profitability of a modular, series-built product is
evaluated in detail.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 105/153
APPENDIX A REFERENCES
Gaspar, H., Rhodes, D., Ross, A., & Erikstad, S. O. (2012). Addressing Complexity Aspects in
Conceptual Ship Design: A Systems Engineering Approach. Journal of Ship Production
and Design Vol. 28 No. 4, 145-159.
Gidado, K. (1993). Numerical Index of Complexity in Building Construction with Particular
Consideration to its Effect on Production Time. Ph. D. Thesis, University of Brighton.
Hall, R. H. (1979). Organisations: Structures, Processes and Outcomes. New Jersey: Prentice-
Hall.
Hickson, D., Pugh, D., & Pheysey, D. (1969). Operations Technology and Organisation
Structure: an Empirical Reappraisal. Administrative Science Quarterly Vol. 14, 378-397.
Hubka, M., & Eder, W. (1988). Practical Studies in Systematic Design. Oxford: Butterworth-
Heinemann.
Kast, F., & Rosenzweig, J. (1979). Organisation and Management: A Systems and
Contingency Approach. New York: McGraw-Hill.
Klein Woud, H., & Stapersma, D. (2002). Design of Propulsion and Electric Power Generation
Systems. London: IMarEST.
Nieuwenhuis, J. (2012). Evaluating the appropriateness of product platforms for Engineered-
To-Order ships. Ph. D. Thesis, Delft University of Technology.
Thompson, J. D. (1967). Organizations in Action. New York: McGraw-Hill.
Valerdi, R. (2005). The Constructive Systems Engineering Cost Estimation Model
(COSYSMO). Los Angeles: University of Southern California.
van Lunteren, R., & Wiersum, S. (2016). Estimate 6110. Vlissingen: Damen Schelde Naval
Shipbuilding.
Williams, T. M. (1999). The Need for New Paradigms for Complex Projects. International
Journal of Project Management Vol. 17, 269-273.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 107/153
Table 1 Division of systems into categories based on complexity (Hubka & Eder, 1988) 31
Figure 1 Sketch of ship model used for casco property estimations 51
Table 2 Input variables for casco estimation model, based on Sigma 10514 52
Figure 2 Frame area in function of transverse frame spacing 55
Figure 3 Example of applied regression analysis for multiple output values 58
Table 3 Results of cable estimation model for Sigma 10514 (PKR1) 62
Table 4 Interdependency factors for component Large structural members 64
Table 5 Interdependency factors for component Main engine power 64
Table 6 Interdependency factors for component Power grid group 1 65
Figure 4 Schematic representation of the applied complexity calculation method 66
Figure 5 Arrangement of worksheets in the demonstrative Excel calculation model 67
Figure 6 Layout of complexity_graph sheet 68
Figure 7 Layout of input sheet for benchmark project PKR1 (man hour data were erased) 69
Figure 8 Layout of Interdependency_factors_fi sheet 70
Figure 9 Distribution of work content for benchmark project PKR1 (Sigma 10514) 77
Figure 10 Relative complexity factors per system for Sigma 6110 78
Figure 11 Distribution of work content for Sigma 6110 79
Figure 12 Man hours per ton: comparison between Sigma 10514 and Sigma 6110 80
Table 7 Sigma 6110: comparison between DSNS and model estimates 81
Figure 13 Relative complexity factors per system for PSV 5000 83
Figure 14 Distribution of work content for PSV 5000 84
Figure 15 Man hours per ton: comparison between Sigma 10514 and PSV 5000 85
Table 8 PSV 5000: comparison between DSNS and model estimates 86
Figure 16 Complexity factors for Sigma 6110 and PSV 5000 87
Figure 17 Relative complexity factors per system for PSV 3300 88
Figure 18 Distribution of work content for PSV 3300 89
Figure 19 Man hours per ton: comparison between Sigma 10514 and PSV 3300 90
Table 9 PSV 3300: comparison between DSNS and model estimates 91
Figure 20 Evolution of estimated work distribution for Sigma 6110, relative to PKR1 94
Figure 21 Total relative project complexity for Sigma 6110 in function of input categories 95
Figure 22 Evolution of estimated work distribution for PSV 5000, relative to PKR1 96
Figure 23 Work content distribution for Sigma 10514 in function of varying frame spacing 98
Figure 24 Work content and LSW in function of average plate thickness, Sigma 10514 99
Figure 25 Sigma 10514 silhouette 111
Figure 26 Modular division of Sigma 10514 113
Quantification of Manufacturing Complexity in Shipbuilding Projects page 109/153
Basic specifications of the Sigma 10514 surface combatant. More detailed data are found in
(Damen Schelde Naval Shipbuilding, 2012).
General
Basic designation Surface combatant
Crew 118 persons
Passengers 2 persons
Design speed 28 kn diesel-mechanic
14 kn diesel-electric
Propulsion CODELOD, two main diesel engines and two electric propulsion engines
Two CPPs
Dimensions
Length 105.11 m
Beam 14.02 m
Displacement 2365 t
Armament
Oto Melara 76 mm SR Strales deck gun
Close-in weapon system: FFBNW
2 triple torpedo launchers
2 ship-to-surface missile launchers
12 cell VL MICA anti air missile launchers (1 group)
Remarks
The first of the two PKR vessels was used as source for man hour data; in other words, no
learning effects were applicable for the production force during construction of the vessel used
as a reference in this report.
Both PKR frigates were built in six modules, which were joined only in a late stage of
production at the PT PAL yard in Surabaya.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 113/153
All weights are including a 3% margin. During transport of the individual modules, weights can
be somewhat less due to incomplete outfitting. (Damen Schelde Naval Shipbuilding, 2014)
Module IV – hangar
Weight: 81 t
Module VI – funnel
Weight: 77 t
Commercial in Confidence
2nd Generation Modular Building concept Sigma
Quantification of Manufacturing Complexity in Shipbuilding Projects page 114/153
Basic technical specifications of the Sigma 6110 surface combatant. More detailed data are
found in (Damen Schelde Naval Shipbuilding, 2015).
General
Basic designation Surface combatant
Crew 33 persons
Passengers 0 persons
Design speed 26 kn diesel-mechanic
Propulsion CODAD, four diesel engines in “father-and-son” configuration
Two CPPs
Dimensions
Length 59 m
Beam 10 m
Displacement 750 t
One surveillance and one fire control radar and navigation X-band radar
Four Electronic Support Measures modules (ESM), three optical surveillance and target
designation modules, IFF transponder
2 decoy launching installations
Armament
Oto Melara 76 mm SR Strales deck gun
Close-in weapon system: two .50 secondary guns
2 ship-to-surface missile launchers
8 cell VL MICA or 21 cell RAM anti air missile launchers (1 group)
Remarks
The specific Sigma 6110 surface combatant used for calculations in this report is the first out of
a series of eight ships for which the DSNS project department has made a production estimate.
This implies that no learning curve effects are to be taken into account, as is the case with the
PKR1 (Sigma 10514) vessel.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 116/153
Basic technical specifications of the PSV 5000 platform supply vessel. More detailed data are
found in (Damen Shipyards Gorinchem).
General
Basic designation Platform supply vessel
Crew 16 persons
Passengers 28 persons
Design speed 14 kn diesel-electric
Propulsion Diesel-electric, two azimuth thrusters
Two bow thrusters, optional third retractable thruster
Dimensions
Length 85.6 m
Beam 22 m
Displacement 5000 t
Armament
(Not applicable)
Remarks
-
Quantification of Manufacturing Complexity in Shipbuilding Projects page 118/153
Basic technical specifications of the PSV 3300 platform supply vessel. More detailed data are
found in (Damen Shipyards Gorinchem).
General
Basic designation Platform supply vessel
Crew 16 persons
Passengers 6 persons
Design speed 13.5 kn diesel-electric
Propulsion: Diesel-electric, two azimuth thrusters
Two bow thrusters
Dimensions
Length 80.1 m
Beam 16.2 m
Displacement 3300 t
Armament
(Not applicable)
Remarks
-
Quantification of Manufacturing Complexity in Shipbuilding Projects page 120/153
To provide a demonstrative model of the mathematical method proposed in the research, the
PKR1 project was analysed both in terms of technical specifications and man hour data,
resulting in the breakdown of systems and components as given in Table 10 is presented.
Additionally, components specific for the Damen PSV 5000 were added to increase the
flexibility of the demonstrative calculation model.
In Table 10, primary and secondary components are considered as one group in order to
present the components is an accessible fashion. In the demonstrative Excel model, this
presentation is maintained as the level of detail for the man hour data are not such that an
explicit subdivision in primary and secondary components is required.
It must be noted that the fact that components are located next to or below a given subsystem
does not necessarily indicate that these components are (only) part of said subsystem.
ER exhaust system
Gearboxes
Rudders
Propellers
Bow thrusters
HFO system
Propulsion and power plant weight
Steel work engine room
ER quality control
ER commissioning
ER preservation
ER deck covering
ER insulation
Shock attenuating mountings 2000
3000 ELECTRIC PLANT Power supply Auxiliary engines
Auxiliary engine power
Generator cabling
Power grid cabling
Power grid group 1
Power grid group 2
Power grid group 3
Lighting
Power sockets and small electrics
Shock attenuating mountings 3000
4000 COMMAND AND Platform network Hard-wired instrumentation I/O
SURVEILLANCE redundancy
Combat network Network cabling
redundancy
Platform related I/O
Platform related network cabling
Combat sensor I/O
Quantification of Manufacturing Complexity in Shipbuilding Projects page 123/153
Armament I/O
Combat related network cabling
Degaussing system Degaussing cabling and control
Bridge, control room and mast sensor
and comm. installation
Research sensor systems
Navigation sensor systems
Navigation radars
Military grade sonar systems
Combat radar and sensor systems
Hull mounted sensors and decoy
systems
Sensor system weight
Sensor and communication system QC
Sensor and communication system
commissioning
Sensor system preservation
Shock attenuating mountings 4000
5000 AUXILIARY SYSTEMS Helicopter hangar Heli hangar plates
Heli hangar stiffeners
Heli hangar welding
Heli hangar outfitting
Piping heli hangar
Deck covering heli hangar
Deck cranes (2.3 t)
Tug winches
Mooring winches
Cargo pumps
External firefighting system
Auxiliary equipment
Piping propulsion and power supply
Quantification of Manufacturing Complexity in Shipbuilding Projects page 124/153
General piping
Engine room ventilation system
Accommodation HVAC
Helicopter hangar HVAC
Bridge and control room HVAC
RHIBs
Tenders
Shock attenuating mountings 5000
6000 OUTFIT AND Outfit and furnishing Bridge and control room outfitting and
FURNISHING furnishing
Accommodation outfitting and furnishing
7000 ARMAMENT Armament Deck gun assembly
Secondary guns
Torpedo launcher installations
SAM installation
Anti-ship missile installations
Armament weight
Shock attenuating mountings 7000
8000 INTEGRATION (Factor 1) (Factor 1)
ENGINEERING GENERAL
9000 SHIP ASSEMBLY AND General preservation
SUPPORT SYSTEMS
General quality control
Ship commissioning
Modular strategy Modular construction additional cabling
Number of modules Deck welding provisions
Yard transport
Module joining
The tables in this appendix list the extrapolation factors for the interdependencies between the
systems given in Appendix I, so for both components installed on the Sigma 10514 and PSV
5000 designs. For an elaboration on how these factors were determined, reference is made to
chapter 5.
General remarks:
Displacement is assumed to be a function of hull steel weight and the amount and types
of propulsion and combat systems.
(design)
Crew benchmark Crew 1 1
factor
Passengers and Passengers and 1 1
crew benchmark crew
factor
1. Normally, displacement is influenced partly by the size of the propulsion plant. In the
demonstrative model, this influence is neglected to avoid calculation loop errors, hence
the value 0 for weight factor sp. Replaced by Hull and accommodation plates, which is
assumed to be related with auxiliary systems to some extent.
General remarks
The coefficients given in Table 12 were determined using the curve fit function in
Microsoft Excel. Coefficients were manually tuned to deliver the best fit; this was often
necessary to ensure that for an input value of 1, the output is 1 as well.
A coefficient without digits indicates that the number was chosen as a constant without
influence, while values with three digits are the result of a curve fit. This is a minor detail
with no consequences for the actual calculation.
stiffeners
Hull and Decks in hull 1.000 0.530
accommodation
plates
Ship length 1.000 0.259
Ship beam 1.000 0.775
Length of 1.000 0.606
accommodation
Decks in 1.000 0.515
accommodation
Deck height 1.000 0.437
Longitudinal 1.000 0.226
compartments
Transverse 1.000 0.218
compartments
Hull and Frame spacing 1.000 -0.257
accommodation
stiffeners
Ship length 1.000 0.519
Ship beam 1.000 1.000
Decks in hull 1.000 0.501
Decks in 1.000 0.580
accommodation
Length of 1.000 0.507
accommodation
Welding hull and Frame spacing 1 -1
superstructure
Automated welding -0.282 1 1
hull and
superstructure
Casco weld seams 1 1 1
Heli hangar welding 0.021 1 1
Automated welding Automated welding 1 1
hull and
superstructure
Casco weld seams 1.000 1 1
Quantification of Manufacturing Complexity in Shipbuilding Projects page 128/153
accommodation
Hull commissioning Hull steel weight 1 1 0.5
Total heavy Ship length 1 1
machinery space
Ship beam 1 1
Deck height 1 1
Technical room Ship length 1 0.5
volume
Ship beam 1 0.5
Deck height 1 1
System space Deck height 1 3.857
below deck
Bridge and top deck Length of 1 0.25
area accommodation
Ship beam 1 1
Sonar room volume Deck height 1 1
Armament space Ship length 1 0.25
Ship beam 1 0.25
Deck height 1 1
power
Steel work engine Propulsion engines 0.45 1 1
room
Auxiliary engines 0.35 1 1
Gearboxes 0.15 1 1
Rudders 0.3 1 1
Propellers 0.4 1 1
Bow thrusters 0.3 1 1
ER quality control Propulsion and 1 1 0.8
power plant weight
ER commissioning Propulsion and 1 1 0.5
power plant weight
ER preservation Propulsion and 1 1 1
power plant weight
ER deck covering Total heavy 1 1 1
machinery space
ER insulation Hull insulation 1 1 1
Shock attenuating Shock and noise 1 1
mountings 2000 requirements
Propulsion and 1 1 1
power plant weight
Component-specific remarks
1. “Propulsion engines” refers to the engines with their ancillary equipment, including fuel,
starting air, cooling water and lubrication systems. It is a virtual figure, related to the
dimensions of the propulsion train and the amount of time required for installation.
2. Power in kW per bow thruster.
3. Also comprises deck equipment mentioned under category 5000.
General remarks:
Quantification of Manufacturing Complexity in Shipbuilding Projects page 133/153
Required electric and hence auxiliary engine power is assumed to be related to the
power supply redundancy factor, number of sensor and weapon systems, complement
and electric drive power. The weights of these factors when determining required
auxiliary power were deducted from an electric load balance.
Component-specific remarks
1. “Auxiliary engines” refers to the engines with their ancillary equipment, including fuel,
starting air, cooling water and lubrication systems. It is a virtual figure, related to the
dimensions of the power generation prime movers.
2. Relation between system space below deck and deck height.
3. To allow for an accessible integration in the demonstrative model, relevant primary
components for power grid dimension estimation were divided into three groups.
4. This number was derived from the ratio accommodation volume to total manned room
volume.
systems
Sensor and Hard-wired 1 1 1
communication instrumentation I/O
system QC
Sensor and Hard-wired 1 1 1
communication instrumentation I/O
system
commissioning
Sensor system Sensor system 0.5 1 1
preservation weight
Sensor and 0.5 1 1
communication
system QC
Shock attenuating Shock and noise 1 1
mountings 4000 requirements
Sensor system 1 1 1
weight
Table 15 Horizontal and vertical interdependencies for command and surveillance systems
General remarks:
Some auxiliary systems were already taken into account in main systems 2000, 3000
and 4000.
hangar
Piping propulsion Ship length 1 0.22
and power supply
Ship beam 1 0.22
Propulsion and 1 1 0.5
power plant weight
General piping Ship length 1 0.22
Ship beam 1 0.2
Decks in hull 1 1
Deck height 1 0.22
Deck height 1 -3.857
Combat radar and 0.15 1 1
sensor systems
Combat system 0.2 1 1
weight
Passengers and 0.3 1 1
crew benchmark
factor
Hull and 0.35 1 0.2
accommodation
plates
Auxiliary equipment 3 1 1
Piping heli hangar 0.15 1 1
Auxiliary equipment3 Propulsion and 0.4 1 1
power plant weight
Deck cranes 0.15 1 1
Tug winches 0.1 1 1
Cargo pumps 0.05 1 1
External firefighting 0.15 1 1
system
Heli hangar outfitting 0.1 1 1
Engine room Propulsion and 1 1 0.5
ventilation system power plant weight
Accommodation Length of 1 0.5
HVAC accommodation
Decks in 1 0.8
Quantification of Manufacturing Complexity in Shipbuilding Projects page 140/153
accommodation
Ship beam 1 0.5
Passengers and 1 0.5
crew
Longitudinal 1 0.75
compartments
Transverse 1 0.75
compartments
Deck height 1 -3.857
Helicopter hangar Helicopter hangar 1 1
HVAC
Bridge and control Deck height 1 -3.857
room HVAC
Hull and 0.5 1 1
accommodation
plates
Sensor system 0.5 1 1
weight
Shock attenuating Shock and noise 1 1
mountings 5000 requirements
Piping propulsion 0.5 1 1
and power supply
Piping general 0.5 1 1
Component-specific remarks
1. The work connected to construction of the helicopter hangar is taken into account in
section 1000 HULL STRUCTURE; the input variables are however listed under 5000
AUXILIARY SYSTEMS.
2. It is assumed that with increasing vessel dimension, the length of individual pipe spools
also increases. Work content for piping hence is only related to vessel dimensions with
a limited degree.
7000 ARMAMENT
As this category refers to engineering activities which are not physically related to production, it
is not taken into account in this research.
General remarks:
For components related to transport and joining of modules, factors listed in Table 19
are based on personal judgement following from the modular building strategy
described by DSNS in the context of the PKR project (Damen Schelde Naval
Shipbuilding, 2014).
Table 19 Horizontal and vertical interdependencies for ship assembly and support systems
Quantification of Manufacturing Complexity in Shipbuilding Projects page 144/153
General remarks
piping and electrics): not all of the components in this differentiation group are physically
located in the same space, and hence are better considered as individual
interdependency factors.
In general, work content for equipment inside a differentiation group is assumed to
increase in a square relation with system density; refer to section 2.2.2 for
substantiation. Only if the individual components are located far away from each other,
or located in different physical rooms, exponential factor e3 is assumed to be 1 (for
example, auxiliary systems: this category also comprises deck equipment, which is
usually not located in a confined space).
Quantification of Manufacturing Complexity in Shipbuilding Projects page 146/153
APPENDIX L ASSESSMENT OF AVAILABLE MAN HOUR DATA FOR PKR1 (YN 414)
In section 6.2.2, it was stated that the project used as a demonstrative benchmark would be the
PKR1 project. Details regarding this vessel can be found in Appendix E.
As this ship was built using a modular strategy, with only two out of six modules being built at
the DSNS yard in Vlissingen, an overview of registered man hours for each module is not
readily available.
To obtain a sound man hour estimate for the complete project, data for modules III and V (built
in the Netherlands) were analysed and compared to estimates made by the project department.
Remarks:
As only one project was evaluated, no remarks can be made regarding statistical significance
of these figures. It can however be said that, with a deviation of only 3.7%, the estimates made
by the project department form a sound basis for an estimation model designed for use during
the pre-contract phase; experience by the project department shows that deviations in high-
level estimations can be expected not to exceed 5% (van Lunteren & Wiersum, 2016);
variations mainly lay in different categorisations.
Another issue of interest is the division of man hours per activity. After evaluation both the
registered and estimated man hours for modules III and V, the charts displayed in Figure 30
can be drawn, allowing for conclusions regarding conformity of the historical and estimated
data.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 147/153
20%
Casco work
General outfitting
Piping
3%
53% Supervision and coordination
Commissioning
Cabling
20%
Casco work
41% General outfitting
Supervision and coordination
56%
Commissioning
Figure 30 Work content division: planned and registered hours for PKR1 modules III and V
Quantification of Manufacturing Complexity in Shipbuilding Projects page 148/153
Taking into account these remarks, the following ratios between estimations and registered
hours become clear:
Conclusions
Both the totals as well as the proportions of work content compare well between the project
department estimates and the actually registered production hours. The assumption has to be
made that man hours are properly registered; there is no way to accurately check the man hour
values in a detailed manner as no other projects within DSNS offered a similar overview of
work content, and competitors, understandably, treat such information as highly confidential.
Moreover, as the estimation method is based on relative values and work content evolutions,
accuracy of the total amount of man hours is not critical, as long as information regarding
proportions between different work categories is reliable.
Taking into account the conformity between estimations by the project department and
registered man hours, it is assumed that DSNS estimates are sufficiently accurate to predict
work content for ships and modules built the at the Damen Schelde yard.
The given division of man hours is quite general. Work content divisions on a more detailed
level partly depend on personal judgement; an aspect which could be improved by more
detailed hour registration and further fine-tuning of the complexity estimation model.
Quantification of Manufacturing Complexity in Shipbuilding Projects page 149/153
APPENDIX M ADAPTING DATA FOR PKR1 (YN 414) FOR USE AS INPUT DATA
To make the results of the model relevant to the DSNS project department and easy to
interpret, the high-level division of the Damen SWBS structure is maintained for definition of the
applied complexity quantification model. This structure is devised as follows:
In historical data and estimations, the abovementioned structure is not always followed; for the
man hour estimation for PKR1 for example, expected work content is given per general task
type (e.g. furnishing) and per module. The actual hours were registered in a more detailed, but
differently categorised fashion. Combining these data for verification (as presented in Appendix
L) and for sufficiently detailed definition of an estimation model requires considerable
processing.
The following are the categories maintained in the DSNS estimations for PKR1:
Steel processing
Section building and hull assembly
Small steel work
Outfitting
Piping
Quality control
Project department DSNS
Quantification of Manufacturing Complexity in Shipbuilding Projects page 150/153
For each of these categories, work hours are specified per module. As it is known which major
components are located in which modules, and most modules are included to meet a specific
function (for example, module III is the engine room module), a substantiated, more detailed
categorisation can be made, taking into account the even more detailed, registered man hour
data for module III and V, and data regarding components (for example expressing the
proportions between cable lengths for different systems).
In Table 21, tasks and components which can be considered determining for the amount of
man hours, and which could be deducted from the source data, are listed with their processed
estimated work content. Not all components of influence are listed, only those who could be
directly linked to a figure concerning work content are given. The complexity quantification
model is based on more components which have influence (through differentiation or
interdependency) on the items listed in Table 21.
Remarks:
5. Historical data: ratio mechanical work auxiliary plant/(power plant + auxiliary plant) =
0.69. Also takes into account ancillary equipment. It can be assumed that most of the
mechanical work in module III is related to power supply and propulsion
6. Assumption: exhaust system/RHIB mountings 50/50
7. Historical data: ratio mechanical work auxiliary plant/(power plant + auxiliary plant) =
0.31. Also takes into account ancillary equipment. It can be assumed that most of the
mechanical work in module III is related to power supply and propulsion
8. File estimating cable lengths: 800 m cable reduction if one diesel generator is
eliminated, work content fraction 0.005147 of electrical installation for all generators
9. Assumption: all remaining cabling module III = power grid, remaining cabling modules
IV and V = networks, other modules: 50/50. Based on cable estimation file
10. File cabling, fraction 0.115815
11. File cabling, fraction 0.059304. Includes emergency stops, earthing, tank level
measuring, signal panels, watertight door system
12. File cabling, ratio: instrumentation/total cable length = 0.142839. Assumption: I/O
fraction constant for all modules
13. File cabling: degaussing = 0.081978 of total cable lengths. Assuming a linear relation
between cable length and work content
14. Assumption: 40% of outfitting for Exocet (anti-surface missile) installations
15. Includes outfitting, quality control, commissioning, preservation, furnishing, insulation
16. Factor 0.8: some work is connected to accommodation part (galley) of module III
17. Assumption: work on tenders equals 2 times the work on RHIB installations
18. Proportional to number of systems, complement and design displacement
The “cable estimation file” to which is reference is made is the total cable calculation for the
JSS Karel Doorman, a project in which DSNS was involved as well. It can be said that this
vessel has a higher degree of automation than the PKR1, however many of the inputs and
outputs related to this aspect are virtual or combined. In other words, increased levels of
automation do not necessarily entail fundamental increments in physical cable lengths, hence
the relative work content and divisions of cabling within the domain of electrical installations are
assumed to be in the same order of magnitude.
On themselves, the remarks do not shed full light on the subject; for instance, the work
fractions mentioned in the remarks refer to man hours per source category and these are not
given in this report. Study of the calculation files is necessary to obtain full insight in the way
man hour estimations were processed. In this report, no further elaboration will be given for
reasons of confidentiality.