Design Modelling
TMR4115
Thursday, August 28 2006
Prof. Stein Ove Erikstad
Reality
Decision Models
Design Methods Prof. Stein Ove Erikstad
Agenda
Recapturing last week
characteristics marine systems design
Modelling the design process
basic building blocks
Modelling in design
basic model elements
modelling examples
2
Design Methods Prof. Stein Ove Erikstad
Models and Methods in
Design
Reality
understand
improve
DECIDE
understand the
problem
select the
correct model
focus on what is
important
simplify
Decision models
(and their respective
method)
Design Methods Prof. Stein Ove Erikstad
The Design Paradox
Postpone decisions
Increase knowledge
4
Design Methods Prof. Stein Ove Erikstad
Syntax vs Semantics
(hull form & propeller) (required SHP)
(hull form) (seakeeping behaviour)
(hull form, propeller, machinery) (ship speed)
(all ship systems) (total cost)
5
Design Methods Prof. Stein Ove Erikstad
Activities and Events
Clarification of
task
Design
initiation
Concept definition
Outline
Specification
Request for
tender
Conceptual
design
Problem description
(Statement of need)
Outline specification
(Tender invitation)
Tendering/sales
Project development
Embodiment
design
Contract
specification
Build project
Class design
Engineering design
Fabrication engineering
Procurement
Materials management
Fabrication
6
Design Methods Prof. Stein Ove Erikstad
The basic design process
Reference model
GENERATE
DECIDE
ANALYSE
EVALUATE
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
7
Design Methods Prof. Stein Ove Erikstad
Design vocabulary
V
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
8
Design Methods Prof. Stein Ove Erikstad
Example Vocabulary Elements
Name
Unit
[Link]
[m]
Bd
Breadthofdeck
[m]
BM
Distancebetweenverticalcentre
[m]
Bp
Breadthofpontoons
[m]
dc
Diameterofcolumns
[m]
Dp
Depthofpontoon
[m]
KB
DistancekeelandVCB
[m]
KG
DistancekeelKandVCG
[m]
Ld
Lengthofdeck
[t]
Lp
Lengthofpontoon
[m]
Ifunc
V
GENERATE
Kgen
Symbol
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
9
Design Methods Prof. Stein Ove Erikstad
Semisub example
Ld
Bd
h ag
Column
height hc
Top
dc
dc
Dp
Tsur
Ttr
Lp
Bp
Lp
Design Methods Prof. Stein Ove Erikstad
10
Syntactic Knowledge Ks
Design alternatives
materials
components
variants
Design ranges
Logical constraints
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
11
Design Methods Prof. Stein Ove Erikstad
Syntactic Knowledge Ks
Semisub sample
Design alternatives
4, 6, 8 columns
circular versus quadratic columns
Design ranges
column diameter 8-15 m
Pontoon length 80-120m
Logical constraints
Bp < B
Bc < Bp
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
12
Design Methods Prof. Stein Ove Erikstad
Design description SemiSub
Decision space
*INITIAL POINT**
Current Point:
Values of design variables:
x( 1)= 80.000000
x( 2)= 12.000000
x( 3)= 10.000000
x( 4)= 10.000000
x( 5)= 25.000000
x( 6)= 50.000000
x( 7)= 70.000000
x( 8)= 60.000000
x( 9)= 40.000000
x(10)= 5.0000000
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
13
Design Methods Prof. Stein Ove Erikstad
Other examples Design
Descriptions
Contract specification
CAD model
Equipment data sheets
P/ID diagrams
Design Methods Prof. Stein Ove Erikstad
14
What is then the next
step?
We have now described what
the design look likes
We now need to describe
what it can do
i.e. from FORM to FUNCTION
15
Design Methods Prof. Stein Ove Erikstad
Design Interpretation SemiSub
VDL
B
Dc
Analysis
Dp
Bp
Steel weight of the pontoons:
Steel weight of the columns:
Steel weight deck structure:
Total steel weight:
Displacement, oper. cond.:
Heave period operation:
Variable Deck load
Ifunc
V
GENERATE
Kgen
307.17
380.26
961.80
2099.23
9251.03
20.32
1200 t
ANALYSE
GEN
Kana
Igoal
ANA
EVALUATE
EVAL
Ieval
DECIDE
DEC
Keval
16
Design Methods Prof. Stein Ove Erikstad
Analysis knowledge
Knowledge that relates
FORM and FUNCTION
Examples
functional relations
finite element analysis
comutational fluid
dynamics
systems simulations
Both the relations
themselves, and how to
derive such relations
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
Keval
Design Methods Prof. Stein Ove Erikstad
EVALUATE
EVAL
Ieval
DECIDE
DEC
17
Semisub example
Ld
Bd
h ag
Column
height hc
Top
dc
dc
Dp
Tsur
Ttr
Lp
Bp
Lp
18
Design Methods Prof. Stein Ove Erikstad
Whats next?
We now have derived some
measure of the performance
of our design. How can we
use this to make decisions?
-> We need to evaluate this
against our design goals and
requirements
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
Keval
Design Methods Prof. Stein Ove Erikstad
EVALUATE
EVAL
Ieval
DECIDE
DEC
19
Design interpretations
I
Intended interpreations
goals, objectives, constraints
I.e. what we want to achieve, requirements
Inferred interpretations
I.e. the actual performance of the current
design based on analysis
20
Design Methods Prof. Stein Ove Erikstad
Semisub example
High Variable Deck Load (VDL): The main purpose of the
vessel is to provide a working area for drilling operations. To
perform the drilling operation a number of consumables are
needed, e.g. drill pipes, drilling mud, fuel, water supplies. A high
VDL is advantageous because the time interval for supplies can be
longer.
Large deck area: A large deck area will provide space for
storage of consumables and equipment, and working space.
Alternatively, the deck area can be handled as a user requirement,
and hence be modeled as a constraint.
Low construction costs: the construction costs of the vessel
is difficult to model. A common approach is to approximate these
costs as a linear function of the vessels Light Weight (LW) (steel +
machinery + outfit weight). Since the Machinery and outfit
weights are assumed fixed, the cost will be a function of the steel
weight only.
High availability: in order to carry out the drilling operation,
the motion of the deck area must be within certain range. Within
this range, the motions of the platform can be compensated for by
the use of a heave compensator. Outside this range, the drilling
operation must be stopped, and eventually the vessel must be deballasted to survival draft.
The two main governing factors of the availability will be the
motion characteristics of the vessel, and the airgap between the
water surface and the platform deck.
21
Design Methods Prof. Stein Ove Erikstad
Example
Semisub goals and reqs
Maximize deck load and minimize building cost
VDL
WS p WSc WSd
min f wVDL 1
wSW
1
tar
VDLtar
WS
Requirement for max heave period
2L p Bp D p cm 1 dc 2 T D p
op
9.81 dc2
treq 0
22
Design Methods Prof. Stein Ove Erikstad
Representational spaces
the design space, capturing the description of the design object.
The ultimate goal of the design process
the performance space, holding the functional interpretation of the
design object, that is, objective measures of performance
usually derived from design analysis
the goal space, representing the intentions and requirements of
the designer, and gives the design process a direction towards
purposeful solutions. The goal space contains modal design
relations instead of factual statements (e.g. Ship has length 50
m) , the relation is modified to express statements about
requirements and expectations about the design object (e.g. Ship
should have capacity 3000 TEU, Ship must have B < 32.2 m)
the value space, holding an interpretation of the design object
with respect to subjective measures of performance, based on
the synthesis of the functional interpretation and the design goals
performance
space
value space
design
space
goal space
23
Design Methods Prof. Stein Ove Erikstad
DECIDE
Based on the evaluation of
the design performance vs.
goals, requirements...
... the current design solution
is either selected, or a new
iteration is started
Ifunc
V
GENERATE
Kgen
ANALYSE
GEN
Kana
Igoal
ANA
Keval
Design Methods Prof. Stein Ove Erikstad
EVALUATE
EVAL
Ieval
DECIDE
DEC
24
Summary
Classification building blocks
ShipXKnowledge
Entities
Control level
knowledge
Object level
knowledge
Design Object
Entities
GEN
Domain Knowledge
Entities
ANA
DEC
Ifunc
Igoal
Ieval
Kgen
Kana
EVAL
Keval
25
Design Methods Prof. Stein Ove Erikstad
Why?
The PURPOSE has been to identify
the main BUILDING BLOCKS in
the design PROCESS
These are found in most design
processes, though their sequence
and relations may vary
Thil will be the subject throughout
this semester
26
Design Methods Prof. Stein Ove Erikstad
Basic inference processes
Design Methods Prof. Stein Ove Erikstad
Inference processes in
design: Deduction
I = 1(Ki, D)
(2.1)
Ki:
x DW(x) V(x) Resistance(x) = f(DW(x), V(x))
D:
DW(ShipA) = 200.000 V(ShipA) = 15
I:
Resistance(ShipA) = 1200
E.g. design analysis
28
Design Methods Prof. Stein Ove Erikstad
Inference processes in design:
Generating designs
D = 3(Ks,V) (2.3)
E.g. systematic parameter
variation
29
Design Methods Prof. Stein Ove Erikstad
Inference processes in design:
Deriving design descriptions
D = 2(Ki, I) (2.2)
Ki:
x DW(x) V(x) Resistance(x) = f(DW(x), V(x))
I:
Resistance(ShipA) = 1200
D:
DW(ShipA) = 200.000 V(ShipA) = 15
E.g. linear optimisation,
simplex from outside the
system
30
Design Methods Prof. Stein Ove Erikstad
Inference processes in design:
Acquiring design knowledge
Ki = 5({D1, I1}, {D2, I2}, ...)
D1, I1: DW(ShipA) = 200.000 V(ShipA) = 15, Resistance(ShipA) = 1200
D2, I2: DW(ShipB) = 212.000 V(ShipB) = 14.8, Resistance(ShipB) = 1240
D3, I3:
Ki:
x DW(x) V(x) Resistance(x) = f(DW(x), V(x))
D1:
V(ShipA) = 15.0 IsType(ShipA, TypeX)
D2:
V(ShipB) = 14.8 IsType(ShipB, TypeX)
D3:
Ks:
x IsType(x, TypeX) V(x) isTypically around 15 knots
E.g. towing tank
Design Methods Prof. Stein Ove Erikstad
31