Вы находитесь на странице: 1из 27

Design of High Performance Craft

from a Human Factors Perspective

T.R. D’Arcy, I.R. House, P.W. Kingsland, R.J.H. Spink,


M. Yuceulug, D.A. Hudson, R.A. Shenoi

School of Engineering Sciences, Ship Science, University of


Southampton, Southampton, UK.

ABCD Symposium, Panama City, Fl


25-26th April, 2006

1
Overview
Overview
• Background
• Design procedures
• Design tools
– Theoretical method, model tests, full-scale tests
• Human performance measure
• Design outcomes
• Conclusions
• Further work

2
Motivation
Motivation
• Modern small, very high-speed vessels:
– Fatigue, injury, long-term pain
• Quantify effects on operator
– Heart rate, blood chemistry, muscle
fatigue, oxygen uptake
• Link to naval architecture attributes
– Hull form, sea-state, operating manner
• Include at design stage
3
Design
Design Process
Process
Market need

Ship owner

Shipyard

Hullform

Comfort requirements

Model Basin

Final Design

Sea Trials 4
Design
Design Process
Process
Ship owner

Statement of
requirements
Preliminary Comfort
parameters requirements

Genetic algorithm

Preliminary hullform(s)
Tank testing Numerical code
Performance prediction

Final Design

5
Sea Trials
Genetic
Genetic Algorithm
Algorithm
• Numerical model of the theory of evolution
• Search space optimisation routine
• Fast
• Multi-variate
• Flexible, adaptable
• Proven in many fields
• No need for continuous design space
• 3 distinct operators dependent on a measure of merit:
– Selection
– Crossover
– Mutation (1/10000)

6
Genetic
Genetic Algorithm
Algorithm
• Sectional area curve to define hulls
• Fix draught - vary beam, deadrise
• Create ‘generation’ of hulls
• Predict motions
• Calculate vibration dose values
• Define ‘measure of merit’ – ’selection’
• Highest values to next ‘generation’ –
‘crossover’
• Iterate process
7
Genetic
Genetic Algorithm
Algorithm
Sectional area curve
Create generation Fix draught - vary
beam, deadrise
of hullforms

Highest values of
‘measure of Predict motions
merit’

Calculate
Vibration Dose Values 8
Theoretical
Theoretical Method
Method
• 2D Strip theory (after Zarnick)
– Wedge impact model for added mass/damping
• Non-linear, time domain
• Regular or irregular wave simulations
• Well known for ‘conventional planing hulls’
– Validated vs Fridsma, etc.

• Limits for speed?


• Limits for hull form (RIBs, VSV, etc.)?

9
Case
Case Study
Study
• Assisting ‘Kali’ Challenge
– Gas turbine propelled wave-piercing RIB
– Attempt Round Britain <30ft record
• Compare to existing record-holder
– ‘Titan’ – conventional RIB form

10
Model
Model Testing
Testing
WAL/GKN tank – up to 12 m/s
– Calm water and regular/irregular waves
– Heave and pitch displacements at CG
– Vertical acceleration CG and Bow

CALM REGULAR IRREGULAR


MS = Model Scale
FS = Full Scale
MS Speed MS Speed MS Sig Wave Ht MS Modal Period MS Speed FS Sig Wave Ht FS Modal Period
m/s m/s m s m/s m s

KALI 1,2,3,4,5,6,7,9,11 8.11 0.035 1.08, 1.53, 2.16, 2.65 8.11 0.5 6
12 12.17 0.035 1.53, 2.16, 2.65 12.17 0.5 6
12.17 0.75 6
12.17 1 6
TITAN 8.41 0.5 6
6.91 0.8 4
3.45 0.94 2.75

11
Model
Model Testing
Testing

Conventional RIB Kali at 52kts


form at 45kts
12
Results
Results
Kali Wave Height & CG Acceleration Trace
FS Sig Wave Ht 0.5m FS Modal Period 6s
Vs = 30 knots
80 3

60 Wave Height (mm) 2.5


Acc CG

40 2

20 1.5

CG Acceleration (g)
Wave Height (mm)

0 1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

-20 0.5

-40 0

-60 -0.5

-80 -1

-100 -1.5
Time (s)

13
Results
Results
Titan Wave Height & Acceleration Trace
FS Sig Wave Ht 0.8m FS Modal Period 4s
Full Scale Speed 30 knots

150 12

Wave Height
CG Acceleration 10
100 Bow Acceleration

50
6
Wave Height (mm)

Acceleration (g)
0 4
0 2 4 6 8 10 12

2
-50

-100
-2

-150 -4
Time (s)

14
Validation
Validation
Comparison of Code and Tank RAO Results for Kali at 12.17m/s with Wave Height 0.035m

1.8

1.6

1.4

1.2
Code Pitch
R.A.O.

Tank Pitch
1
Code Heave
Tank Heave
0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14
Wavelength/Boatlength

15
Full
Full Scale
Scale Testing
Testing –– why?
why?

• Develop measurement capability


• Further validation of code through RMS
comparisons
• Establish ‘link’ between human
performance and boat motions
• Acquire data in beam/following seas

16
Full-Scale
Full-Scale Testing
Testing
General Arrangement
8.8m RIB, 30Kts (max.)
Full
Full Scale
Scale Testing
Testing
• Robust measurement system
– 11 channels accelerations
– Wave buoy data
– GPS track
– Heart-rate of crew

Conventional RIB
form at 30kts

18
Full
Full Scale
Scale Testing
Testing

Conventional RIB form at


20kts, head seas
19
Full
Full Scale
Scale Testing
Testing
PSD of z-axis Titan full scale, run num ber 18
3.5

2.5
(ms-2)²/Hz

1.5

0.5

0
0 50 100 150 200 250
Effect of 50Hz Low Pass Butterworth Filter Hz
1
10
PSD of z-axis Titan full scale, run number 21
PSD of BUTTERW ORTH filtered z-axis Titan full scale, run number 21
0
10

-1
10

-2
10
(ms-2)²/Hz

-3
10

-4
10

-5
10

-6
10

20
0 1 2
10 10 10
Hz
Full
Full Scale
Scale Testing
Testing
10
Vertical accleration at bow
Vertical acceleration at CG

slam impact
Acceletation (g)

slam at bow,
no slam no slam at CG
4

-2
100 102 104 106 108 110 112 114 116
Time (s)

Ave. of 1/3 highest Ave. of 1/10 highest


Peak (g) RMS (g) (g) (g)
CGZ BowZ CGZ BowZ CGZ BowZ CGZ BowZ
Head 5.54 9.02 0.42 0.72 3.75 8.11 4.89 9.02
Beam 5.09 8.79 0.41 0.61 2.81 6.68 3.77 7.92
Follow 2.71 7.57 0.37 0.43 1.78 5.37 2.24 6.81
Vibration
Vibration Dose
Dose Value
Value (VDV)
(VDV) –– why?
why?
• No direct physical measurement (no ethics approval)
• Uses full acceleration history for each run
• Apply specific weightings to each axis
• Current British Standard for whole body vibration
• Other modes of transport
• Reliable comparisons between similar data sets
• Applicable for full continuum of motions from isolated
shocks to long duration vibration
• Assess cumulative effects of motions, including those with
high crest factors

22
Measure
Measure of
of Human
Human Comfort
Comfort
• Calculate VDV from full-scale data
• Limiting time from experienced crew
• Define limiting VDV
• Combine with limiting RMS measures in Weighting
‘measure of merit’ factors
   
1  3   1 
Measure of merit = time × ×   +  a 
2  a 1 LCG
 10   110 BOW 
Time to
limiting VDV
Ave. of highest 1/10th values
Apply to rank designs in each ‘generation’
23
Conclusions
Conclusions

• Design procedure for human performance


– Genetic algorithm
• Supported by
– Theoretical predictions
– Model testing
– Full scale testing
• Develop single ‘measure of merit’
• Readily adaptable for improved data

24
Further
Further Work
Work

• Hull design method ?


• Genetic algorithm – vary more parameters
• Extend validation database
• Improve motions prediction tool
• Include more ‘human factors’ data/methods –
i.e. improve ‘measure of merit’
• Multi-criteria optimisation

25
Thanks
Thanks –– and
and questions
questions
• T.R. D’Arcy • J.I.R. Blake
• I.R. House • S. Lewis
• P.W. Kingsland • W. Wang
• R.J.H. Spink • ‘Kali’ Challenge
• M. Yuceulug – Simon Sanderson
– Jon Aldiss

26
Thanks
Thanks -- and
and questions
questions
• UK Engineering and Physical Sciences
Research Council
– University College Chichester
– Institute of Sound and Vibration Research,
University of Southampton
• School of Engineering Sciences, University of
Southampton
• Royal Academy of Engineering
• ‘Kali’ Challenge
– Simon Sanderson
– Jon Aldiss

27

Вам также может понравиться