Customer Needs Driven Engineering Design and Shape Optimisation Hakki Eres Computational Engineering and Design Research Group Faculty of Engineering and the Environment University of Southampton Hakki.Eres@southampton.ac.uk BCAM Conference on Turbomachinery and Aerodynamic Shape Optimisation 28 March 2014 - Bilbao, Spain
Outline Value-Driven Design (VDD) effort in the CRESCENDO project (~15 minutes) Condition of supply (COS) shape optimisation work in the SAMULET project (~15 minutes) Concept Design Analysis (CODA) method (~10 minutes) Conclusions (~5 minutes) Questions (~10 minutes)
VALUE-DRIVEN DESIGN - CRESCENDO
CRESCENDO WP2.2 CRESCENDO is an EU co-funded R&T project Total budget: 55 million Euros Launched in May 2009, ended November 2012 Led by Airbus and brings together 59 organisations from 13 different countries Workpackage 2.2 is the Requirements Establishment Task 2.2.2 is the Value generation modelling and simulation Demonstrator: The Simulation of value for alternative engine architectures
Research Motivations Research Motivations: Demonstration the integration of unit cost, maintenance cost, and surplus value models into engine preliminary process Feasibility studies of modelling overall design merit in early concept design Challenge: Can we employ value based methods and models to assess and rank order different design solutions? Conventional Engine More Electric Engine (MEE)
Simulation Models All models are Vanguard implementations They are all accessible through web services Integration uses Isight web services component Unit Cost Maintenance Cost Unit Cost Model Automatic model generation Hierarchical model Maintenance Cost Model Stochastic and hierarchical model Development and labour costs Surplus Value Model Fleet level, economics based profitability model Used as the design goodness metric
Model Hierarchy Component level model Engine Hierarchy Modules Components Fictitious data Fan blades Module 01: Low Pressure Compressor Fan Disc Module level cost splits Engine Module 02: Intermediate Pressure Compressor IPC blades IPC discs Fictitious data Module 03: Intermediate Case Interchange Struts Bearings Engine level Monte Carlo simulations Fictitious data
Stochastic Maintenance Cost Model Auto generation of cost models based on cost data Module level exposure and component level maintenance rates Module level exposure rates Component level maintenance rates Fictitious data Constant module exposure rates (Fictitious data) Triangular distributions on component maintenance rates (Fictitious data)
Component Model (Video) Fictitious data
Whole Engine Model (Video) Fictitious data
Browser View (Video) Fictitious data
Surplus Value (SV) Model SV model uses engine, aircraft and operational parameters as inputs and calculates the profit generated by a fleet of aircraft for a given operational period Web services based integration to workflow execution environments (Isight) is possible Fictitious data
Simulation Models in Isight Two scenarios: Conventional Engine More Electrical Engine Unit and maintenance cost results are fed into the SV model through data links
Running the Isight Workflow (Video) Fictitious data
SHAPE OPTIMISATION - SAMULET
SAMULET WP5.6.3 SAMULET is funded by TSB and EPSRC Total budget: About 90 Million GBP Led by Rolls-Royce Workpackage 5.6.3 Optimisation of condition of supply (COS), black forging (BF) and heat treatment (HT) profiles based on a given disc profile Ultrasonic forging inspection and cost modelling Isight integration
High Pressure Turbine Discs
Ultrasonic Scanning Model Forged profiles of given discs can be analysed to find the ultrasonic scan success rates Fictitious data
Integrated Workflow Conditional branches for each optimisation strategy Conditional branches for each N={6,.., 15}
Initial Optimisation Conditional branches are used to run one of the following optimisation tasks Cost and Inspectability: Objectives: Minimise total unit cost and maximise full scan rate Variable: Number of edges Forging Volume Objective: Minimise black forging volume Variable: Number of edges Inspectability Objective: Maximise full scan rate Variable: Number of edges
Cost and Inspectability This optimisation task has five sub-tasks COS generation, ultrasonic scanning analysis, BF/HT generation, calculation of geometric parameters, and cost analysis
Further Optimisation Conditional branches are used for every possible value for OptimumNumberOfEdges parameter Currently for each N={6,.., 15} Each sub-task contains another optimisation process specifically configured for the value of N Adding and configuring a new optimisation task takes about 10 minutes
Sample Optimisation Task (N=12) Most of the optimisation tasks are identical to the one in the Initial Optimisation task Only COS Generation task is different
Sample Results Fictitious data Scenario N Initial Optimisation Further Optimisation Cost and Inspectability 14 Forging Volume 15 Inspectability 14
Concept Design Analysis (CODA) CODA METHOD
CODA Overview CODA (Concept Design Analysis) method aids the conceptual design and selection phase within new product development CODA evolved from QFD (Quality Function Deployment) method In order to calculate the value of a design we need two orthogonal measures: The cost of delivering the product Performance (i.e. How well the product will meet the needs of customer) 26
Design Merit CODA Method CODA uses a Merit Function to measure the overall worth of the design. 100% f(x) 1 and 2 Three types of merit functions are used 1. Maximise 3 and 4 2. Minimise 3. Optimise 0% Increasing levels of 1. Maximum takeoff weight 2. Sea level static thrust of the engines 3. Specific fuel consumption of the engine 4. Cruise viscous drag coefficient of the aircraft 5. Cockpit illumination level 6. Legroom in the economy class cabin 5 and 6 x
Maximise function Examples Reliability Corrosion resistance Life MTBF 28
Minimise function Examples Fuel consumption Noise Pollution 29
Optimise function Examples Tension in racquet strings Oversteer /understeer Suspension damping Mobile phone keypad size Cockpit lighting 30
Assessing Customer Needs A binary weighting matrix or Analytical Hierarchy Process (AHP) can be used to assess the importance of CNs
Overall Design Merit Overall design merit ECs ECs ECs
CODA Driven Workflow Problem: Find a COS profile based on a given disc profile by considering ultrasonic inspection rate, black forging volume, and unit cost. Design merit is the single objective
CODA Driven Optimisation
Customer Needs
Overall Design Merit Cheap to manufacture Better scanning rates Minimise forging volume Design Merit Calculations Overall design merit is a measure of how well the customer needs are satisfied on a percentage scale. Fictitious data 50.95% Engineering Characteristics Unit cost Scan success rate Forging volume Normalised needs data 20.00% 60.00% 20.00% Correlation 0.9 Value 120000 Relationship Type Min Upper Limit 150000 Neutral or optimum point 125000 Lower Limit 50000 Tolerance Merit value 51% 0% 0% Correlation 0.9 Value 0.9 Relationship Type Max Upper Limit 1 Neutral or optimum point 90% Lower Limit 0 Tolerance Merit value 0% 50% 0% Correlation 0.9 Value 50000000 Relationship Type Min Upper Limit 70000000 Neutral or optimum point 55000000 Lower Limit 40000000 Tolerance Merit value 0% 0% 53% Relevant engineering characteristics (EC) are mapped to corresponding customer needs through correlation factors and relationship functions (Max, Min, Opt)
Excel Model Model outputs Fictitious data Variable inputs
CODA Analysis Unit Cost Scan Success Rate Design Merit Forging volume
Final COS Profiles Fictitious disc profile and data Better scanning rates 90% Better scanning rates 5% Better scanning rates 5% Better scanning rates 40% Cheap to manufacture 5% Cheap to manufacture 90% Cheap to manufacture 5% Cheap to manufacture 30% Minimise forging volume 5% Minimise forging volume 5% Minimise forging volume 90% Minimise forging volume 30%
Conclusions Surplus value model for a fleet of aircraft can be utilised to assess and rank order different design solutions Conventional versus more electric engine Concept Design Analysis (CODA) model allows interactive exploration of different design alternatives CODA models can be used to steer multi-objective design optimisation problems by considering different customer needs
Questions