American Institute of Aeronautics and Astronautics1

Internet Based Automated Design Optimization Setup to Support Integration of Customer in the Design of User

Customizable Products.

Jiju A. Ninan* and Zahed Siddique†

School of Aerospace and Mechanical Engineering, University of Oklahoma, Norman, 73019

Integration of customers is a necessary element to design and produce customer centric products. Some of the challenges faced during the customer/user integration into the design process include: (1) in many instances customers do not have any knowledge on how to design the product, and (2) the customer specified configurations needs to be feasible and optimized for performance. To achieve customization of products, on real time, the design evaluation tools need to be automated and enhanced to enable operation over the internet. The focus of this paper is to present a framework that uses finite element based optimization tools to integrate the customer into the design process via internet for delivering user customized products. For the purpose of automating the analysis and optimization a product family CAD/FEA template was developed. The template generalizes the geometry building and analysis of each configurations developed using a product platform approach. The proposed setup is demonstrated through the use of a bicycle frame family. In this paper the focus is on the application of optimization and FEA to facilitate design of customer centric products.

I. Introductionass customization features customer centric products which can be altered or changed by the customer to fit his/her needs. The idea is to satisfy an increasingly heterogeneous market by offering customizable products. To achieve customization it is necessary to integrate the customer into the design process so that the product

can be tailored to his/her needs. Some of the challenges faced during the customer/user integration into the design process include: (1) in many instances customers do not have any knowledge on how to design the product, and (2) the customer specified configurations needs to be feasible and optimized for performance. Design tools and methodologies need to be altered to accommodate customer into the process of designing customized products. New tools need to be developed which can communicate with the customer and collect important information regarding the customization of the product. This requires a user interface to communicate with customers, who are often the lay man, in a non technical manner.

Internet can be used to communicate and integrate the user in a real time basis. Commercially used design evaluation and optimization tools are resource intensive and cannot be operated through internet. Moreover the user may not be skilled to operate engineering software. To achieve customization of products, on real time, the design evaluation tools need to be automated and enhanced to enable operation over the internet. The focus of this paper is to present a framework that uses finite element based optimization tools to integrate the customer into the design process via internet for delivering mass customized products.

The idea of mass customization and allowing the customer to specify his/her requirements can be a very challenging task. Before initiating mass customization for a set of products, questions that need to be answered include: Is the configuration specified by the user feasible? Is the customer specified configuration optimum? Consider a product where key dimensions can have large structural implications, allowing a person with no knowledge in engineering to choose the specifications can create unstable structures. In this case, the user specified configuration has to be analyzed and checked for feasibility. Moreover user specified design needs to be optimized, subject to design and manufacturing constrains. Having a design analyst analyze each specified configuration is not an efficient solution to the problem. In this paper we suggest a framework, where user selections and specifications regarding the customization of the product are collected using a web based interface. These parameters and selections are then used to automatically build the CAD model of the product variety. The customer specified

* Graduate Research Assistant, School of Aerospace and Mechanical Engineering, 865 ASP Ave, Norman, OK.† Assistant Professor, Department School of Aerospace and Mechanical Engineering, 865 ASP Ave, Norman, OK.

M

10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference30 August - 1 September 2004, Albany, New York

AIAA 2004-4317

Copyright © 2004 by Zahed Siddique. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

American Institute of Aeronautics and Astronautics2

configuration is optimized for weight and performance. If the optimization procedure does not yield a feasible design the user is asked to select the dimensions that can be altered. The optimization is then performed using the selected dimensions as design variables. In case of a feasible design CAD model of the configuration, geometric dimensions and specifications of the customized design are shown to customer as feedback. Figure 1 shows the employed optimization and iteration procedure.

The process of optimization, using finite element analysis based techniques, consist of several steps. First a CAD model of the configuration is generated using the dimensions and options specified by the customer. The geometry is meshed, material properties and boundary conditions are applied to the model. The analysis is then performed, results evaluated, and the analysis file passed to the optimizer. The configuration is optimized subject to the design and manufacturing constrains. Utilizing a Finite Element specialist to optimize each and every configuration specified by the customer is not a practical solution. The most obvious solution is to automate the process of analysis. Since the products family concept features commonality, general rules can be specified for the analysis of all family members. For the purpose of automating the analysis and optimization a product family CAD/FEA template was developed. The template generalizes the geometry building and analysis of each configurations developed using a product platform approach. Suitable algorithms, which show the implementation of the template to perform automated Finite Element Analysis, are also presented in the paper. In this paper the focus is on the application of optimization and FEA to facilitate design of customer centric products.

II. BackgroundA product platform is defined as “a collection of the common elements, especially the underlying core

technology, implemented across a range of products,”1. Product family using a platform, for both top-down and bottom-up approach, is provided through addition/substitution/deletion of modules from the platform2-5 or by stretching/shrinking the platform6. Industrial applications of both modularity based and scale-based approaches7have been presented in literature. Several optimization approaches have been developed by researchers to determine the best values design variable settings for the product platform Simpson, et al.8developed a web-based customization system for refiner plates for pulp and paper processing. Flores, et al.9 presented a similar web-based system for customizing, based on parametric modifications, for coated steel belt sheaves. Zha and Lu10 developed a web-based knowledge system to support product family design for a power supply product family.11-16

Saxon and Beaulieu17 developed a web based engineering system which can automatically perform finite element analysis of stabilizer bars. The system was developed as an in house design tool at ArvinMeritor Light vehicle systems, Simulation and Analysis (LVS, S&A) department to perform automated analysis of stabilizer bars. The

No

Collect user specified

parameters and options

Build FE Model & Optimize

Resulting Configuration

feasible?

Build CAD model of resulting geometry, communicate results and document

Ask for dimensions that can be

varied

Build FEModel Add

DVs & Optimize

Resulting Configuration

feasible?

Build CAD model of resulting geometry, communicate results and document

Terminate

yes

yes

No

Figure 1. Optimization and iteration procedure

American Institute of Aeronautics and Astronautics3

web interface takes user inputs for the dimensions of the stabilizer bar and performs an automated finite element analysis of the configuration. The system returns an HTML report of the analysis to the user.

Siddique and Boddu18 presented an agent based framework to automatically 3D models of customer specified designs. Siddique and Zhou19presented a template based approach to represent the product family. A Product Family CAD (PF-CAD) module was developed to implement the design templates and automatically generate solid models of family members associated with the product platform. Siddique and Ninan20 extended the PF CAD template to accommodate the automated Finite Element Analysis of product families for internet based architecture to automated FEA of family members. The research presented in this paper extends this work.

III. General Framework for Internet Based Framework for Customer Centric Design and Optimization

The general architecture of the Internet based framework for customer centric design and optimization system is shown in Figure 2 .The main components of the system are: (1) Web based user interface to gather user preferences, (2) CAD software and API, (3) CAD database (4) Common database, (5) FEA software and API, (6) FE database, and (7) Design optimization module. The user interface is a web page that takes the user’s selections regarding the product. The user selected parameters are categorized as structural parameters, geometric parameters, or a combination of both. For example, the material preferences are categorized as structural, while dimensional parameters are a combination of both. The geometric parameters are passed to the CAD system, where the API for the CAD systems receives these parameters, invokes the CAD software, and builds the solid model of the customized product. The component files are then automatically exported as IGES file into a common a folder accessible to the FE software. To generate CAD models from user selections, and perform Finite Element Analysis, and optimization on these models, we need a set of well defined guidelines and instructions, which tell the software in a structured manner how to perform these tasks in different cases. To achieve this, a MC CAD/FEA/Opt template,

which generalizes the geometry construction and analysis information for the entire family members, was developed. To demonstrate the capability of the system, a case study on the customization of bike frames is presented in Section IV. The frames are custom designed to the suit the physique of the rider. The customer specified design is automatically checked for structural feasibility and optimized in real time. The results from the feasibility study and optimization are passed back to the user as feed back.

Figure 2. System architecture for internet based framework for customer centric design and optimization.

American Institute of Aeronautics and Astronautics4

IV. Case Study-Customizable Bike FramesThere is a simple, inarguable truth about cycling: Your performance, endurance and comfort on a bicycle are a

direct function of how well that bicycle fits riders proportions (http://www.cannondale.com/bikes/custbike/c2.html). The customized bikes were initially used by the pro racers to maximize the comfort and lessen the fatigue which enabled them to ride faster and longer. Due to the success of the customizable frames in increasing the efficiency of the cyclist, companies like Cannondale started offering customizable frames in the market for the general customers. The process of customization starts with specifying the key dimensions by the customer to fit his/her physique. The process of customizing starts with the customer specifying the dimensions of the frame. A virtual model of the frame is generated by using CAD software; the tubes are cut and welded to precisely match the custom geometry, treated, painted, finished and shipped. The interesting observation is that most of the companies offer only certain models to customize. To guarantee the structural rigidity of the customized designs the designers for example in Cannondale use CAD and Finite Element analysis tools. Finite element analysis of the possible models will be performed before hand anticipating all the possible configurations. Then possible range of dimensions to choose by the customer will be worked out. Therefore while specifying the dimensions the customers can only chose from a predefined set of choices. This raises certain key questions: How customizable are the designs? How to provide more freedom to the customer in the customization process while maintaining the structural rigidity of the product? Can optimization techniques be performed on the user specified configurations to improve the design thus improving the efficiency of the customized product?

Using the general architecture for Internet based Mass customization shown in Section III a MC CAD/FEA/Optimization set up for customization of bike frames was set up (Figure 3). The user interface which is a web based user interface built using ASP technology takes the following user inputs.

� The weight of the rider� Desired Arm Reach (Seat to Handle)� Total length of the frame� Total height of the frame� Floor to seat height� Wheel Radius� Clearance

These parameters are then passed to the FE software from the web page. Finite Element analysis and optimization of the user configuration is carried out in real time. If the user specifications lead to an infeasible configuration then the user is prompted to select the dimension that can be varied to generate a feasible configuration during the second iteration. A second iteration is carried out including the selected dimensions also as the design variables so that a feasible configuration can be derived at. In case of feasible configuration the dimensions and geometry of the resulting product is shown to the user as feedback. Three types of frames available for customization are: Men, Feminine, and Tandem. Section A explains the mass customization information setup in detail. The optimization macro used for automating the analysis and optimization of the user specified configuration is explained in Section E

A. Implementation of the General Setup using ANSYS and ASP The information sever set up consists of two servers running on Microsoft Windows 2000 operating system.

Server I holds the installation of the Internet Information systems . The user interface web page and associated files are located at the root folder of the IIS. The second server holds the installation of ANSYS. The user selections and parameters are collected by ASP. ASP then modified the MC optimization macro and passes it to Server II. Sever II checks for the appearance of the macro file and then invokes ANSYS running the Macro in Batch mode. After the analysis ANSYS outputs a results text file which is read by ASP, interprets and communicated it to the user through the web page. ASP plays a critical role as an agent between the user and the engineering software.

B. Web Based User InterfaceFigure 4 shows the web based user interface of the MC CAD/CAE/Customization system. Customization of

frames is a four step process. First step involves selecting the model to customize. Three models are available for customization- Men’s, Feminine and Tandem. In the second step user is prompted to enter information regarding the dimensions of the intended design and also riders weight. This information is then submitted to ANSYS . ASP plays a critical role in submitting these information as parameters to ANSYS and Invoke the software to start the analysis and optimization procedure. The role of ASP in the system is explained in Section C. If the user inputs do not lead to a feasible configuration after optimization in Step 3 user is prompted to select the dimensions specified in step 2 that

American Institute of Aeronautics and Astronautics5

can be varied. Then a new optimization study is performed with the selected parameters also as the design variables. They are allowed to vary within a range of + 10 % of its initial value. In case of a feasible design the dimensions of the resulting model and the geometry are shown to the user as a feedback. Optional analysis information can also be included in the HTML result file generated By ANSYS. Currently the system only allows two iterations. If the second optimization run (Step III) leads to infeasible configuration the system communicates the result and then prompts user to change geometry and exits out of the loop.

C. Role of ASPASP acts as the agent between the

user/customer and the engineering analysis software. The information collected through the web user interface need to be passed to ANSYS and then invoked to perform the analysis. The macro which automates the analysis and optimization resides in the working directory of ANSYS. The user inputs are passed to the macro using a parameter file. The parameter file is a text file which is modified by ASP each time user submits a configuration for analysis. The file monitoring system which checks for a new parameter file in server II (shown if Figure 3) invokes ANSYS passing these parameters. After performing the analysis ANSYS puts out a text file which contains the results of the analysis along with JPEG picture files of the configuration. ASP searches for new files to appear, opens and passes the information to the user. In case of optional Step 3 ASP modifies the part of the macro file which specifies the design variables to include those parameters check marked by the user. Section E explains the MC CAD/FEA/Optimization macro in detail.

D. CAD/FEA Template for Bicycle product family

A well built product family architecture forms the core for mass customizable products. Customization can be achieved in two ways. The first method involves breaking down the design into modules. Modules common to all family members are called the product platform. Product

Modify Macro

WEB Interface

IIS +ASP

SERVER I

ANSYS FEA &

Optimization Module

Perform optimization and O/P results

Results Folder

MC Optimization Macro

Read Results and communicate to user

SERVER II

Figure 3. Information set up and implementation of the system using ANSYS and ASP.

Figure 4. Web based user interface for customization showing Steps 1 and 2.

American Institute of Aeronautics and Astronautics6

variety is provided by adding optional modules requested by the customer. The second method of achieving customization is through scaling. Customers can scale certain dimensions or parameters to customize the product to suit their needs. A combination of these methods can be utilized to specify a product family with lots of variety. The framework proposed in Section III can handle both modular product family architecture and scaling. To generate CAD models and perform Finite Element Analysis from the user selections we need a set of well defined guidelines and instructions, which tell the CAD/FEA software in a structured manner how to perform the tasks in different cases. To achieve this, a product family CAD/FEA template, which generalizes the geometry construction and analysis information for the entire family members was developed (Table 1)

The template uses a top down approach to define the family members. On the top is the product family which comprises of the different family members PFMi.. Each family member can be generated by addition of optional modules to the product platform. For the purpose of generating the CAD and the Finite Element model, the assembly, geometry, material, mesh and loading information need to be included into the template. The elements of PFMi include - PP, PO, AssemCon and LoadCon. It means a member of product family should embrace the product platform (PP), utilize product options (PO) to provide the variety; assembly constraints (AssemCon) to specify spatial relationships among them; and the loading conditions (LoadCon) and boundary conditions (BoundaryCond). In the template it is assumed that the product platform (PP) is not null or empty set. To provide varieties across the product family the product option (PO) set cannot be empty either. The AssemCon is the assembly relation set which

Table 2. CAD/FEA Template for family of bicycle frames

PF= {Men’s, Feminine, Tandem}PF= {PP, PO, AssemRel, LoadCondA}PP= {ST, CST, CSB}PO= {.}

PFM1=Men’s= {PP, LTM, HTTM, HTM, AssemRel, LoadCondA}PFM2=Feminine= {PP, LTL, HTTL, HTL, AssemRel, LoadCondA}PFM3= Tandem= {PP, LTT, LTTF, HTTT, HTTF, HTT, STF, DTT, AssemRel, LoadCondA}}

LoadCondA=Loads applied at the assembly levelLoadCondA= {Ø} ! Since no extra loads are applied at the assembly level in this case

AssemRel =Define the geometric relation between each of the components, for example where each part is joined to each other etc

BoundCond1 = {Pt. of intersection of CST & CSB, all DOF, 0}BoundCond2 = {Lower End of HT, all DOF, 0}

LoadCond1= {Top of ST, net Y-direction, Point Load, weight of the rider}Loadcond2= {Point of intersection of Lower Tube and ST, neg Y-direction Point Load, Weight of the rider}LoadCond3= {Top o f FST, neg Y-direction, Point Load, weight of the rider}LoadCond4= {Point of intersection of Lower Tube Tandem and STF, neg Y-direction, Point Load, Weight of the rider}

Element Type= {Beam}

Material= {Aluminum T6}

Element Size= {6}! The Same Material, Element and Element Size assumed for all components

Product Platform Components (PF)Seat Tube STChain Stay Top CSTChain Stay Bottom CSB

Product Optional Components (PO)Lower Tube Ladies LTLLower Tube Men’s LTMLower Tube Tandem LTTLower Tube Tandem Front LTTF

Horizontal Top Tube Ladies HTTLHorizontal Top Tube Men’s HTTMHorizontal Top Tube Tandem

HTTT

Horizontal Top Tandem Front

HTTF

Head Tube Lades HTLHead Tube Men’s HTMHead Tube Tandem HTT

Seat Tube Front STF

Diagonal Tube Tandem DTT