Report CopyRight/DMCA Form For : Improving The Aircraft Design Process Using Web Based
Improving the Aircraft Design Process Using vision of an aircraft design process using Web based can be divided into three phases conceptual design
Designing and developing new aircraft systems is time consuming and expensive. Computational simulation is a promising means for reducing design cycle times but requires a. flexible software environment capable of integrating advanced multidisciplinary and muitifidelity. analysis methods dynamically managing data across heterogeneous computing platforms and. distributing computationally complex tasks Web based simulation with its emphasis on. collaborative composition of simulation models distributed heterogeneous execution and. dynamic multimedia documentation has the potential to meet these requirements This paper. outlines the current aircraft design process highlighting its problems and complexities and. presents our vision of an aircraft design process using Web based modeling and simulation. Page 2 of 35,1 Introduction, Intensive competition in the commercial aviation industry is placing increasing pressure on. aircraft manufacturers to reduce the time cost and risk of product development To compete. effectively in today s global marketplace innovative approaches to reducing aircraft design cycle. times are needed Computational simulation such as computational fluid dynamics CFD and. finite element analysis FEA has the potential to compress design cycle times due to the. flexibility it provides for rapid and relatively inexpensive evaluation of alternative designs and. because it can be used to integrate multidisciplinary analysis earlier in the design process 171. Unfortunately bottlenecks caused by data handling heterogeneous computing environments and. geographically separated design teams continue to restrict the use of these tools In order to fully. realize the potential of computational simulation improved integration in the overall design. process must be made The opportunity now exists to take advantage of recent developments in. information technology to streamline the design process so that information can flow seamlessly. between applications across heterogeneous operating systems computing architectures. programming languages and data and process representations. The World Wide Web has emerged as a powerful mechanism for distributing information on a. very large scale In its current form it provides a simple and effective means for users to search. browse and retrieve information as well as to publish their own information The Web continues. to evolve from its limited role as a provider of static document based information to that of a. platform for supporting complex services Much of this transformation is due to the introduction. of object technologies such as Java and CORBA Common Object Request Broker Architecture. 36 within the Web The integration of object technology represents a fundamental some would. say revolutionary advancement in web technology The web is no longer simply a document. access system supported by the somewhat limited protocols Rather it is a distributed object. system with which one can build general multi tiered enterprise intranet and internet. applications,Page 3 of 35, The integration of the Web and object technology enables a fbndamentally new approach to. simulation Meb based simulation A Web populated with digital objects models of physical. counterparts will lead to model development by composition using collaborative Web based. environments 9 Model execution will occur across networks using Web based technologies. e g Java and distributed simulation techniques e g C O M A Finally simulation execution. models and other related data will be documented using forms of hypermedia hypertext video. virtual models etc, Web based simulation has the potential to provide the necessary tools to improve the aircraft. design process through integration and support for collaborative modeling and distributed model. execution In the remainder of this paper we examine how this might be achieved In Section 2. we provide a brief overview of the aircraft design process drawing attention to the complexities. of the process and its inherent problems Section 3 provides a review of the area of Web based. simulation and singles out several principles of Web based simulation that we believe are. important in the aircraft design process In Section 4 we present an example scenario illustrating. how Web based modeling and simulation might be used in that process and discuss aircraft. model development and distribution using the Onyx simulation framework Onyx s object. oriented component model visual environment for model assembly and support for both Web. based and distributed object execution are explained in context of the integration of a jet engine. within the aircraft Lastly in Section 5 the relationships to the Web based simulation principles. outlined in Section 3 are identified and discussed as are general implications of Web based. simulation on the design process,2 The Aircraft Design Process. The aircraft design process can be divided into three phases conceptual design preliminary. design and detailed design The conceptual design phase identifies the various conditions of the. mission and synthesizes a set of initial aircraft configurations capable of performing the mission. For commercial aircraft the mission is defined by airline company demands which typically. Page 4 of 35, include payload requirements city to city distance along a proposed service route traffic volume. and frequency and airport compatibility If the conceptual design effort confirms the feasibility of. the proposed mission management may decide to proceed with one or more preliminary designs. In the preliminary design phase more detail is added to the aircraft design definition Here the. aerodynamic shape structural skeleton and propulsion system design are refined sufficiently so. that detailed performance estimates can be made and guaranteed to potential customers In the. final design phase the airframe structure and associated sub systems such as control systems. landing gear electrical and hydraulic systems and cabin layout are defined in complete detail. The design of an aircraft is an inherently complex process Traditional preliminary design. procedure decomposes the aircraft into isolated components airframe propulsion system control. system etc and focuses attention on the individual disciplines fluid dynamic heat transfer. acoustics etc which affect their performance The normal approach is to perform disciplinary. analysis in a sequential manner where one discipline uses the results of the preceding analysis. see Fig 1 In the development of commercial aircraft aerodynamic analysis of the airframe is. the first step in the preliminary design process Using the initial Computer aided Design CAD. geometry definitions resulting from the conceptual design studies aerodynamic predictions of. wing and fbselage lift and drag are computed Key points in the flight envelope including take off. and normal cruise are evaluated to form a map of aerodynamic performance Next performance. estimates of the aircraft s propulsion system are made including thrust and fuel consumption rate. The structural analysis uses estimates of aerodynamic loads to determine the airframe s structural. skeleton which provides an estimate of the structure weight. Complicating the design process is the fact that each of the disciplines interacts to various. degrees with the other disciplines in the minor analysis loop For example the thrust requirements. of the propulsion system will be dependent on the aerodynamic drag estimates for take off climb. and cruise The values of aerodynamic lift and yaw moments affect the sizing of the horizontal. and vertical tail which in turn influence the design of the control system For an efficient design. Page 5 of 35, process fully updated data from one discipline must be made accessible to the other disciplines. without loss of information Failure to identify interactions between disciplines early in the minor. design cycle can result in serious problems for highly integrated aircraft designs If the coupling is. not identified until the system has been built and tested experimentally then the system must. undergo another major cycle iteration further increasing the time and expense of product. development, There are many factors that can make the design process less efficient These include. 1 Lack of interoperability Numerous software packages CAD solid modeling FEA. CFD visualization and optimization are employed to synthesizs and evaluate designs. These tools are often use different possibly proprietary data formats As a result they. generally do not interoperate and require manual manipulation Lvha passing data. between applications Although in some cases custom translation tools are available to. massage the data into the appropriate format users still spend considerable time and. effort tracking data and results as well as preparing submitting and running the computer. applications 28, 2 Heterogeneous conzpriting environments The aircraft design computing environment is. extremely heterogeneous with platforms ranging from personal computers to Unix work. stations to supercomputers To use the various software required in the design process. users are forced to become familiar with different computer architectures operating sys. tems and programming languages, 3 Geographically separated design groirps Multidisciplinary design and analysis is fie. quently carried out by geographically dispersed engineering groups In special cases. entire subsystems may be designed and developed by third party contractors or compa. nies The propulsion sub system for example is designed and built separately by the pro. pulsion company and delivered to the aircraft company for installation in the aircraft In. any case geographic separation places pressure on the designers to maintain a high level. of interaction during the design process so that loss of data is minimized. Page 6 of 35, Improving the design process therefore requires the development of an integrated software. environment which provides interoperability standards so that information can flow seamlessly. across heterogeneous machines computing platforms programming languages and data and. process representations We believe that web based simulation tools can provide such an. environment,3 Principles of Web based Simulation, Since its inception in 1990 the World Wide Web WWW or Web has quickly emerged as a. powerful tool for connecting people and information on a global scale Built on broadly accepted. protocols the WWW removes incompatibilities between computer systems resulting in an. explosion of accessibility 2 301 Within the simulation community this proliferation has led to. the establishment of a new area of research Web basedsimulation involving the exploration. of the connections between the WWW and the field of simulation Although the majority of work. in web based simulation to date has centered on re implementation of existing distributed and. standalone simulation software using Web related technologies there is growing. acknowledgement that web based simulation has the potential to fundamentally alter the practice. of simulation 1 11, In one of the first papers to explore the topic of web based simulation Fishwick 8 identifies. many potential effects of web based simulation with attention given to three key simulation. areas 1 education and training 2 publications and 3 simulation programs He concludes that. there is great uncertainty in the area of Web based simulation but advises simulation researchers. and practitioners to move forward to incorporate Web based technologies Building on Fishwick s. observations Page and Opper 25 present six principles of web based simulation which capture. the vision of future simulation practice 1 digital object proliferation 2 software standards. proliferation 3 model construction by composition 4 increased use of mal and error. approaches 5 proliferation of simulation use by non experts and 6 multi tier architectures and. multi language systems,Page 7 of 35, In the remainder of this section we briefly review several of these principles In the following. sections we will examine in more detail how each apply to both the development of a simulation. environment and to the improvement of the aircraft design process. 3 1 Digital Objects, In the mid 1960 s a pioneering simulation language called Simula 67 3 was developed to. more faithfully model objects in the physical world Simula 67 introduced many of the core. design concepts e g classes and objects which form the foundation for the object oriented. programming paradigm Since that time object oriented technologies such as object oriented. programming OOP design OOD and analysis OOA have had a major impact on the field of. simulation Today the majority of simulation languages as well as many of the most successful. general purpose languages are object oriented, The importance of objects in simulation applications naturally leads us to consider their use as. part of the WWW infrastructure The WWW however is currently based on documents rather. than objects In the future though it is envisioned that the Web will be populated by digital. objects with documents being just one type of object The objects representing models and data. for use in simulation environments will be made available for use through publication on the. Indications of a transition to an object based WWW are currently evident in the successful. application of mobile code and distributed object technologies Mobile code programs which. can be transmitted across a network and executed on the client s computer make it possible to. deliver digital objects in either executable or serialized form across the WWW Several. programming languages which can produce mobile code have been developed 3 32 33 34 the. most well known and widely supported is Java 11 Compiled Java code known as byte code can. be downloaded across the Web to the client where it is executed by a Java Virtual Machine The. Java run time system incorporated within the Java Virtual Machine provides an extensive class. library that can be accessed by the compiled code,Page 8 of 35. Component Object Model COM 29 and High Level Architecture HLA 2 11 Alternatively a. component architecture may be defined by the particular simulation application in which the. objects are to operate This is often the case in domain specific simulation environments where. the component architecture must be crafted to meet specific requirements of the domain The. Onyx simulation environment described in the following section is such an example it defines a. component architecture which is oriented towards physical modeling of aerospace systems. 3 4 Heterogeneous Modeling and Simulation, The digital objects of our Web based simulation future will populate a Web that is highly. heterogeneous Digital objects will certainly be developed using different programming languages. and programming styles e g object oriented procedural functional etc The digital objects. will themselves be highly variable Some will be based on mobile code which can move across the. Web e g agents while others will form object busses which provide services from specific. locations on the Web Applications will become more complicated as a result with complex. multi tier architectures becoming the standard In order to operate effectively in such an. environment Web based simulation will need extensive enabling technologies such as search. engines to locate appropriate digital objects and models translators to convert models and data to. appropriate formats and expert systems to guide non experts in the use of Web based simulation. 4 An Example Scenario, In this section we present a scenario illustrating how Web based modeling and simulation can. be used in the aircraft design process Our goal is to discuss both the technical issues related to the. design development and publication of digital objects as well as organizational issues. concerning the roles engineers and programmers play in the Web based design process Although. the discussion is oriented towards the aircraft design process we believe that it is applicable to. engineering processes used in many fields,Page 10 of35. The modeling and simulation environment for our research is the Onyx simulation system 26. 271 The major features of Onyx include the following. A set of object classes and interfaces for representing the physical attributes and topology of. the aircraft system is included These classes comprise an object oriented component architec. ture capable of housing the analytical and geometric views of the various aircraft components. employed in the design process The architecture facilitates and ensures object interoperability. among separately developed software components, A visual assembly irtrerface is included for graphical creation and manipulation of aircraft. system models It enables users to establish model design control model execution and visu. alize simulation output, A dynamically defined run time simzdation executive is included to control complex multi. level simulations, A persistence engine capable of transparently accessing geometry and data stored in either. relational or object database management systems is included. A connection service provides access to federated model and data repositories using standard. internet protocols r arious connection strategies to access Web and server based distributed. objects are included, Our goal in creating Onyx is to develop a simulation based design system that promotes. collaboration among aerospace designers and facilitates sharing of models data and code Special. emphasis is placed on developing a distributed system which fosters reuse and extension in both. the models and the simulation environment To achieve these goals we have made extensive use. of object oriented technologies such as object oriented frameworks sofnvare components and. design patterns, An object oriented framework is a set of classes that embodies an abstract design for solutions. to a family of related problems 191 Onyx is designed as a layered collection of frameworks with. individual frameworks for the visual assembly interface persistence engine connection services. Page 11 of 35, simulation executive and component architecture The set of classes in each framework define a. semi complete structure that captures the general functionality of the application or domain. Specific functionality is added to Onyx by inheriting from or composing with framework. components In the example in the next section we will illustrate this by denying new classes to. represent the components in an aircraft engine then assembling instances of those classes to form. a complete engine model, A key characteristic of Onyx and object oriented frameworks in general is its inverted control. structure In traditional software development the application developer writes the main body of. the application which defines a series of calls to various libraries of subroutines These libraries. provide reusable code while the main body is customized by the application developer In. framework design the control structure is defined by the framework with predefined calls going. to methods that the application developer writes In this approach the design or structure of the. application which is domain specific is reused and the specific iunctionality of the. application is provided by the developer Using this approach Onyx reduces the burden for. aircraft engineers and modelers allowing similar aircraft component mod to be developed. faster and more efficiently The concept of reuse is best illustrated for modtij that are assembled. from a library of components i e composition and for models that are made in several versions. with minor differences i e inheritance, A major product of object oriented design is the identification of sofnvare components self. contained software elements which can be controlled dynamically and assembled to form. applications The central step in identifying them is recognizing rscumng fundamental. abstractions in the domain By identifLing these abstractions and standardizing their interfaces. these components become interchangeable Such components are said to be plug compatible as. they permit components to be plugged into frameworks without redesign Onyx s software. components use a variant of the JavaBeans 7 component architecture to define standard. interfaces and abstractions These components represent the plug compatible digital objects. with which the Web based models of the aircraft and its subsystems are de eloped. Page 12 of 35, Throughout the Onyx environment design patterns recurring solutions to problems that arise. when building software in various domains 131 are used to achieve reuse Patterns aid the. development of reusable software components and frameworks by expressing the structure and. collaboration of participants in a software architecture at a level higher than source code or object. oriented design models that focus on individual objects and classes 31 Patterns also are. particularly useful for documenting software architectures and design abstractions They provide. a common and concise vocabulary which is useful in conveying the purpose of a ensoftware. The Onyx simulation environment is designed to be both multi tiered and platform. independent so as to provide the greatest flexibility when modeling complex aircraft systems Java. was chosen as the implementation language as it offers extensive class libraries a distributed. object model Le Java RMI and byte code interpreters on a wide range of computer. architectures among other benefits As a result the Onyx system is extremely portable and. accessible The visual assembly interface described below for example can be run in the. context of a Web browser which are widely available while computationally intensive. components run on dedicated distributed servers, Java is also the preferred language for programming Onyx software componenrs as models. written in Java are easily downloaded across a network and dynamically loaded into the Onyx. environment In cases where it is desirable or necessary to use a programming languagz other than. Java software components may be accessed from Onyx using C O M A CORI3A s ability to deal. with the heterogeneous nature inherent in distributed computing environments makes it. particularly suitable for leveraging legacy applications not written in Java This is especially. useful for simulation of aerospace systems in which the majority of existing analysis programs. have been written in procedural languages such as FORTRAN and C The use of C O M A adds. flexibility to the Onyx system allowing it to wrap these existing programs rather than having to. replace or abandon them,Page 13 of35,Reed J I Foiien G J und Ajjeh A i. 4 2 Engine Aircraft Integration Scenario, This scenario illustrates our vision of how Web based modeling and simulation may be used in. the process of development and integration of an aircraft subsystem within the complete aircraft. As stated earlier the aircraft design process generally follows a hierarchical decomposition of the. aircraft system see Fig 2a into major airframe components e g Fuselage Rudder Wing and. Propulsion System i e Engines Individual engineering groups are responsible for establishing. the conceptual and preliminary designs for each respective component These teams work. together exchanging information as necessary to develop the individual component designs and. as the process progresses to integrate them into a final design. We have selected for our example the integration of the propulsion subsystem into the aircraft. because it represents one of the more complex aspects of aircraft design Propulsion system. performance size and weight are important factors in the overall aircraft design Engine size and. thrust for examp e influence the number and placement of engines which in turn affects aircraft. safety performance drag control and maintainability Furthermore because the engine is. designed and developed by an external manufacturer Le an engine company this example. illustrates the challenges faced by designers separated both geographically and organizationally. We intend to show how Web based modeling and simulation can address these and other issues. 4 2 1 Model Authoring As in the aircraft company engineering design groups in the engine. manufacturer are generally organized according to a physical decomposition of the engine with. individual teams responsible for developing the major engine components Fan Compressor. Combustor Turbine Mixer etc see Fig 2b In each team a model author having expertise in. the given design area establishes a conceptual model of the component During early phases of. design model resolution is kept relatively coarse to speed simulations and enable more complete. exploration of the design space Such a model typically consists of a set of algebraic andor. linearized ordinary differential equations which describe the component s gross behavior At this. stage in the design knowledge of component characteristics is incomplete so empirical data. gathered from rig testing of previously developed components are scaled to approximate the. Page 14 of 35, current model These data commonly referred to as performance maps attempt to capture. component characteristics within their operating range and serve to provide closure to the. 4 2 2 Component Authoring Once a conceptual model is validated a component author. working closely with the model author maps the model to the computational domain creating a. software component which encapsulates the model abstraction As pointed out in section 3 the. mapping is largely dependent on the choice of component architecture being used The Onyx. component architecture used here is based upon a control volume abstraction The use of control. volumes is standard engineering practice wherein the physical system is divided into discrete. regions of space control volumes which are then analyzed by applying conservation laws. e g mass momentum energy to yield a set of mathematical equations describing physical. behavior see Fig 3 A component archtecture predicated on this approach provides a. convenient and familiar mapping mechanism for modeling physical systems and ensures that a. simulation component resembles the conceptual model developed by the model author A brief. overview of the Onyx component architecture is presented below a complete description can be. found in ref 26, 4 2 3 Overview of Onyx Component Architecture There are four basic entities in the Onyx. architecture Element Port Connector and DomainModel see Fig 4 The Java interface. Element represents a control volume and defines the key behavior for all engineering. component classes incorporated into Onyx It declares the core methods needed to initialize run. and stop model execution as well as methods for managing attached Port objects Classes. implementing this interface generally represent physical components such as a compressor. turbine blade or bearing to name a few see Fig 3b However they may also represent purely. mathematical abstractions such as a cell in a finite volume mesh used in a CFD analysis This. flexibility permits the component architecture to model a variety of physical systems. Consider for example a component author in the Compressor design team wanting to develop. a representative Compressor digital object for uie in simulations during preliminary design The. Page 15 of 35,Reed J A Folleri G J and Ajjeh I 4, author begins by defining a concrete implementation of the Element interface such as. Simplecompressor see Fig 4 Here the author extends the abstract class DefaultElement. which captures common implementation aspects of the Element interface as well as maintaining. a list of Port objects associated with its subclasses Alternatively the author could implement the. interface directly explicitly defining each interface method This feature is used through the. architecture to provide flexibility the component author may select to utilize the default. functionality of the common abstract class or inherit from another class hierarchy and implement. the interface directly, An Element may have zero or more Port objects associated with it The interface Port. represent a surface on a control volume Le Element through which some entity e g mass or. energy or information passes Ports are generally classified by the entity being transported. across the control surface For example the SimpleCompressor has two FluidPort objects. representing the fluid boundaries at the Compressor entrance and exit and a StructuralPort. object representing the control surface on the Compressor through which mechanical energy is. passed Le from a driving shaft The Port interface defines t vo methods to set and retrieve the. data defined by the Port These data may be stored in any type of Java Object such as Hashtable. or Vector The common abstract class Defaultport defines default functionality for these. methods and maintains a reference to the Connector object currently connected to the Port. The common boundary between consecutive control volumes is represented by a Connector. object The interface Connector permits two Element objects to communicate by passing. information between connected Port objects see Fig 3c It is also responsible for data. transformation and mapping in situations where the data being passed from Ports of different. type The need for such data transformation can range from simple situations such as conversion. of data units to very complex ones involving a mismatch in model fidelity e g connecting a 2 D. fluid model to a 3 D fluid model or disciplinary coupling e g mapping structural analysis results. from a finite element mesh to a finite volume mesh used for aerodynamic analysis.