Estrategia de diseño de la distribución en planta y automatización de las tareas de diseño en ingeniería aeronáutica

CONGRESO INTERNACIONAL DE INGENIERÌA GRÀFICA 31 de mayo 1y2dejunio XVIII Barcelona 2006 ACTAS DEL CONGRESO Diseño e innovación INGEGRAF UNIVERSITAT POLITÈCNICA DE CATALUNYA UNIVERSITAT POLITÈCNICA DE CATALUNYA DISEÑO AUTOMATIZADO, MEDIANTE TÉCNICAS DE SISTEMAS EXPERTOS EN CATIA V5, APLICADO A COMPONENTES ESTRUCTURALES AERONÁUTICOS MARTINEZ PALACIOS, Juan; MARTINEZ LOMAS, María Eugenia; LÓPEZ CARRERAS, Ramón; DEL RIO CIDONCHA, María Gloria; Universidad de Sevilla, España Escuela Superior de Ingenieros, Departamento de Ingeniería Gráfica [email protected]; [email protected]; ramon.lop[email protected]bengoa.com; [email protected]; RESUMEN En la estructura típica de un avión muchos elementos son idénticos geométricamente: alas, deriva, flaps, alerones, etc.; adoptando la forma de perfil sustentador, tan estudiado por la Aerodinámica. De esta percepción surge la idea de automatizar su diseño. Pero este trabajo, resultado de la colaboración con el departamento de CAD-CAM de EADS-CASA, pretende llegar más lejos, aprovechando las posibilidades que ofrece CATIA V5 para hacer auténticos programas que doten de inteligencia o experiencia al diseño. Las principales herramientas de esta aplicación utilizadas para automatizar el diseño son el Power Copy y el User Feature; y para dotarlo de conocimiento la utilización de parámetros, tablas, fórmulas, reglas y chequeos. En este trabajo se revisa como crear y aplicar todos estos elementos al diseño de las piezas reales: costillas, larguerillos, revestimientos, etc., que componen los conjuntos estructurales de un avión, destinados a la sustentación, y que antes se han mencionado. Palabras clave: CAD, automatización, CATIA, estructuras aeronáuticas ABSTRACT In a plane typical structure, many elements are geometrically identical: wings, vertical tail plane, flaps, ailerons, etc, adapting the supporter profile form, so studied in the aerodynamics. It is from this perception that the idea of automating its design emerges. This work, resulting from the collaboration with the CAD-CAM department in EADS-CASA, aims to go a step further, taking advantage of the possibilities that CATIA v5 offers in the creation of authentic programmes which provide with intelligence or experience the design. The main tools of this application used to automate the design are the Power Copy and the User Feature. To endow the design with knowledge we will use of parameter, lists, formulas, rules and checkups. In this chapter we will revise how to create and apply all these elements to the plain real pieces design such as the covering, of which the structural group, destined to the joint already mentioned, is made up of. Key words: CAD, automatization, CATIA, aeronautical structures. 1. Introducción Se entiende por automatizar a la implantación de los equipos y del software necesario para disminuir el componente humano en la realización de una tarea. Esta tarea puede ser de muchos tipos, desde el mecanizado de una pieza, hasta el control de planta de una nave, o como el caso que nos ocupa, el diseño asistido por ordenador (CAD). La automatización tiene dos grandes ventajas: ahorra tiempo y por consiguiente dinero a una empresa, y permite llevar un control más fiable de los procesos y sistemas. Cada vez el sector industrial evoluciona con más rapidez, sucediéndose vertiginosamente los cambios de productos, modelos, técnicas de fabricación, etc.. Para adaptarse a este mercado cambiante, las empresas tienden a automatizar cada vez más sus sistemas y procesos. En este entorno, al que la industria aeronáutica no es ajena, nació el proyecto de colaboración entre EADS-CASA y la ESI de Sevilla, que se describe en este trabajo; y que ha consistido en el diseño, en CATIA V5, de una de las partes del fuselaje de un avión comercial. Las piezas tenían que empezar a fabricarse en un plazo muy corto, por lo que se debía realizar el diseño con la mayor celeridad posible, y de aquí, surge la necesidad de automatizarlo. Si se observa la estructura típica de un avión, como se muestra en el apartado 2, se puede apreciar que existe gran cantidad de elementos que se repiten. Son estos elementos los que se van a automatizar, de forma que el usuario sólo tenga que efectuar dos o tres operaciones para poder diseñarlos, liberándolo de carga de trabajo, y asegurando que el elemento automatizado va a cumplir con las exigencias de la norma. También en este segundo apartado, se revisan las etapas por las que pasa el diseño de un avión. Como los procesos industriales en general, son difícilmente secuenciables, los sistemas de programación tradicional se han mostrado a menudo ineficaces. Por el contrario, los sistemas basados en el conocimiento, entre los que se hayan los sistemas expertos, han experimentado un auge enorme al permitir capturar, y sobre todo almacenar, la experiencia adquirida por los ingenieros durante su trabajo y formación. Como prueba de esto, las más potentes aplicaciones comerciales de CAD-CAM, como es el caso de CATIA V5, han incorporado entre sus módulos, alguno destinado a la gestión y programación del conocimiento. En este proyecto se muestra el desarrollo de un sistema experto, creado con el módulo Knowledge de CATIA V5, utilizando las herramientas: PowerCopy y UserFeature, descritas en el apartado 3; y que tiene como objetivo el diseño de piezas características de un avión. En el apartado 4 se explica, paso a paso, el Diseño de Detalle de cuadernos, larguerillos, revestimientos, angulares y refuerzos de unión; y el montaje de todos ellos. Es de destacar, como dos tipos de personas, con roles diferentes intervendrán: el experto y el usuario. El primero, que suele ser ingeniero, aporta la metodología a seguir en el diseño y los elementos necesarios para automatizarlo. El usuario, que suele ser el delineante, conoce las operaciones de diseño, y elabora las piezas siguiendo las indicaciones del experto. 2. Las estructuras aeronáuticas y su diseño La estructura de un avión se considera generalmente compuesta de tres partes principales: el fuselaje, las alas y los estabilizadores. Está construido fundamentalmente de aleaciones de aluminio y magnesio, si bien se puede encontrar acero y titanio en zonas expuestas a altas temperaturas. Figura 1: Componentes estructurales de un avión El fuselaje es la parte principal o cuerpo del avión, la de mayor volumen y la única que no contribuye a su capacidad de vuelo, sino más bien al contrario, pues es la principal fuente de resistencia aerodinámica. En el fuselaje es donde se aprovisiona el espacio adecuado para la carga, controles, accesorios y otros equipos. Las alas de un avión son las superficies diseñadas para producir sustentación cuando se mueve a gran velocidad a través del aire. El diseño de un ala depende de factores como el tamaño, peso, velocidad deseada para el vuelo y aterrizaje, así como la aceleración de subida que se requiera. En cada semiala se encuentran montadas diversas superficies encargadas de producir algún efecto determinado sobre el vuelo de la nave: alerones, flaps, slats y spoilers. Figura 2: Componentes estructurales del ala de un avión Fijado en el fuselaje se observa un plano montado en forma vertical, llamado estabilizador vertical o deriva. Abisagrado a él se encuentra una superficie móvil, el timón de dirección. Algunas veces, montados en el extremo superior de la deriva, otras a cada lado del fuselaje (depende del fabricante del avión), se sitúan los planos horizontales de cola o estabilizador horizontal. En su borde de salida se haya una superficie móvil, cuyo movimiento controla y maneja el cabeceo de la nave, llamado timón de profundidad. Todas estas estructuras destinadas a la sustentación: alas, alerones, flaps, slats, spoilers, estabilizadores y timones; comparten la misma forma geométrica de perfil sustentador y la misma estructura interna constituida por largueros y larguerillos, dispuestos a lo largo de la envergadura, y por costillas según la cuerda (del borde de ataque al borde de salida). A continuación se procede a su descripción:  Costillas: encargadas de mantener la forma del perfil aerodinámico del ala. Suelen estar formadas de una sola pieza de chapa doblada con los contornos exteriores del perfil del ala. Puesto que son bastante débiles lateralmente, suelen estar reforzadas con perfiles rigidizadores o bordones para prevenir la flexión lateral de las mismas. Así mismo para aligerar el peso del núcleo, se suelen practicar orificios en el mismo. Estos elementos se llaman aligeramientos.  Largueros: están formados por cordones y por el alma. Su perfil tiene la forma de doble T, en dirección perpendicular a la de vuelo. Los largueros son los miembros estructurales más importantes del ala.  Revestimientos: forman una superficie impermeable para soportar las presiones dinámicas. Durante el vuelo, las cargas que actúan sobre el ala están aplicadas casi en su totalidad en el revestimiento. De éste son transferidas a las costillas y de las costillas a los largueros.  Larguerillos: fabricados generalmente por extrusión, estos perfiles que unen costillas y revestimientos, hacen la misma función que las tirantas de una estructura metálica. Figura 3: Detalles de fabricacion de un conjunto sustentador 2.1. Diseño de un avión Los requerimientos o especificaciones de diseño consisten en los objetivos que se desea cumpla la aeronave. Tres son los principales:  Cuánta carga llevar (en aviones civiles se traduce en número de pasajeros) y cuanto espacio se necesita para ello.  A qué distancia se quiere llevar esa carga.  A qué velocidad y a qué altura se quiere volar. Pero no son las únicas, también son importantes: la longitud requerida para despegue y aterrizaje, los costes de operación, el precio del avión, la normativa de certificación que va a regir en el diseño (FAR, JAR,…), etc... Una vez establecidas las especificaciones, el proceso de diseño que permite al fabricante entregar finalmente al cliente un avión ha de pasar por las siguientes fases: Diseño Conceptual, Diseño Preliminar y Diseño en Detalle. En la fase de Diseño Conceptual, un reducido grupo de ingenieros, no expertos en ningún sistema del avión, pero con conocimientos generales de aerodinámica, estructuras, aviónica, motores, dinámica de vuelo, procesos de producción, mantenimiento de aviones y evaluación de costes de fabricación y operación; da rienda suelta a la imaginación y buscan el avión ideal En esta fase surgen las ideas geniales, como los motores en la cola del Caravelle o la "joroba" del B-747. Es aquí cuando se deciden las dimensiones generales del fuselaje, y se hace una estimación del peso total del avión: estructura y combustible. Se hacen hipótesis de eficiencia aerodinámica y consumo específico del motor, y finalmente se determina cual es la carga alar, y la relación empuje de motor/peso al despegue que permite cumplir las actuaciones. A partir de cálculos no demasiado sofisticados se determina la superficie del ala, y los valores aproximados del tamaño de: los estabilizadores de cola, el ángulo de flecha del ala, flaps, timones y alerones. Se decide también, el número de motores y el empuje necesario de cada uno, si el ala es alta o baja, la cola en "T" o convencional, si llevará wingtips, etc. A la fase de Diseño Preliminar se pasa cuando la configuración general del avión se tiene definida de forma global y no se esperan cambios importantes. En este punto entran los ingenieros especialistas, y cada uno se encarga de afinar y optimizar el área tecnológica que le compete, con sofisticados cálculos por ordenador, haciendo pequeños ajustes finales a la geometría del avión. Esta fase termina cuando todos los sistemas del avión están perfectamente definidos: modelo de motor, equipos de aire acondicionado, etc., y además queda establecida la situación de cuadernas, largueros y costillas de la estructura. En la última fase Diseño en Detalle, se definen todas las piezas cuya unión formarán el avión. En definitiva, se plasman en planos y en sistemas CAD, hasta el último remache del avión. En esta fase también se decide cómo debe ser el proceso de producción, y se diseña el utillaje necesario para montar el avión. Esta fase termina cuando las piezas entran en la cadena de producción y se ensambla el avión. Cuando el primer prototipo sale de la fábrica, se inicia una larga serie de ensayos en vuelo, que se encaminan por un lado a verificar el correcto funcionamiento de cada uno de los sistemas; y por otro a comprobar las cualidades de vuelo, de forma que se cumplan adecuadamente todos los requerimientos de diseño. Finalmente una vez se superan con éxito todos los controles, el fabricante obtiene el Certificado de Aeronavegabilidad, que es como la nota de aprobado, requisito fundamental para que el avión pueda volar legalmente. 3. Descripción del módulo KNOWLEDGE de CATIA V5 Una gran novedad qua aporta el CATIA V5 con respecto a versiones anteriores es la introducción de un módulo destinado al Knowledge. El Knowledge de CATIA V5 está en continua evolución, y en cada Release aparecen nuevas aplicaciones, pudiendo llegar a ser un instrumento muy potente de diseño. Hoy en día el Knowledge ayuda a automatizar el diseño, permitiendo insertar herramientas como parámetros, fórmulas, tablas, reglas y chequeos; que se describen a continuación.  Fórmulas: se definen como el elemento de enlace entre el parámetro y el diseño.  Tablas: permiten tabular una serie de parámetros previamente introducidos, de forma que variando un solo parámetro, el resto de los parámetros incluidos adoptan el valor impuesto por la tabla. Éste es el sistema utilizado para introducir elementos normalizados mediante tablas.  Reglas: las reglas se utilizan para dotar al diseño de cierta inteligencia, de forma que pueda adoptar una forma u otra dependiendo de una serie de condiciones.  Chequeos: los chequeos son elementos de verificación. Poseen una estructura idéntica al de las reglas, pero en ellas no se escriben condiciones y sentencias, sino que se imponen condiciones de funcionamiento, de forma que si esa condición no se cumple, el chequeo manda un aviso, advirtiendo al usuario de que esa condición está fallando. Al aplicar estas herramientas de automatización al diseño, no sólo se minimizan los tiempos de construcción, sino que se garantiza que la pieza ha sido construida siguiendo las normas establecidas. 3.1. Herramientas de Automatización: PowerCopies y UserFetures Además de los elementos descritos en el apartado anterior, existen otras herramientas, cada vez más depuradas, que permiten mejorar la automatización en el diseño. Estas herramientas son los PowerCopies y los UserFeatures. El PowerCopy (PC) y el UserFeature (UF) son prácticamente idénticos, y van a ser tratados de forma análoga. En el siguiente apartado se mencionan las diferencias principales entre ambos elementos. Un PowerCopy se utiliza para adaptar elementos previamente creados a las necesidades del usuario. El PC solamente se puede crear e insertar en el interior de un CATPart, y por lo tanto estos elementos son principalmente geometrías (planos, superficies, etc.), que se posicionan de una forma u otra dependiendo de unos datos de entrada que se denominan INPUTS. Los Inputs pueden ser de muchos tipos: un punto, una línea, una superficie, un parámetro, etc. La metodología a seguir para crear un PC y un UF es la siguiente: 1. Se determinan cuales son los Inputs necesarios para que todos los elementos geométricos queden perfectamente definidos. 2. Se estructura el diseño, de forma que los únicos padres de estos elementos sean estos Inputs. 3. Se crean los elementos que hemos elegido como Inputs. 4. Se diseña los elementos que queramos incluir en el interior del PC. 5. Se crea el PowerCopy en el interior de ese CATPart. 6. Se cierra el CATPart que incluye el PC, porque sino no es posible aplicarlo. 7. Se inserta el PC en otro CATPart que posea los Inputs necesarios. 3.2. Diferencias entre ambas herramientas Ambos elementos se crean y se insertan de la misma forma, y el objetivo para el que fueron creados es prácticamente idéntico: ayudar al usuario a dibujar figuras que se repiten en el diseño. Las diferencias principales que hay entre ambas herramientas se describen a continuación:  El PC posee una geometría abierta, y el UF cerrada. Esto quiere decir que cuando se inserta un PC, aparecen en el árbol de especificaciones todas las operaciones necesarias para obtener esa geometría, pudiendo editarlas, cambiarlas, etc. Sin embargo cuando se inserta un UF, aparece en el árbol como una Feature (operación) más del diseño, no se puede cambiar, ni editar ni se puede determinar como ha sido creada.  En un PC se pueden introducir Bodies sin operar, y en un UF sólo se pueden introducir Bodies, si estos han sido operados mediante operaciones booleanas.  Un PC puede ser modificado si se edita haciendo un doble clic sobre él, y un UF no. Además, si se modifica algún elemento incluido en el interior de un UF, es recomendable volver a crear el UF, ya que pueden ocasionarse errores de inserción. 4. Aplicación del módulo KNOWLEDGE de CATIA V5 al diseño automatizado de componentes estructurales aeronáuticos Para ilustrar el alcance de estas herramientas, se va a presentar a continuación una aplicación práctica. Se va a modelar, como ejemplo, el timón de altura del avión T270. Dicho timón no es más que una estructura aeronáutica ficticia, que contiene los elementos estructurales típicos del diseño aeronáutico. El diseño del conjunto de las piezas del timón se ha realizado a partir de una superficie de referencia, también llamada la Master Geometry, que es la geometría necesaria para poder ubicar todas las piezas que lo constituyen. Esta geometría es el dato de partida del diseño: la superficie mojada (es decir, la superficie externa del avión), y tiene la siguiente forma: Figura 4: Master Geometry y Timón de Altura del avión T270 A continuación se detallan los pasos a seguir para la realización de la estructura aeronáutica comenzando por el diseño de las cuadernas. Aquí, se pondrán en práctica varios ejemplos de utilización de PowerCopies y Userfeatures, desde la creación, hasta la inserción. 4.1. Diseño de cuadernas El objetivo será automatizar al máximo la fabricación de cualquier pieza de este tipo. El trabajo se ha desarrollado de forma que todas las cuadernas transversales se realicen de la misma manera, obteniendo piezas diferentes pero construidas con las mismas herramientas de diseño. El aspecto final que tendrá la cuaderna será el siguiente: Figura 19: Montaje de Estructura Una vez que la estructura del elemento ha sido montada es necesario comprobar que las piezas no ocupan el mismo lugar en el espacio, es decir, que están bien situadas unas con respecto a otras haciendo contacto entre ellas y no superponiéndose entre sí. Existe un comando en CATIA V5 que realiza este tipo de tareas. Este comando se llama CLASH, y se ejecuta pinchando en el icono . En teoría todos los elementos deben estar bien colocados debido a la estructura interna de los PowerCopies y UserFeatures que se han creado para la confección de las piezas. Pero no está de más realizar un análisis de interferencias para comprobar que realmente está todo en su sitio. 5. Conclusiones En esta ponencia se ha descrito el resultado de la colaboración entre el área de Desarrollos en CADCAM del departamento de Ingeniería Concurrente en la Factoría de San Pablo (Sevilla), perteneciente a la compañía aeronáutica EADS-CASA; y el departamento de Ingeniería Gráfica de la Escuela Superior de Ingenieros de Sevilla. Y aunque es sobradamente conocido, recordar que EADS es un consorcio europeo dedicado a la fabricación de aeronaves: aviones, helicópteros, cohetes, etc.; y además socio mayoritario del fabricante de aviones de transporte AIRBUS, y cuya filial española CASA, posee sedes en Madrid, Sevilla y Cádiz. Este trabajo, de casi un año de duración, ha fructificado en la elaboración de una aplicación informática (sistema experto), destinado al diseño de componentes estructurales típicos del avión, tales como: alas, alerones, flaps, planos fijos, etc.; que comparten: la misma forma geométrica, la misma función de sustentación en vuelo, y están formados por los mismos tipos de piezas: costillas, largueros, larguerillos, refuerzos, etc. La utilización de un sistema experto ha permitido, no sólo automatizar el diseño, con la reducción de tiempo consecuente; sino que también va a permitir que muchos usuarios menos cualificados, como delineantes, puedan realizar tareas de diseño, recogiendo el conocimiento almacenado por el experto (ingeniero) en el sistema, y con la seguridad de que los resultados cumplen las normas establecidas. Referencias BAUER, K., et al., "Sistemas Expertos", Editado por Nebendahl, D. para Siemens y Marcombo SA, Barcelona, 1988. BRUNET, P. “Diseño gráfico y modelado geométrico”. En Mompín J. (Ed.): "Sistemas CAD/CAM/CAE. Diseño y fabricación por ordenador".Marcombo. Barcelona 1986. Cursos de CATIA: http://www.cadtech.es/cadtech/catis.htm Dassault Systems: http://www.catia.com FRENZEL, L.E., "A fondo: Sistemas Expertos", Edit. Anaya Multimedia, Madrid, 1987. Pagina sobre CATIAV5 de IBM: http://www.ibm.com FERRÉ MASIP, Rafael. “Diseño industrial por computador”. Marcombo Editorial, Barcelona 1987 HUERTA GUIJARRO, Joaquín. et al. “Prácticas de Informática Gráfica para Diseño Industrial”, Colección Material Docente, Publicaciones de la Universitat Jaume I, Castellón, Febrero, 1998, KUNWOO, Lee. “Principles of CAD/CAM/CAE Systems”, Addison-Wesley, Published November 1998 WINSTON, P.H., "Inteligencia Artificial", Edit. Addison-Wesley Iberoamericana, Buenos Aires, 1994. Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha, Juan Martı ´nez-Palacios and Francisco Ortun˜o-Ortiz Graphic Engineering Department, Advanced Engineering School, University of Seville, Seville, Spain Abstract Purpose – The main goal of this paper is to show how the automation of tasks in solid modelling with Catia V5 is approached, by means of macros, working under the Windows operating system and with the Visual BASIC programming language. A macro is a piece of code (written in a certain programming language) which groups a set of operations that define a certain task. Design/methodology/approach – The paper is divided into three parts. In the first one, the necessary concepts to begin working with macros are discussed: the programming languages (object-oriented), compatible with Catia V5, are introduced. The handling of macros, like the one of the fundamental objects of the application, is explained as well. In the second part, the different objects on which the macros for solid modelling are based are described, explaining how they are used and how they are related to one another. In the third and last part, several applications which clarify the use of the studied tools, and which show the great utility of the macros, are included. Findings – Using macros can automate design tasks which are performed repetitively, as well as carry out new operations (features) which facilitate the modelling process and which are not included in the order menu by default. Originality/value – Power and versatility of macros as a task-automation tool is shown, as well as how simple, fast and efficient it is to program them in Visual BASIC. Keywords Automation, Macros, Modelling Paper type Technical paper Introduction The fundamental objective of this paper is to show how to tackle task automation for modelling solids with Catia V5 through the use of macros, with Windows as the operating system and Visual Basic as the programming language. A macro is a code (written in a determined programming language) used to group a set of operations which define a specific task. They are used to automate repetitive tasks as well as to develop new operations (features) that facilitate the modelling process and that are not included by default in the order menu. The first part of this report shows the necessary concepts to start working with macros and also introduces the programming languages (object based) which are compatible with Catia V5. It then describes the distinct objects with which the macros are constructed for modelling solids. The third part shows various applications which clarify the use of the studied tools and clearly demonstrates the usefulness of macros. Instruments for task automation when modelling solids with Catia Catia V5 (Karam and Kleismit, 2004) offers the possibility of automating tasks when modelling solids through the use of three tools: PowerCopy, UserFeature and macros. The objective of these three tools is the automatic generation of results using several inputs which must be totally generic. PowerCopy This tool allows the automation of tasks that have been previously run manually on a solid model (CATPart document). To create PowerCopy you must: define the model elements (geometry, parameters, formulas, ...) which will be taken as generic inputs in the automation process, publish the parameters where the values can be modified by the user and select the resultant elements. Once saved, the model that contains PowerCopy can be run in any other CATPart document. You will have to select the new inputs and define the published parameter values. From then on, both the final process results and the intermediate elements needed to obtain them, will be run automatically along with all the design specification. UserFeature The UserFeature helps to run tasks automatically within a CATPart document in a very similar way to PowerCopy. The difference is that only those intermediate elements which have been selected by the user are created. Therefore, the final results and the intermediate elements will be generated without their design specifications so that they cannot be modified at a later stage. Macros A macro is a code written in a determined programming language used to group a set of operations which define the task to be automated. The essential advantages this tool has over PowerCopy and UserFeature are the following: .automation of all types of tasks for modelling solids; .quick and simple execution through a screen icon; The current issue and full text archive of this journal is available at www.emeraldinsight.com/1748-8842.htm Aircraft Engineering and Aerospace Technology: An International Journal 79/1 (2007) 53–59 qEmerald Group Publishing Limited [ISSN 1748-8842] [DOI 10.1108/00022660710720494] 53 .maximum reduction in the number of inputs, therefore minimising user intervention in the process; .repeats the automated task an indefinite number of times; and .the possibility of working together with CATPart and CATDrawing documents, therefore generating all possible views of the solid model. Catia programming languages Catia V5 allows the creation and execution of macros through the use of distinct object-based programming languages. Any application entity (documents, graphic windows, geometry, parameters, formulas, ...) will be represented and operated by these languages like an “object”. Working on a PC with Windows as the operating system, one can use programmes such as: Visual Basic 5 (Charte, 1998), Development Studio (Hansen and Thomsen, 2004; Dunaway, 2002), Visual Basic for Applications 6.0 (Balena, 1999), Visual Basic Scripting Edition (Cornell, 1998), Java/JavaScript (Arnold and Gosling, 1999) and Visual Cþþ (Sophar, 1999). The creation of macros using Visual Basic The creation of macros from the Catia application is directly linked to the Visual Basic programming language in two distinct ways – two programming environments: CATScript (simplified edition of Visual Basic) and VBScript (derived from Visual Basic Scripting Edition); as well as direct access through the tools menu to the external programming environment – Visual Basic for Applications 6.0 Catia automated object structure The automated objects in the application are organised in a parent-child hierarchical structure. The main object from which all the others descend is an “Application” that directly represents the Catia programme. With this object you can create and operate documents and graphic windows (Figure 1). Object structure in a CATPart document In these documents, the main object from which all the others descend is “Part”. From this object, reference systems, Bodies and Open Bodies (sets of reference elements: lines, planes, surfaces ...), geometrical and dimensional constraints, parameters and all the connections that need to be complied with (formulas, rules, reviews ...) are created and operated (Figure 2). Object structure in a CATDrawing document The drawings of any solid model will be made from the DrawingSheets object (collection of sheets and technical drawings of a document). With this object you can define the format and work scale, create views, carry out cuts or sections and also create texts and 2D geometry. Through the parameters and relations objects you can create parameters, formulas, rules ... (Figure 3). Description of the implemented automated processes The following macros have been developed using the Visual Basic for Applications 6.0 programming language. The user will be able to execute applications on any solid model by only clicking on a screen icon. Repetitive task automation Production of countersunk holes This application automatically and immediately produces an indefinite number of countersunk holes in points situated over different solid faces (this cannot be manually done by multi selection or the use of matrix). By using a simple form you can enter parameters that define the geometry of the holes (diameter, depth, tip angle and its depth) and determine the method to be used when choosing the centre of the holes: pre-selection in the structure tree before executing the macro, direct selection of the points on the geometry or tree, or location of points by creating a search on certain characters. In the following example (Figure 4) holes are created in all those geometry points where the name ends in “TA”. The manual process would require: location of the centres that comply with the search criteria, repetitive use of the tool for the creation of holes and the tool for the creation of matrix (if needed) from the design module Part Design. Figure 1 Initial diagram of object automation Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha et al. Aircraft Engineering and Aerospace Technology: An International Journal Volume 79 · Number 1 · 2007 · 53–59 54 Development of new tools Initial “stock” generation for mechanised parts This application automatically produces the stock or solid block needed to obtain the final part by using different mechanised operations. Inertia analysis of the solid model sets the dimensions, volume and geometry of the “stock” as well as the volume of the removed shavings (Figure 5). The obtained block can be used, for example, to simulate the mechanised process with the Catia V5 NC manufacturing module. On an initial form the excess material from the three directions of the block are entered (the X-axis corresponds to the largest stock dimension and the Z-axis corresponds to the smallest). The manual process would require: interpretation of the data obtained from the model inertia analysis, creation of complete “stock” geometry through mechanic design modules – taking into account excess material and generation of data such as dimensions, “stock” volume and volume of removed shavings. Figure 2 Object structure in a CATPart document Figure 3 Object structure in a CATDrawing document Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha et al. Aircraft Engineering and Aerospace Technology: An International Journal Volume 79 · Number 1 · 2007 · 53–59 55 Generation of planes which including a straight line intersect another given plane at a degrees This application facilitates the generation of planes which including a straight line intersect another given plane at a degrees. (There is no Catia V5 tool that allows this to happen directly.) The straight line and reference plane will be chosen on the screen. Angle a , introduced by the user, is controlled through formulas and reviews. If the angle entered is not within the permitted range (which is between the given line-plane angle and 908) then a screen message will appear informing you of the limit which you cannot exceed (Figure 6). The manual process would require: control of the permitted values for angle a and generation of all the geometry needed to obtain the planes from the reference elements. Control of task execution Automatic generation of standard lightening holes on flat sheets Lightening holes are much used elements in the industry in order to reduce weight and increase rigidity in metal sheets. The automatic generation of these type of holes saves time which is important when modelling metal sheet products. Most of all it ensures that the legislation related to these type of elements is complied with. The designed macro allows you to create standard lightening holes just by selecting their centres on the screen (Figure 7). The manual process would require: the creation of the geometry necessary for a lightening hole profile which complies with the legislation, creation of complete geometry of the lightening hole by revolution of the defined profile around the hole axis, elimination of interior excess material using a boolean subtraction operation, use of matrix if needed or repetition of previous steps to create the rest of the lightening holes for the part. User accessibility to complex tasks Automatic generation of drawings in DXF format This application enables any user (regardless of their amount of knowledge of Catia V5) to automatically generate a document in “dxf” format (legible from other CAD programmes) that contains standard views of any solid model. By using a form the user will define the format and scale they wish to use as well as the standard views to be included in the document (Figure 8). Figure 4 Centre of holes by using a chain of characters Figure 5 Initial “stock” generation of the part Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha et al. Aircraft Engineering and Aerospace Technology: An International Journal Volume 79 · Number 1 · 2007 · 53–59 56 Once the form has been completed, one selects the reference face for the frontal view and define the name and destined location of the dxf document. The views of the drawings will be distributed in a uniformed way within the format. The manual process would require: creation of a new CATDrawing document, the use of tools from the drawing module drafting in order to create the frontal view of the solid model, creation of the rest of the standard views from the frontal view, uniformed distribution of views within the work format and lastly creation of the document in “dxf” format (Figure 9). Parameter analysis The last of the developed applications will allow quick location of those model elements that are affected by the value of a determined design parameter. Once the required parameter has been selected, all the elements linked to it will be shown (geometry, formulas, ...). The manual process would require detailed analysis through parent-child links in all relationships (formulas, rules ...) of the parameter in question. However, it would not be possible to see all of the affected elements at the same time (Figure 10). Figure 7 Standardised geometry and generation of lightening holes Figure 6 Interactive windows and generation of planes Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha et al. Aircraft Engineering and Aerospace Technology: An International Journal Volume 79 · Number 1 · 2007 · 53–59 57 Conclusions Essentially one can affirm that: .Macros are the most powerful and versatile tool that Catia has to offer when automating tasks for modelling solids. .Visual Basic for Applications 6.0 enables a quick and efficient development of any type of macro; .The links between automated objects are basic in the macro creation process. .The control of errors when executing macros will always allow you to define the type of object used and ensure the correct development of the process. .The multiple screen selection system as well as the operation of some properties and methods for certain objects should be improved in higher versions of the application. Figure 10 Parameter information Figure 8 Initial form for the application and selection of reference for a frontal view Figure 9 Definition of name and location of the dxf document and generation of views Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha et al. Aircraft Engineering and Aerospace Technology: An International Journal Volume 79 · Number 1 · 2007 · 53–59 58 References Arnold, K. and Gosling, J. (1999), The Java Programming Language, 2nd ed., Addison Wesley Professional, Boston, MA. Balena, F. (1999), Programming Microsoft Visual Basic 6.0, Bk&CD-Rom ed., Microsoft Press, Washington, DC. Charte, F. (1998), Programacio ´n con Visual Basic 5.0, Anaya Multimedia, Madrid. Cornell, G. (1998), Learn Microsoft Visual Basic Scripting Edition Now, Microsoft Press, Washington, DC. Dunaway, R.B. (2002), The Book of Visual Studio.Net, No Starch Press, San Francisco, CA. Hansen, J.E. and Thomsen, C. (2004), Enterprise Development withVisualStudio.NET,UML,andMSF, Apress, Berkeley, CA. Karam, F. and Kleismit, C. (2004), Using Catia V5, Delmar Learning, Scarborough. Sophar, C. (1999), Aprenda Microsoft Visual Cþþ 6.0, Editorial McGraw-Hill, Mexico. About the authors Marı ´a Gloria del Rı ´o-Cidoncha received the graduated degree in Architecture and PhD in Industrial Engineering from Seville University (Spain). Master degree in TIC’s and e-learning training from University of Seville. Since, 1988 associated Teacher in the area of Graphical Expression, in the Engineering Upper School of Seville. From 1988 to 2002, member of Technological Research and Development group of Andalucia regional government. She is author of three books on graphical engineering and has published 20 papers in this area. Member of the Graphic Engineering Spanish Association (INGEGRAF). Marı ´a Gloria del Rı ´o-Cidoncha is the corresponding author and can be contacted at: [email protected] Juan Martı ´nez-Palacios received the graduated degree in Mechanical Industrial Engineering from Seville University (Spain). He has completed all courses for PhD in Production Management from Industrial Organization and Management from University of Seville. Since, 1989 Associated Teacher in the area of Graphical Expression in the Engineering Upper School of Seville. Since, 1987 work in EADS-CASA and during these years he has collaborated with the most important aeronautical companies: AIRBUS, Boeing, SAAB, Dornier, etc. At the present time he collaborates in the Airbus A-380 and A-400M development programs. Author of two book on graphical engineering and has published 20 papers in this area. E-mail: [email protected] Francisco Ortun ˜o-Ortiz is based in the Graphic Engineering Department, Advanced Engineering School, University of Seville, Seville, Spain. Task automation for modelling solids with Catia V5 Marı ´a Gloria Del Rı ´o-Cidoncha et al. Aircraft Engineering and Aerospace Technology: An International Journal Volume 79 · Number 1 · 2007 · 53–59 59 To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints  Definir y compartir fácilmente la habilidad de diseño entre todos los usuarios.  Automatizar la definición del producto.  Aumentar la productividad  Permitir dar una atención temprana a especificaciones del plan de diseño que previenen un costoso rediseño.  Guiar y ayudar a los usuarios a través de sus tareas del plan de diseño Las aplicaciones que se necesitan para desarrollar soluciones del conocimiento se han dividido en tres partes:  El Consejero del Conocimiento ( Knowledge Advisor)  El Conocimiento Experto (Knowledge Expert)  Plantillas de Producto del Conocimiento ( Product Knowledge Template) 3.1. Knowledge Advisor Los usuarios pueden aportar conocimiento, en un plan de diseño, tales como: conductas, fórmulas, reglas y chequeos; y acudir a ellos cuando se requiera en cualquier momento. Este módulo está compuesto por una serie de herramientas que facilitan el proceso de reutilización de la información. Los Parámetros Son elementos que permiten controlar variables que están destinadas a formar parte de algún aspecto del diseño. Pueden ser de muchos tipos, parámetros de medida, de masa, reales, incluso boléanos. Pueden tener un solo valor, o múltiples valores, o llevar insertada alguna fórmula. Lo más importante que tiene su uso, es la posibilidad que da al diseñador de variar su valor después de ser creado. Las Fórmulas Son los elementos de enlace entre los parámetros y los elementos que conforman el diseño. Se pueden definir por el usuario, aunque Catia trae una gran cantidad de fórmulas integradas: volumen de un modelo, cálculo de la distancia mínima, etc. Las Tablas Cuando tenemos una serie de parámetros que hacen referencia a un mismo conjunto de datos, o que la construcción geométrica a la que hacen referencia esté normalizada, se recogen los resultados en una tabla, podemos crearla de tal forma que sus valores cambien en función de los datos contenidos en ella Las Reglas Las reglas dan la posibilidad al usuario de obligar al modelo a que siga unas determinadas pautas. Son fundamentalmente de dos tipos:  Agrupaciones de fórmulas: se consigue así sobre todo tener un árbol más ordenado y con menos elementos.  Reglas: creadas por el propio usuario con un lenguaje intrínseco de Catia y que siguen la estructura de un lenguaje sencillo de programación, como Visual Basic. Los Chequeos Los chequeos sirven para hacer comprobaciones sobre las definiciones que tienen que cumplir los elementos, así se pueden detectar si se está incumpliendo alguna normativa o alguna regla que lleva incluido el modelo. Las Macros Una Macro, no es más que una aplicación creada con un lenguaje de programación, que se puede construir con un editor de texto, o bien puede ser creada desde un programa en Visual Basic en un documento aparte, y luego llamarla desde Catia a través de una acción. Para construir una Macro el usuario debe conocer el lenguaje de la programación, no es como construir una regla que trae un diccionario con los parámetros y operadores que se utilizan. 3.2. Knowledge Expert Es una herramienta para analizar y manipular todo un producto. Tiene las mismas capacidades que el módulo de Knowledge Advisor, pero tratada de una forma un poco diferente; mientras que en el Knowledge Advisor se crean aplicaciones que contienen características o propiedades del modelo y se almacenan dentro del mismo modelo, en el Knowledge Expert el usuario que llamamos “el experto”, puede generar una Regla Base. Esta Regla Base, es la que va a funcionar como regla estándar para el resto de usuarios que la vayan a usar, y cada vez que quieran crear una regla nueva, tienen que ir añadiéndola a la anterior, teniendo al final un conjunto de reglas. Cada usuario puede manipular a su antojo las reglas que están situadas en el segundo nivel, pero no pueden manipular la Regla Base. Esta sólo puede ser manipulada por el usuario que la creó. Si se desactiva esta regla, quedan automáticamente desactivadas todas las demás. Con esto se consigue ir almacenando información, con la finalidad de ir catalogando todo un conjunto de información y formativas, que posteriormente pueden ser usadas por los usuarios que tenga la empresa. 3.3. Product Knowledge Template Este módulo está compuesto por tres elementos que sirven para automatizar los procesos en el diseño y son en esencia los elementos que componen el conocimiento en Catia. PowerCopy Se utilizan fundamentalmente, cuando se tiene una construcción geométrica que se va a repetir a lo largo del proceso de diseño, así, se construye una sola vez, se inserta dicha geometría en el PowerCopy, y puede ser utilizado en otro modelo tantas veces como sea necesario. Se incluirán como Input (entradas): los Parámetros que vayan a ser constantes en los modelos donde se van a aplicar: espesores, radios, etc., y como Outputs (salidas): el resultado que se quiera obtener. UserFeature La UserFeature tiene la misma estructura que el PowerCopy y su función también es la misma, lo que ocurre es que la estructura interior está oculta y no se tiene acceso a la misma, si no estamos dentro del modelo donde se creó, así se puede decir que es como una caja negra donde sólo se ven las entradas (Inputs) y las salidas (Outputs) Las entradas serán los elementos en los que se puede un diseñador apoyar para la construcción de la geometría. Un PowerCopy puede ser modificado una vez creado, y la UserFeature una vez creada no deja cambiar ningún elemento, sino que hay que construirla de nuevo. Document Template Está integrado en un catálogo, que es una estructura de organización, que permite tener almacenado por familias, las distintas aplicaciones que se van creando en Catia. Así se consigue tener una librería de conocimiento almacenado y ordenado, para su posterior reutilización. Los catálogos permiten crear ligaduras entre ellos con lo cual, se puede tener un solo catálogo raíz, del que cuelguen todos los demás agrupados por familias, y tener totalmente organizadas las distintas aplicaciones que se vayan realizando en un proyecto de una empresa. El Document Template está compuesto por UserFeatures, o PowerCopies formando un modelo; por lo tanto, cuando se quiere volver a utilizar hay que estar dentro de un entorno general formado por un conjunto, un Product en Catia. 4. Normas generales para el tratamiento y cálculo del material de partida en un mecanizado. Todas las piezas que se mecanizan por control numérico parten de un paralelepípedo que se ajusta lo más posible a la pieza final; y que dependiendo del material, el peso y la precisión de mecanizado, recibe una serie de operaciones previas que están normalizadas: Planeado Los materiales empleados para el fresado suelen ser laminados, con imperfecciones en sus caras. Generalmente se pide al fabricante que su espesor tenga 1 ó 2 mm de creces, sobre la cota mayor de pieza. Para garantizar un buen asiento, se necesitará que al menos se tenga una cara perfectamente plana. Taladros de cogida La cogida del material de partida al útil suele ser por tornillos, del tipo de cabeza Allen. Su número y situación dependerá de la geometría de la pieza y del utillaje a emplear. Taladros de centraje Los taladros de centraje sirven para coordinar mecanizados entre las distintas posturas de la pieza o sus montajes en útiles distintos. Son de gran precisión en diámetro y situación. Taladros de izado Cuando el taco sea difícil de manejar por su tamaño o peso, se le darán unos taladros roscados para montar posteriormente unos cáncamos de izado. Para la situación de los taladros, debemos tener en cuenta el tipo de máquina con la que se va a trabajar: si es de mesa horizontal o vertical, y el centro de gravedad del taco. Para las de mesa horizontal y cuando el problema sea de peso, estos taladros nos deben posibilitar el situar la pieza en posición de montaje, para lo cual se darán en los retales y con un margen bueno de seguridad a la pieza. Estos se realizarán a partir de 20 Kg; ya que, por debajo de este peso, se puede manipular el taco con las dos manos. La cantidad de taladros de izado, dependerá del peso o tamaño del material de partida, y siempre se darán por la cara de taladrado de los tornillos. Figura 2: Taladros de izado en una máquina de mesa horizontal Cuando se tenga una máquina de mesa vertical, el taladro de izado se colocará en un lugar adecuado lo más parecido posible a la posición de montaje. Aquí, el peso y el tamaño del material aún son más importantes que en una mesa horizontal, ya que el operario necesitará una mano para atornillar, mientras que con la otra sujeta el material de partida, generalmente piezas de pequeño tamaño. En este caso se harán a partir de 12kg. Figura 3: Taladros de izado en una máquina de mesa vertical Taladro de Descuelgue En las máquinas de mesa vertical, se tiene un problema añadido, y es que todas las creces que se hallan dado a los taladros de cogida, irán hacia el mismo sitio, es decir, abajo. Esto no sería un problema si la pieza sólo tuviera una postura o fuera de poca altura, pero cuando las piezas son altas o la distancia entre redoblonados es pequeña y tienen dos posturas, suele darse. Esto se evita mediante un pasador de los que se tiene que usar en la 2ª postura, y dándole al material de partida un taladro cuyas creces sean, como máximo, la mitad de las dadas a los taladros de cogida, es decir, si para un tornillo de M12 se ha taladrado a  12.5, y el pasador de coordinación es de  12, el taladro de descuelgue se dará a  12.25 mm. Calculo de las dimensiones del material de partida El cálculo de las dimensiones del material de partida se hará de la forma siguiente:  Si el lado que se tiene que calcular forma parte de la pieza terminada, se le dará unas creces de 5 mm aproximadamente por cada lado.  Por el lado donde alojan los tornillos, se darán unas creces de 2 mm más que el diámetro de la herramienta mayor que por allí pase, más el diámetro de la cabeza del tornillo, más 5 mm.  Para el espesor es válido todo lo anteriormente expuesto en el apartado del planeado. Es decir, si lo que se quiere, es terminar la cota de espesor, será la cota de la altura de la pieza; y si no se quiere terminado, se le dará unas creces mínimas de 1 ó 2 mm. Todas estas creces son las mínimas, al final habrá que redondear las cotas del taco. 5. Fases de la construcción del modelo en el sistema experto. El proceso actual de fabricación que se sigue en las piezas de mecanizado es: una vez que se dispone del diseño de la pieza terminada, en un modelo sólido, que suministra el departamento de Proyectos o los departamentos de Fabricación, y en este particular el área de Mecanizado por Control Numérico, se debe establecer la geometría del material de partida, con sus creces y cogidas, envolviendo la pieza origen, y a la vez resultado del proceso final de fabricación que utiliza tecnologías de CAD-CAM. Este procedimiento es lento y muy repetitivo, porque los pasos a seguir son prácticamente los mismos, y hay que hacerlo cada vez que se diseña y fabrica una nueva pieza. La aplicación pretende, con las herramientas que nos proporciona Catia, y su módulo de Knowledge, hacer este proceso sólo una vez y que sirva para el resto de las piezas de la misma familia. La aplicación diseñada, utilizando técnicas de sistemas expertos, en otras palabras basada en el conocimiento, se divide en fases o etapas que se pasan a describir: Construcción del Modelo Base El Modelo Base será la pieza que se quiere construir o material de partida, servirá como patrón para construir la geometría inicial que va a hacer falta para enlazar el resto de etapas Construcción del Material de partida tangente A partir de la geometría, se construirá un paralelepípedo con seis lados tangentes a las seis caras del espacio tridimensional que contenga a la pieza. Construcción del Material de partida con las crecidas Siguiendo las normas de mecanizado de piezas de fresado por control numérico, el material de partida de una pieza no es tangente a las caras, sino que por cada lado tiene unas crecidas para posicionar los taladros de cogida e izado. Construcción de los taladros En esta etapa se construirán todos los tipos de taladros que se tienen que tener en cuenta en la normativa de mecanizado:  Taladros de posicionamiento y cogida del material al útil  Taladros de izado vertical  Taladros de izado horizontal Presentadas las fases de la aplicación, se pasa a explicar la construcción del modelo: Primeramente se dibuja una pieza cualquiera en Catia que servirá como modelo base y que luego se sustituirá por la pieza a la que se quiere calcular el material de partida o entrada. A la hora de dibujar este modelo inicial hay que tener en cuenta la organización interna de un programa de dibujo, puesto que si el apoyo es un plano, la pieza siempre queda vinculada al plano, así que hay que apoyarse en un sistema de ejes totalmente genérico. Este sistema está formado por dos rectas que se cortan para que formen plano, y una tercera que sea perpendicular a las otras dos. Se tienen hasta el momento tres parámetros de entrada: la pieza y dos rectas que pueden ser de la pieza o no, pero que tienen que cortarse para formar plano. Dependiendo de las rectas que se elijan el paralelepípedo que forma el material de entrada variará respecto a la pieza. Ahora, se construirá toda la geometría auxiliar necesaria para delimitar el sólido que contiene el material de partida. Se crean los planos tangentes a las seis caras de la pieza a partir de la creación de los puntos extremos de las direcciones que delimitan el sólido. Se posiciona un plano en cada punto. El plano además será tangente a cada una de las tres rectas que se han definido como las tres direcciones en el espacio. El material de partida con las crecidas puede tener diferente dimensión por cada uno de los lados, por lo que se definirán parámetros distintos para las tres direcciones del espacio. Estos tres parámetros podrán ser modificados por el diseñador porque pueden variar de un modelo a otro. También será necesario posicionar todos los puntos donde podrían ir ubicados los taladros, así como los parámetros que definen la geometría de los distintos taladros. Estos parámetros también serán editables por el diseñador. Algunos elementos geométricos están relacionados con fórmulas siguiendo la normativa de mecanizado. Por ejemplo la separación que existe entre los puntos que definen los taladros de posicionamiento y los taladros horizontales viene definida por: Separación = Diámetro cabeza + Diámetro-Tal-Horizontal + 3mm En la figura 4 se observa la geometría inicial de la que se parte para modelar el estado de entrada. Figura 4: Geometría inicial para modelar el estado de entrada Después de completar la fase inicial (Construcción del Modelo Base) se crea la primera UserFeature para comprobar si esta bien determinada. Se emplean UserFeatures, en vez de PowerCopies, porque cuando se utilice este modelo en otra pieza, el propósito será que el nuevo diseñador sólo pueda cambiar los parámetros que se han definido previamente como editables, y el modo en que se ha hecho la aplicación, quedará oculto al nuevo diseñador. Se consiguen de esta forma dos ventajas fundamentales: primero, que ninguna persona excepto la que ha realizado la aplicación pueda manipularla y estropearla; y el know-how permanece en la empresa que está creando la aplicación, y si esta se vende podrán usar la herramienta pero no sabrán como se ha realizado. Para la creación de una UserFeature en los componentes (components) se van introduciendo todos los elementos geométricos que se tienen en el árbol de especificaciones por orden de creación. En la parte derecha que son los elementos de entrada (Inputs) van apareciendo los padres de los elementos que se van seleccionando. Figura 5: UserFeature Pieza inicial inicialininici al Planos tangentes con crecidas Taladro Posición Taladro vertical Taladro Horizontal Aquí, se inicia la fase Construcción del Material de partida tangente. Es decir, se construye la caja que envuelve a la pieza de tal forma que las caras de la caja y las de la pieza sean tangentes. Para ello, se crea un modelo nuevo con otra pieza base y se inserta la UserFeature que se ha creado en la fase anterior. En la tercera fase, Construcción del Material de partida con las crecidas, el procedimiento es similar al de la fase anterior, pero utilizando una caja cuyos planos son paralelos a los de la caja tangente pero, a la distancia marcada por los parámetros definidos como crecidas en las tres direcciones. Ya solo queda, para terminar el modelo, establecer la posición de los taladros en la cuarta fase. Hasta aquí, se han programado, en la construcción del modelo, cuatro UserFeatures que permiten crear: la geometría inicial, el paralelepípedo tangente, el paralelepípedo con las crecidas y los taladros de posicionamiento. Establecer si el material de partida lleva taladros de izado vertical u horizontal dependerá de dos elementos: el peso y la bancada de mecanizado.  El peso se determinará por el volumen y la densidad del material que lo calcula Catia.  La bancada de izado, será un parámetro editable que tendrá dos valores, horizontal y vertical. Se crearán dos UserFeatures, una que reproduzca la geometría de los taladros de izados horizontales, y otra para los verticales. Después se establece una Regla de Diseño, para precisar si es necesario o no poner los taladros verticales u horizontales dependiendo de lo que pese el Estado de Entrada y de la mesa de mecanizado. Los taladros horizontales sólo son necesarios si la mesa de mecanizado es horizontal y si el Estado de Entrada pesa más de 20 kilos. El taladro vertical es necesario si la mesa de mecanizado es vertical y el Estado de Entrada pesa más de 12 kilos. La Regla de Diseño se crea con un lenguaje de programación muy sencillo que tiene Catia en un editor de texto. Cuando se tiene una UserFeature en un documento se puede tratar como un elemento geométrico más, por lo que se puede desactivar o activar dentro del propio Modelo. Al desactivarla por primera vez en un Documento, se genera un parámetro relacionado con la actividad que cuelga en el árbol de especificaciones de la UserFeature El parámetro es un parámetro interno que Catia lo nombra como Activity y es de tipo Booleano con dos valores, true si la UserFeature está activa, y false si la UserFeature está desactivada. Para terminar la aplicación se tiene que crear el Documento Template, donde estarían incluidos todos los elementos que componen el modelo. Este, se guarda en un Catálogo, y se podrá reutilizar para cualquier pieza de nuevo diseño. Interesa tener una serie de parámetros editables, que puede cambiar el diseñador porque varíe la matriz y el útil de mecanizado. Serán: las crecidas que se dejaran por cada uno de los lados del paralelepípedo, los diámetros de los taladros, tanto de la cabeza como de la caña; la holgura o distancia de separación entre la cabeza del tornillo y el lado del paralelepípedo; y por último la separación entre taladros. 6. Resultados obtenidos. Aplicación del sistema experto a una pieza real A partir de una pieza diseñada en Catia, Figura 6: Pieza diseñada en Catia Se busca la ubicación del documento donde esté guardado el catálogo. Al seleccionar el Document Template aparece una ventana de diálogo, y en esta los elementos de entrada que necesita el modelo, y una ventana de visualización donde muestra sus significados. Figura 7: Ventana de diálogo en el Document Template La elección de las aristas las hace el Usuario, y dependiendo de cuáles escoja la posición del estado de entrada variará, ya que estas dos rectas son las direcciones de posicionamiento de toda la construcción geométrica. Así, si las dos rectas que se eligen no son ortogonales entre sí, el estado de entrada generado no será perpendicular, sino que sus aristas formarán un ángulo distinto de 90º. Esto no es lo habitual, pero puede darse el caso dependiendo de la forma que tenga la pieza. Al elegir los parámetros de entrada el resultado que se obtiene es: Figura 8: Elección parámetros de entrada y resultado El diseñador puede cambiar los valores que determinan la geometría de los taladros, además de precisar si la mesa de mecanizado es horizontal o vertical. En piezas que tienen todas sus aristas redondas, como la pieza de la figura 9, hay que crear en el modelo las dos rectas que sirvan de dirección para formar el estado de entrada, usualmente se elegirán que sean tangentes a las caras y que se corten ortogonalmente. Figura 9: Pieza con aristas redondas Elegidas las rectas se obtiene el siguiente resultado: Vmag1¼SqrððCornerð1ÞPoint1ð1ÞÞ^2 þðCornerð2ÞPoint1ð2ÞÞ^2 þðCornerð3ÞPoint1ð3ÞÞ^2Þ Vmag2¼SqrððCornerð1ÞPoint2ð1ÞÞ^2 þðCornerð2ÞPoint2ð2ÞÞ^2 þðCornerð3ÞPoint2ð3ÞÞ^2Þ Vmag3¼SqrðððPoint1ð1ÞPoint2ð1ÞÞ^2Þ þððPoint1ð2ÞPoint2ð2ÞÞ^2Þ þððPoint1ð3ÞPoint2ð3ÞÞ^2ÞÞ These lines determine the lengths of the two generatrices which join the vertex of the cone to the base points 1 and 2, calculated previously, and the length of the line which joins point 1 and point 2. The generation of the development in the open plane document starts here. The procedure consists of generating and joining triangles which form the approximate surface of the cone. Set objLine1¼objSheet:Lines2d:AddBy2Points ðOrigð1Þ;Origð2Þ;Origð1ÞþVmag1CosðBetaNÞ; Origð2ÞþVmag1SinðBetaNÞÞ A line is drawn with an initial point on the plane document, ending at a point such that the length of the line is Vmag1 and its angle with the horizontal is BetaN. This variable is defined in a subsequent instruction. Set objLine2¼objSheet:Lines2d:AddBy2Points ðOrigð1Þ;Origð2Þ;Origð1ÞVmag2SinðBetaNÞ; Origð2ÞþVmag2CosðBetaNÞÞ A line is drawn with an initial point on the plane document, ending at a point such that the length of the line is Vmag2 and its angle with the horizontal is BetaN þp/2, as it is perpendicular to the previous line. Set objLine3¼objSheet:Lines2d:AddBy2Points ðOrigð1ÞþVmag1CosðBetaNÞ;Origð2Þ þVmag1SinðBetaNÞ;Origð1Þ þVmag1CosðBetaNÞVmag3 SinðBetaNÞ;Origð2ÞþVmag1 SinðBetaNÞþVmag3 CosðBetaNÞÞ A line is drawn with an initial point at the end of objLine1. The length of the line is Vmag3 and its angle with the horizontal is BetaN þp/2, as it is perpendicular to the previous line. The following shows that result of this process for this example, taking the number of divisions as 30 for w¼7. Set objArc1¼objSheet:Arcs2d:AddByCenter StartEndðOrigð1Þ;Origð2Þ;Origð1ÞþVmag2 CosðBetaNÞ;Origð2ÞþVmag2SinðBetaNÞ; Origð1ÞVmag2CosðBetaNÞ;Origð2ÞVmag2 SinðBetaNÞÞ Figure 3 Step 1. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] Figure 4 Step 2. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] Figure 5 Step 3. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] 298 DEL RI ´O-CIDONCHA ET AL. Set objArc2¼objSheet:Arcs2d:AddByCenter StartEndðOrigð1ÞþVmag1CosðBetaNÞ; Origð2ÞþVmag1SinðBetaNÞ;Origð1Þ þðVmag1þVmag3ÞCosðBetaNÞ;Origð2Þ þðVmag1þVmag3ÞSinðBetaNÞ;Origð1Þ þðVmag1Vmag3ÞCosðBetaNÞ;Origð2Þ þðVmag1Vmag3ÞSinðBetaNÞÞ In the previous instructions the arcs objArc1 and objArc2 are drawn, with centers on the initial and final points of objLine1, respectively, and radii Vmag2 and Vmag3 respectively. The intersection of these arcs determines the third vertex of the approximation triangle, and the end points of the lines objLine2 and objLine3 are displaced to this point. These instructions define the intersection point mentioned above, displacing lines objLine2 and objLine3 to their correct position. The previously drawn arcs are deleted. BetaN ¼objLine2:Angle The angle BetaN is defined as that formed by objLine2 with the horizontal of the plane document, and the angle from which the approximation triangle is drawn is wþ1. End of For loop which develops the circular base oblique cones. The result of the development is shown once the For loop has run for all values of w. This is followed by the ElseIf block, which runs if the base of the spline type lateral surface is an ellipse. Else This continues the development of the elliptic base oblique cone... DEVELOPED APPLICATION Solved Cases The practical cases that can be developed by the application are centered on radial surfaces (prism, cylinder, pyramid, and cone), as well as the intersections between them, or adapted surfaces (hoppers). They are as follows: (1) Straight and oblique prisms with any polygonal base. (2) Sections of straight and oblique pyramids, with parallel and oblique polygonal bases. (3) Oblique cones and sections of cones, with parallel or oblique circular or elliptical bases. (4) Oblique and straight cylinders with parallel or oblique circular or elliptical bases. Cylindrical tanks. (5) Adapted surfaces with a lower polygonal base and an upper circular or elliptical base (hoppers). (6) Intersections of straight or oblique cones with other surfaces with any geometry. (7) Intersections of straight or oblique cylinders with a circular base, with parallel or oblique bases, with other surfaces of any geometry. Application to the case of cylindrical pipe bends formed by straight cylinders. Working When the application is run, the main window of the macro appears (Fig. 6). Solid Edge must be open for the application to work correctly, as it is a macro that is run in Solid Edge. The options in the drop-down File menu are: -New: This option allows the user to open a new Solid Edge Part document, in order to generate a new surface to develop. Set objRelations ¼objSheet.Relations2d Set objIntersection1 ¼objRelations.AddIntersection (objLine2, igLineEnd, objArc2, objArc1) Call objIntersection1.Delete Set objIntersection2 ¼objRelations.AddIntersection (objLine3, igLineEnd, objArc2, objArc1) Call objIntersection2.Delete Set objPoint(1) ¼objRelations.AddKeypoint(objLine, igLineStart, objLine2, igLineStart) Set objPoint(2) ¼objRelations.AddKeypoint(objLine, igLineEnd, objLine3, igLineStart) Set objPoint(3) ¼objRelations.AddKeypoint(objLine2, igLineEnd, objLine3, igLineEnd) Call objArc1.Delete Call objArc2.Delete AlfaT ¼AlfaT (objLine2.Angle objLine1.Angle) If w ¼0 Then Call objLine2.Delete Else If w ¼N1 Then Call objLine1.Delete Else Call objLine1.Delete Call objLine2.Delete End If End If Next w EDUCATIONAL SOFTWARE 299 -Open: This option opens a previously saved surface, to edit it or to develop it using the application. -Save: This option saves a Solid Edge Plane (*.dft) document, in that the development of the surface has been obtained. -Close: This option closes the application, although the Solid Edge program remains open. There are also two frames in the main window: Developments and Various §Frame: Developments (Fig. 7). There are four buttons in this frame. -Develop the hopper: This allows the development of hoppers generated in Solid Edge, which are transition surfaces between one circular or elliptical mouth and another that is polygonal, giving rise to flat sides and conical surfaces. Before pressing the button, it is necessary to specify the number of approximation triangles of the conical surfaces (Fig. 8). A window will then appear, in that the user can choose the plane template where the development of the hopper will be made (Fig. 9). The example shown shows the process of development of a transition surface between a rectangle, which is the lower mouth, and a circle, which is the upper mouth. Both mouths are on parallel planes (Fig. 10). By following the steps described above, the development is obtained (Fig. 11). The result is saved in a plane document (*.dft) that can also be saved in other neutral files for interchanging data, such as Iges (*.igs), MicroStation (.dgn), and AutoCAD (*.dwg and *.dxf), and from this point the file can be used to control the cutting trajectory of CNC machines. Figure 7 Development frame. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] Figure 6 Principal window of the application. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] 300 DEL RI ´O-CIDONCHA ET AL. In addition, the application indicates the area of the development and the percentage or error in the approximation, according to the number of divisions. In the Tools menu of Solid Edge Part it is possible to obtain the exact area of the surface, and in Solid Edge Plane, where the development is made, it is possible to obtain the area (by selecting multiple elements using the CRTL key) of all the triangles that make the development. It is therefore possible to calculate the percentage of error, that will be proportionally lower, the larger the number of triangles. -Develop conical or cylindrical surface: This button allows the user to develop the cases of cylinders cones described above, generated in Solid Edge. As before, the user must specify the number of divisions. It is also necessary to choose the template where the development will be represented. -Develop prism or pyramid: This button allows the user to develop the cases of prisms and sections of pyramids described Figure 8 Dialogue box for inputing no. of divisions. Figure 9 Window for selecting plane templates. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] Figure 10 Hopper with rectangular base and circular mouth. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] EDUCATIONAL SOFTWARE 301 above, again generated in Solid Edge. In the same way as before, the number of divisions should be specified beforehand. -Develop intersection or pipe bend: This button allows the user to develop cylinders and cones and their sections, as well as intersection curves with other surfaces generated in Solid Edge. As in this case there are intersections between various parts, it is necessary to select the surface to be developed with intersection (penetration) curves. When the button is pressed, a dialogue box is shown in order to input the number of divisions (number of approximation triangles of the conical surfaces) §Frame: Various (Fig. 12). -Print: This button prints the open document, in this case the Solid Edge Plane document. The user must select the required plane. -Cancel: Closes the application. -Calculate the area inner the hopper: Calculates the interior of the development of the hopper shown in the Solid Edge Plane document. -Calculate the perimeter of the hopper: Calculates the perimeter of the hopper. -Show dimensions: When activated, this function shows on the development of the surface the dimensions of each of the lineal segments that make up the development. ASSESSMENT OF THE APPLICATION In order to obtain information about the results of the use of the macro, a questionnaire was designed and distributed among the students registered in the course of Engineering Graphics in the first year of study of Chemical Engineering at the University of Seville, Spain. The questionnaire was anonymous and was completed by 93 students on the course. It was made up of 6 questions, using a scale of 15(1¼Very unsatisfactory, 2 ¼Unsatisfactory, 3 ¼Indifferent, 4¼Satisfactory, 5 ¼Very Satisfactory). The items used in the study of the macro, as well as their results, are shown below and reflect the success of the development. *Overall assessment (mean ¼4.35). *Improves specificefficacy of Solid Edge (mean ¼4.3). *User friendliness (mean ¼4.25). *Easy browse (mean ¼4.2). *Interaction clear and understandable (mean ¼ 4.27). *Speed of response (mean ¼4.03). CONCLUSIONS—FUTURE WORK *The principal advantage of the macro developed here is that it allows the user to generate surfaces in a 3D CAD environment in a simple and intuitive way, with the dimensions necessary for the user. The user can then access the surface directly using a button on the Solid Edge Part tool Figure 11 Development of the hopper. Figure 12 Various frame. 302 DEL RI ´O-CIDONCHA ET AL. bar, and by indicating the number of divisions can obtain the development in a plane document (*.dft) that can then be edited to adjust it to the format of the plotter. There is currently no other application on the market that can do this. *Another important advantage is that the programming in Visual Basic allows for a fast and high quality graphic result, even when the number of divisions is greater than 500. The errors are less than 1%, and in the majority of cases less than 0.5%. Therefore, accuracy is optimized and the working time is decreased. *Future lines of work include the creation o fan application that determines automatically the cutting trajectory by creating numerical control instructions. Also, it will be interesting to increase the number of surfaces that the application is able to develop or intersect. ACKNOWLEDGMENTS The authors wish to thank Ricardo Morales Duarte for the work carried out in the conception and development of the application, as well as for its user manual. REFERENCES [1] M. F. Iskander, Benefits of virtual teaching, Comput Appl Eng Educ 5 (1997), 12. [2] P. Company, M. Contero, A. Piquer, N. Aleixos, J. Conesa, and F. Naya, Educational software for teaching drawing-based conceptual design skills, Comput Appl Eng Educ 12 (2005), 257268. [3] P. I. A. Penin, M. R. P. Morales, R. P. G. Di ´az, and J. S. Quiro ´s, Multimedia-integrated application for computer-assisted teaching of technical drawing (Aimec-Dt), Comput Appl Eng Educ 12 (2004), 136144. [4] S. C. Ou, W. T. Sung, S. J. Hsiao, and K. C. Fan, Interactive web-based training tool for CAD in a virtual environment, Comput Appl Eng Educ 10 (2003), 182193. [5] M. V. Candal and R. A. Morales, Design of plastic pieces and their molds using CAD/CAE tools. I, Comput Appl Eng Educ 13 (2006), 233239. [6] S. H. Huang, K. R. Bhura, and G. Wang, Cutter path simulation and product visualization using Autocad, Comput Appl Eng Educ 8 (2000), 113126. [7] A. Y. Averkiev, Drawing sheet-metal stepped parts, Russ Eng J 58 (1978), 4546. [8] Z. C. Lin and G. J. Peing, An investigation of an expert-system for sheet-metal bending design, J Mater Process Tech 43 (1994), 165176. [9] Z. C. Lin and J. T. Horng, Sheet-metal products: Database in support of their process planning and surface development, Int J Comput Integ M 11 (1998), 524533. [10] R. Vicente and A. Pinillos, Manual pra ´ctico de Solid Edge v.16, Editorial Servicios Informa ´ticos DAT, Madrid, 2005. [11] F. Balena, Programming Microsoft Visual Basic 6.0, Bk&CD-Rom edition, Microsoft Press, Washington, 1999. [12] Available online at: http://www.idpsoft.com/esp/moddes. shtml. [13] Available online at: http://www.eng-pc.com/Caldereria.htm. [14] Available online at: http://www.lantek.es/index.shtml. [15] Available online at: http://www.arcam.ch/index.htm. [16] Available online at: http://www.revcad.com/Sheet5/ index.html. [17] Available online at: http://dominio-lda.com/19-xes. htm. [18] T. Maekawa and J. S. Chalfant, Computation of inflection lines and geodesics on developable surfaces, Math Eng Ind 7 (1998), 251267. [19] N. M. Patrikalakis and T. Maekawa, Shape interrogation for computer aided design and manufacturing, Springer Verlag, Heidelberg, Germany, 2002. BIOGRAPHIES Gloria del Rı ´o-Cidoncha received her PhD in Industrial Engineering from the University of Seville in 2002, where she is currently assistant professor of several subjects related to computer assisted design. She is a specialist in computers and education, CAD and applied uses of vectorial design. She has been accumulating experience in developing educational software for engineering teaching and over the last years has published papers and works on those subjects. Her major field of research is interactive multimedia systems, specifically innovative ways to teach engineering concepts with multimedia software and to improve spatial perception. Juan Martı ´nez-Palacios received his MS in Industrial Engineering at the University of Seville in 1990. He has been assistant professor in the Area of Graphic Expression in the Engineering Department of the University of Seville since 2003. EDUCATIONAL SOFTWARE 303 Rafael Ortı ´z-Marı ´nobtained his degree in Industrial Engineering from the University of Valladolid in 1989. He is currently working at University of Seville as assistant professor of subjects related to Computer Aided Drawing and Engineering Graphics. His major field of research interests is focused on CAD tools, industrial photogrammetry and the virtual recreation of industrial heritage. Jose ´Ignacio Rojas-Sola is an associate professor of engineering graphics at the University of Jae ´n, Spain. He received an MS (1991) in chemical engineering at the University of Seville and PhD (1995) degree in mechanical engineering at the National University of Distance Education (UNED). He joined the Department of Engineering Graphics, Design and Projects of University of Jaen in 1992, where he has been associate professor since 1996. Therefore, he is the head of the research group Engineering Graphics and Industrial Archaeology from 1996, funded by autonomous government of Andalusia, Spain. He teaches undergraduate students (industrial engineering) and also at the graduate (doctorate) level, and he is the author of more than 85 research papers including 14 referee journal publications, several books and numerous lectures. His research interest focuses on technical drawing, descriptive geometry, computer-aided design, computer graphics, industrial archaeology, history of technology and engineering graphics applied to product design. 304 DEL RI ´O-CIDONCHA ET AL. Edita: ASOCIACIÓN DE PROFESORES DE EXPRESIÓN GRÁFICA EN LA INGENIERÍA Nº 17 Año 2005 Portada: Cartel del III CONGRESO DE EXPRESIÓN GRÁFICA EN LA INGENIERÍA Las Palmas de Gran Canaria,4,5,6 y 7 de junio de 1991. COLABORACIONES: Los trabajos originales deben iniciarse con el título,nombre del autor o autores,Universidad y Centro al que pertenecen, y un resumen del artículo. La presentación será en CD-ROM, preferentemente en Microsoft Word o Quark-XPress para Macintosch o Windows,incluyendo una copia en papel de referencia. En documento aparte se enviará una Ficha de Autor/es,que incluya todas las coordenadas de localización,que en el caso de profesores serán los datos académicos: Universidad, Centro, Departamento, categoría académica, direcciones postal, telefónica,fax y e-mail y los particulares: direcciones postal,telefónica,fax y e-mail. Deberán enviarse a la Redacción: ANALES DE INGENIERÍA GRÁFICA E.T.S.de INGENIEROS INDUSTRIALES (A la atención del Profesor Isidro Zatarain) Paseo del Cauce, s/n. 47011 VALLADOLID Teléf.983 30 90 70 Fax 983 42 33 94 E-mail: [email protected] Edita: INGEGRAF Director : Isidro Zatarain de Dios Consejo de Redacción: Francisco Montes Tubío Miguel Bermejo Herrero Alfonso Cobos Moyano Miguel Ángel León Casas María Luisa Martínez Munueta José María Monguet Fierro Félix Sanz Adán Consejo Asesor : Miguel Bermejo Herrero Agustín Arias Coterillo Fernando Brusola Simón Franco Caputo Vito Cardone Carlos Cobos Gutiérrez José Javier Crespo Ganuza Fernando Fadón Salazar Jesús Félez Mindán Rafael Ferrer Torío Ignacio González Tascón Francisco Hernández Abad Juan Leiceaga Baltar Francisco Montes Tubío Gonzalo Moris Meléndez Jesús Oti Velasco José Carlos de San Antonio Gómez Javier Sánchez Reyes Domicilio: ETS de INGENIEROS INDUSTRIALES Paseo del Cauce,s/n. 47011 VALLADOLID Teléf.983 30 90 70 Fax 983 42 33 94 E-mail: [email protected] Suscripción: Número suelto: 4 e Número atrasado: 5 e Socios de INGEGRAF: Gratuito Preimpresión: EDITO artes gráficas • Valladolid Telf.607 84 27 27 Imprime: Gráficas Santa María,S.L. C/ Cromo,24 • 47012 Valladolid Depósito Legal: M-15091-1995 I.S.S.N.: 1137-7704 Anales de Ingeniería Gráfica n.º 17 Año 2005 CARTA DEL DIRECTOR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 LEICEAGA BALTAR, Xoan DISEÑO E INGENIERÍA NAVAL. OPTIMIZACIÓN TÉCNICO ECONÓMICA DEL PROYECT O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 GALLEGO ÁLVAREZ, Francisco Javier INTEGRACIÓN DE LA INFORMACIÓN RELATIVA A LA CARACTERIZACIÓN TOPOGRÁFICA DE LA COMARCA OLIVARERA “SIERRA MÁGINA ” (JAEN) EN UN SIG VECTORIAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 CAMPO RÁMILA, Francisco Jesús DISEÑO DE MOBILIARIO URBANO SIGUIENDO LA METODOLOGÍA DE E C O D I S E Ñ O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 AGUILAR TORRES, Fernando José EFECTOS DE LA MORFOLOGÍA DEL TERRENO ,DENSIDAD MUESTRAL Y MÉTODOS DE INTERPOLACIÓN EN LA CALIDAD DE LOS MODELOS DIGITALES DE ELEVACIONES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 SERRANO MIRA, Julio ESTABLECIMIENTO DE LA CADENA DE CO TAS FUNCIONALES EN CONJUNTOS MECÁNICOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 DEL RÍO CIDONCHA, María Gloria INTERCAMBIO DE MODELOS SÓLIDOS ENTRE DISTINTOS SISTEMAS DE CAD MEDIANTE EL FORMA TO NEUTRO STEP . . . . . . . . . . . . . . . . . . . 43 1. Introducción Hoy en día, el trabajo del ingeniero está fuertemente unido al ordenador en general, y a los sistemas CAD/CAE en particular. Cada empresa de ingeniería opta por un sistema en concreto eligiendo entre todos los que ofrece el mercado: CATIA, I- DEAS, PRO/Engineer, etc., y siendo común las subcontratas de proyectos entre empresas de ingeniería, es muy posible que no coincidan en el software de trabajo. Y es aquí, donde entra en juego la existencia de un formato neutro de intercambio entre distintos sistemas CAD. La información generada sobre un producto a lo largo de su diseño, fabricación, mantenimiento y venta puede ser La información generada de un producto, a lo largo de su diseño, fabricación, mantenimiento y venta, puede ser tratada con distintos sistemas CAD/CAE/CAM. Este trabajo aborda un tema, tan fundamental y de actualidad, para los usuarios, como es la comunicación entre sistemas. El formato STEP es un estándar desarrollado por la ISO, que hace posible este intercambio de información, y que se está usando exitosamente en diversos sectores industriales. En este artículo se revisa la estructura y funcionamiento de este Protocolo de Aplicación, realizando diversos intercambios de modelos sólidos entre dos sistemas de CAD/CAM: CATIA v5 y SOLID EDGE v9. Son analizadas las limitaciones y ventajas de sus traductores de STEP, para conseguir unas pautas de diseño encaminadas a obtener una transferencia de datos correcta. Palabras clave: CAD, CAM, CAE, STEP, MODELADO, INTERCAMBIO. Abstract Product data generated over its entire life cycle (design, manufacturing, maintenance and sale) can be managed using different CAD/CAE/CAM systems. This paper documents an introduction for data exchange, a chance to transfer data between software application program from differing vendors. STEP format is an international standard for the representation of product data, which provides a neutral format for product data exchange; different companies have succeeded using STEP. How the data models are used to define a formal standard exchange mechanism for product model data is investigated, in addition to test the exchange between two CAD systems: CATIA v5 and SOLID EDGE v9. Both STEP half translators are analysed, so that you can get some design tips in order to succeed in the exchange. Key words: CAD, CAM, CAE, STEP, MODELLING, EXCHANGE. Resumen Intercambio de modelos sólidos entre distintos sistemas de CAD mediante el formato neutro STEP DEL RÍO CIDONCHA, M.ª Gloria MARTÍNEZ PALACIOS, Juan RONQUILLO JAPÓN, Mariano Universidad de Sevilla, España Escuela Superior de Ingenieros tratada con distintos sistemas CAD/CAE/CAM apareciendo el problema de compatibilidad entre los mismos. Para evitarlo, se necesita facilitar la información del producto de forma interpretable por cualquier sistema, es decir, es necesario un formato neutro. El formato STEP es un estándar desarrollado por la ISO que pretende hacer posible el intercambio de información en cualquier punto del ciclo de vida de un producto. Este trabajo muestra en qué consiste la transferencia de datos, así como su evolución histórica, para luego entrar a valorar los formatos neutros más usados actualmente, así como la problemática general asociada al intercambio de información entre distintas organizaciones y sistemas. Se Anales de Ingeniería Gráfica Author's personal copy intentionality, style and balance between what one desires and what is feasible could be mentioned. The complexity which is involved in a design process leads to a decomposition into different tasks which interact with each other, making it necessary to reconsider decisions continuously. Even though the goal of the design can be known from the beginning, the required functionality could possibly not be clear until certain hypothesis are formulated or certain structural solutions are proposed. The complexity and bad structure of design problems are thus clear. Independently of the product or servicetobedesigned,the stages depicted in Fig. 1 are always present. The analysis of the most significant viewpoints of the design process reveals a common link: it is seen as a reasoning process. The most recent solution methods, coming up from the artificial intelligence field, characterize this process as a search along a state space (design partial descriptions), advancing through the application of certain operators (design actions). These operators apply certain knowledge about the design process and bring about transformations, leading in general to more complex descriptions, even though they could result into previous states when the current one happens to be inappropriate and the objective is to reach a goal state (design solution). The floorplan design is a key step when designing a building of industrial plan. The starting point consists of: - An initial outline, defined by its exact shape and dimensions. - The rooms to distribute and their dimensions. - Proximity requirements between the different rooms or activities. The result shall be the definitionof a layoutthat maximizes the efficiency of the building; for example, minimizing costs in the case of industrial plants, or maximizing beauty and comfort in houses and apartments. In this paper, the main concepts, the typical processes and the general goals and restrictions of the design are described. Then, the equation that is capable of modeling the layout problem will be reviewed. This equation is the result of particularizing the general one which describes how man designs in general. Finally, the ways in which architects and engineers have designed layouts are analyzed, emphasizing the advantages and disadvantages of each approach. 2. Architectonic composition strategies Architectonic composition can be defined as the search for physical sketches capable of achieving certain goals while accomplishing with certain restrictions. This process, which is usually applied in complex situations, has two main features: thinking and creative discernment. These two functions are made easier by analytic and representational operators, which take into consideration the sketch that is being created. There are three areas of knowledge in architectonic design: knowledge about the domain, that is, about architectonic objects Fig. 1. Stages in the design process. Fig. 2. Strategies of architectonic composition, from [4]. 560 M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy and architecture in general; knowledge about the representation, that is, about the methods to make a graphical representation of the object being designed; and finally knowledge about the design process, which includes the strategies and criteria used in the process of decision of the architectural shape. The above mentioned concepts have originated different strategies for architectonic composition, as shown inFig. 2.Inthat figure, the traditional methodology is compared with the new strategies born in the eighties. The stages remain approximately the same, but nowadays all of them are computer-aided. 3. Methodologies for floorplan design in industry Floorplan design is a part of the industrial activity planning process. This process consists of four different stages: definition of the product and production process, locating the industrial Fig. 3. Hierarchy of industrial activity planning. Fig. 4. SLP methodology. 561M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy plant, designing the plant and constructing the buildings and their facilities. (Fig. 3) Industrial Architecture takes care of these three last stages of the industrial plant project, and especially the third one: design and project of the facilities required to develop a certain activity. The goal of floorplan design is to order the production means: workers, materials, machinery, furnishing and auxiliary services (maintenance, transport, etc). A good design is the one that finds an order of the different elements that is the most economical one for the work and also the safest and most satisfactory one for the employees. Historically, there are three basic types of floorplan: per product (also know as serial), process or section distribution, and product fixed position. The requirements of the current market have led to the creation of flexible manufacturing cells (FMC), which are an implementation of FMS (flexible manufacturing systems). FMS consist of a set of machines and tools together with a material handling system (everything controlled by a computer). The study of methodologies for floorplan design was carried out intensively in the fifties (Immer [5] and Buffa [6]). In 1961, Muther [7] presented “Systematic Layout Planning”the SLP method (represented in Fig. 4), which includes the flow of materials and works for every kind of floorplan, independently of their nature: hospitals, shopping malls…The wide acceptance of this method are the reason of not other important research having been done in this topic. This does not mean that the problem has lost its importance, but completely the opposite: once an almost unanimous agreement about the methodology to use was reached, many later studies have focused on the two fundamental steps of the method: layout generation and solution optimization. 3.1. Layout generation methods Layout generation methods do not try to exhaustively enumerate all the possible solutions, but also act as an initial filter. Many methods have been proposed, and their study would take a whole thesis. That is why just the main methods and their original creators will be enumerated.(Fig. 5). Three different classifications have been established: depending on the goal function type, depending on the solution generation method and depending on the way of locating activities. 3.1.1. Depending on the goal function type First of all, depending on the type ofgoal function: quantitative functions usually try to minimize the flow of materials, neglecting the cost of locating activities [9]. Qualitative methods are based on evaluating the need for activities to be close to each other from different viewpoints. That is made through a coding (A, E, I, O, U, X) that values the need of such proximity (Fig. 6). This table represents a multi-criteria approach to the problem, as shown in Fig. 7, where the criteria which are considered to be relevant (according to [10]) when designing a floorplan are shown. The most important ones are the multi-criteria techniques, unanimously accepted nowadays, but with different formulations depending on the author: - The most typical implementation is to use a weighted sum of different criteria. Each weight represents the relative importance of each criterion. Fig. 6. Table with relationships between activities, from [11]. Fig. 5. Methodologies for generating solutions, from [4]. 562 M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy - The same thing can be done in a non-additive fashion, as in [8], where a hierarchy is established in order to choose an alternative. 3.1.2. Depending on the solution generation method Another classification, depending on the solution generation method, can be made: -Construction methods: create a new floorplan design, given a floor of known or unknown dimensions. The algorithm selects and locates the spaces sequentially. -Improvement methods: the spaces on a given floorplan design are rearranged in order to get an improved distribution. -Hybrid methods: those which combine the previous two. 3.1.3. Depending on the way of locating activities The third and lastclassification depends on the way of locating activities: (a) using discrete techniques, (b) using an analytic formulation of the problem or (c) using slicing algorithms. a) Methods based on discrete techniques separate the initial domain D of area A into a mesh of n sub-domains, each o area a (so that A=n x a). A set of m activities, which require a set of areas a i , is to be located (Fig. 8). The interaction between them is described by the relational intensities w ij . A set of n i squares is assigned to activity i. Then, a distance matrix d, which represents the distances between the squares in the mesh, is created. For instance, d kl would be the distance between the elements k and l. An assignation variable x ij is then defined so that: xij ¼1 if the activity ioccupies position k: xij ¼0 otherwise: The relational cost of the system is defined as: C¼X m i¼1 X m j¼1 X n k¼1 X n l¼1 wij nidnj ddkldXik dXjl with the following constraints: X n k¼1 Xik ¼ni8i;i¼1;2;N;m X m k¼1 Xik ¼18i;i¼1;2;N;n Xik ¼0or1 Xik ¼X2 ik  The goal is to minimize the C function, which represents the cost of the flow of materials between the activities. Some restrictions must be accomplished, though: - The number of squares occupied by activity i must be equal to n i . - Each square must belong to just one activity. b) Analytic formulation (Fig. 9) sets an initial space and n activities with a given geometry. Fig. 10. Distribution obtained by analytic methods. Fig. 9. Analytic formulation. Fig. 8. Discretizing the initial domain. Fig. 7. Qualitative factors that influence floorplan design. 563M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy The goal is once more to minimize the relational cost, given by the expression: min X n i¼1 X n−1 j¼iþ1 CijdfijdjXi−XjjþjYi−Yij  C ij Transport cost (per length unit) between activities i and j f ij Flow of materials between activities iand j X i Distance between the centre of gravity of activity iand the Yaxis Y i Distance between the centre of gravity of activity iand the Xaxis Always taking into consideration the fact that the algorithm cannot generate a superimposition of activities, the result is a distribution which is similar to the one depicted in Fig. 10. c) Slicing algorithms start from an initial domain Dof area A which is rectangular. The goal is to locate nactivities with rectangular shape by recursively dividing the initial domain (Fig. 11). The relational cost is in this case defined as: Cij ¼Wij dij where W ij represents the relational intensities and d ij represents the distances between the activities (usually between their centers of gravity). The goal is again to minimize the relational cost, with the constraint that the sum of the areas is equal to A(the area of the initial domain D). Slicing algorithms divide the initial domain recursively by the means of slicing operators: letters that represent the relative position between activities or clusters. A previous step is to group the activities according to their affinity (dendrogram, Fig. 12), which is accomplished by clustering techniques. The most sophisticated ones are the ones used in [12]. If information about how the recursive cut process is carried out (to obtain the activity distribution) is added to the dendrogram, it becomes a slicing tree (Fig. 13). The letters on the leaves (u, d, l, r) represent the way of making the cuts: up, down, left, right. The slicing tree is a schematic representation of the floorplan design process, as it includes information on how the cuts were made. If a criterion to go along the tree is properly defined, a solution can be univocally defined by a code. For instance, for the tree in Fig. 13, a coding based on pre-ordering would lead to: rr1d23u4d56 This property lets one represent a set of solutions (that one may want to optimise later on) very easily using their codes. 3.2. Techniques for solution optimization There are also many different techniques for solution selection currently available. A classification is represented in Fig. 13. Slicing tree. Fig. 12. Building the dendrogram. Fig. 11. Solution generation by means of slicing algorithm. 564 M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy Fig. 14, which includes the origin of the technique and the first author who applied it to floorplan design. 4. Differences between the composition processes “Until the end of the eighteenth century, the master builder possessed two personalities; the one of the architectonic creation, thought as work of art, and the one of the technique, using its construction methods. But, in the nineteenth century, and almost simultaneously, the Ecole de Pont et Chaussés (where Perronet dreamed of educating new engineers) and the Ecole de Beaux Arts (where the new geniuses of Architecture would be forged). Someone thought that it would be better like that, and the divorce was forced from the beginning, aiming each profession at a different direction.”[21]. The different visions which engineers and architects have from the floorplan design problem are can be appreciated in several aspects. In order to understand these differences, both composition processes are analyzed, reviewing the three most fundamental aspects: the methodology for obtaining solution, the definition goal function which is utilized to compare solutions, and the election of the optimal solution. Fig. 14. Techniques for solution optimisation, from [4]. Fig. 15. Comparison of the composition processes. Fig. 16. Comparison of methodologies for floorplan design. 565M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy 4.1. Methodology In Architecture, the design process has traditionally been a matter for debate. Many theories and thinking lines coexist, there is no consensus. In spite of that, the most recent trends in the composition process have unified themselves, grouping the different steps in the design process into three stages, which have different names depending on the author but still include the same contents: -Analysis stage: consists of the preparation of the information, the listing of requisites, the definition of goals, the planning of needs…Known among authors as: architectonic design,intention,architectonic diagnose, functional level… -Synthesis stage: the one in which solutions are generated. Current authors call it: search for the architectonic object by graphic simulation,planning,layout schematic design,topological level… -Evaluation stage: the different designs are compared and the appropriate one chosen. In order to confirm this theory, it is enough to analyze the historical revision of the processes of architectonic composition that has been made in Section 2. On the other hand, in Engineering, there has been a very short period of time (1950–1961) in which a methodological discussion of the implantation problem has been carried out, andfromthenonageneralacceptancebyallresearchersofthe S.L.P. methodology. 4.2. Goals to achieve Both disciplines differ also radically here. In Architecture, even in the goals lie clearly in its definition (art to the service of the man), they are not quantitatively defined in any of the many existing theories. This is undoubtedly due to the difficulty of parametrizing the aspects of goals as beauty, comfort, functionality… On the contrary, enumerating the goals is natural as Engineering and common to all authors. For engineers, who approach the problem from the industrial floorplan design point of view, it has been easier to parameterize their goals. Authors have commonly used the flow of materials, as well as some other variables more difficult to quantify: covered distance, security, noises, distribution flexibility…This way, the problem acquires a multi-criteria nature, with a great impact over the past years. 4.3. Election of the optimal solution Finally, another differentiating characteristic is the election of the optimal solution. In Engineering, as reviewed in Section Table 1 Comparison of models for architectonic floorplan design Author Technique Input data Roth et al. [13] Lattice theory •Adjacency relationships: desired, undesired and forbidden •Dimensions: Length (max & min.) Width (max & min.) Coyne [1] Expert system •Connections through links, represented by matrices with zeros and ones •Dimensions: Orders the size of the rooms Canivell [14] Lattice theory •Adjacency relationships (1,0) or (yes, no). Orientations are considered here •Dimensions: Surface (minimal) Lengths (minimal) •Partition walls: Directions are fixed (horizontal, vertical) Medjdoub [3] Object-oriented programming and constraint programming •Adjacency relationships (1,0) or (yes, no). Orientations are considered here •Dimensions: Surface (max. & min.) Width and length (minimal) •Corridor length: It is defined as a room Del Río [4] Slicing trees and Fenh-Shui •Architecture version: Binary adjacency relationships (1,0) •Rectangular floor (width×length) Surface of the rooms Author Results Optimization Remarks Roth et al. [13] One solution Minimize the total area of the contour that contains all the spaces. Coyne [1] One initial solution,then modify the resulting layout No function to be optimised Not very suitable for comparing Canivell [14] All the possible solutions Length of walls The outer contour is not rectangular Medjdoub [3] All the possible solutions Length of corridors and walls It has initial space orientation constraints Del Río [4] One solution Maximize required adjacencies (minimal transport cost) Table 2 Classification of architectonic models Goal function Geometric Area Ligett and Mitchell [15], Roth et al [13] Length Canivell [14], Medjdoub [3] Relational Del Río [4] Others Coyne [1],Kovács[16],Maculet[17], Schwarz et al [18],Flemmingetal[2],Charman[19], Smith [20] 566 M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy 3, the use of layout generation methods for the search for all the solutions that fulfill the requirements, involve the appearance of techniques which aim at maximizing the goal function. On the other hand, the latest techniques in architectonic composition show to the architect all the distributions that are compatible with the constraints, so that he can choose according to his personal and hence subjective criterion. All the mentioned differences are summed up in Fig. 15. 5. The search for a common path “The occupations of architect and engineer adopted divergent trajectories, until creating a vacuum between them which few dare to step on. In the few occasions that someone has been talented or decided enough to place himself with authority in such intermediate position (Maillart, Nervi, Nowick and sometimes Wright), the results have been extraordinary”[22]. If the problem is analyzed paying attention to the input data, the used methodology, the output results, and the pursued objectives, and how architects and engineers approach the problem, the following conclusions (which show up the similarities between both approaches) can be reached. In general, the input data consists of an initial contour, certain rooms to be located (with certain dimensional requirements) and some relationships between rooms or activities. The results are the position or distributions of these activities or rooms. This is common to both disciplines. The employed methodologies tend to get closer to each other. In fact, the last trends in Architecture, represented by ARCHIPLAN [3], and the traditional engineering technique SLP are rather similar, as it can be inferred from Fig. 16. Regarding the obtaining of solutions, and as a consequence of the current wide use of numerical methods among architects, the utilization of modern optimization techniques has spread. Some cases are shown in Table 1. The main differences between the approaches by engineers and architects have to do with the goals.The importance of composition and aesthetic factors in Architecture lead to goals as functionality, habitability, balance, beauty, etc, which are in many cases subjective and hard to turn into parameters. On the other hand, efficiency, cost reductions, etc., are the indices used in Engineering. But it is also important to mention the convergence between both disciplines: architects have understood the importance of quantifying how good a design is, (in Table 2, some studies in which different numerical goal functions are used are classified) and engineers have understood that the efficiency of a design is closely related to the employees' satisfaction and motivation, so that one must take care of the design of working position, considering issues as security, ergonomics, and beauty of the industrial facilities. 6. Conclusions “Independently from the simplistic idea that the architect is the master of the shape and the engineer is the master of the technique, the goal of looking for a global quality must be reached: a good adjustment between the project and its goals, the obtaining of satisfactory shapes that make one feel, as well as the economy of the construction.”[23]. Floorplan design stays as a current problem, both in Architecture and Engineering, especially since its solution is a key step towards the design of a building or an industrial plant. In this paper, two ways of historically approaching the problem have been reviewed, but the authors (two engineers and an architect) intended to emphasize the similarities that both visions show. Section 5 concludes that there are many more aspects joining rather than separating engineers and architects. It is firmly expected that technological advances will lead to a less subjective architecture and to a more human industry in which both visions about the floorplan design problem should converge. References [1] R.D. Coyne, Logic Models of Design, Edit., Pitman Publishing, London, 1988. [2] U. Flemming, et al., SEED-Project, Journal of Architectural Engineering 1 (1995) 147–203. [3] B. Medjdoub, Methode de conception fonctionnelle en architecture: une approche CAO basée sur les contraintes: ARCHIPLAN, Ph.D. Thesis, Ecole Centrale de Paris, 1996. [4] M.G. Del Río Cidoncha, Un modelo para el diseño de distribuciones en planta en arquitectura. Ph.D. Thesis, Escuela Superior de Ingenieros, Universidad de Sevilla, 2002. [5] J.R. Immer, Layout Planning Techniques, Edit, Mc Graw Hill, New York, 1950. [6] E.S. Buffa, Secuence analysis for functional layouts, The Journal of Industrial Engineering (March 1955) 12–25. [7] R. Muther, Systematic Layout Planning, Industrial Education Institute, Boston, 1961. [8] J.J. Cano, Implantación de la Planta Industrial: Planteamiento racional de la implantación, report within his competitive exam for the position of university professor, Universidad de Valencia, 1987. [9] J.A. Tompkins, J.A. White, Facilities Planning, Edit, Wiley & Sons, New York, 1984. [10] C.M. Harmonosky, G.K. Tothero, A multi-factor plant layout methodology, International Journal of Production Research 30 (8) (1992) 1773–1789. [11] R.L. Francis, J.A. White, Facility Layout and Location, an Analytical Approach, Edit, Prentice- Hall Inc., Englegood Cliffs, New Jersey, 1974. [12] K.I. Tam, A simulated annealing algorithm for allocating space to manufacturing cells, International Journal of Production Research 30 (1) (1992) 63–87. [13] J. Roth, R. Hashimshony, A. Wachman, Turning a graph into a rectangular floor plan, Building and Environment 17 (3) (1982) 163–173. [14] J. Canivell Rodriguez, Procedimiento interactivo para la generación y optimización de organizaciones espaciales en planta, Ph.D. thesis, Universidad de Sevilla, 1986. [15] R.S. Ligett, W.J. Mitchell, Optimal space planning in practice, Computer Aided Design 13 (5) (1981) 277–288. [16] L.B. Kovacs, Knowledge based floor plan design by space partitioning: a logic programming approach, Artificial Intelligence in Engineering 6 (4) (1991) 162–185. [17] R. Maculet,Représentationdesconnaissances spatiales (Algébre de Manhattan) et raisonnement spatial avec contraintes, Ph.D. thesis, Université Paris 6, France, 1991. [18] A. Schwarz, D.M. Berry, E. Shaviv, On the use of the automated building design system, Computer Aided Design 26 (10) (October 1994) 747–762;. Representing and solving the automated building design problem, Computer Aided Design 26 (9) (September 1994). [19] P. Charman, Gestion des contraintes géométriques pour l'aide à l'aménagement spatial, Ph.D. thesis, ÈcoleNationaledesPonts et Chaussées,November 1995. 567M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Author's personal copy [20] I. Smith, Creative Design Objects From Cases For Interactive Spatial Composition, 4th International Conference on Artificial Intelligence in Design (AID'96), Stanford University, 1996. [21] J.L. Manzanares Japón, La perplejidad: un Ingeniero entre Arquitectos. Inaugural lecture of the 97/98 academic year at University of Seville. [22] F. Candela, En defensa del formalismo y otros escritos, Xarait Ediciones, Madrid, 1985. [23] O. Arup, Building design for energy economics, Longman Group, United Kingdom, 1981. Further reading [1] J.C. Lebahar, Le dessin d' architecte, Edit. Roquevaire: Parenthese, Marseille, 1983 and Etude en CFAO, Architecture et batiment, Edit. Hermes, Paris, 1985. [2] P. Quintrand, et al., La conception assisteé par ordonnateur en Architecture, Edit, Hermes, Paris, 1985 y Perspectives de la CAO en architecture, Revue Internationale de CFAO et d' Infographie, vol. 7, 1992, pp. 63–69. [3] R.D. Coyne, J. Gero, Knowledge-Based Design Systems, Edit, Addison- Wesley, New York, 1991. [4] G.C. Armour, E.S. Buffa, CRAFT: a heuristic algorithm and simulations approach to the relative location of facilities, Management Science 9 (2) (January 1963) 294–309. [5] M.J. Rosenblatt, The facilities layout problem: a multi-goal approach, International Journal of Production Reseach 17 (4) (1979) 323–332. [6] J.M. Seehof, W.O. Evans, Automated Layout Design Program (ALDEP), The Journal of Industrial Engineeering 18 (12) (1967) 690–695. [7] C.E. Donaghey, V.F. Pire, Solving the Facility Layout Problem with BLOCPLAN, Industrial Engineering Departament, University of Houston, TX, 1990. [8] P.C. Gilmore, R.E. Gomory, Multistage cutting stock problems of two and more dimensions, Operation Research 13 (1963) 94–120. [9] S.S. Heragu, A. Kusiak, Machine layout: an optimization and knowledge based approach, International Journal of Production Research 28 (4) (1990) 615–635. [10] L. Stockmeyer, Optimal orientations of cells in slicing floorplan designs, Information and Control 57 (2/3) (1983) 91–101. [11] S. Kirkpatrick, C.D. Gelatt, M.P. Vecchi, A heuristic algorithm and simulation approach to relative location of facilities, Optimization by Simulated Annealing 220 (4598) (1983) 671–680. [12] F. Glover, Tabu Search: Tutorial, Technique Report, Center for Applied Artificial Intelligence, University of Colorado, 1989. [13] K.I. Tam, Genetic algorithms, function optimization y facility layout design, European Journal of Operational Research 63 (1992) 322–346. [14] J. Grobelny, On one possible fuzzy approach to facilities layout problems, International Journal of Production Research 25 (8) (1987) 1123–1141. [15] O. Akin, Models of architectural knowledge: an information processing view of architectural design, Ph. D. Thesis, Carnegie-Mellon University, Pittsburgh, 1979. 568 M.G. Del Río-Cidoncha et al. / Automation in Construction 16 (2007) 559–568 Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 International Journal of Production Research, Vol. 45, No. 15, 1 August 2007, 3457–3476 A multidisciplinary model for floorplan design G. DEL RI ´O-CIDONCHAy, J. MARTI ´NEZ-PALACIOSy and J. E. IGLESIAS*z yDepartamento de Ingenierı´a Gra ´fica de la Universidad de Sevilla, Escuela Superior de Ingenieros, Av. de los Descubrimientos s/n, 41092 Seville, Spain zDepartamento de Fı´sica Aplicada III de la Universidad de Sevilla, Escuela Superior de Ingenieros, Av. de los Descubrimientos s/n, 41092 Seville, Spain (Revision received April 2006) The aim of this work is to develop and implement a model for facility layout design. The first step will be to study the traditional process of positioning, locating, or distributing the different facilities, which will be done with the help of slicing trees. Afterwards, the connections between the facilities will be handled, looking for the shortest path that communicates them all, by means of an expert system. Finally, the orientation of the original surface will be treated, using rules from feng shui. These three aspects and their different techniques will be integrated into a single application, with the addition of knowledge and experience from architects and engineers. The space-distribution algorithm of the application itself provides excellent results. Apart from that, the application considers the routing problem, especially important in architecture (but not so important in industrial plan design, when there are no walls between facilities). The application also considers the orientation of the surface, which helps to improve the comfortableness of the building and power saving. The main advantage of the application is that, while being a good tool for solving the facility layout problem, it also considers the routing and orientation of the building. Keywords: Architecture and industry; Layout optimization; Slicing trees; Expert system; Routing 1. State of the art in floorplan design The general design process is first explained in this section. Then, the different approaches to the floorplan problem by engineers and architects are reviewed. Finally, the different methodologies for floorplan generation and solution optimization techniques are also reviewed. The goal of the design process is to obtain a description of a product or service satisfying certain needs and wishes. Independently of the way in which the process is characterized and the type of product or service, the stages depicted in figure 1 are always present. *Corresponding author. Email: [email protected] International Journal of Production Research ISSN 0020–7543 print/ISSN 1366–588X online ß2007 Taylor & Francis http://www.tandf.co.uk/journals DOI: 10.1080/00207540600889550 Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 .Dynamic programming: improves branch-and-bound, suppressing redundant partial trajectories. .A: combines dynamic programming and subestimation incorporation. 3. Development of a method for solving the floorplan design problem There are three steps necessary to make a floorplan design. In the first step, the relationships between different modules, their distribution or location, are established. In a second step, the corridors that communicate the facilities with each other in the simplest and fastest way are found. This task is known as routing. In the last step, the spaces are geographically oriented. For the locating problem, slicing trees were used, extending a technique for VLSI circuits to the floorplan design problem. For the routing problem, an expert system based on rules for production was chosen as the most appropriate for the functional requirements. For the orientation problem, a computer assisted numerical method was used for handling the recommendations provided by feng shui. These three modules are synthesized in the proposed Cut-Expert-feng shui (CEF) methodology. 3.1 Floorplan distribution or locating It is important to bear in mind that floorplan design involves ‘ordering in the most efficient way certain spaces between which certain interactions exist: flow of materials, resources or information’ (Tompkins and White 1996). Some criteria must be set to evaluate which alternative is the best. These criteria aim at maximizing habitability and beauty in architecture and at maximizing efficiency and minimizing costs in engineering. Figure 8. Search tree from the problem in figure 7. 3464 G. del Rı´o-Cidoncha et al. Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 3.1.1 Problem statement. In general, the starting point of the floorplan design problem consists of: .a domain floor D with a given geometry and area, D(A); .the spaces or activities to distribute, with area a i and geometry D i (a i ); .proximity requirements between certain spaces or activities, know as relational intensities w ij . The goal is to minimize the total relational cost of the system: C¼X m i¼1X m j¼1 wijdij, while keeping the area under the limits of the original floor: X m i¼1 aiA: The distance d ij is the distance between the centres of gravity of the activities iand j. Moreover, the following constraints must also be satisfied: qi, minqiqi, max 0oioi, max i,j¼1, 2, ...,m, where q i is the aspect ratio of activity i: length divided by width. Setting a limit for this value (Del Rı´o-Cidoncha 2002) prevents the algorithm from generating overly elongated spaces, which are especially inappropriate in architecture. On the other hand, o i is the dead area ratio of activity i. This parameter indicates the existence of occupied areas in the initial domain floor Dand the effect this has on the shape of the solution. 3.1.2 Utilized methodology. The proposed CEF methodology divides the process into three steps: partitioning, orienting and ordering (described in section 2.1). For partitioning the initial rectangle, as CEF can handle just this type of geometry, the ratio-cut algorithm was used because it prioritizes the balance between the areas to both sides of the cut figure 9, in which A i and B i are the subgroups of rooms or activities created by the cut, shows that the design parameter for this algorithm is the ratio between the weight of a cut and the product of the sizes of the sub-trees. The heuristic of the algorithm is to minimize that parameter. When the value of the parameter is equal for two distributions, the distribution with a larger denominator or the distribution with a more similar area is chosen. For orienting, the Lather method (Lengauer 1990) was used (figure 10). This proposes orienting the cut so that it is parallel to the shortest side of the partitioned surface. It requires a continuous analysis of the shape of the partitioned surface, so that all the resulting shapes are under control. For ordering, the purpose of the CEF program is to minimize the function that represents the transport cost between spaces (figure 11). All the possible distributions are calculated, and the distribution that provides the minimum cost function value is spared. Multidisciplinary model for floorplan design 3465 Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 3.2 Routing Once the locating problem is solved, the next step is to find the smallest net that interconnects all the departments or modules. A method for solving this problem has been searched for. The algorithm for looking for the tree that provides a minimum laying was adapted, at the same time analysing the connection between square spaces with equal size and the best way of joining sub-trees that interconnect groups of modules. Synthesizing these three analyses, an algorithm that reaches a local minimum in a reasonable time was obtained. The algorithm will be translated to production rules, so that an expert system for module connection is available. According to the terms defined in figure 12, the method operates in the following way: .Mark the nodes which are not desired to be part of any corridor as forbidden. Mark the input node as compulsory. .Mark the branches that lead to forbidden nodes also as forbidden. Figure 9. Ratio-cut algorithm. 3466 G. del Rı´o-Cidoncha et al. Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 Figure 11. Minimization of the goal function, that is, the sum of distances between the centres of gravity of the related spaces or activities. Figure 10. Lather method. Spaces with elongated shapes are avoided. Multidisciplinary model for floorplan design 3467 Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 .Look for the nodes that are marked as compulsory or have been configured thus in the process of forbidding nodes or branches. This way, there will be no isolated modules. .Check the different alternatives for sub-corridors, and in case any of them have just one alternative, make that node and the branch that communicates it part of the sub-corridor. .When no more branches can be added, the longest non-forbidden branch is eliminated. Once a branch is discarded, the process continues until all the sub-corridors are merged into one, thus connecting all the spaces. At this point, a solution has been reached, but it has to be revised in order to look for local improvements (for example, branches that form loops). 3.3 Orientation in floorplan design The third and last stage of the methodology stage is the orientation. The empirical rules depicted in figure 13, which depend on the type of space and are related to the cardinal points, were obtained from feng shui. In order to put together the feng shui recommendations on relationships between spaces, a diamond-tipped relational diagram (figure 14) has been used. Upper-case letters represent the relationships that are directly derived from previously drawn conclusions. Lower-case letters represent the relationships that are introduced from the PAKUA anagram. The A, U, and X categories are used in the first case. The rest are used for the inducted categories. When these recommendations are analysed, their parallelism with the occidental house conception can be appreciated. 4. Description and analysis of the results The purpose of this section is to validate the CEF method that has been proposed as a solution to the floorplan design problem, for both industry and architecture, by Figure 12. Definition of terms used in the heuristic. 3468 G. del Rı´o-Cidoncha et al. Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 Figure 13. Orientations recommended by feng shui. Figure 14. Diamond-tipped relational diagram. Multidisciplinary model for floorplan design 3469 Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 applying it to different problems that have already been studied by other authors. As the CEF method consists of three modules (locating, routing, and orienting) that work independently, their efficiency will be initially tested one by one. A later experiment will test how the system works on the whole. 4.1 Locating module in engineering In order to develop the experimental part of the locating module in engineering, the problem posed by Francis and White (1974) is used. For validating the proposed method, the results will be compared with those from other authors that have studied the same floorplan design problem. The selected methods are .Francis and White method; .Kar Yan Tam (KYT) method (Tam 1992); .ARBOGEN method (Santamarina Siurana 1995). The Francis and White method, based on a CRAFT (Armour and Buffa 1963) improvement method, requires an initial solution as a starting point. KYT is a hybrid method, based on genetic algorithms and slicing trees. It utilizes clustering techniques to generate a dendrogram that represents the structure of the tree. ARBOGEN is also a hybrid method based on genetic algorithms and slicing trees. As opposed to KYT, it also admits different slicing tree structures, applying the genetic algorithms in a double level in order to look for the most adapted species. The different results are presented in figure 15. The results are compared according to the following criteria: .Resulting geometric shapes: the comparison with the Francis and White solution is not possible, as its geometry is not rectangular. The other represented solutions use slicing trees and have rectangular geometries. In these cases, it is interesting to calculate the length–width ratio for the spaces, setting certain limits for acceptable values. This ratio indicates the quadrature of the spaces. The mean values are shown in table 1. .Transport costs: to make a quantitative comparison, these were calculated by multiplying each flow by the corresponding transport cost and by the distance between each pair of activities. The sums of all the partial costs are shown in table 2. CEF 1 is exactly the same as CEF but with binary relational intensities. It can be deduced in table 2 that the distribution obtained with the method proposed in application 1 achieves a transport cost 17.7% and 41.3% less than KYT and ARBOGEN, respectively. Application 2 achieves an even lower cost, showing the importance of utilizing integer relational intensities in the design program, as the approximation to the real problem is larger. 4.2 Locating module in architecture It is important to emphasize first the impossibility of quantizing aesthetic and composition factors, such as habitability, balance, and beauty, which are important goals of an architectonic distribution. Moreover, as opposed to what happens in engineering, there are several strategies to approach the floorplan design problem. 3470 G. del Rı´o-Cidoncha et al. Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 This review, that is the design process itself, shows the composing process (the main discussion point) and its formalization and conceptualization (most of the researchers’ aims). There are two differentiating aspects between the different models: the number of solutions that are obtained and the type of goal function (table 3). Most of Figure 15. Obtained results. Table 1. Mean quadrature ratio. Method KYT ARBOGEN CEF Mean ratio 3.39 2.04 1.75 Table 2. Transport costs. Method CRAFT KYT ARBOGEN CEF-1 CEF-2 Cost 4521.5 6324.2 7470.5 4379.6 4246.2 Multidisciplinary model for floorplan design 3471 Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 the architecture-specific applications provide all the possible solutions, leaving the definitive choice in the professional’s hands. It is in the goal function where more diversity between methods can be found, even if many researches focus on optimizing dimensional aspects (such as wall lengths, areas, and so on), summed up in different parameters. Fewer try to maximize the proximity between spaces with related needs, the single goal for every industrial procedure. 4.3 Routing module The problem of connecting a set of spaces, whereby each can be connected to the net by any point of its perimeter, is a rather particular problem, and there are few such studies or resolution algorithms, as its most important application is floorplan design. Thus, programs outside floorplan design had to be searched for to compare the routing module application. Another field in which trees that interconnect points are searched is microelectronics. In this context, the ORCAD-PCB program is presented. This is a commonly used program for printed-circuit wiring. Comparing both solutions, search times are found to be on average 62% longer in CEF than in ORCAD-PCB. On the other hand, CEF takes advantage of this extra time to yield a shorter solution (4.5% of the length of the net on average). 4.4 Complete method All the abilities of the application will now be used at the same time: locating the spaces, orientating the floor, and obtaining the corridor that communicates all the modules with a minimum length. In order to evaluate the results, the floor of an already-built house is compared with the solution from the CEF program. The construction of the house being studied was finished in 1994 and was designed by the architect Jaime Lo ´pez de Asiaı´n y Martı´n. Its floor is shown in figure 16. The solution provided by the CEF method is shown in figure 17. Table 3. Different approaches based on the number of solutions and type of cost function. Number of solutions obtained Single solution Roth et al. (1982), Del Rı´o-Cidoncha (2002) All possible solutions Ligett and Mitchel (1981), Canivell (1986), Coyne (1988), Kovacs (1991), Maculet (1991), Schwarz et al. (1994a, b), Charman (1995), Flemming et al. (1995), Medjdoub (1996), Smith (1996) Type of goal function Geometric Area Ligett and Mitchel (1981), Roth et al. (1982) Length Canivell (1986), Medjdoub (1996) Relational Del Rı´o-Cidoncha (2002) Other Coyne (1988), Kovacs (1991), Maculet (1991), Schwarz et al. (1994a, b), Charman (1995), Flemming et al. (1995), Smith (1996) 3472 G. del Rı´o-Cidoncha et al. Downloaded By: [Universidad de Sevilla] At: 17:54 17 February 2008 The CEF solution is quite similar to the floor designed by the architect. Considering the locations of the rooms, only the kitchen–dining-room cut (via a vertical wall) is different to the separations on the real floor. This is due to the weak relationship established between the kitchen and the other rooms by feng shui, which makes the separation between the kitchen and the other rooms the first cut to be performed. The routing module of the application was employed to establish the path that communicates all the rooms. The results are shown in figure 18. Figure 17. Solution with CEF. Figure 16. Floor of the house studied. Multidisciplinary model for floorplan design 3473