Stochastic Analysis Computation User Manual
Contents
1. E E Vigal al C XwINNTSProfileskqjantonioXE scritorioX5tac 30 1 Miel ES E Ficheros de Daf SINT RECALCULO em viga cal c Opel nb 36PESO PROPIO RECALCULO is i p IERI 38 RECALCULO RECALCULO Hombre Variable young E NO ACTIVO ls Modificar Variable DENES E Eliminar Variable DONNA 3 a D E 46 E AA S 2 Hombre Variable young o el ac Modificar Variable u Bo a Analisis E Eliminar Variable AZ Vista opci n c L Donais 7 TO Dibun Meda Sey Ts Than nad Normal 210000000000 21000000000 45 D Nombre Variable young 1 240510E 5 167 s E Modificar arable Eliminar Variable Fara ayuda pulsar F1 Fil 45 Col 29 2 1 3 4 1 Modifying a Variable If the Modify option was selected STAC will present the adequate window depending on the variable type the PDF window the Expression Calculator orthe Output Variable window Then all the parameters can be reentered and changed except for the variable name that can not be modified in any way If an independent variable is modified and it is associated with some dependent variables SIAC will recalculate all these dependent variables informing if the procedure wassuccessful or not Modificar young La variable que va a modificar young est asociada a la variable dependiente dep voung Para que la modificaci n se lleve a cabo se modificar la Dependiente asociada dm If some2. Variables Ant Hill Plat DO desplazamie C Correlacion Datos de la Grafica Info Momentos Estadisticos Max X2 Max Y Media X Media Y Min x Min SDev x 5Dey Y Numero de Tiras Correlaci n Visualizaci n v Mostrar Escala X v Mostrar Escala Y v Ver 5Dev v ver SDev 2x v ver SDev 3x Opciones To display the relationship between two variables it is necessary to select the variables marking the comesponding check box as shown below The first variable will be displa yed on the x axis the second on the y axis Nombre Dist Nombre eee S Seleccione varable para el eje Y Once the two variables are selected the dispersion graph is shown with the statistical moments shown in the next figure and the Pearson linear correlation coefficient Ty y which is calculated asfollows y a PE XY Xy ye Gr fico de Dispersi n Stac Variables desplazamiento r fHBBBBBBz H BBBBBRPH Correlacion Datos de la Grafica Into Momentos Estad sticos Max_X 5 9705e 011 Max Y 6 3848e 007 Media X 2 0897e 011 Media Y 2 1822e 007 Min X 1 7305e 011 Min Y 2 0203e 007 SDev X 2 1800e 010 SDev Y z 3509e 0n amp N mero de Tiros 250 Correlaci n 9 8570e 001 Opciones Salir The statistical moments mean value and standard deviation are evaluate
3. L1 PriBcratch 3 SiSeBatch bat s MCruaBatch bat Stac_O1 Batch bat s PetejanderB atch bat Pe titicanBatch bat pruebaD epsB atch bat fe vigal Batch bat 3 PtBatch bat vigaB atch bat 2 SiaConFiBatch bat Nombre del archiva prueba Vigall Guardar como tipo Ficheros por lates bat Cancelar Abrir como de s lo lectura 3 File Name Helper This is one of the most useful buttons This drop down button consists of the options shown in the figure below and inserts the selected path and the file name into the batch file i a7 aj FEN Agregar Work Dir Tac Agregar Dir Agregar File Agregar Work Dir inserts the Working directory path If it has not been defined nothing is inserted Agregar Dir Inserts the name and path of any directory The directory selection is made using the GUI for file management provided by Windows STAC inserts the short name of a directory to guarantee the compatibility between Win95 98 Me and Nt 2000 Xp platforms Agregar File Insert the name and path of any file The directory selection is made using the GUI for file management provided by Windows It is possible to add the names and path of the input and output variable files to the batch file with a simple click of the mouse left button on the file name that appearsin the file explorer as shown in the figure below E iz Vigal Ficheros para el An lisis Ta
4. STAC Stochastic Analysis Computation User Manual Ant Hill Plat r J 8080882 H BBBBBBH Correlacion Datos de la Grafica Info Momentos Estadisticos _ Max X 5 9705e 011 Max_Y 6 38488 007 Media 2 0897e 011 Media Y 2 1822e 007 Min X 1 7905e 011 Min Y 2 0203e 007 SDev X 2 1800e D10 SDev Y 2 3609e D08 M mero de Tiros 250 Correlaci n 9 8570e 001 International Center of Numerical Methods in Engineering www cimne com Edifici C1 Campus Nord UPC 08034 Barcelona Spain 34 93 4016036 v 11 3 2005 l User Manual 1 1 Introduction This manual will show all the capabilities and functionalities of STAC Some representative examples of the code will be discussed in detail in orderto point out and avoid errors frequently committed by users SIAC stands for Stochastic Analyss Computation and consists of three well defined sages The project definition where the stochastic variables are defined the execution definition of the deterministic code and finally the statistical representation of the generated data 1 3 Project Definition 1 3 1 Introduction A stochastic problem is defined with STAC via the GUI Graphic User Interface called Problem Definition Module PDM In order to define the first stage of any stochastic process the user first needs to analyze a deterministic process and obtain all solution files This means the complete problem has to be executed using all solve
5. delta to obtain the valid boundanes for each class Tous m Num Clases Evoluci n de las Medias Stac Variables Evoluci n de la Media TT A 1 2e 013 1 2e 013 le 013 1e 013 Se 012 Se 012 6e 012 6e 012 4e 012 e 012 e 0lz e 0lz lumi mim lom o o desplazamiento CA Clase 11 Clase 12 M mero de Clases Minimo 1 3613 amp 011 Maximo 2 5025e 011 Incremento 1 2005e 010 Momentos Datos de la Gr fica Info Momentos Estadisticos M mero de Tiras 250 Media z 1822e D07 Var 5 5 40e 016 Visualizaci n v Mostrar Escala X e Mostrar Escala Y v ver Int Conf Sup Max Y 2 8025e 011 SDev 2 3609e 008 Skew 7 7623e 001 Min Y 1 36196 011 ADev 1 8370e 008 Curt 1 7636e 000 C Evoluci n de la Media Histograma N mero de Clases E H Opciones Salir v Wer Int Conf Inf Changing between Histogram and Mean Value Evolution does not affect which variable is analyzed In the Histogram display pressing on any barthe accumulated frequency is shows 1 5 3 2 Statistic al dispersion This option is selected from the main menu with the option Post process gt Dispersion Postproceso gt Dispersi n ascan be seen in the next figure Postpr Momentos Dispersion When this option is selected the following window will appear Gr fico de Dispersi n Stac
6. Dependent variables In both cases the selection and definition of a vanable is very intuitive and simple The next section describes this mechanism 1 3 3 1 1 Independent Stochastic Variables Scalar Variables In order to make the selection of a scalar random variable is necessary to find the comesponding variable field in the proper input file and with the left mouse button pressed select the length of the field Clicking the right mouse button on the selected field will cause a floating menu to appear showing all the probability distributions functions available AZ 10000 fea T o a a A a nm EA E HET ae Escoja Distribuci n me Escoge Distribuci n E LA Discretas i O ig Exponencial A Dizcretaz I EL Gane CO ALA Continues KT P 84 Es m Normal Log ormal Dependiente Rayleigh a l E Uniforme Triangular Beta T1L Gumbel T15 T2L Fr ch t T35 Weibull Chi Cuadrado t de Student LIM P ELESEM A73 Wht aut bw i Tu TS nT n Escoja Distribuci n MAA ALA Discretas Hi Binomial AA Continuas t Geom trica ELEMENT Dependiente Poisson E up CONECT 167 2 12 Laplace 54 rg an Logistica Meeooscooseoccccsosesecosescseceseoseceocesesesocesceoseoceseeosecesosesebcesoccesosceceecosceoceceescososesesoseceseseseceseceseceecseseeeo D l eesccenoseosecsocsooceccosooce
7. file to run the process A batch file consists of a sequence of MS DOS commands to define the working directories the solver commands and all the instructions to complete a deterministic analysis STAC will launch this batch file as many times as the sample size defined by the user to perform a stochastic analysis The ADM providesa specific window in which the batch commandscan be specified Ejecuci n Edici n del Fichera por lotes rel ki cd C CALSEF 1 EXE Viga cal The batch file will be saved in the Working directory This file will be recovered if the project isopened to avoid reentering the commands There is a tool bar available see figure below to simplify writing the batch file A brief description of the functionality of each button will be given next Ela 3 a 2 1 Importa batch file Asthe name indicates this button allows to include in the ADM a batch file previously written the file extension must be bat Two options exist The imported file is a new sequence of commands or the imported file is added to the commands already written Stac v1 0 22 Desea agregar el fichero coma datos adjuntos Would you like to add the file to the current data 2 Save a batch file This option savesthe batch file under another name allowing the userto recover it later With this option the written batch file can be used for another project or program Guardar Fichero Batch Guardar err 3 PriScratch hdd
8. input or output variables can not be defined asan auxiliary file Deleting auxilia ry files is also possible with this option 1 4 4 2 Definition of the Input variable conelation The random number generator implemented in STAC not only produces numbers according to the PDF selected but is also able to establish a correlation between these random numbers This capability is a powerful tool to reduce the number of runs The correlation value between two variables is given in the spread sheet shown below This table isobtained with the Definir Comelacion Define Correlation button of the ADM im Definir Correlaci n Definir Correlaci n entre las Variables Ind Tabla de Definici n Variables densidad espesor Variables Aceptar Cancelar If no value is given a correlation zero is applied Zero correlation is also the default value The comelation between two variablesisa number between 1 and 1 If the introduced absolute value is greaterthan 0 85 STAC suggeststo use dependent variables Tabla de Definicion Variables 2 Stac v1 0 X Li La correlaci n es superior a O85 Para tal caso es recomendable el uso de variables DEPENDIENTES Definicion iables Stac v1 0 x e La correlaci n entre variables ha de estar entre y T 1 4 5 Step 4 Batch command definition The ADM includes many toolsto assist in the creation and testing of the batch
9. 0000 y 273000000000 0 PL E10000000000 B BE1000000000 Nombre Modulo de Ybung Aceptar Cancelar Block or Vector Variables In orderto define a Block or Vector variable it is necessary to follow the steps described next Once the Block has been marked locate the mouse on a colored zone in the Block case the zone is pink and pressthe right button The same pop up menu like that shown for scalar variables will appear The variables included in the Blocks have the same PDF but are characterized by distinct random variables With this option a uniform sample can be created where all the variables have the same probability distribution In this way only one PDF must be defined for all the variables in the Block and the check box Uniformizacion de la muestra must be selected If every variable in the block should be defined with the same PDF shape and different PDF parameters the check box Uniformizacion de la muestra must be unchecked In this case the values of the variables selected in the Block define the PDF para meters for every variable im Distribuci n Normal v 9 659E 00 Datos Distribuci n ES 22 A Mombre Bloque Normales Bloque W Uniformizaci n de la Muestra Aceptar Cancelar Aplicar When the Block is created the name given to every variable in the Block contains the Block Name and an index indicating the position in the Block In the example sh
10. Ficheros de Datos Viga cal CHUTE Ia Ficheros de Resultados E viga sal CM IMNTSPrafiles y El a Ficheras Auxiliares E Petejander kab C WINNTSP Explorador de Proyec 4 Test the Batch File This option is offered to test the sequence of commands in the batch file This utility can serve to validate and correct errors in the deterministic execution 5 Help This button gives accessto a small guide and some suggestions useful for the batch file commands edition The figures below show the batch file for running the example and the DOS window shown when the Test Batch File button was pressed echo oft cd C S5TAC30 1 gt PRJISCR IN CoNCALSEF 1 EXE Viga cal cl 858 B 0 nnnnmegnnnnm 888 8 dts nh EE n 888000888 888 2688b 888 0 888 Y8hood3P 7888 o 08820 888888P Y8hbhod8P 08880 CALCULO DE SOLIDOS Y ESTRUCTURAS POR EL METODO DE ELEMENTOS FINITOS CENTRO INTERNACIONAL DE METODOS NUMERICOS EM INGENIERIA Version 5 1 5 fc CI M H E Copyright BARCELONA ESPADA ESTOY REALIZANDO LOS CALCULOS PRELIMINARES Forrtl El sistema no ha encontrado la ruta de acceso especificada Forrtl severe 29 gt file not Found unit 65 file Presione cualquier tecla para continuar 1 5 Analysis Execution When the process definition is complete the stochastic analysis can be performed In the example shown in the figure below all the process data isspecified Now to
11. Trabajo se guardar n las archivos temporales creados durante la ejecuci n En el Directorio de Servicio se guardar n los resultados selecci n del directorio CMWINNT Profiles jantonioi ESCRIT 15T4 amp C30e 11 obtenidos durante la ejecuci n Directorio de Trabajo Directorio de Servicio 1 4 3 Step 2 Definition of the number of runs Ci WINNT ProfilestjantonioESCRIT 115T4C30e 1 Projects VIGG1 Dez Cancelar Ael The size of the process population or the number of runs in the stochastic analysis is defined in this phase It is also necessary to st the number of wrong runs A wrong run occurs when itis impossible to read any of the output variablesforany reason A good practice Is recommended to consider some processes to fail because the random nature of the input variables can cause a run to fail When the maximum number of wrong runs is reached the stochastic process will stop automatically There are two ways to define these numbers One is via the main menu Ejecutar 2Opciones General and the other is using the ADM In this case the number of runs must be entered in the sample size boxasshown beside Opciones de Ejecuci n Directorios General Ejecuciones 5 Defina el N mero de Ejecuciones que desea realizar 200 Bs Defina el N mero M ximo de Ejecuciones Erroneas permitidas Archivos Temporales Silo desea puede eliminar los archivos temporales generados durante
12. and an output file When defining a variable the user should keep the following points in mind Any violation of these rules will cause the variable creation to be canceled e Only numerical characters must be selected e When there is more than one variable in a row only one can be selected at a time e lf the variable can have a postive and a negative sign and the variable appears as positive in the input data file it is necessary to add a space to considerthe possible negative values Stac v1 0 EXAMINAN TA Proffes jantoro E scribonoxStac 30 10 2001 WWiga cal gt F3 Brchwo Ficheros Edicion Ventanas Ejecutar Fostproceso Administracion Amdo BY dmg dh J8rROR BECALCULO 2 3SLTRIZ MASA RECALCULO FN ars de Des SULLIDA DATOS NO ACTIVO Viga cal CO WINNT Profile ALLIDA TENSIONES NO ACTIVO A LIDA TENSIONES GAUSS WO ACTIVO q31LIDA ERROR ELEMENTAL NO ACTIVO ALIDA FUERZAS INTERNAS NO ACTIVO 45 TERIAL TOUNG EMERBRBEDER POISS0N 0 2 AREA Z5N SET o ASTOMETRIA E 38 CONECTIVIDADES ELEMENTALES SQ ELEMEN SET CONECTIVIDADES pS zi 1 i 64 59 60 5 EM 2 1 6d 60 65 Ef J 3 1 65 60 62 E 4 1 E 54 62 60 57 m li Stac v1 0 E XW INHTXProfilezXjantonioXE scritono Stac 30 10 2001 viga zal gt 1 Ep Pi Archivo Ficheros Edicion Ventanas Ejecutar Postproceso Administraci n Ayuda la xj bd d 152 BS D s3s83z2lE LU OO LS6B199F 06 E Elf v
13. ble de Analisis Descripci n CCar cteres restantes fiat 1 3 4 How to modify vanables Once a vanable has been defined the user has the possibility to modify or unselect tt The procedure isthe same forall kinds of vanablesand isdescnbed below 1 Select the variable to modify or unselect The selection can be made in three different ways In the figure below the three possibilities are shown In all cases it is necessary to pressthe left mouse button a Selecting by the marked zone When pressing the left button on a marked zone blue zone or on the variable field itself STAC selectsthe variable Selecting by the variable name When pressing the left button on variable name shown on the Input Output variable window The variable name will be displayed in the hierarchical structure shown Selecting by the variable descriptor This is perhaps the more interesting option offered in STAC The PDM contains a window called Data Overview Vista de Datos where all the selected variablesare listed and presented with useful information like the variable type size PDF type etc When pressing the variable name on this window STAC will show the variable selected in the corresponding file 2 Once the variable is selected pressthe right mouse button and a pop up menu will Show the options Modify or Delete li Stac v1 D Archivo Ficheros Edicion Ventanas Ejecutar Postproceso Administracion Ayuda S A
14. d to show the results graphically he GUI provided by STAC allows to define and execute all the processes needed to perform a single shoot Additionally STAC offers the possibility to verify the process definition before launching the stochastic analysis This functionality allows to reduce the time needed to define the execution process 1 4 1 Initial considerations In this document the process definition is presented step by step The order followed is not relevant so the usercan complete the steps presented in any order A good practice before beginning the execution process isto declare the location of the Working and Service Directories Both are indispensable for camying out the stochastic execution The user can define and redefine these directories at any time except forduring the processexecution IEEE Postprocezo Administre Ejecutar Ltrl E Wer log Ejecuci n Wer Tiras Efoneos Opciones ADM Analysis Definition Module When the option execute ctrl4E is selected the window shown below appears This window contains all the tools needed to define the execution of a process Ei Stac v1 0 Archivo Ficheros Edicion Ventanas Ejecutar Postproceso Administracion Ayuda co Eel d E E Vigal Ficheros para el An li El viga cal OC WINNT Profile mm E s Ficheros de Resultados E viga sal C WINNT Profile A adir Borrar Ping Editar Ejecuci n Edici n del Fichera por lot
15. d for each variable selected Additionally the correlation is shown in the area below the graph Visualizaci n v Mostrar Escala X Jw Mostrar Escala Y v ver SDew v ver SDev 2x v ver SDev 3x Forthe dispersion post process the PPM allowsto select any two variables no matter if they are input or output variables The order of selection only affects the definition for the x and y axes For the dispersion graph the PPM calculates and show the probability ellipses for 45 60 and 95 of occurence probability These curves can be useful to detect unusual valuesthat can disturb the sample 1 5 3 3 Other functionalities of the PPM Several functionalities ha ve been included in the PPM to manage the data generated asa graphic report orto share these results with other programsusing the Window GUI Pressing the right mouse button on the graph zone causes the following menu to appear Opciones Copiar Guardar Grafica Imprimir Gr fico The options Copy Copiar Save Guardar Gr fico Print Imprimir Gr fico are common to all kinds of graphs displayed The option Options Opciones contains commands corresponding to the type of graph displayed The available options comesponding to the type of graph displayed are described next Options The options window menu can also be displayed using the Option menu from the main window of the PPM The facilities provided with this option are shown in the figu
16. ed by the new analysis in order to maintain the meaning of the vanables Should this not be the case the following eror message will appear The old data can not be used Stac v1 0 9 El formato del fichero de resultados previa es incompatible con el nueva que se va a crear Desea continuar se sobreezcribir n los datos previas In order to keep the generated results it is possible to export the data in two different formats One is Excel compatible and the other one is ASCII format In any case the data is available to run a post process with an extemal application like Excel as shown in the next figure eee Edicion ventanas Ejecutar Postpraces Agregar Fichera k Quitar Fichero Proyecto P Archivo Excel 1 5 2 Cancel an analysis The user can cancel the execution of an analysis when he considers it opportune i e when the number of obtained samples is statistically sufficient The EXM automatically cancels the process of analysis when the number of emoneous samples is equal to the limit defined in the ADM These emors can be caused by very different factors but in all the casesthe symptom isthat the EXM cannot find the valuesof the output variables Cancelar Ejecuci n An lisis Cancelado N mero de errores sobrepasado The EXM also cancels an analysis when the EXM itself is closed This can be done with the button provided to cancel the process or usng the Windows cancel button located atthe top
17. es amp amp bel sl 3 cd C WINNT EFrofiles jentonio ESCRIT 1 STAC3U 1 PE C XWINNT Profiles jantonio ESCRIT 1 STAC3U 1 PRISC Ficheros Generaci n de variables Aleatorias Incluyalelimine aquellos archivos que considere necesarios para la ejecucion A adir Guitar Muestreo Tama o de la Muestra 200 Numero maximo de 5 Tiros Fallidaz Explorador de Proyecto Ei Datos E Resultados iw Analisis Vista de Datos Para ayuda pulsar F1 Is necessary to fulfill all the requirements in this window if not trying to run a process will result in an error message indicating the missing steps as shown in the next figure Stac 1_0 AN Se ha de definir previamente el numera de Tiros Error message It is necessary to set the number of runs 1 4 2 Step 1 Definition of the Service and Working directories Selecting the Options Opciones choice on the Execute Ejecutar menu of the top bar The Service and Working Directory window will appear In the folder marked Directorios there is a button for each directory which allows to change orredefine a new path When a new project is created the default directories are shown However the user can specify any directory including those located in a local area network Opciones de Ejecucion Directorios General Defina los Directorios de Trabajo y de Servicio necesarios para la ejecuci n del an lisis En el Directorio de
18. igat 153 69 0 194754E 07 0 194925E 06 Ea Feheros de Resultados 154 90 0 136650E 07 0 195724E 06 viga sal CNW INNTSProfiled fee 91 0 346640E 07 0 203771E 06 156 az O 466f06nF 05 O Z 03716 06 157 93 0 348166E 07 0 203782E 06 15 94 0 192917E 07 0 214034E 06 159 95 0 129644E 07 0 217833E 06 d 160 arc 0 32459259E 0 DO 22135 7SE 06 E 161 97 O 471680E 06 0 221434E 06 H 162 98 0 351781E 07 0 221710E 06 163 99 0 348833E 07 0 238948E D6 164 100 O 174296F 07 0 238987E 06 i 165 101 0 790144E 10 0 239188E 06 E 165 102 0 173965E 07 0 239644E 06 H L 167 103 0 355926E 07 Ea Fi zr F EjDatos E Resultados on 168 A STAC allowsto define two different types of variables e Scalar Variables e Vector Variables An example of how to select a scalar variable is shown in the previous figures where the user selects only one variable at time A vector variable or better known as a Block is a set of at least two scalar variables of the same length located in the same position in consecutive lines SIAC offers a simple mechanism to select a Block A Block is defined by two points the upperleft comerand the lowerright comerasshown in the figure below The steps necessary to select a Block are l Position the mouse on the upper left comer of the first variable of the Block and then pressthe right button 2 Selectthe option mark the beginning of the Block Marcar Inicio de Bloque 3 A new w
19. indow appearsto select the periodicity of the variables in a Block This functionality is useful since it allows to select only certain fields with a repetitive pattem as shown in the figure below 4 Position the mouse on the lower right comer of the last variable of the Block and pressthe right button again 5 Finish the selection with the option mark end of the Block Marcar Fin Bloque The beginning and end block marks must be in the same file Canceling a Block definition is possible clicking the right button of the mouse outside the selecting area An pop up menu wil appear giving the option to cancel the selection Dehnr par metros selecci n ES AA ca NE r Dehna periodicidad de selecci n Hooo gt Seleccion por bloques lt Marcar Inicio bloque u z aI Due T gH ean Meteat inicio Hogue 7 pd z D Marcar Fin Bloque l 5 poro hero Edicion enanas Ejecutar Fo pre Amrn on Ayuda ous Hi Wigal El po Ficheros de Cabos SOS CODADEDADAS DE LOS HOD El Vigs cal Co wren er ofile zug MODO COORDENADAS 03 1 0000D 0 25000 00000 FONDA e SO0000 d 00560 gaopnocoggoono E E F 5 Lm 11 12 13 1d pE IE 17 iu Li DLS DE 06 Pes ajuda pare FO AA AS HA HA Eu 1 3 3 1 Random variables Input variables In SIAC two typesof stochastic variablescan be defined Independent variables these can be Discreet Continuous
20. ing the stochastic process iy WINNT siapdsim mdl Lista Archivos Agregados x Profiles mE da ma i inn v bu Nombre Fichero Tipo Fichero a tac DEP 9k lina cal CNWINNTSPra File In EEA Stac exe rm Stac 30 10 2001 3 BEDO 1 Bitmaps Shac bp J viga sal CAWINATT Pro File Out Stac vbw 5Lac zip Calculadora stacy D mdl 3 Clases Viga cal P QE Vigaz cal atos CQ Exirmidsnes Insertar coma 2 Fotos Diagramas File In Fite Qu 3 Modulos Tipo de Archivo E Once the project has been created the user has the possibility to add or delete input and output files using the menu command Ficheros Files 1 3 3 Defining random variables The main functionality of the PDM isto allow in a simple and fast way to select and to define the variables of the problem Both the input and the output variables are treated like ASCII strings and are located in a absolute positions in the data and results files These absolute positions are defined by the line and column number in the corresponding file Therefore the user only has to select a value to use as a random variable The specification of the random variables with SIAC is as simple as selecting the text numerical value of our interest with the mouse This process must be repeated for every variable to be defined More details of this one process wil be described in later sections The following figures show this processin an input
21. la ejecuci n del an lisis Eliminar SET Aceptar Cancelar Muestreo Tama o de la Muestra 200 Numero m ximo de 5 Tiros Fallidas Using the main menu command besides the possibility to define the number of runs there is the option to delete temporary files created during the previous analysis for the same project During the firt run of every analysis execution the button Delete Files Eliminar Archivos is disabled however once the analysis has ended the temporary files created can be deleted Only files created by STAC are deleted files created by the solvers are not included in this option 1 4 4 Step 3 Add auxiliary files and Define Input variable conelation 1 4 4 1 Add auxiliary files If the solver orthe deterministic process needs some additional files not produced by the process itself different from those defined the Input and Output variable section is necessary to provide them beforehand in order for them to be available during the sochastic process Examples are files containing material properties groups of nodesor a data structure etc Ficheros Generaci n de variables Aleatorias Incluya elimine aquellos archivos que considere necesarios para la ejecucion A adir Quitar Definir Correlaci n If an existing file in the project is added with this option an error message will inform of this fact It is evident that a file containing
22. ml r rr N mero de Tiros Visualizaci n Momentos v Mostrar Escala X e Mostrar Escala Y Datos de la Grafica Momentos Estadisticos Media 2 1097e4 011 SDev 2 1049e 010 ADev 1 6336e 010 Info M mero de Tiras 91 Max Y 2 47 4e 011 Min Y 1 7560e 011 Evoluci n de la Media Histograma ver Int Conf Sup v ver Int Conf Inf INCISO k pulsar para actualizar datos WOW mu C OH Opciones Intervalo Inferior 95 2 47 741e 011 2 d0527e 011 e 33313e 011 j2ble 011 2 1060be 011 1162e8 011 044508e 01 1 1 97245e 011 1 20051e 011 1 82817 amp 011 1 7560 38 011 em Var 4 405e 00 Skew 1 7547e D01 Curt 4 5259e 001 Salir The PPM offers the possibility of displaying the results in real time taking into account the results already obtained to generate the graphs shown To update the graph the useronly needsto pressthe left mouse button on any of the available variables In the following section more details about the PPM will be explained 1 5 3 Finishing an analysis An analysis can be finished for different reasons a Ihe usercancelsthe execution b The EXM cancels the analysis because the maximum number of allowed wrong processes wasreached C The execution of all the runs was completed In these cases and if the number of correct executions is greater than two the EXM gathersand processesthe results obtained and acti
23. ndow that shows all the details of the execution state as shown in the next figures The information shown includes State of the temporary files necessary to boot the execution inicializando variables State of the evolution of the stochastic analysis for each run and statistics about the numberof completed and emoneousruns Miser e MEE IT ESTER de MonteCarlo Estado de la Ejecucion del Analisis gt Recogiendo valores variables de salida gt Agregando resultado ok Fin tiro 4 250 gt Ejecutando tiro 5 250 gt Generando Fichero Nileatorio ok gt Reemplazando valores variables File In gt Ejecuci n simulador Progreso amp Tiros Fallidas Inicializando analisis Generando File Cabeceras wars File Out Generando File Correlaciones oak Generando File Cabeceras vars File In Generando File Aleatorios ok Generando File Batch ok Copiando Files Entrada y Cabeceras al Workbir Fin inicializacion Ejecutando tiro 177 250 Generando Fichero Nileatorio ak Reemplazando valores variables File In Ejecuci n simulador ak Recogqiendo valores variables de salida Agregando resultado ok Fin tiro 177250 Ejecutando tiro 178 250 gt Generando Fichero Nileatorio ok gt Reemplazando valores variables File In gt Ejecuci n simulador ok gt Recogiendo valores variables de salida Fin tiro 178 250 As can be seen in the previous figure the EXM window contains tw
24. new colorto apply The changesof colorare applied automatically once selected Calor Serie Seleccionada Color BE Default 00000000 OBODODOmO mi i imi miT mn E de la Gr fica BENHENBENMH LALA ero de Tiras 250 EHEHEHEH Y 3 0531 011 Y 1 74232e 011 5 Mas Colores It is also important to consider that when the PPM presents a graph both scales are automatically calculated If the user changesthe scale there is an option which allows to restore the original scale Copy This option allows to the PPM to integrate SIAC with other applications when copying the data generated it is transfemed into a matrix form fully compatible with all of the Windows environment like Excel or gluing the image asa bit map into a Word document Evoluci n Promedios Intervalo Superior 9596 Intervalo Inferior 9596 2 2804e 4011 3 05309e 4011 2 0298e 4011 2 54144e 4011 3 05309e 4011 2 0298e 4011 2 30028e 011 2 85 67 e4011 1 4288e 011 2 19144e 4011 2 6396e 011 1 4328e 011 2 13979e 011 2 5014e 011 1 7819e 011 Save This option allows to the user to save in a file in bit map format A standard dialog boxfor specifying the file name will appear Guardar Imagen J Bbdd Bitmaps Fotos Diagramas zs Calculadora Modulos L Clases PostProcezo _ controles Procratch 2 datos L Projects Hombre del archivo Guardar cama tipo Windows Bit aF file brp Cancelar Abrir como de s lo lect
25. o buttons Post Process and Cancel Execution CancelarEjecuci n The meaningsare obvious The EXM window contains a check boxto show or hide the DOS execution window for the solver The EXM detects that the solver has finished the execution when the DOS window closes In systems based on the MSDOS platform i e Win9x the execution window is not always closed automatically like on NT platforms If this happens the EXM can not detect when the execution of the batch file has finished The solution to this small disadvantage is simple The user must activate the check box close automatically when leaving in the properties menu of the MSDOS window when the first run hasexecuted Forthe next runsthis configuration is no longerneeded ig Ejecuci n del Analisis de MonteCarlo Estado de la Ejecuci n del An lisis Tiros Fallidos Numero maximo de Tiros Fallidos gt Error tiro 176 250 The EXM offers the possibility to restart a previous analysis This functionality is important since it allows to consider previously gathered data The next figure shows a dialog box with the analysis resume question Pressing the Ok button a new data analysis will be saved otherwise the previous data will be used Stac v1 0 m 9 Existe un fichero de resultados de una ejecuci n anterior Desea sobreescribir los datos The previously gathered data hasto consist of the same number and type of vanables as the data that would be creat
26. o ways Real Time Gathering In this way the PPM shows the results of the simulation while the process is in execution In this way the PPM allows to monitor statistically the data generated Analyzing the evolution of the variable s mean helpsto decide if the simulation can be suspended or more runsare needed In this mode the dispersion graphical mode can not be used Differed Time Gather In this way the PPM is used asa statistical data processor Can be executed repeatedly once the execution has finished In both cases the main window of the PPM showsthe input variables independent and dependent and output variables The user can select any of them to analyze their statistic al moments orthe statistical dispersion as will be shown in the next section Nombre Mombre Dist m desplazamiento EN 1 5 3 Statistic al Functions The PPM only supports basic statistical functions These functions can be grouped into two sets functions that operate on one variable statistical moments and functions that operate on two variables statistical dispersion 1 5 3 Statistic al moments 414 This option is selected from the main menu with the option Post process gt Moments Postproceso gt Momentos as can be seen in the next figure This option can also be selected from the EXM as wasexplained in previous sections Po stor Dispersion h When this option is selected the following window
27. ocosocceocooccosoocescososeseceocceeoccescceeecceoecesoscocceoococecceceoson Once the probability distribution function has been selected PDF a new window appears describing the PDF chosen its shape and suggested parameters The user can change the parameters function to fit the PDF at his her requirements A brief example is shown in the next figures In the case of a Normal PDF Gaussian function SIAC automatically assigns the selected value to the average p and a value of 0 luto the standard deviation o Also the interval where the PDF shape wil be shown is defined by the interval u 30 u 30 These values can be modified by the userand are described in detail in ANNEX A LIt Stac v1 0 Archivo Ficheros Edicion Ventanas Ejecutar Postproceso Administraci n Ayuda UP E Es Ef Vigal nfileskjantonioXE scritorioX5tac 30 1 Mal ES E 3 Ficheros de Datos TEMS IONES NO ACTIVO E E Viga cal C WINNT Profil ENSIONES GAUSS NO ACTIVO 0 23 Variables RROR ELEMENTAL MO ACTIVO Em pl young UERZAS INTERNAS MO ACTIVO YOUNG z i1 000008911 POISSON a campo marcado L TIVIDADES ELEMENTALES n Normal 270000000000 45 21 000000000 Para ayuda pulsar F1 Fil 47 Col 13 AL w Distribuci n Normal x Gr fica ES E Distribucion Nore Datos Distribuci n Hi 210000000000 o 21000000000 14700000
28. own if a Block of 2 variables is created the name of these variables will be Bloque Normales 1 and Bloque Normales 2 respectively 1 3 3 1 2 Dependent Stochastic Variables The process to define a dependent variable starts with the field selection This procedure is the same as the procedure shown for independent variables When the floating menu appears select the option Dependent Dependiente and the Expression Calculator will be accessible The Expression Calculator allows to define a function using any type of mathematical expressions involving the Independent Stochastic Variables to create a dependent variable ruled by this function For example to create a dependent variable ruled by the SIN of the independent variable Young Pressthe SIN button in the Expression Calculator Select an independent variable from the list of variables shown in the Expression Calculator The variable selected appears as part of the mathematical expression Pressthe left parenthesis button to finish the expression The expression generated can be evaluated any time pressing the Evaluate button If the user introduces an expression and it is not evaluated STAC will evaluate it when the user pressesthe OK Aceptar button im Variable Dependiente Distribucion Dependiente Calculadara pem Archivo young 10000000000 D SWINNT SProfiles jant bloque normales 0 5 CAWINAT Profiles jant bloque normales 0 25 CAMINATA Profiles a
29. re below Opciones Escala de Valores Escala del eje de valores 1 Escala autom tica Restablecer Escala Maxima fo Minima fo Divisiones Principales Divisiones Secundarias Aceptar Lancelar As mentioned before each type of graph includes different options Mean Value Evolution The user can only modify the scale of the Y axis and the color of each curve generated The tab Color de serie has the following format Opciones x Escala Eje Y Escala Eje x Color de Serie Color de las Series Series Evoluci n de Fromedios a Evoluci n de Promedios Color Serie Sele Intervalo Conf Superior 95 Intervalo Conf Inferior 95 NENNEN Aceptar Cancelar Histogram Similarto the previoustype of graph Only the scale of the y axis can be modified The tab Colorde serie hasthe following format Escala Eje Y Escala Eje amp Color de Serie Color de las Series Series Histograma Color mmm Aceptar Cancelar Dispersion In this case the user can change the scale of the x and the y axis The tab Colorde serie hasthe following format Opciones Escala Eje Y Escala Eje amp Color de Serie Color de las Series Series Grafica de Dispersi n LAT ici Color Serie Sele Elipse SDev 45 Elipse 5SDiev xz 60 Elipse SDev x3 95 Aceptar Lancelar In orderto change the color of the series is necessary to select the series and then the
30. ri bloque normales 0 25 CAMINATA Profiles ani bloque normales 0 5 CWNN T Profiles jant bloque_normales 0 25 CAMINATA Profiles ani bloque normales 0 25 CAMINATA Profiles ani bloque normales 0 5 CAWINA T Profiles jan ir Nombre Dep Young Hesultada 8550272510194 33 Aceptar Cancelar Evaluar Error Evaluating the expression with the Evaluate button it is possible to obtain an idea of the values that the dependent variable will be able to assume however it does not guarantee that the function defined always gives a real number and in some cases the numbercan be indefinite As an aid the Expression Calculator always shows the number of parentheses that remain open see next figure Distribucion Dependiente Le Generada Calculadora ISIN young 1 3 3 2 Results variables Output variables The process to select the output variables is similar to selecting the input variables The user selects a field or a Block and presses the right button of the mouse A pop up menu appears with the option to create an output variable When accepting a new window appears querying the variable name and optionally a comment Pesososoossosesosesocssocooosscosoocsceoossccsoocsosososesoscseoosocovesese D 2329644E 06 Mombre Desplazamiento E Hn constante en el nodo 103 Lrear variable de an lisis Desplazamiento de una T al aplicar una fuerza ig Variable de Analisis Varia
31. right comerasseen in the next figure i Ejecuci n del Analisis de MonteCarlo Estado de la Ejecuci n del An lisis gt Recogiendo valores variables de salida gt Agregando resultado oak gt Ejecutando tiro 5 250 ok gt generando Fichero Mileatorio0 ak gt Reemplazando valores variables File In gt Ejecuci n simulador gt Progreso amp gt Tiros Fallidos Post proceso Cancelar Ejecuci n 1 5 3 Post Proc ess The user can follow the evolution of the analysis The EXM offers the possibility of activating a Post Process Module PPM pressing the corresponding button i Ejecuci n del Analisis de MonteCarlo Estado de la Ejecucion del Analisis gt Recogiendo valores variables de salida gt Agregando resultado ok gt Fin tiro 4 250 gt Ejecutando tiro 5 250 wido Fichero Nileatorio ak zando valores variables File In n simulador One of the more interesting values presented in the PPM is the average evolution of ok ok any of the vanables defined in the analysis as shown in the figure below Evoluci n de las Medias 5tac Evoluci n de la Media Variables Evoluci n Promedios Intervalo Superior 95 Hombre WW Our J J 2 33313e4011 2 261e 011 2 18886e 011 5 2 11672e 011 gt 04458e 011 1 97245e 011 1 90031e 011 1 52617e 011 1 75603e 011 D Mombre desplazamiento E e
32. rs involved and all I O files have to be generated All this information deterministic analysis is necessary so that laterthe stochastic analysis can be defined with STAC The deterministic I O files are the basis to define the stochastic variables their format and the distribution type for the input variables For the output variables a statistical register is defined and initialized li Stac v1 0 Archivo Elcheros Edicion ventanas Ejecutar Postprocezo Administracion Ayuda Ga El E Explorador de Proy acto Para ayuda pulsar Fl m MDP All the information necessary to define and execute a stochastic analysis is collected in a file called ProjectHle This isa small file in ASCII format that contains among others e the name ofthe Project unique forthe system e the numberof samples e additional information for the solver execution The Project Hle can be visualized with any text editor but it is highly recommended not to change tt The following figure shows the main window of the STAC system There are several optionsto choose from e New Project Nuevo Proyecto to create orto define a new project e Open Project Abnr Proyecto to open a recent or existing project e Delete Project Eliminar Proyecto to delete a project Archive Ficheros Edicion Wentanas Ejecutar Postproceso Administracion Ayuda Nuevo Proyecto Abrir Proyecto nada derat Cerrar Proyecto Eliminar Proyecto Proyecto
33. run an analysis it is only necessary to press the button Comenzar Ejecuci n Begin Execution or in the menu bar submenu Execute the option Ejecutar Execute Any of the options presented is valid L i Stac v1 U Archivo Ficheros Edicion Ventanas Ejecutar Postproceso Administraci n Ayuda cor El a viga cal C WINNT Profile BS Ficheros de Resultados Cluster E viga sal C WINNT Profile E E Vigal Ficheros para el Anali G E Ficheros de Datos near Barrar Ping Editar Ejecuci n Edici n del Fichero por lotes amp amp bl a cd C WINNT Profiles jantonio ESCRIT 1 STAC3U 1 PR DC XWEINNT Profiles jantonio ESCRIT 1 STAC3U 1 PRISC sf Ficheros Generaci n de Variables Aleatorias Incluyalelimine aquellos archivos que considere necesarios para la ejecucion A adir Guitar Muestreo Y Tama o de la Muestra 200 Numero maximo de 5 um Tiros Fallidos Comenzar Ejecucion E Datos E Resultados iw Analisis Er Para ayuda pulsar F1 E Explorador de Proyecto The PDM and the ADM are the modulus in charge of the problem definition Once the execution of the analysis begins the Execution Module EXM isactivated This module isin charge of the analyss management from the random variable generation to the gathering of the output variables b LC Comenzar Ejecucion When the stochastic analysis starts the EXM activates a wi
34. s Recientes Ctrl L E ea Para ayuda pulsar F we 1 3 2 Creating a new Project When the user selects the option New Project a new window is displayed to query the new project s name If the name already exists as a project previously stored in the SIAC s data base an enor wil be generated indicating such an incidence nevertheless the usercan change the name of his project orcancel the creation From this point the system s operation will be described using an example called Vigal i Huevo Proyecto Nombre Prayecto viga E Agregar Archivos Ld A Additionally as an added functionality at this point the user has the possibility of adding to the project the input and output files that define the deterministic process This functionality is presented later in the project definition w Huevo Proyecto Nombre Proyecto Agregar Ficheros vigal o Seleccione Unidad ee Seleccione Fichero y arr strelo hasta E c la lista de Archivos Agregados O bien E haga doble Click er e A EJCA SialonFrs rs_ Ed In this example a new project is created Viga 1 with a single input file Viga cal and a single output file Viga sal It is important to remark that the results files must comespond to those obtained using the deterministic solveroperating with the selected input data files In the case of not following this recommendation unexpected or erroneous results can occur dur
35. ser presses the mouse over a point in the curves generated the PPM presents information about the run numberand the mean value atthat point ON AAA AAA Plo5sex A A E euin 2 303638 007 a 2 194718 007 Tio 71 2 O858e 007 fal deor 2 O5be 007 Moments The PPM also calculate and showsthe following statistic al moments e Mean Value 5 J e Variance iJ EP Var Xy 3000 Xn D X X 2 N 153 e Standard Deviation o X1 Xp Mar X4 Xp 3 e Deviation of the mean value N ADEV Xo Xn y 2710 X 4 J e Kurtosis Coefficient 44 14 xj x 5 Kurt X X 4 uL EN j l e Skew coefficient Nix x T 6 Skew Xq yo Xn 2 Y N i o Momentos Datos de la Grafica Info Momentos Estadisticos N mero de Tiros 95 Media 2 106e 01 1 Var 4 2741e 020 Max Y 2 4484011 SDev 2 0674e 010 Skew 2 0412e 001 Min Y 1 560e 01 1 ADev 1 599881 Cure 5 6356e 001 Histogram This option is selected choosing the Histograma radio button In the same window the usercan setthe number of classesto be visualized Evoluci n de la Media Histograma N mero de Clases E gt By default the PPM assigns JN classes where N isthe number of runs The user can change the number of classes introducing the value into the comesponding text box The amowscan also be used The maximum number of classes is 60 The PPM calculates the interval width
36. type of error occurred during the dependent variable actualization division by zero etc the changesdefined in the dependent variable will not take effect 1 3 4 2 Deleting a Variable If the Delete option was selected STAC will eliminate the variable selected If an independent variable is eliminated it can happen that this variable forms part of one or several dependent variables Should this be the case a message indicates it and STAC wil delete the independent variable and all the associated dependent variables Eliminar young dep young Para que la eliminaci n se lleve a cabo se han de eliminar las dependientes asociadas Desea eliminar young v las dependientes asociadas 3 La variable que se va a ELIMINAR young est asociada a una o varias variables dependientes L4 1 4 Definition of the execution process In orderto complete the definition of a project it is necessary to declare The sample size orthe number of runs The maximum number of wrong runs This is the number of times that solver execution ends without producing output variables The necessary commands forthe solver execution The additional files needed by the solver The Working Directory where all the generated files will be saved temporarily during the analysis process The Service Directory where the data corresponding to the results obtained during the different executions will be saved Later this data will be processe
37. ura Prnt As it indicates this option prints an associated graph and all its data like the statistical moments and other additional information
38. vates The Post Process Module PPM this allows to visualize the ESSE c statistic al moments and the dispersion of the collected data Momentos k Dispersion The options in the Execution menu to report the log SEE B Postproceso Administr files forthe process execution and wrong runs The log files report in detail the whole processof the analysis execution and erroneous runs if there Ver Log Ejecuci n were any Ver Tiros Err neoz Ejecutar Opciones The Delete temporary files button described in section 1 4 3 This option is activated independently of the number of completed runs 1 5 Result Visualization 1 5 1 Introduction The results obtained are presented using a simple statistical post process The module involved is called Post Process Module PPM The PPM is activated when the analysis is finished or when a previous project is opened In any case to show the PPM window the userhasto select the option post process from the main menu Momentos k Dispersi n Initially the most common statistical moments are evaluated on all the variables defined in the project Also a small regression analysis between two of the used variables can be made Since SIAC was not designed to perform a full statistical descriptive analysis it is possible to export the generated data so that it can be processed by applications specially designed for such ta sks The PPM can operate in tw
39. will appear Evoluci n de las Medias 5tac Variables Evoluci n de la Media Nombre Visualizaci n Momentos V Mostrar Escala x Datos de la Gr fica AAA v Mostrar Escala Y s Info Momentos Estadisticos v Yer Int Conf Sup Media Var I ver Int Conf Inf dorm SDev Skew Min Y ADev Curt C Histograma Opciones Salir The functionalities offered in this window are the following Mean evolution This option shows a simple representation of the variable s mean value evolution as the number of runs is increased Additionally it Shows two curves that define the boundary of the confidence interval with a probability of 95 An example is shown in the next figure Later sections will explain in detail how the user can change the graphical elements like the scales line width or the colors of the different curves in the graph Evoluci n de la Media Evoluci n Promedios Intervalo Superior 95 Intervalo Inferior 3595 2 47 4le oll 247741811 d ee cal mm el 2 4052 e 011 2 e4 011 2333138011 zd33138 011 j ble 011 j ble 011 2 18885811 2 18885811 2 116 2e 011 1116284011 4M T z U44588 U011 1 97245e 011 04450e 01 1 1 92458 011 1 200 518401 1 1 200 51e 011 1 0 lt 01e4 011 1 8281 e4 011 1 ob e 01 1 1 fo603e 01 1 q T my rr 17 in a in id N mero de Tiros If the u
Download Pdf Manuals
Related Search