Download - 150 - Understanding Optimization Design Studies

Transcript
  • 8/13/2019 150 - Understanding Optimization Design Studies

    1/23

    OptimizationStudiesUnderstandingOptimizationDesignStudiesLecture

    UnderstandingOptimizationDesignStudies.mp3

    Understanding Optimization Design Studies

    Optimization Studies adjust design variables to best achieve a

    specified goal while satisfying specified design limits.

    Components:

    Goal Design Limits Design Variables

    Optimization versus Feasibility

    Algorithm

    1. Initial Values2. Local Sensitivities3. Base Analysis4. Repeat until convergence or Max Iterations

  • 8/13/2019 150 - Understanding Optimization Design Studies

    2/23

    Optimization Model

    Maximum Von Mises Stress during an Optimization Design StudyLectureNotes

    The Optimization Design Study

    An Optimization Design Study adjusts one or more design variables to best achieve aspecified goal or to test feasibility of a design, while respecting specified limits.Mechanica adjusts the model's variables in a series of iterations through which it triesto move closer to the goal while respecting specified limits.

    The following components are required in order to define an Optimization Study:

    Goal: During the course of an Optimization, you specify a default or user-defined Mechanica measure that you wish to maximize or minimize.

    Design Limits: Design Limits can be specified for any default or user-definedmeasures. During the Optimization, Mechanica attempts to find a solution thatdoes not violate any of the design limits while trying to improve the goal. Youcan specify whether a design limit should be greater than, less than, or equalto a specified value.

    Design Variables: Similar to a Global Sensitivity Design Study, you select thedesign variables that you want Mechanica to adjust during the study. Therange and initial values for these design variables can be specified as

    necessary.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    3/23

    A good example of an Optimization problem might be trying to minimize the massand material used for a loaded structural member (the Goal) by varying keydimensions such as thicknesses (the Design Variables) while keeping the stress below

    acceptable levels to avoid failure (the Design Limits).

    Note the following key aspects of the Optimization Design Study:

    The Goal and Design Limits are optional, but you must have at least one Goalor one Limit.

    If you have not specified a Goal, Mechanica simply searches for the firstfeasible design that satisfies the defined Design Limits.

    When defining an Optimization Design Study, you can change the type

    to Feasibility and the goal section of the dialog box becomes

    unavailable. As such, it becomes an Optimization Design Study without

    a goal. As the name suggests, a Feasibility Study determines whether

    the design variables can be changed to values such that the model is

    feasiblea state defined where none of the design limits are

    compromised.

    The Optimization Algorithm.

    The following steps summarize the procedure deployed by Mechanica when runningand Optimization Design Study:

    1. Mechanica starts the optimization by using the initial values specified forDesign Variables.

    2. For the initial Design Variable's values, Mechanica runs Local SensitivityStudies in the background and tests the slopes of every Design Variable.

    Depending on the outcome of these local sensitivity studies, some of thevariables may or may not contribute dramatically satisfying the limits. As such,some variables values will stay the same during the study (flatter slope localstudy graphs) and some will adjust (steeper slope local study graphs).

    3. The base analysis (or analyses) is run at the current Design Variables. Thenecessary measures (used for Design Limits and Goal) are extracted,

    evaluated, and investigated against the imposed conditions.

    Design Limits have precedence over the Optimization Goal.

    4. If the Goal or the maximum number of iterations has not been reached, theprocess repeats starting with Step 1 above.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    4/23

    Best Practices

    Optimization typically requires more iterations as the number of DesignVariables increases. Therefore, use as few Design Variables as possible. LocalSensitivity Studies are helpful in deciding which design variables are important.

    The smaller the design space being considered, the shorter the optimizationrun time. Therefore, use the results from the Global Sensitivity Studies tonarrow the range of each design variable being considered and thus limit theoverall design space.

    The Optimization takes less time the closer the baseline model is to theoptimum. You might find it useful to review results from Sensitivity Studies toimprove the candidate design prior to optimization.

    It if often helpful to start an Optimization from a feasible design state. If youhave trouble getting an solution from an Optimization problem, considerrunning a Feasibility Study first and using the resulting design variable valuesas the starting point for an Optimization.

    UnderstandingOptimizationDesignStudiesDemonstrationUnderstandingOptimizationDesignStudies_demo.mp4UnderstandingOptimizationDesignStudiesProcedure

    Procedure: Understanding Optimization Design Studies

    ScenarioDefine an Optimization Design Study.

    Optimization support_bracket.prt

    Task 1.Open the Mechanica Application and define an Optimization Design Study.

    In this example, you are defining an Optimization Study that will help you

    identify the design changes required to find the lightest weight bracket able

    to sustain specific von Mises stress values. Notice that some of thesimulation features (material, load, constraints, surface regions, static

    analysis) are already defined in the model. Investigate them and identify the

    key aspects of the constraining/loading scheme.

    1. Click Applications > Mechanica.

    2. Click Mechanica Analyses/Studies from the main toolbar.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    5/23

    3. Click File > New Optimization Design Study....

    4. In the Name field, type SP_BRCK_OPTM.

    Because you have only one analysis defined in the model, Mechanica will

    automatically select it as the base analysis. The same is true for the single

    load set defined in the model.

    5. In the Goal area of the dialog box, verify that the drop-down menu is set to

    Minimizeand that the measure is set to total_massas shown in the dialog box.

    These settings are the Mechanica defaults and just happen to match our

    objectives in this procedure.

    6. Click Add Row in the Design Limits area of the dialog box.

    7. Select max_stress_vmfrom the Measures dialog box and click OK.

    8. Verify that the operator is set to

  • 8/13/2019 150 - Understanding Optimization Design Studies

    6/23

    10. Click Select Dimension in the Optimization Study Definition dialog box.

    11. Select the Round 1feature from the model tree and select the 1.5dimensionas shown in the figure.

    12. Click Select Dimension in the Optimization Study Definition dialog box.

    13. Select the CROSS_SECTIONfeature from the model tree and select the 10dimension as shown in the figure.

    14. Type the following variable names and values in the corresponding cells in the

    variables section of the dialog box as shown in the figure below.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    7/23

    15. Click Options...to open the Design Study Options dialog box.

    16. Verify that the Optimization Algorithm drop-down menu is set to Automatic.

    Based on the computation needs, the Automatic setting enables the software

    to change between the SDP and GDP algorithms as necessary.

    17. Type 5.0for the Optimization Convergence.

    Note that the Optimization Convergence value you set above is for the

    Optimization algorithm and not for the multi-pass adaptive-style polynomial

    convergence.

    18. Verify that the Maximum Iterations is set to 20.

    19. Select the Repeat P-Loop Convergenceand Remesh after each shape

    updatecheck boxes.

    20. The dialog box should appear as shown in the figure. Click Closeto close the

    Design Study Options dialog box.

    21. Click OKto complete the Optimization Study Definition and close the dialogbox.

    1.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    8/23

    Task 2.[Optional] Run the optimization.

    You can actually run the optimization if you have time. It should take about

    510 minutes to complete. If you do not have the time to actually run the

    Optimization, you can review the results in the Completedirectory, or skip

    to Step 4 and end the exercise.

    1. Verify that SP_BRCK_OPTMis selected in the Analyses and Design Studies

    dialog box and click Start Run > Yesto start the design study.

    2. Click Display Study Status once the analysis is started.

    3. When the analysis is complete, review the Run Status window. You should beable to make the following observations:

    Four iterations of the optimization algorithm were required.

    Four base analyses were required. Nine perturbation analyses were required. In order to reduce stress below the design limit, the FILLET dimension grew to 191

    and the BRCKT_HEIGHT dimension shrank to 8.

    The final von Mises stress finished at 235.79 N/mm2, less than the MaximumDesign Limit of 240 N/mm2(MPa).

    4. Close any open dialog boxes.

    5. Return to the Standard Pro/ENGINEER mode by clicking Applications > Standard.

    6. Click Save from the main toolbar and click OKto save the model.

    7. Click File > Erase > Current > Yesto erase the model from memory.

    8. If necessary, click Closefrom the Summary window and click Closeto close the

    Diagnostics window.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    9/23

    This completes the procedure.

    UnderstandingOptimizationDesignStudiesExercise

    Exercise: Running a Feasibility Study

    Objectives

    After successfully completing this exercise, you will be able to:

    Understand the difference between optimization and feasibility studies. Create and run a feasibility study.

    ScenarioIn an optimization study, the user has a goal of maximizing or minimizing a quantity

    specified as the optimization goal. In a feasibility study, there is no desire to achieve a

    specific goal; the user only wants to know if a specific design is feasible.

    Tuning forks produce a sound by vibrating at specific resonant frequencies. If the resonant

    frequency is within the audible range of 20 Hz to 20,000 Hz, then the frequency can beclassified as a musical note. For example, the note, middle-C, a vibrates at 256 Hz,

    whereas the note, E, vibrates at a frequency of 320 Hz. Piano tuners and guitar playersuse tuning forks to tune their instruments to certain musical notes. In this exercise, you

    tune a fork to the musical note "G," which vibrates at 384 Hz.

    Feasibility fork.prt

    Task 1.Open the Mechanica application and define a new modal analysis.

    1. Click Applications > Mechanica.

    Note that the Mechanica analysis model has already been started. The model

    already has a material definition and a constraint set.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    10/23

    2. Click Mechanica Analyses/Studies from the main toolbar.

    3. Click File > New Modal....

    4. In the Name field, type Fork_Mode.

    5. Type 1in the Number of Modes field.

    6. The dialog box should appear as shown in the figure. Click OKto complete theModal Analysis Definition and close the dialog box.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    11/23

    7. Verify that Fork_Modeis selected in the Analyses and Design Studies dialog box

    and click Start Run > Yesto start the design study.

    8. Click Display Study Status once the analysis is started.

    9. When the analysis is complete, review the Run Status window. You should be ableto make the following observations:

    The modal frequency is reported as 406 Hz, which is higher than the desired 384Hz.

    10. When you are through inspecting the Run Status window, click Closeto close theRun Status window and Closeto close the Diagnostics window.

    Task 2.Create and run a Feasibility Study.

    1. Click File > New Optimization Design Study....

    2. In the Name field, type fork_opt.

    3. Select Feasibilityfrom the Type drop-down menu.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    12/23

    Note that the Goal area of the dialog box is removed when you set the type

    to Feasibility.

    4. Click Add Row in the Design Limits area of the dialog box.

    5. Select modal_frequencyfrom the Measures dialog box as shown in the figure andclick OK.

    6. Click on the

  • 8/13/2019 150 - Understanding Optimization Design Studies

    13/23

    12. Click Options...to open the Design Study Options dialog box.

    Note that the optimization convergence is set to 1%. This means that the

    study will find a feasible design that will report a modal frequency of

    3841%.

    13. The dialog box should appear as shown in the figure. Click Closeto close the

    Design Study Options dialog box.

    14. The dialog box should appear as shown in the figure. Click OKto complete theOptimization Study Definition and close the dialog box.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    14/23

    15. Verify that fork_optis selected in the Analyses and Design Studies dialog box and

    click Start Run > Yesto start the design study.

    16. Click Display Study Status once the analysis is started.

    The Design Study should complete in less than five minutes.

    Task 3.Review the Optimization/Feasibility Design Study results.

    1. When the Optimization/Feasibility Design Study finishes, examine the results in the

    Run Status window. Scroll up in the Run Status window until you find the initialDesign Status.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    15/23

    Note that the modal_frequency design limit was violated for the Initial

    Design Status, but after the Design Variable was changed to 94.0221, the

    modal_frequency design limit was satisfied. Recall that you set the

    convergence percentage to 1% the resulting 384.73 is within 1% of the

    384 design value you specified.

    2. When you are through inspecting the Run Status window, click Closeto close the

    Run Status window and Closeto close the Diagnostics window.

    3. Verify that fork_optis selected in the Analyses and Design Studies dialog box andselect Info > Optimize History.

    4. Press ENTERwhen prompted if you want to review the next shape.

    5. Press ENTERagain when prompted if you want to review the next shape.

    6. Press ENTERwhen prompted if you want to leave the model at the optimized

    shape.

    The feasibility study found the first solution to satisfy the limits and ended

    the analysis. Had the analysis been for optimization, then the analysis would

    have continued after satisfying the design limit and attempted to minimizeor maximize the specified goal. The feasibility study takes considerably less

    time and can tell the user quickly if a desired scenario is possible or not.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    16/23

    7. Return to the Standard Pro/ENGINEER mode by clicking Applications > Standard.

    8. Click Save from the main toolbar and click OKto save the model.

    9. Click File > Erase > Current > Yesto erase the model from memory.

    10. If necessary, click Closefrom the Summary window and click Closeto close theDiagnostics window.

    This completes the exercise.

    UnderstandingOptimizationDesignStudiesExerciseExercise: Optimization Design Studies

    Objectives

    After successfully completing this exercise, you will be able to:

    Use the info from Local and Global Sensitivity Studies to set up an OptimizationDesign Study.

    Use Mechanica in order to find the optimum design. Interpret the results from an Optimization Study and visually inspect the optimum

    solution.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    17/23

    ScenarioInitially, all the design variables in the model were examined and investigated using Local

    Sensitivity Studies in a previous exercise. This enabled you to select only the variablesgreatly affecting the Mechanica measures of interest (such as maximum displacement and

    stress to name a few).

    After this preliminary step, you examined the range settings for the remaining variablesand their influence on the Mechanica measures using the Global Sensitivity Study in

    another previous exercise.

    Once all the initial input is established, you can start evaluating the changes in the model

    such that an optimum solution is found. This solution has to satisfy certain designrequirements (also known as design limits) and be within a user-defined convergence

    value.

    You will apply this workflow and investigate the optimum solution for the bracket of the

    car jack. Our goal is to find the lightest weight bracket able to sustain the external loads

    and develop stresses less than 240 MPa.

    JackOpt up_link.prt

    Task 1.Open Mechanica mode and define an Optimization Study.

    1. Click Applications > Mechanica.

    2. Click Mechanica Analyses/Studies from the main toolbar.

    3. Click File > New Optimization Design Study....

    4. In the Name field, type LINK_OPTIM.

    The Mechanica default is to set the Goal to Minimizethe total_mass

    measure. Since this happens to be your objective for this Optimization, you

    do not need to make any changes to the goal area of the dialog box.

    5. Click Add Row in the Design Limits area of the dialog box.

    6. Select max_stress_vmfrom the Measures dialog box and click OK.

    7. Verify that the operator is set to

  • 8/13/2019 150 - Understanding Optimization Design Studies

    18/23

    9. Click Select Dimension from the Variables area of the dialog box.

    10. Select Cut id 91from the model tree.

    11. Select the 12.7dimension from the display area as shown in the figure.

    12. Click Select Dimension from the Variables area of the dialog box.

    13. Select Shell id 528from the model tree.

    14. Select the resulting 3.81dimension from the display area as shown in the figure.

    15. Adjust the values in the Variables area of the dialog box to match the figure.

    16. Click Options...to open the Design Study Options dialog box.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    19/23

    17. Type 5.0as the Optimization Convergence.

    18. Verify that Maximum Iterations is set to 20.

    19. Select the Repeat P-Loop Convergenceand Remesh after each shapeupdatecheck boxes.

    20. The dialog box should appear as shown in the figure. Click Closeto close theDesign Study Options dialog box.

    21. Click OKto complete the Optimization Study Definition and close the dialog box.

    Task 2.Run the Optimization Study and create Results windows for the OptimizationDesign Study.

    1. Verify that LINK_OPTIMis selected in the Analyses and Design Studies dialog box

    and click Start Run > Yesto start the design study.

    2. Click Display Study Status once the analysis is started.

    The Optimization Design Study should complete in approximately 23

    minutes. If you do not have time to wait for the Design Study to complete,

    you can still complete the rest of the exercise using the results in the

    Completedirectory.

    3. When the analysis is complete, review the Run Status window. You should be able

    to make the following observations:

    Four iterations of the optimization algorithm were required. Four base analyses were required. 9 perturbation analyses were required. In the course of the Optimization Design Study, the radial_offset dimension grew to

    18 mm and the THICK dimension shrank to 3.34 mm.

    The final von Mises stress finished at the Maximum Design Limit, 240 N/mm2(MPa).

  • 8/13/2019 150 - Understanding Optimization Design Studies

    20/23

    The mass of the part was reduced from .172 kg to .167 kg.4. When you have completed inspecting the Run Status window, click Closeto close

    the Run Status window and Closeto close the Diagnostics window.

    5. Verify that LINK_OPTIMis selected in the Analyses and Design Studies dialog boxand click Results to start Results mode.

    6. Type VM_OPTIMin the Name field and Max Stress Von Mises vs. OptimizationPassin the Title field.

    7. From the Display type drop-down menu, select Graph.

    8. From the Graph Ordinate (Vertical Axis) area, select Measurefrom the topmost

    drop-down menu.

    9. Click Define Measure to open the Measures dialog box.

    10. Select max_stress_vmfrom the list of Measures, and click OKto close theMeasures dialog box.

    11. The dialog box should appear as shown in the figure. Click OK and Showto show

    the resulting graph.

    12. Click Copy from the main toolbar in the Results window.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    21/23

    13. Type MASS_OPTIMin the Name field and Total Mass vs. Optimization Passin

    the Title field.

    14. Click Define Measure to open the Measures dialog box.

    15. Select total_massfrom the list of Measures, and click OKto close the Measuresdialog box.

    16. The dialog box should appear as shown in the figure. Click OK and Showto showthe resulting graph.

    17. Click Copy from the main toolbar in the Results Window.

    18. Type VM_FRINGEin the Name field and Max Stress Von Misesin the Title field.

    19. From the Display type drop down menu, select Fringe.

    20. Verify that the drop-down menus on the Quantity tab are set to Stress, the units

    field is set to MPa, and the Component field is set to von Mises.

    21. The dialog box should appear as shown in the figure. Click OK and Showto showthe resulting graph.

  • 8/13/2019 150 - Understanding Optimization Design Studies

    22/23

  • 8/13/2019 150 - Understanding Optimization Design Studies

    23/23

    2. When you are through examining the results, click File > Exit Results > Noto exit

    the Result window without saving any results.

    3. If necessary, return to the Standard Pro/ENGINEER mode by clicking Applications> Standard.

    4. Click Save from the main toolbar and click OKto save the model.

    5. Click File > Erase > Current > Yesto erase the model from memory.

    6. If necessary, click Closefrom the Summary window and click Closeto close theDiagnostics window.

    This completes the exercise.