The Engineering Lab's mission is to enable engineers worldwide with the web applications and training necessary to use optimization and machine learning methods to improve the performance of structural and mechanical systems.

What does The Engineering Lab do?

  • Web Apps for Optimization and Machine Learning - The Engineering Lab develops web applications that enable engineers to optimize finite element models with either gradient-based optimizers or machine learning technology. The primary finite element analysis solver supported is MSC Nastran. These web apps are available for purchase.
  • Training for Optimization and Machine Learning - The Engineering Lab offers training lectures and tutorials detailing the necessary steps to utilize MSC Nastran's optimizer or machine learning technology. With training provided by The Engineering Lab, engineers will be better prepared to use the latest computational methods to improve the performance of structural and mechanical systems. Inquire today about training pricing.
  • Custom Web Apps for Engineers - The Engineering Lab builds custom web applications for engineers. These custom web apps are built using the latest web development methods and are rapidly deployed to meet a broad range of requirements. Examples include custom user interfaces to FE solvers, desktop automation, simulation results processing and more. Contact The Engineering Lab today to inquire about your custom web application.

Inquiries may be forwarded to christian@ the-engineering-lab.com.

The SOL 200 Web App will convert existing SOL 1xx bdf files to SOL 200, the design sensitivity and optimization solution sequence.

The SOL 200 Web App automatically generates, updates and validates the necessary entries. Below are some of the entries supported:

  • Design Variables and Regions (DESVAR, DVXREL1, DVXREL2, DEQATN, DLINK, DDVAL, TOMVAR, TOPVAR)
  • Design Objective (DRESP1, DRESP2, DEQATN, DESOBJ)
  • Design Constraints (DRESP1, DCONSTR, DRESP2, DCONSTR, DEQATN, DCONADD, DESSUB)

The SOL 200 Web App also features capabilities to perform Machine Learning.

A datasheet with a full list of capabilities is available at this link.

A free trial of the SOL 200 Web App is available at the link below.

Free Trial - SOL 200 Web App

1. Minimize Errors – All optimization applications are prone to errors during optimization. An example of such an error is USER FATAL MESSAGE. Such errors are exacerbated if BDF files are hand edited. The SOL 200 Web App takes a unique approach to minimizing errors. With each keystroke, mouse click and configuration, the web app actively validates entries and immediately notifies the user of invalid entries. This instant "Status" feedback enables users to fix issues in real time before performing an optimization.

Watch the validations in action by referring to any tutorial video in the Tutorials Section.

When considering a new optimization tool, make sure to ask these questions:

  • What is your tool’s approach to mitigate errors?
  • I have this very difficult optimization. May I see a demonstration of this using your tool?
  • May I have an evaluation copy of your software to test this difficult problem?

2. View Entries Immediately – With traditional tools, the bulk data entries needed for optimization are hidden and are only visible after export. The SOL 200 Web App displays and updates the bulk data entries as users prepare the optimization.

View entries generated in real time by watching any tutorial video in the Tutorials Section.

When considering a new optimization tool, make sure to ask these questions:

  • Does your solution generate bulk data entries for Nastran SOL 200?
  • Does your solution use a separate optimizer or the Nastran optimizer?

3. Use Advanced Capabilities – Many optimization tools are characterized by highly complex interfaces, often making advanced optimization capabilities impractical. The SOL 200 Web App’s interface is streamlined by using the latest web technology. This streamlining makes the most advanced optimization capabilities practical, such as multi-model optimization and machine learning.

Click here to view a datasheet with a full list of the SOL 200 Web App's capabilities.

When considering a new optimization tool, make sure to ask these questions:

  • I have this very difficult optimization. May I see a demonstration of this using your tool?
  • May I have an evaluation copy of your software to test this difficult problem?

4. Access Through a Web Browser – Traditional applications require an installation on each desktop and can be time consuming to deploy company wide. The SOL 200 Web App requires only a single installation on either a desktop or server, and any other desktop on the local network may access the web app via a web browser.

Today request a Free Trial of the SOL 200 Web App and begin using your web browser to configure optimizations for Nastran SOL 200. Visit the Free Trial page.

When considering a new optimization tool, make sure to ask these questions:

  • Does this interface support the MSC Nastran optimizer in SOL 200 or a separate optimizer?
  • May I see what the installation process is like?
  • Do I have to install this tool on every desktop?

5. Leverage Built-in Training – Some optimization tools require separate training courses to begin performing optimizations. The SOL 200 Web App comes with over 50 tutorials detailing simple to advanced optimization scenarios, including size optimization, global optimization, multi-model optimization, machine learning and more.

A complete set of available tutorials is available in the Tutorials Section.

When considering a new optimization tool, make sure to ask these questions:

  • Do I need a training course before I use your software?
  • How many training tutorials are included in the software? May I see a copy of these tutorials?

6. Spend Less – Optimization solutions, which often feature both an optimizer and interface, sometimes cost up to $100,000 USD. The SOL 200 Web App is priced at $350 USD per month. A majority of Nastran users already have licensing to use Nastran’s optimizer. The next step is to select an interface, such as the SOL 200 Web App.

For a Free Trial, Student or Professional version of the web app, visit the Free Trial page.

When considering a new optimization tool, make sure to ask these questions:

  • I already have a license for MSC Nastran’s optimizers. Why do I need a second optimizer?

Over 25 step-by-step tutorials are available regarding fundamentals of optimization and how to properly use the SOL 200 Web App. Click on any of the following links to jump to the section.

Please contact christian@ the-engineering-lab.com if you are interested in further Nastran SOL 200 guidance or training.

Basic Optimization Tutorials

Title and Description YouTube Tutorial
Unconstrained Optimization with MSC Nastran SOL 200

Part of Calculus involves finding maximums or minimums of functions. The process of finding maximums or minimums is the essence of optimization.

In this video, MSC Nastran Optimization is used to find the optimum point or minimum of a two-variable function, f(y1, y2) = y1^2 + y2^2.
Link
Constrained Optimization with MSC Nastran SOL 200

This video demonstrates the use of MSC Nastran Optimization to find the minimum of f(y1, y2) subject to a constraint g(y1, y2).
Link
Side Constraints on Design Variables - MSC Nastran Optimization

This video demonstrates the use of MSC Nastran Optimization to find the minimum of f(y1, y2) subject to limits on the design variables y1 and y2.
Link
Best Compromise Infeasible Design - MSC Nastran Optimization

In constrained optimization, an optimum solution may not exist that satisfies all the constraints. This video demonstrates such a scenario and walks through a fix.
Link
What is Global Optimization? MSC Nastran SOL 200 / Optimization Tutorial 

This video discusses the meaning of Global Optimization and walks you through the process of setting up a Global Optimization with MSC Nastran SOL 200.
Link
What is size optimization? What is shape, topology, topography and topometry optimization?

In this short video, the following MSC Nastran optimization types are described.
  • Size Optimization
  • Shape Optimization
  • Topology Optimization
  • Topography Optimization
  • Topometry Optimization
Link

Size Optimization Tutorials

Title and Description Lecture Notes PDF Tutorial YouTube Tutorial
Structural Optimization of a 3 Bar Truss - MSC Nastran Optimization

A truss structure is optimized with MSC Nastran. The design variables are the cross-sectional areas of the rod elements. The objective is to minimize the weight of the structure while ensuring the stress and displacements are within specified constraints.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Sensitivity Analysis of a 3 Bar Truss - MSC Nastran Optimization

A structural optimization was previously performed on a 3 bar truss. In this tutorial, the process to perform a sensitivity analysis is detailed.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link
Automated Structural Optimization of a Stiffened Plate with MSC Nastran SOL 200/Design Optimization

This example demonstrates the use of MSC Nastran to optimize the thickness of the plate and the thickness of a beam section to minimize weight. Constraints are imposed on the stresses in the shell and beam elements. Additional constraints are imposed on deflections.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Vibration of a Cantilevered Beam (Turner's Problem), MSC Nastran Optimization

This example demonstrates the use of MSC Nastran to optimize the rod areas and shell thicknesses such that the structure's weight is minimized and the first natural frequency is above 20 Hz.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Dynamic Response Optimization with MSC Nastran Optimization

This example is from the MSC Nastran Design Sensitivity and Optimization User's Guide.

"This example demonstrates structural optimization when the structural loads are frequency dependent. The system considered is a flat rectangular plate clamped on three edges and free along the fourth, as shown in Figure 8-21 . The problem investigates minimization of the mean square response of the transverse displacement at the midpoint of the free edge, while constraining the volume of the structure (and hence, weight) to be equal to that of the initial design. A pressure loading with an amplitude of 1.0 lbf ⁄ in^2 is applied across a frequency range of 20.0 to 200.0 Hz. A small amount of frequency- dependent modal damping has also been included."

MSC Nastran 2016 Design Sensitivity and Optimization User's Guide. Chapter 8: Example Problems. Dynamic Response Optimization.
Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Model Matching, Frequency Response Analysis

A frequency response analysis has been performed, but the results do not match experimental results.

This tutorial discusses the model matching procedure in order to correlate Finite Element Analysis and test results.

Starting BDF Files: Link
Solution BDF Files: Link
Link
Using MSC Nastran Optimization for Model Matching / System Identification

In this example, the cross section of a rod is designed such that the analysis modes match experimentally measured data. MSC Nastran Optimization is used to minimize the root sum of squares for Mode 1.

This example is an adaptation of the example found in the UAI/Nastran User's Guide for Version 20.1 - 252.6.6 System Identification. The following is an excerpt from the guide describing this example. Keep in mind this video is an adaptation and will not match all the values in the following description:

"An important area of research is the tuning of finite element models to experimental test results. This is often called system identification. This example problem illustrates how optimization may be used to address these requirements. It features:

  • Normal modes optimization
  • Constraints on RMS error in mode shapes
  • Frequency constraints
  • Using an analytical response as the objective

Example Problem 25-6

Consider the model shown in Figure 25-13. It is a simple cantilever beam which is composed of 10BAR elements with circular cross sections having a constant diameter of 4.0 in. The cross-sectional area of the three BAR elements at the root of the beam may be varied. You wish to design this area such that it matches known test results.

The finite element model, which is found in file MDO6, is shown in Figure 25-13. Table 25-19 presents the test results and a definition of the design constraints for the problem. These data reflect that two natural modes, mode 1 (first bending) and mode 3 (first extensional) have been measured experimentally at three locations which correspond to GRID points 3, 6 and 9 in the finite element model. The frequency of the first mode, 61.912, has also been measured.

The design model is simple having a single design variable which represents the root cross-sectional area."

UAI/Nastran User's Guide for Version 20.1 - 252.6.6 System Identification
Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Automated Optimization of a Composite Laminate with MSC Nastran Optimization (SOL 200)

This example details the use of MSC Nastran Design Optimization (SOL 200) to optimize the weight of a tube composed of a composite laminate.

The ply thicknesses and orientations are allowed to vary during the optimization process. The orientation angles are limited to 5 degree increments. Constraints on the failure indices are applied.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Optimizing for Buckling - Twenty-Five Bar Truss with MSC Nastran Optimization

This example is from the MSC Nastran Design Sensitivity and Optimization User's Guide.

"This problem, often seen in the early design optimization literature, calls for a minimum weight structure subject to member stress, Euler buckling, and joint displacement constraints. The structure is shown in Figure 8-25 . The formulation of the buckling constraints is a good example of constructing normalized constraints based on user-defined structural responses."

MSC Nastran 2016 Design Sensitivity and Optimization User's Guide. Chapter 8: Example Problems. Twenty-Five Bar Truss, Superelement and Discrete Variable Optimization


Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Acoustic Optimization, Beta Method

A fluid is enclosed in a structural box and subjected to an acoustic source. The goal is to minimize the peak acoustic pressure while letting the structural thicknesses vary and preventing the weight from significantly changing.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Acoustic Optimization, Nastran BETA Function

This tutorial is a repeat of the previous Acoustic Optimization example, but highlights an alternative method to setting up the optimization for Nastran SOL 200. The BETA method is used and reduces the work that was previously required.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link
Optimization for Multiple Load Cases or SUBCASEs

The web app makes simple configuring design constraints for dozens or hundreds of load cases. This tutorial guides you through the process.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link
Buckling Optimization of a Cantilever Beam

This example demonstrates the procedure to configure Nastran SOL 200 for buckling optimization. This example also covers how to optimize for multiple buckling scenarios.

Starting BDF Files: Link
Solution BDF Files: Link
Link
Global Optimization

This example demonstrates the procedure of performing a Global Optimization with MSC Nastran SOL 200.

Often, optimization problems have multiple local minimums, or maximums, when starting from different initial design variables. To find the global optimum, multiple local optimizations must be performed, then the best of the local optimizations is taken to be global optimum. This process can be performed with the Global Optimization capability available in MSC Nastran SOL 200.


Starting BDF Files: Link
Solution BDF Files: Link
Link Link
Parameter Study

This tutorial details the use of MSC Nastran SOL 200 to perform a "parameter study."

What is a parameter study?

A common engineering technique is to try different structural configurations, for example, changing structural dimensions, and review the impact on structural responses such as displacements and stresses. Dozens, possibly hundreds of structural configurations would ideally be evaluated. This is termed "parameter study." MSC Nastran SOL 200 includes a capability to automatically generate multiple structural configurations and perform static or dynamic analyses. The outcome are results from multiple structural configurations that can be compared.



Starting BDF Files: Link
Solution BDF Files: Link
Link Link
Multi Model Optimization

Multi Model Optimization (MMO) is the process of optimizing multiple design models concurrently. Design variables across multiple models can be linked and simultaneously optimized. A merged or combined objective can optimize the objective of each design model. The design constraints of each design model are also included in a multi model optimization.

This tutorial details the procedure to configure a multi model optimization.

Starting BDF Files: Link
Solution BDF Files: Link
Link

Topology, Topometry and Topography Optimization Tutorials

Title and Description Lecture Notes PDF Tutorial YouTube Tutorial
Topology Optimization - Minimization of mass while satisfying stress constraints MSC Nastran Topology Optimization - Minimizing mass with stress and displacement constraints

A solid block of material composed of 3D or Hexahedral elements is subjected to two load cases. Topology Optimization is used to minimize the mass of the structure, while satisfying both stress and displacement design constraints.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Topology Optimization - Manufacturing Constraints MSC Nastran Topology Optimization Manufacturing Constraints

A cantilever beam is composed of 3D or Hexahedral elements and a load is applied at the free end. Topology Optimization is used to identify regions of material to remove. This example discusses options to produce a symmetric design and a design that can be manufactured via casting.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Topology Optimization - Mirror symmetry constraints MSC Nastran Topology Optimization Mirror Symmetry Constraints

A plate composed of 2D finite elements, is simply supported and has a load applied at the midpoint. Topology Optimization is used to identify material to remove. This example focuses on satisfying weight, stiffness and symmetry targets.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
MSC Nastran Topology Optimization - Multidiscipline - Static Loading and Natural Frequency

A simply supported plate is composed of 2D finite elements and a load is applied at the midspan. The MSC Nastran Topology Optimization capability is used to determine which portions of the plate should be kept while satisfying weight, stiffness and first natural frequency constraints. This example also showcases the ability to optimize for multiple analysis types, e.g. static and normal modes analysis.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Topometry optimization example MSC Nastran Topometry Optimization of a Cantilever Plate

This tutorial is an introduction to Topometry Optimization. A simple cantilever plate is used to demonstrate element-by-element optimization of thickness.

Starting BDF Files: Link
Solution BDF Files: Link
Link Link Link
Topometry optimization example MSC Nastran Topometry Optimization of a Composite Panel

This tutorial covers the use of Topometry Optimization to determine ply shapes.

Starting BDF Files: Link
Solution BDF Files: Link
Link
Topometry optimization example MSC Nastran Topography Optimization - Bead or Stamp Optimization

This tutorial covers the use of Topography Optimization to determine optimal bead or stamp patterns. MSC Apex is used afterwards to review the results of the optimization.

Starting BDF Files: Link
Solution BDF Files: Link
Link

Advanced Optimization Tutorials

Title and Description PDF Tutorial YouTube Tutorial
Manually Starting MSC Nastran and Uploading Results

This tutorial discusses how to manually start MSC Nastran.

This tutorial also discusses how to upload result files (.f06, .csv or multiopt.log) to the SOL 200 Web App.
Link Link
CSV Export and Import for Design Variables, Responses and Constraints

Large design models may have thousands of design variables and constraints. The web app supports the export and import of CSV files. With the aid of Excel, hundreds of entries can be rapidly configured. This tutorial discusses the CSV export and import functionality.
Link
Responses in Design Model

Responses are the outputs following a structural analysis. Some examples include displacements at nodes, stresses in elements, etc. Each design cycle during the optimization process involves a structural analysis and the production of responses. This tutorial discuses the use of the Responses App to carefully inspect responses found in the .f06 file.
Link
The Design Cycle Process of MSC Nastran SOL 200/Optimization

MSC Nastran Optimization or SOL 200 takes multiple design cycles to shift the design variable values, for example dimensions of your structure, until an optimum of your objective is reached.

This video walks you through the design cycle process in MSC Nastran Optimization.
Link
Summary of Design Cycle History in the .f06 file - MSC Nastran Optimization

At the end of an optimization with MSC Nastran, the final summary of the optimization is available at the bottom of the .f06 file.

This video discusses how to interpret the final summary.
Link
Viewing Optimization results in Excel - MSC Nastran Optimization

The results of an MSC Nastran Optimization can be viewed in excel. Information such as the change of objective and design variables can be viewed for each design cycle.

This video walks you through the process of viewing optimization results in excel.
Link
How to create a new bdf file with optimized properties - MSC Nastran Optimization

Once MSC Nastran Optimization has produced optimized design variables, e.g. optimized structural dimensions, the original .bdf/.dat file must be updated with the new property values.

This video establishes two methods of updating the original .bdf/.dat file.
Link
How to fix 'RUN TERMINATED DUE TO HARD CONVERGENCE TO A BEST COMPROMISE INFEASIBLE DESIGN'

MSC Nastran SOL 200 or Design Optimization employs an intelligent method of handling hundreds of design constraints. This video covers the following:
  1. Discusses a solution to fix this message found in the .f06 file: 'RUN TERMINATED DUE TO HARD CONVERGENCE TO A BEST COMPROMISE INFEASIBLE DESIGN'
  2. Normalized constraints
  3. Constraint screening
  4. Maximum constraints
  5. Interpreting normalized constraints and constraints in the .f06 file
Link
How to perform a Sensitivity Analysis in MSC Nastran SOL 200

This video details 3 methods of performing a sensitivity analysis with MSC Nastran.

The following points are discussed:

What is sensitivity analysis?
How to perform a sensitivity analysis with MSC Nastran
Interpreting the sensitivities
Link
How to verify design variables - MSC Nastran Optimization

Configuring a design model for MSC Nastran Optimization is a simple process, but care must be taken to ensure the configuration was properly done.

This video details how to verify that design variables have been properly configured.
Link
How to verify design constraints - MSC Nastran Optimization

A simulation of a Finite Element model can produce an overwhelming amount of output such as displacements, stresses, strain, etc.

Design constraints are imposed on specific outputs. This video outlines a procedure to ensure the design constraints are applied to the correct responses such as displacements, stresses, strains, etc.
Link

HDF5 Explorer Tutorials

Title and Description PDF Tutorial YouTube Tutorial
Use the HDF5 Explorer to Create Plots

Starting with MSC Nastran 2016, results can be outputted to the HDF5 file type. This tutorial introduces the HDF5 Explorer and the following concepts:
  • Acquiring datasets from the HDF5 file (.h5)
  • Creating plots
    • Applied Loading vs. Frequency
    • Displacement vs. Frequency
    • Mode Participation Factors vs. Frequency
    • Natural Frequency vs. Mode Number
  • Exporting data to CSV
Starting Files: Link
Link

Machine Learning Tutorials

Title and Description PDF Tutorial YouTube Tutorial
Introduction - MSC Nastran Machine Learning Web App

There are many machine learning methods available, such as neural networks, decision trees, genetic algorithms, supervised learning, etc.

Watch this presentation to gain an understanding of the machine learning methods implemented in the MSC Nastran Machine Learning web app. The MSC Nastran Machine Learning web app makes use of the following methods:

  • Gaussian Process Regression - Used to predict the output of MSC Nastran
  • Bayesian Optimization - Used to select candidate designs


Link Link
Machine Learning, Structural Optimization of a 10 Bar Truss with MSC Nastran SOL 400

Machine learning methods are used to optimize a truss structure. This example features the optimization of nonlinear responses from MSC Nastran SOL 400. MSC Nastran is used to evaluate the FE model. The design variables are the cross-sectional areas of the rod elements. The objective is to minimize the weight of the structure while constraining the axial stresses and displacements. Most machine learning examples are lower dimension problems with 1 to 5 parameters. This tutorial demonstrates a higher dimension scenario with 10 parameters.

Starting Files: Link
Solution BDF Files: Link
Link Link
Machine Learning, Structural Optimization of a 3 Bar Truss

Machine learning methods are used to optimize a truss structure. MSC Nastran is used to evaluate the FE model. The design variables are the cross-sectional areas of the rod elements. The objective is to minimize the weight of the structure while constraining the axial stresses.

Starting Files: Link
Solution BDF Files: Link
Link Link
Machine Learning, Nonlinear Buckling (Post-Buckling) Optimization of a Reinforced Cylinder with MSC Nastran SOL 400

A reinforced cylinder is fixed at the base and a load is applied laterally at its top. Machine learning is performed to determine the optimal thicknesses of the reinforcements to achieve a minimum eigenvalue of 30 while minimizing the weight. This example features the optimization of nonlinear responses from MSC Nastran SOL 400 and involves nonlinear buckling (post-buckling) analysis.

Starting Files: Link
Solution BDF Files: Link
Link Link
Parameter Study, Global Optimization with a Latin Hypercube Design

Consider the optimization of a generic engine model that represents a V6 engine and automatic transmission as well as other components such as engine mount, transmission mount, crankshaft, drive axles, and flywheel. The Parameter Study web app is used to configure multiple local optimizations, each with different initial values for the variables, and MSC Nastran is used to perform each optimization. Frequency response plots are created afterward. The process of performing multiple local optimizations, then taking the best, or optimal, design is known as Global Optimization.

Starting Files: Link
Solution BDF Files: Link
Link Link
Prediction Analysis, Gaussian Process Regression

This tutorial demonstrates the use of Gaussian process regression.

Link Link
Prediction Analysis, Dynamic Impact of a Rigid Sphere on a Woven Fabric

Consider a transient analysis of a rigid sphere impacting a woven fabric. The parameters allowed to vary include the friction coefficients. The response of interest are the displacements.

This tutorial describes how to configure multiple MSC Nastran runs to generate training data. Gaussian process regression is used to train a surrogate model and make predictions. The prediction performance of the surrogate model is also evaluated. Also discussed are instructions to create displacement vs. time plots.

Starting Files: Link
Solution BDF Files: Link
Link
Prediction Analysis, Frequency Response Analysis (SOL 111)

Consider a frequency response analysis of a ground vehicle. For different configurations of the ground vehicle, there is a desire to rapidly determine the frequency responses, including accelerations and pressures, while keeping the number of Finite Element (FE) solver runs to a minimum.

This tutorial describes the procedure to use Gaussian progress regression as a surrogate model for computationally expensive Finite Element (FE) based simulations. This tutorial walks users through the process of acquiring training data, fitting the surrogate model, making predictions and quantifying uncertainty. In addition, the process to screen variables/parameters via Automatic relevance determination (ARD) is discussed.

Starting Files: Link
Solution BDF Files: Link
Link
Prediction Analysis, Buckling

Consider a linear buckling analysis. The parameter allowed to vary is a spring constant. The response of interest is the buckling load factor.

This tutorial describes how to configure multiple MSC Nastran runs to generate training data. Gaussian process regression is used to train a surrogate model and make predictions. The prediction performance of the surrogate model is also evaluated.

Starting Files: Link
Solution BDF Files: Link
Link
Prediction Analysis, Post-Buckling

Consider a post-buckling analysis. The parameter allowed to vary is a spring constant. The responses of interest include the applied load and displacements.

This tutorial is similar to the previous tutorial named Prediction Analysis, Buckling. Like the previous tutorial, this tutorial discusses the use of Gaussian process (GP) regression to train a surrogate model. To expand your experience with GP regression, this tutorial purposely demonstrates a scenario where a poorly fitted model is obtained and what the procedure is to remedy this issue. Also discussed are instructions to create load vs. displacement plots .

Starting Files: Link
Solution BDF Files: Link
Link