Esteco engineering logo
Book a demo
Book a demo
Technology

Design of experiments

Leverage our design of experiments (DOE) algorithms to better understand the design space and extract as much information as possible with a limited number of design evaluations.

Book a demo Contact us
Design of experiments reduce the complexity of your design problem

Reduce the complexity of your design problem

Gain deeper insight into the design space, pinpoint the most important design variables, and reduce the number of evaluations needed to find the optimal solution. Explore key regions of certain designs without testing every possible combination of design variables using our DOE methods.

What is design of experiments

Defining an engineering problem and selecting variables frames the region of interest, called the design space. With design of experiments (DOE), the design space can be explored efficiently by considering the combined influence of all variables and modeling the system’s response across a wide range of values. The goal is to explore the design space in a way that maximizes understanding of its properties while minimizing experimental effort, ensuring that the selected sample points provide a reliable representation of the entire design space.

How to choose the design of experiments methodology

The choice of the most suitable DOE depends largely on the problem at hand, the purpose of the exploration (correlation analysis, RSM training or optimization) and the available computational resources. What you can do with DOE optimization:

  • identify the most important design variables,
  • spot relationships between variables,
  • build a dataset for RSM training or AI-data driven predictive modeling such as reduced order models (ROM),
  • generate a suitable starting population for optimization algorithms.

Choose the most suitable DOE algorithm

Space fillers

Initial exploration of the design space or global optimization.

  • Random
  • Sobol
  • Uniform Latin Hypercube
  • Incremental Space Filler
  • Constraint satisfaction

Design for Robustness and reliability

Identify factors that influence robustness and reliability.

  • Latin Hypercube – Monte Carlo
  • Taguchi Orthogonal Arrays

Statistical Designs

Fit response surfaces and study factor effects.

  • Full Factorial
  • Reduced Factorial
  • Central Composite Designs
  • Box - Behnken
  • Placket Burman

Adaptive DOE

Incrementally and adaptively build DOE for your problem.

  • MACK
  • Lipstichts sampling
  • Adaptive Space Filler
  • Random
  • Sobol
  • Uniform Latin Hypercube
  • Incremental Space Filler
  • Constraint satisfaction
  • Latin Hypercube – Monte Carlo
  • Taguchi Orthogonal Arrays
  • Full Factorial
  • Reduced Factorial
  • Central Composite Designs
  • Box - Behnken
  • Placket Burman
  • MACK
  • Lipstichts sampling
  • Adaptive Space Filler

Leverage modeFRONTIER to predict designs faster

Learn more about modeFRONTIER

modeFRONTIER’s DOE methodology

Gain deeper insight into design space, identify the most important design variables and reduce the number of design evaluations required to find the optimal solution. modeFRONTIER provides a wide selection of DOE algorithms for sampling the design space.

  • Space filler DOEs serve as a starting point for subsequent optimization processes or to create a database for training response surface models (RSM).
  • Statistical DOEs create samplings for sensitivity analysis, allowing a more in-depth understanding of the problem by identifying the sources of variation.
  • Robustness and reliability DOEs help create a set of stochastic points for robustness evaluation.
  • Optimal design DOEs help you effectively reduce the dataset volume in a suitable way.

How our customers use design of experiments

Our DOE technology helps our customers plan and execute experiments to maximize knowledge acquisition.

1/4

Key challenges solved by a design of experiments approach

Optimize inefficient search methods

Traditional one-factor-at-a-time and trial-and-error testing is inefficient, leading to an excessive number of costly and time-consuming evaluations.

Input variables affecting the experiment

Testing all combinations of input parameters in a multi-variable problem is often impractical.

Overcome poor optimizer initialization

Optimizers like MOGA-II or FAST struggle without good starting points, risking local optima. DOE provides diverse initial designs that uniformly explore the space, seeding robust optimization workflows.

Inadequate RSM training sets

RSM need representative data to approximate system behavior accurately. DOE constructs optimal training sets with balanced coverage and low correlation, enabling reliable metamodels for faster predictions.

Benefits of DOE optimization techniques

DOE tools are essential for tackling the complexity of design problems in engineering simulation by maximizing the knowledge gained from experimental data through smart placement of points in the design space. This allows you to:

Save time and resources for experiments

Quickly gain a solid statistical understanding of the problem by identifying the sources of variation.

Check for robust solutions

Identify sources of variation and variable interactions at an early stage of the design process.

Better decision making

Use models of the problem informed by detailed knowledge of the design space.

Leverage VOLTA to democratize design space exploration

VOLTA is the only SPDM solution with fully integrated design of experiments capabilities. It democratizes DOE execution by combining web-based collaboration, process automation, and data management. This simplifies access for non-simulation experts and empowers your teams to:

  • exchange simulation data across teams,
  • share DOE-enabled simulation workflows with other subject matter experts,
  • connect simulation to the digital thread of product data.

Scale and democratize DOE workflows with VOLTA

Learn more

Frequently asked questions

How do I set up a Latin Hypercube DOE in modeFRONTIER and link it to a CFD solver?

Steps:

  1. Start with a new workflow and define your parameters
  2. Insert the Latin Hypercube DOE node
  3. Connect the DOE to your CFD solver
  4. Define outputs from the CFD solver
  5. Run the workflow
How can I visualize the sensitivity results after running a DOE?

You can explore sensitivity results in modeFRONTIER right after your DOE completes. The platform gives you several visual tools that help you understand which inputs drive your outputs the most. You don’t need extra setup—everything lives in the Run Analysis environment.
Here’s a simple path you can follow.

  1. Open the run and go to the analysis dashboard
  2. Use the sensitivity-specific tools
  • Correlation Matrix
  • Scatter Plots
  • Parallel Coordinates Chart
  • Sobol / FAST / Non-Parametric Sensitivity (if available)
  1. Fit a response surface for deeper insights
  2. Quick sensitivity workflow (step-by-step)
How do I launch a modeFRONTIER DOE from VOLTA and monitor it remotely?

You can launch a modeFRONTIER DOE directly from VOLTA and monitor everything remotely through VOLTA’s web interface. The flow is simple once you know where things live. Below is a clear, step-by-step path that matches how most teams use VOLTA for distributed DOE execution.

  1. Upload your modeFRONTIER project to VOLTA
  2. Configure your DOE run inside VOLTA
  3. Select a cluster or execution engine
  4. Launch the DOE
  5. Monitor the DOE remotely in real time
  • Run Status Board
  • Live Charts
  • Workflow & execution logs
  1. Explore results directly in VOLTA
Can VOLTA help me compare two DOE plans created by different engineers?

Yes — VOLTA is great for comparing two DOE plans, even if they were created by different engineers, use different variable ranges, or run separate solver setups. You can place both DOEs in the same VOLTA project (or keep them separate) and use VOLTA Analytics to compare them side by side.

Here’s how you can do it in a simple, practical way.

  1. Upload both DOE runs into VOLTA
  2. Assign both experiments to the same Analytics workspace
  3. Use side-by-side comparison tools
  • Parallel Coordinates
  • Scatter Plots & Density Plots
  • Correlation Matrices
  • Performance Comparison
  1. Create dashboards to present comparisons
  2. When is this especially useful? VOLTA helps a lot when engineers use different approaches, like:
  • One uses 100-sample LHC; another uses 40-sample optimized LHC
  • One modifies variable ranges
  • One introduces constraints; the other does not
  • One uses a DOE that fails often in the solver VOLTA makes these differences visible in minutes.
What’s the recommended workflow for DOE → RSM → multi-objective optimization?

Here’s a clean and practical workflow you can follow when you want to move from DOE → RSM → multi-objective optimization. It keeps things simple but still respects good engineering practice. You’ll see how each step feeds the next so you build accuracy and confidence along the way.

  • Define the problem clearly
  • Build the initial DOE
  • Fit your response surface model (RSM)
  • Refine the DOE iteratively (if needed)
  • Launch the multi-objective optimization (MOO)
  • Validate the Pareto solutions on the real model
  • Rebuild the surrogate and re-optimize (optional loop)
  • Present and interpret the results
How do I publish and reuse DOE experiments from modeFRONTIER in VOLTA?

You can publish DOE experiments from modeFRONTIER into VOLTA and reuse them across teams, projects, and future studies. VOLTA acts as a shared library where your DOE runs, data, workflows, RSMs, and dashboards become reusable digital assets.

Below is a clear step-by-step guide that shows how to publish, organize, and reuse these experiments.
Upload your modeFRONTIER workflow (.mfw) into VOLTA

  1. Run (or import) your DOE experiment inside VOLTA
  2. Publish the experiment as a reusable asset
  3. Reuse the DOE data in new VOLTA experiments
  • Reuse the DOE as a dataset for new RSMs or analytics
  • Reuse the DOE as a starting point for new optimization workflows
  1. Share DOE results across teams
  2. Version-control your DOEs (optional but recommended)

Experience VOLTA and modeFRONTIER software

Whether you have a question about our products, licensing options or pricing, our solution experts are here to assist you.

Book a demo