Skip to contentScholarGate
LibraryBookshelfDeskReview StudioAssistant
Sign in
On this page
IntuitionHow it worksWhen to use itStrengths & limitationsCommon pitfallsApplicationsFrequently asked🔒 Read the full methodSourcesRelated methods
Cite this pageSpotted an issue on this page? Report or suggest a fix →
Home›Experimental design›Simulation-Assisted Fractional Factorial Design
Process / pipelineEngineering methods

Simulation-Assisted Fractional Factorial Design

Also known as: SA-FFD, virtual fractional factorial design, computer-aided fractional factorial design, simulation-based FFD

Simulation-assisted fractional factorial design (SA-FFD) combines the statistical efficiency of fractional factorial experimentation with computerized simulation models to screen and estimate factor effects when physical experiments are too costly, hazardous, or time-consuming. A carefully chosen subset of factor-level combinations — the fractional factorial array — is executed inside a validated simulation model instead of (or alongside) a real process, dramatically reducing resource requirements while preserving the ability to identify main effects and low-order interactions.

ScholarGate
  1. Process / pipeline
  2. v1
  3. 2 Sources
  4. PUBLISHED
Cite this page →
Tools & resources
Download slides
Learn & explore

Read the full method

Members only

Sign in with a free account to read this section.

Sign in

Method map

The neighbourhood of related methods — select a node to explore.

Simulation-assisted fractional factorial design
Central Composite DesignDesign of experimentsResponse Surface Methodo…Simulation-assisted desi…

When to use it

Use SA-FFD when you need to screen many factors efficiently but physical experiments are too expensive, slow, dangerous, or destructive to run at the required scale. It is especially powerful in early design phases — process design, product development, system architecture — where a validated simulation model exists and rapid iteration is needed. Also appropriate when physical replication is infeasible but computational replication is cheap. Do NOT use when no validated simulation model is available or when the simulation is so computationally expensive that running the required number of design points is impractical; in those cases, consider space-filling designs (Latin hypercube, Sobol sequences) or surrogate modeling (Kriging, response surface metamodels) as alternatives. Avoid if the true underlying system is strongly nonlinear across the studied range and you do not add center points or augment to a response surface design afterward.

Strengths & limitations

Strengths
  • Eliminates or drastically reduces costly, time-consuming physical experiments during early design phases.
  • Preserves statistical efficiency of fractional factorial designs — main effects and key interactions are estimable with a minimum number of runs.
  • Full experimental control: simulation allows precise factor settings that may be impossible or unsafe in physical systems.
  • Easy replication and variance estimation in stochastic simulators at negligible marginal cost.
  • Enables rapid design-space exploration and supports iterative redesign before committing to physical prototypes.
Limitations
  • Validity of conclusions depends entirely on the fidelity of the simulation model; errors in the model propagate directly into factor-effect estimates.
  • Fractional designs alias (confound) certain effects; higher-order interactions cannot be estimated without augmentation or additional runs.
  • Screening designs (Resolution III) cannot distinguish main effects from two-factor interactions if both are active.
  • Deterministic simulators produce identical output for the same input, making replication-based variance estimation impossible without adding artificial noise or using metamodel uncertainty.

Frequently asked

How is this different from a standard fractional factorial design?

The statistical structure — the fractional array, aliasing patterns, resolution — is identical. The only difference is that each run is executed in a computer simulation model rather than in a physical experiment. This affects how variance is handled (deterministic vs. stochastic simulators), how model validation is done, and what additional confirmatory experiments are needed, but the design and analysis workflow follows the same principles.

What resolution should I choose for a simulation-assisted screening study?

Resolution IV or higher is generally preferred when simulation runs are cheap, because it keeps main effects clear of two-factor interactions. Resolution III is acceptable only if prior domain knowledge strongly suggests that interactions are negligible. Avoid Resolution III designs when the system is known to be interactive, because aliased effects cannot be disentangled without follow-up runs.

My simulator is deterministic — do I still need replication?

With a deterministic simulator, repeated runs at the same settings produce identical output, so classical replication adds no information about variance. Instead, add center points to detect curvature, or reserve a small set of confirmation runs. If you need variance estimates (e.g., for prediction intervals), introduce metamodel uncertainty quantification or use stochastic simulation wrappers.

How many physical confirmation runs are needed after the simulation study?

There is no fixed rule, but best practice is to run at least 3–5 physical confirmations at the predicted optimal settings and at one or two other design points. These confirm that the simulation model captures real-world behavior accurately enough for the decisions being made. Large discrepancies signal model inadequacy and require model recalibration before conclusions are acted upon.

Can I use SA-FFD when my simulator takes hours per run?

If each simulation run takes hours, a 32- or 64-run fractional design may still be feasible over days or weeks, especially with parallel computing. If the total compute budget is the binding constraint, consider instead a Latin hypercube design analyzed with a Kriging (Gaussian process) metamodel, which extracts more information per run for continuous factor spaces. SA-FFD is best when run counts are feasible and linear effects are the primary target.

Sources

  1. Kleijnen, J. P. C. (2008). Design and Analysis of Simulation Experiments. Springer. ISBN: 978-0387718125
  2. Montgomery, D. C. (2017). Design and Analysis of Experiments (9th ed.). Wiley. ISBN: 978-1119113478

How to cite this page

ScholarGate. (2026, June 3). Simulation-Assisted Fractional Factorial Design. ScholarGate. https://scholargate.app/en/experimental-design/simulation-assisted-fractional-factorial-design

Related methods

Central Composite DesignDesign of experimentsResponse Surface MethodologySimulation-assisted design of experiments

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

  • Central Composite DesignExperimental design↔ compare
  • Design of experimentsExperimental design↔ compare
  • Response Surface MethodologyExperimental design↔ compare
  • Simulation-assisted design of experimentsExperimental design↔ compare
Compare side by side →

Similar methods

Simulation-assisted full factorial designSensitivity Analysis with Fractional Factorial DesignSimulation-assisted Box-Behnken designSimulation-assisted design of experimentsSimulation-assisted response surface methodologyOptimization-assisted fractional factorial designSensitivity analysis-integrated full factorial designFractional Factorial Experiment

Related reference concepts

Variance Reduction TechniquesStatistical Simulation Methods: GeneralSimulationMonte Carlo MethodsComputational Techniques • Simulation ModelingHigh-Performance Statistical Computing

Spotted an issue on this page? Report or suggest a fix →

ScholarGate — Simulation-assisted fractional factorial design (Simulation-Assisted Fractional Factorial Design). Retrieved 2026-07-21 from https://scholargate.app/en/experimental-design/simulation-assisted-fractional-factorial-design · Dataset: https://doi.org/10.5281/zenodo.20539026
Quick facts
Originator
Box, Hunter & Hunter (FFD basis); Kleijnen and others (simulation integration)
Year
FFD: 1950s; simulation integration: 1980s–2000s
Type
Experimental design with computational augmentation
DataType
Simulated output responses (continuous or discrete); factor settings from fractional factorial array
Subfamily
Engineering methods
Related methods
Central Composite DesignDesign of experimentsResponse Surface MethodologySimulation-assisted design of experiments
ScholarGate

A content-first reference library for research methods — what each one is, how it works, and where it comes from.

Open data (CC-BY)

Explore

  • Library
  • Search the library…
  • Browse by field
  • Fields
  • Journey
  • Compare
  • Which method?

Reference

  • Subjects
  • Atlas
  • Glossary
  • Methodology
  • Philosophy

Your tools

  • Bookshelf
  • Desk
  • Chat

Company

  • About
  • Pricing
  • Contact
  • Suggest a method

Entries are compiled from published sources for reference. Verifying the accuracy and suitability of any information for your own use remains your responsibility.

© 2026 ScholarGate · A research-method reference library
  • Privacy
  • Cookies
  • Terms
  • Delete account