Risk-based Central Composite Design
Also known as: Risk-informed CCD, CCD with risk assessment, Uncertainty-aware central composite design, Risk-integrated RSM
Risk-based Central Composite Design (Risk-based CCD) integrates formal risk identification and uncertainty quantification into the classical CCD framework. By coupling the rotatable second-order experimental structure of CCD with probabilistic risk metrics, engineers and scientists can simultaneously optimize process responses and characterize the risk of unacceptable outcomes — making it particularly valuable in regulated industries such as pharmaceuticals, chemical engineering, and advanced manufacturing.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Risk-based CCD when you need to optimize a process or formulation while explicitly bounding the probability of producing out-of-specification or unsafe outcomes — especially in regulated industries (pharmaceutical, chemical, food, aerospace) where demonstrating risk-aware optimization is required. It is appropriate when: (1) continuous quantitative responses can be measured at each experimental run, (2) the design space has 2–6 controllable factors, (3) a quadratic model is plausible, and (4) input uncertainties or manufacturing tolerances are non-negligible. Do not use it when factors are purely categorical (use factorial or mixture designs instead), when the number of factors exceeds roughly six (the number of CCD runs grows rapidly), when the relationship between factors and responses is highly non-smooth or discontinuous (invalidating the polynomial surface), or when no reasonable quantification of input uncertainty is available (which makes the risk evaluation step unreliable).
Strengths & limitations
- Combines process optimization and risk quantification in a single experimental campaign, reducing cost versus running separate studies.
- The rotatable CCD geometry provides equal prediction variance at all points equidistant from the design center, yielding balanced insight across the factor space.
- Risk propagation through the fitted surface is computationally efficient compared to risk analysis on first-principles simulation models.
- Aligns with regulatory guidance (ICH Q8/Q9/Q10, FDA Process Analytical Technology) requiring demonstrated risk-aware design spaces.
- Center-point replicates enable pure-error estimation and lack-of-fit testing, supporting rigorous model validation.
- Validity depends on the adequacy of the quadratic response surface model; highly nonlinear or discontinuous responses are poorly captured.
- Run count grows substantially with the number of factors — a 5-factor CCD requires at minimum 27 runs — making it resource-intensive for large factor spaces.
- Risk metrics are only as reliable as the assumed input uncertainty distributions; poorly characterized variability leads to misleading risk estimates.
- The approach addresses factor-response risk in a known operating region but does not account for unknown-unknown failure modes outside the design space.
Frequently asked
How is Risk-based CCD different from standard CCD?
Standard CCD optimizes mean process responses by fitting a quadratic surface and locating its optimum. Risk-based CCD adds two steps: upfront risk identification (which factors and limits matter for safety or compliance) and post-fit risk evaluation (propagating input uncertainty through the surface to compute the probability that outputs violate those limits). The experimental structure is identical; the difference lies in the framing, scope, and analysis that surround it.
How many factors can I study with a risk-based CCD?
CCD is practical for 2 to 6 continuous factors. Beyond six factors the run count becomes prohibitive (a 6-factor face-centered CCD already requires 77 runs including replicates). For higher-dimensional factor spaces, consider a fractional factorial screening step first to reduce the factor set before applying CCD.
What software can run this analysis?
The CCD structure and quadratic model fitting are available in JMP, Minitab, Design-Expert, and R (rsm package). Monte Carlo risk propagation through the fitted surface can be implemented in R (sensitivity, mc2d), Python (SALib, scipy.stats), or dedicated reliability software such as Crystal Ball or @RISK. Design-Expert integrates Monte Carlo simulation directly for design space probability analysis.
Is a risk-based CCD the same as Quality by Design (QbD)?
Not exactly, but the two are strongly aligned. QbD (per ICH Q8) is a broader pharmaceutical development philosophy requiring systematic risk-aware development. Risk-based CCD is one of the experimental and statistical tools used within a QbD program to define and verify the design space — the region of factor settings within which the process reliably meets quality specifications.
When should I prefer risk-based CCD over a robust CCD approach?
Both approaches address variability, but differently. Robust CCD (Taguchi-inspired) selects operating conditions that minimize response sensitivity to noise factors, treating robustness as an optimization objective. Risk-based CCD quantifies the actual probability of exceeding defined specification limits given known input distributions. Use risk-based CCD when you have regulatory or contractual requirements expressed as probability thresholds (e.g., P[failure] < 0.001) and when you can characterize input distributions from historical data or tolerance specifications.
Sources
- Box, G. E. P., & Wilson, K. B. (1951). On the experimental attainment of optimum conditions. Journal of the Royal Statistical Society: Series B, 13(1), 1–45. DOI: 10.1111/j.2517-6161.1951.tb00067.x ↗
- Central composite design. Wikipedia. link ↗
How to cite this page
ScholarGate. (2026, June 3). Risk-based Central Composite Design. ScholarGate. https://scholargate.app/en/experimental-design/risk-based-central-composite-design
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.
- Box-Behnken DesignExperimental design↔ compare
- Central Composite DesignExperimental design↔ compare
- Failure Mode and Effects AnalysisExperimental design↔ compare
- Reliability AnalysisReliability↔ compare
- Response Surface MethodologyExperimental design↔ compare