Design of Experiments: Testing synergistic interactions of design factors
Engineering the Solution Space
Once the challenge is rigorously defined and framed—a condition achieved by the Problem Definition phase—the emphasis shifts entirely to creation. Albert Einstein advised that problems cannot be solved using the same kind of thinking that created them. This insight demands innovative, yet structured, thought during the Solution Design phase to develop alternatives that directly address the stakeholder objectives articulated in the Value Model.
Solution Design is the deliberate process of composing a set of feasible alternatives, or candidate solutions, for presentation to the decision maker. This is an iterative phase that cycles frequently across the spectrum of define, design, and decide, with a constant focus on refining ideas and ensuring practical feasibility. The objective is not simply to select from available options, but to generate the best possible solution set that maximizes value.
Generating Creativity Through Structured Techniques
Innovative thinking must be pursued deliberately, not haphazardly. The transition from raw ideas to viable solutions requires applying structured ideation techniques that deliberately counteract common cognitive biases and group limitations.
Foundational Techniques for Ideation
The most common method for idea generation is Brainstorming, a technique that prioritizes quantity and spontaneous contribution while strictly deferring all criticism. Its variations, such as structured and unstructured sessions, aim to generate a high volume of ideas, which can then be organized using tools like Affinity Diagramming. However, traditional brainstorming risks being compromised by dominant individuals or groupthink.
To mitigate these risks, solution designers employ techniques like Delphi and Brainwriting. Delphi methods rely on anonymous, iterative surveys among subject matter experts, minimizing social pressure and personal bias. Brainwriting achieves similar anonymity by requiring ideas to be written down silently, thus eliminating the influence of vocal majorities. Edward DeBono’s Lateral Thinking encourages creativity by fundamentally questioning the basic assumptions and context of the problem, avoiding the trap of simply “digging the same hole deeper”.
Morphological Analysis and Alternative Generation
Morphological Analysis (MA) provides a systematic, highly structured method for combining components to explore the total possible solution space. The technique, often visualized using Zwicky’s Morphological Box, requires the team to identify all critical design parameters and their possible variables, mapping every conceivable combination into a multi-dimensional matrix. For a complex system, this quickly generates hundreds or even thousands of potential alternatives, even if many are logically or empirically inconsistent.
MA is essential for ensuring that no potential solution architecture is overlooked, enabling the team to select solution architectures that span the design space. The resulting alternative generation process often relies on existing organizational knowledge or external search tactics, although solutions generated via original design tactics—which stress innovation and custom-made alternatives—account for approximately 24% of strategic decisions studied.
The Analytical Core: Screening, Costing, and Improvement
Feasibility Screening of Candidate Ideas
The large pool of ideas generated must be ruthlessly filtered against mandatory stakeholder requirements. Feasibility Screening acts as a series of increasingly fine filters, quickly eliminating ideas that fail to meet non-negotiable criteria. The filtration levels correspond to the requirements identified during Problem Definition: Needs (“must have” requirements), Wants (“should have” requirements), and Desires (“nice to have” features).
An alternative is immediately rejected (No-go) if it fails any Needs criterion, although the design team may choose to refine or modify the idea to make it feasible for a subsequent screening. Only alternatives that pass this screening become true solution candidates for quantitative evaluation. This process confirms that resources will be allocated only to demonstrably feasible options.
Cost Analysis and Affordability
Cost constraints are a non-negotiable factor in system feasibility. During this phase, cost analysts develop or refine the life cycle cost (LCC) model, ensuring it completely covers the estimated development, manufacturing, and operational costs for every candidate solution. Systems engineers must verify the cost model’s completeness and then compute the estimated LCC for each candidate.
Affordability is a continuous consideration, driven by the principle that solutions must provide the best value for the resources. Consequently, LCC estimates must be developed in detail to support a potential tradeoff against the projected value, which is critical for moving into the Decision Making phase.
Refining Solutions with Design of Experiments
Once costed, candidate solutions are evaluated quantitatively using predictive models and simulations to understand their performance over time. To gain maximum information efficiency, the systems engineer employs Design of Experiments (DOE), a methodology for simultaneously studying the individual (main) and combined (interaction) effects of system factors (variables).
DOE utilizes structured tests, such as 2-level factorial designs, to move factors (e.g., engine thrust, number of fins) from a low level to a high level to calculate the average change in the performance measure (response). A key strength of DOE is identifying interaction effects, which show how the impact of one factor depends on the level of another—a crucial insight into system synergy. For example, combining two changes may yield a superior result not predictable from studying the two factors in isolation.
For high-complexity systems involving many factors, Fractional Factorial Designs allow the team to achieve nearly the same results with a fraction of the design points required for a full factorial design. This efficiency comes at the cost of potential precision loss (confounding), which the concept of design resolution measures. Lower-resolution designs are highly effective as screening designs, quickly eliminating factors that prove not to contribute significantly to the outcome. Tools like Pareto Analysis visually support this screening, separating the “vital few” factors that contribute most significantly from the “trivial many”.
From Potential to Preferred
Solution Design is the laboratory of the SDP, where the imaginative ideas generated by systems thinking are subjected to rigorous engineering discipline. By applying structured idea generation, LCC analysis, and experimental design (DOE/Pareto), the solution design team reduces a vast possibility space into a manageable, highly refined set of candidate solutions. These preferred solutions, now fully documented with cost and predicted performance data, are ready for the final quantification and tradeoff analysis required in the Decision Making phase.