Research tools for analysing design margins in complex engineered systems
Three integrated modules — parametric margin valuation, top-down margin allocation, and probabilistic margin risk — built around the Margin Value Method.
Build calculation graphs for an engineered system, identify the margins that exist between decided values and their thresholds, and quantify what each margin is worth.
The Margin Value Method (MVM) evaluates each margin in a design on two axes: how much performance you would lose if the margin were eliminated, and how much external change the margin can absorb before it is consumed.
Inputs and decisions are connected through calculation nodes into a Margin Analysis Network (MAN). Margins arise wherever a decided value — for example a catalogue selection or a discrete choice — exceeds the threshold required by the design. The tool computes margin metrics for every such pair and visualises them on the Margin Value Plot.
Each margin in the network is summarised by three quantities, computed automatically once the parametric model is built.
How much the decided value oversatisfies the threshold, measured as (decided − threshold) / threshold. The visible “size” of the margin.
The performance loss that would occur if a given margin were eliminated, propagated forward through the calculation network. The cost of carrying the margin.
The extent to which a margin is consumed when an input parameter deteriorates to its maximum absorbable value. The benefit of carrying the margin under change.
Impact and absorption are plotted as a bubble chart, with bubble size set by local excess. The four quadrants give a designer a quick read on which margins to keep and which to reduce.
Provides change absorption with little performance cost. Keep, and consider growing.
Significant absorption, but at a meaningful performance cost. Worth a deliberate decision.
Neither costs much nor buys much. Generally safe to ignore.
High performance cost, negligible absorption benefit. Prime candidate for redesign.
Read the tutorial for a step-by-step walkthrough, browse the case studies for worked examples, or jump to the literature for the underlying papers.