Coming to Terms: The Design
By Daniel I. Newman
Vertiflite, July/August 2021
This series focuses on terms used in our expanding vertical flight community, addressing the uses of terminology that threaten to become routine expressions or idioms — or already are — but are misleading or erroneous. Here, we address the fact that “the design” is only an ideal.
A product or system has a form, a fit, a method of fabrication and a function. Form is geometry and materials. Fit is internal assemblies, tolerances and interfaces to other parts. Fabrication is formation, assembly and inspections. Function is performance both absolute and relative to the requirements. The first three can be considered “the design” and function is “the capability.”
A dichotomy for the engineers who develop and define products is designers and assessors (both analysts and testers). The designers define the form and fit, and so document what the product is. Analysts predict and testers measure performance to see if it is good enough. Trades and decisions at all levels during development seek a design that best balances all requirements and constraints. And if the design is not good enough, there is iteration of the design until it is.
Historically, large engineering firms relied more on analysis for product development, and testing to confirm compliance. Garage tinkerers rely on cut-and-try testing. At a high level, these are just variations on this same “design, then assess” theme. The explosion of high-fidelity, physics-based modeling, on the shoulders of Moore’s Law for computing power, has enabled expanded use of simulation for product development — even for tinkerers — and perhaps soon for compliance as well.
A challenge for complex products, and even simple ones with a lot of iteration, is to keep track of what is “The Design.” The management of the evolving product/system configuration is critical, as detail is added and early definition changes. This is true for both the physical configuration and the functional architectures. Each version of each component and assembly must be named and documented (e.g., “Best Technical Approach 3.2”), so that the analysis and test data can be correlated and used to mature the design to closure efficiently, effectively and seamlessly. Note that the time required for assessment of any part or assembly is an issue, as late data isn’t helpful unless it confirms prior decisions.
Perhaps least understood is how “The Design” is really just a plan. Every part of the development process uses the design as a point of departure, but each deviates in its own way.
Production: Fabrication of parts and assemblies encounters tolerances, with the acceptable tolerances for each part based on affordability. So, the as-built product represents the design, but every one differs. This is very pronounced for early helicopters with aluminum parts with compound-curvature built from hand-drafted lofted surfaces. The mechanics knelt on the parts to successfully buck the rivets to fit the frames, and the aircraft worked well enough. No one worried that the as-built product didn’t match the as designed lofts, until subsequent computer-aided design (CAD) models from digitized mylar drawings were used for new parts that didn’t fit the as-built aircraft.
Analysis: Analysis of parts is done with approximations of the design as well. Whether computational structural dynamics (CSD), computational fluid dynamics (CFD) or computational electromagnetics (CEM), the surfaces and solids are gridded into finite elements to represent the design. Ideally, these include high-density meshes when the geometry is complex or attributes (e.g., loads, turbulence or scattering) are intense to capture the important details that drive the relevant performance. Engineers hope to finish the analysis or test in time to affect the relevant decisions with a design that fits with all the other changes. But the analysis only explores the phenomena and interactions the engineers know and try to model, so they generate as-analyzed performance to predict that of “The Design,” with confidence that comes from correlation with testing.
Material Testing: Testing relies on approximation as well. Every material specimen cannot be tested to failure to be sure of its properties, or there’d be none left to use. So, material allowables are developed using a small sample of test data with “knock-downs” from the tested sample to account for material variation, with a strong relationship between sample size and knock-down. Analyses of new and complex phenomena (e.g., shape memory alloys) and complex joints (e.g., bolted bonded composites) and innovative configurations (e.g., multi-rotors) are without the depth of test data to provide the same confidence in predictions as for conventional designs.
Product Testing: Product or component testing is an approximation as well; only final compliance has the actual finished product available to test. For early tests that are constrained by cost, schedule and knowledge of the design, parts and interfaces are represented as well as they can be. So, the as-tested configuration also differs from “The Design.”
Confidence in analysis is based upon correlation of as-tested data with the as-analyzed predictions, both approximations of “The Design.” This lack of as-tested data is one of the challenges with innovation, whose delays and costs terraform the so-called “Valley of Death” between “good ideas” and fielded products.
It is said that “no one believes analysis but those who did it” (as they know the actual level of detail that was used) but “everyone believes test except those who did it” (as they know the amount of putty and tape used to get the testing done on time and budget). Compliance is confirmed by comparing the tested (or predicted) performance to the requirements. So, “The Design” is still only the plan.
Promotion: Articles, brochures and proposals fuel the imagination with artists’ renderings that present the wish list for design, the performance and the operational capability. These are not a commitment but a vision — making claims of performance are just that, claims. Caveat emptor: beware the promise that “it does” or “it can.”
At every stage in the creation and development and production of a product or system, there are many designs, from initial to current to hoped. The important information comes from all the subsequent activities that use the design but depart from it — all for good reason.
So, when someone talks about “The Design,” keep in mind that it’s a reference point… but it’s only the plan.