Data precision: recalibrating expectations

The engineering properties of materials—their mechanical, thermal, and electrical attributes—are well characterized. They are measured with sophis­ticated equipment according to internationally accepted standards and are reported in widely accessible handbooks and databases. They are not exact, but their precision—when it matters—is reported; many are known into three-figure accuracy, some to more. A pedigree like this gives confidence. This is data that can be trusted.

Additional properties are needed to incorporate eco-objectives into the design process. They include measures of the energy committed and carbon released into the atmosphere when a material is extracted or synthesized— its embodied energy and carbon footprint—and similar data for processing of the material to create a shaped part. There are more properties, intro­duced in a moment. But before the introductions it helps to know what to expect.

Take embodied energy as an example. It is the energy to produce unit mass (usually, 1 kg) of a material from, well, whatever it is made from. It is a key input to any eco-tool. Unlike the engineering properties, many with a provenance stretching back 200 years, embodied energy is an upstart with a brief and not very creditable history. There are no sophisticated test machines to measure it. International standards, detailed in ISO 14040 and discussed in Chapter 3, lay out procedures, but these are vague and not easily applied. There is no pedigree here; it’s a mongrel. So just how far can values for this and other ecoproperties be trusted? An analysis, documented in Section 6.3, suggests a standard deviation of ±10% at best.

Bad news? Not necessarily. It depends on how you plan to use the data. Methods for selecting materials based on environmental criteria must be fit for their purpose. The distinctions they reveal and the decisions drawn from them must be significant, meaning that they must stand despite the imprecision of the data on which they are based.

The data sheets of Chapter 12 deal with this issue by listing all proper­ties as ranges: aluminium: embodied energy 200-240MJ/kg, for example. The ranges allow "best-case" and "worst-case" scenarios to be explored. When point (single-valued) data is needed, take the mean of the range.

Updated: September 29, 2015 — 12:14 am