Nigel Howard recently criticised an article published on The Fifth Estate concerning a study that looked at embodied versus operational energy consequences in the context of house star ratings. In this article he explains his critique of hybrid life cycle assessment in more detail.
The traditional approach to life cycle assessment is process LCA, conducted according to the ISO14040/4 international standard. Process LCA requires careful consideration of the goal of a study focusing on the key questions to be addressed. A boundary is then drawn around the system that defines the physical scope and timeframe over which the study is to be conducted.
The boundary typically spans from extraction of the raw materials and energy sources right through to the end-of-life disposal and waste consequences. Process LCA omits very many remote consequences of the product lifecycle, since it would not be practical to trace the impacts of all of these. Moreover they are immaterial to the decisions being informed by the LCA.
In recent years, a new approach to LCA has been devised that is unbounded in scope or over time – consequential LCA. Consequential LCA is especially valuable for investigating policy making or planning decisions where the decisions that are informed will themselves perturb the supply/demand balance of markets or where it would be hard to define either a physical or temporal boundary that could contain the trail of consequences informed by the LCA. Consequential LCA relies on scenarios predicting into the future to guide its results and conclusions.
Both methods are needed to address different problems. The analogy that I like to use is between a map and a plan. Imagine we were building a new highway linking two towns and we needed to find the site to start work. We would take out our map, which is based on the existing road network, and use this to navigate to the start of the new road site to start our work. The map of the existing road system is analogous to process LCA – the map was drawn based on recent past information for the area that we were interested in. The map would be bounded to an appropriate scale to see the detail that we needed (a globe, for instance, would not be helpful).
Having arrived at the construction site we would take out our plan for the works – the plan would show the route of the new road, but until the road is built it would not enable us to get from one place to the other.
The responsible council/government will have examined all kinds of scenarios and routes for the new road – they would have considered the effect on traffic flows in the new road and all connected feeder roads, the effect on signalling, the effect of noise, pollution and run-off, the effect on businesses and property values in the area. The plan is analogous to the consequential LCA.
If I want to get from A to B now, I need the map. If I want to assess the environmental merits of a product or service compared to alternatives available to me now, I need an attributional LCA bounded in scope to focus on the features of the alternatives.
If I want to understand the full implications of a new road then I need a plan. If I want to assess the environmental implications of a new government policy or a major new piece of infrastructure then I need an unbounded consequential LCA and scenarios forecasting the most likely implications.
To practically conduct a consequential LCA it is common to use input-output data. Input-output tables show the industrial/commercial economic sectors as both rows and columns in a matrix – the values within the matrix show the value of product/service sold from the row sectors to the column sectors. They map the flow of money and value through an economy. By correlating the financial values with physical quantities of products, energy sources etc. and inverting the matrix, the environmental consequences of products and services can be estimated. Provided the tables also account for inputs and outputs, they are only bounded globally.
There are pros and cons to both attributional and consequential LCA. Consequential LCA reveals many of the remote consequences arising from products/services from say insurance, legal, marketing – activities that would typically be omitted from a process LCA.
On the one hand, these can add up to a significant omission, but on the other, they are remote from the production activities and don’t inform a practical decision that anyone is responsible for taking. Additionally, they tend to add a similar overhead onto competing products, somewhat marginalising the distinctions that a process LCA would highlight.
In recent years the idea of merging process LCA data with I/O consequential data has been proposed to fill gaps in available process data and to bridge the gap between process and consequential LCA. This seems an attractive idea, but the moment you mix bounded scope data with unbounded scope data you have neither the clarity of a bounded scope study nor the consistent full reach of a consequential study.
Of course, unbounded scope data has much higher impact consequences than bounded scope data for the same product/service, so what do mixed bounded/unbounded scope hybrid LCA results actually mean? It’s like taking six apples and two apple trees and trying to talk meaningfully about eight sources of fruit!
Referring back to the University of Melbourne study, the authors are making a good point, but my concern is that the hybrid LCA may be exaggerating the significance of the embodied energy components measured unbounded and consequentially compared to the operational energy consequences estimated by energy modelling, with a relatively narrow scope.
Using my ENVEST prototype design tool (universally consistent process LCA methodology) suggests that in most Australian climates there is still room for improved standards of energy efficiency before encountering perverse outcomes from materials embodied energy increases in materials like insulation – but the devil is in the details.
In 2016, LCA has lost a lot of credibility because many of the fundamental scientific principles that it is based on are now being routinely compromised – hybrid LCA is one example of many problems in LCA that I am blowing the whistle on.