I would like to group the values of different impact categories into a single index. What are the most common practices for this? Can anyone suggest me literature for this?
The use of a single index to express the results of multidimentional evaluation is dangerous - important information is often lost! It is better to express multidimentional results graficaly in a radial chart - at least this should always accompany the index.
Multidimentional evaluations, being multiobjective, involve systems thinking - an index is a symptom of reductionist thinking...
This is a very important question, and, as you might guess, the answer is not simple (if there is any proper one at all).
I think it is important to understand why you pretend to join categories that do not have much in common (normally there are different measurement units, scopes, detail levels, etc.). I ask because I do think that the answer and possible solutions depend on what you plan to do with the result. When assessment systems like LEEDS or DGNB give you their final output, it is a merged result of many analysis (not only but to a certain degree including LCA), it is exactly with the scope of the simplification and ease of use for commercial publication of the result. Is that what you are looking for?
In my ongoing research I do prever to leave things apart in order to avoid "comparing apples with bananas".
I know my answer is more of a question, but I hope it can help you.
Well, you have to do your homework studying existing literature and methodology: in LCA, end point indicators over the 3 areas of protection (ecosystems, humans, resources) may be combined into a single score (expressed in eco-points) according to assumptions and the philosophy you select (which area do you favor etc.). Another option, especially if you have only few impact categories which cannot lead to end-point indicators, and that you have a number of production units under study, is that you calculate an environmental efficiency score with the DEA methodology. A set of inputs (costs or physical inputs to production) lead to a set of outputs (your impact indicators). Data Envelopment Analysis will calculate one single efficiency score per production unit (so called Decision-Making Unit DMU). See relevant literature on my page. Finally, you may carry out your own research and compute one single indicator from the ones you have, based upon acceptable and justified assumptions and choices (+ proper normalization). All the best.
Thanks for the question and answers provided. Instead of grouping/aggregating different impact categories, consider identifying the important impact categories and study them (e.g., through multi-objective optimization). In this regard, our following papers on chemical processes may be of interest to chemical/environmental engineers/researchers.
Sharma, Chua and Rangaiah, "Economic and Environmental Criteria and Trade-Offs for Recovery Processes", Materials and Manufacturing Processes, 26, p. 431-445 (2011).
Lee and Rangaiah, "Optimization of recovery processes for multiple economic and environmental objectives. Industrial and Engineering Chemistry Research, 48, p. 7662–7681 (2009).
Dear all, thank you for your answers. In my PhD thesis, I'm developing an sustainable index. Specifically in the environmental sphere, I've been working with LCA and already have impacts (mid and endopoints), using Umberto software. Right now I'm researching the best way to aggregate LCA results to a single index, according to my goal and scope. Your helpful comments are helping me with this.
That depends on which method of characterization you use...By impact2002+, there is a single score available directly in simapro, that the common single score used in LCA. But that doesn't give some much interested... You should find, in Joliet 2003, the aggregation of mid point to end point then single score.
The use of a single index to express the results of multidimentional evaluation is dangerous - important information is often lost! It is better to express multidimentional results graficaly in a radial chart - at least this should always accompany the index.
Multidimentional evaluations, being multiobjective, involve systems thinking - an index is a symptom of reductionist thinking...
I don't think there is a One Best Way - there are several ways that work well depending on the situation. If you think of each method as a 'transformer' that conveys a particular type of information, then you can more easily assess the meaning and usefulness of the method.
For example, the ecoscarcity method (that somebody suggested) conveys the extent to which the environmental impact exceeds limits of ecological critical limits. With EPS, one gets a measure and ranking based of the willingness-to-pay to avoid the environmental impact. And then there is distance-to-law-targets, and then there are the rankings of panels, and so on… None of these methods are objectively equivalent, but taking a more interpreting stance, it is possible to say something about the information they convey. I have much of this summarized in the textbook I wrote (The Hitch Hiker's…) and also an old paper from 1993 together with Thomas Rydberg.
If your work is to understand the different available methods, then there is a lot to peruse. If your work is to develop another and new method, be specific about the meaning it is to stand for.
I agree with the two former reactions, you have the risk to loose information, so the purpose of your work needs to fit the solution. So for what purpose do you want to use your indicator. A s said a lot of different indexes can be constructed, depending on your objective it can be simple (define a baseline/year/situation and replace all values by an index (i.e. a value of 100), you can now add the different outcomes, relative changes work directly on the index. More complex one's is by using an threshold or a weighting factor for each factor.
Good question and to echo some of the other replies, you should make sure that the scope of your study will warrant something like computing a single score. My colleagues and I have used multi-criteria decisions analysis along with LCA to derive single LCA scores, just as you were asking about. There is plenty of research on MCDA in the literature and a bit of research on its use with LCA. We recently published an article demonstrating this for treated lumber options. The article information is on my page. Feel free to message me if you have further questions or comments.
I hope I'm not too late to answer you about this. Plenty of perfect answer have already been given so I will only give you one more interesting lead that I read few months ago. It is the PhD thesis of a guy named Perez Gallardo. His purpose is not really to obtain a single indicator but he tried to group several indicators together thanks to Principal Component Analysis (PCA) in order to reduce their number. It might be an interesting reading for you.
Here is a link: http://ethesis.inp-toulouse.fr/archive/00002416/
You must set weighing factors on the normalized damage categories. This is a very difficult choice. The easiest way is to integrate the environmental results (in term of damage categories) in an optimization procedure
There are several methods to get a single value, but I do not recommend it because the weighting factors could, and usually are, not be the most adequate for your system, taking into account geographical validiy. If you are interested in just the single value you have to estimate the valid factors.