The Monte Carlo principle is simple. A calculation is repeated many times, each time choosing a different value for each parameter. A parameter can for example be an emission or raw material input. The results form a distribution from which uncertainty information can be derived with basic statistical methods.

The values that are being chosen in the Monte Carlo analysis must be within a specified distribution. To support this, you can specify the uncertainty on the inputs and outputs of a process or product stage in all new SimaPro versions.

SimaPro supports 4 types of distributions:

Monte Carlo Results

In the figure above the 95% interval is shown per impact category.  

SimaPro stores all the outcomes of each calculation and these results form a distribution themselves, as can be shown in the figure below. In this figure, we display the eco-indicator value of electricity Low voltage in Europe.

SimaPro can display such ranges for every impact category and even for every emission, both as graph and in tabular form.

Comparing products and dealing with correlations

When products are compared, we must observe the important issue of correlation.

Suppose we have two products. Product A is made of 20 kg of steel, while product B is made of 21 kg of steel.

Also suppose that the uncertainty in the CO2 output of steel production is very high, lets say +/-  100%.

If we calculate the Monte Carlo distributions for the CO2 emissions for product A and B, we would conclude that we cannot say product A is better than B.

However, since both products use the same steel, the uncertainty is completely correlated. In order to determine the difference in CO2 output this is not relevant. We can conclude the obvious fact that product A will have a 5% lower CO2 production than product B, because it simply uses 5% less of the same steel.

This example shows that there is a real danger that Monte Carlo calculations overestimate uncertainty if products are compared where correlations are not observed.

The new SimaPro versions consider correlations in a very sophisticated way. It will not show two overlapping distributions, as this can easily give a very wrong interpretation.

Instead, it will show in how many calculation product A scored lower than product B on a certain indicator or LCI result. Another way of representing the difference is to show the distribution of the value A/B, for each calculation.

The figure above displays the distribution of the ration of product A and B. If (for every calculation run) A/B is smaller than one, we can be sure that A has a lower load than B In the example 19% of the runs give a value larger than one, shown as red columns.

This means there is some uncertainty about the preference of product A over B.

Uncertainty data are being progressively added to the Australian LCA Database.