Uncertainty

The modeling of LCAs is related to many different types of uncertainties. openLCA enables to determine the uncertainty of data with the aid of a Monte Carlo Simulation. With this simulation you can check how robust your results are. To calculate the uncertainty you can assign every input and output of every process in your product system a distribution and conduct a Monte Carlo Simulation.

First, open a process by doublecklicking it and go to the tab „Inputs/Outputs”. Click once in the field „Uncertainty distribution type” of every flow where you want to add a distribution (If you click twice you will open the flow in the editor). In the field „Properties” at the bottom of the process editor you can indicate geometric or arithmetic mean value, (geometric) standard deviation, Max or Min value depending on the selected probability distribution and save the changes.

Monte Carlo simulation

To start a Monte Carlo simulation click on the uncertainty symbol in the main menu and select your product system.

Now you can determine the calculation framework by selecting a calculation method (for small product systems, you should use the sequential method as it is then much faster than matrix inversion) and determining amount of simulation. You can analyse LCIA categories or flows; the selection can be changed during the simulation. Finally, click on „Start simulation” and the program will run. The simulation can be canceled by clicking on „Stop simulation”. A simulation run can be saved, and previous uncertainty calculations can be loaded.

Approximation

For an estimation of the uncertainty for flows openLCA provides another faster calculation option, by using Gaussian error propagation formulas. The approximation formulas assume normally distributed random errors for these flows. According to literature, Gaussian approximation formulas provide the same results as Monte Carlo Simulation provided that the relative error (i.e. the variation coefficient, standard deviation divided by the mean) is below 20% (Ciroth 2001). This condition is checked during the calculation; if the variation coefficient is found to be higher during the calculation, a warning message is issued together with the calculation results.
To use approximation for uncertainty calculation, you can edit settings for the calculation in the properties field of the inventory of a process, and change the default entry use uncertainty calculation to true. Note that you can use only the sequential methods for the calculation. As default uncertainty approximation is inactive, to activate it set „true” instead of „false” in the properties field.