Correlations/Dependencies in LCA UQ
-
- Posts: 2
- Joined: 20 Apr 2011 19:19
Correlations/Dependencies in LCA UQ
Is anyone in the LCA community considering dependencies/correlations between a unit process's flows for the purposes of uncertainty quantification (UQ)? I am wondering if any common LCI datasets include such information. -Mark Campanelli
Re: Correlations/Dependencies in LCA UQ
Hi Mark,
I am not aware of any attempt - could anyone else help? Interesting thought and question.
Best wishes,
Andreas
I am not aware of any attempt - could anyone else help? Interesting thought and question.
Best wishes,
Andreas
-
- Posts: 2
- Joined: 20 Apr 2011 19:19
Re: Correlations/Dependencies in LCA UQ
Hi Andreas,
I will separately email you the slide presentation that goes along with the paper (it was too large to upload). This presentation has a graph (not shown/discussed in the paper) that shows the potential effects of the correlation of CO2 and NO2 emissions in a CO2-e characterization step for a single unit process. We used data for burning anthracite coal from the U.S. LCI database. However, because there was no uncertainty information included in the dataset, we assumed four levels of relative uncertainty in the emissions data (as measured by the coefficient of variation). For each level, we then analyzed the resulting error in the computation of the standard deviation of the CO2-e that occurs from ignoring correlation, for a range of potential correlation coefficients in [-1,1]. The potential size of these errors (which are non-conservative for positive correlation, and larger when the relative uncertainty in the data is larger) suggests that correlations/dependencies may be important to account for in datasets and computational techniques.
-Mark
I will separately email you the slide presentation that goes along with the paper (it was too large to upload). This presentation has a graph (not shown/discussed in the paper) that shows the potential effects of the correlation of CO2 and NO2 emissions in a CO2-e characterization step for a single unit process. We used data for burning anthracite coal from the U.S. LCI database. However, because there was no uncertainty information included in the dataset, we assumed four levels of relative uncertainty in the emissions data (as measured by the coefficient of variation). For each level, we then analyzed the resulting error in the computation of the standard deviation of the CO2-e that occurs from ignoring correlation, for a range of potential correlation coefficients in [-1,1]. The potential size of these errors (which are non-conservative for positive correlation, and larger when the relative uncertainty in the data is larger) suggests that correlations/dependencies may be important to account for in datasets and computational techniques.
-Mark
Who is online
Users browsing this forum: No registered users and 1 guest