Approaches to Data Acquisition

• The LCA depends upon data acquisition

• Qualitative vs. Quantitative– While some quantitative analysis is appropriate,

inappropriate quantification can obscure issues

• LCA Information Systems– Non-prescriptive – global solution may depend on

locally “non-green” choices– Must describe both data and uncertainty

Treatment of uncertainty

• The nature of uncertain parameters– Random– Unknowable

• We can use descriptive statistics to characterize uncertain parameters

• Or we can simulate uncertain systems– Monte Carlo sampling methods

Monte Carlo Simulation

• Uncertain independent variables are described by probability density functions

• A cumulative distribution function is calculated

Monte Carlo Simulation

• The y axis is sampled randomly, and corresponding x axis values are selected

0

0.2

0.4

0.6

0.8

1

1.2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Series1

Monte Carlo Simulation

• Multiple independent variables can be sampled for each model run

• N such calculations produce N model results - itself a distribution, from which descriptive statistics and probabilities can be determined

Using MC in Data Acquisition

• Since many of the data values in our industrial systems are uncertain, we can use Monte Carlo sampling to gain estimates of these values– e.g. we can estimate the average amount of lead

used in an iBook by combining the individual components’ probability distributions

– Then running the calculation many times results in a composite distribution of lead in the iBook