05.03.2014 Views

addressing uncertainty in oil and natural gas industry greenhouse

addressing uncertainty in oil and natural gas industry greenhouse

addressing uncertainty in oil and natural gas industry greenhouse

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

where<br />

r = the correlation coefficient between U X , U Y , (discussed further <strong>in</strong> Sections 4.2.2 <strong>and</strong> 4.6).<br />

However, the IPCC Good Practices guidance states, “Once the summation exceeds two terms <strong>and</strong><br />

the covariance occurs, the use of the Monte Carlo approach is preferable where data resources are<br />

available” (IPCC, 2001).<br />

b. Uncerta<strong>in</strong>ty Propagation for a Product (or Quotient)<br />

The equation for propagat<strong>in</strong>g uncerta<strong>in</strong>ties from the product or quotient of two or more measured<br />

<strong>and</strong> <strong>in</strong>dependent quantities is similar to Equation 4-4. However, <strong>in</strong> this case the relative<br />

uncerta<strong>in</strong>ties are used, as shown <strong>in</strong> Equation 4-6. When multiplied by 100, the result<strong>in</strong>g<br />

comb<strong>in</strong>ed <strong>uncerta<strong>in</strong>ty</strong> (U(Rel) XxYxN ) is expressed as a percentage.<br />

2 2<br />

⎛UX<br />

⎞ ⎛UY<br />

⎞ ⎛U<br />

N ⎞<br />

Urel ( )<br />

X× Y× ... × N= Urel ( )<br />

X÷ Y÷ ... ÷ N= ⎜ ⎟ + ⎜ ⎟ + ... + ⎜ ⎟<br />

⎝ X ⎠ ⎝ Y ⎠ ⎝ N ⎠<br />

2<br />

(Equation 4-6)<br />

Equation 4-7 is used to estimate the <strong>uncerta<strong>in</strong>ty</strong> of a product or quotient of two parameters (X <strong>and</strong><br />

Y) where the uncerta<strong>in</strong>ties are correlated <strong>and</strong> positive values. Here also, relative <strong>uncerta<strong>in</strong>ty</strong><br />

values are used <strong>in</strong> the equation <strong>and</strong> the result<strong>in</strong>g comb<strong>in</strong>ed <strong>uncerta<strong>in</strong>ty</strong> is on a relative basis.<br />

2 2<br />

⎛UX ⎞ ⎛UY ⎞ ⎛UX UY<br />

⎞<br />

Urel ( )<br />

Correlated X × Y<br />

= ⎜ ⎟ + ⎜ ⎟ + 2r⎜ × ⎟<br />

⎝ X ⎠ ⎝ Y ⎠ ⎝ X Y ⎠<br />

(Equation 4-7)<br />

c. Comb<strong>in</strong><strong>in</strong>g Uncerta<strong>in</strong>ties<br />

It may be necessary to comb<strong>in</strong>e multiple <strong>uncerta<strong>in</strong>ty</strong> parameters associated with a s<strong>in</strong>gle<br />

measured value, such as comb<strong>in</strong><strong>in</strong>g uncerta<strong>in</strong>ties for precision <strong>and</strong> bias. For <strong>uncerta<strong>in</strong>ty</strong><br />

parameters that are <strong>in</strong>dependent, the comb<strong>in</strong>ed <strong>uncerta<strong>in</strong>ty</strong> is calculated us<strong>in</strong>g the absolute<br />

uncerta<strong>in</strong>ties as shown <strong>in</strong> Equation 4-4. Similarly, for <strong>uncerta<strong>in</strong>ty</strong> parameters that are related to<br />

each other, Equation 4-5 applies.<br />

4.2.2 Correlation Coefficient<br />

The correlation coefficient, r, used <strong>in</strong> Equations 4-5 <strong>and</strong> 4-7, is a number between -1 <strong>and</strong> 1 that measures<br />

the l<strong>in</strong>ear relationship between the errors or uncerta<strong>in</strong>ties of two measured parameters. The value of r is<br />

zero when the parameters are <strong>in</strong>dependent. As stated previously, once the <strong>uncerta<strong>in</strong>ty</strong> propagation<br />

exceeds two terms <strong>and</strong> covariance occurs, the use of the Monte Carlo approach (described further <strong>in</strong><br />

Section 4.6.1) is preferable (IPCC, 2001). Additional details on calculat<strong>in</strong>g the correlation coefficient are<br />

provided <strong>in</strong> Section 4.6. A simplified explanation follows.<br />

Pilot Version, September 2009 4-7

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!