EC Pro: Why is cost avoidance higher in Pro 11 than in Pro 09?
Posted by Matthew Heinz, Last modified by Joel Brickell on 14 July 2015 05:12 PM
Problem: The total cost avoidance in Pro 11 is slightly higher than it was in Pro 09. Why is it higher?
Solution: A correction made to the Pro 11 cost avoidance processor provides more accurate and appropriate cost avoidance calculations for meters that (1) have a baseline bill that straddles the baseline start date and (2) are weather-sensitive in that month. This issue is documented as ticket #34 on the Pro 11 New Features list.
Here is a step-by-step analysis of an actual case:
1. Opened database in Pro 09 SP1 (811)
2. Re-calculated cost avoidance for all meters, all bills
3. Ran reports: CAP01C and CAP01E
4. Made backup
5. Opened backup in Pro 11 (622)
6. Upgraded DB from Pro 09 to Pro 11
7. Ran report CAP01C
8. Verified CA is exactly the same before re-processing in Pro 11 (total CA =$5,793,141, 16.88%)
9. Re-calculated cost avoidance for all meters, all bills
10. Ran report. Now total CA per CAP01C is $5,886,959 (17.11%), an increase of $93,818 which represents a 1.62% increase over Pro 09. Most of the change (94% of the change) is in electricity, the other 6% is in irrigation. Gas, propane, sewer and water are unchanged.
11. Which meters have changed the most? By merging data from both CAP01E reports in Excel, we get this list of meters with most significant changes. Of 281 meters, 269 are exactly the same; only 12 have changed. About 50% of the entire change is in one meter, E5876764440 (High School Main), so that will be primary point of investigation.
12. Which months have changed for this meter? Comparison of CAP10 for Pro 09 vs Pro 11 shows that all bills have the exact same cost avoidance except for bills that contain July days.
13. Does the base year have bills that straddle the baseline start date? Yes. The baseline starts on July 1, 2006 and that is a straddle bill.
14. Is the baseline start month weather sensitive? Yes
15. Since it is both a straddle AND weather sensitive, that means the baseline touches TWO July bills, July 2006 and July 2007. Is either on a statistical outlier? Yes, July 2007.
16. You can see above that July 2007 is much lower than July 2006, and also much lower than expected by the statistical model. This implies that the energy management program was underway, at least partially, in July 2007, resulting in a significant summer energy reduction. Using July 2007 as a baseline month is therefore inappropriate because it already reflects energy management initiatives.
17. Is the July baseline inappropriately low in Pro 09? The Pro 09 CAP06 report shows that 28 days of the July baseline are incorrect. The overall baseline of 232,797 is too low and does agree with the actual 2006 electric bills (see #13 above).
18. This baseline creation bug was corrected in Pro 11 on ticket #34. It is listed in the Pro 11 release notes:
19. The correction in Pro 11 causes the July baseline to be properly populated. The Total Use of 335,921 is reasonable as compared with the actual July and August 2006 electric bills. The July 2006 actual bill was 347,796, but it gets reduced slightly because July 2006 has two days represented in the Aug 06 bill, which is lower and pulls down the baseline slightly.
20. A quick review of the other meters with most significant cost avoidance changes (see #11 above) shows the same situation to be present. Heres the meter with the second greatest change; its situation is identical to the prior meter.
21. In conclusion, it has been shown that the correction made to the Pro 11 cost avoidance processor provides more accurate and appropriate cost avoidance calculations for meters that (1) have a baseline bill that straddles the baseline start date, (2) are weather-sensitive in that month, and (3) further exacerbated when of the two bills that touch the baseline start month is a statistical outlier.
Created By Steve Heinz, 07/19/2011