Department of Energy

Advanced Manufacturing Office

EnPI V5.0 Tool Algorithm

Updated Feb 5, 2018

1
Contents
Definition of Symbols ................................................................................................................................................4
Facility Level Calculations .........................................................................................................................................5
Calculation Methods when Actual Values are used ...............................................................................................5
Energy Intensity (MMBtu/Unit of Production or MMBtu/sq ft) ........................................................................5
Cumulative Improvement (Total Change) in Energy Intensity from Baseline Year (%) ...................................5
Annual Improvement (Annual Change) in Energy Intensity for Current Year (%) ...........................................5
Annual Savings (Total Energy Savings) since Baseline Year (MMBtu/year) ...................................................5
New Energy Savings for Current Year (MMBtu/year) ......................................................................................5
Estimated Annual Cost Savings ($/year) ............................................................................................................5
Calculation Method when Normalized Values are used.........................................................................................7
SEnPI (unitless) ..................................................................................................................................................7
Cumulative Improvement (Total Change) in Energy Intensity from Baseline Year (%) ...................................7
Annual Improvement (Annual Change) in Energy Intensity for Current Year (%) ...........................................8
Annual Savings (Total Energy Savings) since Baseline Year (MMBtu/year) ...................................................8
New Energy Savings for Current Year (MMBtu/year) ......................................................................................9
Estimated Annual Cost Savings ($/year) ............................................................................................................9
SEP Calculations ..................................................................................................................................................... 10
SEP CUSUM (MMBtu) .................................................................................................................................. 11
Corporate Level Calculations .................................................................................................................................. 11
Current Year Total Primary Energy Use (MMBtu/year)................................................................................. 12
Adjustment for Baseline Primary Energy (MMBtu/year) ............................................................................... 12
Adjusted Baseline Primary Energy (MMBtu/year) ......................................................................................... 12
Cumulative Improvement (Total Change) in Energy Intensity from Baseline Year (%) ................................ 12
Annual Improvement (Annual Change) in Energy Intensity for Current Year (%) ........................................ 12
Annual Savings (Total Energy Savings) since Baseline Year (MMBtu/year) ................................................ 12
New Energy Savings for Current Year (MMBtu/year) ................................................................................... 12
Least Squares Regression Analysis ......................................................................................................................... 13

2
3
Definition of Symbols
The following symbols and abbreviations are used in the Facility Level Calculations and Corporate Level
Calculations sections of the document.

AI Annual Improvement (Annual change) in Energy Intensity for Current Year

(%)

AEC 5, 10, 15 Energy Consumption for 5%, 10%, 15% Improvement Projections

BY Baseline Year

CI Cumulative Improvement (Total Change) in Energy Intensity from Baseline

Year (%)

CS Energy Cost Savings

CY Current Year

Corp Corporate level total

�
𝑬𝑬𝑬𝑬 Modeled Source (Primary) Energy Consumption (MMBtu)

EC Actual Source (Primary) Energy Consumption (MMBtu)

EI Energy Intensity or Energy Normalized for Relevant Factors (MMBtu/ Unit of

Production or MMBtu/sq ft)

Energy Cost p Energy cost for a given time interval (e.g. day, week or month)

Energy Savings i The energy savings for a given time interval (e.g. day, week or month)

𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐄𝐓𝐓𝐓𝐓𝐓𝐓 SEP Trailing Twelve Month Energy Performance Indicator

ES TTM SEP Trailing Twelve Month Energy Savings

I Interval of time in which the data is entered (e.g. daily, weekly, monthly, etc.)

MY Model Year

N Number of plants within the company

PY Previous Year

SEnPI Superior Energy Performance Indicator

Unit Cost Unit energy cost

4
Facility Level Calculations
The following section describes the calculation methods used for the facility level calculations.

Calculation Methods when Actual Values are used

When “Use Actual” is selected as the calculation method in the EnPI tool, the following calculations are used.

Energy Intensity (MMBtu/Unit of Production or MMBtu/sq ft)

𝐸𝐸𝐸𝐸
𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 =
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃

𝐸𝐸𝐸𝐸
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 =
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹

Cumulative Improvement (Total Change) in Energy Intensity from Baseline Year (%)
When using Production Energy Intensity:
(𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 − 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 )
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = �𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐸𝐸𝐸𝐸
𝐵𝐵𝐵𝐵

When using Building Energy Intensity:

(𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐵𝐵𝑌𝑌 − 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 )
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸
𝐵𝐵𝐵𝐵

Annual Improvement (Annual Change) in Energy Intensity for Current Year (%)
𝐴𝐴𝐴𝐴𝐵𝐵𝐵𝐵 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃

Annual Savings (Total Energy Savings) since Baseline Year (MMBtu/year)

The following equation is used to calculate the savings for each energy type (e.g. electric, natural gas, etc.) and for
the total of all the energy sources entered into the tool.

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 − 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶

New Energy Savings for Current Year (MMBtu/year)

𝑁𝑁𝑁𝑁𝑁𝑁 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 − 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃

Estimated Annual Cost Savings ($/year)

First, the unit cost for each energy type (e.g. electric, natural gas, etc.) for each time interval is calculated by
dividing the energy cost by the energy consumption.

𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖
𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖 =
𝐸𝐸𝐸𝐸𝑖𝑖

Next, the daily, weekly or month energy savings is calculated for each energy type by subtracting the current year
consumption for the time interval from the baseline year consumption for the same time interval (e.g. September
2010 (baseline) electric – September 2013 (current year) electric].
5
 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 = 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 𝑖𝑖 − 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 𝑖𝑖

The unit cost for each energy type and time interval is then multiplied by the energy savings for each energy type
and time interval.

𝐶𝐶𝐶𝐶𝑖𝑖 = 𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖 × 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑖𝑖

Finally, the cost savings for each energy type (e.g. electric, natural gas, etc.) and time period is summed to
determine the annual cost savings.
𝑁𝑁 𝑁𝑁 𝑁𝑁

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = � 𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑖𝑖 + � 𝐶𝐶𝐶𝐶𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺 𝑖𝑖 + � 𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑋𝑋 𝑖𝑖

𝑖𝑖=1 𝑖𝑖=1 𝑖𝑖=1

6
Calculation Method when Normalized Values are used
When “Regression Analysis” is selected as the calculation method in the EnPI tool, the following calculations are
used.

SEnPI (unitless)
Forecasting (Model year = baseline year)

Calculation for the model year:

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀 = 1

Calculation for all years after the model year:

� 𝑀𝑀𝑀𝑀
𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 𝐸𝐸𝐸𝐸
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = ×
� 𝐶𝐶𝐶𝐶
𝐸𝐸𝐸𝐸𝑀𝑀𝑀𝑀 𝐸𝐸𝐸𝐸

Chaining (Model year does not = baseline year, and does not = last reporting year)

Calculation for all years prior to the model year, including the baseline year:

� 𝐶𝐶𝐶𝐶
𝐸𝐸𝐸𝐸
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 =
𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶

Calculation for the model year:

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀 = 1

Calculation for all years after the model year:

� 𝐵𝐵𝐵𝐵
𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 𝐸𝐸𝐸𝐸
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = ×
� 𝐶𝐶𝐶𝐶
𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸

Backcasting (Model year = last reporting year)

Calculation for all years prior to the model year, including the baseline year:

� 𝐶𝐶𝐶𝐶
𝐸𝐸𝐸𝐸
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 =
𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶

Calculation for the model year:

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀 = 1

Cumulative Improvement (Total Change) in Energy Intensity from Baseline Year (%)
Forecasting (Model year = baseline year)

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = (1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 )×100

Chaining (Model year does not = baseline year, and does not = last reporting year)

Calculation for the model year, and all years prior to the model year:
7
 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = {[(1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 ) − (1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 )] × 100} + 𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃

Calculation for all years after the model year:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = (1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 ) × 100

Backcasting (Model year = last reporting year)

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = {[(1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 ) − (1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 )] × 100} + 𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃

Annual Improvement (Annual Change) in Energy Intensity for Current Year (%)
Forecasting (Model year = baseline year)

𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃

Chaining (Model year does not = baseline year, and does not = last reporting year)

Calculation for the model year, and all years prior to the model year:

𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶 = [(1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 ) − (1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 )] × 100

Calculation for all years after the model year:

𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃

Backcasting (Model year = last reporting year)

𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶 = [(1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 ) − (1 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 )] × 100

Annual Savings (Total Energy Savings) since Baseline Year (MMBtu/year)

The following equations are used to calculate the savings for each energy type (e.g. electric, natural gas, etc.) and
for the total of all the energy sources entered into the tool.

Forecasting (Model year = baseline year)

� 𝐶𝐶𝐶𝐶 − 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐸𝐸𝐸𝐸

Chaining (Model year does not = baseline year, and does not = last reporting year)

Calculation for the model year, and all years prior to the model year:

� 𝑃𝑃𝑃𝑃 � − (𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 − 𝐸𝐸𝐸𝐸

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 + ��𝐸𝐸𝐸𝐸𝑃𝑃𝑃𝑃 − 𝐸𝐸𝐸𝐸 � 𝐶𝐶𝐶𝐶 )�

Calculation for all years after the model year:

� 𝐶𝐶𝐶𝐶 − 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐸𝐸𝐸𝐸

Backcasting (Model year = last reporting year)

� 𝑃𝑃𝑃𝑃 � − (𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 − 𝐸𝐸𝐸𝐸

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 + ��𝐸𝐸𝐸𝐸𝑃𝑃𝑃𝑃 − 𝐸𝐸𝐸𝐸 � 𝐶𝐶𝐶𝐶 )�

8
New Energy Savings for Current Year (MMBtu/year)
𝑁𝑁𝑁𝑁𝑁𝑁 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 − 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃

Estimated Annual Cost Savings ($/year)

First, the unit cost for each energy type (e.g. electric, natural gas, etc.) for each time interval is calculated by
dividing the energy cost by the energy consumption.

𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖
𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖 =
𝐸𝐸𝐸𝐸𝑖𝑖

Next, the daily, weekly or month energy savings is calculated for each energy type and time interval using the
equations below. In the equations “I” denotes the current time interval (e.g. day, week, or month). The subscripts
“i-1” indicates the previous time interval. For example, if “I” is referencing September 2013, “i-1” indicates
August 2013.

Forecasting (Model year = baseline year)

� 𝑖𝑖 − 𝐸𝐸𝐸𝐸𝑖𝑖
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 = 𝐸𝐸𝐸𝐸

Chaining (Model year does not = baseline year, and does not = last reporting year)

Calculation for the model year and all years prior to the model year:

� 𝑖𝑖−1 � − �𝐸𝐸𝐸𝐸𝑖𝑖 − 𝐸𝐸𝐸𝐸

𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖−1 + ��𝐸𝐸𝐸𝐸𝑖𝑖−1 − 𝐸𝐸𝐸𝐸 � 𝑖𝑖 ��

Calculation for all years after the model year:

� 𝑖𝑖 − 𝐸𝐸𝐸𝐸𝑖𝑖
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 = 𝐸𝐸𝐸𝐸

Backcasting (Model year = last reporting year)

� 𝑖𝑖−1 � − �𝐸𝐸𝐸𝐸𝑖𝑖 − 𝐸𝐸𝐸𝐸

𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖−1 + ��𝐸𝐸𝐸𝐸𝑖𝑖−1 − 𝐸𝐸𝐸𝐸 � 𝑖𝑖 ��

The unit cost for each energy type and time interval is then multiplied by the energy savings for each energy type
and time interval.

𝐶𝐶𝐶𝐶𝑖𝑖 = 𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖 × 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑖𝑖

Finally, the cost savings for each energy type (e.g. electric, natural gas, etc.) and time period is summed to
determine the annual cost savings.
𝑁𝑁 𝑁𝑁 𝑁𝑁

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = � 𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑖𝑖 + � 𝐶𝐶𝐶𝐶𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺 𝑖𝑖 + � 𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑋𝑋 𝑖𝑖

𝑖𝑖=1 𝑖𝑖=1 𝑖𝑖=1

9
SEP Calculations
The following outputs are displayed when “Regression Analysis” is selected as the calculation method in the EnPI
tool. These calculations were developed for the SEP Program but may be valuable to general users and Better
Plants Participants as well.

SEP Trailing Twelve Month Energy Performance Indicator

In the following calculations trailing 12 months indicates the sum of the current plus previous 11 months values.
For example, if the value is calculated for December 2013, the trailing 12 month sum would be the sum of the Jan
2013 through Dec 2013 values.

Forecasting (Model year = baseline year)

∑𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 EC𝐶𝐶𝐶𝐶

𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑇𝑇𝑇𝑇𝑇𝑇 =
� 𝑪𝑪𝑪𝑪
∑𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 EC

NOTE: Energy consumption above is the total source consumption for all energy sources.

Chaining (Model year does not = baseline year, and does not = last reporting year)

� 𝐵𝐵𝐵𝐵 ∑𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶

𝐸𝐸𝐸𝐸
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑇𝑇𝑇𝑇𝑇𝑇 = ×
� 𝐶𝐶𝐶𝐶
𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 ∑𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 𝐸𝐸𝐸𝐸

NOTE: (1) The energy consumption in above equations refers to total source consumption for all fuel
sources. (2) Trailing 12 months refers to 12 month period which includes month 1 after the mid-term
model year.

This value is not calculated when backcasting is selected.

SEP Trailing Twelve Months Energy Savings

Forecasting (Model year = baseline year)

𝐸𝐸𝐸𝐸𝑇𝑇𝑇𝑇𝑇𝑇 = � �
𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 − � 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠

NOTE: monthly energy consumption above is the total source consumption for all energy sources.

Chaining (Model year does not = baseline year, and does not = last reporting year)

� 𝐵𝐵𝐵𝐵 � + (�
𝐸𝐸𝐸𝐸𝑇𝑇𝑇𝑇𝑇𝑇 = �𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 − 𝐸𝐸𝐸𝐸 � 𝐶𝐶𝐶𝐶 − �
𝐸𝐸𝐸𝐸 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 )
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑛𝑛𝑛𝑛 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠

NOTE: (1) Energy consumption in above equations refers to total source consumption for all fuel
sources. (2) Trailing 12 months refers to 12 month period which includes month 1 after the mid-term
model year.

This value is not calculated when backcasting is selected.

10
Energy Consumption for 5%, 10%, 15% Improvement Projections
The following values indicate the amount of energy that should be used in order to save 5%, 10%, or 15%.

Forecasting (Model year = baseline year)

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇
𝐴𝐴𝐴𝐴𝐴𝐴5,10,15 = �1 − � �� × � 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀ℎ𝑙𝑙𝑙𝑙 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶
100 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀ℎ𝑠𝑠

Chaining (Model year does not = baseline year, and does not = last reporting year)

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵
𝐴𝐴𝐴𝐴𝐴𝐴5,10,15 = (1 − ) × � 𝐶𝐶𝐶𝐶
× ∑𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑠𝑠 𝐸𝐸𝐸𝐸
100 � 𝐵𝐵𝑌𝑌
𝐸𝐸𝐸𝐸

This value is not calculated when backcasting is selected.

5%, 10%, 15% projected energy savings and actual energy savings
Forecasting (Model year = baseline year)

𝐸𝐸𝐸𝐸5,10,15 = � � 𝐶𝐶𝐶𝐶 − 𝐴𝐴𝐴𝐴𝐴𝐴5,10,15

𝐸𝐸𝐸𝐸
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀ℎ𝑠𝑠

Chaining (Model year does not = baseline year, and does not = last reporting year)

� 𝐵𝐵𝐵𝐵 � + (�
𝐸𝐸𝐸𝐸5,10,15 = �𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 − 𝐸𝐸𝐸𝐸 � 𝐶𝐶𝐶𝐶 ) − 𝐴𝐴𝐴𝐴𝐴𝐴5,10,15
𝐸𝐸𝐸𝐸
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 12 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀ℎ𝑠𝑠

These values are not calculated when backcasting is selected.

SEP CUSUM (MMBtu)

For all regression methods, the following equation is used to calculate CUSUM.

� 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 � + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = �𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 − 𝐸𝐸𝐸𝐸

In the equation above, “Interval” represents the time interval entered by the user (e.g. month, week, day). For the
first data point after the model year when chaining is selected as the regression method the “previous interval”
refers to the last data point before the model year.

Corporate Level Calculations

This section lists the calculations for the corporate roll up. These calculations are appropriate whether all
participating plants selected “use actual”, “regression analysis”, or a blend of the two methods.

11
Current Year Total Primary Energy Use (MMBtu/year)
The following equation is used to calculate the savings for each energy type (e.g. electric, natural gas, etc.) and for
the total of all the energy sources entered into the tool.
𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐸𝐸𝐸𝐸 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 1 + 𝐸𝐸𝐸𝐸 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 2 + ⋯ + 𝐸𝐸𝐸𝐸 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑁𝑁

Adjustment for Baseline Primary Energy (MMBtu/year)

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝑓𝑓𝑓𝑓 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

Adjusted Baseline Primary Energy (MMBtu/year)

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝑓𝑓𝑓𝑓 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

Cumulative Improvement (Total Change) in Energy Intensity from Baseline Year (%)
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶
� 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 1 × 𝐶𝐶𝐼𝐼𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 1 � + �𝐸𝐸𝐸𝐸
��𝐸𝐸𝐸𝐸 � 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 2 × 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 2 � + ⋯ + �𝐸𝐸𝐸𝐸 � 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑁𝑁 × 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑁𝑁 ��
=
� 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 1 + 𝐸𝐸𝐸𝐸
𝐸𝐸𝐸𝐸 � 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 2 + ⋯ + 𝐸𝐸𝐸𝐸 � 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑁𝑁

*When a plant that selects “use actual” as the calculation method is included in the roll-up, 𝐸𝐸𝐸𝐸𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃
� 𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 .
should be used in place of 𝐸𝐸𝐸𝐸

Annual Improvement (Annual Change) in Energy Intensity for Current Year (%)
𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

Annual Savings (Total Energy Savings) since Baseline Year (MMBtu/year)

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

= 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 1 + 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 2 + ⋯
+ 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑁𝑁
New Energy Savings for Current Year (MMBtu/year)
𝑁𝑁𝑁𝑁𝑁𝑁 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

12
Least Squares Regression Analysisi
This section outlines the equations that are used for the regression analysis in the EnPI V3.0 tool. The tool uses a
Microsoft Excel add-in to perform the regression analysis. The add-in uses the equations listed below to predict
the energy consumption based on the independent variables entered by the user.

Regression analysis is a statistical method for predicting the behavior of a dependent variable based on the
independent variables. The table below defines the symbols that will be used in the regression analysis equations
outlined in this section.

Equation Symbols and Definitions

Symbol Definition
Dependent variable predicted by the regression model (i.e.
y
predicted energy use)
Measured dependent variable (i.e. measured energy use)
y*
Number of independent variables and coefficients
p
The ith independent variable from total set of p variables (i.e.
x i (i=1, 2… p)
production, HDD, CDD, etc.)
The ith coefficient corresponding to x i
b i (i=1, 2… p)
Intercept or constant
b0
Total number of parameters including intercept
k=p+1
n Number of observations
i=1, 2… p Independent variables’ index
j=1, 2..n Data points index
R2 Coefficient of determination
r Residual (or error, or deviation)
SS E Residual (error) sum of squares (or regression sum of
squares)
SS R Regression sum of squares
SS T Total sum of squares

For the EnPI V3.0 tool, the dependent variable is the energy consumption by the facility. The independent
variables can be production, cooling degree days (CDD), heating degree days (HDD), etc. If the user selects more
than one independent variable, a multivariable linear regression equation is needed to predict the dependent
variable or energy consumption at the facility.

Regression analysis determines the formula that can be used to predict the dependent variable based on the
independent variables. The general formula for a multiple linear model is:

𝑦𝑦 = 𝑏𝑏0 + 𝑏𝑏1 𝑥𝑥1 + 𝑏𝑏2 𝑥𝑥2 +. . . 𝑏𝑏𝑝𝑝 𝑥𝑥𝑝𝑝

In this formula, y is the predicted dependent variable. The measured dependent variable is depicted by the y*. The
difference between the predicted and measured dependent variable is called the residual (also known as error or
deviation).

𝑟𝑟𝑗𝑗 = 𝑦𝑦𝑗𝑗∗ − 𝑦𝑦𝑗𝑗

13
The goal of regression analysis is to determine the coefficients (b 1, 2… i ) that result in a minimized error sum of
squares. The error sum of squares (SSE) is calculated by:
𝑛𝑛 2
𝑆𝑆𝑆𝑆𝐸𝐸 = � �𝑟𝑟𝑗𝑗 − 𝑟𝑟𝑎𝑎𝑎𝑎𝑎𝑎 �
𝑗𝑗

Microsoft Excel calculates the error sum of squares (SS E ) for each combination of coefficients and determines the
combination of coefficients that minimize the SS E . The coefficients that result in the lowest error sum of squares
are entered into the regression model to produce an equation that can be used to estimate the dependent variable
given the independent variable(s).

Along with determining the model that best predicts the relationship between the independent and dependent
variables, Microsoft Excel also calculates the coefficient of determination. The coefficient of determination (R2) is
a measure of how well future outcomes are likely to be predicted by the model. A regression model is a good fit
for the data if the R2 value is close to 1. In order to calculate the R2 value, the regression sum of squares (SS R ) and
the total sum of squares (SS T ) first need to be defined. The regression sum of squares is defined as:
𝑛𝑛 2
𝑆𝑆𝑆𝑆𝑅𝑅 = � �𝑦𝑦𝑗𝑗 − 𝑦𝑦 ∗𝑎𝑎𝑎𝑎𝑎𝑎 �
𝑗𝑗

And the total sum of squares is calculated using the following equation:
2
𝑆𝑆𝑆𝑆𝑇𝑇 = ∑𝑛𝑛𝑗𝑗�𝑦𝑦 ∗𝑗𝑗 − 𝑦𝑦 ∗𝑎𝑎𝑎𝑎𝑎𝑎 � , where 𝑦𝑦 ∗𝑎𝑎𝑎𝑎𝑎𝑎 = �∑𝑛𝑛𝑗𝑗 𝑦𝑦 ∗𝑗𝑗 �/𝑛𝑛

The total sum of squares can also be defined:

𝑆𝑆𝑆𝑆𝑇𝑇 = 𝑆𝑆𝑆𝑆𝐸𝐸 + 𝑆𝑆𝑆𝑆𝑅𝑅

Now that the residual, regression, and total sum of squares have been defined, the coefficient of determination can
be defined as:

𝑆𝑆𝑆𝑆𝑅𝑅
𝑅𝑅 2 = = 1 − 𝑆𝑆𝑆𝑆𝐸𝐸 /𝑆𝑆𝑆𝑆𝑇𝑇
𝑆𝑆𝑆𝑆𝑇𝑇

i
Source: Cameron, Colin. "Excel 2007: Multiple Regression." UC-Davis Economics. University of California Davis
Department of Economics, n.d. Web. 3 Oct 2012. <http://cameron.econ.ucdavis.edu/excel/ex61multipleregression.html>.

14