Energy Intensity Baseline of the Northwest Food Processing Industry

Energy Intensity Baseline of
the Northwest Food
Processing Industry
Prepared by:
With support from:

Northwest Food Processors
Education and Research Institute
8338 NE Alderwood Road, Suite 160
Portland, OR 97220
Telephone: +1 (503) 327-2200
Fax: +1 (503) 327-2201
E-mail: foodipc@foodipc.org
Web: http://www.foodipc.org
Northwest Food Processors Association
8338 NE Alderwood Road, Suite 160
Portland, OR 97220
Telephone: +1 (503) 327-2200
Fax: +1 (503) 327-2201
E-mail: info@nwfpa.org
Web: http://www.nwfpa.org
© 2010 Northwest Food Processors Education and Research Institute
© 2010 Northwest Food Processors Association
All rights reserved. It is permitted to download this electronic file, to make a copy and to print out the
content for the purpose of preparing reference copy documents only. You may not copy or “mirror” the file,
or any part of it, for any other purpose without permission in writing from the publishers.
2 Energy Intensity Baseline of the Northwest Food Processing Industry

Table of Contents
1 Program Overview .................................................................................... 2
2 Project Overview ...................................................................................... 2
3 Proj
ect Methods ....................................................................................... 3
3.1 Data Collection and Data Security ..................................................... 3
3.2 Selection of Units ............................................................................. 4
3.3 Time Period of Evaluation & History .................................................. 4
3.4 Sub-Cluster Characterization ............................................................ 4
3.5 Use of Aggregate, Mean and Median Energy Intensity ....................... 6
3.6 IAC Data Illustration .......................................................................... 7
3.7 Estimating Data Error ....................................................................... 9
4 Proj
ect Results ....................................................................................... 11
4.1 Data ............................................................................................... 11
4.2 Comparison with IAC Historic Data .................................................. 12
4.3 Energy Intensity by Sub-Cluster ...................................................... 13
4.4 Sources of Energy Consumption ..................................................... 14
5 Recommendations .................................................................................. 15
5.1 Data Security .................................................................................. 15
5.2 Encouraging Participation ............................................................... 15
List of Tables
Table 1: Major Food Processing Sub-Clusters ................................................ 5
Table 2: Example NAICS+3 Assignments for Cheese Manufacturing ............... 6
Table 3: Results of Baseline Data ................................................................ 11
Table 4: Comparison of 2008 Baseline Data to Historic IAC Data .................. 12
Table 5: Comparison of 2008 Baseline Data to Historic IAC Data .................. 13
Table 6: Sample Breakdown and Comparison by Sub-Cluster ....................... 13
List of Figures
Figure 1: Example of Median Energy Intensity ................................................ 7
Figure 2: Histogram of Historic Northwest Food Processing Energy Intensity
Data ........................................................................................................ 8
Figure 3: Average Estimated Sampling Error vs. Total Population Size ......... 10
Figure 4: Histogram of 2008 Northwest Food Processing Energy Intensity
Data ...................................................................................................... 12
Figure
5: Sources of Energy in the Northwest Food Processing Industry ....... 14
2 Energy Intensity Baseline of the Northwest Food Processing Industry

Appendices
Appendix A: Energy Intensity Data Collection Process .................................. 17
Appendix B: Visual Illustration of Mean, Aggregate, and Median Energy
Intensities .............................................................................................. 18
Appendix C: IAC Northwest Food Processing Data Set ................................. 19
Appendix D: Alternate Method of Estimating Sample Error ............................ 23
Appendix E: Detailed Project Energy Intensity Data ...................................... 27
Appendix F: Sub-Cluster Characterization .................................................... 28
Appendix G: Glossary .................................................................................. 30
Appendix H: References .............................................................................. 31
Energy Intensity Baseline of the Northwest Food Processing Industry 3

1 Program Overview
The food processing industry has been a cornerstone of the Northwest regional economy for
more than 100 years. It employs, directly and indirectly, more than 280,000 people, has a current
total regional economic impact of $42.5 billion and a payroll of $2.4 billion. As the only
manufacturing sector in Oregon to enjoy positive job growth in 2008, the importance of food
processing to the rural Northwest economy cannot be overstated, particularly in these challenging
economic times.
One of the keys to the food processing industry’s ability to remain competitive will be its ability to
use energy efficiently. The Northwest Power and Conservation Council’s 6 th Power Plan
indicates that the food processing industry in the Northwest accounts for 12 percent of the overall
non-direct service industry electricity demand and is the second largest sector of electricity
consumption behind pulp and paper. Further, U.S. Department of Energy’s Energy Information
Administration data shows food processing to be the fifth largest sector of fossil fuel consumption
in the U.S. This provides a sense of the industry’s reliance on energy and the importance that
improvements in energy efficiency present.
In response to the needs and priorities of its members, the Northwest Food Processors
Association (NWFPA) has focused on energy efficiency programs for over five years, partnering
with the Northwest Energy Efficiency Alliance (NEEA). NWFPA and NEEA recently convened two
high-level regional events, an Executive Visioning Forum at which food processing executives
identified the vision and goal for energy within the industry and an Energy Efficiency Roadmap
Workshop, at which the food processing industry, electric and gas utilities and government
participants identified and categorized ideas to achieve that vision and goal. These ideas are
driving the development of a Northwest Food Processors Energy Efficiency Roadmap. In
February 2009, at the U.S. Department of Energy’s (U.S. DOE) Northwest Industrial Energy
Efficiency Summit, NWFPA and U.S.DOE signed an historic memorandum of understanding to
memorialize NWFPA’s ambitious goal: a 25 percent reduction in member-wide energy intensity
over the next 10 years.
Thus, Northwest food processors became the first U.S. industrial segment to voluntarily commit
to such a sweeping energy efficiency goal. The NWFPA Energy Efficiency Roadmap will drive the
industry’s energy efficiency implementation efforts over the next decade. A four-minute video on
this leading-edge energy intensity efficiency initiative can be viewed at
http://www.youtube.com/watch?v=esMEwC7jzFg.
2 Project Overview
As a fundamental step towards reducing the Northwest food processing industry’s energy
intensity, it is important to establish a baseline energy intensity from which to gage progress. This
project is the first attempt of its kind to measure and characterize the energy intensity of an entire
regional industrial sector. Many key learnings have emerged as a result of this process that are
very applicable to other industrial sectors.
The primary objective of this project is to measure and best characterize the energy intensity of
the Northwest food processing industrial sector. Real world energy intensity data coupled with
comparative analysis of historical data will provide not only a baseline for energy use in the
sector, but also a characterization of trends that will help shape the sector’s energy efficiency
decision making.
Specific objectives of this project include:
Analyze Northwest food processing industrial facility’s historical energy intensity data estimates
available from the Industrial Assessment Centers (IACs) database. This analyzed data will:
Ensure the collected project data is of the correct magnitude.
2 Energy Intensity Baseline of the Northwest Food Processing Industry

Help understand which sub-sectors (frozen foods, dehydrated products, canned products, etc.)
consume the most energy for a given unit of finished product.
Help understand the nature of the distribution of individual food processing facility energy
intensity values. This in turn, shapes how many actual project plants will need to be sampled to
best characterize the entire sector.
Collect actual individual food processing industrial facility data in calendar year time intervals for
a period of 2-3 years. Because the collected data represents potentially sensitive manufacturing
information that could compromise competitive advantage, the process for collection and
safeguarding the data is nearly as important as the data itself.
Conduct a final analysis of the project data and compare with historical data. Based on the
analysis, set a baseline energy intensity value that best represents the entire industrial sector.
Document the entire baseline energy intensity collection and characterization process so that it
can be easily repeated to measure the sector’s progress towards meeting its reduction goals.
3 Project Methods
3.1 Data Collection and Data Security
The data collection process for each participating company is detailed in Appendix A. Because
energy use and production data for each site is confidential, the initial request for a company to
participate in the study requires top executive dialogue between the NWFPA staff and the client
company. This is primarily to ensure there is top-down support within the client company for data
collection and to assure the client that all data will be handled with the utmost of security.
Once top executive permission was obtained from the company to participate in the project,
NWFPA entered into a binding Non-Disclosure Agreement (NDA) to provide written proof to the
client company of NWFPAs data security commitment. The NDA provided that NWFPA would only
release information in the aggregate and that individual company data would not be discernable.
During the initial contact, the client company executive typically provided the name and contact
information for the responsible person on the client’s staff to assist NWFPA research staff with
gathering the data.
Actual data collection was normally conducted by the client company and included filling out a
formatted blank data collection spreadsheet file with production and energy use data fields.
To ensure confidentiality, the completed data collection spreadsheets were stripped of the
individual plant identification and assigned a 3-digit identification code. A separate file of
company name, plant location, and the corresponding 3-digit identifier were maintained on a
separate written file stored in a secure location.
Consideration was given to engaging an independent 3 rd party firm capable of handling sensitive
data (CPA, etc.) to receive the completed raw data sheets, assign a random 3 digit ID, and strip
off any specific company identifying data. This approach was not taken due to available time and
resources, but could be considered for future data collection efforts.
Energy Intensity Baseline of the Northwest Food Processing Industry 3

3.2 Selection of Units
Energy Intensity is measured as the quantity of energy required per unit output or activity, such
that using less energy to produce a product reduces the intensity1. This measure can be
expressed mathematically as shown in Equation 1 below which is consistent with the approach
used by the U.S. Department of Energy, Energy Efficiency and Renewable Energy (EERE).2
Equation 1
I e,i =
E t,i
A t, i
Where:
I e,i : Energy Intensity for a given facility, or industry “i”
E t,i : Total delivered energy used at a given facility, or industry “i”
A t,i : Total activity, or total output, from facility, or industry “i”
The units of energy chosen were British Thermal Units (BTU) because the units are easily
understood in the industry by most plant and corporate personnel. All forms of consumed energy
were converted into BTUs.
Activity can be measured as any useful output from the plant. Because of the large variation in
types of product, a mass value was used. As with BTUs, for ease of understanding, the mass unit
chosen was pounds. The resulting units of energy intensity are thus BTUs per pound of finished
product expressed as “BTU/Lb.”
3.3 Time Period of Evaluation & History
Energy and production data from each plant were collected in one-year time periods coinciding
with the calendar year January 1 to December 31. Evaluation of one-year periods mitigated the
seasonal variation of energy consumption and production volumes. The food processing industry
has relatively large variations in energy consumption due to outside ambient temperature and
production volumes, which are timed with harvests and/or affected by seasonal consumer
demand. Additionally, this annual approach was relatively easy for companies’ data collection
efforts.
Data collection normally spanned three years at each sampled plant in order to track trends or
identify any significant changes in operations that may have been noteworthy. Note that in order
that energy use data and production data coincide in time, the company often needed to make
direct contact with its utility provider(s) to obtain energy consumption data within the specified
temporal window.
3.4 Sub-Cluster Characterization
Energy intensity values for each facility were sub-characterized by the plants’ primary North
American Industrial Classification System (NAICS) identifier. Baseline project participants were
recruited such that the data were representative of the NWFPA membership population within the
11 major food processing sub-clusters shown in Table 1 below.
1 http://www1.eere.energy.gov/ba/pba/intensityindicators/efficiency_intensity.html
2 http://www1.eere.energy.gov/ba/pba/intensityindicators/pdfs/index_methodology.pdf
4 Energy Intensity Baseline of the Northwest Food Processing Industry

Table 1: Major Food Processing Sub-Clusters
Sub‐Cluster NAICS Description Sub‐Cluster NAICS Description
Grain &
Oilseed
Sugar
Frozen Foods
Canning
Dehydrators
Dairy, Milk,
Cream,
Cheese
Seafood
311211 Flour Milling
311611 Animal (not Poultry)
311212 Animal
Rice Milling 311612 Meat Processed fm Carcasses
Slaughtering
311213 Malt Mfg & Processing 311613 Rendering and Byproduct Processing
311221 Wet Corn Milling 311615 Poultry Processing
311222 Soybean Processing
311230 Breakfast Cereal Manufacturing
311223 Other oilseed processing 311812 Commercial Bakeries
311311 Sugarcane Mills 311813 Frozen Cakes, Pies, & Pastries
311312 Cane Sugar Refining Baking 311821 Cookie & Cracker
311313 Beet Sugar Manufacturing 311822 Mixes & Dough from Purch. Flour
311411a Frozen Fruit, Juice, & Veg. (Ex. Potatoes) 311823 Dry Pasta Manufacturing
311411b Frozen Potato Products 311830 Tortilla Manufacturing
311412 Frozen Specialty Prep. Refrig. 311941 Mayonnaise, Dressing, & Other
311520 Ice Cream & Desserts
Foods 311991 Perishable Prepared
311421 Fruit & Vegetable
311320 Choc./Confectionery Mfg fm Cacao
311422 Specialty Canning 311330 Confectionery Mfg from Purch. Choc.
311423 Dried & Dehydrated 311340 Nonchocolate Confectionery Mfg
311514 Dry, Condensed, & Evap. Dairy Prepared 311911 Roasted Nuts and Peanut Butter
Non
311511 Fluid Milk Mfg 311919
Refrigerated
Other Snack Food Manufacturing
311512 Creamery Butter Foods 311920 Coffee & Tea Manufacturing
311513 Cheese 311930 Flavoring Syrup & Concentrate
311711 Seafood Canning 311942 Spice and Extract Manufacturing
311712 Fresh & Frozen Seafood 311999 All Other Misc. Food Mfg
In Table 1, each major sub-cluster is further broken out by the appropriate 6 digit NAICS
identifiers. In most cases, each of the 11 major sub-clusters is composed of the subset NAICS
identifiers within each cluster, however there are some exceptions which include:
1. Dried, Condensed, and Evaporated Dairy products were moved into the Dehydrators subcluster
based on experience showing that the dehydrated dairy products, NAICS 311514,
are very energy intensive.
2. Breakfast Cereal Manufacturing, NAICS 311230 was placed into the Baking sub-cluster
because the baking of the finished product is the process that consumes the most energy.
3. Frozen Potato Products (NAICS 311411b) were separated out from Frozen Fruit, Juice,
and Vegetables (NAICS 311411a). Normally, Frozen Potato Products would be included
in the NAICS 311411 identifier, but because potato products are very energy intensive
(there is normally a cooking step to reduce moisture--frying, blanching, etc, followed by a
freezing step--the additional identifier was added to allow a more thorough analysis of this
sub-cluster.
An additional 3 digits were added to each plant’s NAICS identifier to provide an even greater
level of detail, as required. These internally-developed NAICS+3 codes build upon the existing
level of detail identified within the NAIC system, but do not have numbers assigned. Table 2
below is an example of the NAICS+3 assignments, which further classify NAICS 311513, Cheese
Manufacturing according to the detailed NAICS listing.
Energy Intensity Baseline of the Northwest Food Processing Industry 5

Table 2: Example NAICS+3 Assignments for Cheese Manufacturing3
311513000 Cheese Manufacturing
311513001 Cheese (except cottage cheese) manufacturing
311513002 Cheese analogs manufacturing
311513003 Cheese products, imitation or substitute, manufacturing
311513004 Cheese spreads manufacturing
311513005 Cheese, imitation or substitute, manufacturing
311513006 Cheese, natural (except cottage cheese), manufacturing
311513007 Curds, cheese, made in a cheese plant, manufacturing
311513008 Dips, cheese based, manufacturing
311513009 Processed cheeses manufacturing
311513010 Spreads, cheese, manufacturing
311513011 Whey, raw, liquid, manufacturing
3.5 Use of Aggregate, Mean and Median Energy Intensity
To understand the energy intensity value of many facilities within an industry, it is important to
understand the definition of the energy intensity values that can be used to characterize a group
of facilities. The Aggregate Energy Intensity value is simply the sum of all the energy consumed
by the industrial sector divided by the sum of the finished product produced by the sector. This is
illustrated in Equation 2 below.
Equation 2: Aggregate Energy Intensity
I e,s = E t,s
A t,s
Where:
I e,c Energy Intensity for the cluster sample “s” (BTU/Lbs)
E t,s Total delivered energy used by the cluster sample “s” (BTU)
A t,s Total activity, or total output, from the cluster sample “s” (Lbs)
I e,i Energy Intensity for a given facility “i” (BTU/Lbs)
E t,i Total delivered energy used at a given facility “i” (BTU)
Total activity, or total output, from facility “i” (Lbs)
A t,i
=
ΣE t,i
ΣA t,i
Another approach is to use a Mean Energy Intensity value, which is the mean of the sample of
each individual plant’s energy intensity and is shown in Equation 3 below.
Equation 3: Mean Energy Intensity
I e,s
=
1
n s
Σ
E t,i
A t,i
Where:
I e,s Mean Energy Intensity for the cluster sample “s” (BTU/Lbs)
n s Number of sites (facilities) in the cluster sample “s”
E t,i Total delivered energy used at a given facility “i” representing the sub-cluster (BTU)
Total output, from facility “i” representing the sub-cluster (Lbs)
A t,i
3 As listed at http://www.naics.com/censusfiles/ND311513.HTM#N311513.
6 Energy Intensity Baseline of the Northwest Food Processing Industry

Yet another approach is to use the Median Energy Intensity value, which in some non-normal
distributed samples may better describe the central tendency of the sample processor facilities.
Simply described, the median value is the center of the ordered set of individual site energy
intensity values. The median value of an ordered sample population set is shown in Figure 1
below.
Figure 1: Example of Median Energy Intensity
A more visual comparison of all three energy intensity measures are contained in Appendix B.
Factors to consider in selection of a measure that represents the true central value of the
industrial cluster are:
1. Accuracy in characterizing the entire cluster with a limited sample set,
2. Consistency with historic data,
3. Which value will be most influenced by efforts to energy consumption reduction efforts.
Because of the ability to better describe the central tendency of the distribution of individual plant
energy intensities, the Median Energy Intensity value will be used to report the NW food
processing cluster energy intensity baseline value. Additionally, the aggregated energy intensity
of the sample set (of all individual plants) will also be computed and used to estimate sample
error.
The IAC data illustration in the next section will practically demonstrate why the selection of the
median value best suits the energy intensity data.
3.6 IAC Data Illustration
To illustrate which measure to use, the Industrial Assessment Center (IAC) data for the
Northwest food processors is examined.
There is a wealth of historical industrial energy assessment data available through the IACs’
database4. The IACs are part of the U.S. Department of Energy's (USDOE) Office of Energy
Efficiency and Renewable Energy (EERE) and contribute to its efforts by partnering with U.S.
industry in a coordinated program of research and development, validation, and dissemination of
energy efficiency technologies and operating practices.
The IACs are staffed by engineering faculty at 26 universities throughout the U.S. and are funded
by USDOE to conduct 8-12 energy efficiency assessments at small to medium industrial facilities
each year. The assessments are typically free (or low cost) if the company and/or facility meet
certain qualification criteria5:
• Within Standard Industrial Codes (SIC) 20-39.
• Within 150 miles of a host campus.
• Gross annual sales below $100 million.
• Fewer than 500 employees at the plant site.
• Annual energy bills more than $100,000 and less than $2 million.
• No professional in-house staff to perform the assessment.
4 http://iac.rutgers.edu/database/
5 http://www1.eere.energy.gov/industry/bestpractices/iac_eligibility.html
Energy Intensity Baseline of the Northwest Food Processing Industry 7

A typical plant assessment takes 1-2 days with 1-2 faculty team leaders and 4-5 engineering
students gathering technical data and energy measurements. About 60 days is required after the
visit to produce the assessment report with energy improvement recommendations.
Over the past 22 years, IACs at Oregon State University and the University of Washington have
completed roughly 150 energy efficiency audits at food processing facilities throughout Idaho,
Oregon, and Washington. The assessment data are available online, but the identifying names
and other proprietary data have been removed. Project research staff has analyzed these data
and the following summarizes the Northwest food processing plants that were assessed:
• 136 plant assessments in Idaho, Oregon, and Washington – of these, 129 plants were
used in the analyzed data set because of the ability to readily estimate finished
production by pounds.
• 20 years worth of data - the earliest assessment was in 1987, the latest was in 2007.
• Average assessment age is about 12.7 years old.
Each of the assessments contains pre-assessment data, which details annual plant energy
production, energy consumption, costs, standard industrial classification (SIC) (the forerunner to
NAICS). From this data, a historical characterization and model of the Northwest food processing
industrial sector can be made. The full set of IAC Northwest food processing sector data is
contained in Appendix C.
Figure 2 below is a histogram of all of the energy intensity values from the IAC data set. The light
blue bars represent the frequency of occurrence for the IAC data within each successive range of
values or “bins.” Each bin is 200 BTU/Lb wide.
Figure 2: Histogram of Historic Northwest Food Processing Energy Intensity Data
Median:
1,316 BTU/Lb
Aggregate:
1,529 BTU/Lb
Mean:
2,763 BTU/Lb
Approximate
Lognormal
Distribution
The distribution of data in Figure 2 is positively skewed which results in the mean value not being
centered on the central tendency (highest frequency) value as in the classic “bell” shaped normal
distribution. The distribution is positively skewed because energy intensity must be a number
larger than zero and extremely high values of energy intensity are not economically feasible. In a
8 Energy Intensity Baseline of the Northwest Food Processing Industry

positively skewed distribution, the median better describes the central tendency of the
distribution, while the mean will be larger than the median. Note that the median value is closest
to the central tendency of the distribution while the aggregate is about 16% larger in value. The
mean is over 100% larger than the median. This illustrates why the median value was chosen as
the value that will represent the data in the project. It will be shown that this project has
accumulated energy intensity data distributed in a manner similar to that of Figure 2.
Lastly, the red curve approximates the best-fit lognormal distribution. Using statistical analysis of
the IAC data, it does match closest to a lognormal distribution (p=0.108, α=0.05). This will be
important in understanding how to estimate error.
3.7 Estimating Data Error
The computed aggregate value of energy intensity in Figure 2 (1,529 BTU/Lbs) is less than 16%
different from the median value (1,316 BTU/Lbs) of each individual plant energy intensity. This is
not a coincidence. In fact, for an infinite population of data, the median will be equal to the
aggregated computed value. This relationship allows for comparison between the aggregate and
median energy intensity values to develop a relative measure of accuracy of the sampled plants
to estimate the entire cluster. This relative measure of Sample error is shown in Equation 4
below.
Equation 4: Measure of Sample Error of Cluster-Wide Energy Intensity
ε
°
= I e,s — I e,s
°
I e,s
Given:
ε Measure of Sample Error
I°
e,s Median Energy Intensity for the cluster sample “s” (BTU/Lbs)
Aggregate Energy Intensity for the cluster sample “s” (BTU/Lbs)
I e,s
Using Equation 4 above, sample error of the sample population can be estimated. Using the IAC
data of 129 food processing plants, a simulation was run in which a random set of plants with
their associated total energy BTU values, total pounds of production values, and energy intensity
values were recorded. In the simulation, a sample of 5 plants, then 10 more, then 10 more, and
so on until all 129 plants were computed for aggregate and median intensity values. The error at
each sample size was then computed. Lastly, this simulation was trialed 20 times, and Figure 3
below is the graphical representation of these trials.
From Figure 3, it can predicted that approximately 1/3rd of all of the Northwest Food Processors
Association plant population (about 44 plants out of 134 total plants) will need to be sampled in
order to obtain an average estimated sample error below 30%. Likewise, another 50 plants (94
plants total) will need to be sampled in order for error to be less than 20%.
An alternate method of predicting sample error, which will be used to confirm the primary method
above, is detailed in Appendix D.
Energy Intensity Baseline of the Northwest Food Processing Industry 9

Figure 3: Average Estimated Sampling Error vs. Total Population Size
30%
33%
3.8 IAC Historical Data Limitations
The average assessment age is 12.7 years old.
The IAC database is further limited because its plant/company size is somewhat smaller than the
average plant size of the NWFPA member plant population (about 134 plants out of a Northwest
food processing industry plant population of about 546 plants). The IAC assessment qualification
criteria would likely include about 90% of the NWFPA membership, however the 10% that would
be excluded are large facilities that likely have much more opportunity for energy intensity
reduction.
10 Energy Intensity Baseline of the Northwest Food Processing Industry

4 Project Results
Thus far, data has been collected from 47 plants and the resulting energy intensities are
summarized in Table 4 below. A detailed summary list of the individual plant energy intensity data
is contained in Appendix E.
4.1 Data
Data were requested from the participating plants for the years 2006 through 2009, with 2009
data being year-to-date data at the time of submission. Due to variability in company record
keeping, not every participating plant was able to provide all the data for all the requested project
years. However all plants did submit data for 2008 and thus 2008 was designated as the baseline
year for the study. Table 3 below is a synopsis of the collected project data.
Table 3: Results of Baseline Data
Measures 2006 2007 2008 2009
Number of Plants 32 42 47 30
Aggregate Energy Intensity (BTU/Lb) 2,428 2,279 2,256 1,625
Median Energy Intensity (BTU/Lb) 2,129 1,589 1,994 2,157
Mean Energy Intensity (BTU/Lb) 3,703 3,335 3,530 2,752
Est. Error % Primary Method 14.0% 43.4% 13.1% 24.7%
Est. Error % Secondary Method
Confidence = 95% 27.3% 58.4% 23.6% 23.1%
Confidence = 90% 17.2% 48.0% 16.1% 13.8%
Confidence = 85% 10.4% 41.1% 11.0% 7.5%
There are some items to be noted from this limited data set in Table 3:
Aggregate energy intensity (the sum of all finished product pounds for the sample plant divided
by the sum of all BTUs of energy consumption for the sample plants) has decreased every year
from 2006 to 2008 (for a total of 8.3%). There is also a decrease in 2009 as well, but with only 19
of the 36 plants having reported their 2009 data, this figure is likely not statistically sound.
The mean energy intensity has also shown a general downward trend and has decreased 9.6%
from 2006 to 2008. Again, with a limited data set, the 2009 figure has been discounted. It should
be noted that the median figure is statistically unchanged during this period.
The primary method of error estimation is below expected levels for the size of the data set.
Recall, that the primary method of error estimation compares median to aggregate energy
intensities with the median as the reference. It is expected that a data sample of about 44 plants
(33% of the 134 NWFPA member population) would yield approximately 30% in primary method
error, which is based on the IAC Northwest food processing data set.
The secondary method of error estimation is likewise below expected levels of 30% to 40%
(depending on confidence level) for the sample size. This is most likely due to the “un-smooth”
distribution of plants. This error method assumes that the individual plant energy intensity values
are distributed log-normally. The collected project data is close to being log-normally distributed
but not close enough to bring the error estimation levels below 100%. Refer to Figure 4 for a
graphic representation of the project data distribution.
The 2008 project data is distributed log-normally (α=0.05, p=0.551) and an approximate ideal
distribution of data (for an infinite population of data) is shown on the blue curve on Figure 4.
Similar to IAC data set, the median, aggregate, and mean energy intensity values are in the same
order of increasing size.
Energy Intensity Baseline of the Northwest Food Processing Industry 11

Figure 4: Histogram of 2008 Northwest Food Processing Energy Intensity Data
Median:
1,994 BTU/Lb
Aggregate:
2,256 BTU/Lb
Mean:
3,530 BTU/Lb
Approximate
Lognormal
Distribution
4.2 Comparison with IAC Historic Data
Table 4 shows the key measure comparison of IAC historic data and the 2008 baseline data. In
particular, note that the median energy intensity is approximately 51% in the 2008 baseline data
set. There are some possible reasons for this difference.
Table 4: Comparison of 2008 Baseline Data to Historic IAC Data
Measures
IAC
Historical
2008
Baseline
Number of Plants 129 47
Aggregate Energy Intensity (BTU/Lb) 1,529 2,256
Median Energy Intensity (BTU/Lb) 1,316 1,994
Mean Energy Intensity (BTU/Lb) 2,763 3,530
Est. Error % Primary Method 16.2% 13.1%
Est. Error % Secondary Method
Confidence = 95% 27.3% 24.9%
Confidence = 90% 17.2% 19.0%
Confidence = 85% 10.4% 14.9%
Lognormal Dist. Fit P-Value (α=0.05) 0.551 0.108
Lognormal Dist. Peak (BTU/Lb) 390 340
Lognormal Dist. Mean 7.21 7.51
Lognormal Dist. Std. Deviation 1.08 1.23
To better understand the primary reason why the 2008 baseline data median energy intensity is
larger, Table 5 shows how the components of the energy intensity values differ between the two
12 Energy Intensity Baseline of the Northwest Food Processing Industry

data sets. As was discussed in section 3.8 above, the 2008 baseline data is made up of larger
plants that produce more product and likely have more complex processing that uses more
energy per unit of finished product.
Table 5: Comparison of 2008 Baseline Data to Historic IAC Data
Production
(x1,000 Lbs)
Total Energy
(MMBTU)
Measures
IAC
Historical
2008
Baseline
IAC
Historical
2008
Baseline
Median Value 26,000 106,107 33,416 109,071
Mean Value 59,394 127,005 90,842 286,487
Standard Deviation 86,378 114,232 207,819 340,226
One other small contributing factor is that there is more variability in the 2008 baseline data as
shown by the larger values of standard deviation. The last factor contributing to the difference will
be discussed in the next section.
4.3 Energy Intensity by Sub-Cluster
Recall in section 3.4 that the baseline data would also be analyzed according to the 11 major
sub-clusters. Table 6 below is a breakdown of the sub-clusters according to number and median
energy intensity within each sub-cluster.
Table 6: Sample Breakdown and Comparison by Sub-Cluster
IAC NW Food Processor Data
Median Ranking
Percent Energy of Median
of IAC Intensity Energy
Sample (BTU/Lb) Intensity
2008 Baseline
Ranking
of Median
Energy
Intensity
Sub‐Cluster
Number
Dehydrators 10 7.8% 4,656 1 1
Prepared Non Refrigerated Foods 12 9.3% 3,257 2 2
Animal Slaughtering & Processing 9 7.0% 2,570 3 5
Canning 21 18.6% 1,913 4 7
Seafood 15 16.3% 1,182 5 6
Prepared Refrigerated Foods 6 11.9% 1,153 6 3
Baking 9 4.8% 1,125 7 9
Frozen Foods 24 4.8% 1,073 8 4
Grain & Oilseed 6 4.8% 473 9 NA
Dairy, Milk, Cream, Cheese 17 13.5% 375 10 8
Total 129 100.0%
For comparison, the ranking of median energy intensities from both the IAC data and the 2008
Baseline data are shown. The top two median energy intensities in both data sets are similar in
relative size among the sub-clusters. After the top two, both data set rankings diverge mostly due
to the different 3 rd and 4 th largest median energy intensity sub-clusters in the 2008 Baseline data
set. It is believed that this divergence is a consequence of (1) a smaller total sample in the 2008
Baseline data set as compared to the IAC sample (minor effect) and, (2) the existence of larger
plants in the 2008 Baseline data sample (major effect).
It should be noted that the sub-cluster median energy intensities and the number of plants in
each sub-cluster of the 2008 Baseline data could not be published due to the possibility of
compromising the identity of the contributing companies. Due to the low data size among some of
the sub-clusters, it may have been very easy for those familiar with the regional industry to
identify individual contributing companies.
Also note that the “sugar” sub-cluster was removed for clarity – there were no plants classified
into this category in either the IAC data or the 2008 Baseline data.
Energy Intensity Baseline of the Northwest Food Processing Industry 13

A detailed listing of the number of plants is included in Appendix F. This listing also shows the
NWFPA membership and the food processing industry as a whole in Idaho, Oregon, and
Washington.
4.4 Sources of Energy Consumption
Similar to the IAC data set, the 2008 baseline data also shows that natural gas is the largest
source of energy used by the Northwest food processing sector. Figure 5 below is an illustrative
comparison of the two data sets with energy sources broken out. This also served as an
additional validation check of the 2008 baseline data.
Figure 5: Sources of Energy in the Northwest Food Processing Industry
IAC Northwest Food
Processor Data
2008 Baseline Data
4.5 Conclusions
The collected 2008 baseline data behaves as predicted by the IAC data in most instances, and
where divergence occurs, there is sufficient understanding of the underlying reasons for the
divergence. It can be concluded that the 2008 baseline data accurately represents the Northwest
food processing industry based on the evidence presented above, which includes:
1. Both the 2008 baseline and the IAC historic data are lognormal distributed because of the
finite nature of energy consumption and production.
2. Both data sets share the same ascending order of median, aggregate and mean energy
intensities.
3. Both data sets have similarly behaving estimated errors using two different methods.
Additionally, both data sets have estimated error within expected ranges.
4. Both data sets have similar percentages of energy types consumed, namely electricity
and natural gas.
5. The 2008 baseline data has a larger median value due to:
a. The IAC eligibility criteria favoring smaller, less energy intensive plants
b. Larger variability in the 2008 baseline data as a consequence of fewer plants
c. A larger percentage of very energy intensive dehydrating plants in the 2008
baseline data
14 Energy Intensity Baseline of the Northwest Food Processing Industry

5 Recommendations
Three key learnings emerged during the course of this data collection project:
1. Treat participating plant’s data with the utmost of confidentiality at every step.
2. Keep the data request small and simple.
3. Encouraging voluntary participation is difficult – do everything possible to engage.
5.1 Data Security
The project staff established a well-coordinated system for protecting participant data and identity
from inadvertent disclosure. A well thought data security plan is key to following through on this
commitment. Elements of this plan included:
1. Storing all project related materials on secure disks – an open source encryption program
was used to block out a portion of the hard drive storing the data and also on the portable
USB memory device. The program used by the research staff was Truecrypt.
2. Stripping all company names off data files upon receipt and replacing each data file with a
3-digit identifier. The correlation between company name and the 3-digit number ID were
maintained on a secure hardcopy file.
3. Being careful not to disclose identifying information that might closely link the company
with its data. For example, where a company operated the only plant within a specific
NAICS category, the name of the category had to be modified to protect the plant’s
identity.
5.2 Data Simplicity
During the course of the data collection, it was determined that collection of energy and
production data in monthly time increments, as originally planned, was unnecessary given the
objective to characterize the baseline energy intensity of the Northwest food processing industry
cluster. Additionally, managers and other data-reporting individuals within each company have
limited time to accomplish their regular duties. Therefore, data collection requirements were kept
simple by using a calendar year time increment for energy and production data. A sample data
sheet is shown in Appendix G.
5.3 Encouraging Participation
The research staff was fortunate enough to have access to a retired food processing industry
executive who still had many contacts with the regional industry executives. Research staff found
that the top-to-top executive contact between the project staff and each individual processor
company was absolutely critical to success. This interchange set the appropriate project
expectation and ensured that all concerns regarding data confidentiality, project objectives, and
other concerns were fully vetted. On balance, it was this on-staff executive resource that primarily
encouraged participation, but other avenues of project participation encouragement should be
explored.
Plant participation is voluntary and therefore the participating companies set the pace for
providing the requested data. This responsiveness may impact the timing of overall project
completion. Again, the key is to keep the process simple from the participant’s point of view
A recommended alternative to improve volunteer participation would be to implement a
mechanism for providing anonymous feedback to the participating companies so they can
understand how they compare with their anonymous peer plants in energy intensity. In the 1980s,
the former Oregon Productivity Center (OPC), maintained a “double blind” cross-industry
productivity survey for many years.
The OPC process consisted of a 3rd party accounting firm mailing a prepared blank survey to
each participating plant. The plant then completed the survey and returned the survey back to the
accounting firm. Once a completed survey was received, the accounting firm replaced the name
Energy Intensity Baseline of the Northwest Food Processing Industry 15

of the plant with a randomly assigned identification number and maintained the cross-reference of
plant name to identification number in a secure location. The survey was then sent to the OPC for
data processing and analysis, identified only by the random identification number. The OPC staff
then completed the analyses of each industry sector and the unidentified plant shown for
comparison purposes. The analyses were sent back to the 3rd party accounting firm, which then
placed only the subject individual plant name back on the analysis sheet and mailed it back to the
original plant. In this way, OPC never knew the names of the plants and each individual plant
only saw its own plant named in a full comparison of how it ranked against its anonymous peer
plants in several key measures.
It can be reasonably hypothesized that a similarly constructed “double blind” energy intensity
survey with feedback, would elicit more volunteer plant participation. A critical element of this
process is the independent 3rd party firm that has experience handling sensitive data. This
approach would add much credibility to the data security assurances and greatly mitigate any risk
of inadvertent sensitive data disclosure, thus protecting each participating company and the
research staff.
16 Energy Intensity Baseline of the Northwest Food Processing Industry

Appendix A: Energy Intensity Data Collection Process
Energy Intensity Baseline of the Northwest Food Processing Industry 17

Appendix B: Visual Illustration of Mean, Aggregate, and Median Energy Intensities
18 Energy Intensity Baseline of the Northwest Food Processing
Industry

Appendix C: IAC Northwest Food Processing Data Set
IAC
Assessment ID Visit Date NAICS Products
Production
(Lbs)(1)
Energy Used
(MMBTU) (2)
Energy Intensity
(BTU/Lb)
OR0003 3/15/1987 311612 DRIED MEAT PRODUCTS 6,252,000 42,586 6,812
OR0004 3/15/1987 311423 FREEZE DRIED FRUITS 14,500,000 87,857 6,059
OR0011 6/15/1987 311411 FRUIT CONCENTRATE 59,400,000 77,430 1,304
OR0012 8/15/1987 311511 DAIRY PRODUCTS. 57,000,000 15,678 275
OR0014 9/15/1987 311712 PACKAGED SEAFOOD 11,200,000 12,623 1,127
OR0015 9/15/1987 311712 PACKAGED SEAFOOD 12,300,000 10,814 879
OR0017 10/15/1987 311421 PUREE CONCENTRATES 14,000,000 60,327 4,309
OR0019 12/15/1987 311421 MARASCHINO CHERRIES 16,000,000 95,711 5,982
OR0023 3/15/1988 311511 MILK,ICE CREAM & YOGURT 9,400,000 5,852 623
OR0024 3/15/1988 311511 VARIOUS DAIRY PRODUCTS 33,000,000 10,674 323
OR0025 3/15/1988 311511 MILK 32,000,000 11,655 364
OR0028 4/15/1988 311712 PACKAGED SEAFOOD 10,500,000 7,158 682
OR0029 5/15/1988 311411 FROZEN FRUIT 15,800,000 21,440 1,357
OR0030 6/15/1988 311919 POTATOE & CORN CHIPS 58,000,000 288,512 4,974
OR0031 6/15/1988 311920 ROASTED COFFEE, SOUPS 11,200,000 32,204 2,875
OR0033 7/15/1988 311712 PACKAGED SEAFOOD 29,000,000 34,272 1,182
OR0034 8/15/1988 311712 PACKAGED SEAFOOD 12,000,000 16,278 1,356
OR0040 8/15/1988 311411b FROZEN FRENCH FRIES 35,000,000 57,258 1,636
OR0041 9/15/1988 311511 MILK PRODUCTS 18,000,000 8,618 479
OR0043 9/15/1988 311423 POTATO FLAKES & STARCH 24,000,000 110,395 4,600
OR0044 9/15/1988 311411 FROZEN CORN & CARROTS 94,000,000 100,231 1,066
OR0045 9/15/1988 311611 FABRICATED LAMB 12,200,000 17,114 1,403
OR0046 9/15/1988 311611 BEEF CUTS & POULTRY FEED 85,000,000 74,217 873
OR0047 10/15/1988 311411 FROZEN PRODUCE 62,200,000 67,175 1,080
OR0048 10/15/1988 311421 PICKLES,RELISH & MUSTARD 31,500,000 23,762 754
OR0051 1/15/1989 311221 MARGARINE AND SHORTENING 60,000,000 43,096 718
OR0052 1/15/1989 311412 FROZEN SOUPS, CAKES 52,800,000 97,480 1,846
OR0054 3/15/1989 311812 BREADS 46,600,000 48,352 1,038
OR0056 3/15/1989 311991 SALADS,VEGGIES,SOUR CRM. 52,800,000 31,806 602
OR0070 8/15/1989 311423 DEHYDRATED FOOD 3,900,000 46,949 12,038
OR0072 8/15/1989 311919 POTATO CHIPS,NUTS,BUTTER 1,700,000 14,110 8,300
OR0073 8/15/1989 311421 CANNED FRUIT & VEGETABLE 73,690,000 81,867 1,111
OR0077 10/15/1989 311421 MARASCHINO CHERRIES 11,000,000 46,632 4,239
OR0080 11/15/1989 311941 SALADS, SOUPS, & SAUCES 10,000,000 10,901 1,090
OR0081 12/15/1989 311615 FRESH TURKEYS 13,000,000 33,416 2,570
OR0086 4/15/1990 311812 COOKIES 16,000,000 22,827 1,427
Energy Intensity Baseline of the Northwest Food Processing Industry 19

IAC
Assessment ID Visit Date NAICS Products
Production
(Lbs)(1)
Energy Used
(MMBTU) (2)
Energy Intensity
(BTU/Lb)
OR0088 5/15/1990 311422 CANNED CHILI 30,000,000 39,497 1,317
OR0116 3/15/1991 311514 CONDENSED SKIM MILK 46,000,000 26,042 566
OR0122 6/15/1991 311520 ICE CREAM, FROZEN YOGURT 24,615,450 15,988 650
OR0127 7/15/1991 311411 FROZEN FRUIT 6,000,000 3,086 514
OR0130 7/15/1991 311611 BEEF AND PORK PRODUCTS 20,000,000 24,098 1,205
OR0143 12/19/1991 311221 MARGARINE,SHORTENING,ETC 81,000,000 18,480 228
OR0144 1/14/1992 311411 FROZEN FRUIT, CORN 44,000,000 72,903 1,657
OR0156 6/15/1992 311513 CHEESE, BUTTER 4,000,000 12,729 3,182
OR0164 9/2/1992 311411 Frozen Vegetables 577,000,000 133,390 231
OR0165 9/17/1992 311520 Frozen Yogurt 10,700,000 11,155 1,043
OR0168 11/20/1992 311411 Frozen Pumpkin 11,700,000 10,894 931
OR0181 6/22/1993 311411 Frozen Vegetables 100,000,000 148,450 1,484
OR0184 7/13/1993 311411 Fruit Juice Concentrate 65,000,000 75,132 1,156
OR0188 7/16/1993 311511 Yogurt 6,250,000 6,322 1,012
OR0190 8/18/1993 311411 Vegetables 142,000,000 109,393 770
OR0193 8/26/1993 311411 Frozen Vegetables 100,000,000 171,175 1,712
OR0194 9/1/1993 311411 Frozen Vegetables 214,562,000 210,991 983
OR0204 2/10/1994 311511 Milk 65,000,000 15,304 235
OR0208 4/21/1994 311612 bacon, ham, sausages 7,000,000 26,909 3,844
OR0211 6/2/1994 311511 Milk, Cottage Cheese, Sour Cream, Buttermilk 76,689,105 17,399 227
OR0214 6/15/1994 311712 Fish Meal 10,000,000 37,298 3,730
OR0221 8/24/1994 311919 Tortillas, Corn Chips 20,600,000 51,922 2,521
OR0225 9/21/1994 311712 Fish Meal 20,600,000 5,051 245
OR0230 12/15/1994 311991 prepared salads 7,695,000 8,519 1,107
OR0235 3/22/1995 311421 packaged juice 45,959,317 27,028 588
OR0243 7/7/1995 311421 Aseptic Beverages 47,617,407 32,168 676
OR0255 9/28/1995 311511 Fluid Milk, Ice Cream 156,000,000 45,594 292
OR0258 12/13/1995 311712 Frozen Fish 110,000,000 14,646 133
OR0261 12/14/1995 311421 Pickles 37,000,000 42,889 1,159
OR0271 6/25/1996 311421 fruit juice, concentrate, jelly 119,406,000 75,947 636
OR0272 6/26/1996 311421 apple juice concentrate 69,772,000 92,262 1,322
OR0281 8/14/1996 311514 powdered milk 80,000,000 260,644 3,258
OR0283 8/19/1996 311421 maraschino cherries 16,117,786 79,496 4,932
OR0284 9/4/1996 311421 maraschino cherries 6,759,452 22,253 3,292
OR0285 9/5/1996 311421 Brine and Fresh Pack Cherries 12,805,150 26,131 2,041
OR0286 10/1/1996 311421 Pitted Brined Cherries 8,370,100 16,012 1,913
OR0291 10/22/1996 311919 potato chips, corn chips, nuts 4,719,113 25,486 5,401
20 Energy Intensity Baseline of the Northwest Food Processing Industry

IAC
Assessment ID Visit Date NAICS Products
Production
(Lbs)(1)
Energy Used
(MMBTU) (2)
Energy Intensity
(BTU/Lb)
OR0296 4/17/1997 311991 Meatless sandwich patties 17,060,941 20,443 1,198
OR0297 4/17/1997 311999 Steamed Rice 18,900,000 7,917 419
OR0300 6/16/1997 311423 Dehydrated Soup Mix 2,132,000 4,220 1,979
OR0302 6/18/1997 311823 Dried pasta and beans 38,000,000 34,232 901
OR0308 8/7/1997 311999 Processed eggs 16,000,000 17,730 1,108
OR0309 8/28/1997 311511 Milk, Ice Cream, Cottage Cheese 164,989,620 61,900 375
OR0311 10/1/1997 311411 Frozen Vegetables 34,259,124 33,378 974
OR0315 12/16/1997 311812 Baked goods 480,000,000 38,978 81
OR0327 6/14/1998 311812 Cookies 25,289,628 57,471 2,273
OR0338 9/6/1998 311919 Potato Chips 18,000,000 65,500 3,639
OR0351 6/21/1999 311712 Artificial crab meat 16,380,000 47,871 2,923
OR0354 6/24/1999 311712 Imitation crab meat 10,000,000 17,942 1,794
OR0357 8/10/1999 311712 Surimi, Fish, Shrimp, Crab 16,790,000 19,447 1,158
OR0358 8/11/1999 311712 Surimi, Shrimp, Crab 4,800,000 8,471 1,765
OR0359 8/12/1999 311712 Fish, crab, shrimp 18,200,000 11,846 651
OR0360 9/16/1999 311941 Soy Sauce 14,466,890 27,505 1,901
OR0371 7/17/2000 311411 Frozen Onions 30,100,000 27,198 904
OR0372 7/18/2000 311211 Flour 230,000,000 44,628 194
OR0377 8/10/2000 311423 Dehydrated Potato Flakes 38,435,809 412,293 10,727
OR0381 8/16/2000 311421 Canned and pureed fruit 8,300,000 16,448 1,982
OR0382 9/12/2000 311421 Fruit Puree 100,000,000 305,335 3,053
OR0385 9/15/2000 311712 Fresh/Frozen Seafood 10,000,000 13,541 1,354
OR0387 12/11/2000 311711 Artificial Crab/Surimi Seafood 12,341,692 33,326 2,700
OR0410 3/26/2002 311942 Hops extract 1,170,000 16,515 14,116
OR0411 3/26/2002 311942 Hops pellets 14,200,000 15,106 1,064
OR0414 5/19/2002 311520 Frozen Yogurt 3,552,830 8,526 2,400
OR0416 6/18/2002 311511 Fluid Milk 354,117,000 62,789 177
OR0422 6/20/2002 311421 fruit juice and puree 11,605,932 45,533 3,923
OR0428 12/16/2002 311230 Granola, Tea 15,400,000 17,322 1,125
OR0440 6/23/2003 311411 Frozen Onions, Peppers 62,900,000 63,402 1,008
OR0441 6/26/2003 311514 Powdered Milk 86,000,000 405,209 4,712
OR0442 6/27/2003 311513 Cheese 42,536,000 85,155 2,002
OR0459 4/5/2004 311411 Frozen Vegetables 150,000,000 97,990 653
OR0462 6/15/2004 311612 Sausages and Pepperoni 2,895,000 10,747 3,712
OR0467 6/22/2004 311941 Food dressings, mayonnaise, margarine 75,936,284 224,543 2,957
OR0474 8/31/2004 311225 Margarine 105,000,000 145,654 1,387
OR0475 9/9/2004 311421 Canned vegetables and fruit 86,000,000 136,090 1,582
Energy Intensity Baseline of the Northwest Food Processing Industry 21

IAC
Assessment ID Visit Date NAICS Products
Notes:
Production
(Lbs)(1)
Energy Used
(MMBTU) (2)
Energy Intensity
(BTU/Lb)
OR0478 9/16/2004 311411 Dried & Frozen Apples 51,246,000 529,210 10,327
OR0479 11/16/2004 311512 Butter 18,000,000 14,788 822
OR0484 3/22/2005 311942 Baking Powder, Cinnamon and other spices. 20,000,000 13,003 650
OR0490 6/28/2005 311211 Bulk flour & sugar 110,000,000 11,134 101
OR0491 6/29/2005 311421 Fruit Filling and Sweeteners 26,000,000 78,058 3,002
OR0493 7/19/2005 311911 Coffee Beans 6,500,000 24,412 3,756
OR0499 1/26/2006 311812 Bread and Buns 55,806,000 67,205 1,204
OR0503 6/26/2006 311421 Fruit Puree 160,000,000 229,940 1,437
OR0506 6/27/2006 311613 Sausage and sausage products 1,900,000 4,232 2,228
OR0507 6/28/2006 311514 Dairy 292,000,000 93,404 320
OR0508 8/8/2006 311423 Frozen Fried Potatoes 280,000,000 449,040 1,604
OR0511 8/30/2006 311812 bread and buns 50,000,000 511,615 10,232
OR0518 5/24/2007 311511 fluid milk 134,737,200 25,007 186
UU0067 5/25/2004 311213 barley malt 170,000,000 308,057 1,812
UU0079 10/7/2004 311513 cheese 94,000,000 137,701 1,465
UU0080 10/8/2004 311514 whey 25,000,000 2,063,375 82,535
UU0081 10/8/2004 311513 cheese 216,000,000 638,709 2,957
UU0093 7/15/2005 311612 tempura beef, chicken & pork 5,000,000 13,697 2,739
UW0023 10/3/2008 311230 Cereal Products 100,000,000 63,824 638
Total 7,661,827,830 11,718,573 1,317
(1) Production expressed as equivalent pounds when not explicitly listed as being in pounds.
(2) Energy used is the sum of all energy consumed on an annual basis.
Aggregate 1,529
Median 1,317
Mean 2,763
22 Energy Intensity Baseline of the Northwest Food Processing Industry

Appendix D: Alternate Method of Estimating Sample Error
This section details another method of predicting sample error based on sample size as related to the total
population size. The IAC data distribution (histogram) of individual plant energy intensities is shown in Figure
D1 below.
Figure D1: Distribution of IAC Northwest Food Processor Historical Energy Intensity Data
Median:
1,316 BTU/Lb
Aggregate:
1,529 BTU/Lb
Mean:
2,763 BTU/Lb
Approximate
Lognormal
Distribution
The blue bars represent the actual number (or frequency of occurrence) of individual plants within each of the
100 BTU/Lbs analysis bins. The red curve approximates the shape of a best fit lognormal distribution.
The next step is transforming the entire population of IAC data energy intensities to a logarithmic value and
building a similar log histogram of individual plant energy intensities. This “transformed” distribution is shown in
Figure D2 below.
Once the individual intensity values have been transformed into log values and plotted, the resulting
distribution in Figure D2 transforms to a more traditional bell shaped normal distribution shape with the red
curve approximating the theoretical normal distribution shape. Based on the results of the transformation
above, normal confidence interval evaluation tools can be used.
Energy Intensity Baseline of the Northwest Food Processing Industry 23

Figure D2: Histogram of Transformed Logarithmic Individual Plant Energy Intensity Values
Population Mean, μ = 7.21
Population Std. Deviation, σ = 1.08
Population Size, N = 129
Approximate
Normal
Distribution
The co nfidence interval approach is simply an expression based on sample mean, standard deviation, and
sample size that measures how far the sample average and population mean can be reasonably expected to
be apart.
Figure D3: Graphic Representation of Confidence Interval
As shown in Figure D3, the sample average will lie in the confidence interval which is the area bound by curve
between ± ε, or Margin of Error values. “ε” will represent the sampling error.
From statistical texts it can be shown for normal (Gaussian) distributed populations, the margin of error is
given by Equation D1 below.
24 Energy Intensity Baseline of the Northwest Food Processing Industry

Equation D1: Margin of Error
ε = z σ
√n
Where:
ε: Margin of error of the sample
σ: Population standard deviation
α: Confidence Interval – the area under the curve in Figure C3
z α /2 : Distance from mean expressed as multiples of standard deviation
n: Sample size
The “zα/2” value is determined by which confidence interval is chosen. For this analysis, confidence intervals
of 95%, 90%, and 85% will be evaluated.
Sample size can also be expressed as a percentage of population size as shown in Equation D2 below.
Equation D2: Sample Size Expressed as a Fraction of Population Size
%N = n / N
Where:
%N: %Fraction of population size
N: Population size
n: Sample size
From the transformed normal histogram in Figure D2, Equations D1 and D2 can be evaluated at several
different confidence intervals and many sample values. One set of values are demonstrated in Figure D4.
Figure D4: Example Error Range Calculation
Given:
m (mean) = 7.21 (From the transformed distribution)
s (Std. Dev.) = 1.08 (From the transformed distribution)
α = 0.05 (i.e. Confidence = 95%)
z α /2
= 1.96 (this value looked up from a Z score table at α=0.05)
n = 86 (note this is about 66% of the total population of 129)
N
= 129 (this is the full population size)
= 7.21
ε = z σ
√n
= 1.96
1.08
√86
= 0.228
= 0.228 This point on x axis is:
7.21+0.228 = 7.438
Now un-transform 7.438 using the reverse of the natural log function: e x .
Thus, Mean + ε = e 7.438 = 1,699 BTU/Lbs.
Recall, the actual untransformed median value was 1,316 BTU/Lbs thus,
estimated sample error is 1,699 – 1,316 = 383 BTU/Lbs.
Thus, sample error as a percent of the median is 383 BTU/Lbs ÷ 1,316 BTU/Lbs = 29%.
Energy Intensity Baseline of the Northwest Food Processing Industry 25

Using the same approach as Figure D4, if the same process is repeated at every sample size from 1 to 129
(expressed as % of total population size) for confidence levels of 95%, 90%, and 85%, the results are as
shown in Figure D5.
Figure D5: Plot of Sample Error as a Function of Sample Size and Confidence Interval
Note that the shape of the curves in Figure D5 follows the same shape as the sample error using the
previously developed median to aggregate difference approach (shown in green) which follows the 85% (blue)
confidence interval curves. Additionally, note that even if the full 100% of the IAC population were sampled (as
it models the NWFPA plant population), the best expected error will be between 18-24%. This alternate method
does provide a moderately accurate method of estimating error.
26 Energy Intensity Baseline of the Northwest Food Processing Industry

Appendix E: Detailed Project Energy Intensity Data Confidential
Energy Intensity Baseline of the Northwest Food Processing Industry 27

Appendix F: Sub-Cluster Characterization
28 Energy Intensity Baseline of the Northwest Food Processing Industry

Appendix G: Example Data Collection Worksheet
Company Name:
NAICS:
Finished Product Lbs
Total 06 Total 07 Total 08 Total 09*
Electricity KWH
Gas MMBTU
Other Energy#
Electrical BTUs 0 0 0 0
Gas BTUs 0 0 0 0
Total BTUs 0 0 0 0
BTUs/Pound
Data entry cells
* YTD as noted
#If applicable
Company Name: ABC Foods Company
NAICS: 311511
Total 06 Total 07 Total 08 Total 09*
Finished Product Lbs 70,001,377 100,773,989 103,438,359 43,002,247
Electricity KWH 1,438,900 1,818,640 1,929,200 880,400
Gas MMBTU 8,531 10,263 10,153 4,877
Other Energy# 0 0 0 0
Electrical BTUs 4,910,965,700 6,207,018,320 6,584,359,600 3,004,805,200
Gas BTUs 8,531,000,000 10,263,000,000 10,153,000,000 4,877,000,000
Total BTUs 13,441,965,700 16,470,018,320 16,737,359,600 7,881,805,200
BTUs/Pound 192 163 162 183
Data entry cells
* YTD as of 8/22/2009
#If applicable
Energy Intensity Baseline of the Northwest Food Processing Industry
29

Appendix H: Glossary of Terms
Cluster
Confidence
Double-Blind
EERE
Energy Intensity
Aggregate Intensity
Mean Intensity
Median Intensity
IAC
MMBTU
NAICS
NAICS+3
NDA
NEEA
Normal
Log-Normal
NWFPA
OPC
SIC
Sub-Cluster
USDOE
A regional collection of companies within the same industry
A measure of not making the incorrect decision when choosing a level of
error based on the statistical data available with in a sample – based on
the Gaussian distribution
A study method in which respondents identities are kept confidential and in
turn the respondent companies can see their company data compared with
other companies’ data in which the identities have been removed KWH
Kilowatt-hours, measure of real electric energy consumption
Office of Energy Efficiency and Renewable Energy, a branch of the USDOE
A measure of energy consumed per unit of finished product or economic
activity
Energy intensity among an entire cluster, which is computed by summing
the entire energy consumed in the cluster and dividing by the sum of the
finished product of the cluster
The arithmetic mean of all individual plant energy intensities within a
sample
The geometric center of all individual plant energy intensities within a
sample
Industrial Assessment Center, one of a collection of university based
energy efficiency research centers funded by the USDOE
Millions of British Thermal Units, a common measure of natural gas energy
North American Industrial Classification System, a unified method of
assigning a classifying code to a company depending on the type of
finished product or service provided
An internally developed method of further classifying the primary finished
product of a manufacturing facility
Non-disclosure agreement, a legally binding agreement to protect another
organizations sensitive or proprietary information
Northwest Energy Efficiency Alliance
A distribution of data which conforms (or is close) to the shape of the
classic Gaussian (i.e. “bell”) curve shape
A Gaussian (e.g. Normal) distribution in which each individual sample value
is now the logarithm of the original value.
Northwest Food Processors Association
Oregon Productivity Center, a publicly funded organization which operated
near Oregon State University in the 1980s and provided benchmarking data
among Oregon manufacturing industry companies
Standard Industrial Classification, the method of assigning a classifying
code to a company depending on the type of finished product or service
provided which was the forerunner to the NAICS
A further classification among a cluster depending on the type of finished
product that is manufactured.
United States Department of Energy
30 Energy Intensity Baseline of the Northwest Food Processing Industry

Appendix H: References
USDOE, EERE, Energy Intensity Indicators,
http://www1.eere.energy.gov/ba/pba/intensityindicators/efficiency_intensity.html
USDOE, EERE, Energy Intensity Indicators,
http://www1.eere.energy.gov/ba/pba/intensityindicators/pdfs/index_methodology.pdf
Industrial Assessment Center (IAC) Database, http://iac.rutgers.edu/database/
Energy Intensity Baseline of the Northwest Food Processing Industry 31