United States
 Environmental Protection
 Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
 Research and Development
EPA/600/SR-93/096   July 1993
  Project Summary
 Development of  Residential
 Wood  Consumption  Estimation
 Models
 Walid Ramadan and Mark G. Smith
  Data on the distribution and usage
of firewood were obtained from a pool
of household wood use surveys. Based
on a series of regression models de-
veloped using the STEPWISE  proce-
dure in the SAS statistical package,
two variables appear to be most pre-
dictive of wood use per household: (1)
heating degree days, and (2) percent-
age of households that burn wood as a
main heat source. The average number
of cords  burned in fireplaces is esti-
mated as a function of the total num-
ber of cords burned  per  household,
availability  of wood, and  population
density. Models  were also developed
to estimate the distribution of  wood-
burning devices. Variables that appear
to be most predictive of the percentage
of wood stoves are percentage  of
households that burn wood as a main
heat source, heating degree days, avail-
ability of  wood, and percentage of ur-
ban population. Input data, predicted
values, and deviations from actual val-
ues are  tabulated  for  each  model.
Graphs for each  model show both ac-
tual and predicted values for the vari-
ables being estimated.
  This Project Summary was developed
by EPA's Air and Energy Engineering
Research  Laboratory, Research  Tri-
angle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).

Introduction
  This report presents several statistical
models that have been developed to pre-
dict wood  fuel use in residential applica-
tions based on a review of available resi-
dential wood use surveys  conducted for
local area  planning or State Implementa-
tion Plan (SIP) inventory development
These models could be used to improve
the emissions  estimates  for residentia
wood fuel consumption in  areas that aro
not required, and therefore are  not likely
to conduct household surveys. The mod
els may also be included in the Area an; I
Mobile Source  Subsystem (AMS) of tru-
Aerometric Information Retrieval Systen
(AIRS).

Database Development
  Data on the  distribution and  usage o
firewood  are needed  in  order to  accu
rately characterize the resulting air pollu
tion impacts. Currently, the most accurate •
means of estimating residential wood com
bustion levels in a given area is througT
direct surveys of households. Wood use
surveys were obtained from several states
including  Colorado,  Connecticut, Idaho
Maine, Montana, Nevada, New Hampshire
Oregon, Vermont, and Washington. Whik
most surveys  reviewed were oonductec
specifically to  evaluate wood use  in ar
area, some were conducted for other pur
poses, but still included questions pertain
ing to wood burning. Most  of  the survey?
reviewed were from and within  Montana
Colorado, and  Washington. The geo
graphic levels at which the surveys wen
conducted ranged from small  towns tc
entire states. Only a fraction of the sur
veys clearly defined the boundaries of the
survey area. Of the surveys, 58% wen
conducted prior  to  1985;  the remaining
42% were conducted during or after 1985
  State-level survey results were collectec
for Colorado, Michigan, and Washington
These surveys were not used in the anaiy
sis because the focus of this effort was tc
develop models applicable at the county
level. Many other surveys collected were
not usable due primarily to a lack of infer
mation on  the surveys' boundaries, and/or
                                                  Printed on Recycled Pape<

-------
the lack of questions in the Surveys that
would  result  in estimates of fuel  wood
use. Finally,  in areas where more than
one survey had been conducted, only the
most recently complete was  used. Over-
all, 68 surveys were obtained, of which 40
were used in the analysis.

Statistical Analysis
  Before designing new models, several
existing models were examined that quan-
tify residential wood combustion. Two stud-
ies in particular were reviewed which esti-
mate firewood usage at the county  level.
One, conducted by Brookhaven National
Laboratory, performed a regression analy-
sis on  survey data collected by the U.S.
Forest  Service. The other study, by Skog
and Watterson, performed  a regression
analysis on the survey data  collected as
part of the New England Fuelwood Sur-
vey. The  models  and  their results  are
briefly  described and their application to
the surveys collected presented.

Conclusions
  Based on a series of regression models
developed to  estimate wood use and  dis-
tribution of wood-burning devices, the  two
variables that appear to be most predic-
tive are heating degree days and percent-
age of households that burn wood as a
main  source  of heat. A variety of local
factors can directly impact the number of
wood-burning households and the quan-
tity of wood burned by each. Such factors
include the historical availability and rela-
tive price of other heating fuels, and  the
heat content of firewood in the area where
the  survey was conducted. The latter fac-
tor is important because wood use is mea-
sured by cords, a volume measurement.
A cord of pine contains approximately 65%
of the heat content of a cord of oak.
  Limitations  on the estimates  of wood
use presented here do not necessarily lie
in the methodology but with the data used
as inputs for the models. The accuracy of
the  data collected is in doubt  because
there  was no independent verification. It
is likely that the variability in the data is
due in large part to the respondents' inac-
curate impression of how much wood is in
a cord. Another limitation is that the geo-
graphical boundaries of many of the sur-
vey areas were  poorly defined. Finally,
the  validity of the surveys themselves is
questionable especially in those areas that
showed large discrepancies between  the
reported  estimates of households using
wood  as a main heating fuel and the cor-
responding Census statistics.
  The residential wood consumption mod-
els  developed here may be improved by
 Walid Ramadan and Mark G. Smith are with TRC Environmental Corp., Chapel Hill,
   NC27514.
 Robert C. McCrillis is the EPA Project Officer (see below)
 The complete report, entitled "Development of Residential Wood Consumption
   Estimation Models," (Order No.  PB93-199 248/AS; Cost: $19.50, subject to
   change) will be available only from:
         National Technical Information Service
         5285 Port Royal Road
         Springfield, VA 22161
         Telephone: 703-487-4650
 The EPA Project Officer can be contacted at:
         Air and Energy Engineering Research Laboratory
         U.S. Environmental Protection Agency
         Research Triangle Park, NC 27711
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268

Official Business
Penalty for Private Use
$300
including a number of socioeconomic vari-
ables such as cost of other energy forms
in comparison to wood energy and aver-
age household  income. The models can
also be improved by conducting more sim-
plified and specialized wood use surveys.
Future guidance on the development and
use of surveys of residential wood use
should attempt  to gather  more detailed
information specifically on appliance char-
acteristics (i.e., catalytic,  EPA-certified),
wood characteristics,  and wood-burning
practices. Simplified surveys  reduce the
number of nonrespondents and the num-
ber of unanswered questions. Missing an-
swers in surveys present problems when
the data collected in a survey  are subject
to  statistical analysis  because statistical
procedures cannot be immediately used if
there  are missing values. In addition, the
models can be further improved by incor-
porating cross-sectional data  from areas
that conduct surveys periodically.  Cross-
sectional data may be used  to develop
wood use projection models and will  cap-
ture the effects of changing socioeconomic
factors on wood use patterns.  Finally, the
wood use models presented in the report
can be  validated, using field  surveys to
show how accurately the models estimate
wood use in areas where no surveys are
planned.

   'U.S. Government Printing Office: 1993 — 750-071/80027
                                                        BULK RATE
                                                  POSTAGE & FEES PAID
                                                           EPA
                                                     PERMIT No. G-35
EPA/600/SR-93/096

-------