Replace with appropriate Cover

'?' twi'
12000
(i960
— Il970
l^fiBjJmfflfW
I items (in terms of
_^ waste content)
- 	 - _ -n^-ft*

                      J discarded I
                      1 materials |
                      V from LRT items _X

-------

-------
     METHODS OF
      PREDICTING
     SOLID WASTE
CHARACTERISTICS
         Environraentai ;.-•-**.
         Region V, Library
         230 South Dearborn Street
         Chicago, Illinois 6060X
          This report (SW-23c) was written by
         GAIL B. BOYD and MYRON B. HA WKINS
         URS Research Company, SanMateo, California
        for the Federal solid waste management program
           under Contract No. PH 86-68-98
         U.S. ENVIRONMENTAL PROTECTION AGENCY
                1971

-------
        An environmental protection publication
     in the solid waste management series (SW-  23c) .
For sale by the Superintendent of Documents, U.S. Government Printing Office
                Washington, D.C. 20402 - Price 40 cents
                     Stock Number 5502-0048

-------
                   FOREWORD
INFORMATION ON  THE  QUANTITY  AND COMPOSITION  of solid  waste  is
indispensable in designing, implementating, and operating the solid waste management
systems of today and in forecasting  the requirements of the  solid  waste systems of
tomorrow.  Presently, methods of estimating  and  predicting  the quantity and  com-
position  of solid waste  are  most commonly based upon direct  measures  such as
sampling of solid waste itself. There  are, however, several disadvantages in the direct
approach:  (1)  Sampling  at  disposal  sites  does  not provide  adequate information
regarding the source  of the waste. (2)  It is difficult to recognize and identify solid
wate  components in  samples. (3) Discard  rates  for certain  solid wastes  are too
discontinuous, relative to time, to provide meaningful  samples unless the  sampling itself
is  done over  an adequate period of time. (4) Sampling  does not provide the insights
necessary to predict future solid waste quantities and compositions.
   The project summarized by this report involved a preliminary design of a model for
estimating and predicting the quantity  and composition of solid  waste and a determina-
tion  of its feasibility. The concept employed is, however, a  novel  one. Rather than
utilize  direct measures,  the model estimates  and predicts on the basis of knowledge of
materials and quantities before they become a part of the solid waste stream,  together
with an understanding of the  process  by which materials become solid  waste. Such a
model  accomplishes by synthesis  what direct measures seek to accomplish by  analysis.
The  synthetic  approach partially or  totally overcomes the disadvantages previously
mentioned for direct sampling of solid waste.
   The project  was directed  to a study of  residential solid waste, the  most  complex
area  of all in regard  to  this new technique  because of the countless items that  enter
the average home, many of which are small or insignificant. Thus, the project elected
to test feasibility in the area most likely to  highlight the weaknesses of  the technique.
In point of fact, however, the preliminary design performed quite well. Although the
approach  involves  certain  difficulties  in implementation  and  maintenance  in the
residential sector at this  time, this is not necessarily true for  the commercial sector.
Stores  deal more in bulk than residences do, and consequently, complexity is  reduced.
Further,  commercial  establishments  maintain detailed  inventories, something the
homeowner does not do.  The model developed is, therefore, especially attractive for
estimating and predicting the quantity and composition of commercial solid waste on a
community-wide basis, hitherto a  most intractable problem.
   The contributions  of  the  project  officer,  Albert  J.  Klee, Bureau of Solid  Waste
Management, to this study are gratefully acknowledged, as are those  of Professor John
Heer, University of Louisville, and the  editors of Life magazine.
                                            -RICHARD D. VAUGHAN
                                             Deputy Assistant Administrator
                                                 for Solid Waste Management
                                      in

-------

-------
                   CONTENTS
METHODS OF PREDICTING SOLID WASTE CHARACTERISTICS
                                                          1
SUMMARY 	  2

DISCUSSION AND GENERAL APPROACH	  3

HOUSEHOLD-WASTE PREDICTION TECHNIQUE	  8

HOUSE HOLD-WASTE PREDICTION MODEL  	12

DESIGN OF MODEL AS USED FOR TEST CASE	17

RESULTS OF TEST OF PROTOTYPE MODEL	20

CONCLUSIONS AND RECOMMENDATIONS	25

REFERENCES	27




FIGURES i

1 A CONCEPTUAL OVERVIEW OF THE BASIC SOLID WASTE PREDICTION MODEL  ...  5

2 THE HOUSEHOLD-SOLID-WASTE-PREDICTION MODEL, WHICH WAS EXPANDED
  TO ACCOUNT FOR LONG-RESIDENCE-TIME AND SHORT-RESIDENCE-TIME
  ITEMS SEPARATELY  	  7

3 HOUSEHOLD-SOLID-WASTE-PREDICTION MODEL	14

4 SOLID WASTE GENERATION RATES OBSERVED BY UNIVERSITY OF
  LOUISVILLE SAMPLING STUDY	23

5 SOLID WASTE GENERATION RATES PREDICTED BY USING URS PROTOTYPE	23
 TABLES

 1 QUANTITIES OF SELECTED WASTE MATERIALS GENERATED IN TEST
  COMMUNITY (AREAS OF JEFFERSON COUNTY, KENTUCKY) 	 2

 2 SIGNIFICANT COMMUNITY ACTIVITY CLASSIFICATIONS 	 4

 3 MAJOR EXPENDITURE CATEGORIES	 9

 4 HOUSEHOLD CHARACTERISTICS	10

 5 MATERIAL CLASSIFICATIONS BY WHICH WASTES ARE DESCRIBED IN
  PROTOTYPE MODEL 	18

 6 PREDICTION OF WASTE GENERATION RATES BY MATERIAL AND
  ORIGIN	22

 7 COMPARISON OF WASTE GENERATION RATES	23

 8 COMPARISON OF OUTPUT DATA FORMATS  	24

 9 PREDICTION METHOD STATUS AND PROPOSED WORK	26

-------

-------
                                     w^                              Jiii*P^         ^^HrJU*
                                       PREDICTING


                                                  JL                        Jl
  The  design,  implementation,  and  operation  of
efficient  and  economical   solid   waste  collection,
handling, transport, and disposal systems require accu-
rate information on the quantities and  characteristics
of the solid waste lo be processed. This information is
needed for the present and for the lifetime of a waste
management system. The overall objective of the study
reported herein was to develop methods that  can be
applied to real (or hypothetical) cities,  counties, and
regions for  estimating the quantity  and character of
the  solid wastes generated  at  present and at  various
times in the future.
  The work was planned  to be implemented  in two
phases.  Phase  1 was  the  preparation of  preliminary
design of the prediction  model and a determination of
its feasibility. The second  phase was the development
of the model, preparation of the associated computer
programs, assembly of  appropriate  data  banks, and
conduct of test runs for  specific areas.  This report
describes the work undertaken in  the first phase, in-
cluding  the results, our conclusions,  and  our  recom-
mendations  regarding the concepts and approaches for
the  work to be performed in the second phase of the
study.
  Within the scope of work as specified  in the con-
tract for phase 1, URS Research Company undertook
four major tasks. The first was to develop specifications
for, and to design,  the basic waste prediction model.
This necessitated establishing an output  information
format pertinent to the anticipated use of the data and
placing particular importance upon the degree of detail
in description of materials, location of waste, and the
time  at  which waste is  produced.  Concomitantly,
sources  of  information on  input  commodities and
activity  characteristics were identified  and investigated,
and the type and quality of available information were
defined.  Output  specifications  were  compared with
input  data availability, and  a  compatible model was
developed.
  The second task was to collect, develop, and formu-
late selected  standard data, descriptors, and functions
for  input commodities and activities. This involved an
analytical study of materials, commodities, community
activities, and waste production. Although much in-
formation  was presumed available,  it needed modifi-
cation  to  the format  used  in  the  model.  Various
sources were to be evaluated relative  to prediction of
the  character of commodities in the future. One of the
information items anticipated to be  somewhat difficult
to obtain was the "residence time"  data.  In the initial
study,  only a selected spectrum of  input  commodities
and activities was to be evaluated.
  The third  task was to test the model manually in a
small community. This test was to be used to evaluate
the  general performance of  the model. Manual com-
putation was  to be used to avoid costly  programming
and computer time while the model was  in a prelimi-
nary stage of development. A small community  or a
segment of  a  community was to be  selected  with
information on the quantity of input materials to be
obtained. The waste  characteristics would be predicted
and compared with data collected at disposal  or other
sites. The basis for the selection of the sample was to
be  its  capability of  providing  a significant sample of
the  most important prediction parameters.
  Finally, the test  results were to  be  evaluated in
order  to judge the practicality of  the  model, the
possible accuracy and precision of the system, and the
limitations to the system.

-------
SUMMARY
   The  basic  concept  of the solid waste  prediction
model developed in this project is that the waste gen-
erated by a community is derived primarily  from the
goods  and  materials  consumed  by the  community;
therefore  waste  quantities and characteristics can be
estimated   from  information concerning  the   con-
sumption  habits  of the community and the  manner in
which it  uses  and consumes the material  obtained.
Such an approach facilitates  the prediction  of future
waste characteristics because  of the widespread availa-
bility of information  concerning the  production, use,
and  composition of goods and materials in the future.
   The  study  consisted primarily of  determining the
availability of usable information and developing  a pre-
liminary prediction model for residential-household solid
wastes.  This model was restricted to the prediction of
present-day,  short-residence-time  wastes.  Its perfor-
mance was  tested by  comparing its predictions  for  a
given locality with the results of an actual study of solid
waste generation in that locality. The areas studied were
in Jefferson County, Kentucky. Pertinent  demographic
data were collected and  the  solid  waste  quantities of
various materials estimated by manual application  of the
prediction method. The results compared favorably with
the measured values (Table 1).
        Table 1. Quantities of selected waste
        materials generated in test community
        (areas of Jefferson County, Kentucky)
              (Ib/HOUSEHOLD/wk)
Measured by Univ.
of Louisville study
Material
Paper
Glass
Metal
Low
24.9
3.8
3.9
High
37.8
67
6.0
Estimated by URS
prototype model

23.43
2.54
4.23
   It is recommended  that  additional  work  be  per-
formed to  achieve further results. The  first objective
would  be   to  complete  the  development  of   the
residential-household-prediction model, to modify it to
include institutional and commercial activities,  to com-
puterize the model, and to run a test case. The second
objective of the additional work would  be to  perform
a  preliminary  development and feasibility  test with
methods similai to those used  for predicting household
wastes  and  foi predicting industrial, agricultural, and
other selected wastes.

-------
                  DISCUSSION AND
                                              GENERAL
                                      APPROACH
  An intimate knowledge of the composition  of the
waste is needed  as a basis for  the design  of waste
management systems  and for the planning of research
and  development for solid waste-handling and treat-
ment processes. The  technical feasibility of incinera-
tion, composting, and other processes; the operational
practicality of collection,  segregation, and secondary-
material recovery systems; and the economic feasibility
of all aspects  of waste handling require detailed in-
formation regarding the source, quantity, and quality
of all waste materials. In order to perform such studies
effectively, the changes  in  characteristics  of waste
materials (resulting from  population growth, land use
modification,  and technological  changes in material
properties and industrial processes) must be  predicted
for several decades.
  The usual  method of obtaining  data  on waste
characteristics and quantity is by physically inventory-
ing the materials delivered  to the  disposal site. Whereas
such an approach provides information that can be
useful in formulating immediate solutions to current
problems,* its use as the primary basis for defining and
solving long-range problems is suspect for four major
reasons.  First, inventorying present and past  wastes
cannot serve  as  a basis  for predicting future waste
characteristics or quantity, since it does not  account
for any of the technical, social,  economic, or other
factors that influence waste generation. Second, the
difficulty of segregating and  identifying waste  com-
ponents (after  they have  been  stored,  compacted,
transported, and repeatedly handled) severely reduces
the accuracy of any quantitative description. Third,
studies conducted at the disposal site generally do not
provide adequate information  concerning  the sources
of waste. Fourth, the discard  rate for many types of
waste is too discontinuous (relative to time) to allow
conventional surveys  to obtain sufficiently representa-
tive samples.
  The present  study is concerned with a waste  pre-
diction technique based on the following hypothesis:
that the wastes discarded by a community derive
primarily from the materials purchased, consumed, and
used by that community (i.e., "inputs"); therefore, one
can predict the amount and nature of these waste
"outputs"  by  identifying  what  goods  and  materials
  *Data derived from conventional waste measurement and
analyses have been valuable to the present study in that, they
provide baseline waste generation information and indicate
where better definition is needed.

-------
constitute  the  community's inputs  and by knowing
how the community acts upon these inputs.
   It  is  implicit in this hypothesis that studying the
characteristics of commodities, materials, and products
with  their "consumption"  by society will  reveal  in-
formation  about  the resulting  waste. Note that the
time required for a given material to pass through the
community is  an  important consideration  and varies
widely for different  materials (or products) according
to  the  uses  to  which  they  are  put. Some input
materials, such  as foods, are received, distributed, and
consumed  and become  waste in a mattei of hours or
days. Other  products, such as those  used in construc-
tion,  may be  resident  in the community  for half a
century or longer before they become waste.
   In a discrete  community, the input  materials may
be acted on  by the community by consuming, process-
ing,  or  generating  (Table  2).  The  community  may
consume  them,  eventually converting  all  the input
materials  to waste.  For  example, a case of  canned
vegetables  (consisting of solid food, liquid food, metal
cans,  and a fiberboard box) entering  a community will
be converted  entirely to  waste  within a few  weeks.*
   *For purposes of predicting solid wastes, only the packaging
materials are of interest.
          PREDICTING SOLID WASTE CHARACTERISTICS

The  community may process  them for consumption
elsewhere, with  only a fraction's becoming waste in the
community.   For  example,  most  of  the  steel  and
plastics shipped to a manufacturing plant will leave the
area, their only  trace in the community being the small
amount of production scrap  and the few units sold  in
the area. The community may generate materials,  in
the sense that agricultural crops, vegetation, and live-
stock are grown and resources are obtained from mines
and wells. In most  cases, the materials generated  within
the area are consumed outside the area, although waste
materials will  remain in the area.
   Obviously,  not  all  activities are  exclusively in one
category  or another. For instance,  all processing and
generating activities are also  consumers  of some  input
materials but  to a lesser extent.
   Estimation of the solid waste materials produced by
any  given community,  through  use of the concepts
proposed  herein,  primarily  involves  inventorying  all
waste-producing activities  in the community; compiling
standard  information  on  the materials  consumed per
unit  size  per  unit  time  by  each type  of activity and
information on which of  these materials become solid
wastes; and  determining  the rate of solid waste pro-
duction from  this information.
                              Table 2. Significant community activity classifications**

                       Consumption activities
                            Residential household (single-and multiple-family dwellings)
                            Financial and business establishments (including government activities)
                            Institutions
                            Restaurants
                            Utilities (including water and waste treatment)
                            Transportation
                       Processing activities
                            Industrial:
                                  Food processing
                                  Manufacturing of durable and nondurable goods
                                  Chemical manufacturing, processing, refining
                            Construction
                            Commercial (retail and wholesale marketing)
                       Generation activities
                            Agriculture
                            Mining (including petroleum)
                            Commercial fishing
                            Demolition
                            Miscellaneous  (e.g., street sweeping, gardening)
                         **These classifications relate closely to those of the Bureau of Budget's Standard
                      Industrial Classification, and  to  the  system  proposed  by  the Urban  Renewal
                      Administration and the Bureau of Public Roads for land use classification, and thus
                      facilitate the  use  of the vast  amount  of  data   available from  the  Federal
                      Government.! ,2

-------
DISCUSSION AND GENERAL APPROACH
   Information  on the  rate  of material  consumption
(and, therefore, waste  generation)  by  each type  of
activity  is derived from many sources  and  is  based
largely  on  descriptions of  the materral  input-output
characteristics of  the  processes used  by each activity.
Although  the   actual  computations  and  data  bank
formulations vary  somewhat, depending on the type  of
activity  being  considered,  a  basic  computational
approach is applicable (Figure  1).  The four basic data
banks of the model are described here:
   Description of  Community. This is a data bank that
provides an inventory of each  type of activity in the
study community, as  well  as their number,  size, and
location.*  This  is  actually  an  inventory  of  waste
sources.
   Description  of Standard  Activity.  This is a  data
bank that  indicates  for  each  type  of activity the
amount  of waste  items** and  materials  produced per
unit time  per   unit size of  activity.  This data  bank
would  be of a general  nature and  would  apply to the
study of any area  of the country.
   "Location refers here to census tract, enumeration district,
county, urban area, or  some  other segment of the area being
studied.
  **In the terminology used  herein, "waste item" refers to a
recognizable, discrete object made up of one or more materials.
For example, a discarded television set or sofa is a "waste item."
A  "waste material"  is a more specific classification  of waste
content, e.g., glass, copper, paper, wood, or textiles.
   Standard  Waste Composition Data. This signifies a
standard  data  bank  that provides  the  information
needed  to convert a unit of input item into the quanti-
ties of waste  materials it will yield. For  example, N Ib
of  coffee  would be  converted to  X Ib  of granular
solids, Y  Ib  of metal (from  the can), Z Ib of plastic
(from the lid).
   Local Waste Management  Practices.  This is  a data
bank  that  provides  information  on  the  mariner  in
which  specific  wastes  from  specific  sources  are ulti-
mately  disposed of by the community  (i.e., whether
they  are  collected  as  refuse,  segregated  for  salvage,
disposed of  on site, or  otherwise  eliminated).  This
information is specific for  each  community and reflects
both  the  habits of  the  residents and the legal restric-
tions on disposal practice.
   Two  intermediate   compilations   result from  the
various computations (Figure  1).
   Total Waste Inventory.  This  lists the quantity of all
waste items produced in each  location within the study
area. It  is  derived from the first two data banks.
   Descriptive Inventory of Solid Wastes.  This lists the
quantity of each waste  item  and waste material  likely
to end  up  as  a  "solid waste." Solid materials  while
airborne 01  waterborne  are eliminated from considera-
tion.1 This information  is  derived  by using additional
process  data  from  the Activity Description and from
Waste Composition Data to develop the waste material
inventory.
Standard waste
composition
data

,\
; »
      Figure 1. A conceptual overview of the basic solid waste prediction model. The separate models for
    predicting the wastes from each  type of activity would differ somewhat, but  would all share this basic
    format.  Rectangles represent basic data banks. Irregular octagons represent the intermediate compilations
    arising from the various computations. Shaded rectangles are a series of compilations, the final output of
    the model.

-------
                                                                 PREDICTING SOLID WASTE CHARACTERISTICS
   The  final output of  the  model  is a .series of  four
compilations (Figure 1).
   Inventory of Wastes Collected as Refuse. This inven-
tory lists the quantities of  waste  items and materials
normally collected for disposal.
   Inventory of Wastes Segregated for Salvage.  This
lists the quantities of waste  items  and materials segre-
gated (by  the  activity)  and  eithei  recycled within the
process  or collected  separately  for  reuse,  resource
recovery, or salvage.
   Inventory of Wastes Disposed of On Site. This is  a
compilation  of  the   quantities  of waste items  and
materials disposed of by the activity itself. This would
include, for example, materials burned in an incinerator,
ground  in  a  garbage  grinder,  buried, or  otherwise
processed for disposal.
   Inventory of Other Solid  Wastes. This list includes
those  waste items  and materials  not elsewhere  ac-
counted for.
   The  wastes  are  tabulated  in  two  ways: by  the
amount of each item (e.g., pounds  of tin cans, washing
machines)  and  by  the amount of  constituent material
(e.g., pounds  of  paper, ferrous metal, plastic).  This
appears to  meet the requirements of the initial uses of
the  model. It will probably, however, be  useful to
introduce further calculations (not  shown in  Figure 1)
to  convert  the basic  waste inventories so  that  they
reflect other  waste  characteristics, such as  calorific
value, biochemical degradability, chemical composition,
bulk density, real density, and unit  size.
   A more specific version  of  the  basic  prediction
model applicable to residential households would need
to redefine some  data banks in relation to the house-
hold activity function and to incorporate consideration
of  two  important aspects  of the  ultimate  computa-
tional procedure.  The wastes from short-resident-time
(SRT) items must be  predicted separately from  wastes
from long-residence-time (LRT)  items.*  Data  banks
required for predicting waste  quantities and  character-
istics at  selected times in the future must be included
(Figure 2).
   In this  study,  no  specific  useful-life duration was
established  to  differentiate  between  SRT  and  LRT
items. Items  "consumed"  within   a year  or less  are
generally considered, however, to be SRT items. Even
without a  rigidl>  established criterion, little difficulty
was  experienced in assigning items to one category  or
the other.
   Separating  the prediction  processes  for  SRT and
LRT wastes is  necessitated  by two basic factors. First,
household  inventories of LRT  items tend to increase;
therefore, a purchase of such an  item  does not neces-
sarily result in  the immediate discard of  a like item, as
distinguished from SRT items, which  are used essen-
tially immediately and  become  waste  in  the  same
general  period  in which they  are  procured. Second,
because the material characteristics of many LRT items
change  with changes in  technology,  and since  those
being discarded  today  were  manufactured  5  to  15
years ago,  special consideration must  be given to the
material characteristics  of the item. The  means of pre-
dicting  both LRT and SRT wastes  are discussed  in
detail in a later section.
   An important  use of the prediction  model is related
to the estimation  of waste characteristics in the future.
To use the  model  for such a purpose  requires pre-
dictions and forecasts of future changes  in  the follow-
ing:   descriptions of  communities;  household  con-
sumption patterns (and descriptions of standard activi-
ties); composition of SRT  and LRT items;  household
waste management practices; ownership  data on  LRT
items; discard rate data.
   Generally, the  most valid predictions  of changes in
materials, products,  and technology are those made  by
persons having  the best knowledge of the specific tech-
nologies  and industries involved.  Similarly, the most
important  predictions of the growth  and  change of a
given community  are those by the responsible author-
ities within the community. The  method used (that of
having  essentially separate  data  banks  for  each pre-
dictive year) was, therefore,  chosen partly because it
provides a  means of  incorporating independent pre-
dictions  from  different  sources.**  At  present,  it is
planned to use  1965  as the  base year with  1970,  1980,
and  2000  as projection years.  In applying  the model,
only  1  year would be  considered at a time. Although
forecasts are available  for  many  specific items of  in-
formation  (e.g., estimates  of their anticipated growth
made by communities), it  may be possible  and desir-
able  to  develop  specialized  models  (and  computer
   *LRT items are those that tend to remain in use (or at least
in "storage") in the normal household for a relatively long time
Examples would include furnishings, appliances, books, tools,
and the like. Examples of  SRT  items would include food,
periodicals,  most  clothing, and disposable paper and plastic
goods.
   **The problem remains,  of  course,  of ensuring that  in-
dependent predictions are reasonably compatible. This would be
done at  the time the  predictions are evaluated  and the data
banks prepared.

-------
 DISCUSSION AND GENERAL APPROACH
 programs)  to  forecast systematically  the growth of
 selected activities. The specific means  of generating
 these forecasts is not within the scope of this study.
   During  this phase of the study, a detailed diagram
 of the household waste prediction model was develop-
ed. It  followed the  general format  already discussed
but employed  several minor  modifications. This pro-
totype model concentrates on SRT items and considers
but one  time period  (i.e.,  it cannot make  forecasts of
future wastes).




12000
|1980
11970
1965
Household
comumpr ion
patterns

|?000
11980
|1970
1965
a. community






1
\
J



,
L
V
<






\*
T
4


V
A
P

(2000 1
11980
|W70 1
i965 J
0«scnpt>oo of SRT
waste content)


flnvento-yof ^ \± ' Cescnptiv. ^
> »L.O,en',ia,.os,esj *» \ -^ J ^
^7
|2000 /
, I1980 A\
fl970 Y\\
965 ^H
Household's waste Lj
^nanogemenf 1
practices k

12000 1
1 1 980 1
11970 1
1965 I J
Ownership | ^1

r Study area
households1
data an J »* PI prcs,n, ,nven
LRT items ^_ Of LRT „,„,
|I980 1
11970 1
1965 H
Discard rates LJ
For LRT items J

ory j^1

K Study a
d scord
for LRT
|1980 1
|1970
1965 1 LJ
Description of LRT items
(',ri tetms of waste |T~
Campos it on) J

tea's 1 ,
=. J^

Klnvenforyof '
discorded
matena s
From LRT i^emi A
   Figure 2. The household-solid-waste-prediction model, which was expanded to account for long-residence-time and
short-residence-time items separately.

-------

TECHNIQU
   The household-waste  problem was  chosen  as  the
subject for this preliminary development and feasibility
study because in many  ways  it embodies the most
complex  prediction  problems.  In  addition,  the col-
lection,  transport,  and disposal of  household  waste
represent the most costly and  widespread waste man-
agement  problems,  and  their  effective  solution  is
dependent on useful waste characteristic data.
   The  basic  computational   model developed  for
household-waste prediction has  been  described (Figure
2).  Although the  model is large in  that  it handles  a
sizable amount of numerical data concerning expendi-
ture rates, consumer  item descriptions, use  patterns,
and disposal practices, it  is simple in that only basic
arithmetic functions are used in manipulating the data.
   The general aspects  of  the computational model are
discussed  here,  and the  specific application of  the
model is described in a later section.
          Bases for Household-Waste Model
   The fundamental element of the household-solid-
waste prediction  model  is the  data  bank  entitled
"Description  of  the   Community"   (Figure  2).  The
number  of persons living in each area of the com-
munity  is, of course,  the  community  characteristic
most  pertinent to  the household-waste  problem, and
this information is readily  available from  the Bureau of
Census.  Additional  descriptors, such as  number of
families   and  income   distribution   among  families,
provide  information  pertinent  to  the  consumption
habits of the residential sector.
   The  "Description  of the  Standard Activity" is  a
data bank that indicates the  quantitative input-output
characteristics  of  the  activities—in this  case,  the resi-
dential  household. During the  initial  phases of the
study, several  approaches to  determining the  "input"
materials to a  community's  households were investi-
gated, including consideration of the  means of inven-
torying  the shipments into the community, the ship-
ments to wholesalers and distributors, and so on. From
a review of available information, however, it became
obvious  that studies of consumer  buying habits would
be  the most practical sources of data. Of the various
studies of this  type,  the most applicable one encoun-
tered  was  a report entitled "Expenditure patterns of
the American  family."3 The report  was prepared by
the National  Industrial Conference  Board  in  1965
under the  sponsorship of Life magazine, and it will be
referred  to herein as  the  "Life  study." Based upon
statistical data collected as part of an unusually com-
prehensive  series of surveys of consumer behavior con-
ducted by  the  Bureau of Labor Statistics, U.S. Depart-
ment of Labor, it contains a  vast quantity of data on
the  rate  at which households spend  money on about
700 different  goods and services (Table 3).  The  im-
portant feature of this study  is the information on the
manner  in  which each  of these expenditures varies

-------
  HOUSEHOLD-WASTE PREDICTION TECHNIQUE
                                      Table 3. Major expenditure categories*
  Major expenditure categories
 Number of
items listed**
Major expenditure categories
 Number of
items listed * *
  Food, beverages, and tobacco
       Cereals and bakery products               23
       Meats, poultry, and fish                   18
       Milk, cream, and cheese                   11
       Fruits and vegetables                     47
       Other foods                             56
       Alcoholic beverages                       5
       Tobacco                                 4
  Housing and household operations
       Shelter and other real estate                17
       Household operations                     22
       Household supplies                      20
  House furnishings and equipment
       Household textiles                       14
       Furniture                               11
       Floor coverings                           8
       Major appliances                         18
       Small appliances                          4
       Housewares                             12
       Miscellaneous items                      11
  Clothing and accessories
       Clothing (aggregated by sex and age)       142
       Clothing materials                         1
       Clothing upkeep                          3
  Transportation
       Automobile purchase                      2
       Auto-operating expenses                 10
       Auto insurance                           5
       Repairs and parts                         1
       Other auto expenses                       1
                         Local transportation
                         Intercity transportation
                         Other transportation expenses
                   Medical and personal care
                         Group plans and insurance
                         In-hospital care
                         Professional services
                         Drugs and medicine
                         Other medical care
                         Personal care services
                         Personal care supplies
                   Recreation and equipment
                         TV, radios, and musical instruments
                         Spectator admissions
                         Participant sports
                         Club dues and memberships
                         Hobbies
                         Toys and play equipment
                         Recreation out of home city
                         Other recreation
                   All other goods and services
                         Reading
                         Education
                         Other current expenditures
                         Additional disbursements
                                             1
                                             6
                                             5

                                             3
                                             3
                                             3
                                             6
                                             1
                                             5
                                            14

                                            10
                                             4
                                             3
                                             1
                                             5
                                             7
                                             1
                                             1

                                             6
                                             4
                                             4
                                             7
                                     Total number of items and
                                     aggregations of items listed = 566
      *These catagones are  handled as separate parameters in the Life study (Ref. 3).

     **Some of these "items" are actually aggregations of related individual items. For example, the following aggregation is listed as a
  single "item" of men s clothing, suits, sport coats, and trousers.
relative to  factors  such as family size, family income,
and geographical region (Table 4).
   One limitation of the Life  study,  relative to  its use
for waste prediction, is  that the  results are reported in
dollars expended. Consequently, a considerable portion
of the research  effort was directed toward converting
the  reported  expenditure rates  into material  input
rates. The Life study states how many dollars are spent
(per household per week) on  common consumer items,
such  as canned coffee. Consequently, it was necessary
to obtain  sufficient additional information to convert
this  dollar  expenditure  figure to  pounds  of steel can,
                   pounds of  plastic  lid,  and  pounds  of spent  coffee
                   grounds (per household per week), since these  are  the
                   potential solid  wastes.
                      Other major tasks involved deciding whether or  not
                   these potential wastes actually become waste, how  this
                   occurs, and what  form the wastes take. These factors
                   depend upon how the household processes the "input"
                   material. For   most items, information concerning  use
                   patterns came  from  a  general knowledge of household
                   operations,  rather  than from  formal data sources. The
                   details on the  means of obtaining and manipulating the
                   various types  of information  concerning each  input in

-------
10
                                      PREDICTING SOLID WASTE CHARACTERISTICS
          Table 4. Household characteristics*
Household characteristics
Number of categories listed
Age of head of household                   6
Stage in life cycle (indicates ages of children)    6
Family size                               5
Family income                            6
Earner composition (identifies breadwinners)    4
Occupation of head of household             6
Geographical region                        4
Race                                    2
Market location (proximity to metropolitan area) 5
Home tenure (owned or rented)              2
Education of head of household              5
      Total number of categories listed =
            51
   *These  household characteristics are handled as  separate
parameters in the Life study (Ref. 3).
order to obtain final  waste  generation predictions are
discussed in the following paragraphs.
  Prediction of LRT Items.  A  major problem, as pre-
viously  indicated,  arises  from the necessity of predict-
ing  the disposal  of items that have  a  relatively long
useful  life. In  actuality, today's household waste con-
sists of three  major fractions: SRT  items,  including
their  packages and containers,  that  have  been pur-
chased recently; the containers and packaging materials
accompanying  the   LRT  items  that  have   been
purchased  recently; LRT items purchased in the past.
Although  the Life  study data provide information that
can  be used  as a basis  for the  predictions of waste
from the  first  two  items, they are not useful for the
prediction  of  the  third, since  the purchase of a new
LRT  item is  not  automatically  accompanied by the
discard  of an older one.
   The  following   characteristics,  common  to  many
LRT items, make  prediction of their waste characteris-
tics  difficult:  their discard rate is difficult to estimate
because  of the tendency to  increase the  per  capita
resident  inventory  of many  items;  many  items dis-
carded  by one population group are reused by  other
population groups  in  different locations before finally
being  discarded  as refuse;  technological  changes  in
manufacturing  processes, product design, and materials
have  resulted  in  changes in  the composition of many
items.
   The  first of  these characteristics  is related  to a
number  of factors,  including:  a general  increase  in
disposable  income;  a continuing reduction in the base
price of many LRT items;  the  occurrence of buying
trends based  upon status and affluence  motivations;
the low  salvage value of outmoded  but still operable
items.
   Television sets are a good example of this tendency.
When  television was first introduced,  ownership was
quite limited, whereas now virtually  all families own a
set, and  many own  two or more. Television ownership
has risen from 20  percent of all  families  in  1950 to
more than 93 percent  at  present.4  Many  other  items
(e.g., radios,  small  appliances,  boats)  also  show an
increased per  capita inventory.5 The tendency of  many
families  to  maintain  more  than  one  residence  (e.g.,
vacation  homes)  also complicates the  prediction pro-
cedure.
   The fact that many LRT items can be put to use by
someone other than the original purchaser complicates
waste  prediction  in  that it interferes with  determining
the location  at which  the item will  ultimately be dis-
carded as waste.  This  problem  also   interferes with
estimating the useful life of many items.
   The third  problem, that of introducing composition
changes into  the program, is not overwhelming. Actual-
ly,  most of  the'  significant changes  appear  to have
occurred  rather recently, and their general characteris-
tics can  be inferred from trade literature. The  problem
is  largely  related  to  obtaining  sufficiently  accurate
"average"  composition  information  for  the  various
items.
   Specific methods  of predicting the quantities and
characteristics of  yesterday's LRT waste have not been
developed to  date.  The major problems have, however,
been identified and several concepts  have been briefly
explored. The computational procedure  is believed to
be  conceptually  correct  (Figure  2).  The data  bank
"Ownership Data for LRT Items" is used to indicate
the  quantities of  durable  goods  used  or stored in
typical  households  of various economic  levels. This
type of  information is available  from  the Bureau of
Labor  Statistics.  Bureau  of Census,  and  various non-
governmental  sources,  although it must be evaluated
and coordinated before use.
    The  data bank  "description  of  LRT  Items" will
be  more difficult   to  develop.  The  desired  output
"number  of each type  of items of X years' average age
discarded per household  per unit  time" can probably
be derived for many items from production and owner-
ship data. The derivation, however,  will be somewhat
laborious and t-me  consuming  unless a computer pro-
gram is developed to assimilate the available data.
   The data bank "Discard Rates for LRT Items" will
provide  information  on the  composition  of  the dis-
carded LRT  items.  The format of the information is

-------
HOUSEHOLD-WASTE PREDICTION TECHNIQUE                                                             11

dependent  on the output of the "discard  rate" data    dure for prediction of LRT waste was developed and
and  the  assumptions concerning  the composition of    used as part of this study.  Its development will be an
LRT  items  of various sizes,  types, ages, and brands.    important initial aspect of our subsequent work.
This information  will be costly to collect if the  accu-       Quantities of solid waste derived from SRT house-
racy requirements are demanding.                        hold items have been estimated by means of a com-
   As  previously  indicated,  no computational proce-    putational approach.

-------
  HOUSEHOLD
 PREDICTION
   The following discussion examines various portions
of the household-waste prediction model and explains
their  content, function, and relationships (Figure 3).
To  some  extent, the  description relates more directly
to the methods used for the manually pei formed case
study,  rather than those that  wil)  be  developed for
eventual computer solution.
            Prediction of Input Items
  Summary of Expenditures. Block c is the result of
the  initial  computation.  It indicates the amount of
money  all households in  a study area will spend col-
lectively on each of a large number of consumer items.
The expenditures  are expressed in  terms of "dollars
spent per unit time" and  are tabulated on an item-by-
item basis (e.g., expenditure rates for bread, soap, and
floor wax are expressed separately). This expenditure
summary is based  upon information regarding the way
an individual  household of a particular type spends
money  on  various items  (Consumer Expenditures by
Household, block  b) combined  with information on
how many households of  each type are located within
the  study area (descriptive information on households
in study area,  block a).
  Consumer Expenditures by Household. These  data
(block b) were obtained primarily from the Life study
described previously.3  Data are expressed in terms of
"dollars  spent per household per unit time" and are
tabulated on an item-by-item basis. The data in the
Life study are presented in the foim of a large matrix,
the row  headings  being  items, the columns being
household characteristics. The body of the matrix  itself
consists of the  expenditure rates each type of house-
hold will have for each particular item listed (there are
tens of thousands of such  rates listed, all given to the
nearest cent per week). In the present study the major
groupings of the Life  report row and column headings
were listed. In summary, a wide variety of information
is  available, but generally only those data relatable to
income and section of the country were found to be
directly applicable.
   The Community Description.  Block a provides in-
formation  on the  demographic  and socioeconomic
nature of the sludy  area.  The following factors  were
found  to  provide the most useful information on the
people and the activities in the study area: population;
population density; average size of household; distribu-
tion of income, number, size, and  type of establish-
ments  (wholesale and retail  trade, selected services,
manufacturers, mineral industries, agriculture). For any
given study area, the most desirable sources of  such
information might differ somewhat but would general-
ly  include the U.S. Department of Commerce (especial-
ly  the Bureau  of Census and the Business Defense
Services  Administration);  the  U.S.  Department  of
                                              12

-------
HOUSEHOLD-WASTE PREDICTION MODEL
                                                                                                           13
Labor  (Bureau of Labor Statistics);  the  State  depart-
ments  of  commerce;  local chambers of commerce;
various State  and local  agencies concerned with  com-
mercial and industrial development.
   Inventory of Input Items. Block  / is  computed by
multiplying the  expenditure per unit time  for  each
item (from Summary of Expenditures, block c) by the
unit cost  of  the  corresponding item  (Price  per Unit
Amount  Conversion  Factors, block d). Thus, block /
gives the number of units of each item  entering the
study area per unit time. The amounts are expressed in
terms  of whatever units of measure  are  most appro-
priate  for the particular  item  (e.g.,  dozens of  eggs,
loaves of bread, pounds of coffee, pairs of shoes).
   Price/Amount  Conversion Factors. These data (block
d) are derived in a subroutine that establishes the average
price per unit amounl,  the different  sizes and types of
units  commonly  produced and marketed being taken
into account. This subioutine is also used to estimate the
"average  composite  package unit," which is used in a
subsequent computation. The data for these subroutines
come from a wide variety of sources, including surveys.
Care must  be taken to ensure that costs are taken for a
common year; in addition, regional variations may have
some effect on cost.
   The preceding  discussion has been  limited to that
portion of the model pertaining to the development of
the inventory of input items to the household and is
applicable  to all  inputs  other  than LRT  consumer
items,  i.e.,  SRT  consumer  items; the containers and
packaging  materials  that accompany  SRT items; the
containers  and  packaging  materials  that accompany
LRT items (Figure 3).
            Prediction  of  Potential Wastes
    The  following paragraphs are concerned  with  the
means of  using the  input  inventory  to  develop  waste
predictions.
    Block i of Figure 3 represents a  tabulation of in-
formation that   describes   each  item  in  terms  of its
material   composition  (i.e.,  pounds  of glass,  steel,
plastic, paper, etc). It is at this point in the model that
each  item's consumption  rate  would  be defined  in
more  detail.  For example, W  Ib  of coffee  per  week
would be  defined as X Ib  of granular solids, Y Ib  of
tin-plated  steel  cans, Z  Ib  of plastic lids. These values
will be the "average" for a composite unit size package
and will be based on the market coverage (i.e., distri-
bution)  of sizes  and types of packages.  For example,
50 Ib  of coffee  cannot be interpreted as 50 cans  of
1-lb size.  Rather, it  will consist of some mix of 1/2-,
1-,  2-, and  3-lb metal  cans, plus some mix of special
paper bags. Since each  of  these alternative containers
will  affect  the household  waste  differently,  it  was
necessary to devise a  simple  method to account  for
them. The method used  here was merely to construct a
hypothetical "pound of coffee,"  m  percent  of which
comes in one size and  type of container, n percent of
which comes in another size and  type, and so  on.  An
alternative method  would be to consider every size and
type  available  on the market  as  a  separate item, but
the problems of data storage (and availability) rule out
this  approach  immediately. In this  study, descriptive
information was  obtained  primarily  by collecting, dis-
assembling,  and  weighing  a  multitude of  items  and
their  associated containers and  packaging. The market-
ing information of blocks /, g, and h  provides a basis
for assigning weighting factors  that  reflect the  relative
importance  of various sizes  and  types of items available
(i.e.,  the example  of ground  coffee  being  considered
again,  marketing  information^  indicates  that  major
emphasis should  be placed  upon  the 1-  and 2-lb cans
and only minor emphasis on bags).
   Combining the Input Item Inventory (block /) with
the description of each  item (block /) yields a Descrip-
tive Inventory  (block k) of  the  material that enters  the
household. This inventory indicates, in separate tabula-
tions, both materials and items that enter the house-
hold.
   The Waste/Item Conversion Factor (block /) is intro-
duced to indicate  the  primary fate  of materials that
enter  the household.  The  various alternatives are as
follows:  solid waste; airborne wastes (e.g., aerosol-type
hair  spray,  air deodorants); liquid, i.e., sanitary waste
(e.g.,  cleansers, food, tissue); distributed within resi-
dence so as to be  "lost" (e.g., floor wax); distributed
outside  residence so as to be "lost"  (e.g., fertilizer,  pet
food); prolonged  "storage."
   Those items not  entering  the  solid  waste  stream
must  be  carried  in  the original inventory because of
the need to include their containers as solid waste. It is
anticipated  that m the computerized model this func-
tion will be included  in block h, the data bank that
gives  the composition and  content of items.  Applying
the waste/item conversion  factors of block / to  the
inputs  (i.e., combining blocks k  and  /) yields  the
Descriptive  Inventory of Potential  Waste Items  and
Materials (block «).*
   *Note that the waste materials listed in this inventory are still
keyed to the parent item from which they originated.

-------
14
                                                             PREDICTING SOLID WASTE CHARACTERISTICS
                          
-------
HOUSEHOLD-WASTE PREDICTION MODEL
                                                                                                   15
  Ill
        i ss
         3i g
         o j I
          o
            § 5
            i S.
                O i "
                £ o ~ ~
                £ E g
                  =" ? -5   -
                           *1
                            S_J S
                        5 ±-2
                        Hi
                        n ^ 9"
                         ° D -C
                         ^,-C +-
                                      &-£
                                      
o -
  ~D .
-a ' -
"o S
1-8
2,-

                       3
                       ^ ^
                       D i
•|
i i-
O M_
£ S
g £
"c ^
D >*-
S g
J °
5 ^3

1^
O v
£ ^
^ '"
^-^
-O o
D ^i;


IS 5>
— 0)
Of O


J, 1
J3
D
•5 °>
> QJ
O ^_
i^^
? =-S
| — _Q

a> ~ oj
«T 5 S
_c ^ —
£ R ?
-a
1> -Q
^ i
g >
O 4i
u -O
OJ C
£ ^

E Z S
a) u "D
i I §
"° — a.
^ ai r-
r~ ~a ^
l!i
c *• s
c "o -
111


; S E

01 ~
E IfE "o
^ — i_
>T . -Q
^ -^ I
Q."o C
i- E _£
"?:| ^>
a Li c
QJ ^O *-
S g g
jr ji c
u o o
S - u

"° | oi
01 D C
z « "?
"£ > c
- - p
- g b
T) £ "*-
S °J
- Q. >-
_£- i" _C
U L, m
O O jz
fc w O
,2^3
^ k_ £
- 3 _Q
"c - T3
•B If

Ps
-^ 1 1
" t -
El"fe •£
S1 C v
~ S
•B o-g
OJ 0 -
a- aj

*- -c p
7 ^ «
§^ S
S -C -E
-s «> *"-
8*|
*O tj °
cu u
S>= -c
? £ S
S _c ^t
O - n
U -C J?
~ * ~
-£ F* en
3 £ «
> •*- ^_,
                                               ll
                                                     « « ^ &>-a
                                                     *-   o o ^
                                                      E  a) c c (5
                                                    -£ a ° I ^

                                                    ^ x S !J^
                                                     c^rr c ^ -a

                                                     - _c E
                                           11 — JD a

-------
16
         PREDICTING SOLID WASTE CHARACTERISTICS
   Block  m  has  been  included  to  allow  the  direct
introduction of data on wastes that are best estimated
outside the  model.  Examples  of wastes handled this
way include: newspapers and periodicals; mail; grocery
bags and cartons  used  to transport groceries; leaves,
grass,  and  garden cuttings;  floor  sweepings.  These
wastes are considered separately because their quanti-
ties  are not  dependent upon the household's expendi-
ture on a particular consumer item.*  Newspapers and
mail are estimated on the basis of information  from
local publishers and postal authorities; grocery sacks on
the  basis  of national production  figures; leaves,  grass,
garden cuttings, and sweepings on  the basis of informa-
tion concerning population density, income,  and the
degree to which the study area is built up. To date, we
have not been  able to develop sufficient information to
allow us to  make quantitative predictions of how the
generation  of  these  materials  will  vary with   given
demographic factors, although the experimental studies
of waste composition conducted and sponsored by the
Bureau  of  Solid  Waste  Management  should  be very
useful.
          Prediction of Solid Waste Quantity
   Block  o consists  of  a series of distribution factors
that  indicate the   ultimate fate of  solid wastes and
reflect  the  manner  in which  households  typically
handle  waste  materials.  These  factors  indicate  the
portion of each spent item that will be disposed of via
each of the various disposal alternatives.  For example,
it  might state  that,  from an  input of 1  Ib of ground
coffee,  20 percent  of the  spent grounds would  enter
the sanitary  sewer (via  the  sink, presumably),  80 per-
cent  of the  grounds  would  be disposed  of  via  the
garbage can, and  100 percent of both  the metal  can
and  its  plastic lid would  leave  the garbage can. The
influence  of  other  local practices,  such  as  whether
backyard  incineration  is  permitted,  would  also  be
incorporated in this  computation.  These  distribution
factors  of block o  are  based upon the  following: in-
formation  concerning the waste management practices
of various typical  types of households  (block p); in-
formation  giving the number  of households  of each
such  type  located  in the study  area, as given by  the
community descriptors (block a).**
   Applying these distribution factors to  the inventory
of potential wastes  (i.e., combining blocks o  and ri)
yields a series  of  new  tabulations (blocks q,  q,  q",

-------
                                               ^(f
              MODEL
                              FOR
   The primary objective of this study was to develop
selected parts of the waste prediction concept into an
operable technique and to test its feasibility by apply-
ing it to a real community. In the interest of achieving
these  goals  as fully as possible within existing time and
funding constraints, it was necessary to assign priorities
to those portions of the  model that  relate to the field
data against which it was to be compared.
   After several possible  alternatives were examined, a
study of solid  wastes in parts  of  Jefferson County,
Kentucky,  was selected  for  comparison.  This  study,
directed by  Professor John  Heer at  the  University of
Louisville, was conducted in such a  way  that  residen-
tial-household solid wastes were isolated and measured
independently of wastes from other sources.7 Since the
primary  emphasis  in the  URS study  was  in  the
household-waste-prediction  model, it  was  decided to
concentrate  on predicting  the  equivalent waste com-
ponents.*
   The following paragraphs discuss  how the particular
functions  diagrammed  in  the  basic  household-solid-
waste-prediction  model were performed in developing
and applying the model for the test case.
         Community and Expenditure Data
   Data for block a  of Figure  3 (the description of
households   in  study area) were  obtained primarily
from Bureau  of Census publications. The most impor-
tant information  included  the  population, the average
size of the ho use ho Id unit,** and the general geographi-
cal location.!
   Data  for  block  b  (Consumer Expenditures  by
Household)  were obtained  from  the  Life study dis-
cussed  previously. For our purposes, not all the detail-
ed data were of value, since a large portion pertained
to expenditures on items that do not contribute signifi-
cantly  to  household solid wastes. If the Life data are
considered as comprising  a  matrix,  the  portion drawn
upon most heavily by our prototype model related  to
the following: food, beverages, and tobacco; household
supplies;  personal  care  supplies; clothing and acces-
sories;  other SRT items.  Some 130  consumer  items
were considered  in  detail  to determine how each con-
tributes  to  waste  generation. The major household
characteristic considered  in  the  test  case was  "geo-
graphical  region."  Expenditure data  for the North
Central region of the United  States were  adjusted  to
reflect  the fact  that Jefferson County  tends  to  have
somewhat larger  households than the national  average.
It was originally  intended to  consider several income
levels independently, but  for the reasons discussed on
page 20 ,  only the  average  income  for  all households
was used.
   *Exammation of University of Louisville's preliminary results
indicated  LRT  items  do not contribute significantly to resi-
dential waste.  For  this  teason  and because of the  problems
discussed in a previous section, no attempt was made to estimate
waste derived from these ilems
  **The average  household  size tor  Jefferson County is 3.42
persons The national average is 3.36 persons.8 The conversion
factors 3.36/3 42 was applied, to adjust data represented on a
national basis.
   tThe portion  of the Life study  data  that applied to the
North Central section of the United States (as defined therein)
was used in this study
                                                    17

-------
 18
          PREDICTING SOLID WASTE CHARACTERISTICS
        Material Price and Composition Factors
   Data  for  block  d  (the  price/amount conversion
factors) were generated with information  such  as  that
contained  in  blocks  e, f,  and g but did not employ a
formalized  mathematical  model.  Rather,  information
concerning  the  market  coverage  and  prices of various
sizes  and types  of consumer items were  manipulated
by  hand  to  obtain  price-versus-amount conversion
factors. Coverage  and price information were obtained
from  various sources, including  production  statistics,
wholesale  and retail  trade statistics,  and  communica-
tion with  persons employed in the  appropriate  retail
businesses.6'9  The  decision to generate  the conversion
factors  by hand rather  than  by a mathematical model
was made  in the  interests  of  maintaining  as much
flexibility as possible during this developmental stage.*
   Block  i  (the  description  of  each  item's  physical
characteristics and material composition) was generated
with information  such as  that contained in blocks/ g,
and h.  Block h,  the descriptions  of size  and type of
each item, was  developed by actually  dissecting several
hundred common consumer  items and weighing  (or
otherwise measuring) their respective component parts
and materials.
   Since the present  study is concerned primarily with
the development and demonstration  of  a  prediction
technique, it  was  decided to describe wastes in terms
of  material  classifications (Table 5).** The particular
classifications (and aggregations) used  were selected to
allow  the  model's  predictions to  be  compared with
measured waste generation data from the University of
Louisville study. The result of this item-by-item analy-
sis  is a large and rather  comprehensive tabulation of
characteristics (block /).
   Data for block / (the waste/item conversion factors)
were developed entirely on the basis of personal exper-
ience,  rather than  from specific  data  sources.  The
purpose  of  including these  factors   is to  provide a
means of accounting  for the  fact  that many items are
   Table 5. Material classifications by which wastes are
 described in prototype model
   *Because  of their size and complexity, the tabulations of
data and results of computations are not included in this report.
Copies can be made available to authorized requesters.

  **The degree of detail with which items are described should
be dictated by the use to which the waste prediction information
will be put. Thus, in a comprehensive survey, pursuant to design-
ing systems  for waste collection, transport, treatment, and ulti-
mate disposal, it  may be necessary to describe all waste con-
stituents in terms of their material composition as  well as their
physical,  chemical, and  biological properties. In  less  than
comprehensive studies, it is not necessary to describe wastes m
such detail.
Combustible materials
      Paper and cardboard
      Plastics
      Rubber
      Textiles
      Leather
      Wood
      Leaves and grass
      Other combustibles (excluding garbage)
Food-derived gaibage
Metals
      1-errous metals
      Aluminum and Us alloys
      Copper and its alloys
      Other metals
Glass
Refractory materials (exeluding glass)
purchased  in one  form  but  discarded  in a very  dif-
ferent  form. For  example firewood is  converted  to
ashes,  fertilizei  to  garden cuttings, and clothing to  lint
and rags.
             Extraneous Input Information
   Block m (the mechanism for introducing extraneous
information  to  supplement  the  model)  was  most
valuable  in  accounting for the following  newspapers
and  periodicals; mail  (excluding  periodicals);  grocery
sacks and  other paper bags; and  aluminum  foil (that
amount  not  elsewhere  accounted   for);  aerosol-type
metal  cans;  miscellaneous  plastic bottles (those  not
elsewhere  accounted  for); and  textile materials. The
advantage  of handling the first three entries in  this
manner  has already  been discussed.  The  remainder
were  introduced in  Ihis  way  primarily  because they
tend to enter the household in association with a great
variety of consumer items. This approach,  that of using
production  statistics  for  the materials themselves, will
provide reliable estimates of their rate of  consumption
by  the  household,  provided the  materials  are used
somewhat uniformly  by all classes  of  households.
   In this study, the data  for block o (the distribution
factors  to  reflect  household waste management  prac-
tices) were developed on  an  intuitive basis rather than
on the basis of blocks a and p. As mentioned previous-
ly, at the present state of the art, sufficiently detailed
information  is  not  available  for predicting  how  a
household's  waste  management  practices  vary  relative
to measurable demographic and socioeconomic factors.

-------
MODEL DESIGN FOR TEST CASE
                                                   19
For the purpose  of this  study,  however, such  a  defi-
ciency  appears  to  be of little significance. In  order to
yield results that could be  compared with those from
the University of Louisville study (which accounted for
the wastes that were found in home garbage cans), the
assumption was made that wastes would be distributed
throughout the various disposal  alternatives in a man-
ner that the investigators assumed  was typical, judging
from their own experience. It was assumed that virtual-
ly  all disposable packaging  materials, small containers,
and paper goods would be  placed  in the garbage can.*
In  addition, it  was  assumed  that  the  fractions  of the
items  that could  conceivably be  segregated and saved
for some other form of disposal  (e.g.,  15 percent of
returnable glass bottles, 75  percent of old clothing, 50
percent of newspapers)  would also be discarded via the
garbage  can.  These  assumptions are only  approxima-
tions but must  serve until  more definitive information
becomes available.
   The prototype's output takes the form of a series of
waste inventories (blocks q,  q , q", and q'"),  one for
each  of  the  major  alternative  means by which house-
hold  wastes  are  disposed of.  The  format  of  these
inventories is  the weight of materials corresponding to
each of the material classifications given  in Table 5. -
   *Some of these disposable materials will not be disposed of
via the home garbage can. The materials associated with lunches
packed in the home but eaten elsewhere are examples of this. In
this study, the only materials for which the quantities discarded
at  home  were  reduced  below the total quantities were wax
paper, beverage  cans, and  "no-deposit, no-return" beverage
bottles.

-------
   As indicated  in the previous section, the solid waste
generation in residential  areas of Louisville was  esti-
mated to  check the feasibility of the waste prediction
technique.
   The University of Louisville (UL) study involved the
periodical  collection  of  the  contents of  refuse  con-
tainers  from  some  144 households throughout  the
study area. The particular households sampled and the
collection  schedule  employed  were  both  carefully
selected to obtain representative  samples of residential
solid  waste.  These  samples  were  then  assayed to
determine both quantity and character, as expressed by
a number  of physical  and chemical indices.
   Even though our study and the UL study were both
directed  toward solid  waste  generation from house-
holds in  the same community,  certain  factors  were
expected to interfere with their direct comparison. For
example,  although  the UL  study considers  the  same
waste  items  considered  in  the  URS  study,  it  can
account for them only if the householder disposes of
them via  the refuse  container. That  is,  if the house-
holder disposed of the waste in  some  other  fashion
(e.g.,  by  separate  collection  or  perhaps on-site com-
posting  or  burning),  it  was  not  counted. Similarly,
wastes disposed of by  any  route other  than the com-
mon  refuse container (e.g., those  disposed  of  via the
sink  garbage  grinder, segregation  for salvage,  on-site
incineration) would not show up in  the UL samples.
These  differences  between  the two  studies  are cited
primarily  to  indicate that each method is  based upon
simplifying assumptions that  impose limitations, some
of  which  complicate  the  task  of  comparing  their
respective results.
   In many  ways  Jefferson County  was a good test
community:  it constitutes part of an SMS A,* an area
for which a rather large amount of statistical  data are
readily  available;  it  is  neither  entirely isolated  geo-
graphically nor is  it submerged within a megalopolis; it
has had a reasonably stable population  growth pattern
and  does  not,  therefore, exhibit  some of the  diffi-
culties of areas that have "boomed" or are declining in
stature.
   A  primary goal of the UL study was to determine
the relationship  between  a household's income and its
waste  generation.  To  accomplish this,  the  sample
households  were divided into three income brackets,
and  waste  samples were collected  and  analyzed  to
determine the manner in which their  amount  and con-
tent correlated with income level.
   The  initial plans for this  study included a similar
goal.  It had been  intended  to take advantage of the
fact  that our  expenditure data include information on
the manner in which household consumption  patterns
vary as a  function  of income level.3 It was planned  to
use this information in our model to  develop  separate
waste  predictions  for  each  of three  or four income
levels.  The  preliminary results of  the -UL study indi-
cated,  however,  that, though waste generation is affect-
ed by income level, the statrstical  correlation  between
the two was barely significant in the Jefferson County
   *U.S.  Department  of Commerce has designated Jefferson
County, Kentucky, and both Clark and Floyd Counties, Indiana,
as the "Louisville, Kentucky-Indiana  Standard Metropolitan
Statistical Area (SMSA). The principal cities included are Louis-
ville, Kentucky, and New Albany, Indiana
                                                     20

-------
TEST RESULTS OF PROTOTYPE MODEL
                                                                                                            21
households  considered. For  this reason, it was decided
not  to spend the limited available time and funding on
a  similar effort. Rather  it was  decided  to  develop
income-independent predictions.* The basic household
consumption information actually selected was that for
all nonfarm households  in the North Central  quadrant
of  the  United   States.  For the  purpose of  drawing
comparisons, the UL  data for  all income groups were
averaged.
                    URS Results
   As  described  previously,   consumer   items   were
identified as  being either SRT  or  LRT. During the
study the waste content  of approximately 120 items
was determined.** The amount of household waste con-
tributed  by  each   of the  purchased  items  was de-
termined  for each  of the  constituent materials.  For
example, the amounts of  paper, plastic, and aluminum
foil  attributable to the consumption  of  two  separate
items, say breakfast cereal and  laundry detergent, were
determined   and  tabulated  separately  in  terms  of
pounds  of  waste  material  per  household per  week
(Table 6).
   Block m in the  model  was described earlier as pro-
viding  a means  of incorporating generation  data for
waste  components  whose  amounts  are  not  closely
related  to   any  particular consumer  expenditure rate.
The final row of entries  in Table 6 consists  of totals
calculated  for  "block m" wastes, such as mail, leaves,
grass, garden cuttings, and miscellaneous wastes.f
   *This  study of expenditure pattern data has  led to the
conclusion that income  level is perhaps the  most  significant
single factor influencing wastes. It is recommended that further
studies be directed toward determining the impact of income on
waste characteristics.
  ** The list of items comprising Table  3 was pared down by
eliminating LRT items (and the packages for same), items with
insignificant residential solid waste content, and those for which
the expenditure per household was less than  $0.05 per week.
   •fBased on  U.S. Post Office statistics, it was estimated  that
some 7.1  Ib of paper and cardboard per  week enter the typical
Jefferson County household via the mails.10'1' On the basis of
preliminary  figures  obtained from  Bureau   of Solid  Waste
Management, it is estimated that some t  7 Ib per household per
week of yard wastes and 1.1 Ib per household per week of waste
wood from various sources are to be expected. Circulation in-
formation and newsprint usage estimates were provided by the
publisher of Louisville's principal newspapers.'1  2
        University of Louisville Study Results
   Because  each of  the  households  in  the  UL study
was  sampled several times during the study period, the
observed waste generation data were expressed in terms
of a statistical distribution (Table  7 and Figure  4).
   The  UL results also  differed in  format from  ours.
They segregated waste materials into 5 categories  (con-
sumer items in  Figure 4), whereas the  present study
employed   15   material  categories.  The  relationship
between the two  systems has been indicated (Table 8).
The  lack  of conflict  between the two  systems stems
from the  fact that we were aware  of their format and
designed  ours  to  be  merely an expansion of  it. The
expanded format  is,  of course,  desirable in  that  it
provides the degree of flexibility necessary to make the
data applicable to a wide range of  uses.
                Comparison of Studies
   The waste  generation rates  of Table  6  have  been
summed  into   five  categories   (see  Table  8).  which
correspond  to  those  in the UL  report, and are listed  in
Table 7. Those summations estimated in  this study have
been plotted  as   bold   dots  in  Figure  5 to  allow
comparison with the  UL results. Note that each of the
predicted values is smaller than  the corresponding  value
measured by UL.  This would be expected, since none of
the predictions are complete to date (i.e., for each, there
are items that have not yet been  fully accounted for),
   We believe  that the  performance  of  the prototype
computation is satisfactory. The discrepancies between
the URS study and the  sampling study are well within
the  expectation  for  this  preliminary trial of  a novel
technique.  Furthermore,  the  discrepancies  are  ex-
plained  by  the fact   that the present model is based
upon only  a minimum amount of information. It was
stated previously  that the prototype model used for
the  comparison study considers  only SRT  items and
packaging materials and excludes  LRT items. It is be-
lieved that, given  the  opportunity to include these and
other excluded items and  to refine various  other pre-
liminary  estimates  and   approximations,  the   waste
generation rates predicted by the model would progres-
sively approach the actual average  rates.*
    t The actual average rates are not necessarily equal to those
measured since measurement techniques usually suffer from
sampling and other errois.

-------
22
                                                                   PREDICTING SOLID WASTE CHARACTERISTICS
                                                                                                  2 £  ~
                 *
                  e
                 •u

                 03
                 aj

                 "es



                 X
                 .0
                 V
                 e
                 u
                 00
                 c
                 _o
                 '•*-!
                 o
                 •3

                 £

SJEU3JCIU
iOBFTOl
-0
S]CJ3UI
puB wddoj
sAo[(r sji
ptm uinuiuinjv
stcjaiu

snoiMj
p3AutqpooS,
S3|qijsnquioo
J3l{JO
SSCJ3 pUE


pooA\


J3i;jE31


SSJfJX'O ]

»«.»
SJ.li.IJ

pwoqpreo
pire isdc,]

Consumer
items


0
on


0)
O
^.
o
fN
.^













2
o
?
-^

F'ood, beverages,
and tobacco


-




m
._
0













4—
O
O
rJ
vD
O

Household
Supplies


ON
6



•i—
^
Q
0


"o
c
o
E
rt
C
C
M

O



^

(N
O

Personal care
supplies


















-i^

o
o
"-1
-





Clothing and
accessories





^0
o
o
10

o


o
r--
-1

~




VO
^t
o

~
o
01
2

Wastes from other
sources (block m


3


oo
O
SO
rn
0)


O
r-
^^
o

""



^
'^i'
^

en
O
00
CO

c5

o
3
^
,





O







?










                                                                                                       o ^
                                                                                                  "53 ^ ^ E

                                                                                                    ,2 c B
                                                                                                  s    M S
                                                                                                  "*"* "*^ %

                                                                                                  b "^ c S
                                                                                                  Z 5  2 ».


                                                                                                  fill
c  c ° 'S
^  cd « ?3
3  S ^ S


- 'I ^
rt 'S ^ !>>

5 O -™ O
-u ^j -,, be

Isll

2 -s-e °

S£ 3>-P
     (D
                                                                                                  ° ^
 '•a a"3
 1  c c S
   6J) 5 ^-
  •a S «
                                                                                                  - .S -2 S

                                                                                                  £ 60." 1*

                                                                                                  H C T3 «
                                                                                                  
-------
  TEST RESULTS OF PROTOTYPE MODEL
                                                                                                         23
                     Table 7.   Comparison of waste generation rates (Ib/HOUSEHOLD/wk)
Material
Paper
Garbage
Glass
Metal
Minerals

Low
249
4.8
3 8
3.9
0
Measured by Univ. '
of Louisville study
Medium
29.7
9.9
53
4.7
0.7

High
37.8
12.6
6.7
6.0
3.1
Estimated by URS
prototype model
23.43
Not estimated
2.54
4.23
Not estimated
PAPER
GARBAGE
GLASS
METAL
MINERALS
                   10
                             20
                                                       PAPER
                                                       GLASS
                                                       METAL
                                                                       10
                                                                                 20
                                                                                           30
                                                                                                     40
                                       30
                                                  40    HOUSEHOLD SOLID WASTE GENERATION RATES (Ib/HOUSEHOLD/wk)
HOUSEHOLD SOLID WASTE GENERATION RATES (Ib/HOUSEHOLD/wk)

             |     |   RANGE OF ALL MEASURED VALUES

               *      MEAN
•     VALUE PREDICTED BY URS PROTOTYPE MODEL

   ]   RANGE OF ALL MEASURED VALUES

*     MEAN
    Figure 4.  Solid waste  generation rates observed by      Figure 5. Solid waste  generation rates  predicted by
  University of Louisville sampling study.                   using URS  prototype.  URS results are superimposed
                                                        over University of Louisville results.

-------
24
                                                            PREDICTING SOLID WASTE CHARACTERISTICS
                          Table 8. Comparison of output data formats
University of Louisville URS prototype
waste categories waste categories



Paper fS^^""


Garbage — =dl____

Metal <^^
Paper and
cardboard
Plastics
Rubber
Textiles
Leather
Wood
Leaves and
grass
Other combustibles
(excluding garbage)
F'ood-denved
i 	 garbage
Ferrous metals
Aluminum and
its alloys
Copper and
its alloys
\ 	 Other Metals
Glass -~=^r^H^ Glass
,. . 	 (~ Refractory materials
Minerals -=dl~_~ , , , ,
	 1 	 (excluding glass)

-------
                                CONCLUSIONS
                                                                         ^m mHli^ffi ^H wH^V^
   The solid waste prediction  method  investigated in
 this project is technically feasible and appears to offer
 a unique and effective means of estimating waste quan-
 tities  and  characteristics.  In addition, it  provides  a
 framework for the prediction of future wastes. Table  9
 indicates the relation of the  progress to date (phase 1)
 to the overall requirements  of the program and sug-
 gests  that  subsequent work be undertaken in two
 phases. Phase 2, the next logical step, would be
 devoted to completing the  capability to predict the
 solid wastes from residential, commercial, and institu-
 tional activities in the community. These wastes make
 up  a  large fraction of the solid  wastes  arriving  at
 disposal sites and  are very  susceptible to changes (as  a
 result   of  changing  technology)  that  affect waste-
 handling and disposal operations. Moreover, pertinent
 input   data are  generally  available  concerning these
 wastes. For these  reasons,  it appears desirable to give
 priority to these  aspects  of the  work. In phase 2,
 however, the problems  of predicting solid wastes from
 industrial, agricultural, arid  other sources would also be
 investigated, at least through the stage of a test of the
 approaches. The major and final development of pre-
 diction methods for  wastes from these  sources would
 constitute the third phase of the work.  There are two
 reasons for  relegating  these to  phase 3. First, current
 work  (being done by other contractors) as well as
 planned studies concerning industrial and agricultural
wastes  should be used  as a basis  for  the  proposed
 prediction methods. The results  will not be available to
 an appreciable  extent during  the phase  2 period.
Second, the  requirements  for  a computerized solid
waste  estimation procedure for industrial, agricultural,
and possibly other wastes  have not been  defined  or
delineated. Both of these  deficiencies would be re-
solved by the  preliminary studies  in  phase  2, and
decisions concerning the need  for phase 3 and specific
approaches to be taken therein would thereby be per-
mitted.

   The major tasks  that should be undertaken in phase
2 are the following:
   1.  Complete the development of the  residential-
household-waste-generation  model,   including  those
aspects related  to  LRT items and  the prediction  of
future wastes.
   2.  Adapt the model to  handle the  wastes gen-
erated by commercial, institutional, and other activities
that are found to be major contributors.
   3.  Perform  preliminary  model design and conduct
a feasibility study for prediction of industrial, agricul-
tural, and other  wastes.
   4.  Establish computation specifications; reevaluate
preliminary  residential  model; relate desired model
specifications to computer capacity and operating cost;
firm up computer model specifications; establish input,
output, and function formats.
   5.  Develop  computer program;  prepare  flow dia-
grams  and design overall program; prepare and  debug
programs; verify running times.
   6.  Expand standard data banks;  collect, evaluate,
and collate as many standard  activity and commodity
descriptions as are pertinent to the test area (see task
7); put data in computer format.
                                              25

-------
26
          PREDICTING SOLID WASTE CHARACTERISTICS
                               Table 9.  Prediction method status and proposed work
Waste types
Residential, SRT
Residential, LRT *
Residential, future
Commercial
Institutional
Industrial
Agricultural
Other
Development stages
Conceptual
design





Preliminary
data collection
Phase 1 (complet




Manual
test
ed)




Final design
program for
computer

Phase 2




Complete
data collection



Phase 3

Computer
test





     *SRT-short-residence-time items
      LRT-long-residence-time items

   7.  Select test area; develop test  area data; collect,
 evaluate,  and compile input  commodity and  activities
 inventories  and  schedules for  area,  including popula-
 tion, land use,  and  other  changes;  put  data  in  com-
 puter format.
   8.   Run waste  prediction  for  test area; compute
several pertinent predictions  for a range of conditions.

   9.   Evaluate   results;   compare   predictions  with
independent estimates or surveys; prepare report.

-------
                    REFERENCES
  1. Office of Statistical Standards, U.S. Bureau  of the Budget. Standard industrial
        classification manual. 1967. Rev. ed. Washington, D.C.. 1967. 615 p.
  2. Urban Renewal Administration, Housing and Home  Finance Agency, and Bureau of
        Public Roads,  U.S. Department of Commerce. Standard land use coding manual.
        1st ed. Washington, U.S. Government Printing Office, 1965. 111 p.
  3. Linden, F., ed. Expenditure patterns of the American family. New York, National
        Industrial Conference Board, 1966. 175 p.
  4. Sindlinger and Company,  Inc.  1967 brand inventory  of household durables in
        Delaware Valley, U.S.A. Philadelphia Inquirer, June 1967. 30 p.

  5. Automobile  Manufacturers Association.  1968 automobile  facts  and  figures.
        [Detroit], 1968. 70 p.

  6. Personal communication.  T.  Healy, Safeway Stores, Inc., to  G.  B.  Boyd, URS
        Research Company, July 1968.
  7. University of Louisville.  Louisville, Ky. -Ind. metropolitan region   solid  waste
        disposal  study: interim  report on a solid waste demonstration project; volume I:
        Jefferson County, Kentucky.  [Cincinnati].  U.S.  Department   of Health,
        Education, and Welfare, 1970. 205 p.

 8. Sales management's survey of buying power. Sales Management, 98 (12): 1,3,5...
        15. June  10, 1967.

 9. Bureau of Census,  U.S. Department  of  Commerce.  Special report on household
       ownership and purchases of  automobiles and selected  household durables,
        1960-1967.  Current  Population  Reports  Series  P-65:  Consumer Buying
       Indicators,  No. 18. Washington, U.S.  Government Printing  Office, Aug.  11,
        1967. 24 p.

   *Perhaps one of  the most valuable results of performing the present study was the collection and
evaluation of a large body of statistical information concerning community inputs and activities.
More  specifically,  this information consists of industrial and agricultural production rates,
commercial sales activity, personal consumption patterns, and the like. Whereas the  amount and
detail of information are not readily  describable, some feeling may be obtained by considering that
more than 200 documents constitute the basic bibliography (all of these  contain information that
has direct bearing on predicting waste generation). For simplicity, however, only those sources of
information that precisely document the text of this report have been cited herein.
                                   27

-------
10.  Personal communication. B. Ehrler, Postmaster, Louisville, Ky., to G. B. Boyd, URS
        Research Company, Oct. 1968.

11.  Personal  communication.  R.  R.  Germain,  U.S.  Post Office  Department,  San
        Francisco, to G. B. Boyd, URS Research Company, July 1968.
12.  Personal  communication.  L.  A. Kern,  Traffic Manager,  Louisville  Times  and
        Courier-Journal, .o G, B. Boyd, URS Research Company, July 1968.
                                                                              14
                                     28
                                           II. s. (,OYJ KTxMl NT PIUNTD.f. PI-FICK  IH71 O - 447- 2if4

-------