-------
-------
-------
-------
1. PURPOSE OF SURVEY
OSW's purpose in conducting a national mail survey of
RCRA-regulated hazardous waste handlers was essentially three-fold:
The primary purpose of the survey was to charac-
terize the populations of hazardous waste generators
and treatment, storage, and disposal (TSD) faci-
lities regulated under Subtitle C of RCRA. The
RCRA regulations apply to nearly every segment of
American industry and to a wide variety of waste
management processes. Due to the lack of available
data, the survey was conducted to identify and
assess the attributes and characteristics of
these industries and processes.
A second purpose of the survey was to develop a
national data base on hazardous waste management
practices for use by OSW and others in the con-
tinuing development of the RCRA regulatory program
and assessing its impact on the regulated community.
OSW was required under Executive Order 12291 to
conduct regulatory impact analyses of its major
hazardous waste regulations. The development of
a national statistical data base on hazardous
waste management practices is integral to the
completion of these required analyses.
The final purpose of the survey was to estimate
the magnitude and "scope of hazardous waste genera-
tion and its treatment, storage, and disposal in
the United States. Previous estimates of the
total quantity of hazardous waste generated annu-
ally have varied substantially. This survey was
intended to provide a baseline estimate of
hazardous waste generation and treatment, storage,
and disposal quantities in 1981.
-------
Figure 1
PURPOSE OF SURVEY
Characterize Hazardous Waste Handlers
Regulated by RCRA:
Generators
Treatment, Storage, and Disposal Facilities
/
Develop Data Base to Support:
Regulatory Development
Regulatory Impact Analyses
Regulatory Review
Estimate 1981 Hazardous Waste
Quantities:
Generated
Treated
Stored
Disposed
-------
-------
SURVEY SCOPE
National Survey of Hazardous Waste
Handlers
1981 Hazardous Waste activities only
Only those who registered with EPA
(Motif lets & Part A permit)
i
Only RCRA covered hazardous waste
management:
Therefore design excludes:
Small generators
90-day Accumulators
Wastewater treatment
in tanks
Deiisted wastes
Other Exempted Handlers
and Waste Streams
-------
-------
Figure 3
SURVEY METHODOLOGY
Pre Survey Screening:
-TSD process type in 1981 verified
-Proportion of generators not producing
hazardous waste in 1981 estimated
i
Sample Design:
-For national projections
-Stratified (region, SIC, hazardous
waste quantity)
-Separate samples for generators &
9 different TSD process types
Data Collection Method:
-Mail Survey in fall 1982
-Extensive telephone call backs
Quality Control Efforts:
-Computer Edit
-Telephone data gap/correction
-Largest handlers rechecked
-Internal consistency checks
-Units of measure standardized
-------
SURVEY OPERATIONS; MAIL OUT, FOLLOWUP, RESPONSE AND
DATA CODING AND EDITING
Each TSD facility received a TSD general questionnaire
and one or more questionnaires specific to the technology type(s)
for which it was sampled. Most generators received only the
generator questionnaire. The questionnaires were developed by
Westat in conjunction with EPA staff analysts. They were pre-
tested in June of 1982 and received approval from OMB.
The questionnaire package included instructions and
i
lists of hazardous waste codes for use in the questionnaire.
Packages were sent by registered mail in September of 1981. By
RC5A regulations, facilities had 45 days to respond. Both Westat
data collection specialists and the RCRA Hotline at EPA provided
technical assistance via telephone to the respondents throughout
the response period. Many respondents 'were granted time exten-
sions to dates up to February 1, 1983. All TSD facilities that
did not respond in 45 days were sent followup letters, and,
later, were contacted by telephone. Nonresponding generators
were mailed followup reminder "letters in January, 1982.
-------
Statistical tabulations were run on edited data to
minimize errors that might have been caused by respondents,
coders or key entry operators. The editing process included:
manual editing and visual examination of the coding work; computer-
ized .edit to check all data items for legitimate range of values;
computerized "logic edits" to check for consistency between many
(but not all) questions; and visual examination of frequency
distributions to identify any suspect cases. In addition, a
major effort was made to resolve questionable responses to impor-
tant variables. -This was done by. a "data.retrieval" effort that
involved telephone call-backs to approximately 30 percent of the
respondents.
/
In spite of this substantial editing effort, it is
inevitable that some errors might remain in such a large national
data base representing a wide-diversity of industries reporting
detailed technical data on complex waste handling operations.
Even the telephone call-back efforts intended to resolve apparent
conflicts were not always successful in obtaining the requested
data on selected items. Nevertheless, we believe that, the many
stages of data editing and cleaning, together with the carefully
worded questions, have produced an edited data base, substantially
'free of processing, coding and entry errors.
-------
4. SAMPLE SIZES AND RESPONSE RATES
Figure 4 shows the number of questionnaires mailed out
and the number of responses received (both completed questionnaires
for eligible respondents, and statements that no hazardous waste
was generated/managed for ineligible respondents. The ratio of
column 2 to column 1 in Figure 4 is the response rate shown in
Figure 4a. The sample sizes shown in Figure 4a are the number
of completed questionnaires received. These sample sizes along
with the estimated universe size can be used to determine the
statistical reliability or confidence intervals on estimates
from the sample (see Section 5).
The response rates achieved in the hazardous waste
surveys were high, falling in the vicinity of 90 percent for
most process types as shown in Figure 4a. With response rates
this high there is little concern about non-response bias.
Response rate is defined as. percentage of completed question-
naires among eligible facilities that were contacted. Since
eligibility of non-respondents was unknown, it was estimated
using the same eligibility rate as found among respondents.
This is generally a conservative_approach since the nonrespond-
ents are often ineligible. Thus the true response rates are
likely to be somewhat higher than shown in Figure 4a. This is
especially true of generators, which had particularly high ineli-
gibility rates (i.e., a high percent of installations that pre-
viously notified EPA of their intent to generate hazardous waste
were not actually generating in 1981).
-------
Figure 4 (Preliminary Data)
* RESPONSE TO RCRA SURVEY
f
Generators
TSD's (General
questionnaire)
TSD Components
Injection Well
Landfill
Land Treatment
Surface Impoundment
Waste Pile
*
Incinerator
Storage Container
Storage Tank
Treatment Tank
Questionnaires)
^ mailed out^/
11220
2599
115
20_2_
\ 122
\ 327^
243
265
423
847
Responses
.Total receive^
9877
2348
103
172 -
99
298
215
239
389
772
* Eligible
2O16
1462
73
79
37
146
73
125
191
290
* Ineligible
7864
884
30
93
62
152
142
114
198
482
-------
-------
5. STATISTICAL RELIABILITY
- The statistical accuracy achieved in the survey of
hazardous waste facilities varies among the process types which
were sampled separately. For a number of facilities or, proportion
or percentage type measure (e.g. the proportion of landfills
that are lined), the accuracy at the, 95 percent confidence level
was good for all TSD facilities combined (is%). For generators
it was even better (±2%).
/
Although highly accurate estimates of numbers of faci-
lities and proportions of facilities with various character-
istics was achieved, it was not possible to achieve similar
accuracy for quantities of waste. The reason quantity estimates
are subject to so much more estimation error is that: (a) quan-
tities of hazardous waste vary dramatically among facilities and
(b) the sample design was optimized for facility characteristics
rather than quantities. The reason for (b) v/as partly due to
importance of determining facility characteristics (e.g. what
proportion of landfills are lined) and partly due to the absence
of reliable facility size infonnation-'-at the time that the sample
was drawn, which prevented effective probability proportional to
size sampling.
An extremely large variation was found in facility
size (where "size" is measured by the. quantity of hazardous
waste generated in 1981.) Figure 5 shows that the size distribu-
tion is very skewed due to some very large generators. This
also results in a mean value which is 350 times the median.
Figure 5a is an alternative way of looking at the same data. If
all hazardous waste generators were ranked from the largest to
the smallest, then only 1% of the generators would account for
nearly 90% of the hazardous waste, as shown in Figure 5a (i.e.,
-------
-------
Figure 5a (Preliminary Data)
CUMULATIVE HAZARDOUS WASTE GENERATED In 1981
100%
TJ 80%
0)
0) *-
0) J>
2 £
O 0)
k. -j-*
60%-
£ w 40%-
^ o
3 -O
i «
08
20%-
0
\.
1%
40%
60%
80%
100%
Cumulative Percentage of Generators in U.S
(from largest to smallest generator)
-------
the largest 140 generators produced about 35 billion gallons,
while all 14,100 generators produced only 40 billion gallons).
Similar skewed distributions were found for TSD faci-
lities. For TSD facilities the mean quantity of hazardous waste
managed per year was about 200 times the median.
To confirm the quantities reported by the largest
facilities, EPA made independent verification phone calls to the
largest cases and also determined through in-depth discussions
that the wastes and processes reported were indeed RCRA regulated
hazardous wastes.
/
Because the statistical reliability of quantities of
waste is much lower than the accuracy for proportion or number
of facilities, we have chosen to use the 67% confidence interval
(i.e., +1 standard deviation) when discussing quantities, which
is only half as wide as the 95% confidence interval would be.
Figure 5b shows the large amount of variability found
in two measures: the total quantity of hazardous waste generated
'in the U.S. during 1981 and the total quantity of waste that was
managed as hazardous waste in the' U.S;. during 1981. To inter-
pret the plus or minus factor indicated in the table for gener-
ators, the true value of the total quantity of hazardous waste
generated is covered with 67 percent confidence by an interval
centered at the estimated value and extending on either side of '
the estimated value by as much 'as 40 percent of the estimated
value. For example, if the estimated value is 100 million gal-
lons, the corresponding 67 percent confidence interval would be
from 60 million gallons to 140 million gallons. Clearly, the
confidence interval is quite wide and the estimated quantity of
total hazardous waste generted is not very precise. A similar
interpretation holds for the t25% factor for TSD facilities.
-------
Figure 5b (Preliminary Data)
Statistical Reliability of Estimates
from the Generator and
TSD General Questionnaire
Statistical Reliability of Estimates
95% Confidence 67% Confidence
Interval on a Interval on
Proportion or Quantity of
nn«tinnn*ii.* Number of Hazardous Waste
Questionnaire Faci|gties Handled
Generator ±2% ± 40 % of the total
fSD General ± 5%. _ ± 25 % of the total
-------
Various other quantity estimates from the generator
sample and TSD sample show similar sampling error, but usually
somewhat better than the ±40 percent confidence interval for
hazardous waste generated. For example, the estimate for
hazardous waste generated and shipped off site has a ±20 percent
confidence interval at the 67 percent level.
For individual components or process types the accuracy
on numbers of facilities or proportions of facilities with same
characteristics ranged from 6 to 11 percent as shown in Figure
5c, but was better than 10 percent for all components except
Land Treatment. .
/
The 95 percent confidence intervals presented in
Figure 5b and 5c are based on all -respondents to the question-
naire. If it is desired to obtain 95 percent confidence inter-
vals for subsets of the respondents, the "plus or minus" term
shown for the 95 percent confidence interval should be
multiplied by the factor, K, shown below:
Subpopulation as a Factor, K, to multiply
fraction of the population ± Term in Figure 4
.80 1.1
.60 ~- - ~~ 1.3
.40 1.6
.20 2.2
.10 3.2
This is a reasonable approximation as long as the subsample size
is fairly large, say 30 or more.
-------
Statistical Reliability of Estimated
Proportions by Process Type
Process Type
Injection Wells
Landfills
Land Treatment
Surface
Impoundment
Waste Piles
Incinerators
Storage
Containers
Storage Tanks
Number of
Respondents
Statistical
Reliability of
Proportions
at the 95%
Confidence
Level
73
79
34
146
73
125
- 191
235
+
±
±
±
±
±
±
±
5%
9%
11%
7%
9%
6%
7%
7%
Treatment
Tanks
123
± 9%
-------
PRELIMINARY
HIGHLIGHTS
OF FINDINGS
-------
Figure 6 (Preliminary Data)
REGIONAL DISTRIBUTION OF HAZARDOUS WASTE
GENERATORS IN 1981
JTIMATED TOTAL NUMBER OF
rNERATORS ACTIVE IN 1981: , 14,100
-------
Figure 7 (Preliminary Data)
QUANTITY OF HAZARDOUS WASTE GENERATED
IN 1981
80
60
h 40
67%
Confidence
Intervals
95%
L- 20
40 Bilfion gallons = 150 Million metric tonnes
Known lower limit on quantity generated
-------
Figure 7a (Preliminary Data)
QUANTITIES OF HAZARDOUS WASTE
GENERATED BY TYPE OF WASTE
BILLION % OF ALL
, TYPE OF WASTE GALLONS GENERATION
ent Solvents & Processed
Sludges, Listed Industry
J/Vaste ("F" or "K" prefix)
Reactive (D003)
Corrosive (D002)
EP Toxic (D004-D017)
Qisposed Toxic Products,
Jff-spec. Products, Chem.
itermediates ("IT prefix)
Listed by State-.,
Acute Hazardous Wastes
("P" prefix )
Ignitable (DO0 1 )
Jelf-defined as Hazardous
25.9
15.6
13.5
3.7
2.3
^ 1.9
0.3
Q.3
0.1
/^5$
39%
34%
9%
6%
5%
1%
1%
less than
.5%
TOTAL QUANTITY GENERATED = 40 BILLION GALLONS
-------
%igure 7b preliminary Data)
NUMBERS OF GENERATORS AND
QUANTITIES GENERATED BY INDUSTRY
Quantity Generated
Industry (SIC code)
Chemicals (28)
Machinery (35)
Transportation Equipment (37)
Motor Freight Transport (42)
Petroleum Refining (29)
Primary Metals (33)
Fabricated Metal (34)
Electrical Machinery (36)
Electric, Gas, and
Sanitary Services (49)
Other
All Industries
Number of Generators
2,440
700
680
80
370
1 850
s
2,640
1,510
250
4,580
14,100
Bill. Gal.
28.4
2.3
2.3
1.7
1.3
1.O
.8
.7
.5
1.0
40.0
Pet.
71%
6%
6%
4%
3%
3%
2%
1%
1%
3%
100%
-------
Figure 8 (Preliminary Data)
ON SITE VS. OFF SITE
MANAGEMENT OF GENERATED HAZARDOUS WASTES
IN 1981
NUMBER OF GENERATORS
ALL ON SITE
2,300
16%
3,100
8,700
SOME ON SITE
SOME OFF SITE
ALL OFF SITE
o o o
o
22%
62%
TOTAL = 14,100 Generators
-------
Figure 9 (Preliminary Data)
NUMBER OF GENERATORS RECYCLING
HAZARDOUS WASTES
Percent of All
Generators
lOO-i
90-
80-
70-
60-
50
40-
30-
20-
10-
0
7,800
5,700
6,100
PRIOR TO 1981
DURING 1981
AFTER 1981
Total Generators: 14,100
-------
Figure 10 (Preliminary Data)
HAZARDOUS WASTES RECYCLED
IN 1981
(% on site vs. off site)
RECYCLED
1.6 Bill. Gal
(4%)
-------
Figure 11 (Preliminary Data)
REGIONAL DISTRIBUTION OF TREATMENT, STORAGE,
DISPOSAL FACILITIES IN 1981
IX
HI -A
*"=> $
^TIMATED TOTAL NUMBER
)F ACTIVE TSD'S IN 1981: 4820
-------
Figure 12 (Preliminary Data)
NUMBER OF FACILITIES WITH TREATMENT, STORAGE,
AND/ OR DISPOSAL IN 1981
4,500
4,000
3,500
3,000
2,500
1,-
1,000
500
0
Total TSD Facilities = 4820
4,300
1,500
430
TREATMENT
STORAGE
DISPOSAL
* T+S+0 exceeds 4820 due to multiple processing at facilities
-------
Figure 13 (Preliminary Data)
NUMBER OF FACILITIES USINO
SELECTED TREATMENT , STORAGE,
AND DISPOSAL PROCESSES
NUMBER OF FACILITIES
PROCESS TYPE IN 1981*
Storage Container 3,580
Storage Tank 1,430
Surface Impoundment 770
Treatment Tank 610
Incinerator 240
Landfill 200
Waste Pile 170
Injection Wells SO
Land Treatment 70
Other Processes 320
Total TSD's: 4820*
*Sum of process types exceeds 4820 due to multiple
processing at facilities
-------
Figure 14. (Preliminary Data)
NUMBER OF COMMERCIAL VERSUS
OTHER TSD FACILITIES
Total TSD's - 4820
COMMERCIAL
FACILITIES
320 (7%)
MORE THAN 50% of waste
from other firms and publicly
owned Q£ operated
MORE THAN 50% of waste
from other firms and privately
owned and operated
50% OR LESS waste from
other firms
-------
/' Figure 15 (Preliminary Data)
QUANTITY OF WASTE MANAGED AS
HAZARDOUS WASTES BY TSD FACILITIES IN 1981
I- 120
- 100 v
SO
>- 60 *
67%
I- 40
Confidence
Intervals
' 95%
- 20
80 Billion gallons = 300 Million metric tonnes
-------
Figure 16 (Preliminary Data)
QUANTITY GENERATED
VERSUS MANAGED
MANAGED
GENERATED
BBL&aL
QUANTITY MANAGED EXCEEDS GENERATION
BECAUSE OF :
Out-of-system wastes
Inter facility shipments
Multiple stage processing
Existing stocks carried over
Non-hazardous wastes
"managed as hazardous wastes"
Respondent error/imprecise questions
Sampling variability
-------
*J -*-
Rgure 17 (Preliminary Data)
QUANTITIES OF HAZARDOUS WASTE
DISPOSED IN 1981
Underground Injection
Impoundment
Landfill 3%
Land Treatment
1.4%
TOTAL QUANTITY DISPOSED:
15.6 Billion gallons (58 Million metric tonnes)
-------
Figure 18
DDlTlONAL STUDIES UNDERWAY
Mail Survey Report on Generators
& TSD's
( Nov 83)
^Mail Survey Follow-Up
A. Design - Identify gaps, uncertainties
- Develop methodology & design
data gathering plan
B. Implement design & r'eport on
^- revised estimates <
1* Special Analyses by Process Type
V. On-going Analysis File Update
-based on EPA office feed back &
further facility confirmations
(Fall 83)
(Spring 84)
(Fall 83-
Winter 84)
(83-84)
Small Quantity Generators
- National Survey
(Fall 83)
f. Waste as a Fuel
-Track 1 Rept. (convenience sample)
-Track 2 Rept. (National survey)
-(Fall 83)
(Winter 84)
7.
Analyses of RCRA Annual Reports
Reports submitted from handlers
State summaries to EPA
EPA summary report
(Winter 84)
(Fall 84)
(Winter 85)
------- |