Transcript Advanced Hazardous Waste Inspector - Calcupa

11th Annual
California Unified Program
Training Conference
Advanced
Hazardous Waste
Inspector
Training
01/29/2009
11th Annual California Unified Program Conference
1
What is a Valid
Waste Determination?
Part II
Analysis or
Knowledge of Process?
01/29/2009
11th Annual California Unified Program Conference
2
Most of the Time…
It’s simple
01/29/2009
11th Annual California Unified Program Conference
3
But when it isn't simple,
who makes the
Waste Determination?
The Generator


“The person whose act or process produces
hazardous waste or whose act first causes a
hazardous waste to become subject to
regulation.”
A hazardous waste Generator must comply
with the requirements of Title 22 CCR,
Division 4.5, Chapter 12.
01/29/2009
11th Annual California Unified Program Conference
4
An inspector
is not a generator
Enforcement Sampling



01/29/2009
A regulator does not necessarily
need a representative sample to
support an enforcement action.
The primary reason is that the data
quality objectives (DQOs) of the
enforcement agency often may be
legitimately different from those of a
waste handler.
A sample taken for enforcement is
used to demonstrate that the waste
exceeds a standard (e.g. STLC).
11th Annual California Unified Program Conference
5
Enforcement Sampling
An enforcement official, when conducting a
compliance sampling inspection to evaluate a
waste handler’s compliance with a “do not
exceed” standard, take only one sample. Such a
sample may be purposely selected based on
professional judgment. This is because all the
enforcement official needs to observe – for
example to determine that a waste is hazardous –
is a single exceedance of the standard.
EPA530-D-02-002 (draft), August 2002,
11 {RCRA Online # 50940}
01/29/2009
11th Annual California Unified Program Conference
Page
6
Hazardous Waste
Determination
Is the generator’s responsibility.
It is not the inspector’s
responsibility!
01/29/2009
11th Annual California Unified Program Conference
7
Hazardous Waste Determination
§66262.11


First, the generator must determine if it is a waste.

§66261.2 Definition of a waste

§66261.3 Definition of a hazardous waste

§66261.4 Materials which are not waste (Is it exempted?)

§25143.2 Excluded recyclable materials
Next, the generator must determine if it is a hazardous
waste.

Is it listed in article 4 or in Appendix X of Chapter 11?

Or does it exhibit any of the characteristics set forth in article 3
of Chapter 11?
01/29/2009
11th Annual California Unified Program Conference
8
Hazardous Waste Determination
§66262.11


The Generator can make a hazardous waste
determination by:

(1) Testing; or

(2) Applying knowledge of the hazard
characteristic of the waste in light of the
materials or the processes used.
This is also called Waste Analysis.
01/29/2009
11th Annual California Unified Program Conference
9
Waste Analysis
The cornerstone of a hazardous waste program is the
ability of facility personnel to identify properly, through
waste analysis, all the wastes they generate, treat, store,
or dispose of. Waste analysis involves identifying or
verifying the chemical and physical characteristics of a
waste by performing a detailed chemical and physical
analysis of a representative sample of the waste, or
in certain cases, by applying acceptable knowledge of
the waste. (OSWER WAP Guidance Manual)
01/29/2009
11th Annual California Unified Program Conference
10
Testing

Accurate analytical data is required to comply
with Chapter 18, LDR requirements.

A written Waste Analysis Plan (WAP) is
required for:



01/29/2009
TSDFs,
PBR Treatment and,
Generators treating hazardous waste to meet LDR
standards.
11th Annual California Unified Program Conference
11
A Waste Analysis Plan




Establishes consistent internal management
mechanism(s) for properly identifying wastes on site.
Ensures that waste analysis participants have
identical information (e.g., a hands-on operating
manual), promoting consistency and decreasing
errors.
Ensures that facility personnel changes or absences
do not lead to lost information.
Reduces your liabilities by decreasing the instances
of improper handling or management of wastes.
01/29/2009
11th Annual California Unified Program Conference
12
Waste Analysis
Plan?
http://www.epa.gov/epaoswer/hazwaste/ldr/wap330.pdf
01/29/2009
11th Annual California Unified Program Conference
13
Data Quality Objectives

A generator should sample and analyze with
the data quality objective of determining with
a confidence interval of 80% if a waste is
hazardous or not.

A waste exhibits the characteristic of
ignitibility, corrosivity, reactivity or toxicity
if representative samples of the waste have
any of the following properties…
01/29/2009
11th Annual California Unified Program Conference
14
Article 3. Characteristics
of a Hazardous Waste
§66261.20 General
(a) A waste, as defined in section 66261.2, which is not
excluded from regulation as a hazardous waste pursuant
to section 66261.4(b), is a hazardous waste if it exhibits
any of the characteristics identified in this article.
(c) Sampling and testing pursuant to this article shall be
in accord with the chapter nine of SW-846, the
Department will consider samples obtained using any of
the other applicable sampling methods specified in
Appendix I of this chapter to be representative samples.
01/29/2009
11th Annual California Unified Program Conference
15
Characteristic Wastes




§66261.21(a) A waste exhibits the characteristic of ignitability if
representative samples of the waste have any of the following
properties:
§66261.22(a) A waste exhibits the characteristic of corrosivity if
representative samples of the waste have any of the following
properties:
§66261.23(a) A waste exhibits the characteristic of reactivity if
representative samples of the waste have any of the following
properties:
§66261.24(a) A waste exhibits the characteristic of toxicity if
representative samples of the waste have any of the following
properties:
01/29/2009
11th Annual California Unified Program Conference
16
Representative Sample
§66260.10 Definitions.
"Representative sample" means a sample
of a universe or whole (e.g., waste pile,
lagoon, ground water) which can be
expected to exhibit the average properties
of the universe or whole.
01/29/2009
11th Annual California Unified Program Conference
17
EPA publication SW-846
Test Methods for Evaluating Solid
Waste, Physical/Chemical Methods
 OSW's official compendium of approved
analytical and sampling methods for use
in complying with RCRA regulations.
 SW-846 primarily is a guidance document
that sets forth acceptable, although not
required, methods for the regulated and
regulatory communities to use for RCRArelated
sampling
and
analysis
requirements.
01/29/2009
11th Annual California Unified Program Conference
18
SW 846
http://www.epa.gov/sw-846/sw846.htm
01/29/2009
11th Annual California Unified Program Conference
19
01/29/2009
11th Annual California Unified Program Conference
20
SW 846, Chapter 9
Sampling Plan

SW 846 assumes that:

The concentration of a contaminant in individual
samples will exhibit a normal distribution.

Simple random sampling is the most appropriate
sampling strategy.

As more information is accumulated, greater
consideration can be given to different sampling
strategies.

Start with simple random sampling and assume a
normal distribution.
01/29/2009
11th Annual California Unified Program Conference
21
A short and sweet
Statistical Analysis
 Known
01/29/2009
population

Random samples

Plot in a Histograph

Compare actual candy
population with
sample population
11th Annual California Unified Program Conference
22
Candy Population
90 bags of candy

10 bags contain 0 pieces of
(0 pieces)

20 bags contain 1 piece of
(20 pieces)

30 bags contain 2 pieces of
(60 pieces)

20 bags contain 3 pieces of
(60 pieces)

10 bags contain 4 pieces of
(40 pieces)

Population mean is 180/90 = 2
01/29/2009
11th Annual California Unified Program Conference
23
Candy Population Histograph
(normal distribution)
Population mean
= 2
30
20
10
0 1
01/29/2009
2
3
4
11th Annual California Unified Program Conference
24
Random sample

Four samples from web 14, 37, 40, 81 (90 bags)

Four samples from web 7, 19, 35, 41 (50 bags)

Six samples from web 3, 24, 64, 71, 76 , 90
01/29/2009
11th Annual California Unified Program Conference
25
Histograph of Samples
3
2
1
0 1
01/29/2009
2
3
4
11th Annual California Unified Program Conference
26
SW 846
Normal distribution
 Terminology
 Concepts
 An exercise
 Some more terminology

01/29/2009
11th Annual California Unified Program Conference
27
Normal Distribution
In a normal distribution a
Samples
bell shaped curve is used to
represent the boundaries
of the population. The
“true” population (under the
blue curve) is never known,
but precise and unbiased
samples will provide an
accurate estimate of the
The sample population under the
true population.
magenta curve is an estimate of
the true population.
01/29/2009
11th Annual California Unified Program Conference
28
A Bell Curve has Tails!
I left the tails off most
of the diagrams because
I couldn’t figure out
how to draw them!
 The X axis is the concentration.
 The Y axis is the number of samples.
 The tails are where the people who
got 100% or 0% on an exam are found.
01/29/2009
11th Annual California Unified Program Conference
29
Reliable Waste Analysis

Reliable information concerning the chemical
properties of a solid waste is needed for
comparison with applicable regulatory thresholds.

If chemical information is to be considered reliable,
it must be accurate and sufficiently precise.

Accuracy (no bias) is usually achieved by
incorporating randomness into the sample
selection process.

Sufficient precision is most often obtained by
selecting an appropriate number of samples.
01/29/2009
11th Annual California Unified Program Conference
30
Sample size


01/29/2009
Small samples (A) cause the
constituent of interest to be
under-represented in most
samples and overrepresented in a small
proportion of samples.
Larger samples (B) more
closely reflect the parent
population.
Sometimes you sample a
large portion or even the
entire population, so you
don’t need statistics to
determine a confidence
interval.
11th Annual California Unified Program Conference
31
Terminology
Precise, Accurate & Biased



Precise means all of the samples are similar; they form
a “tight group” on the graph. Taking more samples or
taking larger samples will increase sample precision.
Accurate or unbiased means that you’re taking truly
random samples. Properly planned random samples are
accurate and unbiased samples.
Inaccurate samples are synonymous with biased
samples. They are not representative samples. Poor
tool selection or calibration can cause sample bias.
01/29/2009
11th Annual California Unified Program Conference
32
Biased & Imprecise Samples
Biased samples do not
represent the true
population. The biases
could result from poor
tool selection or
contamination.
Imprecise samples
have a lot of
variation. More
samples should
decrease variation.
01/29/2009
0
Mean
1012.5
11th Annual California Unified Program Conference
2000
33
Biased & Precise Samples
A poor sampling plan
could lead to biased or
inaccurate samples.
Poor tool selection,
poor sampling design
or contamination are
some causes.
Biased sampling shifts
the population curve.
0
Sample True
Mean Mean
2000
Who can think of another cause for biased samples?
01/29/2009
11th Annual California Unified Program Conference
34
Unbiased & Imprecise Samples
Unbiased samples are
Random samples. Random
samples fall inside the bell
curve that represents the
true population.
Take more
samples to
increase the
precision.
01/29/2009
0
Mean
1012.5
11th Annual California Unified Program Conference
2000
35
Unbiased & Precise Samples
Unbiased samples
are a function of
randomness. Random
sampling requires proper
plan design and tool
selection.
Precise samples
are a function
of the number
of samples.
01/29/2009
0
Mean
1012.5
11th Annual California Unified Program Conference
2000
36
Waste Analysis (Testing)
To evaluate the physical and chemical
properties of a solid waste



The initial -- and perhaps most critical -element is the sampling plan.
Analytical studies, with their sophisticated
instrumentation and high cost, are often
perceived as the dominant element.
But analytical data generated by a
scientifically defective sampling plan have
limited utility.
01/29/2009
11th Annual California Unified Program Conference
37
SW 846




Waste characterization requires a
representative sample.
At least two samples of a material are required
for any estimate of precision.
SW 846 uses an 80% confidence interval as an
acceptable degree of sampling accuracy and
precision.
Normally data from four representative
samples is the minimum required to achieve
an 80% confidence interval.
01/29/2009
11th Annual California Unified Program Conference
38
How many samples are enough?
An example
A business wants to dispose of a pile of used blast
medium. It has been reused and it is well mixed.
It might have been used to remove paint with lead
pigment.

Is it hazardous?

Testing or knowledge of process?


It might contain lead. – Knowledge??
How many samples do need for testing?

01/29/2009
Four?
11th Annual California Unified Program Conference
39
Sampling Plan

Make a 3-D grid of the pile.

Number each area of the grid.

Select four numbers randomly. Random number
generators are on the web, tables or in textbooks.

Sample from the four areas represented by the
number.

Analyze the samples using TTLC.
01/29/2009
11th Annual California Unified Program Conference
40
Sample Results

The TTLC for lead is 1000 mg/kg.

Sample A contains 1000 mg/kg. Is sample A
hazardous waste?

Is the waste pile hazardous?

Sample B contains 1050 mg/kg, sample C
contains 980 mg/kg and sample D contains
1020 mg/kg.
Is the waste pile hazardous?
01/29/2009
11th Annual California Unified Program Conference
41
Is it Hazardous?
A. Yes, 3 of 4 is good enough.
B. No, it’s 100% or nothing.
C. More analysis and maybe more samples are
required.
The answer is C!
01/29/2009
11th Annual California Unified Program Conference
42
More Analysis?



Yes, more analysis.
The samples were pretty close:

A contains 1000 mg/kg,

B contains 1050 mg/kg,

C: 980 mg/kg, and

D: 1020 mg/kg
A range of only 70 mg/kg.
Do we need more samples?

Yes, well…
01/29/2009
11th Annual California Unified Program Conference
43
Guess how many samples
4
5
15
20?
The answer is
15.31
Where did that number come from?
01/29/2009
11th Annual California Unified Program Conference
44
Seven Step Statistical Process
Used to determine number of samples
(SW 846 Table 9-1)
1.
2.
3.
4.
5.
6.
7.
01/29/2009
Determine the mean
Determine the variance
Determine the standard deviation
Determine the standard error
Determine the confidence interval
Determine if the variance is > the mean
Determine the appropriate number of samples
11th Annual California Unified Program Conference
45
Statistics
The last time…
I would have gotten a PHD
if I liked math…
Give it a chance!


01/29/2009
It’s just addition,
multiplication and
division.
Oh, and square roots, but
you can use a calculator.
11th Annual California Unified Program Conference
46
If you really hate numbers
Pretend to listen, it’s the polite
thing to do, and remember:

You need at least four (4)
samples.

More samples may be
required if the waste is:


01/29/2009
Heterogeneous, or
Close to the regulatory
threshold
11th Annual California Unified Program Conference
47
Step 1: The Mean
Samples
A: 1000ppm B: 1050ppm
C: 980ppm D: 1020ppm
The sample mean
is the average
value of the
samples. It’s an
estimate. The
true mean is
never known.
01/29/2009
0
Sample
Mean
1012.5
11th Annual California Unified Program Conference
2000
48
Normal Distribution
Variance
The variance is
the sum of the
differences
between the
sample values and
the mean, squared.
The variance sets
the boundaries of
the distribution.
01/29/2009
variance
0
Mean
11th Annual California Unified Program Conference
2000
49
Normal Distribution
Standard Deviation
Standard
Deviation
The standard deviation
is the square root
of the variance.
variance
0
Mean
1012.5
01/29/2009
11th Annual California Unified Program Conference
2000
50
Normal Distribution 80%
Confidence
CI
Interval (CI)
If you take 100
samples, 80 should
fall inside the
boundaries of the
80% CI.
01/29/2009
variance
0
Mean
1012.5
11th Annual California Unified Program Conference
2000
51
Normally you would
evaluate all four samples


All four randomly selected samples must be
considered in a valid statistical analysis.
In the following example, four sets of two will
also be analyzed to illustrate the effects of:



Decreasing the variance in concentration in the
samples.
Increasing number of samples.
The relationship of the mean to the Regulatory
Threshold (RT).
01/29/2009
11th Annual California Unified Program Conference
52
Step 1. The Mean

Add the results of all samples and divide by
the number of samples
Sample A=1000ppm
Sample C=980ppm
Sample B=1050ppm
Sample D=1020ppm
MEAN

A+B+C+D =(1000+1050+980+1020)/4 = 4050/4 =
1012.5 ppm




01/29/2009
A+B
C+D
B+D
A+C
= 2050/2 = 1025 ppm
= 2000/2 = 1000 ppm
= 2070/2 = 1035 ppm
= 1980/2 = 990 ppm
11th Annual California Unified Program Conference
53
Step 2. The Variance
Variance = (sample A - mean)2 + (sample B - mean)2 +(..)
Number of samples – 1
= (1000-1012.5)2+(1050-1012.5) 2 +(980-1012.5) 2 +(1020-1012.5)2
3
= (12.5)2 + (37.5)2 + (32.5)2 + (7.5)2 = 2675/3 = 891.67
3
A+B: (1000-1025)2 + (1050-1025) 2 =1250
1
C+D: ( 980 - 1000) 2 + (1020 - 1000) 2 = 800
1
B+D: (1050 - 1035) 2 + (1020 - 1035) 2 = 450
1
A+C: (1000 - 990) 2 + ( 980 - 990) 2 = 200
1
01/29/2009
11th Annual California Unified Program Conference
54
Step 3. The Standard Deviation
A=1000 ppm, B=1050 ppm, C= 980 ppm, D=1020 ppm
Standard Deviation = Variance ½


The variance of A+B+C+D is 891.67
The square root of 891.67 (standard deviation) = 29.86




A+B: Variance = 1250; standard deviation = 35.35
C+D: Variance = 800; standard deviation = 28.28
B+D: Variance = 450; standard deviation = 21.21
A+C: Variance = 200; standard deviation = 14.14
01/29/2009
11th Annual California Unified Program Conference
55
Step 4. The Standard Error
A=1000ppm, B=1050ppm, C= 980ppm, D=1020ppm
Standard Error = Standard Deviation
(Number of samples) ½

Standard error ABCD = 29.86/(4)1/2 = 14.93




Standard error A + B = 35.35/1.41 = 25.07
Standard error C + D = 28.28/1.41 = 20.06
Standard error B + D = 21.21/1.41 = 15.04
Standard error A + C = 14.14/1.41 = 10.03
01/29/2009
11th Annual California Unified Program Conference
56
Step 5. Confidence Interval
A=1000ppm, B=1050ppm, C= 980ppm, D=1020 ppm
Confidence Interval = Mean ±(student “t”)(standard error)

A+B+C+D: 1012.5 ± (1.638)(14.93) = 1012.5 ± 25.46
(988 to 1038). 80 of 100 samples should have
concentrations between 988 and 1038 ppm.




A+B: 1025 ± (3.078)(25.07) = 1025 ± 77 (948 to 1102)
C+D: 1000 ± (3.078)(20.06) = 1000 ± 62 (938 to 1062)
B+D: 1035 ± (3.078)(15.04) = 1035 ± 46 (989 to 1081)
A+C: 1010 ± (3.078)(10.03) = 1010 ± 31 (979 to 1041)
01/29/2009
11th Annual California Unified Program Conference
57
Step 6
Is the Variance > the Mean?
If the variance is not greater than the mean, go to
step 7.
 A + B + C + D: 891.67 is not > 1012.5
 If the variance is greater than the mean , you have
to transform the data. An example follows for
samples A & B.




01/29/2009
A+B:
C+D:
B+D:
A+C:
1250 is > 1025
800 is not > 1000
450 is not > 1035
200 is not > 990
11th Annual California Unified Program Conference
58
Step 6
Is the Variance > the Mean?
Mean
0
Variance
0
If variance is > mean then part of the population is less than zero, i.e.
with samples A & B the population is between -225 and 2275. You can’t
have a concentration of less than zero so you have to transform the data.
01/29/2009
11th Annual California Unified Program Conference
59
Transform the data if the
variance is > the mean
Usually data is transformed into a smaller number by
taking either the log or the square root of the value.
Step 1a. Transforming the mean

10001/2 = 31.62

10501/2 = 32.40
Total
64.02 = 32.01
2
01/29/2009
11th Annual California Unified Program Conference
60
Transforming the Variance
& Standard Deviation

Step 2a. Transforming the Variance

Variance = (sample A - mean)2 + (sample B - mean)2
Number of samples - 1




A+B: (31.62 – 32.01)2 + (32.4 - 32.01) 2 = 0.304
1
Step 3a. Transforming the Standard Deviation
Standard Deviation = Variance 1/2
A+B = (0.304)1/2 = 0.5515
01/29/2009
11th Annual California Unified Program Conference
61
Transforming the Standard
Error and Confidence Interval


Step 4a. Transforming the Standard Error
Standard Error = Standard Deviation
(Number of samples) 1/2
A+B:
0.5515/1.41 = 0.39
Step 5a. Transforming the Confidence Interval
 Mean ±(student “t”)(standard error)
 A+B: 32.01 ± (3.078)(.39) = 32.01 ± 1.20

01/29/2009
11th Annual California Unified Program Conference
62
Step 6a. Variance > Mean

A+B: The transformed variance (0.304) is not
greater than the transformed mean (32.01).

Now go to the last step #7.
01/29/2009
11th Annual California Unified Program Conference
63
Step 7. Determine the
number of samples (n)

The Regulatory Threshold (RT) using TTLC for
lead is 1000 ppm.
n = (student t)2(variance)
(RT – mean)2d

Use the square root of the RT for lead (36.62)
for transformed data.
01/29/2009
11th Annual California Unified Program Conference
64
n =
2
(student t) (variance)
(RT – mean)2
A + B + C + D: (1.638)2 (892) = 15.31
(1000 – 1012.5)2

A+B:
(3.078)2 (0.304) = 7.73 samples


C+D:
(3.078)2 (800) =


B+D:
4.27
(1000 – 1035)2
(3.078)2 (200) = 18.94
A+C:

01/29/2009

(1000 – 1000)2
(3.078)2 (450) =


(32.62 – 32.01)2
(1000 – 990)2
11th Annual California Unified Program Conference
65
So, fewer samples are
required if,
The waste is essentially homogenous
or
Well above or below the threshold
01/29/2009
11th Annual California Unified Program Conference
66
Other Types of Sampling
01/29/2009

Stratified random sampling

Systematic random sampling

Authoritative sampling
11th Annual California Unified Program Conference
67
Stratified Random Sampling




Stratified random sampling is appropriate if a batch of
waste is known to be non-randomly heterogeneous.
An example is a pile of blast media. One layer is from
blasting lead paint, the next layer is from blasting new
aluminum parts prior to painting. Another example is a
stripping tank that is used to clean different parts and is
periodically changed. The waste could vary from batch to
batch.
Stratification may occur over space (locations or points in
a batch of waste) and/or time (each batch of waste).
The units in each stratum are numerically identified, and a
simple random sample is taken from each stratum.
01/29/2009
11th Annual California Unified Program Conference
68
Systematic Random Sampling




Systematic random sampling, in which the first unit to
be collected from a population is randomly selected but
all subsequent units are taken at fixed space or time
intervals.
An example of systematic random sampling is the
sampling along a pipeline at 20 feet intervals.
The advantages of systematic random sampling are the
ease with which samples are identified and collected
and, sometimes, an increase in precision.
The disadvantages of systematic random sampling are
the poor accuracy and precision that can occur when
unrecognized trends or cycles occur in the population.
01/29/2009
11th Annual California Unified Program Conference
69
Authoritative Sampling

Sufficient information is available to accurately assess
the chemical and physical properties of a waste,
authoritative sampling (AKA judgment sampling) can be
used to obtain valid samples.

This type of sampling involves the selection of sample
locations based on knowledge of waste distribution and
waste properties (e.g., homogeneous process streams).
The rationale for the selection of sampling locations is
critical and should be well documented.

An example is an inspector taking one sample a
discarded liquid that appears to be gasoline (color &
odor) to verify that it is gasoline and has a flash point
below 140 °F.
01/29/2009
11th Annual California Unified Program Conference
70
Enforcement Sampling
RCRA Waste Sampling Draft Technical Guidance
EPA530-D-02-002 (draft), August 2002, page 10 & 11, RCRA Online # 50940


2.2.4 Enforcement Sampling and Analysis
The sampling and analysis conducted by a waste handler during
the normal course of operating a waste management operation
might be quite different than the sampling and analysis conducted
by an enforcement agency. The primary reason is that the data
quality objectives (DQOs) of the enforcement agency often may be
legitimately different from those of a waste handler. Consider an
example to illustrate this potential difference in approach: Many of
RCRA’s standards were developed as concentrations that should
not be exceeded (or equaled) or as characteristics that should not
be exhibited for the waste or environmental media to comply with
the standard. In the case of such a standard, the waste handler and
enforcement officials might have very different objectives.
01/29/2009
11th Annual California Unified Program Conference
71
Enough on Sampling?


01/29/2009
What about
knowledge of
process?
But first, a questionIs this good enough?
11th Annual California Unified Program Conference
72
Quick Break

Take 5
01/29/2009
11th Annual California Unified Program Conference
73
Hazardous Waste
Determination
01/29/2009
11th Annual California Unified Program Conference
74
Waste Determination by:
Two options:
CCR 66262.11
 Process Knowledge
 Analysis
01/29/2009
11th Annual California Unified Program Conference
75
Hazardous Waste Determination
CCR 66262.11
(b) the generator may determine that the waste is not
a hazardous waste by either:
 (1) testing…; or
 (2) applying knowledge of the hazard characteristic
of the waste in light of the materials or the processes
used and the characteristics set forth in article 3 of
chapter 11 of this division.
(A generator may also conservatively declare their
waste hazardous)

01/29/2009
11th Annual California Unified Program Conference
76
Knowledge of Process
Why Use Knowledge
Listed Waste is a function of how the
waste is generated (knowledge)
 Know that it is Hazardous
(however, I, C, R wastes are may not be
hazardous if they no longer have that
characteristic)

01/29/2009
11th Annual California Unified Program Conference
77
Knowledge of Process
OSWER 9938.4-03
RCRA Online 50010
01/29/2009
11th Annual California Unified Program Conference
78
01/29/2009
11th Annual California Unified Program Conference
79
01/29/2009
11th Annual California Unified Program Conference
80
Knowledge of Process
OSWER 9938.4-03
 Process Knowledge (Acceptable
Knowledge)
 What goes in + contaminants introduced
= What comes out.
 Waste Analysis Data from other facilities
 Old Analytical Data
 A lot of the information that is acceptable
to demonstrate Knowledge Of Process
(KOP), looks a lot like Analytical Data.
01/29/2009
11th Annual California Unified Program Conference
81
Knowledge of Process
OSWER 9938.4-03
Process knowledge should include detailed
information on the wastes obtained from
existing documented waste analysis data or
studies conducted on hazardous wastes
generated by processes similar to that which
generated the waste.
01/29/2009
11th Annual California Unified Program Conference
82
Knowledge of Process
OSWER 9938.4-03
Situations where using KOP may be appropriate:
 Constituents are well documented such as for
F or K listed waste
 Wastes are discarded unused chemicals
(P & U listed)
 Health & safety issues in sampling
(too dangerous to sample)
 Physical nature of waste (construction debris)
makes sampling impractical
01/29/2009
11th Annual California Unified Program Conference
83
Knowledge of Process
Faxback 11918
Conservative Classification:
The regulations allow a generator to characterize
its waste based on process knowledge, and it is
understood that generators may at times
characterize their wastes as hazardous
conservatively, rather than incur the costs of
testing every batch or stream.
01/29/2009
11th Annual California Unified Program Conference
84
Knowledge of Process
Faxback 11592, 11579
Limited analytical
Labs being unable to determine
conclusively that the waste is or is
not hazardous . . .
It would probably be prudent for
the generator to manage those
wastes as hazardous waste.
01/29/2009
11th Annual California Unified Program Conference
85
Seeking Concurrence with DTSC?
CCR §66260.200
(m) A person seeking Department concurrence with a nonhazardous
determination or approval to classify and manage as nonhazardous a
waste which would otherwise be a non-RCRA hazardous waste shall
supply the following information to the Department:
(5) laboratory results including results from all tests required
by chapter 11 of this division and a listing of the waste's
constituents. Results shall include analyses from a minimum
of four representative samples as specified in chapter 9 of
"Test Methods for Evaluating Solid Waste,
Physical/Chemical Methods," SW-846, 3rd Edition, U.S.
Environmental Protection Agency, 1986 (incorporated by
reference in section 66260.11 of this chapter);
01/29/2009
11th Annual California Unified Program Conference
86
01/29/2009
11th Annual California Unified Program Conference
87
Is ONE sample good
for anything?
• Faxback 11907 - Representative sampling
(Fluorescent Tubes)
“it appears that you tested one spent fluorescent tube to
conclude that all of your spent fluorescent tubes are non
hazardous. . . . Based on one tube, we have no way to
assess the variability between fluorescent lamps. . . A
representative selection of lamps randomly chosen
should be analyzed to make this determination.”
01/29/2009
11th Annual California Unified Program Conference
88
KOP Documentation
OSWER 9938.4-03
“…EPA looks for documentation that
clearly demonstrates that the information
relied upon is is sufficient to identify the
waste accurately and completely.”
01/29/2009
11th Annual California Unified Program Conference
89
KOP Documentation
•
•
•
•
•
The generator is very familiar with the waste generation
process and the California and Federal hazardous waste laws
and regulations.
Detailed chemical information for all the chemicals and
materials utilized in the process is available.
A detailed review of the generating process has been
completed and the point of generation has been properly
identified.
All documentation utilized to make the determination is
included in the operating record associated with the waste
stream.
The generator has evaluated the information gathered and
made a written determination.
01/29/2009
11th Annual California Unified Program Conference
90
What if generator does not have
waste determination available?

Inspector has the authority to enter and inspect,
sample, photograph and copy records. (HSC
25185)

Inspector may proceed with an administrative,
civil or criminal enforcement action.

If the inspector suspects a generator to be wrong,
they may request additional information.
01/29/2009
11th Annual California Unified Program Conference
91
Knowledge of Process
OSWER 9938.4-03
Conclusion (EPA Guidance Doc):
Although EPA recognizes that sampling and
analysis are not as economical or convenient
as using acceptable knowledge, they do
usually provide advantages. Because
accurate waste identification is such a critical
factor for demonstrating compliance with
RCRA, misidentification can render your
facility liable for enforcement actions.
01/29/2009
11th Annual California Unified Program Conference
92
Real Life Application

A generator must
determine if waste rags
with used oil are a
hazardous waste.

Initial testing
determined Cadmium is
present and is the only
metal that exceeds
TCLP.
01/29/2009
11th Annual California Unified Program Conference
93
Let’s say 15 rags with used oil
are sampled from a drum of rags
Step 1:
Determine the
mean of the
sample results:
Add the results
of all samples
and divide by
the number of
samples.
01/29/2009
N
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Cd mg/L
1.20
1.40
1.60
0.68
0.52
0.93
0.50
0.40
0.40
0.20
0.30
0.30
1.20
1.30
TCLP for
Cadmium is
1.0mg/L
Total:
11.80/ 15 =
0.786667
11th Annual California Unified Program Conference
94
Step 2: Determine the variance of
the sample results
Variance = (sample A - mean)2 + (sample B - mean)2
+(..) and divide by number of samples - 1
N
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
01/29/2009
Cd mg/L
1.20
1.40
1.60
0.68
0.52
0.93
0.50
0.40
0.40
0.20
0.30
0.30
1.20
1.30
0.87
Mean difference squared
-0.787
0.41 0.170569
-0.787
0.61 0.375769
-0.787
0.81 0.660969
-0.787
-0.11 0.011449
-0.787
-0.27 0.071289
-0.787
0.14 0.020449
-0.787
-0.29 0.082369
-0.787
-0.39 0.149769
-0.787
-0.39 0.149769
-0.787
-0.59 0.344569
-0.787
-0.49 0.237169
-0.787
-0.49 0.237169
-0.787
0.41 0.170569
-0.787
0.51 0.263169
-0.787
0.08 0.006889
11th Annual California Unified Program Conference
Total:2.951935/14
=
0.210853
95
Steps 3 & 4: Determine the Standard
Deviation and Error of sample results

Step 3: Determine Standard Deviation
Standard Deviation = square root of the Variance
Square root of 0.211 = 0.459

Step 4: Determine the Standard Error of the
sample results:
Standard Error = Standard Deviation divided by
the square root of the number of samples
Standard Error = 0.459/square root 15= 0.118
01/29/2009
11th Annual California Unified Program Conference
96
Step 5: Determine Confidence Interval
Is it a Hazardous Waste?
Step 5: Determine the 80% Confidence Interval
for the sample results: 0.69 through 0.89
Confidence Interval (CI) = Mean plus and minus the
product of Student "t" and Standard Error
Student "t" for 80% CI and 15 samples = 0.866




0.866 x 0.118 = 0.102188
Upper CI = 0.787 + 0.102 = 0.889
Lower CI = 0.787 - 0.102 = 0.685
01/29/2009
NOT HW!
11th Annual California Unified Program Conference
97
BABYKING
SQUEEZE FISH
Components†
Orange fish
Purple fin of
orange fish
Yellow fish
Green fish
Pink fin on
green fish
Lead
Cadmium
Chlorine/PVC
Arsenic
Mercury
0
0
140
184
309,995
216,709
0
0
0
0
0
0
0
124
82
146
327,995
299,320
155,789
0
0
0
0
0
0
Note: numbers in this table represent parts per million of the given chemical by
XRF testing. Cadmium yellow is cadmium sulfide (CdS), cadmium red is cadmium
selenide (CdSe) and cadmium orange is an intermediate cadmium sulfoselenide.
01/29/2009
11th Annual California Unified Program Conference
98
So, what now?




XRF is not an SW-846 test for evaluating a solid
waste.
We know it has Cadmium and possibly Selenium.
RCRA or non-RCRA? TCLP vs. TTLC
A minimum of 4 representative samples
(a minimum of four representative samples) as specified in chapter 9 of
"Test Methods for Evaluating Solid Waste, Physical/Chemical Methods,
SW-846, 3rd Edition, U.S. Environmental Protection Agency, 1986
(incorporated by reference in section 66260.11 of this chapter);

"
What test method and for what do we test?
01/29/2009
11th Annual California Unified Program Conference
99
So How Do We
Test the Fish?
Based on what we just learned and with the help of
SW-846…
 Test Two or more fish for metals. TTLC metals
show: Cd, Se, ???
 Rule out ND or near zero? Based on Two tests?
 Just call if HW for those way over limit?
 Determine the 80% CI for the rest?
01/29/2009
11th Annual California Unified Program Conference
100
Waste Classification
Resources






http://ccelearn.csus.edu/wasteclass/intro/intro_01.
html
www.calcupa.net/conference/2007/terracetuesday/CUPA07wcfinal.ppt
www.epa.gov/epaoswer/hazwaste/ldr/wap330.pdf
www.dtsc.ca.gov
http://www.epa.gov/epawaste/hazard/index.htm
McCoy’s RCRA Unraveled
01/29/2009
11th Annual California Unified Program Conference
101
01/29/2009
11th Annual California Unified Program Conference
102
Ten
minute
Break
01/29/2009
11th Annual California Unified Program Conference
103