ML22031A056

H-121S Marssim Course Jan 2022

ML22031A056
Person / Time
Issue date:	01/31/2022
From:	Office of the Chief Human Capital Officer
To:
References
Download: ML22031A056 (35)

Text

Table of contents Module 6: Radiological Survey Types in Support of Decommissioning Module 7: DCGLs Pt. 1 Module 8: DCGLs Pt. 2 Module 9: DCGL Problem Set Module 10: RESRAD Module 11: RESRAD-BUILD Module 12: DandD Module 13: Visual Sample Plan (VSP)

Module 14: Classification and Survey Units Module 15: Reference Areas Module 16: Surface Activity Assessment and Detection Sensitivity of Survey Instrumentation Module 17: Statistical Design of Final Status Surveys Module 18: FSS Design Exercises - WRS Test Module 19: FSS Design Exercises - Sign Test Module 20: Integrated Final Status Survey Module 21: Performing the Statistical Tests Module 22: Data Quality Assessment, Power Curves, and the Assessment of Multiple Radionuclides Module 23: Evaluating the Cushing Data Set Module 24: Special Survey Issues Module 25: Indistinguishable from Background Pt. 1 Module 26: Indistinguishable from Background Pt. 2 Module 27: Final Status Survey Reports MARSSIM Cheat Sheet Module 1: MARSSIM Overview, Part 1 Module 2: MARSSIM Overview, Part 2 Module 3: MARSSIM Overview, Part 3 Module 4: MARSSIM Overview, Part 4 Module 5: MARSSIM Overview, Part 5

MARSSIM CHEAT SHEET

• MARSSIM (NUREG-1575) is a document that provides guidance for conducting final status surveys at radiologically contaminated facilities undergoing decommissioning.

• The final status survey is conducted by the licensee (or a subcontractor) after they have concluded that further remedial action is not necessary. In other words, the site should be relatively clean and is expected to meet the release criterion established by the regulator.

• A single final status survey is not conducted over the entire site. Instead, final status surveys are conducted in discrete areas of the site known as survey units. The planning, implementation and data assessment for the final status surveys proceed independently for each survey unit. Ultimately, every survey unit at a site must be demonstrated to meet the regulators release criterion.

• The Nuclear Regulatory Commissions criterion for unrestricted release of a property can be summarized as follows: residual contamination that is distinguishable from background should not result in more than 25 mrem in a single year to an average member of the critical population. Many individual states employ lower criteria (e.g., 10 mrem),

but we will assume for the purpose of this discussion that the criterion is 25 mrem.

• The concentrations (typically in soil or on building surfaces) that result in 25 mrem are referred to as derived concentration guideline levels (DCGLs).

MARSSIM OVERVIEW Abelquist, E. Decommissioning Health Physics. A Handbook for MARSSIM Users. Second Edition. CRC Press, Taylor & Francis. 2014.

NUREG-1505. Nuclear Regulatory Commission. A Nonparametric Statistical Methodology for the Design and Analysis of Final Status Decommissioning Surveys. Revision 1. 1998.

NUREG-1507. Nuclear Regulatory Commission. Minimum Detectable Concentrations with Typical Radiation Survey Instruments for Various Contaminants and Field Conditions (pre-published draft). June 1998.

NUREG-1575. Multi-Agency Radiation Site Survey and Investigation Manual (MARSSIM). Revision 1. August 2000.

NUREG-1576. Multi-Agency Radiological Laboratory Analytical Protocols Manual. July 2004.

NUREG-1757 (three volumes). Nuclear Regulatory Commission. Consolidated Decommissioning Guidance.

IMPORTANT GUIDANCE DOCUMENTS The seven step DQO process is an integral component of the planning phase of the data life cycle. Its primary purpose is to ensure that all the important issues are addressed. These seven steps can be boiled down to the following:

1. State the problem: Identify the planning team, decision makers, deadlines, resources and a concise description of the problem.

2. Identify the decision: for a final status survey this would be Is the level of residual contamination in a given survey unit below the release criterion. Then, the alternative actions are identified e.g., further remediation, reevaluation of the DCGLs, restrictions on release, etc.

3. Identify inputs to the decision: Identify the specific questions to be answered, e.g., What physical characteristics of the site need to be evaluated, What chemical characteristics of the contamination need to be determined, etc. The chosen means to answer these questions are identified. The information needed to establish the DGCLs is identified. What methods will be used to provide the necessary data is determined.

4. Define the study boundaries: Areas of the site to be evaluated are defined, and the time frame in which the survey will be performed is defined.

5. Develop a decision rule: the statistical method for describing the residual activity is identified e.g., the mean, median for the survey unit, etc.

The action levels are identified.

6. Specify limits on decision errors: Estimate the likely variation in the measurements for the survey unit, identify the null hypothesis and define the consequences of Type I and Type II errors in terms of health, political, and resource issues. Specify acceptable values for Type I and II error rates (alpha and beta). The formal DQO process does not reference the LBGR, but the latter should be specified when beta is specified.

7. Optimize the design of the survey for obtaining the data. Evaluate data collection design alternatives, develop the mathematical expressions that will be necessary to implement the alternatives and select the optimal options.

DATA QUALITY OBJECTIVES (DQO) PROCESS (MARSSIM Appendix D)

MARSSIM DATA LIFE CYCLE (MARSSIM 2.3)

In all four phases of the data life cycle, communication between the licensee and the regulator is essential.

1. Planning Phase (design the survey)

2. Implementation Phase (perform the survey)

3. Assessment Phase (evaluate the measurements) 4. Decision Making Phase (what to do if the survey unit fails to meet the release criterion)

1. PLANNING PHASE OF THE DATA LIFE CYCLE 1. Determine the DCGLw for Individual Nuclides (outside scope of MARSSIM)

• It might be possible to use screening levels published by the NRC (NUREG-1757 Vol 2.) with minimal justification.

• If a screening level is not available for a particular radionuclide, one might be calculated using default input parameters and the DandD code.

• A less conservative (i.e., higher) DCGL might be calculated using site-specific input parameters and the computer code RESRAD or RESRAD-Build.

2. Determine the Gross DCGLw for Multiple Nuclides when Performing Gross Alpha or Beta Measurements (MARSSIM 4.3.4)

• Use the most restrictive (lowest) of the DCGLs for individual nuclides, or

• Determine the fraction of the total activity (alpha or beta) contributed by the various nuclides and use the following equation:

DCGLGROSS is the gross alpha or beta activity DCGL for the specified mix of nuclides.

f1, f2, etc. are the fractions of the total alpha or beta activity contributed by nuclide 1, nuclide 2, etc.

DCGL1, DCGL2, etc. are the individual DCGLs for nuclide 1, nuclide 2, etc.

3. Adjust (lower) the DCGLw for Surrogate Nuclides (MARSSIM 4.3.2)

It is possible to use measurements of a nuclide that is inexpensive to analyze (e.g., Cs-137 by gamma spec) as a surrogate for the measurements of one or more expensive to analyze nuclides (e.g., Sr-90 by radiochemistry) if a ratio can be established between the nuclides. When this is done, the DCGL for the nuclide that is measured (the surrogate) must be adjusted downwards.

DCGLADJ SURR is the adjusted (lowered) DCGL for the surrogate nuclide.

R2, etc. are the expected ratios of the activities of the non-detected nuclides to the activity of the surrogate nuclide.

DCGL2, etc. are the individual DCGLs for nuclide 2, etc.

4. Determine the DCGLEMC (NUREG-1505 Chapter 8)

The DCGLEMC is the maximum permitted average concentration in a hot spot. It is the concentration of a specified nuclide in a specified area (smaller than the survey unit) that is assumed to result in 25 mrem in a year (i.e., the release criterion). It is calculated as follows:

DCGLEMC = DCGLW x AF AF is an area factor that is specific to the nuclide and area.

Although tables of example area factors have been published, there are no default area factors (or DCGLEMC for that matter). To calculate the area factor, divide the dose predicted with RESRAD (or RESRAD-Build) for the survey unit (or default area) by the dose predicted for the area of the hot spot. In the planning phase of the data life cycle (e.g., step 15), a worst case hot spot size must be assumed. This assumed area is that bounded by four measurement/sampling points.

During data assessment, actual hot spot areas are used to determine the DCGLEMC.

5. Classify the Site According to Contamination Potential (MARSSIM 4.4, NUREG-1505 2.2.3, 2.2.4)

Each area of the site is assigned one of the following classifications according to its potential for contamination

• Class 1 impacted areas have or had a potential for individual measurements above the DCGL. Remediated areas generally considered Class 1.

• Class 2 impacted areas have or had a potential for contamination at a significant fraction of the DCGL. Individual measurements should not exceed the DCGL.

• Class 3 impacted areas have little or no potential for contamination. Individual measurements should not exceed a significant fraction (e.g., 10-20%) of the DCGL.

• Non-impacted areas have no potential for contamination. If possible, reference area(s) are established in non-impacted area.

Note that the definitions of impacted areas in NUREG-1757 are more fIexible than the above which are given in MARSSIM.

6. Establish the Survey Units (MARSSIM 4.6)

The site is divided up into areas of similar contamination potential known as survey units. The survey unit is the fundamental unit of compliance. Planning, implementation and data assessment are conducted independently for each survey unit. Maximum recommended survey unit areas:

Structure (e.g., building)

Land Area

Class 1

100 m2 floor area

2000 m2

Class 2

1000 m2 floor area

10,000 m2

Class 3

no limit

no limit 7. Determine if Scenario A or B will be Used

• Scenario A is the most commonly employed approach in final status surveys, and the only approach described in MARSSIM (NUREG-1575). The object is to demonstrate that the average/median level of residual radioactivity in a survey unit is less than the DCGL.

• In Scenario B, the object is to demonstrate that measurements in the survey unit are indistinguishable from those in background. Scenario B is used when the DCGL is low, the radionuclide is in background (or the measurements are not nuclide specific) and background is variable. The primary source of guidance for Scenario B is NUREG-1505.

8. Determine if the Sign Test or Wilcoxon Rank Sum (WRS) Test will be Used to Assess the Data

• Analysis is nuclide-specific and the nuclide is not in background (e.g., Co-60 in soil) use the Sign test.

• Analysis is nuclide-specific and the nuclide is in background (e.g., Pu-239 in soil) at a small fraction of the DCGL use the Sign test.

• Analysis is nuclide-specific, and the nuclide is in background (e.g., Ra-226 in soil) at a significant fraction of the DCGL use the WRS test.

• Analysis is not nuclide-specific (e.g., gross beta on surfaces) and background is a small fraction of the DCGL use the Sign test.

• Analysis is not nuclide-specific (e.g., gross alpha on surfaces) and background is a significant fraction of the DCGL. This situation can be problematic if surface types and background vary from one survey unit to another. In this case, there are several options to account for background (Chapter 12 NUREG-1505):

use the WRS test, or

subtract average background for surface types from survey unit gross measurements and use Sign test on net values, or

subtract paired background measurements for surface types from gross measurements and use Sign test on net values.

9. Determine if the Unity Rule will be used in the Statistical Tests (NUREG-1505 Chapter 11)

The unity rule is used when two or more nuclides are analyzed in each soil sample (or alpha and beta measurements are performed at each location). The unity rule can be thought of as a sum of the fractions approach wherein the concentrations of multiple radionuclides are expressed as fractions of the DCGLs. When a DCGL for multiple nuclides is used in these tests, it is assigned a value of 1.

n n

GROSS DCGL f

DCGL f

DCGL f

DCGL 1

2 2

1 1

+

+

=

n n

SURR SURR ADJ DCGL R

DCGL R

DCGL DCGL 1

1 2

2

+

+

=

1 etc.

DCGL 2

DCGL Conc. nuclide 1 Conc. nuclide 2 Combined concentration of multiple nuclides

+

+

=

Remediated areas generally considered C

1. PLANNING PHASE OF THE DATA LIFE CYCLE continued 10. Select the Type of Detection Equipment (MARSSIM Chapter 6)

• Performing scans and static measurements on structural surfaces gas fIow proportional counters usually preferred.

• Performing scans for gamma emitters in soil sodium iodide detectors usually preferred.

• Measuring contaminants in soil collect soil samples and analyze by gamma spectroscopy and/or radiochemistry. Contaminants in soil might also be measured by in-situ gamma spectroscopy.

11. Determine the Measurement Protocols (MARSSIM 5.5.3)

• Class 1 Survey Units Scan coverage: 100%. Measurements/samples collected in systematic (e.g., triangular) pattern.

• Class 2 Survey Units Scan coverage: 10 to 100%. Measurements/samples collected in systematic (e.g., triangular) pattern.

• Class 3 Survey Units Scan coverage: judgmental. Measurements/samples distributed randomly.

• Non-impacted area No scan or measurements required.

13. Determine the Scan and Measurement Investigation Levels (MARSSIM 5.5.2.6)

The investigation level is the instrument response (e.g., cpm) that triggers an investigation when exceeded. MARSSIM suggests the following:

• Class 1 survey unit. The instrument response corresponding to the DCGLEMC for the area bounded by four measurement points.

• Class 2 survey unit. The instrument response corresponding to the DCGLw. If this is exceeded, the survey unit might have been misclassified.

• Class 3 survey unit. The instrument response corresponding to some fraction of the DCGLw. If the scan MDC exceeds the DCGLw, the instrument response at the scan MDC might be used.

14. Determine the Acceptability of Type I and Type II Errors and Set the LBGR (MARSSIM Appendix D, NUREG-1505 Chapter 2)

For the purpose of the statistical tests, the Null (working) Hypothesis is that the median level of contamination in the survey unit exceeds the DCGLw.

• Regulator establishes maximum acceptable probability (alpha) of the statistical test falsely concluding that the median level of contamination (above background) in the survey unit is below the DCGLw when it is actually at or above it. The most likely value for alpha is 0.05 (i.e., 5%).

• Licensee establishes acceptable probability (beta) of the statistical test falsely concluding that the median level of contamination (above background) exceeds the DCGLw when it is at a concentration known as the lower boundary of the gray region (LBGR).

• The licensee sets the LBGR at some concentration below the DCGLw. In general, the LBGR should be set at the expected average/median concentration in the survey unit.

15. Determine the Appropriate Number of Measurements or Samples (MARSSIM 5.5.2, NUREG-1505 Chapter 9)

Calculate the relative shift. Given the relative shift and values for alpha and beta, the required number of measurements or samples can then be found in Table 5.3 of MARSSIM if the WRS test is to be used. If the Sign test is to be used, the number of measurements is found in Table 5.5.

The relative shift is a unitless number (often between 1 and 4) related to the chance that individual measurements will exceed the DCGLw. The smaller the relative shift, the greater the likelihood some measurements exceed the DCGLw and the greater the number of measurements that should be made.

is the expected variability of the measurements. It, like the LBGR, is based on earlier characterizations.

If the unity rule is employed:

• use 1 as the value for the DCGL in the relative shift calculation, and

• use the following equations to determine the appropriate values for the LBGR and.

16. For Class 1 Survey Units, the Number of Measurements might need to be Increased (MARSSIM 5.5.2.4)

This is because the scan must be sufficiently sensitive to detect a hot spot exceeding its DCGLEMC. The largest (worst case) potential hot spot area is assumed to be that bounded by four measurement points (the survey unit area divided by the number of measurements/samples). If the scan MDC is below the DCGLw, the scan MDC is also below the DCGLEMC and there is no problem. If the scan MDC is above the DCGLw, it must be compared with the DCGLEMC for a hot spot of that area. Then, if the scan MDC is above the DCGLEMC, the number of fixed measurements/samples must be increased so that the increased DCGLEMC equals the scan MDC. To do this, we first divide the actual scan MDC by the DCGLW. This gives the area factor for the new smaller hot spot where the scan MDC equals the DCGLEMC. Then the hot spot area corresponding to this area factor is determined. Dividing this new area into the total survey unit area gives the new required number of measurements/samples.

17. Establish Reference Grid and Determine Measurement/Sample Locations (MARSSIM 5.5.2.5, NUREG-1505 Chapter 3.5)

MARSSIM does not recommend a particular type of reference grid. When measurements/samples are to be distributed in a systematic pattern (Class 1 and 2 survey units), MARSSIM recommends a triangular (equilateral) pattern. The reference grid coordinates of the starting point for a systematic pattern are determined using random numbers. If the measurements/samples are to be distributed randomly (class 3 survey units), the coordinates for all the locations are selected using random numbers.

12. Determine the Measurement and Scan MDCs (MARSSIM 6.7, NUREG-1507 3.1 and Chapter 6). One of the few absolute requirements in MARSSIM is that the measurement MDC be below the DCGLW. Nevertheless, MARSSIM recommends a MDC that is 10-50% of the DCGLW. Typical measurement MDCs for gas flow proportional counters found in NUREG-1507 Table 5.2. Example lab sensitivities found in MARSSIM Table 7.3.

CB is the background count (counts).

t is the count time (min). This equation assumes background and sample count times are identical.

Ei is the instrument (2 pi) efficiency (counts per particle leaving the surface).

Es is the surface efficiency (fraction of the decays that a detectable particle leaves the surface).

Default is 0.5 for betas with maximum energies above 400 keV and 0.25 for alphas and betas with maximum energies between 150 and 400 keV.

A is the probe area (cm2).

Typical NaI scan MDCs can be found in Table 6.7 of MARSSIM and Table 6.4 of NUREG-1507.

d is 2.32 if the acceptable probability of false positives is 0.25 and the acceptable probability of a correct detection is 0.95 (see MARSSIM Table 6.5).

CBi is the background count during the time interval i.

i is the time interval (sec) that the probe is over the hot spot (of an assumed size).

p is the surveyor efficiency (MARSSIM recommends 0.5 be used).

Scan MDC (dpm/100cm )

2 =

100 A

E E

i p

s CBi i

60 d' Measurement MDC (dpm/100cm )

2 =

100 A

E E

t s

CB i

3 4.65

+

DCGL DCGL etc.

2 nuclide 1 2

nuclide 2

+

=

+

)

(

)

(

2 1

etc DCGL nuclide 2 conc Expected DCGL nuclide 1 conc Expected LBGR

+

+

=

Relative Shift = DCGLW LBGR

2. IMPLEMENTATION PHASE OF THE DATA LIFE CYCLE

3. ASSESSMENT PHASE OF THE DATA LIFE CYCLE 1. Scan Surfaces for Contamination (MARSSIM 6.4.2)

The detector probe is slowly moved back and forth over potentially contaminated surfaces while the surveyor listens to the detectors audio output for indications that the investigation levels have been exceeded. The primary purpose of the scan is to locate small areas of elevated activity, i.e., hot spots. If located, the latter are characterized by additional measurements/samples to determine that the contamination is below the DCGLEMC.

• For alpha and beta scans, the probe is usually 0.5 to 1.0 cm above the surface. A typical scan rate might be one half to one probe width per second.

• For gamma scans, a NaI probe is swung over the ground in a 1 m arc approximately 10 to 15 cm above the surface. A typical scan rate is 0.5 m/s.

2. Perform Static Measurements on Surfaces and/or Collect Soil Samples (MARSSIM 6.4.1, 7.5)

This is done to obtain accurate determinations of the contamination levels at a number of unbiased locations. This data will be assessed statistically to determine if the contamination levels in the survey unit are below the DCGLw and used to determine that the survey unit was properly classified.

• Static measurements of alpha or beta concentrations are performed with the probe directly on, or just above, the surface (in the final status survey, the surfaces should be clean with little to no removable activity). Measurements are typically performed using scalars set to one minute count times.

C is the surface concentration (dpm/100 cm2).

RN is the net count rate (cpm).

Ei is the instrument efficiency (2 pi efficiency).

Es is the surface efficiency.

A is the probe area (cm2).

• Soil samples are generally collected to a depth of 15 cm. One kilogram samples are usually collected for gamma spec analysis, while smaller samples might be obtained if a radiochemical analysis is performed.

Default values for surface efficiency (Es):

• 0.5 for betas with maximum energies above 400 keV.

• 0.25 for alphas and betas with maximum energies

between 150 and 400 keV.

1. Data Verification (MARSSIM 9.3.1, NUREG-1576 Chapter 8)

It is determined whether or not the laboratory and field personnel did what they were supposed to do. For example:

• were the correct instruments used and were daily QC checks on the instruments performed.

• were the count times and sample masses what they were supposed to be (i.e., were the requisite MDCs obtained).

• were the requisite number of spilt, duplicate, blank samples performed.

• is there missing documentation.

2. Data Validation (MARSSIM 9.3.2, App. N, NUREG-1576 Chapter 8)

Data points are assessed and flagged as necessary. In essence, this is a reality check on individual measurements. Typical qualifiers (fIags) are:

• U measurement less than critical level.

• J measurement very uncertain or questionable.

• R data point rejected.

3. Preliminary Data Review (MARSSIM 8.2.2)

• Determine the following: number of valid measurements, lowest measurement, highest measurement, mean, median, and standard deviation.

• Based on the measurements and scan data, determine if the area classification appears correct.

• Determine that the requisite number of measurements are made.

• Determine if the mean is above or below the LBGR.

• Survey unit fails if the average measurement (Sign test) or the difference between the average survey unit and reference area measurements (WRS test) exceeds the DCGLw. Note, this does not include biased measurements, only those collected for the purpose of the statistical tests.

• A statistical test is not necessary if all the measurements (Sign test), or the difference between the highest survey unit measurement minus the lowest background measurement (WRS test), is below the DCGLw. As before, this does not include judgmental and biased measurements.

• A statistical test is necessary if the average measurement (Sign test) or the difference between the average survey unit and reference area measurements (WRS

test) is below the DCGLw but some measurements are above the DCGLw. This doesnt include biased measurements.

4. Data are Plotted/Graphed (MARSSIM 8.4, NUREG-1505 4.2.2)

• A posting plot is produced in which the measurements (e.g., pCi/g or dpm/100 cm2) are indicated on a drawing of the survey unit at the locations where the measurements were taken.

• A histogram might also be generated.

5. If necessary, the Sign Test is Performed (MARSSIM 8.3, NUREG-1505 Chapter 5)

This test only employs the unbiased randomly distributed or systematic measurements. Judgmental or otherwise biased data are not used.

• The total number of measurements being evaluated in the survey unit is N. The number below the DCGLw is the statistic S. If a measurement is the same as the DCGLw, it is not counted and the total number of measurements (N) is reduced by one.

• If S is above the appropriate critical value in Table I.3 of Appendix I in MARSSIM, the Null Hypothesis (that the survey unit exceeds the release criterion) is rejected.

• If S is tied with or below the critical value, we fail to reject the Null Hypothesis and a decision must be made as to how to proceed.

6. If necessary, the Wilcoxon Rank Sum (WRS) Test is Performed (MARSSIM 8.4, NUREG-1505 Chapter 6)

This test only employs the unbiased randomly distributed or systematic measurements. Judgmental or otherwise biased data are not used.

• Add the DCGLw to each of the reference area measurements.

• The adjusted reference area measurements are then pooled with the survey unit measurements.

• The pooled measurements are ranked (sorted) from lowest to highest (1, 2, 3, etc.). Tied values are assigned an average rank (e.g., 2.5).

• The sum of the ranks of the adjusted reference area measurements is the statistic Wr.

• If Wr is above the appropriate critical value in Table I.4 of Appendix I in MARSSIM, the Null Hypothesis (that the survey unit exceeds the release criterion) is rejected.

• If Wr is tied with or below the critical value, we fail to reject the Null Hypothesis and a decision must be made as to how to proceed.

7. Perform an Elevated Measurement Comparison (MARSSIM 8.5.1, NUREG-1505 Chapter 8)

This test involves all the survey unit measurements, i.e., biased and unbiased measurements.

• Every measurement (above background) above the Investigation Level triggers an investigation.

• The investigation involves confirming the measurement and then determining the area, average concentration, and DCGLEMC for each hot spot.

8. Determine that the Total Dose from All Sources is Below the Release Criterion (MARSSIM 8.5.2, NUREG-1505 8.1) 100 A

E E

R C

s i

N

=

is the average concentration in the survey unit determined from unbiased measurements.

DCGLW etc.

DCGLEMC for hot spot 1 DCGLEMC for hot spot 2 (ave.conc.hot spot 1 )

(ave.conc.hot spot 2 )

< 1

+

+

+

In Scenario B, two statistical tests are performed for each survey unit: the WRS test and the Quantile test. The Null Hypothesis in these tests is that the survey unit measurements are indistinguishable from background. As such, the goal is to fail to reject the Null Hypothesis in both tests.

1. Perform the Kruskal-Wallis Test The purpose of this test is to show that significant variability exists in background. This can be considered a justification for employing scenario B. The Null Hypothesis is that no significant variability exists.

• Obtain measurements/samples in at least four reference areas. NUREG-1505 recommends at least 10 measurements in each, whereas NUREG-1757 recommends at least 15.

• Rank the pooled measurements from all the reference areas. Sum the ranks in the individual reference areas.

• Calculate the Kruskal-Wallis statistic (K) using the following equation:

N is the total number of measurements in all the reference areas.

ni is the number of measurements in a given reference area.

Ri is the sum of the ranks in a given reference area.

• Compare K with critical value in Table 13.1 of NUREG-1505. In this table, k is the number of reference areas (usually 4). NUREG-1505 recommends an alpha of 0.1 whereas NUREG-1757 recommends a value of 0.2.

• If K exceeds the critical value, the null hypothesis is rejected and the conclusion is that there is significant variability in the background data.

2. Determine a Concentration that is Indistinguishable from Background (e.g., 3 )

• The concentration that is indistinguishable from background is some multiple of. NUREG-1505 and NUREG-1757 both use three as the multiple.

sB is the mean square between the reference areas.

sW is the mean square within the reference areas.

n0 is related to the number of measurements in the reference areas.

N is the total number of measurements in all the reference areas.

ni is the number of measurements in a given reference area.

k is the number of reference areas.

xi is an individual measurement in reference area i.

3. Perform Wilcoxon Rank Sum Test on the Survey Unit Data (NUREG-1505 6.3)

• Adjust survey unit data by subtracting the concentration that is indistinguishable from background (3) from each survey unit measurement.

• Rank the adjusted survey unit data and the unadjusted reference area data.

• Sum the ranks of the adjusted survey unit data. This statistic (Ws) is compared with the critical value in Table A.7 of NUREG-1505 (m and n are the numbers of survey unit and reference area measurements respectively).

• If Ws is less than or equal to the critical value, the null hypothesis is not rejected and the survey unit is assumed indistinguishable from background.

4. Perform Quantile Test on the Survey Unit Data (NUREG-1505 Chapter 7)

• Each rank is identified as being from the survey unit (S) or a reference area (R). These identifications are then sorted in order from the lowest to highest rank.

• The number of the r highest ranks that are from the survey unit is compared with the value k. Values for r and k are indicated in NUREG-1505s Table A.7.

Note that the meanings of m and n are the reverse of those in the WRS test.

• If this number is less than k, the Null Hypothesis is not rejected and the survey unit is assumed indistinguishable from background.

n i

in N

N k

k 1

2 1

1 0 =

=

2

0

(

)

B S

2 W

S n

2

=

DECISION MAKING PHASE OF THE DATA LIFE CYCLE STATISTICAL CALCULATIONS IN SCENARIO B INDISTINGUISHABLE FROM BACKGROUND (NUREG-1505 Chapter 13)

A decision must be made as to how to proceed if the survey unit was misclassified, failed the elevated measurement comparison, failed the statistical assessment, or the total dose from all radiation sources exceeded the release criterion.

• Measurements in Class 2 or 3 survey units exceeding the DCGLW indicate that the area might have been misclassified. If so, the area (including nearby survey units) may have to be recharacterized, reclassified, subdivided into smaller survey units, and resurveyed.

• If the average concentration (above background) in a hot spot is determined during the elevated measurement comparison to exceed the DCGLEMC for that hot spot, the hot spot is remediated and resurveyed.

• If the average concentration (above background) in the survey unit exceeds the DCGLW, or the statistical test fails to reject the Null Hypothesis, the entire survey may have to be remediated and resurveyed.

• The statistical test might fail to reject the Null Hypothesis because the test was performed incorrectly. Check to see if this was the case.

• The statistical test might fail to reject the Null Hypothesis because not enough measurements/samples were obtained. If there is reason to believe that this is the case, the regulator might permit one round of double sampling. Some indications that not enough measurements/samples were taken include an observation that the actual standard deviation of the measurements was greater than that estimated when the relative shift was calculated, and/or the average of the measurements was higher than the LBGR. In double sampling, additional measurements/samples are obtained at randomly selected locations. These measurements are added to the pool of measurements already obtained and the statistical test is redone.

• The statistical test might fail to reject the Null Hypothesis because the reference area measurements were lower than appropriate for the survey unit. Assess the suitability of the reference area.

• The DCGLW or DCGLEMC might have been too conservative (low). Consider reevaluating the assumptions/parameters in the dose modeling. Note that this is a last resort. Evaluations of the dose modeling should have been done much earlier.

• Consider releasing the site under restricted conditions.

4 s

k i=1 2

B =

k i=1 k

i=1 n

j=1 2

xi xij n i ( )

2 n i k 1 k

i=1n i K

N(N+1)

=

12 i

R2 3(N+1) s k

k k

ni i=1 i=1 i=1 j=1 2

w =

x2 2

ij xi n i( )

n i

(

)

1

Table 5.3 from MARSSIM Values of N/2 for Use with the Wilcoxon Rank Sum Test.

To achieve the desired DQOs (acceptable rates of Type I and Type II errors), the total number of measurements is N.

N/2 measurements are taken in the survey unit, and N/2 are taken in the reference area.

/ is the relative shift.

Table 5.5 from MARSSIM Values of N for Use with the Sign Test.

To achieve the chosen DQOs (acceptable rates of Type I and Type II errors),

N measurements are taken in the survey unit. There are no reference area measurements.

/ is the relative shift.

Critical Value for Wilcoxon Rank Sum Test Critical Value for Sign Test m is the number of measurements in the reference area (WRS test) n is the number of measurements in the survey unit (WRS test)

N is the number of measurements in the survey unit (Sign test) z = 1.96 when = 0.025 z = 1.645 when = 0.05 z = 1.282 when = 0.1

MARSSIM IS FLEXIBLE - TAKE ADVANTAGE OF IT!

)1

(

12

)1

(

2

+

+

+

+

+

m n

m n

z m

n m

N z

N 2

2 +