Ask an Expert

Ask an Expert
Question and Answer Page
MARSSIM (survey design)

Submitted 8/31/04
Q: How many soil sampling locations are required for H-3 or Sr-90 when scanning options do not exist and area factors can not be calculated?

A: Good question - and no easy answers. Fortunately. in many cases the hard-to-detect radionuclides can be scaled to radionuclides that can be detected via scanning. But I'm assuming that's not the premise of your question.

Perhaps the most appropriate response would be to say, "Follow the DQO Process." This may not seem like much help, but it does provide a context for answering the basic question involved - "How much contamination are you willing to miss?" The answer of course will be dependent on the acceptable dose or risk of leaving behind contamination. The survey approach also must consider the acceptable errors of missing the specified amount of contamination deemed important not to miss. Also, it should be noted that contrary to the statement in your question, area factors can be calculated for any radionuclide, regardless of whether you can scan for them or not.

As mentioned above, the sampling design should be based on the data quality objectives process. When you have no scan capability, one possible approach to consider is as follows. Based on characterization data, determine the necessary MARSSIM final status survey sample size. Perform systematic sampling in the survey unit and analyze samples. Assess the data with a posting plot - results of the posting plot indicate potential locations for contamination. Determine if contamination levels may exceed those determined in the DQO Process to be important not to miss. Perform second stage sampling based on results of first sampling stage at locations where samples exceed DCGLw.

Submitted 3/26/04
Q: I am sampling a building to turn it over for public re-use. Can I use general area dose measurements say with a Bicron Microrem meter or do I need to do something different? The following is stated in DOE Order 5400.5: "Chaper IV, 4.c External Gamma Radiation. The average level of gamma radiation inside a building or habitable structure on a site to be released without restrictions shall not exceed the background level by more than 20 m R\hand shall comply with the basic dose limit when an "appropriate-use" scenario is considered. .... External gamma radiation levels or open lands shall also comply with the basic limit and the ALARA process, considering appropriate-use scenarios for the area.

A: Exposure rates historically have been measured during the final status survey to satisfy the appropriate exposure rate guideline (e.g., 20 μR/h in your example). While the industry standard for environmental (low level) measurements of exposure rate may be the pressurized ionization chamber (PIC), both the Bicron Micro-Rem meter and Ludlum model 19 micro-R meter can be used to measure exposure rates. A favorable attribute of the PIC is its relatively flat energy response across a wide gamma energy range; the response of the hand-held Bicron and Ludlum meters mentioned above are energy dependent to varying degrees.

With the move to dose-based release criteria, exposure rate measurements are not usually performed during the final status survey (since DCGLs are expressed in terms of building surface activity or soil concentrations). However, guidance in NUREG/CR-5849 "Manual for Conducting Radiological Surveys in Support of License Termination" (http://www.orau.org/ptp/5849/nureg5849.htm) states that exposure rate measurements should be conducted at 1 m above the ground at systematic and random locations, and at locations of elevated direct radiation as indicated by scans.

Submitted 3/16/04
Q: I am using DOE order 5400.5 table 1 release values for my DCGLs. I potentially have Group 1, 2, 3, and 4 radionuclides. My plan is to take gross measurements and apply the unity rule using Group 1 and 2. Is this allowable?

A: The MARSSIM Manual in Section 4.3.4 addresses this issue: "For situations where multiple radionuclides with their own DCGLs are present, a gross activity DCGL can be developed. This approach enables field measurement of gross activity, rather than determination of individual radionuclide activity, for comparison to the DCGL."

DOE 5400.5 provides surface activity guidelines for four groupings of radionuclides. Let's assume that from characterization data for a survey area you know that 20% of the contamination is from radionuclides with a guideline of 5,000 dpm/100 cm2, and 80% is from radionuclides with a guideline of 1000 dpm/100 cm2. Equation 4-4 in MARSSIM can be used to calculate the gross activity DCGL for this radionuclide mixture (1200 dpm/100 cm2). Note: The efficiency calculated for the detector must also be based (weighted) on the same radionuclide mixture used to determine the gross activity DCGL. In this example the unity rule would NOT be used; rather, you would calculate a gross activity DCGL to account for the radionuclides potentially present.

Now consider a situation where your contaminant mix includes both alpha- and beta emitting radionuclides - and where the survey strategy is to make separate measurements of alpha and beta radiation. In this circumstance you would likely end up determining gross alpha and gross beta DCGLs (as described above) - as well as using the unity rule. A general rule of thumb is that the unity rule is necessary whenever more than one measurement is performed at a specific location. So considering DOE 5400.5, alpha measurements might be performed for Group 1 (alpha emitters) and beta measurements performed for beta emitters from the other groups. Examples of this common situation can be found in Chapter 12 of Decommissioning Health Physics (2001).

Submitted 7/31/03
Q: This is a question of how to determine if a site is impacted or unimpacted. Complete historical records are not always available for military bases. The AF had a radiochemistry laboratory that began operations in the 50''s. No airborne stack release records have been found, although it appears to have done mostly environmental level work (although no one can be 100% sure of this because of the classified nature of the mission). The AF would like to back up their assessment that airborne releases did not impact surrounding property by collecting soil samples at 3 or 4 undisturbed locations (non random), which doesn't meet a Class 3 survey. My read on MARSSIM is that a site can't be declared unimpacted by a scoping/characterization survey alone. The AF would use the survey data along with the available historical information (NRC license records, process knowledge) to conclude the site is unimpacted. Do you think this is reasonable?

A: The submitted question relates to the impacted/non-impacted determination for a site. Fundamentally, non-impacted areas are those areas that have no reasonable potential to be contaminated. They are areas that would be considered acceptable as background reference areas precisely due to the virtually non-existent potential for contamination.

Submitted 6/30/03
Q: A couple of sanity check questions. If a Class 3 area had only one sample at depth above the DCGL, should it be reclassified as Class 1? Also, if a Class 1 area has 2 or 3 samples above the DCGLemc, would there be a situation where remediation of these areas was not appropriate (given the fact that it is established that remediation did not occur with sampling, no hot particle)?

A: Your first question concerns contamination above the DCGL identified in a Class 3 area. Reclassification is certainly warranted in this case, and likely to a Class 1 area. Perhaps more importantly is the need for additional characterization, as well as a review of the historical site assessment data, to determine the radiological condition in this and possibly other similar areas. Finding contamination above the DCGL in a Class 3 area is a flag that casts doubt on the validity of the classification process.

Submitted 2/4/03
Q: Should the scanMDC for a class 3 area be below the DCGL? Page 6-7 of MARSSIM (and page B-1 of Draft Vol 2 NUREG 1757, 9/02) recommends a scanMDC of 10-50% of the DCGL. Is this for class 1 only? If so, why? Page 5-46 of MARSSIM states: "Because there is a low expectation for residual radioactivity in a class 3 area, it may be prudent to investigate any measurement exceeding even a fraction of the DCGLw." Shouldn't this investigational criteria include scans and not just static measurements? If the scanMDC is allowed to be greater than the DCGL in a class 3 area, is there guidance as to how much greater?

A: This is a very good question, and the answer is certainly not as satisfying as one would like. Please allow a bit of background to develop a response to your insightful question. MARSSIM recommends two approaches for demonstrating compliance with release criteria: 1) a statistical approach to assess the average contamination level in a survey unit, and 2)scanning to identify potential hot spots in the survey unit. These two survey methods complement each other, and are equally important.

Submitted 10/16/02
Q: A site developed overly conservative DCGLs (assumed 6 feet depth)
and then applied them to survey units that they believed were Class 3.
When the survey unit failed to comply with the definition of a Class 3 area
(i.e. no samples over DCGLs), they modified the DCGLs, but only
for those sample locations that were above the conservative DCGLs.
Once they showed compliance with those sample locations, they declared
the site had passed. Is there anything wrong with this approach other than
making sure that the background reference area is also adjusted for the new less conservative DCGLs?

A: First, if the contamination is generally present to a depth of 6 feet at this site, then you have subsurface contamination and the guidance in MARSSIM does not strictly apply. Many sites that have subsurface contamination use characterization surveys to determine the extent of the contamination, and then remediate based on the characterization data. Following remediation the soil areas are largely uneven ("moon-scape" is the term often used), but the MARSSIM survey guidance can be used assuming that only surface soil contamination remains, even if the new surface is 2 m below grade. If this is not the case at your site, then you must rely on the DQO Process to develop appropriate, agreed upon, survey procedures.

Indeed, some of the MARSSIM guidance may still be warranted. For instance, the general rule for soil sampling is that the sampling depth should be consistent with the contaminated soil depth modeled to derive the DCGLs. MARSSIM states on p.7-11, "Sample collection procedures are concerned mainly with ensuring that a sample is representative of the sample media, is large enough to provide sufficient material to achieve the desired detection limit, and is consistent with assumptions used to develop the conceptual site model and the DCGLs." Furthermore, the manner in which the samples are collected from the survey unit should be the same for sampling in the background reference area.

In your example, the site developed DCGLs based on a contamination depth of 6 feet. Following the guidance in MARSSIM, each sample should have been collected over this depth to be consistent with the modeling. For example, core samples to a depth of 6 feet should have been collected and mixed to ensure that the soil sample results were consistent with the derived DCGLs.

A major limitation in this approach is that you have virtually zero scan sensitivity for contamination that is buried. For example, if the contamination in some site areas was present at a depth of 2 to 5 feet below the surface, but not present at the surface, then potentially large volumes of soil contamination could be missed due to essentially zero scan sensitivity.

Based on the limited information that could be reasonably presented about this site in this forum, it appears that the described approach has flaws. First, soil samples that exceed the DCGLw are not only problematic for Class 3 areas, but also Class 2 areas (depending on the number of samples greater than the DCGLw). Furthermore, changing the DCGL during the course of the final status survey to demonstrate that samples collected meet this new, modified DCGL is inconsistent with the MARSSIM process.
 

Submitted 10/7/02
Q: I design and perform a survey and later during the assessment phase, I see my real sigma is larger than first determined, indicating I should have taken 22 samples rather than than 20 as previously determined. However, this new number of samples includes 20% more than necessary unrelated to statistical issues. Since I have already taken 20 do I have to redo the sampling design and take a new set of 22 samples or can I assume I have satisfied the sampling design with the original 20 samples? Also, is there a method for adding a few random samples to make-up for a shorthanded survey design.

A: Your question is similar to the MARSSIM question submitted on October 8, 2002; please refer to the answer to that question for additional information. To answer your first question directly, if you pass the statistical test (i.e., reject the null hypothesis in MARSSIM), then the survey unit passes regardless of the fact that you underestimated the standard deviation. Now if you had underestimated the standard deviation to a greater degree, e.g., let's say that your retrospective analysis indicated that you should have collected 30 samples, then it's very likely that you wouldn't have passed. The 20% increase in sample size provides a bit of insurance against underestimated standard deviations, but only up to a point.

Your second question raises the issue of double sampling. Double sampling would be considered in the event that you failed the statistical test, and a retrospective analysis indicates that your survey unit mean is less than the DCGLw. In that case, it may be possible to collect additional random samples to supplement the initial sample size, and to perform the statistical test again. The regulator will need to provide approval for this practice beforehand because double sampling increases the Type I error from that planned. See Appendix C in the recently published draft NUREG-1757, volume 2. On p. C-6 the NRC states, "...double sampling should not be used as a substitute for adequate planning. If it is to be allowed, this should be agreed upon with NRC staff as part of the DQO process." Well said.
 

Submitted 10/8/02
Q: My question is, given any particular N, can I ignore up to 20% of the data for whatever reason and still pass the survey unit? MARSSIM says the additional data points are to account for unuseable data. If I have some unuseable data, how does this affect my power? Can I accept the data as long as I have at least enough data equivalent to N-20% per equation 5-2?

A: Thank you for submitting this question. Believe it or not, the MARSSIM guidance to collect an additional 20% above the statistically required sample size is for your (the MARSSIM user's) benefit. Your principal question seems to be "can I ignore the up to 20% of the data and still pass the survey unit?" The simple answer is 'yes'. However, allow me to elaborate on this point.

Submitted 9/6/02
Q: Here's a question or two. Let's say we establish our scan rate for a structural surface and then tell our field people to pause for a full minute over an area when a certain count rate has been reached. This is in effect static measurements of one minute duration that have not been located randomly but are in effect, biased measurement points. We will not include these values in the Compass program during the assessment phase. However, I would want to re-examine the variability of the survey unit using these points to determine if the correct number of randomly selected static points were previously calculated. If there were a sufficient number of random points originally selected then no further actions would be necessary. What happens when there is an increased variability due to the inclusion of these biased data points? Would this represent a re-do in Compass and a re-survey? Let's further suppose that these biased points indicate the survey unit would fail but the randomly selected points would indicate the unit passes? I assume this could be a possibility because the biased locations were located strictly based on elevated concentrations in the survey unit. What actions would you recommend?

A: You bring up a number of interesting points that I will attempt to answer. First, there are two types of measurements expected in each survey unit: random (or random-systematic) measurements and judgmental measurements. Only the random measurements should be used in the statistical assessment (via the hypothesis test) of whether the DCGLw (average guideline) has been met. Scanning is used to identify the presence of any hot spots, and subsequent measurements at these locations are not random, but judgmental measurements. Judgmental measurements are NOT included in the statistical test, but rather are compared to the DCGLemc for each hot spot identified. Note: It is acceptable for judgmental measurements to be greater than the DCGLw; however, all of the hot spots in the survey unit must be considered together, using Eqn 8-2 in MARSSIM.

Please visit this page again soon for additional questions submitted to our D&D experts and the answers that may assist you!

Return to Q&A Page