p070 A Comparison of Transient Dose Model Predictions and Experimental Measurements

Paper A COMPARISON OF TRANSIENT DOSE MODEL PREDICTIONS AND EXPERIMENTAL MEASUREMENTS R. L. Steinman,* R. F. Weiner,† and K. J. Kearfott* version 5, contains routines to estimate radiological and non-radiological consequences and risks to various populations of interest, including potentially exposed populations along the transport route, populations that share the transport route, occupationally exposed populations, and the maximally exposed individual. In addition, RADTRAN has the capability to analyze different transportation route combinations, allowing the analyst to break a route into up to 20 segments that share similar road and population characteristics. Researchers at Argonne National Laboratory (ANL) developed RISKIND (Yuan et al. 1995) in 1993 to perform scenario specific individual transportation risk analysis calculations, primarily to provide an efficient means to address specific public concerns. Although RISKIND performs a set of calculations similar to those performed by RADTRAN, the models used to represent the radiation field within a few meters surrounding an intact transport package and potential accidental dispersion scenarios are significantly different (Steinman and Kearfott 2000). RISKIND is capable of calculating up to 20 individual dose scenarios in a single run of the code. Additionally, it can calculate scenario specific population-averaged doses for subpopulation groups such as the elderly, school children, or prison inmates. However, the RISKIND model can accommodate only truck and rail transport modes. Although both RADTRAN and RISKIND have been evaluated for user friendliness (Brumburgh and Alesso 1993) and their output has been compared to other codes that perform similar calculations (Maheras and Pippen 1995; Biwer et al. 1997), none of the mathematical models have been validated against experimental data. This paper evaluates the RADTRAN and RISKIND models used to estimate the transient dose to an individual, specifically comparing the results of the experimental measurements to the established mathematical models in an effort to determine which model more closely reflects actual dose measurements from a moving package. The experimental measurements were initially performed for the Department of Energy sponsored National Transportation Program to verify the transient dose model prediction that dose from a passing, or transient,

Abstract—The RADTRAN and RISKIND transportation risk analysis computer codes are the primary tools used to estimate dose consequences and risks associated with the transport of radioactive material. Over the years, some of the mathematical models used within the two computer codes have been updated and the methodologies to calculate input parameters have been improved. In addition, both codes have been evaluated for ease of use and appropriateness of application and verified against other computer codes that perform similar calculations. However, neither code has been validated against experimental data. This report discusses the results of five sets of experimental measurements used to partially validate the specific mathematical models used to predict the dose to an individual due to a passing shipment of radioactive material within the RADTRAN and RISKIND computer codes. Based on the comparisons it was found that RISKIND most closely predicted the measured dose in the majority of the investigated scenarios and that 12 out of 14 cases demonstrate the expected inverse relationship between the measured dose and the distance of closest approach. Only half of the data demonstrated the expected inverse relationship between dose and speed of travel. Health Phys. 83(4):504 –511; 2002 Key words: computer calculations; risk analysis; dose assessment; transportation

INTRODUCTION THE RADTRAN and RISKIND transportation risk analysis computer codes are important tools for the assessment of the potential risks associated with the transportation of radioactive material (RAM). Sandia National Laboratories (SNL) first developed RADTRAN in 1977 to perform transportation risk analysis calculations for the first comprehensive environmental assessment of radioactive materials transportation, the Final Environmental Impact Statement on the Transportation of Radioactive Material by Air and Other Modes (U.S. NRC 1977). RADTRAN, now in * University of Michigan, Nuclear Engineering and Radiological Sciences, Ann Arbor, MI 48109-2104; † Jason Technologies, Las Vegas, NV 89146. For correspondence or reprints contact: R. L. Steinman, Advent Engineering Services, Inc., Prairie House, Domino’s Farms, PO Box 555, Ann Arbor, MI 48105-0005, or email at [email protected]. (Manuscript received 14 March 2001; revised manuscript received 5 February 2002, accepted 6 June 2002) 0017-9078/02/0 Copyright © 2002 Health Physics Society 504

Transient dose model predictions and experimental measurements ● R. L. STEINMAN

shipment of radioactive material is inversely proportional to both the distance of closest approach and the speed of travel. BACKGROUND In the incident-free transient dose model, the dose experienced by a receptor is a function of (1) the external radiation field surrounding the package, (2) the distance between the vehicle centerline and the receptor of interest, and (3) the speed at which the vehicle passes the receptor. The RADTRAN model essentially transforms the non-uniform radiation field around an actual package into an isotropic point source, such that the radiation field strength at a distance of 1 meter plus one half the characteristic package dimension (CPD) is equal to the highest measured dose rate (DRPKG) at 1 m away from the actual package surface or from the outermost surface of a closed vehicle, as shown in eqn (1):

DR i 共 r 兲 ⫽ f i ⫻ DR PKG ⫻ e ⫺ ␮ ir ⫻ Bi共r兲

冋

册

共1 ⫹ 0.5de 兲 r

ET AL.

Since the sources that were measured during the experiments emit only gamma radiation, for the purposes of this discussion, the neutron component of eqn (1) can be ignored. Thus the dose to an individual from a single shipment of gamma emitting material can be simplified to

k o ⫻ DR PKG ⫻ e ⫺ ␮ ␥r ⫻ B ␥ 共 r 兲 D ␥共 r 兲 ⫽ , r2

(1)

where i ⫽ ␥ for photons or n for neutrons; fi ⫽ the fraction of DRpkg attributed to radiation type i; e⫺␮ir⫻Bi(r) ⫽ the term representing attenuation and build-up in air; m ⫽ 2 when r ⱖ 2de or 1 when r ⬍2de; r ⫽ source-to-receptor distance; and de ⫽ effective package dimension, which equals the CPD for packages ⱕ4 m and 2(1 ⫹ 0.5⫻CPD)3/4 ⫺ 0.55 for packages ⬎4 m. The RADTRAN model presented in eqn (1) neglects potential contributions to dose from groundscatter and skyshine. For transient doses to receptors at distances from the source greater than the source dimension itself, the RADTRAN model assumes that the exponent, m, equals 2, i.e., the source geometry is represented as a point rather than a line. Since the largest population in incident-free dose calculations (i.e., those scenarios that do not include accidents) is along the side of the road and typically at distances of 30 m or greater, this approximation is reasonable. However, since members of the public are becoming increasingly concerned about specific doses, such as those to passengers in passing cars, pedestrians at intersections where the shipment stops, or hitchhikers who tend to be closer than 30 m, it seemed prudent to validate this approximation for closer sourceto-receptor distances.

(2)

where ko is used to represent the package shape factor, (1 ⫹ 0.5de)2. Eqn (2) is then integrated from ⫺⬁ to ⬁ with respect to time, which is related to the change in r via the speed of travel, to arrive at the mathematical model used to predict the integrated dose to an individual due to a passing shipment of radioactive material. The integrated gamma dose equation is given by eqn (3), where x represents the perpendicular distance between the receptor and the package line of travel (distance of closest approach) and v represents the speed at which the shipment passes the individual:

m

,

505

D ␥共 x 兲 ⫽

2 ⫻ k o ⫻ DR PKG v

冕

⬁

x

e ⫺ ␮ ir Bi 共r兲 dr. r 冑r2 ⫺ x2

(3)

For gamma radiation energies typically of interest to a transportation risk analyst, 0.4 – 0.9 MeV (Sandquist et al. 1985; Weiner and Neuhauser 1991), the product of gamma attenuation and build-up in air is generally less than 1. The RADTRAN model conservatively assumes this product equals 1, reducing the integral to a form that has an analytic solution (Neuhauser et al. 2000). The simplified result of the integration, provided in eqn (4), is actually used to evaluate how well the RADTRAN model fits the experimentally measured data presented in this report:‡

D ␥共 x 兲 ⫽

k o ⫻ ␲ ⫻ DR PKG . vx

(4)

The RISKIND external radiation field model is based on a 3-D Monte Carlo transport calculation for a representative spent nuclear fuel (SNF) cask (Chen and Yuan 1988). The dose rate expression used in RISKIND is an empirical fit [provided in eqn (5)] to the results of a MORSE-SGC/S (Emmett 1984) calculation that includes the effects of groundscatter and skyshine, as well as energy dependant attenuation and build-up. The coefficients, Aji, are normalized such that the dose rate at a ‡ Note that the RADTRAN model, as available in the RADTRAN code, incorporates the effect of a distributed population along the roadside. Since the code does not easily allow (it can be done, just not easily) the calculation for a single individual, eqn (4) as listed above was used in a spreadsheet to calculate the RADTRAN model values.

506

Health Physics

October 2002, Volume 83, Number 4

source-to-receptor distance of 2 m is equal to the regulatory limit of 0.1 mSv h⫺1 (10 mrem h⫺1):

冘A 关log r兴 , 7

log关DRi 共r兲兴 ⫽

j

(5)

ji

j⫽0

where i ⫽ ␥ for photons or n for neutrons, and r ⫽ source-to-receptor distance measured from the package surface. Actually, r ⫽ ro ⫹ TIOFF, where ro is the distance from the vehicle edge to the receptor and TIOFF is the distance from the vehicle edge to the surface of the package. The RISKIND individual transient dose model is essentially identical to eqn (3), except that in RISKIND the external dose rate, DR, is a function of source-toreceptor distance, r, so it must remain inside the integral. Thus, the RISKIND model is not easily reduced to a simple analytical solution. The RISKIND equation for total integrated dose to an individual at a perpendicular distance of closest approach, x, is given by eqn (6). All of the variables in eqn (6) represent the same quantities as previously described for the RADTRAN equation:

D ␥共 x 兲 ⫽

2 v

冕

⬁

x

DR ␥ 共 r 兲

冑r

r 2

⫺ x2

dr.

(6)

EXPERIMENTAL METHODOLOGY Five sets of experimentally measured transient dose data were recorded for on-site shipments of radioactive material occurring between November 1998 and March 2000 at the Hanford Site in Richland, Washington, by employees of Waste Management Federal Services, Inc., Northwest Operations. All of the measurements were made using the General Electric Reuter-Stokes RSS-112 PIC Portable Environmental Radiation Monitor, which consists of a tripod mounted spherical 7.2-L, argon filled high-pressure ion chamber and the battery-powered RSS112 data acquisition hardware (Reuter-Stokes 1995) as shown in Fig. 1. During the experiments, 5-s timeaveraged dose rate data were collected and written to a memory cartridge. Later the data were downloaded from the cartridge into EXCEL and summed to determine the total integrated dose experienced by the detector during a single pass of the radioactive source term. The essential parameters for each set of experimental data, such as the transport index, speed, distance of closest approach, and characteristic package dimension (CPD), are summarized in Table 1. In all cases, (1) the detector was located on top of the tri-pod assembly, such that the detector remained approximately 1 m above the ground (or bed of the truck); (2) the distance of closest

Fig. 1. The General Electric Reuter-Stokes RSS-112 PIC portable environmental radiation monitor.

approach, denoted by x, was measured from the package surface to the centerline of the detector; and (3) the package dose rate (DRPKG) was determined from the maximum value recorded for 5 dose rate measurements made at various radial positions 1 m from the surface of the actual package. Since measurement cases B through E were performed on paved roads using similar open bed trucks, the groundshine conditions were assumed identical for those cases. Additionally, a PATRAM ’92 paper by Weiner and Neuhauser (1992) demonstrates that effects of groundshine are negligible at the distances of concern for the measurements described in this paper. Thus, groundshine contributions were not explicitly included in the RADTRAN model calculations. Case A represents transient dose measurements recorded in July 1998 for a train pulling two noncontaminated tanker cars and one empty single shell 137 Cs contaminated liquid waste tanker that was located at the end of the train (McFadden and Weiner 1998). Two sets of Reuter-Stokes instruments were staged perpendicular to the track at a distance of 10 m. The instrument located in the 200 East Area at Hanford

Transient dose model predictions and experimental measurements ● R. L. STEINMAN

ET AL.

507

Table 1. Important parameters of the five measured scenarios. Data set designation

Waste type (isotope)

TI (␮Sv/h)

A

Rail tank car (137Cs)

700

B

208-L drums (TRU)

70a

C

208-L drums (TRU)

71

D

208-L drums

E

Well logging source (241Am and 137Cs)

6.8 12

Speeds (km/h) 16 32 16 32 16 32 16 8 16

Closest approach (m)

CPDb (m)

10

5

7 12 17 3 5 7 8 1 3 1 3

3.5 1.8

1.8 1

a

This value is an estimated TI based on the health physics survey reports for individual drums, since the value was not actually measured during the experiment. b The CPD is the characteristic package dimensions, typically the largest dimension is used. In the cases examined here, the CPD was the length of the cylindrical package.

recorded transient data as the train passed at 16 km h⫺1 (10 mph). The other instrument was located to the west of the 200 East Area and recorded data as the train passed at 32 km h⫺1 (20 mph). The remaining four measurements were for truck shipments of contaminated waste material. The detector moved past a staged (stationary) source in the case of the first two truck shipments (B and C). Essentially, the RSS-112 was secured in the bed of a pick-up truck as depicted in Fig. 2 and driven past the staged waste truck. This is mathematically equivalent to a stationary receptor and a moving source, with the exception of the change in detector height that occurred when the detector was placed in the bed of the truck (increased height by approximately 0.25 m). However, it was determined that the height of the detector had negligible effect on the measured doses compared to other sources of error described below and thus was not explicitly modeled. In the remaining truck measurements (D and E), the source was driven past a stationary detector as depicted in Fig. 3.

Fig. 2. RSS-112 Detector System secured in pick-up truck bed.

Fig. 3. Experimental set-up for transient truck shipment measurements.

Case B represents transient dose measurements made in April 1999. The source term for this case consisted of twelve 208-L (55-gallon) drums staged on a flatbed trailer parked east of Building 327. Eleven of the drums contained transuranic contaminated waste material; the twelfth drum was empty and used solely to maintain a full transport array configuration. The drums were arranged in a 6 by 2 array such that the line of six drums were parallel to the long side of the trailer. The overall dimensions of the array were measured to be 3.5 m long by 1.2 m wide by 87.5 cm high. The array was located at the front of the trailer approximately centered on its width. Since the truck was not hooked up to the trailer, scatter from any surfaces near to the array could be ignored. Measurements were made by strapping the Reuter-Stokes tripod in the open bed of a pick-up as described previously.

508

Health Physics

Case C again represents a staged array of 208-L (55-gallon) drums in an open truck parked in a lot in the 300 Area during July 1999. In this case, three drums with measurable surface dose rates were aligned along the edge of the truck bed. The measurements were made in a manner similar to that described for Case B above. Case D represents an array composed of five 208-L (55-gallon) drums and a wooden box staged on a flatbed truck parked east of Building 327 in November 1999. The drums contained transuranic contaminated materials; the wooden box was filled with non-radioactive waste material. The box was located directly behind the driver cab compartment, followed by two rows of drums. Dose rate data were measured for two speeds as the truck moved past the stationary detectors along Route 240. The final set of preliminary measurements (Case E) were performed in March 2000 using a well logging calibration source containing 1.85 GBq 241Am and 0.37 GBq 137Cs in a DOT 7A, Type A package. Data were collected at 1 and 3 m distances from the edge of Albany Avenue as the truck drove past the stationary detectors. The package was left partially open to allow a higher than normal level of radiation to escape the package,

October 2002, Volume 83, Number 4

which helped to distinguish the transient dose data from the 0.001 mGy h⫺1 (0.1 mR h⫺1) background in the area. It was estimated that the variation in speed could be as much as 10% on either side of the value designated in Table 1. The error in the distance of closest approach, x, was estimated to be ⫾0.5 m. The error associated with the detector is ⫾5%, which is the highest relative percent error associated with the instrument calibration reports for the isotopes of interest (primarily 137Cs) to the measurements presented here. RESULTS AND DISCUSSION Fig. 4 compares the measured dose values to the doses calculated using the RADTRAN and RISKIND models previously provided in eqns (4) and (6), respectively. However, one must note that since RADTRAN, unlike RISKIND, measures the distance of closest approach, x, from the point-source (i.e., geometrical center of the package) rather than the package surface, the value of x provided in Table 1 must be manually modified prior to use in eqn (4). The modification to the value of x simply requires the addition of the distance from the

Fig. 4. Comparison of transient dose models and experimental measurements.

Transient dose model predictions and experimental measurements ● R. L. STEINMAN

509

ET AL.

Table 2. Case-by-case experimental measurement scenario parameters. Case number

Radial distance, x (m)

Speed (km/h)

Package length (m)

Package radius (m)

Dose rate at 1 m (␮Sv/h)

Measured dosea (␮Sv)

A1 A2 B1 B2 B3 B4 B5 C1 C2 C3 C4 C5 C6 D1 D2 D3 D4 E1 E2 E3 E4

10 10 7 12 17 12 17 3 5 7 8 5 8 1 3 1 3 1 3 1 3

16 32 16 16 16 32 32 16 16 16 16 32 32 16 16 32 32 8 8 16 16

5 5 3.5 3.5 3.5 3.5 3.5 1.8 1.8 1.8 1.8 1.8 1.8 1.8 1.8 1.8 1.8 1 1 1 1

0.5 0.5 0.6 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.4 0.4 0.6 0.6 0.6 0.6 0.3 0.3 0.3 0.3

700 700 70 70 70 70 70 71 71 71 71 71 71 6.8 6.8 6.8 6.8 12 12 12 12

4.90 ⫻ 10⫺2 4.20 ⫻ 10⫺2 1.30 ⫻ 10⫺2 8.00 ⫻ 10⫺3 5.00 ⫻ 10⫺3 4.20 ⫻ 10⫺3 2.40 ⫻ 10⫺3 6.55 ⫻ 10⫺3 3.84 ⫻ 10⫺3 2.28 ⫻ 10⫺3 2.03 ⫻ 10⫺3 2.32 ⫻ 10⫺3 1.24 ⫻ 10⫺3 9.49 ⫻ 10⫺4 9.38 ⫻ 10⫺4 7.00 ⫻ 10⫺4 2.63 ⫻ 10⫺4 1.01 ⫻ 10⫺3 4.36 ⫻ 10⫺4 6.46 ⫻ 10⫺4 2.53 ⫻ 10⫺4

a

Note the measured value represents a single recorded measurement for all of the cases, except cases E1 thru E4. In Cases E1 and E2 the reported dose is the average of 14 individual measurements of the integrated dose at the specified speed and distance of closest approach. The value reported for cases E3 and E4 represent the average of 15 individual measurements.

geometric center to the edge of the actual package to the value of x. By virtue of the way the RISKIND cylindrical approximation was created, the package radius provided in Table 2 provides the appropriate modification of x for use in eqn (4). Both the RADTRAN and RISKIND models predict doses within an order of magnitude of the measured integrated individual dose value. There are no RISKIND values reported for data set A because RISKIND automatically forces the dose rate to comply with the regulatory limit of 10 mrem h⫺1 at 1 m, if a higher dose rate is entered, as in the case of data set A, RISKIND automatically resets the dose rate to 10 mrem h⫺1. In data set B, which is for relatively “large” distances of closest approach, the RADTRAN model provides dose predictions closer to the measured dose and RISKIND consistently under predicted the dose. In data sets C and D, the RISKIND model generally predicts doses that are closer to the measured values than the RADTRAN model. In data set E both models over predict the measured dose, leading to the belief that some other confounding factor that has yet to be identified affected the model results. Thus, data sets A and E are excluded from further discussion. Table 2 lists the experimental parameters for each specific case that was measured for the five scenarios. In order to create a RADTRAN input file, a dimension to be used as the characteristic package dimension, or CPD, had to be chosen. Since the array of drums formed a rectangular shape, the CPD was initially chosen as the

longest internal diagonal of the array, as suggested by the RADTRAN 5 Technical Manual (Neuhauser et al. 2000). However, the RADTRAN results using this value for the CPD predicted integrated doses that were not consistent with the measured values. Thus, the CPD was revised by assuming a cylinder could adequately represent the array geometry, making the CPD the length of the array. The dimensions for the cylinder approximation to the array were chosen such that the cylinder length equaled the longest dimension of the rectangular array and its radius equaled exactly half the height of the array. The cylinder dimensions are the ones provided as CPD values in Table 2. Table 3 demonstrates the inverse relationship between integrated dose and the velocity at which a shipment passes a stationary receptor by comparing the ratio of integrated dose at 16 and 32 km h⫺1 [10 and 20 mph, respectively] to the expected ratio based on the mathematically derived principle that dose is inversely proportional to velocity. Assuming the velocity varies by no more than ⫾10%, the expected relative error associated with the velocity ratio is 14%, such that the expected ratio of velocities equals 2 ⫾ 0.28. Assuming a detector error of ⫾5%, the relative error associated with the dose ratio is 7%. The expected ratio of doses plus or minus one sigma is provided in the table for each investigated scenario. Fig. 5 graphically depicts the data from Table 3, which shows that the dose ratio agrees with the predicted 1/v relationship in only half of the investigated cases.

510

Health Physics

October 2002, Volume 83, Number 4

Table 3. Investigation of the inverse relationship between dose and speed. Case number

Scenario number

16 km/h dose (␮Sv)

Scenario number

32 km/h dose (␮Sv)

Ratio ⫾ ␴Ratio

1 2 3 4 5 6 7 8 9

A1 B2 B3 C2 C4 D1 D2 E1 E2

4.90 ⫻ 10⫺2 8.00 ⫻ 10⫺3 5.00 ⫻ 10⫺3 3.84 ⫻ 10⫺3 2.03 ⫻ 10⫺3 9.49 ⫻ 10⫺4 9.38 ⫻ 10⫺4 1.01 ⫻ 10⫺3 4.36 ⫻ 10⫺4

A2 B4 B5 C5 C6 D3 D4 E3 E4

4.20 ⫻ 10⫺2 4.20 ⫻ 10⫺3 2.40 ⫻ 10⫺3 2.32 ⫻ 10⫺3 1.24 ⫻ 10⫺3 7.00 ⫻ 10⫺4 2.63 ⫻ 10⫺4 6.46 ⫻ 10⫺4 2.53 ⫻ 10⫺4

1.17 ⫾ 0.08 1.90 ⫾ 0.13 2.08 ⫾ 0.15 1.66 ⫾ 0.12 1.64 ⫾ 0.12 1.36 ⫾ 0.10 3.57 ⫾ 0.25 1.56 ⫾ 0.11 1.72 ⫾ 0.12

CONCLUSION

Fig. 5. Inverse relation between dose and speed of travel.

Table 4 investigates the inverse relationship between integrated dose and distance of closest approach, x. The same data are graphically presented in Fig. 6. In 86% (12/14) of the cases investigated, the measured dose ratio lies within a single standard deviation of the expected ratio. Thus, this data appears to generally support the inverse relationship between integrated dose and the distance of closest approach.

The experimental measurement and transportation model comparison outlined in this paper was designed to demonstrate to concerned members of the public that the models used to predict doses due to the transport of radioactive materials are adequate for that purpose. Since the mathematical discussions are often difficult for people without extensive math backgrounds, an experimental demonstration was devised. The experimental measurements described in this paper show reasonable agreement (86%) with the expected inverse relation between dose and the distance of closest approach; however, the data do not conclusively show the mathematically required inverse relation between dose and speed. Thus, for the purposes of the described demonstration, further measurements are necessary if the inverse speed relationship is to be conclusively demonstrated beyond the mathematical derivations provided in the RADTRAN and RISKIND manuals. Specifically, measurements, which can more accurately control passing speed and source-to-receptor distance, need to be devised. Additionally, it was shown that even though the default RISKIND dose rate curve was derived for spent fuel packages rather than the actual measured waste

Table 4. Inverse relationship between dose and distance of closest approach. x1 (m)

D1 (␮Sv)

x2 (m)

D2 (␮Sv)

v (km/h)

Dose ratio (D1/D2)

Expected ratio (x2/x1)

7 7 12 3 5 7 3 5 3 5 1 1 1 1

1.30 ⫻ 10⫺2 1.30 ⫻ 10⫺2 4.20 ⫻ 10⫺3 6.55 ⫻ 10⫺3 3.84 ⫻ 10⫺3 2.28 ⫻ 10⫺3 6.55 ⫻ 10⫺3 3.84 ⫻ 10⫺3 6.55 ⫻ 10⫺3 2.32 ⫻ 10⫺3 9.49 ⫻ 10⫺4 7.00 ⫻ 10⫺4 1.01 ⫻ 10⫺3 6.46 ⫻ 10⫺4

12 17 17 5 7 8 7 8 8 8 3 3 3 3

8.00 ⫻ 10⫺3 5.00 ⫻ 10⫺3 2.40 ⫻ 10⫺3 3.84 ⫻ 10⫺3 2.28 ⫻ 10⫺3 2.03 ⫻ 10⫺3 2.28 ⫻ 10⫺3 2.03 ⫻ 10⫺3 2.03 ⫻ 10⫺3 1.24 ⫻ 10⫺3 9.38 ⫻ 10⫺4 2.63 ⫻ 10⫺4 4.36 ⫻ 10⫺4 2.53 ⫻ 10⫺4

16 16 32 16 16 16 16 16 16 32 16 32 8 16

1.63 ⫾ 0.11 2.60 ⫾ 0.18 1.75 ⫾ 0.12 1.70 ⫾ 0.12 1.68 ⫾ 0.12 1.12 ⫾ 0.08 2.87 ⫾ 0.20 1.89 ⫾ 0.13 3.23 ⫾ 0.23 1.87 ⫾ 0.13 1.01 ⫾ 0.07 2.66 ⫾ 0.19 2.32 ⫾ 0.16 2.55 ⫾ 0.18

1.71 ⫾ 0.24 2.43 ⫾ 0.34 1.42 ⫾ 0.20 1.67 ⫾ 0.24 1.40 ⫾ 0.20 1.14 ⫾ 0.16 2.33 ⫾ 0.33 1.60 ⫾ 0.23 2.67 ⫾ 0.38 1.60 ⫾ 0.23 3.00 ⫾ 0.42 3.00 ⫾ 0.42 3.00 ⫾ 0.42 3.00 ⫾ 0.42

Transient dose model predictions and experimental measurements ● R. L. STEINMAN

ET AL.

511

of Energy’s Office of Nuclear Energy, Science, and Technology under contract number DE-ACO4-94AL85000.

REFERENCES

Fig. 6. Inverse relation between dose and distance of closest approach.

packages, RISKIND more accurately predicted the experimentally measured transient integrated dose for large packages at close distances. This result is primarily due to the fact that the RADTRAN point source model tends to be overly conservative close to the source, whereas the cylindrical model used in RISKIND more accurately reflects the real package geometry and thus its external radiation field characteristics. However, at distances more typical of residential populations alongside the road, the RADTRAN model provides better agreement with the actual experienced doses. Thus, the results of these measurements show that the on-link component of the transient dose model, the component that estimates the dose to people in vehicles sharing the route with the radioactive shipment, is where the comparison of the RADTRAN and RISKIND models should be applied. The degree to which the on-link models over or under estimate the dose will be the topic of a future paper.

Acknowledgments—The authors would like to acknowledge the assistance of Jason L. Boles, Janet G. McFadden, and Erik Neilsen of Waste Management Federal Services, Northwest Operations, for their assistance in performing the experimental measurements used in this benchmarking project, which were funded by the U.S. Department of Energy’s Office of Transportation (EM-22).

This research was performed under appointment to the Nuclear Engineering/Health Physics Fellowship Program sponsored by the U.S. Department

Biwer BM, Arnish JJ, Kamboj S, Chen S-Y. RISKIND verification and benchmark comparisons. Argonne, IL: Argonne National Laboratory; ANL/EAD/TM-74; 1997. Brumburgh GP, Alesso HP. Comparison of RISKIND and RADTRAN 4. Livermore, CA: Lawrence Livermore National Laboratory; UCRL-ID-115618; 1993. Chen S-Y, Yuan YC. Calculation of radiation dose rates from a spent nuclear fuel shipping cask. Transactions of the American Nuclear Society 56:110 –112; 1988. Emmett MB. MORSE Monte Carlo radiation transport code system. Oak Ridge, TN: Oak Ridge National Laboratory; ORNL-4972 R2; 1984. Maheras SJ, Pippen HK. Validation of the transportation computer codes HIGHWAY, INTERLINE, RADTRAN 4, and RISKIND. Idaho Falls, ID: Science Applications International Corporation; DOE/ID-10511; 1995. McFadden JG, Weiner RF. RADTRAN benchmark project preliminary report. Hanford, WA: Waste Management Federal Services, Inc., Northwest Operations; ENG-RPT-019; 1998. Neuhauser KS, Kanipe FL, Weiner RF. RADTRAN 5: Technical manual. Albuquerque, NM: Sandia National Laboratories; SAND2000-1256; 2000. Reuter-Stokes. RSS-112 PIC Portable Environmental Radiation Monitor Operational Manual, Version 1.9. Twinsburg, OH: Reuter-Stokes; 1995. Sandquist GM, Rogers VC, Sutherland AA, Merrel GB. Exposures and health effects from spent nuclear fuel transportation. Salt Lake City, UT: Rogers and Associates Engineering Corporation; RAE-8339/12-1; 1985. Steinman RL, Kearfott KJ. A comparison of the RADTRAN 5 and RISKIND 1.11 incident-free dose models. Proceedings of the Waste Management 2000 “HHW, LLW, Mixed Wastes and Environmental Restoration—Working Towards a Cleaner Environment” Conference. Tucson, AZ: LOI, Inc.; 2000. U.S. NRC. Final environmental impact statement on the transportation of radioactive material by air and other modes. Washington, DC: U. S. Nuclear Regulatory Commission; NUREG-0170; 1977. Weiner RF, Neuhauser KS. Near-field radiation doses from transported spent nuclear fuel. Albuquerque, NM: Sandia National Laboratories; SAND91-2060C; 1991. Weiner RF, Neuhauser KS. Conservatism of RADTRAN line-source model for estimating worker exposures. PATRAM’92, 10th International Symposium on the Packaging and Transportation of Radioactive Materials. Yokohama City, Japan, 13–18 September 1992. Vienna: International Atomic Energy Agency; 1992. Yuan YC, Chen S-Y, Biwer BM, LePoire DJ. RISKIND: A computer program for calculating radiological consequences and health risks from transportation of spent nuclear fuel. Argonne, IL: Argonne National Laboratory; ANL/EAD-1; 1995. f f