4
Radiation Transport Calculations

Even when DS86 was accepted and implemented, it was acknowledged that thermal-neutron fluence as estimated from cobalt-activation measurements “contradicts the calculated values by an ever-increasing factor that is five at 1000 m” (Roesch 1987). That led the contributors of the neutron measurements in the final DS86 report to say the following (Roesch 1987):

If the measured cobalt activations were accepted as correct representations of thermal fluences and the assumption then made that the calculated fluences on the ground are low by a factor that applies to all energies, then the proportion of neutron kerma in the mixed radiation field beyond 1000 m at Hiroshima would change from insignificant to significant. This leaves the possibility, however unlikely in our collective expert judgment, that the calculated neutron-kerma values are wrong. No known evidence contradicts this hypothesis. Therefore, the conclusion of this chapter on neutron measurements must be that the neutron doses are in doubt until further work is done.

The unsatisfactory performance of DS86 in calculating thermal-neutron activation was also noted in an independent review of the new dosimetry system performed by the National Research Council (NRC 1987).

Despite those problems, DS86 was implemented because of the improvements it offered over the previous dosimetry system, particularly in gamma-dose agreement with TLD measurement and in organ-dose calculation. At the time, those advantages and all the technical advantages of DS86 outweighed the inexplicable discrepancies in some of the thermal-neutron data. DS86 agreed with the limited data on sulfur activation above the threshold of about 2.5 MeV and did not show any important discrepancy with calculations for thermal-neutron activation of



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 4 Radiation Transport Calculations Even when DS86 was accepted and implemented, it was acknowledged that thermal-neutron fluence as estimated from cobalt-activation measurements “contradicts the calculated values by an ever-increasing factor that is five at 1000 m” (Roesch 1987). That led the contributors of the neutron measurements in the final DS86 report to say the following (Roesch 1987): If the measured cobalt activations were accepted as correct representations of thermal fluences and the assumption then made that the calculated fluences on the ground are low by a factor that applies to all energies, then the proportion of neutron kerma in the mixed radiation field beyond 1000 m at Hiroshima would change from insignificant to significant. This leaves the possibility, however unlikely in our collective expert judgment, that the calculated neutron-kerma values are wrong. No known evidence contradicts this hypothesis. Therefore, the conclusion of this chapter on neutron measurements must be that the neutron doses are in doubt until further work is done. The unsatisfactory performance of DS86 in calculating thermal-neutron activation was also noted in an independent review of the new dosimetry system performed by the National Research Council (NRC 1987). Despite those problems, DS86 was implemented because of the improvements it offered over the previous dosimetry system, particularly in gamma-dose agreement with TLD measurement and in organ-dose calculation. At the time, those advantages and all the technical advantages of DS86 outweighed the inexplicable discrepancies in some of the thermal-neutron data. DS86 agreed with the limited data on sulfur activation above the threshold of about 2.5 MeV and did not show any important discrepancy with calculations for thermal-neutron activation of

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) europium. No one knew the magnitude or importance of discrepancies in the calculation of thermal neutrons. Up to the point where they became important as activation markers of the output of the bomb, thermal neutrons were not considered important and were collapsed into a single energy bin in the neutron cross-section used for DS86. Furthermore, the gamma component of dose was considered to dominate the total absorbed dose to the organs of the survivors. Given those circumstances and the obvious advantages of DS86 as an overall dosimetry system for RERF, it was implemented despite unresolved misgivings about thermal neutrons. The radiation-transport calculation technology on which DS86 is based had advanced substantially since the preceding dosimetry system (T65D) was completed, and the technology has continued to advance since DS86. As more advances were made in the technology enabling radiation-transport calculations, as more-sophisticated activation-measurement techniques have been developed, and as more measurements have been made, the discrepancy between the neutrons measured in materials still retaining isotopic markers of the bomb neutrons and the neutron fluence calculated with DS86 has become more widely appreciated. However, recognition of a discrepancy between activation measurement of in situ materials and the activation calculated with DS86 does not identify the cause of the discrepancy. Physicists who have dealt with this type of problem have had and still have differences of opinion about the source of the discrepancy. Given the complexity of the calculations and measurements involved and the inherent uncertainties surrounding the bombings, any of several factors, singly or in combination, could be the source of the problem. The RERF dosimetry system, as already noted, is actually a series of very complex components consisting of, first, the calculated output spectrum of the bomb; the calculated portion of the detonation spectrum that actually escapes from the bomb casing; the transport of that spectrum through the fireball created by the explosion; the calculation of the interactions and geometric distribution of the radiations in the air in the city; and finally the transport of the air-over-ground spectrum through whatever shielding exists. As the spectrum of neutrons released from the nuclear explosion traverses those various environments, it is constantly changing because of the interaction of the neutrons with the elements in their path. The calculation of those interactions depends not only on the elements encountered between the nuclear explosion and the point of measurement, but also on the neutron cross sections of those materials, the abundance of the materials in the path of the neutrons, and how well the neutron cross sections of the materials are known. All those complex nuclear dimensions depend on the number and energy spectrum of the neutrons generated by the fissile material in the bomb. The bombs detonated in Hiroshima and Nagasaki were different from each other in the materials they contained and how they were designed. Therefore, the neutron spectra that were generated and that escaped from those two bombs were different. The Hiroshima bomb was unique in that it was the only one of that exact design ever detonated. In contrast, bombs of the Nagasaki-type have been built and

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) tested numerous times in numerous countries. Thus, the leakage spectrum, yield, benchmark measurements, and calculation codes for the Nagasaki bomb are much better known and understood than those for the Hiroshima bomb. When neutrons interact with elements in materials, they lose energy and the process generates gamma rays and lower-energy neutrons. At the point of nuclear explosion, neutrons can have several million electron volts of energy. By the time they create thermal-neutron activation, the neutron energy is measured in electron volts. The final dose to a person and the production of isotopes by thermal neutrons are determined not only by the neutrons released in the explosion and their transport through air, but also by their interactions with terrain and structures that act as shielding, which further dissipate neutron energy and reduce the quality and magnitude of the radiation exposure at any point of interest. To ensure accuracy in the doses calculated for survivors, all the terms in the calculation, from detonation to dose, must be understood and modeled accurately. The radiation doses for the survivors in Hiroshima and Nagasaki are not known a priori. Just as in any other radiation exposure event, the doses for irradiations of individuals have to be reconstructed. In Hiroshima and Nagasaki the doses for survivors are calculated by DS86. As pointed out in Chapter 2, the gamma-ray component of the organ doses to individuals constitutes the major portion of the total absorbed dose and is in good agreement with direct measurements of gamma-ray signals left by the bombs in the quartz grains in sample cores from exposed bricks and roof tiles. The smaller neutron doses are less certain and are more difficult to verify. Many efforts have been made to confirm calculated neutron doses by measuring the radioactivity induced in elemental material present at the time of the bombing. Benchmark measurements of this type have been made repeatedly for atomic-bomb tests, in radiation-accident reconstruction, and in occupational-radiation monitoring programs. Although the principles of such measurements are well known, their conduct depends totally on finding radioisotopes of sufficient activity to be measurable more than 50 years after the event. The constraints imposed by time and the destruction by the bombing itself make the absolute accuracy of activation measurements difficult at best. Thus, the ability to confirm any radiation-dose reconstruction calculation depends on the accuracy of the measurement with which the calculation is compared. After the implementation of DS86, measurement of neutron activation in material present at the time of the bombings suggested substantial disagreement between measured and calculated values of thermal-neutron fluence. Additional measurements tended to confirm the discrepancy. Given that the activation measurements showed a uniform trend of divergence from DS86, the first efforts to address the discrepancy were directed at the radiation calculations. The concerns that the neutron discrepancy raised for the accuracy of survivor doses and for radiation-transport methods led the radiation-physics and biomedical-research programs at the US Defense Nuclear Agency (DNA) to begin in 1988 to address these problems. The group at DNA had a long-standing involvement with

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) work in this field dating back to the end of World War II. DNA had sponsored the development of the transport method that made DS86 possible and had been involved in initial work pointing to the possibility that advances in radiation-transport technology could yield doses for the atomic-bomb survivors different from those calculated with T65D. Three other considerations drew DNA into the problem at this time. First was an appreciation of the importance of the Hiroshima and Nagasaki dosimetry to estimates of risk for radiation-induced cancer. The radiation-protection community, led by the National Council on Radiation Protection and Measurements (NCRP), established that cancer risk was almost the sole effect of low-dose radiation, and was forming the standards of radiation protection. Second were long-standing professional ties to the group at Science Applications International Corporation (SAIC) that implemented DS86. Third was primary sponsorship of developments in radiation-transport methodology at several of the national laboratories at that time. The broad program that DNA then undertook was carried out at Oak Ridge National Laboratory (ORNL), Los Alamos National Laboratory (LANL), Lawrence Livermore National Laboratory (LLNL), SAIC, the Aberdeen Pulse Reactor Facility (APRF), the Department of Energy Environmental Measurements Laboratory (EML), and the National Institute of Standards and Technology (NIST). Between 1988 and 1993, individual projects were undertaken to address many of the aspects of the discrepancy between calculated and measured thermal neutrons. In retrospect, when the project began, no one at DNA or on the Committee on Dosimetry for the RERF expected the problem to be so intractable or to elude resolution to this day. The first phase of the work concentrated on the validation of the underlying radiation-transport methodology used in DS86. This approach arose out of concern that DS86 was unable to accurately calculate the measured europium or cobalt activation in either Hiroshima or Nagasaki. That inability and inconsistent results in calculating results of controlled-activation experiments at the APRF led to the concern that there might be a fundamental unappreciated flaw in the basic discrete-ordinates (Sn) radiation-transport methodology. This phase of the work—which consisted of numerous projects designed to devise, investigate, and test the impact of possible solutions—was carried out between 1988 and 1991. It included the review of existing neutron cross-section data and evaluations, generation of new neutron cross-section information and evaluations, conduct of proof-of-principle and benchmark experiments, creation of new in situ measurement techniques, and the making of new in situ measurements. The results of this work were to remove the factor-of-10 disagreement between the neutron-activation calculations and measurements in Nagasaki. The results were achieved with the use of new nitrogen and oxygen cross sections for neutrons over a broad energy range, an increase in the number of neutron groups used in the calculation, and the use of different Sn techniques. However, it was possible to confirm agreement in Nagasaki only after new and better measurements were made at greater distances there and a benchmark activation-calculation experiment was completed at APRF (Straume and others

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 1994). Those changes not only removed the factor-of-10 disagreement between Nagasaki calculations and measurements, but also produced better overall agreement with the atomic-test results and supported the validity of discrete-ordinates radiation-transport methodology. However, the many improvements in measurement and calculation technology that resulted in agreement between the Nagasaki measurements and calculations did not resolve the discrepancy in Hiroshima, which gave rise to the second phase of the DNA effort, directed specifically toward efforts to resolve the Hiroshima neutron discrepancy. The Hiroshima neutron work took three forms. First, there was an examination of various hypotheses about unexpected ways in which the Hiroshima bomb might have disassembled during explosion. Second, a composite source-term sensitivity analysis was conducted to determine whether any plausible combination of bomb source term and fission-spectrum neutrons could be found that would match all the in situ measured data sets for detonations at a number of bomb burst heights. Third, a calculation exercise was undertaken to “find” a source term for the bomb that would agree with the in situ measurement data. Many of the suggested changes enhanced agreement between the measured and calculated doses of gamma rays and neutrons, but the funding of this program ended before the Hiroshima problem was solved. THE NAGASAKI DISCREPANCY The ratio of values calculated by DS86 to those measured in situ for europium and cobalt can be clearly seen to disagree in Figure 4–1 for both Nagasaki and Hiroshima. This comparison, made in 1988, is complicated by the fact that there is only one measurement beyond 1000 m in Nagasaki. Given the undue influence of the 1000 m point on comparisons in Nagasaki, a top priority at that time was to obtain additional activation measurements in Nagasaki. The desire to have isotopes other than europium and cobalt for comparison mandated a search for an isotope present in sufficient abundance to assay and with a sufficiently long half-life to still be plentiful. Accelerator mass spectrometry (AMS) was proposed as a method to measure small amounts of 36Cl (Haberstock and others 1986; Straume 1988). First, the development and proof-of-principle for this technique had to be undertaken, because this type of chlorine activation assay had not been done before. After the successful development of the AMS chlorine assay at LLNL, measurements were made from concrete cores taken 822, 1187, and 1261 m from the hypocenter in Nagasaki. The measurements were accepted only after they had been shown to be in good agreement with the results of benchmark experiments conducted first at LLNL and then at distance in an open-air reactor at APRF (Straume and others 1994). They constituted a major step in resolving the neutron discrepancy at Nagasaki. To test the APRF chlorine results, benchmark measurements were also conducted at APRF in 1992 and 1993. These experiments employed a multisphere

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) FIGURE 4–1 Nagasaki and Hiroshima C/M for Eu as a function of slant range. 1988.

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) neutron spectrometer with 12 detectors operating simultaneously at 300 to 1986 m. Neutron fluences and energy spectra were measured by EML using bare, cadmiumclad, and high-density polyethylene BF3 spheres (Goldhagen and others 1996). The thermal-neutron detectors were calibrated by using the NIST thermal-neutron beam (±12% at 95% confidence). The measurements illustrated the crucial role the water content can play in neutron spectral shape and thermal-neutron production. An increase in ground moisture from 8% to 29% produced a measured 1.46×increase in thermal neutrons and increased the percentage of neutrons greater than 0.1 MeV from about 27.5% to 34% of the spectrum. Calculation of those and other benchmark fluences was good overall and diverged from the measurements by only 50% at the worst point of comparison. Thus, the benchmarks support the overall integrity of the radiation-transport calculation methodology used in DS86. The demonstrated ability of Sn to determine benchmark measurements, to calculate nuclear-test results, and to calculate the thermal measurements at Nagasaki makes it highly unlikely that the transport methodology could be responsible for the remaining discrepancy in Hiroshima. That does not rule out the possibility of environments in which thermal neutrons are produced but are not well characterized or not well modeled. Inadequacies in assessing contributions from other elemental isotopes, such as potassium and boron, or the failure to account properly for the water content of the sample could easily create a factor-of-2 error in the thermal-neutron activation in a sample. If that happens, the apparent discrepancies between physical measurement and calculation could create the perception of a greater discrepancy than actually exists in the small fraction of the dose attributed to neutrons as calculated by DS86 for survivors. Calculation Codes and Cross Sections Concurrently with the development of new measurements in Nagasaki, changes were made to the discrete-ordinates calculation method (Kaul and others 1994). Some of the changes resulted from developments in the cross-section libraries available for neutrons—for example, refinements in the oxygen and nitrogen cross-sections between ENDF/B-5 and ENDF/B-6. Concomitant increases in computing capacity made it possible to increase the number of neutron groups from 46 to 174 and the number of gamma-ray groups from 23 to 38 (see Table 4–1). This change also increased the number of thermal-neutron groups from one to four, permitting much better resolution of the critical thermal-neutron calculations. Those changes in combination with a recalculated LANL source term (Whalen 1994; Woolson 1993), changes in the size and details of the air-over-ground geometry of the calculation, the use of newer discrete ordinates calculation and transport codes, as well as changes in the time-dependent source and geometry (see Table 4–1), reduced the calculated thermal activation near the Nagasaki hypocenter by nearly a factor of two and brought the discrete-ordinates calculation into closer agreement with Monte Carlo calculations (Whalen 1994).

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) TABLE 4–1 Feature-by-Feature Comparison of DS86 and Parameters Suggested in 1993 Prompt-Radiation Methodology DS86 Sources - LANL 1983 sources - 27 neutron energy groups - 20 angle bins 1993 Sources - LANL 1990 sources - 46 neutron-energy groups, 20 angles - reformatted to 174 energy groups and continuous angle distribution Geometry - Seven-zone air density profile - maximal radius 2800 m - maximal height 1500 m - maximal radial mesh 25 m Geometry - DS86 material compositions and profiles - continuous vertical density variation - maximum radius 3000 m - maximal height 2000 m - maximal radial mesh 25 m Cross sections - ENDF/B-5 - 46 neutron and 23 gamma-ray energy groups - scattering order P3 - custom weighted by region Cross sections - ENDF/B-6.2 - 174 neutron—38 gamma-ray energy groups - Scattering order P3 Code - DOT-4, 2-D discrete ordinates - first collision source - 240-direction angular quadrature - negative source fixup - convergence criterion 1×10−2 - weighted-difference flux calculation Code - DORT, 2-D discrete ordinates - first collision source - 240-direction angular quadrature - no negative source fixup - convergence criterion 1×10–3 - theta-weighted flux calculation Delayed -Radiation Methodology DS86 Time-dependent source - neutrons: augmented Maxwellian spectra, E to 2.5 MeV - gamma ray: empirical spectra 1993 Time-dependent source - neutrons: ENDF/B-6 spectra, E to 8 MeV - gamma rays: ENDF/B-6 spectra Time-dependent geometry - line-of- sight optical depth (g/cm3) from 2-D air density contours - contours from STLAMB hydrodynamic code Time-dependent geometry - 2-D air density contours - contours from STLAMB hydrodynamic code Transport codes - ANISN: transport in uniform air, 300 time steps - ANISN: transport in hydrodynamically perturbed air - morse: fluence perturbation due to the air-ground interference Transport codes - DORT: transport in 2-D air-over-ground, 12 time steps

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) As work in upgrading the Sn methods developed and the new ENDF/B-6 cross-sections became available, it was increasingly clear that the first visible resonance in the nitrogen cross section was not properly described. Given that that resonance occurs at about 433 keV, its impact on thermal-neutron calculation could be important. To address that concern, high-resolution measurement of the total neutron nitrogen cross section from 0.5 to 50 MeV was undertaken at the Oak Ridge Electron Linear Accelerator (Harvey and others 1992). Measurements indicated that the size and spin parity for the 433-keV resonance had been substantially misrepresented and that there was some degree of shape distortion even in the broader resonances at higher nitrogen energies. These data and data on oxygen cross sections from Germany (Cierjacks and others 1980; Drigo and others 1976) were used to reevaluate 14N and 16O at low energies with a multichannel R-matrix analysis of reactions in the 15N and 17O systems (Hale and others 1994). The incorporation of the newly evaluated cross sections for oxygen and nitrogen led to a revision of the evaluated nuclear data files to produce ENDF/B-6.2 and changed the calculated scattering in air, creating an inelastic scattering above about 3 MeV and more pronounced forward scattering of neutrons below about 1.5 MeV. Together, these factors resulted in a reduction in the calculated sulfur-activation relaxation length from 220 to 207 m and in closer agreement with tests of Nagasaki-like bombs. THE HIROSHIMA BOMB AND POSSIBLE OTHER SOURCE TERMS The combination of the changes in calculation methods, the new chlorine measurement, and the definition and implementation of new oxygen and nitrogen cross sections were all necessary to achieve agreement between the calculations and in situ measurements in Nagasaki. None of those improvements has been implemented in DS86, which has remained essentially unchanged since its inception in 1986. When applied to the calculations in Hiroshima, these same changes have made the agreement between calculations and measurements incrementally better, but have not removed the disagreement as they did in Nagasaki (see Figure 4–2). In Hiroshima, the use of the new nitrogen and oxygen cross sections and the refinement of the Sn techniques have produced calculations that agree more closely with sulfur activation near the hypocenter. The type of chlorine measurements that helped to remove the discrepancy in Nagasaki supports the substantial nature of the discrepancy in Hiroshima (Straume and others 1994). Given apparent resolution of the Nagasaki discrepancy and the intractability of the Hiroshima discrepancy, DNA began systematically investigating various disassembly and detonation hypotheses that had been advanced to account for the measurements and made a first attempt to derive a source from the measurement data themselves. In Hiroshima, the basic problem is that DS86 does not calculate the level of thermal-neutron activation at distance as is present in materials activated by bomb neutrons. It has been known for some time that postulating a source

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) FIGURE 4–2 Nagasaki thermal-neutron activation (1993) revised calculation to measurement ratio as a function of slant range from the hypocenter placed beside Hiroshima thermal-neutron activation (1993) revised calculation to measurement ratio as a function of slant range from the epicenter.

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) for the bomb that contains more neutrons above 1 MeV in energy could bring about better agreement of the thermal-neutron calculation with measured values. Such a source would produce the greater thermal-neutron relaxation lengths necessary to calculate the activation levels measured at distances of 1 km or more. Because the standard nuclear-weapons codes and conventions do not produce the hard spectrum suggested by the measurements, many hypotheses have been advanced as to the mechanisms by which sufficient high-energy neutrons could have been produced to bring about the measured thermal-neutron activation. Most of the suggested mechanisms of high-energy neutron release involve abnormal disassembly of the Hiroshima bomb. Such speculation has been fostered by several circumstances peculiar to this bomb. First, as mentioned above, the bomb dropped on Hiroshima is the only one of its type ever detonated. A number of Nagasaki-type bombs have been tested, but there is no bomb comparable with the Hiroshima bomb from which measurements can be derived. Second, the unique design and structure of the Hiroshima bomb has led to hypotheses about structural failure of the bomb case during detonation that would have permitted fast neutrons to escape unmoderated by the thick iron case that surrounded the core. The fact that the bomb case was a steel gun barrel has led to the suggestion that either the bomb case cracked or the tail of the bomb blew off, creating a streaming path for high-energy neutrons to escape. ALTERNATIVE DISASSEMBLY HYPOTHESIS Cracks had been reported in some gun-tube assemblies during non-nuclear test firing prior to the building of the Hiroshima bomb (Rhodes 1995). This observation led both Auxier (1991, 1999) and Hoshi and others (1999) to suggest that the Hiroshima bomb could have split from the shock of the high explosive before the bomb reached peak nuclear power providing a portal for the release of fission-spectrum neutrons. They suggested that adding such high-energy neutrons might result in a bomb output spectrum that would better match the in situ neutron-activation measurements. Despite the fact that the case for the Hiroshima bomb was test-fired and did not crack prior to being loaded with the nuclear material (Rhodes 1995), the committee considered the possibility of such an unexpected disassembly and concluded that it was both extremely unlikely and incapable of matching both the fast and thermal neutron data had it occurred. Given that those alternative disassembly hypotheses could never be tested and that no direct data support or refute nonstandard modes of disassembly of the bomb, the DNA program sought to explore them to test their feasibility. The first step in the process was to review the bomb hydrodynamics and time course with the design group at LANL. The review produced several observations, all of which are incompatible with abnormal disassembly of the bomb. First, major compromise of the case seems improbable because it would have substantially reduced the observed yield of the bomb (Whalen 1994). Second, the course of development of

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) cracks in the case is measured in milliseconds, whereas the neutrons would have escaped from the bomb within microseconds. So the prompt neutrons would have been released before the explosion began to distort the bomb case. Third, small cracks in the case would have to be oriented in exactly the straight-line paths that neutrons take between collisions for the neutrons to escape unmoderated, as Whalen (1994) has pointed out, this is extremely improbable. Fourth, for neutrons arising late in criticality, small to moderate cracks are annealed by the heating of the bomb. These arguments represent the best judgment of the persons responsible for bomb design and evaluation, but in the end it is just the best judgment of experts. To test directly whether adding fission-spectrum neutrons, whatever their origin to the Hiroshima source term, would improve the agreement with measurements, a composite source study was conducted. The composite-source study was conducted during the summer of 1993 by SAIC (Kaul and others 1994). To test the hypothesis that calculated activation would match the activations measured in Hiroshima better if the source term for the bomb contained more high-energy neutrons, (such as would be the case if the bomb case had cracked prior to nuclear detonation) fission-spectrum neutrons were systematically added to the LANL-calculated Hiroshima output source (similar to the spectrum for inelastic scattering in iron) and compared to the in situ measurements of activation. This study was constrained to determine whether any such combination source could be identified that would agree with the high-energy (sulfur), gamma-ray (TLD), and thermal-neutron activation data. Height of burst for the bomb was also varied to evaluate the influence of height on agreement. In all cases, the addition of fission-spectrum neutrons to the Hiroshima source made the agreement between calculated and measured sulfur deteriorate more rapidly than the agreement between calculated and measured thermal-neutron activation improved (Whalen 1994). This process is strongly governed by the results from the sulfur and thermal-neutron activation and is not strongly influenced by the TLD data, which are relatively insensitive to the assumed spectrum. This can be seen in every analysis that has been undertaken since the inception of DS86, in which there has always been good agreement with TLD data, even in the face of large discrepancies with other measurements. The result from this analysis is that no combination of composite source and height-of-burst could be identified that could simultaneously reconcile the best available 1993 calculation with the TLD, sulfur activation, and thermal-neutron activation data (Kaul and others 1994). The overall implication of all the work done on alternative disassembly hypotheses is that they do not reconcile the Hiroshima calculations and measurements. The implication is that either the constraining measurements or the leakage for the Hiroshima bomb must be altered. The constraining measurements were the sulfur measurement of fast-neutron activation and the entire set of thermal-neutron activation measurements. The implication of the study for the leakage spectrum of the Hiroshima bomb is that the sulfur and thermal-neutron data can be matched only by tailoring an output spectrum in which there are more neutrons of 2–3 MeV

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) and fewer of below 2 MeV and above 3 MeV, because adding neutrons below 2 MeV would cause the overestimation of the measurements at short distances from the hypocenter and additional neutrons above 3 MeV would disrupt agreement with the sulfur measurements. The next phase of this investigation was to see whether such a source could be identified. The effort to define a source term that would cause the calculations to match all the in situ measurements in Hiroshima was undertaken by the Mathematical Physics Division of ORNL in 1993 (Rhodes and others 1994). The first calculation essentially repeated composite-source calculations with the actual leakage of the APRF reactor, which was placed in a 30-degree horizontal band around the midplane of the bomb. Using the APRF at about one-third of the total neutron leakage calculated for the weapons produced an excellent fit—to within a few percent of thermal-neutron activation data. That solution failed for two reasons: it produced a discrepancy of a factor of 10 with the sulfur data under the bomb, and it required the elimination of most of the conventional leakage (which is a softer neutron spectrum) calculated for the bomb to match the short-range measurements. After several other such failed attempts, it was decided to concentrate on leakage in the energy range above the “oxygen window”, which is a large valley in the oxygen cross section at about 2.3 MeV. That was done in two ways. First, 8% of the total neutron leakage of the Hiroshima bomb was concentrated at 2.1–2.7 MeV; this “boost” to the neutron flux was uniformly distributed in a 30-deg band around the horizontal midplane of the weapon. The remainder of the leakage spectrum was adjusted to maintain the number of neutrons, and neutrons above 2.7 MeV were eliminated to match the sulfur data. This output leakage spectrum fit all the data to within 20%. In some of the reports of this work, the leakage spectrum has been referred to as the “pancake source.” One other such source was constructed. It has been referred to as the “funnel-cake source.” In this configuration, 31% of the bomb neutrons were concentrated in an upward-directed 30-deg cone, and some of the neutron output above 2.7 MeV was retained. This configuration produced a good fit with all the in situ measurements; the largest error was 22%. These source spectra were considered informative but could not be adopted because no explanation for such a neutron release could be posited. The exercise causes one to look carefully at the accuracy of the in situ measurements because large errors in the sparse fast-neutron activation data could permit the boost in leakage to be distributed over a wider energy range above 2.7 MeV. It is interesting to note that either of the “boosted oxygen-window spectra” would have only a very small effect on the gamma doses while placing the neutron dose much closer to the estimates of T65D. On the basis of a preliminary spectral unfolding calculation, Pace (1993) suggested that the output spectrum to match the data would require a strong peak above the oxygen window with source reduction both above and below the window. In the adjoint calculations associated with all those calculational exercises, Whalen (1994) sees hope that the Hiroshima neutron-activation measurements can be matched with a source spectrum that is softer, not harder. That hope is based on

OCR for page 60
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) data that show that calculations and measurements of neutron transmission through thick iron (as in the case of the Hiroshima bomb) should be softer than is currently calculated. To follow that lead, the final experiment funded by the DNA program on Hiroshima neutrons was the measurement of neutron transmission through a cross section of the bomb case and tamper. Time-of-flight measurements were made with the LAMPF 800-MeV proton linear acceleration at LANL as a “white” neutron source to measure the transmission of neutrons 0.6–600 MeV. STATUS OF EFFORTS TO IMPROVE DS86 Following the recommendations of this committee (see Chapter 8), a number of projects have been funded and are in progress in an attempt to refine the calculations of DS86. These include a total recalculation of the output of the Hiroshima bomb, the total reevaluation of shielding models for Hiroshima and the factory workers in Nagasaki, and an evaluation of Sn and Monte Carlo calculations as a method for an adjoint determination of a source for the Hiroshima bomb that will agree with the measurements. The recalculation of the Hiroshima bomb will be the most comprehensive ever accomplished. It will include a late-time output spectrum that incorporates new iron cross sections and transmission through the bomb case, a new Monte Carlo source term as a function of energy and angle, and a Monte Carlo transport of the new source to the ground accounting for the tilt and heading of the bomb, and the transport of delayed neutrons over time through air created in a new spherical air blast calculation. The new calculations, which will use the latest ENDF/B-6 cross sections, will be compared with source and DORT calculations. If it is necessary for a satisfactory level of agreement, an adjoint-to-source and forward to free-in-air kerma Monte Carlo calculation will be performed. The next kind of work recommended by this committee is an examination of the adequacy of shielding models in DS86. Two efforts to follow this recommendation are under way. The first is an examination of the nine-parameter, globe, and terrain shielding in Hiroshima. This effort will identify required changes in how shielding is handled in DS86. The second effort is the improved modeling of the shielding environment for the factory workers in Nagasaki. The biological dosimetry for these workers indicates that the shielding could be in error. With better accounting for the shielding provided by these structures and the heavy machinery they contained, the RERF dosimetry system can determine the dose for these workers better. The work to improve the radiation-transport calculations should be completed as quickly as possible. Any improvements derived from the work and other improvements in radiation cross sections and transport methodology achieved since DS86 should be fully implemented in the RERF dosimetry system when they have been reviewed by US and Japanese senior review panels, fully documented, and approved.