Physics, P-DO
|
Design Feasibility and Cost Estimate for a Single-Axis, Multipulse Proton Radiography Facility
J.A. McGill, C.L. Morris (P-25), P.L. Colestock (ISR-6), F. Neri, L.J. Rybarcyk (LANSCE-1), M.E. Schulze (DX-6), H.A. Thiessen (X-4), N.S.P. King (P-23) IntroductionA recent study motivated by the success of proton radiography (pRad) experiments done at Brookhaven National Laboratory (BNL) has examined the design parameters and estimated the costs of proton synchrotrons at both 10 and 20 GeV, which can be applied to quantitative radiography for the weapons stockpile. Potential sites and capabilities at LANL or the Nevada Test Site (NTS) were studied, and design feasibility and cost estimates were completed. The goal of the study was to present options for a new hydrotest capability that provides better position resolution, a factor of 10–100 higher effective dose1, and up to ten time frames. The choice of the two energies is driven by the classified results of static experiments performed using 24 GeV/c and 7.5 GeV/c beams from the Alternating Gradient Synchrotron (AGS) at BNL. These experiments showed that the lower energy is sufficient for measuring weapons-physics processes in scaled experiments, which are important to certification2; and the higher energy is suitable for full-scale hydrotesting on the largest stockpiled systems, with significantly enhanced physics returns3. The Potential SitesThe prospective LANL site was selected to take advantage of the existing accelerator infrastructure at the Los Alamos Neutron Science Center (LANSCE). The 800 MeV LANSCE linear accelerator could be used as an injector, saving the time and money that would be needed to build and commission a new accelerator. In addition, the existing infrastructure of trained people and equipment would simplify commissioning a new accelerator. However, the current site-wide authorization basis would preclude using pRad to diagnose dynamic experiments with plutonium, a key requirement for certification. The potential NTS site studied was selected to take advantage of the U1a firing site and its authorization basis for plutonium experiments. Although the underground construction required at U1a would be more expensive than the site at LANL, and an injector would have to be designed and constructed, once the machine was completed, it would be relatively straightforward to implement an authorization basis for experiments with plutonium.
The number of pulses is driven by the need to measure density at several times normal to infer criticality. Extensive studies by Buescher, Hopson, and Slattery have shown that four pulses spaced at a minimum of 200 ns are sufficient. We have added a fifth time to the design requirements so that early time phenomena can be studied simultaneously and critically. Up to 10 pulses can be used for the 20 GeV ring studied here, limited by the circumference of the synchrotron. The 10 GeV ring can provide up to 6 pulses at 200 ns spacing. The proton dose in Table 1 is twice what has been used in validation experiments described below. This is enough beam to allow a two-Gaussian imaging mode, in which part of the beam is used to image small radii in the object and another part of the beam is used for full-field imaging. pRadTransmission radiography measures the transmission of a penetrating beam through an object and uses the attenuation to measure the areal density profile (thickness) of the object. Typically, the information is used to qualitatively determine the internal structure of the object. Hydrotesting involves imploding a primary from a weapon, in which the nuclear fuel has been replaced with a surrogate in order to not produce a nuclear yield. One goal of hydrotest radiography is to measure densities at late times in the implosion to benchmark numerical simulations that can be used to predict the explosive yield of nuclear systems. Until recently, the only diagnostic available for late-time hydrotesting was flash x-radiography. With the cessation of underground testing, considerable effort has been expended to improve x-ray technology. The most recent manifestations of this effort are the first axis of the Dual-Axis Radiographic Hydrodynamic Test (DARHT) facility and the ARIX facility.5 These machines produce the largest doses and smallest spot sizes ever achieved with flash x?]radiography. Nevertheless, in order to answer questions about stockpile performance and long-term stockpile certification, researchers require data that flash x-radiography cannot provide.4 Transmission radiography is described by the Beer-Lambert law.6 The solution to the differential equation,
where λ is the mean-free path of the probe, z is the distance traversed, N0 is the incident number of particles, and N is the number surviving after a distance (z). For objects of composite materials, z/λ becomes a summation or an integral, but the analysis below remains the same. In quantitative radiography the thickness of an object can be measured as:
where l is the total amount of material traversed. Experimentally, one must consider details that are not reflected in the above equations such as the source spectrum (because λ is a function of energy), and backgrounds, which always seem to be a problem.
The mean-free path, λ, can be chosen to minimize Dl for a given object. Setting An alternative is provided by hadronic rather than electromagnetic probes. The absorption cross section, σA, for hadrons on a nucleus with atomic number A is well-approximated by:
where rA ≈ 1.2A1/3 fm. Using Advantages of pRadIt is possible to take advantage of the long hadronic mean-free path by using high-energy protons as a radiographic probe. High intensities are available in short pulses using current accelerator technology. Because the protons are charged, a proton beam can be distributed across a radiographic object and focused onto downstream detectors using electromagnets.8 In a set of experiments performed using the AGS at BNL, data were taken on a set of test objects to validate pRad and to compare it to x-radiography from DARHT. In all cases that have been studied, the qualities expected from pRad have been verified, and in all comparisons, even modest pulses of 1010 protons have been observed to provide a radiographic advantage over DARHT performance by a factor of between 10 and 100 in units of dose. Even more significantly, pRad has been demonstrated to be quantitative at the percent level, and in high-fidelity mockups of hydrotest experiments, pRad demonstrates the capability to measure some of the underlying physics of cavity shape and mix, topics that have been identified as requirement drivers for hydrotest radiography. The costs are based on a model developed in the Advanced Hydrotest Facility (AHF) project, and are expected to be accurate to ± 20%. It must be emphasized that this study is intended only to show the outlines of a facility that would provide the high-quality radiography discussed above and a rough cost estimate for such a facility. A more definitive design and estimate will be produced during conceptual design. Scope and Cost
A 500 MeV booster synchrotron injects the larger machine (either 10 or 20 GeV) for the NTS siting option. The booster is injected by a small commercially available 11 MeV H(-) linac. As with the larger machines, the dynamics of the booster have been studied to such detail that its final design can be bounded, and to provide a reasonable estimate of its cost.
|
The requirements for these machines were synthesized from a combination of the results from the AGS experiments and from requirements studies carried out over the last decade.4 They are listed in Table 1.
gives an estimate of the hadronic mean-free path in uranium of 220 gm/cm2, or 11.6 cm. The tabulated value is 199 gm/cm2.7 Measurement of thickness is optimized at around 23 cm of uranium. This is much better matched to hydrotest radiography than the x-ray mean-free path. Consequently, the same statistical information can be obtained with a much lower incident flux of hadrons than with high-energy x-rays.
The goals of the accelerator complex are to satisfy the parameters in Table 1. The most demanding machine is the 20 GeV main ring synchrotron, common to both the LANL and the NTS sites. The alternative, a 10 GeV ring, has been studied only to the level of detail necessary to show its proof of principle. Capture, ramp, and acceleration have been studied for the 20 GeV ring, assuming injection could be provided by LANSCE’s 800 MeV linac, or from the 500 MeV “green-field” injection complex at NTS. Detailed studies of the dynamics in the 10 GeV machine have only been done assuming injection from the LANSCE linac.
The overall layout of the ring is a racetrack design as shown in Figure 1. The pair of long straight sections provide for injection, extraction, and acceleration. Parameters of the Main Ring are in Table 2. The 10 GeV design is conceptually similar to the 20 GeV design, and results in an overall smaller circumference (Figure 2). The reduction in size is not precisely a factor of two from the 20 GeV case owing to the required straight sections and other infrastructure for injection and extraction. The ring’s parameters are given in Table 3. The accelerators proposed here are well within the parameters of other accelerator projects that have been proposed and/or built. The reported (or estimated) costs of these accelerators vary considerably from site to site; nevertheless a reasonable estimate can be made, accurate to probably 20%. The costs of several synchrotron projects were studied in the AHF project. Those costs were summarized and reported by Prichard9 and incorporated into a model for costing synchrotrons. The model uses gross measures, such as total weight of dipole and quadrupole magnets, total amount of direct-current power, total length of beampipe vacuum, etc. Cost rates are applied to generate a total equipment cost. Factors are applied for overheads, design costs, and contingency, to arrive at the total project cost (TPC) for the accelerator equipment. A similar breakdown is done for the balance of plant, and cost loading factors specific to the civil construction process is applied. The result of the two exercises is an estimated TPC.
