5
Tensor Technology Report on the Multidimensional Neutron Radiometer Airline Security System
A specific item in the Statement of Task (Box 1-2) for this panel was to review the Tensor Technology report, Advanced Studies on the Multi-Dimensional Neutron Radiometer (MDNR) Airline Security System, (Tensor Technology, 1998a). This chapter reviews the details of the proposed MDNR. The basic MDNR design is a reasonable baseline conceptual design for assessing the airport implementation and acceptability of the PFNTS system.
Technical Capabilities and Physical Attributes
The panel's critique is based on the MDNR design presented in the Tensor report supplemented by discussions with representatives of Tensor. The MDNR conceptual design is shown in Figure 5-1. Tensor has proposed a very innovative approach to the PFNTS neutron source, using a cyclotron rather than a linear accelerator to reduce the footprint of the explosives-detection device. Although this approach has some advantages, it also raises some questions because all of the previous tests on the detection capabilities of PFNTS were conducted on equipment with a linear accelerator as the neutron source. Furthermore, the cyclotron-based design still has weight and space requirements that may preclude integration into the passenger baggage line at existing airports. Although cyclotrons are used in many hospitals for the production of short-lived radioisotopes, the environmental controls and support utilities available at a hospital are different from those of a typical airport baggage makeup room.1
The following discussion covers important aspects of the MDNR technical capabilities. Many of the performance characteristics are summarized in Table 6-1, which compares the performance and operational attributes of MDNR and the InVision CTX-5000 SP.
Performance Levels (Pd and Pfa)
The Pd and Pfa for the cyclotron-based MDNR are not well established. Because of the lack of testing at cyclotron-based neutron sources, this review is based on performance estimates for linear accelerator-based tests performed at the University of Kentucky and the University of Oregon. In the university tests and in the proposed MDNR design, the 9Be(d,n)10B reaction was used for neutron production. One important potential difference between the cyclotron and the linear accelerator involves the energy and spatial stability of the deuteron beam. The cyclotron deuteron beam is stable to within about ± 200 keV, or about one revolution in the cyclotron. The linear accelerators have a beam homogeneity of about < 5 keV while operating at a nominal deuteron energy of 4.2 MeV. The larger spread in the deuteron energies is not expected to be a problem, but temporal variations in the deuteron energy would probably degrade performance. Any variation in the flux of deuterons delivered to the target (the beam current) must be compensated for by other beam diagnostics. Because beam current can usually be easily measured when the beam is not impinging on the beryllium target, the long-term current drift can be compensated for; nevertheless, short-term variations are a matter for concern.
The Ebco Technologies2 cyclotron (proposed for use by Tensor in the MDNR) has a reported current stability of < 1 percent; the University of Oregon linear accelerator has a reported voltage stability of < 0.1 percent. These data, which are included in the facility documentation, are not for equivalent parameters and do not address the temporal dependence of the parameter variations. The voltage is related to the energy of the deuteron and hence affects the neutron energy
spectrum. The current is related to the number of deuterons per second impinging on the target and hence the number of neutrons impinging on the scanned bag. The panel believes the MDNR detection capability should be based on tests conducted with cyclotron neutron sources or on assurances that the short-term (less than a bag scan time, nominally ~7 seconds to support an EDS throughput of 500 bags per hour) cyclotron voltage and current variations are less than those demonstrated in laboratory tests with linear accelerators.
Based on the linear-accelerator blind tests by Tensor at the University of Kentucky (see Chapter 3) to determine the MDNR detection performance, the Pd is ~95 percent for Class B explosives, ~50 percent for Class A explosives, and 0 percent for some configurations of Class A explosives. Tests at the University of Oregon yielded similar results. However, to be valid the Pd of a system must be accompanied by the Pfa. The University of Kentucky tests had a Pfa of ~25 percent. The University of Oregon tests demonstrated a Pfa of ~12 percent using the original detection algorithm and a Pfa of ~5 percent using post-processing analysis. The differences in the Pfa at the two laboratories raise some concerns. Differences in Pfa may be related to the use of an open two-dimensional detector at the University of Kentucky rather than the linear array used at the University of Oregon, to differences in the explosives-detection algorithms, or to differences in the test protocols.
In any event, these laboratory test results indicate that the MDNR would probably not meet the EDS detection performance requirements for Class A explosives.3 Results indicate that the MDNR would meet the EDS detection requirements for Class B explosives and, indeed, for all classes of explosives in the FAA EDS specifications except Class A explosives. Based only on the current demonstrated performance of PFNTS for all explosive classes PFNTS is not very
impressive and would not merit further development. Many explosives-detection systems have come close to the certification-level of detection for Class B explosives but have failed for Class A explosives and often one or two subclasses of Class B explosives. Indeed, most other explosives-detection equipment did not meet the certification-level detection requirements for Class A explosives. PFNTS has little potential for being combined with another detection technology to fill this performance gap. The only possibility for a complementary technology may be nuclear quadruple resonance, which has demonstrated very good performance for Class A-1 explosives but has not demonstrated adequate detection of Class A-2 explosives in large bag configurations.
The panel believes that the post-processed analysis from the University of Oregon tests should be validated better. If the post-processing analysis is representative of future test results (and not unduly tuned to the specific bag set) and if the Pd and Pfa results can be shown to apply to area detectors as well as linear detector arrays, then the PFNTS system performance (Pd > 95 percent and Pfa < 5 percent) may exceed the present and expected detection performance of other EDSs for Class B explosives.
Cost
Tensor has provided details to support a cost of $2.4 million for the initial MDNR unit after the expenditure of an additional $1.8 million for engineering, assembly, and testing of the initial unit. The final units are expected to cost ~ $1.5 million if purchased in quantities of 11 or more. The installation cost is estimated to be an additional $323,000 for the vault enclosure and utility modifications and $175,000 for modifications to the baggage lines. These cost estimates were reasonably substantiated by Tensor for a conceptual design. However, the panel believes that the estimated installation cost is too low, particularly the construction costs for a shielded enclosure and modifications to the baggage line.
The panel's conclusion that the estimated installation cost is unreasonably low is based on the $1 million to $3 million cost for installation alone (i.e., without counting the purchase price) of a CTX 5000 SP fully integrated into the baggage line at some airports (the fully integrated CTX 5000 SP also requires extensive modifications to the baggage line to meet operational requirements). The average installation cost to place approximately one-third of the planned 54 CTX-5000 SP systems in airports as stand-alone units was $20,000 to $150,000 per system. These figures reflect the variability of system integration costs depending on airport design and operational constraints. The MDNR conceptual design and cost analysis are for the simplest airport baggage-line configuration. Of the four airports considered in the MDNR conceptual design study, a baggage-line integration of the MDNR system was feasible in only two. The other two would require that separate facilities be constructed to house the MDNR. Tensor's estimate of installation cost, therefore, should be considered a minimum.
When the CTX-5000 SP systems were installed, the airlines frequently (84 percent of the time) had them installed at the check-in point rather than in the baggage makeup room because it is much easier to resolve alarms when passengers are nearby. Check-in point integration would not be possible for the MDNR system. In fact, MDNR systems would have to be installed on the ground level because of their weight.
Accelerator
One of the notable characteristics of the MDNR conceptual design is the use of a cyclotron, an accelerator with a circular ion path, rather than a linear accelerator, which was used in all previous PFNTS testing and conceptual designs. From the standpoint of integration into an airport facility baggage line, using a cyclotron is a useful selection. However, reduction in the size of the spatial footprint for the system also increases the weight of the MDNR because of the cyclotron's heavy magnets. Even though Ebco Technology's TR9D cyclotron (selected for use in the MDNR) weighs 20 tonnes (22 tons), one could argue that this is not a severe penalty compared to the 109-tonne (120-ton) shielding enclosure or the 528-tonne (581-ton) vault enclosure required for accelerator-based explosives-detection techniques.
A significant advantage of the MDNR accelerator is that it uses commercial off-the-shelf technology. Rather than a unique design, the TR9D cyclotron is basically the same as the commercial cyclotron4 used for the production of radioisotopes. Thus, the use of a commercial off-the-shelf cyclotron would add to the manufacturability of the MDNR, as well as quality control to ensure consistent performance. A potential disadvantage of the cyclotron is that the rotating, accelerating charged particles produce much more gamma radiation than linearly accelerated particles with the same energy and charge. The MDNR overcomes this disadvantage by placing an 80-cm (32-in.)-thick borated concrete shield around the compact cyclotron, in addition to the concrete vault enclosure used to shield both linear and cyclotron accelerators.
Comparing the radiation from different types of accelerators is very complicated. With a linear accelerator, proposals for minimizing the length of the system often involve a two-story configuration, with the accelerator on one floor and the neutron flight path on another. The flooring material between the levels can be used as shielding to protect the detector from radiation. However, bending magnets are
often used (with linear accelerators in this configuration) on the detector floor to focus the deuteron beam on the target. These magnets are the source of intense radiation (as charged particles accelerate around the curved path), which could interfere with the system detection unless shielding is used. But, like the cyclotron magnets, the bending magnets used with a linear accelerator are a localized radiation source that can be shielded efficiently with a small volume of materials.
The potential advantages of a cyclotron over a linear accelerator-size and compact neutron source for easier shielding-may not carry over into an MDNR if it is housed in a separate facility rather than integrated into the baggage line. If a separate facility is required, size may not be a crucial factor, and the design trade-offs with linear accelerator-based systems would have to be reconsidered. The temporal variation of the beam voltage and current of a cyclotron will require more analysis and perhaps validation testing.
Radiation Shielding
The neutron shielding on the outside of the vault enclosure in the MDNR design is intended to provide a work environment for a ''nonradiation worker," that is, a radiation level of less than 2 mrem/hr with a total allowable yearly dose of less than 100 mrem. This level of shielding would simplify and reduce the cost of the radiation safety for general workers in the bag make-up area (outside of the vault enclosure) by eliminating the need for a large thermoluminescent dosimeter (TLD) program. The baseline MDNR radiation shielding consists of a 109-tonne (120-ton) shield around the cyclotron itself and a 528-tonne (581-ton) vault enclosure.
The panel was concerned about the possibility of neutron radiation streaming near the entrance to the baggage line maze in the proposed MDNR design (see Figure 5-2). According to Tensor's radiation transport calculations, the neutron radiation levels would be acceptable. However, if an 8,760-hour year is assumed, the secondary gamma radiation levels in the initial design would exceed 100 mrem per year.5 Assuming that workers would not be in the area for more than 2,080 hours per year, Tensor concluded that the shielding was adequate. Tensor also identified some nominal design changes (e.g., a 50-cm [20-in.] borated polyethylene beam catcher) that could reduce exposure by another factor of four. The fidelity of these calculations was consistent with a conceptual design, but tensor correctly notes that a higher fidelity "engineering design study" should be done in the next phase of development to refine the estimate and uncertainty analysis of the radiation environment along the periphery of the vault enclosure.
In its calculations of radiation shielding, Tensor assumed a neutron source of 7.11 x 109 neutrons per second at the target location. Using the MDNR baseline design with a 50 mamp current, a 2 ns pulse width, and a 1.2-MHz pulse repetition rate, the total target neutron 4π production would be
5.65 x 1012 neutrons per second. This neutron level is higher by a factor of 795 than the one used in the Tensor analysis, which raises concerns about the applicability of Tensor's shielding calculations. The difference in neutron source strength reflects that Tensor's calculations were based only on the neutrons transmitted along the collimated beam and subtended by the bag. Although not obvious in the initial proposal, Tensor used a 93-cm (37-in.)-long lithium-loaded polyethylene collimator between the beryllium target (the neutron source) and the bag. Furthermore, the baseline MDNR design avoids a beam transport line and located the beryllium target at an extracted beam focal point one foot outside the cyclotron itself but within a beam portal in the 80-cm (32-in.) borated concrete shroud. Finally, the outer portion of the cyclotron is filled with borated material, so it too can function as neutron shielding. The presence of all of this neutron shielding around the target lends credence to Tensor's estimate that only the neutrons along the collimated angle subtended by the bag would scatter into the room. However, refined radiation transport calculations will be necessary to validate this estimate.
The panel concluded that the Tensor shielding is a reasonable baseline configuration for a conceptual design. Although much more detailed radiation transport calculations will be required to support an engineering-level design, the MDNR shielding mass of 637 tonnes (109 + 528) (701 tons [120 + 581]) is expected to be reasonably close to what would be determined for an engineering-level design.
Size
The MDNR system will require an 8 x 13 m (26 x 42 ft) footprint, or an area of 104 m2 (1,092 ft2). This footprint includes the vault area but does not include additional support equipment or the cyclotron operations area.
Controlling the temperature and humidity of the cyclotron would require power and water for cooling. In response to questions from the panel, Tensor described five 48-cm (19-in.)-wide rack chassis cabinets that would provide this support. These racks are similar to the support equipment shown in Figure 5-3, a photograph of an Ebco TR19 cyclotron currently installed at a commercial site in Seoul, Korea.
The support equipment adds only about 1.2 m2 (13 ft2) of floor area, much of which could be inside the vault enclosure.
The cyclotron also requires some computer equipment and a cyclotron operations console, which must be located outside the neutron radiation environment of the vault enclosure. The programmable logic controllers should be located no more than 20 m (66 ft) from the other equipment. 6 According to the Ebco literature, the recommended equipment area (support equipment and operations console) requires 9.7 m2 (108 ft 2), a slight increase over the planned MDNR footprint (Ebco Technologies, 1998). The computer equipment may, however, require some temperature and humidity controls beyond the ones normally provided in the baggage makeup room at some airports. This requirement along with the need to protect the equipment from physical damage would increase the required area.
The detector array and most of the support equipment for explosives detection could be located in the vault enclosure. However, this assumes that the explosives-detection algorithm is fully automated and precludes any operator intervention or operator-assisted alarm resolution procedures. Although the blind test results did not show significant added value from operator-assisted explosives detection, an MDNR system in an airport should have the flexibility to include operator-assisted alarm resolution. The area to support operator alarm resolution could probably overlap with the area for the cyclotron operations control.
Weight
The panel concluded that the MDNR design provides a credible estimate of the system mass. The magnets for the cyclotron account for most of the weight of the 20-tonne (22-ton) accelerator. The radiation shielding from the cyclotron enclosure and the exposure vault area contribute another 637 tonnes (701 tons). The weight of the other system support equipment is negligible by comparison.
Operational Capabilities
In addition to the technical capabilities and physical attributes of the MDNR, some operational capabilities will affect the practicality of its implementation into an airport environment.
Commercial off-the-Shelf Equipment
The use of commercial off-the-shelf equipment in the MDNR should ensure quality control, product reliability, and maintenance, as well as minimize costs. The cyclotron and associated control systems are based on commercially available systems that should continue to be available and maintainable. The MDNR neutron detector array, a customized item, is also based on readily available commercial parts.
Radiation Safety
Radiation safety is a critical element of a PFNTS system. For safety reasons, personnel who use the system, and possibly everyone with access to the system, would require radiation safety training. Safety interlocks for access to the vault enclosure area are a part of the MDNR design. TLDs, which would be required for anyone entering the vault area, would probably not be expensive7 but would require administrative support. Procedures for access to the radiologically controlled areas would have to be developed and posted, and a radiation safety officer would have to sweep the vault area every time it was secured to ensure that no personnel were in the area when the cyclotron was turned on. Gaseous effluent from the vault area would probably have to be sampled and monitored for radiation.
The activation of materials in passenger bags is not expected to be an issue under routine operating conditions. However, extrapolating from radiation controls used in thermal neutron analysis systems, a radiation check of each passenger bag as it exits the vault area might be required as a feature of early systems (Jones, 1990).8
The baggage entrance area to the vault enclosure poses special problems from a safety standpoint because an individual entering the vault area could be exposed to unacceptable levels of radiation. Passive marking of radiologically controlled areas is permitted only if all personnel have some level of radiation safety training. More stringent access controls, such as locks or monitored access, are typical of restricted areas adjacent to public areas. The bag access area for the MDNR could be controlled in several ways, but this issue would have to be addressed in an engineering-level design.
Deactivation Procedures
The disposal of activated materials from the cyclotron shield and from the vault shielding is another issue that will have to be addressed. Even if the activation levels are low enough that the shielding materials do not require disposal as activated material, material surveys or clear bounding radiation transport calculations will be necessary to deter-
mine the activation level. An audit trail will have to be maintained for building materials to ensure that no materials9 regulated by the Resource Conservation and Recovery Act (RCRA) are used. The cost of disposing of radioactive material, however, would not be large compared to the purchase price of the equipment. Furthermore, outside companies could be contracted to handle the disposal of radioactive material, which would relieve airport personnel of some of the administrative burden of complying with disposal regulations.
Baggage Flow
The baggage flow through the MDNR is not given sufficient attention in the MDNR conceptual design. This issue should be addressed in much greater detail in an engineering-level design. The sharp turns in the MDNR bag flow were introduced to reduce the neutron radiation levels at the baggage entrance to the vaulted area. However, these sharp turns also increase the potential for baggage jams.
Baggage jams in the MDNR would require that the cyclotron be turned off, a qualified operator or safety officer conduct a radiation survey of the vault area, an operator enter the vault area and clear the jam, the vault area be checked for the presence of other individuals, and the safety interlocks be reengaged. This baggage-clearing procedure would take much more time than clearing a normal jam in another part
Cargo Inspection
The Tensor report discusses the potential of PFNTS for use in scanning cargo but does not provide a conceptual design consistent with cargo-scanning requirements. The panel concluded that PFNTS technology does not have significant potential for the inspection of thick containers of hydrogenous10 materials for explosive amounts consistent with the FAA's EDS detection requirements because neutron attenuation is too great. For thick neutron-attenuating containers, the detection capability of PFNTS would be significantly impaired. The capability of PFNTS to detect explosives concealed in thick containers has not been experimentally verified, so Tensor's predictions are based on theoretical analyses. A cost-effective experimental verification on thick containers could lead to new detection algorithms for use in highly attenuating scenarios.
The Pd for explosives concealed in containers with dimensions not much larger than passenger bags would probably be similar to the Pd for explosives concealed in passenger bags. Once the potential of MDNR for scanning passenger bags has been refined to encompass Class A explosives and a low Pfa has been validated, data should be collected on scanning containerized cargo.