Isaac Maya, Erroll Southers, Ehud Keinan
October 2009 to September 2010
Peroxide based explosives, including TATP, DADP and HMTD, represent a major, growing challenge to homeland security and an emerging risk to the airline industry because of the limited capability of detection. The overall goals of this project are to characterize these peroxide explosives, and to examine deployment strategies for those technologies using risk assessment techniques. This project examined the peroxide explosive threat by: 1.Preparing a broad variety of plastic TATP explosives to develop their characterization, including identifying and characterizing various polymorphic crystals of TATP. 2.Using formal risk assessment methodologies to analyze the comparative costs and benefits of deploying peroxide-based explosive detection technologies in major transportation infrastructure screening locations. The first section of this report focuses on the synthesis aspects and some physico-chemical properties of peroxide based formulated explosives. We prepared various plastic compositions of TATP using conventional binders and various plasticizers. In addition we prepared improvised plastic compositions following the methods and binders used by terrorist, including various greases, petroleum jelly, motor oil and food oils. The experiments are described, and significant aspects of the properties of TATP-class of materials are presented. We used various techniques and methods of high-resolution microscopy to determine the physical appearance of each composition. The resultant information is contained in the body of the report, and would be useful for identifying samples of plastic peroxide explosives captured in terror events. Furthermore, the morphology could be correlated with other properties of these materials, including their sensitivity and detonation velocity. In the second section, we examine a decision tree model for evaluating deployment of peroxide-based explosives detectors. Detection technology may be used at various security layers in and around an airport. Vehicle checkpoints, checked bag and carry-on bag inspections are identified as the potential points of interdiction. We identified 37 critical parameters that collectively determine the level of terrorism risk. Using the decision tree model, we identify the most critical parameters based on an objective of minimizing overall expected economic consequences (EEC). EEC includes the direct and indirect costs of an attack, cost of technology as well as the dollar equivalent of human casualties. Parametric estimates based on open source information and five year time horizon were used to arrive at the conclusion that deployment of plastic explosives detection at all security layers is optimal. The optimal full deployment strategy could result in up to $3.5 billion in reduced expected economic consequences over five years over the current inspections scheme. Even more striking is the observation that the lack of precision in our baseline estimates does not reduce the significance of this result. The optimal decision is robust to the changes in parameter values around their baseline. The optimal decision under the baseline assumptions is to deploy technology at all security layers. The probability of an attempt and the parameters that define the deterrence level, the interdiction capability and the detection probability have influence on the optimal decision in their respective ranges. Cost and economic consequences are minimally important in determining the optimal decision despite their impact on the objective function. Based on the selected assumptions and parameters, the model recommends deployment of technology to reduce the risk from peroxide-based explosives. Parameters related to the cost of new technology are influential on the overall expected economic consequences as well. However, perhaps surprisingly, the optimal decision is insensitive to the cost of the technology. The cost of implementing the optimal alternative is around $432 million, whereas expected savings from reduced economic consequences are over $1 billion in most cases. Cost becomes an issue only when the probability of an attempt in the next five years is very low. There is another reason why full deployment of plastic explosives detection could be recommended. The adaptive behavior of terrorists could reduce the value of improving inspections exclusively at one layer of security. For instance, if checked bag inspections are hardened, then terrorists may look for ways to carry out an attack placing the bomb in a carry-on bag. In this regard as well, a strategy that eliminates all potential gaps in security is preferable. In the last section of the report, we propose tests to be conducted at LAX to evaluate practical aspects (integration strategy, ease of use, etc.) of integrating this technology into the current security strategy, and yield operational data (detection speed, detection accuracy, false positives, etc.) to be used in risk assessments. We would conduct controlled experiments to assess the feasibility of using this technology at various checkpoints, and develop a risk-based recommendation of cost-benefit. We would use an explosives simulant that has the same chemistry as the peroxide explosive without its hazard potential. We would first test how the technology performs in real life situations and its ease of use. Then we would generate estimates of the probability of correctly identifying the threat item and the percentage of false positives using the technology. There are potential three detection nodes: Car checkpoints, carry-on luggage inspection portals and checked luggage inspection stations. These three nodes are where the new technology could be applied to improve the probability of detecting the threat material. The experiments would enable us to understand how we improve detection probability at each node. We would also conduct sensitivity analysis to understand how robust the results are. Expert opinion from TSA, LAWA and others would also be factored into the analysis to identify both the base values and reasonable ranges for parameters used in the model. Before the experiments, we would provide a short training to the operators at each site. The experiments would be conducted in a controlled environment, closed to the general public, and would not be covert, that is, the operator would know in advance of the presence of the simulant. The goal is not to test the operators’ skills, but rather understand how practical it is to use the technology during inspections. Feedback from operators would be sought. Their feedback would be useful in testing the performance of this new technology. We would explore the use of the technology as a supplement to Explosive Trace Detection (ETD) machines.