Simulation: CBRN Detection
© 2009 FrontLine Security (Vol 4, No 1)

Public transportation systems offer significant potential targets for terrorist attack, as the large numbers of people in enclosed environments would contribute greatly to the devastating effects of biological and chemical weapons. However, there are options available to address this type of threat.

One possible option includes a network of sensors for early detection and warning of chemical, biological or radiological agent release. In configuring such a system – in a subway station, for instance – it is critical to position the sensors and configure the alarm system in a way that provides the earliest-possible warning yet avoids false alarms.

Accomplishing this task requires an understanding of how an agent is dispersed through air currents affected by the motion of trains, ventilation and air conditioning systems, along with other secondary influences. It is difficult to obtain this information from physical testing because of the time and cost involved in using an actual subway station. Engineers have found a way to overcome this challenge by using Computational Fluid Dynamics (CFD) to reliably simulate the release of an agent under a wide range of conditions.

Network of Linked Sensors
An effective system would link a network of remote sensors to a command and control station. Designed to detect the presence of a potential biological, chemical or radiological aerosol, each sensor counts particles present in the air to compare with real-time background information. No single sensor can generate an alarm; but, when one sensor alerts, others are signaled via a sophisticated software algorithm to determine whether the ­signature of an agent release is present.

This approach provides for high sensitivity while reducing the potential for false alarms. The system is modular and highly adaptable to meet a wide range of deployment scenarios. The greatest challenge in deploying the system is determining the locations for the sensors and tuning the algorithms used to evaluate the sensor readings and to determine whether an attack may have occurred. To optimize this placement and tuning, engineers need to determine how a released agent will be dispersed through the subway station environment.

Many variables are involved in contaminant dispersion, including the release location and factors that affect air currents in the station. Using experiment to understand the release signature under various conditions would require many physical tests requiring operational time at the ­station and significant manpower. There are also environmental limitations on the volumes of non-toxic simulants that can be released and on how well simulants can represent the characteristics of the real-threat materials.

Air currents in a subway station are primarily influenced by the motion of trains and operation of the ventilation and air conditioning systems. Additional effects include factors such as varying weather conditions at passenger entrances, ventilation shafts and tunnel entrances. Air currents pass numerous corners and obstacles such as structural columns, stairwells and elevator shafts – all of which generate turbulent eddies of varying sizes. The transport and mixing of dangerous gases and airborne particulates is strongly affected by both the bulk motion of the air and the larger turbulent eddies.

Modeling an Agent Release with CFD
Engineers can employ specialized software to perform a Computational Fluid Dynamics analysis to predict dispersion in a subway station. Using technology such as FLUENT from Ansys, Inc., an engineering team can develop a multi-species model that enables them to track two or more gas species as they disperse throughout the computational domain. CFD was used to model releases under a wide range of conditions and to generate the information needed to optimize the deployment of the detection system. CFD provides the ability to capture both bulk flow and turbulent eddies; it also can handle the complex geometries involved in subway stations and other real-world applications.

Figure 1: Cloud spreads slowing with air conditioning turned off.

Figure 2: Cloud spreads more rapidly with air conditioning on.

Figure 3: A train passing through the station increases dispersion.

State-of-the-art modeling software can create the model geometry. The example here shows one subway station containing two tracks at the center, a two-sided platform on each side of the center tracks, and a single track on each side of the double platforms. The station contains stairwells, elevators, structural columns and beams, and service rooms. The platforms included stairwells, elevators and service rooms. Air conditioning re-circulation units above each platform, and a moving train with 10 cars, each with air conditioning intakes and ­outlets, were also modeled.

To minimize the number of simulation cells, and to maximize the accuracy of the solution, the entire station was meshed using hexahedral cells. Ten and a half million cells were produced. A dynamic mesh model was employed so, just as any real train moves through the station, the train in the computer model moves through the meshed domain. As the train moves, layers of cells are removed ahead of it and layers are generated behind it.

In order to resolve the medium and large eddy flow fields, the smallest meshed cells must have about the same size as the medium eddies.

Since turbulent eddies occur in proximity of the structural columns, where Transit MetroGuard System’s chemical and biological sensors are mounted, the columns were resolved sufficiently.

The inlets and outlets of the station’s and subway cars’ air ­conditioning systems were coupled. For example, if a certain mixture of air and toxicant is sucked through the subway car’s air conditioning unit, then that same mixture is assumed to be blown back into the station at the car’s air conditioning exhaust. Similarly, the mixture sucked into the air conditioning units above both platforms at a given track-wise location is blown back into the station at the exhaust end of the air conditioning unit. The platform air conditioners have filters that entrap some of the particles.

Simulation Helps Under­stand Agent Dispersion
Engineers designed a series of typical runs to simulate a release of toxic gas that is ­immediately followed by a train entering the station, stopping and departing. The entire simulation spans five minutes of clock time, including a one-minute stop. Simulations were run with the train’s and station’s air conditioning systems both turned off and functioning.

Figure 1: Dynamic mesh used to model moving trains.

Figure 2: Mesh was sized to resolve Eddies.

The results clearly show that the effects of train motion and air conditioning units on trains and in the station are the major causes of agent dispersion. This is not surprising, considering the train’s speed and volume cause a push–pull effect and that the station and train air conditioning units have very large flow rates. Recirculation zones that further contribute to agent spreading were observed around physical obstructions, such as stairways, columns and rooms throughout the station.

As expected, agent concentration levels at the sensor closest to the release point indicate extremely high airborne particulate counts and toxic gas levels.

During the initial periods after agent releases have started, the matter tends to remain relatively close to the platform floor. But as the train enters the station, agent dispersion is more aggressive in longitudinal, transverse and vertical directions. The agent migrates outside of the station domain, as shown by virtual monitors set up at the tops of stairways leading to the mezzanine, bottoms of stairways leading to a subway station below, and adjoining tunnel sections.

Results show that intake filters for air conditioners located in the station capture approximately 40 percent of the biological particulate matter, as determined at the end of the five-minute simulations. Thus, these filters have a significant beneficial effect on airborne concentration levels, to which the public would be exposed in this attack scenario.

With the air conditioning turned off, the gas cloud spreads slowly before the train passes the release point. The train entrains the cloud, carries it along and spreads it everywhere in its vicinity. The columns, stairwells and booths act as deflectors that generate jets. The wakes of the obstacles also act as turbulent mixers.

With the air conditioning turned on, even before the train arrives, the toxicant spreads across the tracks because of air currents generated by the station’s A/C units. The toxicants are sucked into the air conditioning system through the station’s supply vents.

Agent concentration signatures acquired from the various simulations were used to obtain detection statistics for the Transit MetroGuard System. The information ­contained in these signatures was used to determine agent propagation rate, a key component of the system’s operating ­parameter settings.

Once extended periods of subway station background data were recorded from the various networked sensors and all system parameters were set, system alert thresholds were established.  Signatures obtained from the CFD simulations then were injected on top of the measured background data set, and system level detection statistics were compiled.

When planning for wide-scale system installations, the results analyses such as these can support the generation of system performance predictions for a varied set of subway platforms. CFD simulations significantly contribute to the design planning of the sensor systems, including determination of the optimum sensor placement and quantity along with sensor pickup sensitivity, increasing detection probability while reducing false alarms.  

Edward Vinciguerra is the Equipment Manager for Undersea Systems with Lockheed Martin Maritime Systems and Sensors in Mitchel Field in New York. For more info visit:
© FrontLine Security 2009