Use of Environment Simulation to Support Passive Chemical Sensor Development

Transcription

1 Use of Environment Simulation to Support Passive Chemical Sensor Development David M. Jodeit ITT Industries 2560 Huntington Avenue Alexandria, VA (703) Dennis L. Jones ITT Industries 2560 Huntington Avenue Alexandria, VA (703) Richard McMahon Army Research Laboratory Human Research & Engineering Directorate Aberdeen Proving Ground, MD (410) Robert Ryan ITT Industries 2560 Huntington Avenue Alexandria, VA (703) Dr. John R. White Edgewood Chemical Biological Center Soldier and Biological Chemical Command Aberdeen Proving Ground, MD (410) Keywords: Passive; Sensor; Chemical; Reusable; Reconfigurable; SBA ABSTRACT: The Joint Service Lightweight Standoff Chemical Agent Detector (JSLSCAD) program managed by the Army s Program Manager for NBC Defense, Soldier and Biological Chemical Command, conducted a simulation study in support of the system s critical design review (CDR). ITT and the Army conducted the study in order to provide data on the sensor s field of regard (upper and lower elevation extents) as a function of terrain type and sensor platform. JSLSCAD program personnel will use this data to support system design decisions in conjunction with the CDR held December The JSLSCAD is a 360-degree scanning, detect-on-the-move, passive standoff (up to 5 km) chemical detector that operates on a variety of platforms. The study used the JSLSCAD Distributed Simulation (JLDS), which is an easily configurable, reusable, sensor model that is attachable to any simulated entity or instrumented live entity. An ITT-developed M93 FOX NBC reconnaissance vehicle simulator containing a high-fidelity mobility model of the FOX served as an unstabilized sensor platform, and ModSAF provided a stabilized platform. Study personnel disseminated Sarin (GB) chemical artillery barrages using the Nuclear, Chemical, Biological, and Radiological (NCBR) Environment Simulator. The two-week, 36-trial study included three different terrain profiles (rolling, hilly,

2 mountainous), and each of those profiles had three fields of regard and three agent cloud releases. The data collected in this experiment consisted of data logs and a semicolon-delimited ASCII text file containing all of the positive sensor scans. The Army will use the leave-behind architecture from this study to support virtual human-in-the-loop testing of the JSLSCAD and training for the fielded system realizing the DoD s thrust in simulation-based acquisition (SBA). The topics covered by this paper will include: the JSLSCAD; the simulation architecture; preliminary study results; and possible future follow on work. 1. Introduction and Background The U.S. Army s Edgewood Chemical Biological Center (ECBC) Modeling & Simulation Team and the Army Research Laboratory, Human Research & Engineering Directorate supported the PMNBC / MARCORSYSCOM Joint Service Light Standoff Chemical Agent Detector (JSLSCAD) program by conducting a series of simulation studies aimed at providing predictive data to support various system engineering decisions. These studies were intended to collect data to help define: - the most effective operational scan pattern field of regard (FOR) for the detector when operated as part of a land-based NBC reconnaissance system - possible future support roles of the JSLSCAD simulation. 1.1 JSLSCAD Overview The land-based, vehicle-mounted JSLSCAD operates on the move with a 72-Hz scan rate in a repeating 360-degree horizontal rotation. The scanning begins at the lower elevation limit in counterclockwise (viewed from the top) direction. The scanner is gradually incremented upward in elevation to achieve a 1.5-degree movement per full revolution with a 144-degree/second great circle arc sweep. The resulting path is a continuous smooth spiral from the bottom elevation limit to the top elevation limit. When the top limit is reached, the scanner is indexed to its initial scanning start point, and the scanning is repeated. The JSLSCAD s scan sweep is a 144 degrees/second great circle arc sweep that is constant at any elevation. So, for example, if the scanner is at 30 degrees elevation, the horizontal RPM component of the scanner head is 144/cos(30) = degrees/second. In general, the horizontal motor rotates at 144/cos(elevation) degrees/second. This also affects the speed of the vertical motor as it needs to achieve 1.5 degrees per horizontal revolution. The vertical FOR used for scanning impacts the effective "area covered" or reconnaissance effectiveness. 1.2 Study Goals This study addressed only the FOR issue. Data from the study was intended to support system design decisions related to FOR which were required at the critical design review (CDR) which occurred in December These data will also assist in optimizing scan parameters for detector operations and assess whether considerations for terrain type and JSLSCAD stabilization are worth pursuing. System detection performance was evaluated under two conditions. The first was as a physically stabilized detector. For this condition it was attached to a Modular Semi-Automated Forces (ModSAF) HMMWV vehicle entity and its position and orientation was recorded throughout the mission scenario. The second condition to be evaluated was as an unstabilized detector. For this condition it was attached to a manned virtual FOX vehicle simulator. This joy stick-driven simulator has a highfidelity suspension model that interacts with the digital terrain to represent the FOX's mobility characteristics. 1.3 Study Timeline This study was designed with the expectation that it would take 2.5 to 3 months to plan, execute, and document. Since ITT had solid requirements from the customer and most of the required infrastructure was already in place, it was not too difficult to adhere to the study timeline. 2. Study Architecture The JSLSCAD study employed five distributed simulation tools to create and analyze the required virtual environment. These are the Nuclear, Chemical, Biological, and Radiological Simulator (NCBR) [1], ModSAF, LSLSCAD Distributed Simulation (LDS), ITT s M93 Fox Simulator, and a modified MaK Stealth which displayed false-colored chemical clouds. The MaK Stealth was used to maintain situational awareness during the study trials (Figure 1).

3 Current Sensor Scan Range (distance to ground intersect or maximum range of sensor) Distance to Leading Edge of Cloud Concentration Length (mg / m 2 ) Figure 1. JSLSCAD FOR Study Architecture 2.1 JSLSCAD Simulator The ITT-developed JSLSCAD Distributed Simulation (LDS), which is based upon the architecture of ITT s Biological Standoff Detection System (BSDS) Distributed Trainer software [2], was used to represent the JSLSCAD system. The LDS is a realistic performance-based simulation of JSLSCAD. The LDS is designed to work in a distributed simulation environment and is fully DIS compliant. It receives cloud information from the DIS network via Environmental Process PDUs containing the Gaussian puffs, which comprise the hazard cloud. Currently there is no operator interface to the LDS and the test configuration is done through input files. The input file defines (among other things): entity to attach to; maximum sensor range; and time steps. Output of the LDS is an ASCII-delineated text file (which can be easily imported into a spreadsheet program for analysis) and a DIS Emissions PDU. While the JSLSCAD is a passive sensor and has no emissions it is useful for study purposes to output the Emissions PDU which describes the area that the LDS is currently examining. The Emissions PDU send rate is user configurable through the input files and the output text file is written to only when there is a positive scan return. The output file consists of the following data: Time Stamp (seconds) Vehicle Position (Lat. / Lon. / Elevation) Vehicle Orientation (Roll / Pitch / Yaw) Sensor Orientation (Azimuth / Elevation) The JSLSCAD operates on a 72-Hz cycle, but it is difficult to control the cycle that finely on a standard computer. ITT worked around this issue by running the LDS at a slower rate and calculating multiple sensor lines for each time step. During the study the LDS was run at an 18-Hz cycle with four sensor lines being calculated at each step. The vehicle and cloud were dead reckoned only for the 18 Hz cycle. The LDS cycle and the number of intermediate steps are both configurable in an input file. The LDS scan pattern and sensor characteristics are implemented as C++ class functions and are easily changed. By modifying these functions and recompiling the LDS, it can be easily reconfigured to allow the end user to test out a JSLSCAD with different scan patterns and/or sensor characteristics. 2.2 NCBR Environment Simulator The NCBR Environment Simulator provides simulated prompt and persistent nuclear weapon environments and chemical, biological, and radiological environments using DIS protocols. A NCBR Consortium led by the Army s Soldier and Biological Chemical Command and DTRA has supported the development of the NCBR. The NCBR provides real-time transport and dispersion of chemical, biological, and radiological agents over complex (3- dimensional) terrain with 3D, time-varying meteorological inputs, provides prompt nuclear environments, and tracks ground vehicle, rotary fixedwing aircraft, and naval surface vessel entity hazard exposure and contamination status. The NCBR informs entities of their exposure and contamination status via periodic update. The NCBR can accommodate multiple simultaneous agent events. The NCBR provides 3D chemical clouds to the distributed simulation network. The NCBR uses the DIS Environmental Process PDU to send the data. The clouds are modeled as Gaussian puffs. The NCBR periodically updates the state of the cloud. The Environmental Process PDU contains enough information to allow the receiver to dead reckon the cloud between updates. For agent transport and dispersion in the JSLSCAD study, the NCBR Simulator uses the Vapor, Liquid, and Solid Tracking (VLSTRACK) computer model developed by the Naval Surface Warfare Center in Dahlgren, VA. 3

4 VLSTRACK predicts contamination areas for a wide range of chemical, biological, and radiological agents. To calculate entity exposures to nuclear detonation events, the NCBR Simulator utilizes the DTRA Atmospheric Transport of Radiation (ATRv6). DTRA s XBLAST model will be used to calculate the peak overpressure and dynamic pressure (wind) associated with air blast and the thermal flux from a prompt nuclear burst. Threedimensional wind fields are provided to VLSTRACK from the Wind Over Critical Streamlined Surfaces (WOCSS) mass-consistent flow model. NCBR Simulator Versions 3.4 and greater also include DTRA s Secondorder Closure Integrated Puff (SCIPUFF) transport and dispersion model with the SWIFT meteorological model. The NCBR allows the user to select the hazard dispersion engine and associated meteorological model for the exercise or application. 2.3 Fox Simulator The ITT developed M93 NBC Reconnaissance Vehicle Simulator (FOX) is a high-fidelity, man-in-the-loop simulator that interacts in a realistic manner with the digital terrain and features. For the JSLSCAD Study a joystick-operated version of the FOX simulator was used. This version of the FOX does not have a dashboard or vehicle location indicator, so a ModSAF vehicle traveled the study route and one of the study personnel followed that vehicle with the FOX. 2.4 ModSAF The JSLSCAD study exploits the community-standard ModSAF, a STRICOM-developed, platform-level constructive simulation that represents the outward behavior or performance of simulated entities. ModSAF can play various blue and red ground and air vehicles. JSLSCAD attached the sensor simulator to a ModSAFgenerated constructive vehicle simulation (no human in the loop) for the first testing condition (stabilized sensor platform). In testing the second condition (unstabilized JSLSCAD platform), ModSAF provided a leading vehicle which the man-in-the-loop simulator could follow to drive the appropriate ground route. 2.5 MaK Stealth The MaK Stealth displays a 3D terrain database and the entities in the virtual environment. For the JSLSCAD study ITT used a MaK-modified Stealth to display the 3D Gaussian puffs representing the chemical cloud. The study uses MaK s Stealth to display the terrain, vehicles, chemical cloud, and sensor lines from JSLSCAD (Figure2). Using the Stealth, the study personnel could quickly determine the relative positions of the cloud and the vehicles, and also whether or not the sensor line was intersecting the cloud. 3. Study Design Figure 2. MaK Stealth Display The study consisted of a multi-day effort examining several different scan patterns/extents on different terrain types in stabilized and non-stabilized modes. A JSLSCAD with a FOR of 3 to +30 elevation will certainly scan the named area of interest (NAI) faster than a JSLSCAD with a FOR of 10 to +50. Therefore, the primary measure of effectiveness is whether the JSLSCAD shot lines return CLs with the smaller FOR as readily as it does with the larger FOR. The second measure of effectiveness we will use is the time to return a CL of a given concentration. The study used a stabilized (i.e. ModSAF) JSLSCAD in hilly terrain and an unstabilized (i.e. ITT M93 FOX) JSLSCAD in flat, hilly, and mountainous terrain. The positions of the hazard cloud varied relative to the vehicle s track in accordance with Figure 3 below. The vehicle paths were dependent on the specific terrain chosen in concert with program sponsors for the mission. ITT conducted the exercises on the Ft. Hunter Liggett digital terrain database due to its availability, its representation of the desired terrain types (flat, hilly, and mountainous), and our familiarity with it. The ModSAF mission scenarios for this study incorporate the operational ideas and techniques discussed with personnel from the Reconnaissance Branch, US Army Chemical School (USACMLS). 4

5 Cld 2 7km 5km Cld 3 Cld 1 1km Figure 3. Sample Hazard Laydown Relative to JSLSCAD Platform Track (single cloud per trial) SP Tactics, techniques, and procedures (TTP) for employing the JSLSCAD have not yet been written. Therefore, the study personnel coordinated with the JSLSCAD program office to develop an acceptable route reconnaissance mission. Typically, a security element would accompany the Fox during an NBC route reconnaissance. Movement techniques of the FOX in concert with the security element are situationally dependent. Initial speed for the movements was 25 kph. The flat terrain study scenario was a route reconnaissance by a single FOX NBCRS. The terrain from the start point (SP) to the final checkpoint (CP3), in this example, was essentially flat and open and covered a distance of 10 km. Elevation along this route varied between 281 to 299 m. There were no significant hills within the flat terrain area of interest that would cause FOR obstruction problems. The hilly terrain example (Figure 4) was characterized by rolling hills, which limited the line of sight for standoff detectors. The example route from CP4 to the release point (RP) covered 10 km and the elevation varied between 334 and 472 m. There were numerous hills within the mission area, the highest, located in the southwest corner, was 1100 m in elevation. Figure 4. Hilly route (ModSAF) The mountainous terrain scenario was characterized by high-relief terrain, which limited the line of sight for standoff detectors and also limited the speed at which the FOX vehicle could travel. The chemical strikes in all missions consisted of 180 rounds of 152-mm artillery (the equivalent of a battalion firing 10 rounds). Each round filled with 4 kg of the agent GB (Sarin). These are representative threat rounds from the VLSTRACK threat tables. All chemical laydowns were surface bursts approximately 800 m long by 400 m in depth. 4. Study Results 4.1 Stabilized Platform vs. Unstabilized Platform At first glance it would appear that a stabilized JSLSCAD would be able to more effectively cover the NAI, but upon closer examination of the study results it did not appear that one had a clear advantage over the other. Figure 5 shows a plot of the stabilized platform on the hilly route with the third hilly cloud. The X-axis is the time of the event (in seconds) and the Y-axis is the distance from the sensor to the edge of the hazard cloud (in meters). 5

6 but the longer time it takes to complete one scan cycle offsets that advantage. Also, depending on the movement of the cloud relative to the sensor and the size of the threat, it is very possible that larger FORs will miss the cloud entirely. 4.3 Effects of Terrain Figure 5. Stabilized Platform on the Hilly Route (Cloud 3) Figure 6 shows the same route and cloud, but with the unstabilized platform. The axis and the dimensions are the same as the previous illustration. The terrain effects tie in very closely with the different FORs. On a flat terrain, the operator could use a smaller FOR and still have a good chance of detecting the cloud. But once the sensor vehicle enters terrain where the threat cloud may move up and down hills or through valleys the operator would have to enlarge the FOR to increase the chances of detecting the cloud. Some particular examples of this are: the sensor vehicle traveling along a ridge line would want to have a greater declination and a lower elevation on the sensor in order to look down into the valley; and a sensor vehicle traveling on the valley floor would want to do the reverse. The extent of the FOR though, is still very important in that the smaller the extent the shorter the scan cycle. 4.4 Effects of Cloud Position to Sensor Figure 6. Unstabilized Platform on the Hilly Route (Cloud 3) One thing that we encountered during the study that was surprising was that when the sensor vehicle entered the threat cloud the concentration lengths (mg / m 2 ) returned to the sensor actually started decreasing until the vehicle was leaving the cloud. There are two major reasons for this. The first reason is that the clouds are represented as gaussian puffs and as the vehicle moves through the cloud it no longer has a long integration path through the cloud. The second reason relates to the sensor s elevation angle. At some sensor elevations the sensor is either aiming towards the ground relatively close to the vehicle or is aiming higher than the plane of the vehicle and is then shooting upwards out of the threat cloud. Figure 7 illustrates this. In the images each point represents a positive scans of the sensor. The blank areas are either where the sensor was pointing away from the cloud or where it was aiming under or over the cloud. While the number of bands is the same between the two images there is some difference in the number of positive scans per band. 4.2 Effects of Different Fields of Regard The study looked at three different fields of regard (FOR, or elevation extents): -10 degrees to 50 degrees; -3 degrees to 30 degrees; and 4.5 degrees to 4.5 degrees. A scan cycle is the amount of time required for the sensor to scan the entire FOR. The larger FORs had a potentially greater chance of detecting the agent during a scan cycle, 6 Figure 7. Concentration Length vs. Sensor Elevation (inside a cloud)

7 Figure 7 is a graph of a subset of data from a test run where the sensor vehicle was inside the threat cloud. The Y-axis represents the sensor elevation ranging from 3 degrees to +30 degrees. Looking at the figure it is readily apparent that the highest concentrations occurred the sensor was at an elevation of approximately 0 degrees to +3 degrees. 4.5 Preliminary Results Examination of the data by study personnel provided a few quickly observable results. Of the three FORs that were studied, the one that had the most satisfactory results for detection purposes was 3 degrees to 30 degrees. For flat terrain the benefits of using a smaller FOR were marginal and, in general, the 10 degree to 50 degree FOR was impractical. 4.6 Use In Critical Design Review Final CDR design decisions have not yet been made available to the study personnel, but the preliminary results have shown at least two areas where the JSLSCAD team can realize immediate benefits. The first is in not needing to spend extra development time and funds in building and testing a stabilized JSLSCAD. The data from the study indicate that there is no clear advantage in using a stabilized sensor platform. The second benefit comes from the JSLSCAD team being able to narrow down the range of FORs that will maximize system effectiveness in the operational sensor. It is likely that there will be hardware-imposed FOR limits on the fielded sensor. By examining the results of ITT s study the JSLSCAD team can make more informed decisions on what those limits should be. The JSLSCAD team has already indicated an interest in conducting a follow on study to more analysis using different upper and lower elevation extents. 5. Possible Follow on Simulation Support Work 5.1 JSAF Navy Study This study was primarily concerned with the effectiveness of a land-based JSLSCAD. But considering the joint services nature of the JSLSCAD sea-based effectiveness will also have to be addressed. For this reason ITT has proposed to do a follow on study in which our LDS would be attached to an entity created by a simulation capable of generating realistic oceangoing vehicles. One possibility is to use JSAF to generate these vehicles. In this follow on study the parameters would be very similar to the land-based study with the exception of varying wave height rather than terrain topography. 5.2 Graphical User Interface Front End At the current stage of JSCLSCAD design there is a human operator running the system. When the sensor is ready to be fielded the joint services will need to have trained operators to man those sensors. The JSLSCAD is still in the developmental stage so no user interface yet exists for the sensor. However, once that interface is ready, the LDS could be readily adapted to feed data to the interface outputs and to accept inputs from it as well (see Figure 1). By integrating the resulting training system with a complete set of tools (NCBR, ModSAF, etc.), a complete training suite could then be provided to the customer. 6 References [1] See also: 98F-SIW-140, Fall 98 Simulation Interoperability Workshop, September 1999; 99F-SIW-033, Fall 99 Simulation Interoperability Workshop, September [2] O Connor, Michael J., et al. Developing Biological Hazard Detection Tactics, Techniques, and Procedures Using Distributed Simulation 98F-SIW-140, Fall 98 Simulation Interoperability Workshop, September Author Biographies DAVID M. JODEIT is a Software Engineer for ITT Industries. He was the lead software engineer on the BSDS Training System, the BDT Simulator, the Chemical Standoff Detection Server (CSDS) System and most recently the JSLSCAD Study. He has a 5-year background in software engineering and training systems. DENNIS L. JONES is the Manager of ITT s Simulation Systems Section in Alexandria, VA. He manages programs dealing with modeling and simulation tool development and application for WMD environments, effects, and sensor systems. He has a 14-year background in strategic and theater missile defense, counterproliferation, system integration, and weapons of mass destruction effects. RICHARD MCMAHON works for the U.S. Army Research Laboratory and is the acting chief of the Edgewood Research, Development and Engineering Center (ERDEC) Field Element of the Human Research and Engineering Directorate s (HRED). He is a technical specialist within HRED in simulation-based acquisition 7

8 and the application and promotion of human figure and human performance (HARDMAN III and IMPRINT) modeling techniques to support Army acquisition programs. ROBERT RYAN is a Senior Staff Scientist for ITT Industries Simulation and Training Department in Alexandria, VA. He is a retired Army Officer with a 20- year background in nuclear weapons effects and targeting intelligence and 6 years experience in developing ModSAF-based M&S scenarios. DR. JOHN R. WHITE is a Research Physical Chemist and Program Manager for Distributed Simulation with the Chemical and Biological Defense Command, Aberdeen Proving Ground, Edgewood, MD. Dr. White is responsible for developing and using advanced modeling and simulation to support the US Army's acquisition of chemical and biological defense equipment. 8