Mr. Joseph L. Smith is a director and senior vice president of Applied Research Associates, a 1,500-person engineering and sciences consulting firm. He has played a significant role in the U.S. response to the threat of terrorism. He was intimately involved in the initial development of the Interagency Security Committee Security Design Criteria that applies to all new federal facilities and modernization projects. In addition, he has led development teams in the creation of new technology and computer programs to aid in the protection of people and high value assets during terrorist attack. Computer programs include WINGARD and WINLAC which are recognized national standards for the prediction of window response to blast loads. His teams have provided security and protection consulting and design services for many US icons and symbols including: the Liberty Bell, the Statue of Liberty, the Washington Monument, the White House complex, the Lincoln and Jefferson Memorials, the Smithsonian Institution, and the National Archives vaults that secure the Constitution, the Declaration of Independence, and the Bill of Rights, and many others. Mr. Smith is a member of ASIS and a board-certified Physical Security Specialist (PSP).
On Tuesday, October 12, at 4:30 pm, Smith will lead the session “Active Shooter and Force-on-Force Advancements in Simulations.” The session will demonstrate how virtual simulations can help organizations and facilities prepare for the unlikeliest of scenarios, such as the Mumbai terrorist attacks and the Virginia Tech massacre. Virtual simulations provide cost-effective, valuable insight into scenarios that cannot be realistically played out in real-world exercises. The solution will be demoed for the audience.
What are force-on-force simulations?
Over the last several years, it became apparent that understanding human behavior is very important in emergency situations, especially in cases where you have evacuations of people within damaged facilities. Or you may have an active shooter in a crowd and the crowd would have to react to that shooter as well as any kind of response force reacting to that shooter.
So what we have done is developed, over a period of time, technologies that use what’s called agent-based modeling, where individual humans are modeled in a computer program so they have their own individual characteristics. They can move at different rates depending on their size, age, weight, sex, and other physical characteristics. Their awareness of the facility is also replicated, for example —whether they’re visitors or employees, whether they are awake or asleep in a hotel, or whether or not there’s damage to the facility. And these computerized or simulated human agents can then share information within the computer simulation so if, for example, one of the agents sees an adversary, a damaged hallway, or smoke or fire they can report that to others in the simulation, so they can learn from their own environment.
Simplifying it enormously, this is a training video game?
It’s much more than that. This allows security personnel and planners to look at large numbers of people in cases that cannot be investigated in the real world. You know all of us in school went through fire drills. Imagine trying to do that in a stadium of a 100,000 people, or in a World Trade Center with 50,000 people. They don’t do full-scale evacuations because of the expense and the potential safety issues to the people involved and the disruption of daily life. Imagine what it would cost to empty out a couple of sky scrapers with 50,000 people in it and shut down business for two or three hours.
Do employees of individual tenants in a building undergo this training or is it for security professionals?
Usually it’s done at different levels. The security professionals will probably look at it in-depth for numerous types of scenarios. What we call perimetric analysis is where we vary various parameters and then look at what the effects may be. That sort of information would then be consolidated into training for the day-to-day employees so they wouldn’t have to look at all the detail, but they would benefit from the results.
So the security professionals use the simulations to create best practices for a particular situation?
Yeah, absolutely, I can give you a good example. We did a simulation of a large convention center that had approximately 40,000 people in it. In this particular simulation, our client wanted to look at what would happen if there was a chemical attack by a terrorist in this particular place. And we modeled it with their existing notification and evacuation procedures and found that there were over 1,000 fatalities in the simulation by the exposure of this chemical. By going through and looking at alternate evacuation routes and alternate notification operational procedures, we were able in the simulations to get this down to just a couple of dozens of people exposed to lethal levels of gas. So what we found was that just by looking at the operational procedures through such simulations there is a potential to save a significant number of lives.
How do you deal with the human element in these situations considering no one responds exactly the same during a crisis situation? How do you account for that?
When developing the technology we employed the use of cognitive psychologists. So we looked at what is the range of realistic behaviors. And these agents in these computer simulations have a statistical variation and the computer program allows the agents to randomly have a mix of that behavior pattern within that realm of what the cognitive scientists told us was reasonable behavior. All those types of characteristics can combine to create realistic behaviors when you look at the statistical distribution of behaviors over say hundreds or thousands of these agents.
How long does it take you to model some of these scenarios?
It’s actually getting much faster than it used to be. We can do a fairly sophisticated model of a large building in two-to-three days. And that’s building the geometry of the building. And then there are different ways you can populate that with people. You can populate it statistically so that it’s just spread across a wide population, or you can go in and individually model very specific behaviors of very specific people. Usually what we do is a combination of those two things. So usually it’s another couple of days of modeling the people in the scenarios. And then it takes anywhere from several minutes to several hours to run each scenario.
Of all the scenarios you model, what’s the most difficult kind?
I think probably the most difficult modeling is where people are exposed to time-dependent phenomenon. And let me explain what I mean by that. The case that I just gave you where we had people exposed to say a chemical gas, we have to have the people going through the simulation. They’re learning about their environment. But they’re being exposed to various levels of concentrations of this toxic material over time. And that will depend on where they are in the simulation and what they’re doing in a simulation as well as the spread of that gas throughout a facility based on its HVAC design. That gets very complicated because then you have to track at each time-step the concentrations of whatever gas maybe involved, but then you have to calculate at every time-step what the retained dosage is that these people are accumulating over time and what the effects are on those people—whether they are children, women, men of different sizes—because we all react differently to those kinds of gases.
So are active-shooter scenarios a time-dependent scenario?
I’ve seen the Department of Homeland Security guidelines for active-shooter situations and they don’t leave a person a lot of room for maneuver. How do you deal with another intelligent agent bent on destruction?
There’s a full range of active-shooter scenarios that you can look at. At one end you could look at say an active shooter, even a paramilitary force, trying to take a highly defended target like a nuclear power plant. If that was the case, then the scenarios would probably be played out to look at where you would position guards inside the secured facility, where would you position safety barriers, where would you preposition certain ballistic-resistant doors, and that sort of thing. So you can actually look at the active shooters coming into the secured facility, look at the potential response of your responders and the time for them to take out an aggressor or the time for the aggressor to get to certain targets within that facility. So that would be one extreme.
The other extreme would be an active shooter going into a place which is not defended. Let’s say, for example, we had an active shooter that went into a large venue like a crowded stadium and just started shooting people. The things you could look for is panic and you would want to look at the effect of that large crowd panicking and moving and how can you get them out of that facility and out of harm’s way as fast as you can. So there’s an evacuation scenario. The other part is you would want to look at the prepositioning of response forces and where those response forces should be in order to best respond to a crowd that’s now moving in and around an active shooter as well as what could you expect in terms of lines of fire.
I’d think there are too many variables to run a simulation like that. What if there are five or ten active shooters like in Mumbai?
So what you have to do is play out a number of plausible scenarios. Now obviously you can’t predict every possible scenario. You don’t have time to run every possible scenario but you can postulate from historical events and based on known vulnerabilities of facilities where a likely set of attack scenarios may occur. Again, now it’s not always going to be exact but it can give your security forces at least a good idea of how best to prepare. The better you can define the scenario obviously, the closer to reality the solution will be. And there is some professional judgment that has to be used to set up the range of scenarios. The fortunate thing is that once you set up a structure and set up a population of people, running ten, twenty, thirty, forty, fifty simulations is just a matter of time and computing. So you can determine that if they add more than five active shooters that’s when my response forces can’t respond. If they have blocked certain exits, that’s where I’m really going to have a problem. You can look at the key things you have to worry about and the things you don’t have to worry about in certain scenarios.