A Historical Model of Fallujah as the Precursor to a Possible Shift in Marine Corps Tactics toward Distributed Operations

Written By:

Midn. Edward Brown

17 AUG 2004

Distributed Operations

The threat of terrorism, when combined with the recent increase in the number of asymmetric threats throughout the world, has led the U.S. Marine Corps to contemplate the possibility of a new set of tactics called Distributed Operations (DO). Marine units in Iraq are constantly being attacked by insurgents who operate in small teams scattered all over the battle space, popping out of crowds and buildings to execute quick strikes, only to disappear back into the surroundings. These attackers are constantly searching for the well defined center of gravity that a conventional unit presents. Not only will DO allow the Marines to better conceal that center of gravity, but it will also make it possible for Marine units to operate in much the same way that the insurgents do. In a sense, it gives the Marine Commander the ability to fight fire with fire. Distributed Operations calls for a dispersion of units, spreading themall across the battle spacein squad-sized teams. Dispersed in this manner, the Marines will be able to conduct simultaneous surprise attacks, converging on the enemy from all sides. The Marines will also be able to gather more intelligence from all over the area of operations. These are just some of the expected advantages that DO will provide. In order for the Marines to successfully carry out Distributed Operations, special emphasis will need to be placed on the training of individual squad leaders. Given that squads in a distributed unit will need to be very autonomous, Sergeants will have to be given much of the same training that a platoon commander would have. However, there is still a great deal of work tobe done before this training can take place.

Distributed Operations and Fallujah

It appears that the circumstances of combat operations in Iraq are already forcing the Marines to utilize tactics that a formal concept of Distributed Operations would standardize and train. Squads are going on individual patrols, fire support is being called in at the squad level, and units are being dispersed throughout the countryside in an effort to gather intelligence from the local population. At this point in time, Operation Vigilant Resolve in Fallujah, Iraq is the best historical case of what appears to be Distributed-like Operations. Our first goal in creating a historical combat model of the events that took place in Fallujah was to reproduce the data point that the actual fighting provided. Then, we intended to see what we could learn by sending the model in to data farm over parameters that we thought would make the Marines seem more distributed. By doing this we hoped to see how well a distributed force would have faired in the fighting in Fallujah and what we may be able to change to make them better.

Modeling Fallujah

In order to create a model of fighting in Fallujah we decided to use MANA version 3.0.35. Before modeling could begin we had to do some research on Operation Vigilant Resolve. The majority of this research was done using internet news articles and a few official reports from various internet sources. Thus, the accuracy and quantity of information that we were able to gather is somewhat limited, but we made due with what we could find during a limited time frame.

We started the modeling process by finding a map of Fallujah on the internet and using it as the basis for our terrain map. We colored over the map in Microsoft paint, creating a few new terrain types to better simulate the urban environment. MANA specific colors map to pre-defined terrain features such as hills, valleys, buildings, open field, roads, etc.

Once we had a terrain map to use as a reference for the movement of all our different squads, we began to create our prototype squads. For the blue force, which represented the Marines, we decided that each agent would represent a four man fire team with 5 hits to kill. Using information on actual ranges, accuracy, rate of fire, etc. we created a set of weapons which represented the M16, M203, M249, and the 50cal. (for vehicles). In addition to the mounted and dismounted patrols, we also modeled tanks, helos, AC130s and artillery. For the red force, representing the insurgents, we decided that each agent would represent 3 or 4 people. They were given weapons which were meant to represent AK47s and RPGs. After creating the base units, we then began to replicate them and script their personalities and actions to fit historical accounts.

Model Timeline

We decided to model five days of Operation Vigilant Resolve, from the commencement of operations on 4 Apr 2004 to the tenuous cease fire on 9 Apr 2004. The entire five days is represented by 3600 steps. Therefore, each day is 720 steps long.

On the first day of operations, the Marines push all the way into the center of the city using tanks, helos, and armored vehicles. Later in the day, a squad is fired upon from a mosque and tanks and helos are called in for support. Beginning during the first day and continuing throughout the simulation, Marine sniper teams patrol the southeast boundary of the city and are periodically fired upon from the towers of the central mosque.

The only event that takes place on Day Two is heavy fighting in a northwest neighborhood of the city. Three squads are attacked and call in fire support from an AC130.

Day Three begins with a gunfight at the train station near the northwestern corner of the map. Later on there is some fighting in the southwestern region of the city, followed by some probing of Marine squads dug in along the railroad on the northern perimeter. These dug in squads have the ability to call helos for support.

Day Four begins with a large battle in the northwest corner of the city. The Marines communicate through a surveillance platform to receive fire support from an artillery unit.

At the outset of Day Five a squad of Marines with higher than normal stealth conducts house to house raids, calling in fire support from an artillery unit. A few insurgents in vehicles try to run the roadblocks on the northern road and one of the western bridges. Finally, one last group of insurgents tries to probe the northern perimeter and the Marines call in helos to deal with the problem.

At the end of the five days, there was a real life casualty count of 40 Marines and about 600-800 insurgents. On average, the model tends to produce similar numbers.

Data Farming

Given that the scenario was somewhat scripted to reproduce historical accounts, it was difficult to find parameters to data farm over that didn’t interrupt the some of the necessary elements of the scenario. Such scripting also seems to limit the amount of valuable information that data farming can produce. However we decided to data farm over a few different parameters: Red Stealth, Blue Stealth, Blue Comms Accuracy, Blue Comms Latency, Size of Small Red Squads (ambush squads), Size of Medium Red Squads (skirmish squads), and Size of Large Red Squads (battle squads). We chose these parameters because we decided to look at things that would make Blue be more DO-like. We decided to increase the size of red units to see if they were helped or hindered by massing their firepower. The data farming parameters are listed in the table below.

Parameter / Values Farmed
Comms. Latency / 0, 10, 20
Comms. Accuracy / 80, 90, 100
Blue Stealth / 0, 30, 60, 90
Red Stealth / 0, 30, 60, 90
Size of Small Squads
Size of Medium Squads
Size of Large Squads / 2, 6, 10
15, 30
30, 60

Observations

Data farming results show that although blue stealth greatly reduces blue kills, it doesn’t really lead to more red kills. Red stealth seems to have the seem effect in reducing red kills as blue stealth does in reducing blue kills until red stealth passes 60. At this point, there is a huge decrease in the number of red kills. I believe that these results are due to either an increase in instances were red never makes contact with blue or due to the fact that red kills more of blues helos at 90 stealth, significantly reducing blue’s fire support capability (see charts).

By looking at the red force, which was, both in the model and in real life, the more distributed of the two forces in terms of size, tactics, and dispersion, we found that increasing the amount of troops that you throw into the fight on the size of the distributed force seems to make them less effective. The more troops that the red force has the more their death rate increases disproportionately to blue’s death rate. In the graph below, the blue curve represents red kills and the red curve represents blue kills as they vary across the changing size of the red force.

This may give weight to an argument for distributed operations at the smaller unit level, using squads as the base distributed unit as opposed to a platoon or some set of larger units. Although the concept of using DO at the squad level already seems set in stone at this point, it doesn’t hurt to reinforce the idea.

Data farming over Communications Accuracy and Latency produced no real noteworthy results. However, we determined that this particular model wasn’t really geared towards a test of the effectiveness of communications for a number of different reasons, including the degree to which the actions of the agents were scripted to match up with history.

In summation, in order to get at our first goal of reproducing the historical data point that Fallujah had already provided we had a tendency to script the actions of the agents in the model in such a way that may prevent us from getting any new information of significance from the data farming process. However, data farming did help to reinforce many of the ideas that lessons learned in Fallujah along with previous studies of DO had already created.

Ideas for Future Study

I would like to redo this scenario with the intent of making it less scripted. I believe that my knowledge and experience with MANA now that I have completed my internship is strong enough to allow me to take much of the scripting out of the scenario and allow it to run in a more autonomous manner without diverging too much from the historical truth of the event

After watching my model run numerous times, I noticed that every now and then a helicopter gets shot down. I realize that in real life this would be a catastrophic occurrence as was evidenced by the difficulties that Task Force Ranger had in Mogadishu in 1993. It would be interesting to see what would happen if a distributed force, dispersed all over the battle space, was forced on the defensive in an effort to respond to something like a downed helo. Would they be able to get enough troops on site fast enough? Would the small size of distributed units be able to survive the magnetic effect that a downed helo or a stranded convoy has in an urban environment?

To go along with the previous concept, how effective would a totally distributed force be in a hostile urban environment period (disregarding the effects of a downed helo or a disabled convoy).

Conclusion

I believe that as we delve further into the study of Distributed Operations we will continue to raise more and more questions that need to be answered and create more and more scenarios and possibilities that need to be analyzed. I think we are still in the early stages of understanding the pros and cons of Distributed Operations, but what I have seen so far seems quite promising. Once the concept has been cleaned up and all of the kinks have been ironed out, I think we will find that Distributed Operations has far reaching advantages in terms of the way the Marines conduct operations.