Can we really do less with less? Simulating the potential effects of reducing ICBM alert rates.

Introduction

In the continuing era of ever tighter budgets and creative resource allocation, one initiative currently being considered is reducing the threshold for required intercontinental ballistic missile (ICBM) alert rates. Historically, the existing space wings, which employ the ICBM force, have maintained raw alert rates that normally exceed 98 percent. Some assert that significant savings can be realized by reducing the anticipated alert rate to 90, or even 80, percent. This seems logical on the surface. If units do not have to respond to an off-alert condition until these lower thresholds are reached, surely they can do it with less manpower and resources. But if this is true, what effect do these lower alert rates have on national security objectives?

Background

The ICBM force has formed the bedrock for deterrence since the early 1960s. ICBMs in concert with the two other portions (bombers and ballistic missile submarines) of what was formerly known as the strategic triad, deterred potential aggressors from attacking the United States. The collapse of the Soviet Union and subsequent end to the Cold War is often attributed to the ability of the United States to maintain and modernize the components of the strategic triad, especially the ICBM force. Essentially, the Soviets could not keep pace with the US rate of spending and strategic capabilities. But now that the Cold War is over, what purpose does the ICBM serve in current US national security and military strategies? And of what importance is it to the Air Force?

The creation and implementation of the aerospace expeditionary force (AEF) construct has been a major shift for the Air Force. As a result, it is reducing forward basing of personnel and equipment and has become expeditionary. However, the ICBM force is not deployable and the personnel who maintain ICBM systems have specialized skills that do not seamlessly transfer to an AEF package. How does this force support the AEF construct? Essentially in the same manner it did during the Cold War--deterrence. The ICBM, at some level, still deters potential aggressors from attacking. However, since the United States is the only remaining superpower, does it need to maintain Cold War alert rate levels? That is a concern considering the growing number of nation states joining the nuclear club and with continued regional instability.

Analysis

Initial study team discussions led to the conclusion that creating a simulation utilizing Arena, a product of Rockwell Software, had the potential to yield the analytical information required. Although not a perfect reflection of reality, a simulation can be assumed to closely approximate reality if the input data is properly analyzed and programmed into the simulation. This requires data collection and proper process mapping.

Input data for a 6-month period was gathered and analyzed from Malmstrom Air Force Base, Montana, one of the three remaining ICBM wings and home of 200 Minuteman III ICBMs. This daily data included the types and numbers of available maintenance teams, the types and numbers of different tasks occurring, and the number of security escorts (required for insilo ICBM maintenance) available. Weekend data was excluded because maintenance performance was drastically reduced on the weekends and security escort availability was skewed. Empirical, discrete distributions for all maintenance team availability and for the task types that occurred were developed. A triangular distribution was employed for security escort availability. The process was then mapped using the task types as the arriving entities that required some type of action. Although not all task types have an impact on alert rates, most were modeled because they draw from the same major constraining resource--security escorts. This also helped to assess the impacts on personnel utilization at various alert rate levels.

Several different types of teams and specialists, including civilian personnel, perform maintenance on ICBMs. For the purposes of this simulation, the teams requiring analysis were: missile maintenance teams (MMT), electromechanical maintenance teams (EMT), facilities maintenance teams, periodic maintenance teams, corrosion control teams (CCT), and rivet MILE (Minuteman Integrated Life Extension). These teams encompass the personnel who perform maintenance related to maintaining alert rates and those teams that would benefit or suffer from any alert rate threshold changes.

The next phase of modeling the simulation involved verifying the proper flow for each task process and ensuring the number of tasks and teams generated daily were intuitively reflective of reality. A hierarchy was also established for drawing from the limited security escort pool, based on current practices and policies at Malmstrom. This allows for tasks to be cancelled and resubmitted for completion the next day if there are not enough security escorts or maintenance teams available to perform all maintenance tasks generated during a day (in the simulation). Finally, entities were created to reflect certain types of missiles that are off-alert on a daily basis. Another entity was created that caused complete cancellation of maintenance at established intervals. This last entity was made to reflect the cancellations that occur due to inclement weather or other considerations. This entity occurred approximately six times per year. Once the model became a reasonable representation of reality, statistical data collectors were developed within the model to capture the necessary measures of effectiveness.

The alert rate was calculated on a daily basis assuming any task occurring with alert rate impact resulted in 24 hours of off-alert time. This is fairly representative of reality since some tasks will take more, and some less, time to restore to alert status. Personnel utilization was defined as a daily percentage by team type based on the number of teams used divided by the number of teams available. The simulation was then run 30 times covering a 945-day period, representing 3-work years (weekday only), plus a 180-day warm-up period, which was added to allow the system to stabilize before data analysis. This original model was an As-Is model and was checked to ensure it was representative of current operations. Two additional models were created which allowed the alert rate to drop to 90 percent and 80 percent, respectively, before taking action on most off-alert conditions. Priority I and major program maintenance were allowed to continue as normal. All three models were run with the same replication parameters for comparison of data output.

Figure 1 displays the average alert rate over the 30 replications at each alert rate level (As-Is, 90 percent, and 80 percent). The overall average alert rate in the As-Is model including the 180-day warm-up period, is 97.7 percent. For the 90-percent and 80-percent models, also including the warn-up period, the overall averages are 88.8 percent and 80.1 percent. The story changes slightly when the warm-up period is excluded. The overall averages then become 97.7 percent for the As-Is model (no change), 88.4 percent for the 90-percent model (down 0.4 percent), and 78.4 percent for the 80-percent model (a 1.7 percent decrease). The changes in the 90-percent and 80-percent models reflect the exclusion of the time period when alert rates are being allowed to fall and then have reached a steady state. The As-Is model is always in this steady state. Of course, ICBM units are never able to maintain a 100 percent alert rate because missiles constantly require repair. The same holds true at the 90-percent and 80-percent levels, which become the new 100-percent targets.

[FIGURE 1 OMITTED]

Another area of possible savings is personnel utilization. Surely significant savings at the reduced alert rate levels will be seen, right? Table 1 displays the average utilization rates over all 30 replications during days 180-945 for each maintenance team type in the three models. It is important to note that only maintenance requiring missile site penetration was modeled. Therefore, some team utilization rates may appear low but these teams also perform maintenance not requiring site penetration. Statistically, at the 95-percent confidence interval, the only significant differences occur when comparing EMT at both the 90-percent and 80-percent levels to the As-Is model, and when comparing CCT at 80-percent to the As-Is model. Although statistically significant, these are only minor differences and EMT utilization actually increases at the lowered alert rate thresholds. This may be attributable to the fact that much of their maintenance is often unrelated to maintaining the alert rate. A line-by-line comparison for each team type shows that utilization for most teams remains both statistically and practically the same. This is contrary to what many intuitively believed--why were the anticipated savings not realized?

Several reasons may exist that prevent any tangible personnel utilization savings. One is that all teams that perform alert rate related maintenance also perform maintenance unrelated to maintaining the alert rate. These requirements continue to exist even at the reduced alert rate thresholds. Another factor is the impact that a single off-alert missile can have on the overall alert rate percentage at these various levels. In the As-Is model, there is a baseline of 200 missiles to work with. The impact of one off alert missile in this model is a decrease of 0.5 percent. At the 90-percent and 80-percent models, the baseline is effectively reduced to 180 and 160 missiles, respectively. A single off-alert missile then has an impact of 0.56 percent at the 90-percent level and 0.625 percent at the 80-percent level. Maintenance teams, therefore, have to work harder to try to maintain the status quo at the new threshold levels. But what about the period when the alert rate is dropping? Surely this period will produce significant savings.

A look at the three models shows on average across the 30 replications, the 90-percent model began to stabilize at approximately the 75-day point and the 80-percent model near the 179-day point. Tables 2 and 3 show the utilization rates averages over the first 75 days for the 90-percent and As-Is models and averages over the first 179 days for the 80-percent and As-Is models.

Statistical analysis, again at the 95-percent confidence interval level, shows only MMT utilization differs significantly from the As-Is model at both the 90-percent and 80-percent levels. All other team utilization remains statistically the same. Therefore, the only real personnel utilization savings realized within all model comparisons are the short duration MMT savings (recall MMT utilization is statistically equal beyond 179 days) and a 0.6 percent savings for CCT in the stabilized analysis.

Conclusion

What conclusions can be drawn and what intangible factors exist that were overlooked? First, one can see that realizing personnel savings by reducing required alert rates appears to be a myth. The data simply do not support such a conclusion. Second, the pool of available missiles has shrunk from potentially 500 in the As-Is model (200 at Malmstrom, 150 each at Minot and FE Warren Air Forces Bases) to potentially 450 at the 90-percent level and 400 at the 80-percent level. Decisionmakers at United States Strategic Command, the Air Staff, Joint Staff, and the Office of the Secretary of Defense should be well aware of these changes. Third, personnel proficiency, at least for MMT, can certainly suffer during the period of falling alert rates. Tasks during this period are performed less frequently, which is a major concern for teams that are handling nuclear weapons. Morale is also likely to suffer during this period, since personnel are normally more content when they are gainfully employed, thus eliminating complacency. Additionally, the effect on future reliability of the system after sitting for long periods of time off alert, is not apparent. Finally, what is the impact on spares levels, the effect on the overall supply chain, and what happens if a situation requires bringing all missiles back to alert status?

This last question poses scenarios that should be considered more in depth. Imagine if Malmstrom is sitting with 40 off-alert missiles that all require a missile guidance set replacement. Does it have the capability to store 40 replacement MGS units and also store any failed units? If not, does the depot have the ability to deliver them in a timely manner and receive the failed components? This is just one component that could likely cause an off-alert condition, but what about the others? What happens if the units are directed to bring these missiles back to alert status? Figures 2 and 3, both taken from replication number 16, show the amount of time it takes to regenerate back to within 2 percent of our original threshold (100 percent), first with the 80-percent model, then with the 90-percent model. The gray labels at the low end of the scale represent when an order is given to regenerate to within 2 percent of the maximum alert rate and the gray labels at the upper end show when that goal was achieved. All numbers are in work days and assume all previous model parameters remain the same (personnel availability and task occurrence distributions), and there are no logistical limitations, such as parts availability, transportation, and so forth).

[FIGURES 2-3 OMITTED]

It takes a significant amount of time to reach previous alert rate levels. The 90-percent model seems to show that units can recover very quickly, but further analysis shows that the first four regeneration orders occur at a time when the alert rate is already well above 90 percent. The final regeneration order occurs when the alert rate is below 90 percent and it takes significantly longer to achieve the goal rate. This also assumes there are no additional limitations--a highly unlikely situation.

It is obvious that achieving cost savings from reduced alert rates, while plausible on the surface, will not occur from the standpoint of personnel utilization or operations and maintenance costs such as fuel, parts, vehicle maintenance, training, and so forth. Additional logistical factors also must be considered before undertaking such a dramatic change in philosophy. The situation would likely be different if the United States were to retire a portion of ICBMs from the inventory altogether, as was the case under past treaties. These actions have resulted in savings and maintained the historically high alert rates the ICBM community is accustomed to. The driving and compelling need for any military capability has always been based on the needs and requirements of the warfighter, and this should also be the case here. Clearly, this is not a scenario where we will do less with less.

Article Acronyms

AEF--Aerospace Expeditionary Force

CCT--Corrosion Control Team

EMT--Electromechanical Maintenance Team

ICBM--Intercontinental Ballistic Missile

MMT--Missile Maintenance Team

Colonel Randy Tymofichuk is the Commander, 75th Maintenance Group, Hill Air Force Base, Utah. At the time of the writing of this article, he was a student at the Air War College, Maxwell Air Force Base, Alabama. Major Stephen M. Kravitsky is the Command Lead, Prompt Global Strike, Headquarters Air Force Space Command. At the time of the writing of this article, he was a student at the Air Command and Staff College, Maxwell Air Force Base, Alabama. Senior Master Sergeant Michael C. Dawson is the ICBM/Space Weapons Superintendent, Air Force Logistics Management Agency. Captain Tamiko L. Ritschel is the Chief, Military Analyses Branch, Logistics Analysis Division, Air Force Logistics Management Agency.

Table 1. Average Utilization Rates After 30 Simulation Runs
at Various Alert Rates-Days 180-945

Team    80%     90%      As-Is

MMT     0.78    0.788    0.786
EMT    0.914    0.918    0.891
FMT    0.606    0.605    0.605
 RM    0.558    0.557    0.559
PMT     0.30    0.302    0.301
CCT     0.37    0.372    0.376

Table 2. Average Team Utilization Rates After 30 Simulation Runs
at Various Alert Rates--Days 1-75

Team   90%     As-is

MMT    0.675   0.79
EMT    0.857   0.861
FMT    0.596   0.605
RM     0.559   0.566
PMT    0.304   0.293
CCT    0.362   0.359

Table 3. Average Team Utilization Rates After 30 Simulation Runs
at Various Alert Rates--Days 1-179

Team    80%     As-is

MMT    0.685    0.779
EMT    0.878    0.873
FMT    0.603    0.602
 RM     0.56    0.558
PMT    0.301    0.295
CCT     0.36    0.362