Featured photo by: Airman Justine K. Rho


Resilient Sustainment

By Dr. Daniel W. Steeneck and Maj Timothy Breitbach

Supply Chain Disruptions

Seasoned supply chain managers know that supply chain disruptions are going to happen.  The devastation at Tyndall Air Force Base due to Hurricane Michael is the most recent example of how catastrophic events can wreak havoc on operations, but as a large organization with activities across the globe, recent history in the DoD is littered with disruptions—and not just those of a natural type.

A September 2012 attack on Camp Bastion destroyed or damaged nine US Marine Corps AV-8B Harriers.  On the technical side, Air Force generation of combat sorties has been disrupted due to fleet groundings from safety concerns and common parts failures.  Recent examples include F-15s grounded at Kingsley Field for structural issues and hypoxia challenges across three different airframes—the F-22, F-35, and T-6.

The challenge of dealing with supply chain disruptions are not unique to the military.  For example, a  10-minute fire in an Albuquerque, NM semi-conductor plant “shifted the balance of power between two of Europe’s biggest electronics companies” and has become a classic supply chain case study. Though the negative effects of supply chain disruptions are known, many organizations find themselves woefully unprepared when these disruptions occur.

One challenge is that forecasts based on historical information may not predict the full range of possible future disruption types, timings and severities.  While there may be uncertainty surrounding the exact nature and timing of the disruptions, we can model the consequences of disruptions across a wide range of possibilities, and we can test how different supply chain designs perform against those consequences.

Responding to this challenge of prepping for low-probability, high impact events, we are exploring the concept of resiliency.  Along with a team of students and other researchers at the Air Force Institute of Technology (AFIT), we are addressing the critical issue of resilient sustainment by asking the following questions:

  1. How can supply chain resilience be quantified?
  2. What operational strategies can be employed to improve resilience?
  3. How should sustainment networks be designed for resilience?
  4. How much does resilience cost?

To answer these questions, we draw from academic, commercial and government sources to develop useful insights for Air Force decision makers.  First, we explore what resilience means and how different strategies affect resilience.  Then, we discuss supply chain strategy and the logic behind building resilient, but affordable networks.  Finally, we show model results that use representative data from the Pacific Air Forces (PACAF) Theater to demonstrate how the supply chain responds to disruptions.

Quantifying Resilience

While definitions abound for the term resilience, we find that the USAF definition is sufficient: “the capacity of a force to withstand attack, adapt, and generate sufficient combat power to achieve campaign objectives …despite disruption whether natural or man-made, inadvertent, or deliberate.”

Another way to put it is that a resilient military system both (1) resists any change in performance due to a disruption and (2) recovers quickly from a disruption.  These key concepts are illustrated in Figure 1, in which resistance is determined by severity of the initial dip in the performance graph and ability to recover is indicated by how performance graph increases after its minimum.

Figure 1. Network performance vs. time.

We conclude that resilience is measured relative to the system’s performance metric of interest (e.g., mission capable rate).  However, there is no single measure of resilience; it must be measured in terms of both resistance (e.g., rate of system performance post-disruption) and recovery (e.g., rate of system performance recovery).

Strategies for Adding Resiliency to the Supply Chain

As identified by research conducted by Dr. Pettit, Dr. Fiskel, and Dr Croxton, Supply chain managers and designers have many methods by which resiliency can be increased in a system:

Flexible sourcing deals with that ability to quickly change between suppliers.  In general, flexible sourcing is difficult for AF weapon systems since they are technologically complex, expensive and often produced in batches.  This can make the AF supply chain fragile, e.g., some suppliers for the F-22 supply chain shut down production before the last aircraft was even off the production line.  However, our repair networks do feature flexible sourcing, i.e., many bases have the option of both back-shop repair and Centralized Repair Facility (CRF) repair.

Capacity deals with reserve capacity or back up sources.  Of course, this is expensive due to the cost of our weapons systems, as Lieutenant General Lee K. Levy II mentions in a recent ER article.  Additionally, deciding where to locate the excess capacity presents a unique trade-off for military supply chains: in-theater capacity reduces lead times, however it puts that valuable capacity at risk!

Visibility of disruptions is the ability to quickly know that a disruption has taken place.  One goal of the Air Force Repair Network is to provide “enterprise visibility of like repair capabilities” so that throughput can be improved. One of General David L. Goldfein’s key initiatives is to have a common operational picture for the decision maker, so that we can make decisions at a speed our adversaries cannot match.

Adaptability deals with the ability of a supply chain to modify or change operations in response to some type of disruption.  This could include shifting combat sortie generation to another base or even another service while the disrupted base recovers.  Reduction of lead times has been identified as one way to add adaptability to the supply chain.

Anticipation involves the ability to forecast or discern future threats or disruptions.  Big Data and Artificial Intelligence (AI) promise to help improve risk anticipation.  In fact, former Deputy Defense Secretary Bob Work invested heavily in this effort with the stand-up of the Algorithmic Warfare Cross-Functional Team. The team was stood-up in April 2017 to leverage AI to maintain situational awareness on the battle field.

Recovery is the ability to return to normal state.  Some ways to expedite recovery are based on the ability of the service to manage a crisis, communication, and the ability to mitigate a disruption before damage becomes wide-spread.  For example, Repair Network Integration supports AF repair network recovery through routine communication between node and product repair group managers. This enables high visibility, and therefore quick resolution of repair network issues.

Dispersion involves the distribution of aircraft and repair capacity.  Here, the adage “don’t have all your eggs in one basket” applies.  For example, the attack on Marine Corps’ Harriers shows what happens when aircraft are not dispersed.  Granted, this is difficult to do in some locations.  Also, it is interesting to consider that the trend toward centralization for efficiency’s sake is at odds with dispersion.  This tradeoff is one that we seek to model to provide decision makers more accurate information as to the cost and value of efficiency versus effectiveness.

Network Design and Resilience

In an idealized setting, a network would never be subject to a disruption.  In this case, from an efficiency standpoint, the network would have a few nodes of extremely high degree (called hubs), and many nodes of very low degree (see Figure 2a).  This is called a scale free network structure (Barabási, 2016).  Additionally, in many applications hubs enjoy economies of scale and therefore are overall more efficient at performing work than other, more distributed alternatives.

The disadvantage of scale free networks is that they are susceptible to targeted disruptions (Barabási, 2016), i.e., attacks made by an intelligent adversary who has knowledge of the network’s structure.  Specifically, an adversary would target the hubs.  However, scale free networks are extremely resistant to random disruptions, i.e., disruptions that are equally likely to affect any node of the network.

On the other hand, network structures with fewer hubs, such as random networks (see Figure 2b) are more resilient to targeted attacks since very few, if any, nodes contain a disproportionate number of connections in the network.  However, these network structures are less efficient than scale free networks since they do not have the same economies of scale without hubs.



Figure 2. Network structure: (a) Scale free, (b) Random

The Resiliency Dilemma

Many supply chain scholars have concluded that a more cost-effective approach is to build a resilient supply chain rather than prepare for specific events or disruptions that have a small probability of happening.  Lt Gen Levy highlighted this conundrum in a recent ER issue when discussing the criticality of the sustainment function.

He added that the Air Force no longer has the capacity it once had to surge due to “a diminishing defense industrial base, scarcity of natural resources, and the gap in Science, Technology, Engineering, and Mathematics (STEM)-based human capital” which is both increasing risk and exposing vulnerabilities in “the logistics kill chain”.  Furthermore, current operations are “burning up weapons and ammunition at a ferocious rate, far beyond what the highly consolidated and fragile US defense industry can produce”.  Complicating this problem even more is that congress is reticent to pay for weapons that may be never used.  Lt Gen Levy ends his article with a challenge for the Air Force to “think about and make improvements to our supply chains, maintenance processes, and our ability to project requirements”.

With respect to resiliency, AFIT is looking into the following set of questions:

  • Are we resilient enough?
  • How resilient should we be?
  • How can the Air Force properly measure resiliency?
  • If one or more Air Force bases was incapacitated or destroyed, what would be the impact on our ability to conduct operations in the theater?
  • How can we design resiliency into the supply chain to better plan for disruptions that are difficult to forecast?
  • What are the tools Air Force leaders have available to build more resilient supply chains?

A PACAF Case Study

While this case study is a work in process, AFIT faculty are developing a simulation that measures resiliency and begins to answer the questions listed above.  We focus on the Pacific Theater because of its geographic size and limited number of bases.  Building resilience into such a supply chain is particularly challenging.

The Air Force repair network is a multi-echelon supply chain.  At the top-level, it consists of several Air Force Logistics Complexes (ALCs) which perform specialized and equipment intensive types of repairs/rebuilds of engines, avionics, landing gears, and other repairable aircraft components.  Some ALCs, e.g., Ogden ALC at Hill AFB, have satellite repair centers in forward operating locations such as the Supply Center Pacific (SCP) at Kadena AB, Japan.  The SCP was set up to provide faster support to bases in the geographic region for which PACAF is responsible.  Furthermore, for engine repair, there is a CRF, which acts a hub in the network.  In addition, the repair network consists of back shops that perform intermediate level repairs.

A healthy repair network is critical to sustaining operations.  However, this network is vulnerable to offensive actions by an adversary (targeted) or other disruptions such as manmade or natural disasters (random).  For example, loss of, or diminished capability at the SCP would require affected workloads to be shifted to other facilities at great cost in both time and money.  Additionally, lead times, shipping costs, and inventory costs for replacement parts would increase to maintain the required throughput of repair parts.

Furthermore, we must acknowledge that the likelihood of random and/or targeted disruptions depend on if the Pacific theater is at peace or war.  During peacetime, an efficient repair network is desirable.  However, during wartime a more robust and resilient, albeit less efficient, repair network is needed.

How do we Study and Measure Resiliency?

Simulation is a tool well suited to gain insight into how network design decisions impact resiliency.  We start by setting up a supply chain in Simio® with four bases, two centralized repair facilities, a depot, and appropriate links between them.  The simulation produces broken engines via a statistically representative process intended to capture the historical engine break rate we would expect during combat operations.  The maintenance teams install a spare engine or wait for the broken one to be repaired per the established rules for intermediate versus depot level maintenance.  When the engine is repaired, it is installed on the aircraft to make it mission capable again.  The simulation can measure the mission capable rate over time as well as the average flow time (break to installed) for an engine. A picture of the model repair network is shown in Figure 3.

Figure 3:  Example Repair Network Simulation in Simio®

The notional results show some promise in that it’s feasible, and perhaps useful with the right data and operational insight, to measure effectiveness of different supply chain designs.  The competing designs have an incremental investment value attached to them, thus a return on investment in terms of mission effectiveness metrics, could be at least be proposed given accurate cost data and mission requirements.

Initial Notional Results

The chart in Figure 4 shows a typical performance pattern in any system that becomes tight on capacity following a disruption.  Prior to the disruption, we see that there are peaks and valleys of operational capability due to normal system variation.  This is to be expected, and the fluctuations are indicative of a highly complex supply chain with high variability.  With very little buffer capacity the system cannot respond to the normal variation in engine breaks and repair times.  Then, following a period of normal activity, a disruption was simulated at Day 300.  The disruption took a repair facility offline, thus reducing sustainment capability.

As Figure 4 shows, the up cycle was cut short and the down cycle was exacerbated.  In this case the system pulled out of the nose dive but the new average is well below the normal capacity average.  The number of mission capable aircraft in the system has been reduced.  Figure 5 shows why: engine throughput drops off after the disruption.

Figure 4: Illustrative Mission Capability over Time

Figure 5: Engine Throughput in 25 Day Increments

The results above are simply estimates.  Though realistic and useful for generating insights into how Air Force repair networks function, they are not interpretable without the context of parameters like break rate, repair times, repair capacity, transit times, and other policy decisions.  With that context and proper data input, analyses could be performed to make actual decisions as to where we should add maintenance flexibility, excess capacity, adaptability, and dispersion assets. These decisions could lead to a network design that yields the greatest resiliency.  With a cost model behind each potential network design, the investments can be compared against performance.


As Lt Gen Levy stated, building a more resilient supply chain network depends on the critical thinking skills of future Air Force leaders.  With our adversaries nipping at our heels, and the sky rocketing cost of high tech weapon systems, now is the time to think critically about improving our military supply chains.  We at AFIT are looking for sponsors, data, and people passionate about logistics to help build this case study so that senior leaders can answer important strategic questions regarding building resiliency in the Pacific and across the Air Force.


Daniel W. Steeneck is an Assistant Professor of Supply Chain and Logistics in the Department of Operational Sciences at the Air Force Institute of Technology.  His current research interests are in the use of analytical techniques to solve problems encountered in Supply Chain Management. Prior to joining AFIT, Dr. Steeneck was post-doctoral research associate at MIT’s Center for Transportation & Logistics.  He completed his Ph.D. in Industrial and Systems Engineering at Virginia Tech.  His dissertation, in the area of Reverse Supply Chain Management, won second place in the Council of Supply Chain Management Professional’s dissertation award competition.  While at MIT, Dr. Steeneck’s research included topics in retail operations management and service supply chain inventory management.  His writings have been published in the Wall Street Journal, Sloan Management Review (Frontiers),  Operations Research Letters, International Journal of Production Research, International Journal of Production Economics and the Journal of the Operational Research Society.  Dr. Steeneck has also worked closely with organizations such as Procter and Gamble, OnProcess, Volvo North America, and the U.S. Navy.

Maj Timothy W. Breitbach is the Department of Operational Sciences Logistics Division Chief and an Assistant Professor of Logistics and Supply Chain Management at the Air Force Institute of Technology.  Maj Breitbach received his commission in 2005 as a graduate of the Reserve Officer Training Corps program.  He graduated from logistics officer training in 2006.  His operational assignments include all aspects of US Air Force logistics operations.  As a logistics readiness officer, he has served as a materiel management flight officer-in-charge (OIC), assistant installation deployment officer, vehicle maintenance flight commander, and logistics flight commander.  He also served as the Executive Officer to the Commander of the Defense Logistics Agency (DLA) Energy.  Additionally, Maj Breitbach was deployed as the operations officer for a truck detachment supporting convoy operations in Iraq.

 Special thanks to our sponsors: