TIRED TITANIUM: A FATIGUE‐BASED APPROACH TO AIRCRAFT INVENTORY MANAGEMENT AND ACQUISITION PLANNING

Abstract

Airframe fatigue has emerged as a primary determinant of tactical aircraft service life. To investigate the impact of various operational scenarios on airframe fatigue and aircraft stocks, we develop an econometric model of fatigue and arrest landing accumulation for US Naval aircraft. Model forecasts suggest that fatigue‐related attrition threatens to reduce inventories below the level needed to meet operational commitments before planned replacements are available. Changes to training regimes could mitigate the shortfall, but it is likely that acquisition schedules will have to be accelerated, or current service life extension programs expanded to maintain inventories in the future.

¹This research was conducted while Chad Meyerhoefer was an economist at The CNA Corporation and Robert Trost was at The CNA Corporation while on Summer Break from The George Washington University (GWU). No official endorsement by the Agency for Healthcare Research and Quality or the Department of Health and Human Services, by The CNA Corporation, or by GWU is intended or should be inferred. Programming assistance by Kletus Lawler and Geoffrey Shaw is greatly appreciated.

Keywords:

• Airframe fatigue
• Aircraft procurement
• Acquisition planning

INTRODUCTION

The United States Department of Defense (DoD) is currently in the midst of one of the most significant military aircraft planning and acquisition phases in recent years. Many of the systems acquired during the post‐Vietnam period are reaching the end of their service lives, leaving DoD with the task of replacing or refurbishing a large and diverse set of aircraft operated by the Air Force, Navy, and Marine Corps. As a result, the literature on the economics of aircraft replacement has seen an increasing number of contributions over the past several years. For example, Jondrow et al. (2002) analyze the trade‐off between new procurements and refurbishing older aircraft through a service life extension program (SLEP). They find that SLEPs are cost‐effective provided that they result in moderately lower operating and support costs. Greenfield and Persselin (2003) develop a dynamic programming model to determine the optimal replacement period for an aging fleet by comparing acquisition to operating and support costs. After extending the model to account for aircraft availability, Keating and Dixon (2004) demonstrate its use in forecasting the cost‐minimizing replacement period for the KC‐135 refueling tanker.

However, this dynamic programming model is only strictly appropriate for cargo and re‐fueling aircraft, which can be kept in flying condition over long periods without the imminent threat of airframe fatigue or technological obsolescence. In the case of tactical aircraft, these factors may limit aircraft service life long before the ratio of operating and support costs to acquisition costs suggests an optimal economic replacement time. Indeed, during the post‐Cold War era of continued air supremacy and high operational tempo, all branches of the armed forces have seen airframe fatigue, or the fatigue life expended (FLE) on critical airframe components become the single most important determinant of tactical aircraft service life (Wu et al., 1995). ²

The US Department of Navy (DoN) has developed the most sophisticated and comprehensive airframe fatigue tracking system of any service branch. Under the Navy program the decision to replace an aircraft or initiate a SLEP is based on the computation of an aircraft‐specific fatigue measurement, the FLE index, which ranges between 0 and 100% fatigue. This index is based on known engineering relationships that map information on each aircraft’s flight profile and G‐exceedence levels into a cumulative measure of fatigue for a measured location. When tactical aircraft conduct training or operational flights, the specific maneuvers pilots execute subject the airframe to stress, or G‐forces, that cause metal deformation. ³ Over long periods of time, this process can result in the initiation and growth of airframe cracks that threaten structural failures. By monitoring airframe metal fatigue using the FLE index, the Navy is able to greatly reduce the threat of structural failures. Although this index has been used to calculate FLE rates and make short‐term projections based on historical aggregate flight hour distributions, it has never been correlated to day‐to‐day operational data. Consequently, little is known about how flying different mission profiles impacts airframe fatigue.

In this paper we develop a statistical model that relates the FLE indexes of the F/A‐18 Hornet, EA‐6B Prowler, and S‐3B Viking to their operational determinants, which include the composition of mission profiles flown by the aircraft, ownership unit, and aircraft characteristics. The roles of these three aircraft vary greatly, necessitating that separate models are estimated for each. The F/A‐18 is the Navy’s primary fighter/attack aircraft and is used for a wide variety of potentially stressful missions, from air‐to‐air combat to ground attack and combat air patrols. The EA‐6B is an electronic counter‐measures aircraft that typically accompanies other tactical aircraft during ground attack missions to seek out and destroy enemy radar and air defense installations. As a result, it is often subjected to high‐speed maneuvers at low altitude. In contrast, the S‐3B is used primarily for low‐stress aerial re‐fueling and surveillance missions, although it does still perform some of the anti‐submarine and anti‐ship attack training consistent with its earlier Cold War role.

The three aircraft also have varying procurement costs, operating and support costs and service lives, which are summarized in Table I. Due to budget constraints delaying the timely acquisition of replacement aircraft, the Navy’s tactical aircraft are remaining in service longer than their original design lives, making fatigue monitoring and SLEP scheduling critical to operational readiness. Our model uses data from several Naval Air Systems Command (NAVAIR) databases, including information on planned SLEPs, to make long‐term forecasts of aircraft service through the transition to a fleet that is composed of newly procured F/A‐18 Super Hornets and the Joint Strike Fighter (JSF).

Table I Operating and Procurement Costs per Aircraft (1991–2002)*

	F/A‐18	EA‐6B	S‐3B
Operating Costs (per yr)	1.3	1.5	1.6
Fuel	0.4	0.4	0.2
Consumables	0.3	0.4	0.3
Aviation Depot‐Level Repairables	0.6	0.7	1.0
Total Operating Costs for Design Life	23	33	53
End‐of‐Run Unit Procurement Cost	29	52	27
Average Flight Hours (per yr)	332	337	392
Original Flight Hour Design Life	6,000	7,500	13,000
Source: US Navy Visibility and Management of Operating and Support Cost (VAMOSC) database (operating costs) and US Navy Fact File (procurement costs), 2005*In millions of FY 2005 US dollars

The paper is organized as follows. The next section describes the data sample used to estimate the model and provides some descriptive measures of FLE and the flight hour distributions for each type‐model‐series (TMS). This is followed by a formal presentation of the empirical model and forecasting methodology. The third section gives the estimation results and categorizes the impact of different mission assignments and other operational factors on FLE and arrested landing accumulation. The results of the statistical model are then used in the fourth section to make inventory projections under different operational scenarios. These are compared to procurement schedules for replacement aircraft and planned SLEPs to determine whether shortfalls in aircraft stocks are likely to occur. The final section of the paper concludes with a review of fatigue mitigation strategies and the implications for acquisition planning.

DESCRIPTION OF DATA

Fatigue Data

The Structural Appraisal of Fatigue Effects (SAFE) program at NAVAIR collects cumulative airframe fatigue information and computes the FLE index used in our empirical model. The FLE index is critical to projecting aircraft service life. When it reaches 100% the aircraft is typically either retired or overhauled, unless the index corresponds to a part of the airframe that can be easily monitored for crack initiation through regular inspections (such as the vertical tail of an F/A‐18). ⁴ The F/A‐18 has ten monitored locations on its airframe, but we concentrate on the wing root (WR) and vertical tail (VT) fatigue measurements because these areas accumulate FLE the quickest. In addition, the wing root is the primary measure of fatigue for the overall airframe. When WR FLE reaches 100% the aircraft must be retired or undergo a substantial overhaul, whereas tail components are more easily replaced after reaching structural limits. Both the EA‐6B and S‐3B have only one fatigue index, which is the wing center panel (WCP) for the former and (usually) the rear spar area for the latter.

In addition to data on FLE indexes, the SAFE reports also contain information on cumulative flight hours, arrested landings and catapults. There is a limit on the number of arrested landings and catapults any aircraft can sustain, much like the FLE index limit of 100%. Together, the FLE and arrested landings/catapults limits define an aircraft’s most important structural life limits (SLL), or the usage range determining its service life. Because cumulative arrested landings and catapults track each other closely, we only consider the former measure in the empirical model and service life projections. In order to make this service life measure comparable to the fatigue indexes, the cumulative number of arrested landings, or the arrested landings expended (ALE), is expressed as an index ranging between 0 and 100% of allowable arrests and treated as another FLE index.

FLE and ALE Distributions

The distribution of WR FLE for F/A‐18 aircraft contained in the last year of our data sample (from the 3rd quarter of 2001 to the 3rd quarter of 2002) is given in Figure 1. The distribution is centered at 35% of expended airframe fatigue, indicating that the majority of aircraft have not reached the point where measures must be taken to mitigate high fatigue levels. However, the distribution has a long right tail and there are several aircraft that have WR FLE levels above 80 and even 100%. These aircraft are either scheduled for, or are in the process of receiving fuselage center barrel replacements (denoted CBR+), a SLEP that mitigates both WR fatigue and ALE.

Figure 1 Cumulative F/A‐18 WR FLE (as of Q3 2002)

In general, aircraft of different series are concentrated at different parts of the distribution. F/A‐18D aircraft are typically land‐based – an environment characterized by higher WR fatigue than deployments aboard aircraft carriers. Therefore, they are concentrated in the right tail of the distribution, as are the older F/A‐18A and F/A‐18B flown in training squadrons. The left‐hand side of the distribution is accordingly composed primarily of C series aircraft found in most deployable squadrons.

In comparison to cumulative WR fatigue levels, VT FLE is significantly higher. Nevertheless, the replacement cost of a vertical tail is minor in comparison to the $2 million price tag on a CBR+, so VT FLE is rarely a determining factor in the decision to retire an aircraft. For this reason, we base inventory projections only on WR FLE and ALE. Cumulative levels of the latter index vary greatly by aircraft series, with a high number of arrests on the F/A‐18C due to consistent deployment, and lower ALE for those aircraft typically operating from land bases, such as the F/A‐18A/B/D. The population of F/A‐18C aircraft shown in Figure 2 has an ALE distribution that is fairly symmetric and centered at 65% (excluding the spike at 5% associated with aircraft in non‐deployable squadrons). When this distribution in compared to WR FLE levels in Figure 1, it is clear that many F/A‐18C will reach their ALE limit first, necessitating measures such as the fuselage center barrel replacement or accelerated acquisition be taken to mitigate ALE.

Figure 2 Cumulative F/A‐18C ALE (as of Q3 2002).

In contrast to series C aircraft, the distribution of expended arrested landings for F/A‐18 A/B/D in Figure 3 is concentrated between 0 and 25%, with a long right tail composed almost exclusively of series A aircraft that still deploy with frequency. Although many F/A‐18A/B/D have low ALE, they tend to be the aircraft with the highest wing root FLE, and are not necessarily expected to last much longer than the F/A‐18C under current operating conditions.

Figure 3 Cumulative F/A‐18A/B/D ALE (as of Q3 2002)

The EA‐6B Prowler data sample is slightly more recent than the F/A‐18, reaching through the fourth quarter of 2002, but contains a much smaller number of aircraft. The distribution of wing center panel (WCP) FLE is bi‐modal, with a cluster of low fatigue aircraft centered around 15–20% fatigue, and a separate group of high fatigue aircraft centered at 80% FLE. Many of the aircraft at lower fatigue levels have undergone a SLEP to replace their original WCPs with newer 7050 aluminum wings, while those with high fatigue levels exceeding the structural life limit are currently being “re‐winged”. These SLEPs cost approximately $5 million per aircraft and must continue through 2008 in order to maintain inventories. The distribution of ALE for the EA‐6B indicates the majority of aircraft have significantly lower ALE than WCP FLE levels, suggesting the latter is the greatest threat to inventory sustainability. The S‐3B FLE distribution is more or less symmetric around 40% of expended fatigue, while the ALE distribution is far to the right and centered around 80%. Therefore, arrested landing accumulation is currently a more pressing fatigue problem than FLE for the S‐3B.

Operational and Logistical Data

In order to explain changes in FLE and ALE indexes we compiled data on the operational, logistical, and engineering determinants of airframe fatigue and arrested landing accumulation. Occasional performance upgrades meant to adapt aircraft to evolving threats are one example of an engineering change impacting fatigue. While minor upgrades may not substantially alter airframe life, major upgrades to engines, avionics, or weapons systems can have a significant effect on both the types of missions the aircraft is selected to perform, and their associated stress levels. Because initial service life projections do not take these upgrades into account, manufacturer estimates of aircraft design life can be overly optimistic. Therefore, our models include indicator variables for the F/A‐18C/D Enhanced Performance Engine (EPE) Upgrade and the F/A‐18A avionics upgrade. We also obtained data on the age of each aircraft in order to proxy for a variety of factors related to repetitive metal deformation and corrosion, which influence the level of airframe fatigue expended over time.

The bulk of the operational data used to estimate the model was obtained from the Naval Aviation Maintenance and Material Management (AV‐3M) System. This database contains extensive operational data on individual aircraft, including daily flight records, squadron information and landing records. We merged the AV‐3M data with information on reported fatigue levels from the SAFE database in order to estimate the empirical model. Table II lists the characteristics of the joint AV‐3M/SAFE database. The record of historical SAFE data collected under a consistent methodology limits the length of each sample. This length is smaller for the EA‐6B than the other two TMS.

Table II AV‐3M / SAFE Data Sample

	F/A‐18	EA‐6B	S‐3B
Custody
Navy – Active Duty	347	61	94
Navy – Reserve	88	4	–
Marine – Active Duty	204	28	–
Marine – Reserve	71	–	–
Training Squadron	228	36	22
Blue Angels	28	–	–
Type‐Model Series (Average Age in 2002 in Parenthesis)
F/A‐18A (17.0 yrs)	313	–	–
F/A‐18B (18.1 yrs)	37	–	–
F/A‐18C (11.2 yrs)	451	–	–
F/A‐18D (10.5 yrs)	160	–	–
EA‐6B (20.8 yrs)	–	129	–
S‐3B (27.2 yrs)	–	–	116
Total in Sample	961	129	116
Average No. Reporting per Year	715	89	77
Total Flight Hour Proportions
Training	0.81	0.76	0.62
Support	0.12	0.10	0.30
Operations	0.01	0.01	0.05
Contigency	0.02	0.02	0.00
Combat	0.04	0.12	0.03
Data Range	1/91–9/02	5/97–12/02	1/93–9/02

Nearly all operational aircraft in the DoN inventory over the respective data collection periods are included in the sample, though they may not necessarily have reported data every month or quarter due to lapses associated with depot maintenance or certain deployments. In order to investigate the impact of different operational scenarios on airframe fatigue, historical flight records contained in the AV‐3M data must be aggregated into categories representing different types of operational missions. In general, there are five major categories of flying denoted by the first digit of the total mission requirement (TMR) code. These are: training, support and logistics flights, operations flights conducted in support of tasking not designated as contingency operations, contingency flights, and combat.

The proportion of flight hours in each general mission category by TMS is also reported in Table II. Clearly, training missions make up the majority of flight hours, although there is some variation in the levels across platforms. The S‐3B carries out proportionally fewer training flights and logs nearly three times as many support flights as the other TMS, primarily due to its aerial refueling role. The EA‐6B Prowler has more associated combat hours, reflecting its status as the only electronic attack aircraft available to all branches of the armed forces. As a result, the demand for Prowlers in combat situations is higher than the F/A‐18 and S‐3B.

The 1991–2002 sample period encompasses two large combat operations as well as a variety of smaller conflicts. The number of combat flight hours can be associated with each operation through launch dates from flight records. For the F/A‐18, approximately 21,981 combat hours were flown during Operation Desert Storm (First Gulf War), 38,842 hours were flown during Operation Enduring Freedom (Afghanistan), and 48,309 combat hours correspond to smaller conflicts. Flight hours logged during smaller conflicts appear to reflect Iraq no‐fly zone patrols, and operations in Kosovo, Somalia, Bosnia, and Haiti. ⁵ Although breaking total flight hours into the five categories shown in Table II provides a useful snapshot of historical operational activity, there exists much heterogeneity within these categories with regard to the level of stress placed on the airframe and measured by the FLE index for any given flight. In order to account for this heterogeneity in the statistical model, we further disaggregate total flight hours into between 13 (for the S‐3B) and 18 (for the F/A‐18) different mission categories having similar stress profiles using the TMR codes.

The AV‐3M/SAFE database also contains information on landing records for all types of landings, including arrested landings, airfield landings, and various types of “touch and go” practice landings. Aside from capturing the stress of landings on the airframe, the inclusion of cumulative landing counts is perhaps most important as a proxy for operational environment, and in particular, whether the aircraft was sea‐based or shore‐based during a respective period. Furthermore, counts of different landing categories can be used to account for variation in the fatigue characteristics of missions flown from ships as opposed to land bases that are not captured by TMR classifications.

STATISTICAL SPECIFICATION AND ESTIMATION

Our statistical model of airframe life is specified as a recursive system of two equations, in which the arrested landing index and critical fatigue index of each TMS is related to the operational determinants described above. The recursive specification arises because the change in the FLE index is also a function of the number of accumulated arrested landings, or equivalently, the change in the ALE index over the same period. ⁶

The two‐equation system is defined as:

(1)

(2)

for time periods t = 2,…, T and aircraft i = 1,…, N. The dependent variable in each equation represents the change in the respective index over a given time period, such that in equation (1) ΔFLE_it = FLE_it − FLE _it−1 and in equation (2) ΔALE_it = ALE_it − ALE _it−1. The vector X_it = [X _1,it ,…, X_K,it ]′ represents observable flight hour levels corresponding to K different mission classifications for aircraft i flown between time periods t − 1 and t, and has associated parameter vector β′ = [β_i ,…, β_K ]. Likewise, L_it is a vector of landing variables containing the total number of landings of different types accumulated between periods t − 1 and t, A_it is a vector containing the age and (sometimes) age squared of each aircraft in the respective period, and D_it is a vector of operational control variables, including branch of service, squadron group indicators such as Atlantic and Pacific Fleet, Navy and Marine Corps, performance upgrade indicators, and in the case of the EA‐6B a variable denoting whether the aircraft was subject to a 3‐G maneuver limitation.

Each of the equations in the system contains two intercept terms, one that varies over aircraft and is constant over time (µ_i or α_i ), and another varying over time but not aircraft (κ_t or λ_t ). This two‐way fixed effects specification (Greene, 2000) allows us to control for unobservable aircraft‐specific effects, such as configurational differences or reliability attributes that impact FLE and ALE, as well as unobservable time specific effects, such as minor changes in FLE measurement that are constant across aircraft. Finally, the error terms represent unobservable random shocks to the system and are assumed to be distributed as follows: and

While the process generating changes in the FLE index is based on the relationship between flight hours and metal deformation of the airframe, the process generating changes in ALE is based on the operational environment and launch and recovery schedules associated with different types of flying. Therefore, we can reasonably assume u_it and ϵ_it are uncorrelated and consistently estimate equations (1) and (2) separately using ordinary least squares. The FLE equation is estimated on the entire sample of aircraft in all time periods, but the ALE equation is only applicable while aircraft are deployed aboard ships. Hence, it is estimated on a reduced sample of ship‐based aircraft, determined by taking the ratio of arrested landings, ship touch‐and‐gos and bolters to field landings and field touch‐and‐gos in each time period. If this ratio exceeds 25% the aircraft are included in the ship‐based sample.

The estimation procedure outlined above is adapted slightly in order to model FLE associated with the F/A‐18’s vertical tail. Although there are five separate fatigue indexes corresponding to different zones on the vertical tail, we wish to determine the impact of missions on overall tail fatigue. This is done by jointly estimating five VT FLE equations, each specified as in equation (1), while imposing the restriction that the parameters of all five equations are the same. Allowing the errors terms to be correlated across equations, the five‐equation system is estimated using seemingly unrelated regressions.

Forecasting Procedure

Equations (1) and (2) can be used to project future FLE and ALE levels and to simulate the impact of changing operational requirements on aircraft inventories. By replacing the population parameters in each equation with their consistently estimated counterparts, the following forecasting equations are defined:

(3)

(4)

where is predicted from equation (4). ⁷ If future mission assignments and deployment schedules are assumed to follow historical patterns, the form of the forecasting equations can be modified to reflect this assumption. Let denote the average proportion of flight hours in each mission category, and denote average proportion of non‐arrested landings in each category. If X_TOT,t and L_TOT,t are the total number of flight hours and non‐arrested landings in a given time period then equations (3) and (4) can be equivalently written as:

(3b)

(4b)

The right hand side of equation (4b) must be scaled down by two fifths in comparison to equation (4) to reflect the average amount of time aircraft spend deployed and training aboard ships. ⁸ When the total flight hours and non‐arrested landing levels that the aircraft are expected to log in the future are be inserted into equations (3b) and (4b), the model predicts FLE and ALE given historical mission emphasis, squadron assignments, and associated aircraft age growth. If one expects the mission mix to change in the future, the proportion variables can be changed accordingly so the forecast will reflect the new operational situation.

In the case of the F/A‐18, the forecasting methodology described above requires additional refinement to capture the broad operational environment and specific fleet management techniques unique to this platform. Otherwise, the model may underestimate actual operational lives. In general, the F/A‐18 fleet is composed of mostly A, B and D aircraft that are used almost exclusively for training or expeditionary operations and rarely deploy aboard carriers, and series C aircraft that deploy with frequency. While high levels of FLE and very low ALE characterize the former land‐based training environment, the opposite is true of shipboard operations. Therefore, it is not uncommon to find F/A‐18A aircraft that are near their FLE limit, but far from their ALE limit, while F/A‐18C are close to reaching 100% ALE, but have much lower FLE levels.

In addition to the general categorization of land‐based and ship‐based aircraft, there is also a distinction between Navy and Marine Corps aircraft. ⁹ While Marine Corps squadrons that deploy with the Navy operate much like other deployable aircraft, land‐based Marine Corps aircraft fly a very different mix of missions than the land‐based training aircraft used by the Navy. In particular, they are frequently engaged in expeditionary operational and combat missions, a mission mix that also leads to fatigue indexes.

Maximizing the operating life of each aircraft requires that it reach its FLE and ALE SLL at roughly the same time. In order to mitigate the current misalignment of SLLs for most F/A‐18s and extend the life of the fleet, NAVAIR is instituting squadron rotations for some aircraft. F/A‐18As currently operating from land bases, which have low ALE levels, are given a weapons system and avionics upgrade and sent to deployable squadrons to replace F/A‐18Cs with high ALE levels, but relatively low WR FLE. The F/A‐18Cs are then assigned to land‐based squadrons where they can continue to expend WR fatigue without reaching the arrested landing SLL.

The basic forecasting equations in (3b) and (4b) can be modified as follows to capture the above characteristics of F/A‐18 fleet operations and management:

(5)

(6)

(7)

(8)

This system is written in more general notation, with differences in flight hour categories and flight hour and landing proportions noted. In equations (5)–(8) the population of F/A‐18s is divided into two groups, the first being deployable aircraft and the second land‐based training aircraft. During the forecasting simulation, those in the former group cycle through equations (5) and (6) until their ALE reaches 95%, at which point they are switched with training aircraft in the latter group having relatively low ALE in comparison to FLE. Aircraft are swapped in this manner until the entire population of training aircraft is exhausted, and then remain in their current groups until either WR FLE or ALE index reaches 100%. ¹⁰

DISCUSSION OF RESULTS

Parameter estimates from the statistical model defined by equations (1) and (2) provide a means of comparing the relative stress of each mission assignment in terms of FLE and ALE, net of other operational measures, operator, and aircraft characteristics. Table III contains the coefficient estimates for each flight hour category from the F/A‐18 FLE and ALE equations, re‐scaled so they represent the percentage change in the index per 1000 mission‐specific flight hours. For example, an additional 1000 flight hours of air‐to‐air combat training causes the F/A‐18 wing root (WR) index to increase by 8.5 percentage points, the vertical tail (VT) index to increase by 7.4 percentage points, and the arrested landings fatigue index (ALE) to increase by 26.5 percent, other things being equal. These estimates are of great use to aircraft managers in that they convey the fatigue cost of planned training flights and expected operational commitments.

Table III F/A‐18 Fatigue Indexes and Flight Hour Proportions*

Flight Hour Category	WR FLE	VT FLE	Total FH	ALE	On‐Ship FH
Training Flights
Fundamentals, Instrument	6.4	3.7	16.3	−13.9	7.7
Field‐Carrier Landing Practice	4.9	6.3	4.2	−5.3	5.6
Carrier Qualification	8.9	7.1	2.5	63.9	5.6
Air‐to‐Air	8.5	7.4	35.6	26.5	33.1
Attack (Air‐to‐Ground)	6.2	6.0	22.4	19.0	25.4
Other, Special Equipment	2.3^ns	9.1	0.5	32.8	0.6
Operational/Contingency and Combat Flights
Attack (Target Pre‐assigned)	7.7	7.4	1.2	16.7	3.3
Attack (CAS, Armed Recon.)	4.6	3.2	2.1	14.9	4.5
Attack (SAM / Flak Suppr.)	10.9	10.6	0.5	54.1	1.1
Escort	2.1	1.2^ns	0.4	22.2	1.0
CAP, Anti‐Air, Recon.	6.3	5.8	2.1	22.7	5.3
Intercept	23.4	36.2	0.0	−1.6	0.0
Support Flights
Ferry Flights	6.8	6.7	1.6	−18.9	0.8
Simulated Bogey	7.9	4.8	5.1	24.8	2.1
Check Flts, Experimental, Eval	5.6	4.5	4.0	20.6	3.0
Administrative Support	9.0	8.7	0.4	−5.4^ns	0.1
Non‐combat Patrol, SAR	10.0	1.2^ns	0.6	34.3	0.4
Air Shows	13.9	6.2	0.2	—	—
Model Adj. R‐square	0.32	0.33**	—	0.62	—
*Change in fatigue indexes per 1000 flight hours is reported with % of total or on‐ship flight hours in each TMR category.**Average over five VT FLE equations.^nsNot statistically significant at either the 5% or 1% level (all other effects are significant).

The percentage of total and shipboard flight hours attributable to each mission category is also reported to provide an indication of the relative importance of WR and VT FLE and ALE effects for policy and planning. Certain mission categories may be associated with very high levels of FLE or ALE, but represent only a small proportion of total flight hours, making them relatively unimportant from a fatigue management standpoint. However, if a mission category has a large fatigue impact and accounts for a substantial share of flight hours, planners may want to consider whether the fatigue costs of these missions are justified by their benefits.

Since nearly all of the fatigue effects are significant at the 1% level, we note with a (ns) cases of statistically insignificant estimates. ¹¹ Among training flights, some of the most damaging types of flying for the WR and VT are associated with very small flight hour categories, such as carrier qualification and other/special equipment. However, the largest training category accounting for 35% of the total flight hours, air‐to‐air combat, is also associated with high levels of WR and VT FLE. Given the importance of the former as a SLL, there is an implied trade‐off between maintenance of air‐to‐air readiness and dogfighting skills and other types of F/A‐18 mission training. For example, attack, the second largest training category, has an associated WR FLE that is over two percentage points lower than air‐to‐air training. From an operational standpoint, attack missions accounted for approximately 58% of combat flight hours between 1991 and 2002, whereas air‐to‐air engagements represented less than 1% of total combat hours. Despite the fact that some skills are transferable across a variety of mission profiles, this suggests current training and readiness requirements are designed to provide high levels of readiness against low probability, but potentially catastrophic, airborne attacks, and place less emphasis on developing skills for more typical air‐to‐ground combat operations.

The estimates reveal the training mission associated with by far the highest level of ALE is carrier qualification, followed by other training. ¹² On‐ship flight hours allocated to this type of training are relatively small, but the impact of carrier qualification on ALE is large enough to have a significant effect on aircraft service life. Although the gap between shipboard flight hours in attack and air‐to‐air training is smaller than for land‐based training, air‐to‐air still accounts for the greatest proportion of on‐ship hours. In addition, air‐to‐air training missions appear to be shorter and more frequent, suggesting a shift in training hours away from air‐to‐air would reduce both WR FLE and ALE, resulting in longer aircraft service lives.

Converting the parameter estimates in Table III to elasticities helps provide an indication of how much longer aircraft could remain in service if, for example, air‐to‐air training hours were reduced by 50% and redistributed to attack training, holding total training flight hours constant. The parameter estimates imply a WR FLE elasticity with respect to air‐to‐air and attack training of 0.37 and 0.17, respectively. Therefore, the aforementioned policy change results in an increase in wing root life (reduction in WR FLE) of approximately 5%. Assuming that the typical aircraft service life is 20 years and aircraft attrition is due solely to WR FLE, this translates into an additional year of life for each aircraft. Thus, 665 aircraft could provide as much service life under the new training regime with less air‐to‐air training as the current stock of 700 aircraft under the old training regime, saving at least $1 billion in procurement costs and $805 million in operating and support costs over the design life of the aircraft.

In contrast to the high flight hour levels of training missions, operational/contingency, and combat flight hours represent only 6.3% of total flight hours, so their fatigue impact is comparatively less important. The WR FLE impact of attack flights varies depending on the particular mission profile with an overall weighted average impact of 6.4%, which is slightly higher than the 6.2% level associated with attack training. This similarity in mission stress between combat and training, however, does not hold for air‐to‐air category, with combat intercepts proving to be much more stressful than air‐to‐air training. Surface‐to‐air missile (SAM) and flak suppression missions result in high levels of ALE, but historically account for only 1% of on‐ship flight hours. Nonetheless, ALE from these missions could become significant under sustained combat operations.

The analysis of fatigue effects associated with F/A‐18 support flights leads to some interesting findings. Even after controlling for the effect of squadron assignment to the Blue Angels, the Navy’s flight demonstration team, participation in air shows leads to high FLE. Based on AV‐3M data, we find that the average yearly cost per aircraft (b/w 1992 and 2002) of depot‐level repairs for an F/A‐18A assigned to the Blue Angels is $198,000, which is 58% less than the comparable cost for an active duty F/A‐18A. Both the large fatigue impact of air shows and relatively low depot support cost of the Blue Angels suggests that the Navy may want to re‐consider the inclusion of active duty aircraft in air shows in addition to the Angels. Although the high fatigue effects from air shows are not particularly surprising, administrative support and ferry flights seems to be associated with higher fatigue levels than one might anticipate a priori, considering their primary purpose is to move people or aircraft from one location to another. In fact, administrative support flights are comparable in airframe stress to air‐to‐air training, while ferry flights are slightly more damaging than fundamentals and instrument training.

Predicting the impact of different types of flying on airframe fatigue indexes is particularly important for the EA‐6B community, which has experienced persistent fatigue problems. EA‐6B airframes are older than typical F/A‐18s and earlier aircraft were manufactured with 7079 type aluminum, which demonstrated poor stress corrosion characteristics. As a result, a fatigue management program has been in place since the late 1990s whereby all aircraft are issued a quota of 4G “equivalent” hits ¹³ to be expended on an annual basis, while high wing center panel (WCP) FLE aircraft are subject to a 3G operational restriction. Although it took several years for pilots to adjust to the new restrictions and incorporate them into their flying behavior, there are signs the fatigue management program is successfully extending the life of Prowler airframes.

Because the EA‐6B’s mission is more narrowly defined than that of the F/A‐18, fewer mission categories are included in the model results presented in Table IV. In addition, we found large statistical differences in the fatigue impacts of some training missions in the 1997 to 2001 period when compared with 2002, so the percentage change in WCP FLE per 1000 flight hours is reported separately for the latter year. The time period differences are rather dramatic, with fatigue impacts dropping to less that half their earlier levels in 2002. For example, 1000 flight hours of carrier qualification training in the earlier period lead to an increase in WCP fatigue of 32 percentage points, whereas the fatigue impact was only 9 percentage points in 2002. This clearly demonstrates the greater awareness by pilots of the necessity to mitigate WCP FLE and reduce in‐mission stress in order to preserve aircraft stocks.

Table IV EA‐6B Fatigue Indexes and Flight Hour Proportions*

Flight Hour Category	WR FLE**	Total FH	ALE	On‐Ship FH
Training Flights
Fundamentals, Instrument	7.0	49.5	2.0	41.4
Carrier Qualification	31.5 / 9.0	2.6	0.6	5.0
Air‐to‐Air	56.7 / 21.0	1.4	7.4	1.7
ECM Equipment	10.0 / 4.4	11.0	2.5	18.9
Other (Attack, FCLP)	0.0	5.4	0.2	6.7
Operational, Contingency, and Combat Flights
Attack (Target Pre‐assigned)	0.8	0.2	—	—
Attack (SAM / Flak Suppr.)	2.8	10.7	1.3	13.0
Attack (ECM Support)	2.6	7.8	2.3	4.5
ECM / Photo Recon.	14.5	0.9	2.6	0.7
Support Flights
All Support	4.3	10.3	1.6	7.5
Model Adj. R‐square	0.43	—	0.45	—
*Change in fatigue indexes per 1000 flight hours is reported with % of total or on‐ship flight hours in each TMR category.**Left index change is for the 1997–2001 period, while right change is for 2002.

In addition to reductions in fatigue within particular mission categories, we also found statistically significant decreases in WCP FLE for aircraft subject to 3G restrictions. These aircraft accumulated 0.11 percentage points less WCP FLE per quarter than aircraft not 3G limited. Because this effect has been netted out in the above results, the fatigue estimates imply that changes in pilot behavior have lead to both lower baseline WCP FLE for 3G limited aircraft and recent reductions in mission stress for all aircraft.

The empirical distributions of FLE and ALE for the S‐3B suggest arrested landing accumulation is a far greater threat to inventory sustainability than FLE. Therefore, our empirical analysis concentrates on modeling the dynamics underlying ALE and assumes FLE constraints are unlikely to bind before the S‐3B squadrons are replaced. Aerial refueling has become the S‐3B’s primary mission over the past 12 years and the results in Table V bear this out. Non‐combat and operational/combat refueling accounts for the largest share of non‐training flight hours by a heavy margin and leads to high levels of ALE. Because there must be an S‐3B in the air at all times when aircraft are recovering, aerial re‐fueling missions are frequent but not necessarily long in duration.

Table V S‐3B Fatigue Index and Flight Hour Proportions*

Flight Hour Category	ALE	On‐Ship FH
Training Flights
Fundamentals, Instrument, FCLP	0.0	26.1
Carrier Qualification	47.8	4.6
Air‐to‐Air	22.6	0.6
Attack (Air‐to‐Ground)	9.4	11.4
Anti‐Submarine Warfare (ASW)	5.8	9.2
Other, Special Equipment	13.0	1.1
Operational, Contingency, and Combat Flights
Attack (CAS, Armed Recon, Other)	20.3	2.0
Aerial Refueling	18.3	4.0
Escort, ASW, CAS, Anti‐Air	9.4	0.5
Reconnaisance	11.7	3.9
Support Flights
Non‐combat Aerial Refueling	20.8	24.4
Non‐combat Patrol, SAR	8.2	3.4
Model Adj. R‐square	0.63	—
*Change in ALE index per 1000 flight hours is reported with % of on‐ship flight hours in each TMR category.

The high level of ALE exhibited by the S‐3B for aerial re‐fueling has important implications for future acquisition planning and fleet management. Rather than replacing the S‐3B with a dedicated tanking asset, the DoN plans to add the refueling mission to F/A‐18’s portfolio and have it performed by newly acquired F/A‐18 Super Hornets. Our results suggest that doing so will lead to large increases in ALE for these aircraft and exacerbate the problem of disproportionately high levels of ALE for frequently deployed aircraft. Therefore, in the future, care must be taken to ensure that F/A‐18s performing the tanking mission are rotated out to land‐based squadrons so they do not accumulate ALE too quickly and risk reaching their SLL prematurely. Since the F/A‐18Es used for tanking will have the same performance and design characteristics as other series E aircraft, the logistical cost of rotating them is negligible, while the saving in airframe fatigue could be substantial.

Inventory Forecasts

In addition to determining the impact of mission‐specific flight hours and other exogenous factors on FLE and ALE, the statistical model can be used to forecast future levels of these fatigue indexes for individual aircraft. The model is able to predict when an aircraft will reach its SLL and which SLL (100% FLE or 100% ALE) will be the binding determinant of the aircraft’s service life. Using this information, we are able to construct overall inventory profiles resulting from different assumptions about future operational requirements. ¹⁴

In most cases, the Navy and Marine Corps cannot maintain current aircraft stocks without prolonging the life of the fleet through service life extension programs (SLEPs), such as the $2 million per aircraft fuselage center barrel replacement program (CBR+), and other fatigue mitigation strategies. Our projections can be used to determine whether these strategies are sufficient to sustain aircraft stocks if future operational tempo (optempo) matches historical levels. Since maintaining inventories is even more difficult under higher optempo, the impact of prolonged future combat operations on F/A‐18 stocks is also investigated.

Figure 4 shows the F/A‐18 inventory projection resulting from equations (5)–(8) based on historical flight patterns without any intervention by aircraft managers to mitigate airframe fatigue or extend aircraft service lives. Beginning in 2007, the total inventory level falls short of the required number of approximately 722 aircraft denoted by the dotted horizontal line. ¹⁵ Clearly, the rate of F/A‐18E/F and JSF acquisitions is too slow to keep pace with the fatigue‐induced retirement of existing aircraft. Aircraft in deployable squadrons are retired from service based on either ALE or WR FLE (FLE Ship), while those in land‐based squadrons necessarily reach their WR FLE limit (FLE Land) and must be grounded.

Figure 4 F/A‐18 inventory projection based on historical usage patterns

Attrition in deployable squadrons is due exclusively to ALE until 2007, and it is not until 2010 that FLE begins to surpass ALE as the leading cause of deployable aircraft retirement. Land‐based squadrons maintain the majority of their inventories until 2007, after which inventories begin to decline quickly, causing a substantial drop in available aircraft between 2008 and 2010. Although F/A‐18E/F acquisitions begin to accumulate immediately, a large proportion of total procurements must be used for the aerial refueling mission and to replace F‐14 squadrons, leading to a significant shortfall of aircraft until JSF acquisitions begin to re‐build inventories. Without any intervention by NAVAIR, the inventory deficit exceeds 400 aircraft in 2014 if aircraft are flown in a manner consistent with historical mission assignments.

Due to the large and somewhat anticipated shortfall of aircraft between 2008 and 2023, the DoN has funded a substantial number of F/A‐18 fuselage center barrel replacements (CBR+) to extend the life of the current fleet. In particular, 192 CBR+ have been funded through 2009 at a total cost of $384 million, and another 158 have been planned between 2010 and 2013 at an approximate cost of $316 million. The CBR+ service life extension program gives an aircraft 22% more WR FLE and 35% additional ALE before structural limits are reached. ¹⁶ Figure 5 shows inventory projections after accounting for scheduled and planned CBR+. Although the current fleet lasts much longer than before, a large shortage of aircraft remains between 2008 and 2023. For example, the deficit in 2011 drops from approximately 273 in Figure 5 to 109 aircraft after accounting for CBR+. The maximum shortfall occurs in the last quarter of 2015, but is nearly as critical as before at 364 total aircraft.

Figure 5 F/A‐18 inventory projection based on historical usage patterns with CBR+

Whereas 67 deployable aircraft attrite from ALE in Figure 5, only 21 reach the ALE structural life limit (SLL) in Figure 6. Many of the aircraft closest to their SLL at the beginning on the projection are F/A‐18C with high ALE indexes. These are among the majority of F/A‐18 to receive the CBR+, and since the SLEP gives aircraft proportionally more ALE than FLE, a large percentage subsequently retire from WR FLE, but at a much later date than previously. The CBR+ also extends the life of land‐based F/A‐18D, leading to higher land‐based inventories between 2008 and 2012. Nonetheless, our forecasts predict the number of scheduled CBR+ is not sufficient to sustain inventory levels through the Super Hornet/JSF transition. However, NAVAIR plans to swap high WR FLE and low ALE aircraft from a land‐based squadron with high ALE but low WR FLE aircraft in deployable squadrons in order to maximize their total fatigue expended before retirement or overhaul.

Figure 6 F/A‐18 inventory projection based on historical usage patterns with CBR+ and squadron rotation

This squadron rotation strategy has the potential to increase service life, but it is not costless. The reason why many series A and older series C aircraft are land‐based is because they lack the “smart‐bomb” capabilities and more advanced avionics characteristic of F/A‐18C in deployable squadrons. In order to make them deployable, they must undergo costly upgrades to their avionics and weapons systems. ¹⁷ After the upgrades have taken place, deployable aircraft can be swapped with land‐based aircraft, producing the inventory profile in Figure 6.

Since most of the aircraft with high initial ALE receive the CBR+, only 21 F/A‐18C are left that retire from ALE. These aircraft can be rotated into land‐based squadrons and 21 relatively low WR FLE land‐based F/A‐18A/C placed in deployable squadrons. However, the overall effect on inventories is minimal, although squadron rotations do eliminate the ALE SLL as a service life constraint and extend the life of a few aircraft. Nonetheless, after the CBR+ has been taken into account, the number of available aircraft to rotate is too small to have any significant impact on attrition rates. ¹⁸

Over the past decade there has been an increasing trend in the number and duration of combat operations involving Naval aircraft. Such mission assignments generally result in an increase in flying hours and airframe stress levels that clearly divert from typical peacetime operations. Increased combat operations can have a significant impact on airframe life and must be taken into account by managers to avoid future inventory shortfalls. In order to predict the effect of sustained combat on inventories we use actual operational data from Operation Iraqi Freedom (OIF, Second Gulf War) to determine the effect of engaging the fleet in a different number of conflict years. ¹⁹

The effect on F/A‐18 inventories of 2, 4 and 6 years of sustained combat is reported in Figure 7. Clearly, the higher optempo and stressful missions associated with combat operations lead to shortened aircraft service lives, although inventory depletions appear to be relatively minor under shorter conflict periods. For example, the impact of an additional 2 conflict years on inventories in 2012 is only 20 fewer aircraft, with the shortfall increasing to 44 aircraft under 4 conflict years and 72 aircraft under 6 years of sustained combat. ²⁰ Therefore, the inventory impacts of larger conflicts can be significant and lead to large shortfalls, but shorter operations such as Operation Enduring Freedom (OEF) and OIF do not pose a significant threat to inventory sustainability, provided they do not occur with great frequency.

Figure 7 F/A‐18 inventories in a high‐conflict environment

The statistical model of WCP FLE and ALE defined by equations (3b) and (4b) is used to make inventory forecasts for the EA‐6B similar to those presented above for the F/A‐18. The model simulations reveal that all aircraft eventually retire from WCP FLE, implying the ALE SLL is essentially a non‐binding constraint for the EA‐6B. The projections are based on a starting inventory of 110 aircraft, of which 90 are allocated to the Navy and 20 to the Marine Corps. Since the Marines have made no plans to procure a replacement for their aircraft, we assume they will retain the 20 longest living EA‐6Bs, with the other 90 aircraft slated for replacement by the EA‐18G.

Our simulation predicts that without any intervention by NAVAIR nearly the entire stock of EA‐6B aircraft will reach their WCP FLE SLL before replacement EA‐18G acquisitions enter service in 2006. Of course, airframe fatigue is a well‐known problem in the EA‐6B community and in addition to previously mentioned fatigue mitigation strategies, the DoN has funded 90 WCP replacements valued at approximately $450 million, which effectively set the FLE index on this critical component back to zero. The additional service life gained by “re‐winging” 90 aircraft successfully extends the life of the fleet and avoids a significant inventory shortfall. However, our forecasts still indicate a small drop in EA‐6B stocks below 90 aircraft at several points during the EA‐18G transition. In order to avoid these shortfalls, eight additional re‐wings are necessary, four of which must occur during 2004, followed by three in 2005 and one in 2006. The total cost of the additional re‐wings is approximately $40 million. Re‐winging also extends the life of the 20 longest living aircraft, so the Marine Corps should be able to keep their EA‐6B fleet operational for an additional two years, before aircraft start retiring in 2011. ²¹

SUMMARY AND CONCLUSIONS

In this study, a statistical model of airframe life is developed and estimated using several operational databases maintained by NAVAIR on F/A‐18, EA‐6B, and S‐3B aircraft. Model estimates are used to derive the impact on both FLE and ALE of flight hours in a variety of mission areas; encompassing training, support flights, and combat operations. Results for the F/A‐18 suggest an increase in attack training at the expense of air‐to‐air training, if operationally feasible, would reduce the rate of aircraft fatigue as measured by both indexes. Over the long run, this would allow the DoN to procure fewer aircraft while meeting the same operational needs, provided that alternatives to in‐flight air‐to‐air training could be found so as to maintain current levels of proficiency. Possible training alternatives include the judicious use of simulators to make time spent in the cockpit more productive, or the specialization of pilots in different mission areas.

The EA‐6B community has undergone a similar re‐evaluation of training requirements in recent years in order to extend the life of their airframes to the point when a new EA‐18G electronic attack aircraft is brought online. In addition to educating pilots about the damaging effects of high G maneuvers, managers have also allocated a limited number of 4+ G‐hits to each aircraft on an annual basis. Our EA‐6B WCP FLE model reveals that the community’s fatigue mitigation policies have been effective in recent years, with the fatigue impact of training flights decreasing significantly in 2002 in comparison to the 1997–2001 period.

In addition to quantifying the impact of different mission assignments and other factors on FLE and ALE, the statistical models are used to make forecasts of future aircraft inventory levels to determine whether stocks can be maintained through Naval Aviation’s transition to the Super Hornet and JSF. Inventory projections for the F/A‐18 fleet suggest action must be taken to avoid a significant shortfall in inventories between 2010 and 2022. The DoN has already funded fuselage center barrel replacements (CBR+) to give aircraft additional FLE and ALE, but these appear to be insufficient to maintain F/A‐18 stocks. The forecasts also show that planned squadron rotations have only a marginal effect on service life after CBR+ is taken into account.

Another factor with the potential to accelerate airframe fatigue is higher operational tempo and involvement in future conflicts. We use data from Operation Iraqi Freedom to simulate the impact of sustained future combat operations on F/A‐18 inventories. It is found that short and infrequent conflicts do not lead to large inventory rundowns in future years, but the fatigue impact of long sustained conflicts is significant. Furthermore, participation in combat operations changes the distribution of aircraft that retire from WR FLE and ALE in favor of the latter. In light of the impending inventory shortfalls and continued increase in the proportion of flight hours in combat operations, our findings suggest that either the CBR+ SLEP must be expanded, or acquisition schedules accelerated in order to maintain the current stock of operations F/A‐18 aircraft.

Inventory sustainability does not appear to be as large a problem for the EA‐6B as the F/A‐18, provided the planned SLEP is fully implemented. We find that some additional WCP replacements are necessary to sustain Navy and Marine Corps inventories at desired levels, but by and large, the SLEP is sufficient to maintain inventories through the EA‐18G transition. However, if the WCP FLE SLL is revised downward due to recent findings of high crack initiation levels in current aircraft, then additional WCP replacements will need to be budgeted to maintain inventories. Indeed, of the three platforms we include in the analysis only the S‐3B is found to have sufficient fatigue life to meet its planned retirement from service without intervention by DoN acquisition planners.

Overall, the results show the value of a fatigue‐based approach to acquisition planning and the evaluation of current operational policies. For tactical aircraft, airframe fatigue is proving to be a larger constraint on aircraft service life than previously thought. Although more research is needed to link fatigue forecasts to the age‐related growth in operating and support costs, these forecasts provide policy makers with important information that can be used to foresee large budgetary commitments, allowing them to modify acquisition schedules or seek additional sources of funding in order to avoid inventory shortfalls.

Notes

¹This research was conducted while Chad Meyerhoefer was an economist at The CNA Corporation and Robert Trost was at The CNA Corporation while on Summer Break from The George Washington University (GWU). No official endorsement by the Agency for Healthcare Research and Quality or the Department of Health and Human Services, by The CNA Corporation, or by GWU is intended or should be inferred. Programming assistance by Kletus Lawler and Geoffrey Shaw is greatly appreciated.

²Within US Naval aviation FLE is known as “fatigue life expenditure”; however we use the term “expended” to avoid confusion with money metric expenditure.

³G‐force is a measure of the stress of vertical acceleration on the airframe, where 1G equals the acceleration due to gravity at sea level. See Wu et al. (1995) and the SAE Fatigue Design Handbook (1997) for a discussion of the engineering relationships relating G‐force to fatigue levels.

⁴The Navy uses a safe‐life requirement when designing aircraft, based on both crack initiation and crack growth thresholds, to ensure that the probability a structural crack occurs during an aircraft’s service life is acceptably low. For example, aircraft are designed so that a 0.01‐inch crack will not rupture within one service life.

⁵Note that nearly all flights associated with combat operations are given the combat distinction, not just flights where the pilot directly participates in a combat action.

⁶The reverse, however, is not true since there is little reason to believe that FLE accumulation in the current period affects ALE, especially if the indexes are measured monthly or quarterly.

⁷Note that the time specific controls κ_t and λ_t are omitted from the forecasting equations. Since these terms are measured relative to the last quarter (or year) of the sample, they effectively normalize past fatigue and arrested landing measurements to present values.

⁸This includes regular and surge deployments, composite unit training exercises (COMPUTEX), joint task force exercises (JTFEX), carrier qualifications, and small land‐based arrestment periods. Since equation (4) is estimated using only the sample of deployed aircraft, equation (4b) must be scaled by the proportion of time aircraft are deployed when it is re‐formulated using total flight hours.

⁹Although the Marines also fly the EA‐6B, they operate only about 20 aircraft and mission profiles are more similar to Navy aircraft than for the F/A‐18.

¹⁰Per NAVAIR planning, only F/A‐18A/C aircraft are swapped across the two operational environments, while F/A‐18B/D remain in the land‐based group throughout the simulation.

¹¹The full set of regression results is available from the authors upon request. In theory none of the parameter estimates should be negative, although we do get a few statistically significant negative effects in the ALE equation. This may result from our limited ability to determine precisely when aircraft are at sea.

¹²The high ALE levels attributable to other/special equipment training suggest a large proportion of flight hours in this nondescript category go to practicing carrier landings.

¹³Higher‐level G hits can be converted to 4G equivalents.

¹⁴Although only F/A‐18 and EA‐6B projections are reported below, S‐3B forecasts are available from the authors. They show that inventories are more than sufficient to last through squadron de‐commissioning and the transfer of the aerial refueling mission to the F/A‐18E.

¹⁵This number is based on the somewhat smaller final inventory level that is achieved after all newly acquired Super Hornets and JSF are brought into service. Since the newer aircraft are more productive, the same operational commitments can be fulfilled with a smaller overall inventory.

¹⁶Only series C and D aircraft are eligible for this SLEP.

¹⁷Although we were unable to obtain comprehensive cost estimates of the full package of necessary upgrades, a good proxy is the $1.7 million per aircraft cost of the US‐Israeli Litening‐Advanced Targeting (AT) system, recently endorsed by USMC for their F/A‐18s. Therefore, rotating 21 aircraft would cost $35.7 million in necessary upgrades.

¹⁸We test the sensitivity of these forecasts to variation in the scaling factor of 2/5 used in equation (5), denoting the amount of time aircraft spend aboard ships. If this factor is an underestimate then our results will minimize the beneficial effects of squadron rotations. However, increasing the factor to 3/5 does not significantly influence the results. In addition, the 18 year service life projection for the F/A‐18 (with scheduled CBR+) generated by our model corresponds to forecasts in Wu et al. (1995).

¹⁹A conflict year is defined as one year of sustained combat under the mission assignments and average levels of aggregate flight hours and non‐arrested landings observed during OIF. Since historical flight patterns include the average level of combat operations occurring during the past 12 years, a conflict year should be interpreted as additional tasking beyond historical averages, as would occur during a sustained conflict.

²⁰Since combat operations tend to be highly ship‐based, higher levels of conflict will also change the number of aircraft that retire from ALE as opposed to FLE in favor of the former.

²¹Note that the EA‐6B projections presented are based on historical data. However, recent airframe inspections by NAVAIR have revealed a larger number of cracks than expected. If the WCP FLE SLL is revised downward as a result, then additional re‐wings will be necessary beyond the eight we recommend.