Southwest B737 near Philadelphia on Apr 17th 2018, uncontained engine failure takes out passenger window

Last Update: December 12, 2019 / 17:52:41 GMT/Zulu time

Bookmark this article

Incident Facts

Date of incident
Apr 17, 2018

Classification
Accident

Airline
Southwest Airlines

Aircraft Registration
N772SW

Aircraft Type
Boeing 737-700

ICAO Type Designator
B737

On Dec 12th 2019 the NTSB released their final report concluding the probable cause of the accident was:

The National Transportation Safety Board determines that the probable cause of this accident was a low-cycle fatigue crack in the dovetail of fan blade No. 13, which resulted in the fan blade separating in flight and impacting the engine fan case at a location that was critical to the structural integrity and performance of the fan cowl structure. This impact led to the in-flight separation of fan cowl components, including the inboard fan cowl aft latch keeper, which struck the fuselage near a cabin window and caused the window to depart from the airplane, the cabin to rapidly depressurize, and the passenger fatality.

The findings were confirmed as discussed in the board meeting of Nov 19th 2019.

The NTSB analysed the fan blade fracture:

Fan blade No. 13 in the left engine separated due to a low-cycle fatigue crack that initiated in the blade root dovetail outboard of the blade coating. Metallurgical examination of the fan blade found that its material composition and microstructure were consistent with the specified titanium alloy and that no surface anomalies or material defects were observed in the fracture origin area.

The fracture surface had fatigue cracks that initiated close to where the greatest stresses from operational loads, and thus the greatest potential for cracking, were predicted to occur.

The accident fan blade failed with 32,636 cycles since new. Similarly, the fractured fan blade associated with the August 2016 PNS accident (see section 1.10.1), as well as the six other cracked fan blades from the PNS accident engine, failed with 38,152 cycles since new. Further, 15 other cracked fan blades on CFM56-7B engines had been identified between May 2017 and August 2019, and those fan blades had accumulated an average of about 33,000 cycles since new when the cracks were detected.

After the PNS accident, CFM reevaluated the fan blade dovetail stresses and determined that the cracks in the fan blade dovetail initiated in an area of high stress on the blade because normal operational stresses on the dovetail were higher than the peak stresses that were originally predicted. Before the application of the dovetail coating during manufacturing and before the reapplication of the coating that is stripped during each overhaul, the entire blade, including the dovetail, is shot-peened to provide a compressive residual stress surface layer for the material, which increases the fatigue strength of the material and relieves surface tensile stresses that can lead to cracking. However, higher-than-expected dovetail operational stresses can lead to the loss/relaxation of residual stress (the stress that is present in solid material in the absence of external forces) and premature fatigue crack initiation.

As part of the PHL investigation, five intact fan blades from the accident fan disk were examined to determine their residual stress condition. The results of the residual stress measurements showed abnormal residual stress profiles, including relaxed peak compressive stresses (most with residual tension near the surface), in one or more locations on each of the fan blade dovetails in the shot-peened area.

109 Abnormal residual stress profiles were also measured on fan blade dovetails from the PNS accident engine and on two cracked blades (found during subsequent inspections) that had not been shot-peened (as part of overhaul) since the time that the cracks were estimated to have initiated.

The NTSB analysed with respect to the departure of the engine inlet:

During the development and certification of the CFM56-7B engine, eight FBO rig tests and two engine FBO containment certification tests (under Part 33) were conducted between January 1994 and December 1996. These tests were performed to define the FBO fragmentation, which included determining the mass, quantity, direction, and exit trajectory and velocity of the fan blade fragments. The tests were also performed to define the engine fan case loading and displacements, evaluate the fan case radial containment capability, and assess the response of the proposed production engine hardware. Although the FBO rig and containment certification tests were primarily intended to validate and certify engine hardware (and not airplane structure), a production-representative inlet was installed during one FBO rig test and the first of two engine FBO containment certification tests.

Boeing used the requirements in Part 25 and Special Conditions 25-ANM-132 to develop an initial inlet design configuration. Boeing then incorporated modifications to the inlet design, based on the FBO rig and engine FBO containment certification test results, to ensure that the inlet could withstand the interface loads (that is, the loads on the interface between the fan case A1 flange and the inlet attach ring) resulting from an FBO event, the subsequent structural damage, and the associated post-FBO loads so that the airplane could continue to operate safely and land successfully. Boeing also performed certification structural analyses based on the FBO test results.

The damage to the accident inlet, fan case, and recovered fan blade fragments showed that some aspects of the accident FBO event were consistent with the FBO tests and structural analyses that were conducted as part of the engine and airplane certifications. For example, during the accident FBO event, the fan blade tip and some fan blade mid-span fragments traveled forward of the fan case into the inlet, and the blade root and other fan blade mid-span fragments traveled aft within the fan case. Also, the fan case and the inlet containment shield remained intact, exhibited no pass-through penetrations, and performed as designed. In addition, the initial impact of the separated fan blade with the fan case caused high local fan case deformations, displacements, loads, internal stresses, and the subsequent formation of a high-energy displacement wave that traveled 360° circumferentially and forward and aft, all of which imparted additional deformations, displacements, loads, and internal stresses to the inlet.

Despite these similarities, there were significant differences between the accident FBO event and the FBO tests and structural analyses performed as part of certification. One difference was that the fan blade fragments that went forward of the fan case and into the inlet during the accident FBO event had a different trajectory (a larger exit angle) and traveled beyond the containment shield. Another difference was that the inlet damage caused by these fan blade fragments was significantly greater than the amount of damage that was defined at the time of inlet certification.

With respect to the fan cowl departure the NTSB analysed:

An assessment of engine rundown loads (as part of the fan cowl’s certification structural analyses) determined that the load-carrying capability of the radial restraint fitting would be exceeded after an FBO event. Thus, the overall fan cowl certification loads were calculated with the assumption that the radial restraint fitting would completely fail immediately after an FBO event and provide no path for transmitting loads to the fan cowl structure. As a result, the certification structural analyses indicated that the fan cowl would remain attached to the airplane after an FBO event and throughout the rest of the flight, which was (and is currently) consistent with Boeing’s fan cowl design intent.

During the accident FBO event, the fan cowl fractured, and portions of the fan cowl departed the airplane. After the accident, Boeing used the damage to the recovered fan cowl hardware, the fan case impact damage, and the predicted interface loads for the accident condition to model the fan cowl damage sequence and perform a progressive failure analysis. This analysis, as well as other Boeing postaccident analyses, found that the fan cowl structure is more sensitive and more likely to fail when a separated fan blade impacts the fan case near the six o’clock position (the bottom of the engine), which occurred during the PHL accident, than at the twelve o’clock FBO release position used during the engine FBO containment certification tests. The fan cowl’s higher sensitivity and greater susceptibility of failure from loads resulting from an FBO impact near the six o’clock position were due to the proximity of this FBO impact location to the radial restraint fitting, as shown in figure 21. If the radial restraint fitting does not completely fail or disengage immediately after an FBO event (as designed), the fitting would be capable of transmitting loads from the fan case to the fan cowl.

With respect to the operational factors the NTSB analysed:

At the time of the engine failure (1103:33), the first officer was the pilot flying, and the captain was the pilot monitoring. FDR data showed that the left engine’s thrust lever was moved to the idle position 25 seconds after the engine failure occurred (1103:58) and that the engine start lever was moved to the fuel cutoff position 11 seconds later (1104:09).

At 1109:30, the captain indicated that she would assume the role of the pilot flying and that the first officer (as the pilot monitoring) should begin the Engine Fire or Engine Severe Damage or Separation non-normal checklist (shown in figure 17). The flight crew began the formal execution of that checklist at 1109:52, which was about 6 minutes after the time of the engine failure. However, between the time of the engine failure and the start of the checklist, the flight crew was primarily focused on maintaining airplane control during the emergency descent, navigating toward an emergency landing airport, and communicating with ATC.118 Further, guidance in the SWA 737NG QRH states that “non-normal checklist use starts when the aircraft flight path and configuration are correctly established...usually time is available to assess the situations before corrective action is started.”

The 14-item Engine Fire or Engine Severe Damage or Separation checklist did not have any immediate action (memory) items. Nevertheless, the flight crew recognized the two checklist items that needed to be immediately performed: moving the left engine’s thrust lever to the idle position (step 2) and moving the engine start lever to the fuel cutoff position (step 3). The CVR showed that the first five checklist items were completed by 1111:00, which was 1 minute 8 seconds after the checklist was initiated. The last item that the flight crew accomplished on this checklist was to start the APU (step 10), which occurred about 6.5 minutes (1116:16) after the checklist began. The remaining items on the checklist were to balance the fuel (as needed), change the transponder mode selector from “TA/RA” (traffic advisory/resolution advisory) to “TA,” change the isolation valve switch to auto, and plan to land at the nearest suitable airport. The NTSB notes that, although the flight crew did not discuss the plan to land at the nearest suitable airport as part of the checklist (step 14), the crewmembers had discussed, between themselves and with ATC, landing at PHL before initiating the checklist.120 The flight crew’s nonperformance of the other three items on the checklist had no adverse effect on the outcome of the emergency.

The flight crew did not call for or complete (using the standard checklist procedure) the three other non-normal checklists that applied to the circumstances of this accident: the One Engine Inoperative Landing checklist, which is required after the Engine Fire or Engine Severe Damage or Separation checklist; the Cabin Altitude Warning or Rapid Depressurization checklist, which is required if the cabin altitude warning light or horn annunciates; and the Emergency Descent checklist, which is required by a step in the Cabin Altitude Warning or Rapid Depressurization checklist. However, the captain was permitted to deviate from normal procedures, including the performance of checklists, in an emergency situation. Section 121.557, Emergencies: Domestic and Flag Operations, states the following in paragraph (a):

In an emergency situation that requires immediate decision and action the pilot in command may take any action that [the pilot] considers necessary under the circumstances. In such a case [the pilot] may deviate from prescribed operations procedures and methods, weather minimums, and this chapter, to the extent required in the interests of safety.[121] SWA’s Flight Operations Manual, section 5.1.4, provides the following similar information:

In an emergency situation that requires immediate decision and action, the Captain may take any action necessary under the circumstances. In such a case, the Captain may deviate from Southwest Airlines’ operations procedures and methods, weather minimums, and regulations to the extent required in the interests of safety.

The One Engine Inoperative Landing non-normal checklist (shown in figure 18) had no immediate action items. The first item on the checklist instructed the flight crew to plan for a flaps 15 landing. However, because of airplane controllability concerns, the captain decided to conduct a flaps 5 landing to ensure that the airplane’s airspeed would not get too slow. Even though the checklist and the Aircraft Operating Manual indicated that a flaps 15 landing should be conducted with one engine inoperative, the decision to conduct a flaps 5 landing was within the captain’s authority. The flight crew’s nonperformance of this checklist had no adverse effect on the outcome of the emergency.

The Cabin Altitude Warning or Rapid Depressurization non-normal checklist (shown in figure 19) had two immediate action items—don oxygen masks and set regulators to 100% (step 1) and establish crew communications (step 2)—that the flight crew performed from memory. The CVR recorded sounds consistent with breathing through oxygen masks at 1104:49, which was 1 minute 16 seconds after the engine failure occurred, but the flight crewmembers might have donned their oxygen masks before then given that the CVR recorded unintelligible cockpit communications between 1103:42 and 1104:41. During postaccident interviews, the crewmembers reported some initial confusion about the position of the switch that allowed them to communicate through a microphone in their oxygen mask. According to CVR information, at 1104:54, the flight crewmembers had established communications through their oxygen masks. The nonperformance of the next three items on the checklist, which were quick reference (time-sensitive) items, as well as the deferred items after checklist completion had no adverse effect on the outcome of the emergency. The Emergency Descent non-normal checklist (shown in figure 20) had three quick reference (time-sensitive) items that the flight crew performed independently of the checklist. (The checklist had no immediate action items.) First, the checklist indicated that the flight crew should advise the flight attendants and ATC about the emergency descent (step 1), and the CVR recorded the captain stating to ATC, at 1104:54, “Southwest thirteen eighty has an engine fire descending.”123 Second, the checklist stated that the airplane should descend, without delay, to 10,000 ft or the lowest safe altitude, whichever is higher (step 3). FDR data showed that, 25 seconds after the engine failure and 8 seconds after the airplane’s roll attitude was generally back under the pilots’ control, the emergency descent started, and the airplane descended through
10,000 ft about 8 minutes after the emergency descent began. Last, the checklist indicated that both thrust levers should be moved so that thrust is reduced to minimum (step 5), and the FDR showed that both thrust levers were moved to the idle position 25 seconds after the engine failure.

Given the emergency situation caused by the engine failure, the resulting airplane damage, and the rapid depressurization, the nonperformance of all but one of the remaining items on the checklist did not affect the situation. The one item on this checklist that could have affected the situation involved the deployment of the speedbrakes to the flight detent position (step 6) because that could have allowed a more rapid descent without additional airspeed (which the captain was avoiding due to airplane vibration) and could have facilitated a quicker landing at MDT, which was the geographically closer airport at the time. (The flight crew’s selection of PHL as the emergency landing airport is discussed in the next section.) However, the time that might have been saved would not likely have changed the outcome of the emergency. In addition, speedbrake deployment could have negatively impacted the aerodynamics of the damaged engine and wing, which would have caused additional stability and control issues for the flight crew.

During postaccident interviews, the captain and the first officer indicated that they were focused on maintaining control of the airplane, which was the first action listed in SWA’s Flight Operations Manual for all non-normal situations (see section 1.9.1). Maintaining airplane control was the most critical demand placed on the flight crew once the emergency situation occurred.

Specifically, the crew had to control an airplane that had lost thrust on the left side and was simultaneously experiencing greatly increased drag on the same side because of the damage that resulted from the engine failure. Also, immediately after the engine failure, the airplane entered a rapid roll to the left, which reached 41.3° within 11 seconds, and experienced positive and negative vertical accelerations. FDR data showed that significant control input forces in all three axes of flight were necessary to maintain control of the damaged airplane.

In addition to maintaining airplane control, communications with ATC—beyond those related to headings, altitudes, and clearances—imposed demands on a task-saturated flight crew (even though the intent of those communications was to provide assistance to the crew and follow ATC procedures). For example, according to CVR information, during the 17 minutes between the time of the engine failure and the time that the airplane landed, the crew was given four frequency changes.124 Also, ATC asked the crew multiple times to state the following: the nature of the emergency, including whether the engine was on fire; the number of occupants aboard; the amount fuel on board (which the flight crew provided in hours and was then asked to provide in pounds); and the ARFF support that would be needed after landing.

Even with the additional demands placed on the flight crew as a result of the engine failure, the CVR recorded the first officer’s concern about performing checklists. Specifically, at 1111:45, the first officer stated, “we’re gonna need a few minutes right? To run a couple checklists? Is that right?” The captain replied, 4 seconds later, “nope, just keep goin’.” Also, the first officer stated, at 1113:51, “we got a couple of checklists to run.” However, at 1115:54, after learning about the passenger injury, the captain stated, “let’s get it turned in.” During a postaccident interview, the captain stated that she had initially requested a long final approach to allow time to accomplish checklists but then decided to expedite the approach due to the passenger injury.

The NTSB concludes that performing required checklists according to standard operating procedures is a critical part of safe flight operations. However, given the emergency situation aboard this flight, the flight crew’s performance of most, but not all, of the items on the Engine Fire or Engine Severe Damage or Separation non-normal checklist and the nonperformance of the three other relevant non-normal checklists allowed the crew to appropriately balance the procedural requirement of executing checklists with the high workload associated with maintaining airplane control and accomplishing a safe and timely descent and landing.

Aircraft Registration Data

Registration mark

N772SW

Country of Registration

United States

Date of Registration

C qpbgcf bkgbj Subscribe to unlock

TCDS Ident. No.

Nkmcj Subscribe to unlock

Manufacturer

BOEING

Aircraft Model / Type

737-7H4

Number of Seats

fi Subscribe to unlock

ICAO Aircraft Type

B737

Year of Manufacture

PgAh Subscribe to unlock

Serial Number

Ekcnk Subscribe to unlock

Aircraft Address / Mode S Code (HEX)

Mnggee Subscribe to unlock

Engine Count

A Subscribe to unlock

Engine Manufacturer

BkAmqnAqh Subscribe to unlock

Engine Model

Ki efmbb fl Subscribe to unlock

Engine Type

Fhnkp n q Subscribe to unlock

Main Owner

B hdAjcebhdknpldmpdlAikhdnmnAqcnlnAmbcdpkeeAjlii nAcbApjciAcfmAeAcqfpbngldhhqckiAdibfg Subscribe to unlock

Incident Facts

Date of incident
Apr 17, 2018

Classification
Accident

Airline
Southwest Airlines

Aircraft Registration
N772SW

Aircraft Type
Boeing 737-700

ICAO Type Designator
B737

Photos

Photo from NTSBgov — (Photo credit: NTSBgov / Flickr / License: Public Domain)

You can read 2 more free articles without a subscription.

Subscribe now and continue reading without any limits!

Subscribe now

Are you a subscriber? Login

Southwest B737 near Philadelphia on Apr 17th 2018, uncontained engine failure takes out passenger window

Incident Facts

Aircraft Registration Data

Incident Facts

Photos

Related articles

Southwest B737 near Dallas on Apr 25th 2024, engine shut down in flight

Southwest B738 at St. Louis on Apr 19th 2024, engine failure

Southwest B38M at Washington on Apr 18th 2024, runway incursion forces rejected takeoff

Southwest B738 at Denver on Apr 7th 2024, dropped engine doors, flap damage

Southwest B737 at Baltimore on Apr 8th 2024, bird strike

Newest articles

Red Air MD82 at Miami on Jun 21st 2022, runway excursion causes gear collapse on landing

Southwest B737 near Dallas on Apr 25th 2024, engine shut down in flight

Subscribe today

Partner

AeroInside Blog

Popular aircraft

Popular airlines

Southwest B737 near Philadelphia on Apr 17th 2018, uncontained engine failure takes out passenger window

Incident Facts

Aircraft Registration Data

Incident Facts

Photos

Related articles

Newest articles

Subscribe today

Popular aircraft

Popular airlines

Join 6370 aviation professionals and enthusiasts getting key insights into aviation safety every Monday. Free.