Failure - Interactions

An Engineer's Lesson:

each Engineered-Safety-Feature

eg, the KC-135's  "SYD"

introduces another failure mode

"Shell 77"  / 3May2013,  KC-135R,  63-9977,

Board's Report:  64-pages,  pdf= 3.4MB

See the USAF  investigation  of   KC-135R,  63-9977,   inflight-upset,   slow onset   Dutch Roll,  progressed to   LoC-I    and  inflight-breakup:

KC-135  /  3May2013  Kyrgyz Republic

 " . . .  use of rudder,    while in a Dutch roll,    increased the aircraft's oscillatory instability. . . .   ensuing large side-to-side movements  . . .  inadvertent fluctuations in rudder position.   These fluctuating rudder movements,   coupled with slight right rudder   use    . . .   compounded the Dutch roll severity   and   produced extreme airframe stress   that caused the KC-135's   tail section to separate . . ."

Better:  see the  AW&ST  review  of  this   "Shell-77"  accident --

 . . .  a malfunction   with the aircraft's   series yaw damper  . . . crew identified the  ... source    ... but instead of turning the system  off,    they attempted to correct the yawing motion by   using rudder   and   aileron   inputs   as well as turning  on  the autopilot,   causing the oscillations to worsen.    ​

​Osborne, Tony.   "Factors … Lack of training ... inexperience with 'Dutch roll' … "

Aviation Week & Space Technology (March 24, 2014).

Local Spokane KXLY-TV offered a detailed  web-story  describing the lessons  from the mishap-investigation:

 . . . an aircraft malfunction … aggravated by the autopilot and  …  by some crew inputs, but only because they didn't recognize that it was a Dutch roll  …  only  …  see it in a simulator and  …  the simulator doesn't do a very good job of presenting the Dutch roll …

KXLY-TV,  March 13, 2014

Dutch Roll  recovery  techniques  evolved:

Prior to the Oso-upset-breakup-fire  of  19Oct' 1959,  Boeing had published their  suggestions  for Dutch Roll recovery   in the  Boeing Airliner,   June 1959.

Russel H. Baum,    age 32,     at Boeing  since mid-1957,

total Flight Hrs=  5,015;    time on B707 was 369-hrs.

Boeing  Test-Pilot-B,

Instructor-Pilot for training customer-pilots.

Boeing Airliner  (February 1960) p6,  Figure 5:

During rudder movement of 10 degrees to either side of neutral,  the "dead spot" in the slide valve prevents  ... boosting the rudder

. . . the probable cause  . . .

 was the structural failures induced

during an  improper  recovery  attempt  

from a Dutch roll  which  exceeded the

angle-of-bank  limits  prescribed by the company.

Braniff    Captain - Check Airman,   23,563  Flight-hours,

      John A. Berke,    at  Braniff  since   1936.

The Boeing Instructor-Pilot

    "took the controls . . . the roll stopped . . .

          and then rolled back to the left

              even more rapidly and violently . . . 

 sounds were heard  . . .

      engines separating  from the aircraft. . . .

         the Thrust Levers   . . .   snap

             and the cables go slack."

The Boeing Instructor-Pilot   "then initiated

     another Dutch roll . . .   bank was . . .  40 to 60 deg. . . .

 Berke  allowed  . . .  several oscillations . . .  

       Berke  initiated  recovery . . .   Aileron . . .  

            YAWED heavily to the right  

              and  rolled rapidly . . .   beyond a 90-deg. bank."

_Seattle_Times_,  Oct'  22nd,  1959,  p7:

still searching for a  separated  engine,  investigative groups formed,  

plans a Public Hearing.

Boeing  /   19 Oct 1959,  B707-227,  N7071, 

C/N 17691,  Line #45;  acceptance-flight and customer-crew training,  

violent-recovery from Dutch Roll,   separation of 3 engines, 

FIRE  inside wing trailing edge structure,   

forced  ditching-crash-landing  in 

North Fork of the  Stillaguamish  River,   

near Oso, Washington.  (Fatal.)

Later post-accident  Modifications

Full-Range Hydraulic-Boost

A "powered",  "irreversible"  control surface:  

the Rudder is termed "fixed"  (won't float during Beta-excursions of Dutch Roll  nor  sudden engine-out).

The various modifications,  Tip-Extension of the Fin, and full-range boosted-Rudder,  and  Yaw Damper,  provided B707-pilots with a more effective Fin-Rudder.


  - - Yaw Damper  not even mentioned in AAR  - - 

The C.A.B.'s  Aircraft Accident Report  for the upset of N7071   does NOT include any mention of any Yaw Damper:

This early  B707-227,  N7071,  had the early "parallel"  Yaw Damper installed [an inference from Boeing Airliner, June'59].

The term "yaw damper"  never appears in the CAB's Oso-AAR,   never mentioned:  On one hand that omission suggested that thus the yaw damper was NOT deactivated,  no yaw damper, NOT INSTALLED.   However, in contradiction, there appeared an earlier  Boeing Airliner, June'1959, p3-7, "707 Yaw Damper Operation" --  which cites the "yaw damper" as part of the basic configuration  for B707-227's.

- - No  Hydraulic-Boost until   δR >  10° to 15° - -

The C.A.B.'s  Aircraft Accident Report  for  N7071   does  mention a "rudder boost"  --  but that brief mention is misleading:  The initial hydraulic-boost design,  for that early B707-227 rudder,  ONLY ACTIVATED  hydraulic assist with Rudder-travel beyond  10°.   Essentially,  during the sideslip excursions of Dutch Roll,  even if the pilot held his Pedals "fixed" (pedals linked to only to the Control Tab),  the Rudder itself was "free"  to float, or trail, during the  oscillating  β -excursions.

N7071 - wreckage:  Aft sections,  on left-side, in the river-bed.

The four survivors had escaped from

"ditching stations" located in this aft fuselage-section.

Dutch Roll Oscillation 

Crew training  for customer-pilots:

   First demonstrations,   with Flaps-UP; 

Captain Berke  "made several recoveries

                from Dutch rolls in this configuration."

    Next demonstration:  slowed,   Flaps-40,   155 KIAS;  "Captain Berke  then made several recoveries 

                   from Dutch rolls in this configuration."

Accident occurred late afternoon,

on Monday, October 19th, 1959,   at 1620 PST.

Epistemology  of  the  term   Dutch Roll   in flight dynamics:

Bill Cook (p181)  cites  Hunsacker's early usage:

"... type of  oscillation  ...  possibly derived from the fact that

Dutch ice skaters,  as they negotiated the long, narrow canals,

 typically would  roll  from  the outside edge  of one skate

to the  outside edge  of the other  skate

in a series  of linked,  S turns . . ."

Hunsacker   had  spent his teen years in Michigan  (another "Ghost of Saginaw"),  so  he  was exposed to ice skating  (and perhaps  enjoyed  our   common-sense  of  "flying"  on skates) --  but  his   airplane "Dutch Roll"    analogy-metaphor to  ice skating   will be lost   on most of us   [we grew up  skating for speed,  with a stick in our hands].    Hunsacker's   "outside"  blade-edge  metaphor  misleads  his reader (as un-natural or  "not idiomatic") ,  since we   prairie-grown  skaters  never used that   lazybones- skating technique.

The   Dutch Roll skating technique   is foreign to us,  never used in North America  for hockey,   nor   for  speed skating,   and their  "rolling"  skate-blade is different.  

See the   video  demonstration  of that

              slow-speed "dutch roll" skating method.

There you have it:    Hunsacker's   ice-skating  metaphor  might still be  salvaged,  applied to   airliner  flight dynamics.   But we will leave  that  chore  for   you   writer-skaters.    [At one very large airline it seemed they had hired PILOTS  that included an  entire hockey team from the USAF Academy -- a sort of northern counter-culture to  contrast against their company's  southern mindset].

  Any future   airplane- Dutch Roll  skating  analogy-metaphor  might  better focus   on the   lateral-force  component   used  during skating,  and during airliner  Dutch Roll  testing.

The "explanation"  shown below  appeared as   Attachment "A",  

   near the end of the CAB's twelve-page  Aircraft Accident Report.

Careful:  this   "explanation"   of Dutch Roll  mentions  yaw  and  roll  moments,

       but  it omitted any mention of the   lateral-force   component, 

                 as cited in the Boeing Airliner    (excerpt shown further down this page).

For both the N7071 / 19Oct59, and N7506A /1Mar62, the C.A.B.'s AAR's never define the most-recent configuration-modification of the mishap aircraft (omitted). 

Never discussed in the AAR, nor elsewhere, was the Rudder-response DURING the divergent Dutch Roll of 19Oct59:

Note, deduced from their wording used in the Oso-AAR:

the Braniff B707-227, N7071,   must then have had the "power-augmented rudder" described in the Boeing Airliner, Feb'1960, "707 Rudder Boost".    That term "power-augmented rudder" was employed by Boeing in the July 1960 "Flight Improvement Program." Boeing Airliner. (July, 1960), 3-7.

 Even with that first (hydraulic-) rudder-boost design,   during oscillations of   Dutch Roll,   the Rudder deflections would VARY with the alternating sideslip-angle:

Rudder "floats" with the sideslip airloads-forces,

     "That is, the rudder tends to trail to the downwind side during sideslip. . . ." 

-- Even with the partial (hydraulic-) rudder boost installed and operational,   the Rudder couldn't be "fixed"   (deflection held at zero),     because the hydraulic boost   only activated    after the   Rudder DEFLECTION  had exceeded 10-15 degrees   (the central control region either side of neutral remained unboosted with normal control of Rudder handled by the TAB.   The pilot could only "fix" his pedals,   which only "fixed" the Rudder Control TAB. 

-- If the Rudder "floats"   during sideslip oscillations   of Dutch Roll,   or during sideslip of engine-out thrust asymmetry,    then the mechanically-linked TAB would also   "float"   to a "balanced" tab-direction -- further exciting the Rudder to increased deflection -- a feedback interaction that would further excite the amplitude of the yaw-roll excursion.

Boeing Airliner. Feb'1960 "707 Rudder Boost", 
Figure 5:   "During rudder movement of 10 degrees to either side of neutral, the dead spot in the slide valve prevents the ... PCU from boosting the rudder...."

C.A.B.  Public  Hearing

at the Olympic Hotel,  on Thursday Nov' 19th, 1959,

Story from _Seattle_Times_,   Friday, Nov'20th, 1959, p10.

. . . the Dutch rolls . . . reached angles . . .  far in excess of the limitations established by the company.   Responsibility for safety of this aircraft rested solely on the instructor-pilot. . . .   it was surely less than prudent to permit a pilot with no previous experience in the airplane to attempt a recovery from this extreme maneuver. . . . 

Control of an airplane  is complicated   because . . .  flight takes place in three-dimensional space.   Fortunately,  analysis  is  simplified  by the   mirror  (or lateral) symmetry  of the usual airplane.  By virtue of this symmetry,  the vehicle’s  motions   can be divided into two classes,   longitudinal  and   lateral,    that can be treated independently.   Longitudinal motions   take place  in the   plane of symmetry   of the airplane;  they  include  vertical  and pitching  motions  such as can be brought  about by deflection  of the elevator.

Lateral  motions  act  to  displace  the   plane of symmetry;   they consist of sideways   and  rolling  and  yawing  motions,   such as those  caused by movement of the ailerons and rudder.   The various  lateral  motions   are  interrelated  in a complex  way  and are more difficult to understand  and discuss  than  longitudinal ones. . . .

Vincenti, Walter G.   What Engineers Know and How They Know It :        

 analytical studies from aeronautical history.              

 Baltimore:  The Johns Hopkins University Press, 1993, p55.​

The over-all oscillation,  known as  "Dutch Roll",

results  from the   interaction   of these

moments  and  forces.

The  Dutch Roll  is a   combination   of  

yawing,   rolling,   and  lateral displacement . . .

Bill Cook had   worked  out  a solution to the Dutch Roll  hazard  during

the prior  XB-47  testing.     

Pdf-excerpt,  recollections  from Bill Cook's book,  and  Tex Johnston's.

NO-Rudder   technique    for  Dutch Roll Recovery

In-progress:  standby . . .