Case Study of 767 Horizontal Stabilizer
• Goals of this class
– Carry through the topics of this course on one product
• Look in detail at a real aircraft structural assembly
• Define and flow down KCs
• Compare different assembly methods
– conventional one based on fixtures
– proposed one based on part-to-part mating features
• Draw datum flow chains for them
• Study the economics
767 Case Study
11/30/2004
© Daniel E Whitney
1
History of 767 Horizontal Stabilizer Project
•
•
•
•
Fast/Flexible Manufacturing Project 1996
Coordinated Aircraft and Auto industry projects
Vought Aircraft partner via LAI
Vought’s goal: cut costs, earn more Boeing
business
• Vought’s hypothesis: convert from fixed to
flexible assembly tooling
• Vought’s focus of project: 767 H. S. upper wing
subassembly
767 Case Study
11/30/2004
© Daniel E Whitney
2
Our Challenge: How To Do This
• Available data
– Existing tooling
– No history, people, drawings
– Evidence of errors in tooling
• Our process
–
–
–
–
Understand goals of existing process
Reverse engineer from the top down
Expand scope of study to complete horizontal stabilizer
Look up the supply chain to Boeing to get the requirements
– Generate new process to achieve agreed goals
767 Case Study
11/30/2004
© Daniel E Whitney
3
Structure of Horizontal Stabilizer
767 Case Study
11/30/2004
© Daniel E Whitney
6
Top Level Key Characteristics
GAP
GAP
FTB
PLUS CHORD
SKIN
FTE
SPAR
END
SPAR
END
PLUS CHORD
AERODYNAMICS
(gap betw skin and FTB & FTE)
(gap between skins)
STRENGTH
(based on joining plus chords
and ends of spars)
HORIZ ASSY 2
767 Case Study
11/30/2004
© Daniel E Whitney
7
Horizontal Stabilizer Subassemblies (current decomposition)
STRINGERS
UPPER SKIN
ASSEMBLY
SKIN
D
W
F
FWD SKIN
AFT SKIN
FORWARD
TORQUE
BOX
FOR
PLUS CHORD
D SP
R
A
W
AR
RIB
STRINGERS
RE
A
PA
S
R
R
FIXED
TRAILING
EDGE
STRINGERS
PLUS CHORD
HORIZ ASSY 2
LOWER SKIN
ASSEMBLY
LOWER SKIN
PLUS CHORD
767 Case Study
11/30/2004
© Daniel E Whitney
8
PKCs for Horizontal Assembly
PKC #2 & #3: Aerodynamics affected
by these skin gaps
SPAR END FITTINGS
FWD TO
PKC #1:
Joint Strength
PLUS
affected by
CHORD
alignment between
plus chord ends
and spar end fittings.
RQUE B
OX
FORWARD SKIN
SPLICE STRINGER
AFT SKIN
FIXED TRAILING EDGE
SPAR END FITTINGS
JOINT STRENGTH PKC ACHIEVED
END FITTING
RD
HO
767 Case Study
11/30/2004
© Daniel E Whitney
END FITTING
SPLICE
PLATE
SHIM
SPAR
RD
)
HO
S C NE
D
PLU ALIG
S
(M I
SC
PL U
SPLICE
PLATE
SPAR
SHIM NEEDED TO ACHIEVE
JOINT STRENGTH PKC
9
Current Total Process
PLUS CHORD
ALIGNMENT
TO SPAR
ENDS PKC
PLUS CHORD
ALIGNMENT
TO SPAR
ENDS PKC
SKIN GAPS PKC
FINAL ASSEMBLY
SKIN GAP PKC
DELIVERED
BY SKIN
GAP AKC
PLUS CHORD
ALIGNMENT TO
SPAR ENDS PKC
DELIVERED BY
P.C. ANGLE
TO SKINS AKC
FTB-RIB-FTE SUBASSEMBLY
SKIN SUBASSEMBLY
SKIN GAP PKC
DELIVERED BY
DESIGN OF
FIXTURE TO
HOLD FTB
AND FTE
FORWA
RD TOR
FORWA
Q UE B O
RD SPA
X
R
FORWARD SKIN
S
PLU
RIBS
RD
CHO
767 Case Study
Figures for designing assemblies
SPLICE STRINGER
REAR SPAR
FIXED TRAILING EDGE
AFT SKIN
11/30/2004
© Daniel E Whitney
10
Top-Level KCs
FTB
Skin Gap
& rib align KCs
PKC #2
F Spar
PKC #1
Ribs
FS
PC
Skin Gap
& rib align KCs
AS
SS
PKC #1
PKC #3
A Spar
PKC #2
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
11
Product Decomposition Based on Independent KCs
FTB
F Spar
Ribs
FS
PC
SS
AS
A Spar
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
12
FTB
Ribs
PC
A Spar
FTB
FTE
F Spar
Ribs
SS
FS
AS
PC
SS
A Spar
FTE
FTB
F Spar
PC
Ribs
A Spar
767 Case Study
11/30/2004
FTE
© Daniel E Whitney
FS
AS
Decomposition/Subassemblies
Based on Independent KCs
F Spar
13
Sob
767 Case Study
11/30/2004
© Daniel E Whitney
14
Assembly Access Problem Eliminates an Attractive Assembly Sequence
UPPER PLUS
CHORD
UPPER
SKIN
STRINGER
~24"
SPAR
SHIM
PIVOT RIB
SHIM
~
STRINGER
LOWER PLUS
CHORD
767 Case Study
11/30/2004
If plus chords are assembled
to spar ends before skins
are assembled to plus chords,
then it will be almost
impossible to join skins and
stringers to plus chords
LOWER
SKIN
© Daniel E Whitney
15
Actual Subassemblies
FTB
F Spar
Ribs
FS
PC
SS
AS
A Spar
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
16
FTB
Ribs
PC
SS
Actual Decomposition
F Spar
FS
AS
A Spar
FTE
FTB
F Spar
PC
Ribs
FS
AS
SS
A Spar
FTB
FTE
Ribs
F Spar
767 Case Study
A Spar
11/30/2004
© Daniel E Whitney
FTE
17
Type 1 and Type 2 Methods for One Subassembly - KC Flowdown Seems OK
FTB
FTB
F Spar
F Spar
Ribs
Ribs
A Spar
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
F
A Spar
FTE
18
Second Subassembly Has Lost Its KC Links to Higher Level Assemblies
• Any assembly process for this subassembly must provide proxies for the missing KCs, regardless of
whether the subassembly is
made as a Type 1 or Type 2
• These KCs will be coupled
• Note that no drawing of this
subassembly could be found
767 Case Study
11/30/2004
© Daniel E Whitney
PKC #2
PKC #1
PC
PKC #1
FS
PKC #3
SS
AS
PKC #2
19
Possible Assembly Strategies
767 Case Study
11/30/2004
© Daniel E Whitney
20
Our Challenge
• Current assembly method relies on costly fixtures
• Can a process be devised that does not rely on
fixtures other than for support against gravity?
• Can such a process achieve the PKCs?
• Would it be economical?
• What new worker skills would be needed?
• Can we figure out what the old process was doing
so we can reproduce its objectives using new
methods?
767 Case Study
11/30/2004
© Daniel E Whitney
23
Diagram of Assembly Analysis Process
ORIGINAL PKCs
RD
FORWA
ACHIEVE PKCs
SEPARATELY
(ASSEMBLY IMPOSSIBLE)
SKIN
NGER
STRI
ASSEMBLY POSSIBLEPKCs COUPLED
Skin Gaps
2
1
1
2
PLUS
CHORD
NG
STRI
Forward Skin
ER #3
SK
AFT
IN
CURRENT METHOD
USING FIXTURES
FTB
TWO ALTERNATE METHODS
REQUIRING DIFFERENT AKCs
Plus
Chord
Aft Skin
using edge features
using hole and slot features
FTE
Assembly
Decomposition
Product
Architecture
Joint Strength affected by this alignment.
3
AKCs
1
2
KC Flowdown
PKCs
6
5
4
Assembly Sequence
Generation
Prune into
Families
8
7
Identify most
promising family
Propose Assembly
Features
Analyze
Sequences
FORWARD SKIN
9
10
Equipment
Requirements
SPLICE STRINGER
Process Selection
and Planning
PKCs
PKC #1
PKC #2
AFT SKIN
PKC #3
ASSEMBLY LEVEL DATUMS
PART LEVEL DATUMS
MATING FEATURE (SLOT)
MATING FEATURE (HOLE)
AKCs
VSA RESULTS
767 Case Study
11/30/2004
© Daniel E Whitney
Assembly
Feature AKCs
AKC #1
Plus Chord
aft hole &
forward slot
Aft Skin
aft hole &
inboard slot
Fwd Skin
inboard slot
& fwd slot
Stringer #3
inboard
holes
Stringer #3
holes
24
Liaison Diagram
Str4-11
Fwd
Skin
Plus
Chord
Splice
Str3
Str1-2
767 Case Study
11/30/2004
© Daniel E Whitney
Aft
Skin
25
Current Skin Assembly Process
SPLICE STRINGER
PLUS
CHORD
FWD SKIN
AFT SKIN
FIXTURE
Everything indexes off the fixture
767 Case Study
11/30/2004
© Daniel E Whitney
26
Current Skin Assembly Process - 2
Track f or Rout er t o Tr im Forward Skin Edge
Plus
Chord
Forward Skin
Splice stringer
Aft Skin
Fixture
767 Case Study
11/30/2004
© Daniel E Whitney
27
Assembly KC #1 & #2
#2
#1
767 Case Study
11/30/2004
© Daniel E Whitney
28
Datum Flow Chain for Current Skin Process
Fwd
Skin
Str4-11
Skin Gap
KC
Plus
Chord
Splice
Str3
Plus Chord
Angle KC
EXPLICIT DATUM TRANSFER
CONTACT WITHOUT
DATUM TRANSFER
KC
Str1-2
Aft
Skin
Fixture
767 Case Study
11/30/2004
© Daniel E Whitney
29
New Process #1: Fixtureless (Type 1)
FORWARD SKIN
HOLES CHOSEN
PLU
FOR LOCATION
SLOTS CHOSEN FOR
ORD
S CH
LOCATION, THERMAL &
SPLICE STRINGER
SHOT PEEN GROWTH
ACCOMMODATION
ALL SLOTS DRILLED
OUT TO FULL SIZE
AFT SKIN
ASSEMBLY LEVEL DATUMS
HOLES AT FINAL
ASSEMBLY
BIGGEST THERMAL
DIFFERENCE
PART LEVEL DATUMS
MATING FEATURE (SLOT)
MATING FEATURE (HOLE)
767 Case Study
11/30/2004
© Daniel E Whitney
SMALLEST
FULL SIZE
HOLE
30
PKC Delivery Map for New Process #1
PKCs
PKC #1
PKC #2
PKC #3
AKCs
AKC #1
AKC #2
AKC #2
Assy features
PC aft hole
& fwd hole
Mfr features
Size/shape
of PC and
hole locations
767 Case Study
11/30/2004
Aft skin
aft hole &
inbd hole
Fwd skin
inbd hole
& fwd slot
Size/shape
of skins
© Daniel E Whitney
Str 3 inbd
holes
Hole/slot
locations on
skins & str 3
Skins & str 3
holes & slots
Coord.of
slot lengt h
and hole size
31
Datum Flow Chain for New Process #1
Str4-11
Fwd
Skin
(6)
rd
o
h C
C
K
us le A
l
P g
(1)
An
(6)
Plu
An s Ch
gl e or
AK d
(5)
C
Str1-2
(6)
Skin Gap AKC
Plus
Chord
Note: in existing process,
skin-plus chord is a contact.
In new process it is a mate.
Splice
Str3
(6)
Aft
Skin
EXPLICIT DATUM TRANSFER
CONTACT WITHOUT
DATUM TRANSFER
KC
767 Case Study
11/30/2004
© Daniel E Whitney
32
New Process #2
SKIN
STRINGER
SHIM
PLUS
CHORD
PLU
SC
FORWARD SKIN
HO
RD
SPLICE STRINGER
AFT SKIN
SPACER TOOLS
LAYUP TABLE FEATURES
767 Case Study
11/30/2004
© Daniel E Whitney
33
.
Assembly Features for Process #2
FORWARD SKIN
U
S
P
L
D
OR
CH
SPLICE
STRINGER
AFT SKIN
ASSEMBLY LEVEL DATUMS
PART LEVEL DATUMS
MATING FEATURE (HOLE OR SLOT)
MATING FEATURE (EDGE)
767 Case Study
11/30/2004
© Daniel E Whitney
34
Datum Flow Chain for New Process #2
Str4-11
T
Plus
Chord
P lu
An s Ch (6)
gle or
AK d
(4)
(5)
C
Str1-2
767 Case Study
11/30/2004
(6)
© Daniel E Whitney
Skin Gap AKC
(6)
rd
o
h C
C
K
us le A
l
P g
(1) (1)
T (2)
An
Fwd
Skin
(6)
(5)
Splice
Str3
F
(6)
Aft
Skin
35
KC Deliverability Map - Process #2
PKCs
PKC #1
PKC #2
PKC #3
AKCs
AKC #1
AKC #2
AKC #2
PC aft hole
hole
Assy features
Tool location
at fwd end of
plus chord
Mfr features
767 Case Study
11/30/2004
Aft skin aft
Hole & inboard
surface
CNC
worktable
© Daniel E Whitney
Fwd skin
inboard surf
Skin gap
tool size
Part sizes &
hole
locations
36
Rib-Spar as a Type 2 Assembly
Fwd Torque
Box
Strength PKC
Fairness PKC
Rib
Fixed Trailing
Edge
Step 1: Put FTB and
FTE in Fixture
Step 2: Put in ribs
FTB-Rib-FTE assy
767 Case Study
11/30/2004
© Daniel E Whitney
37
Rib-Spar Assembly - 2
Skin
Plus Chord
Step 3: Add skins and adjust
skin gaps and plus chord alignment
to FTB and FTE
767 Case Study
11/30/2004
© Daniel E Whitney
38
DFC for Rib-Spar as a Type-2
FTB
(6)
SKIN GAP
& RIB ALIGN
PKC
FIXTURE
(1)
RIBS
(5)
(6)
767 Case Study
11/30/2004
FTE
© Daniel E Whitney
39
DFC for Wing Assembly as a Type 2
Two KCs in
conflict
FTB
Skin G
ap
PKC
Fwd
Skin
Str4-11
Plus Chord
Alignment PKC
RIBS
Plus
Chord
Plu
An s Ch
gle or
AK d
C
Plus Chord
Alignment PKC
Str1-2
p PK
a
G
Skin
Skin Gap AKC
SKIN GAP PKC
FIXTURE
Splice
Str3
Aft
Skin
C
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
40
Tolerance Analysis of KC Delivery Using VSA
• VSA was used to on each
candidate new process
• Results show that process
1 is unable to deliver AKC
& PKC 1 all the time
because the holes in the
splice stringer can’t be
placed accurately enough
• This also hurts PKC 2 and
3
767 Case Study
11/30/2004
• Process 2 is able to deliver all 3 PKCs 100% of the time
© Daniel E Whitney
41
Matlab Analysis
(TM)
• Assumed assemblers could maneuver the wing
skin laterally and angularly
• Assumed smaller variation in hole and slot
placement
• Assumed that the rest of the wing was error-free
• Determined that only a few assemblies would fail
767 Case Study
11/30/2004
© Daniel E Whitney
42
3"
G3
S1
FORWARD SKIN
4
G1
12"
P2
336"
5
4
6"
SPLICE STRINGER
PLUS
CHORD
3
63"
3
12"
AFT SKIN
H1
3"
S2
3"
P3
12"
P4
2
Parts and Their Frames
42"
1
3"
767 Case Study
11/30/2004
© Daniel E Whitney
43
FORWARD
FTB
G4
STRENGTH KC
AERODYNAMIC KC
G2
0.045"
G1
P2
G3
S1
5
FORWARD SKIN
RIBS
PLUS
CHORD
63"
4
a
SPLICE STRINGER
3
AFT SKIN
P3
H1
b
S2
2
0.045"
AFT
1
FTE
P4
Parts Assembled and Frames
Aligned
END
FITTING
351"
767 Case Study
INBOARD
11/30/2004
© Daniel E Whitney
OUTBOARD
44
Sample Space for Tolerance Analysis
X2
0.0127"
4
ADJUSTMENT POSSIBLE
BY TRANSLATION
AND CCW ROTATION
0.010"
.005"
ADJUSTMENT POSSIBLE
BY TRANSLATION
NO ADJUSTMENT
NEEDED
3
62.97"
63.00"
-.005"
2
63.03"
Y1
ADJUSTMENT POSSIBLE
BY TRANSLATION
-0.010"
5
ADJUSTMENT POSSIBLE
BY TRANSLATION
-0.0127"
AND CW ROTATION
0.015*5.25/29.25=.0027
767 Case Study
11/30/2004
© Daniel E Whitney
45
Matlab Results
(TM)
f ixt ure and FTB-ribs-FTE modeled as ± 0.008 and plus chord
modeled as ±0.01
0.02
0.015
~120
out of
10000
fall
outside
0.01
0.005
0
-0.005
-0.01
-0.015
-0.02
-0.01
767 Case Study
0
11/30/2004
0.01
0.02
0.03
0.04
© Daniel E Whitney
0.05
0.06
0.07
0.08
46
Pros & Cons of Proposed Processes
Pros
Cons
767 Case Study
Current Process
• Delivers all AKCs
and PKCs
repeatably
• Inflexible fixtures
• Variation absorbed
at stringer-plus
chord interface
11/30/2004
Proposed Process #1
• Delivers AKC #2
and PKC #3
repeatably
• Completely flexible
method
• No dedicated
fixtures
• Uses existing
fab
equipment
• Least costly
• Controls critical
interfaces
• Fails to deliver
AKC #1 on a few
assemblies
• PKC #1 & #2 not
delivered
on those same
assembies
© Daniel E Whitney
Proposed Process #2
• Delivers all AKCs
and PKCs
repeatably
• Completely flexible
method
• Uses existing
fab
equipment
• Controls critical
interfaces
• Requires higherfunctionality tack
fixture (higher
cost)
• Requires a limited
number of small
fixtures
47
Rib-Spar as a Type 1 Assembly
Step 1: Put FTE on
Support
Step 2: Add ribs
Step 3: Add FTE
FTB-Rib-FTE assy
767 Case Study
11/30/2004
© Daniel E Whitney
49
DFC for Rib-Spar as a Type-1
FTB
SK
I
N GAP PKC
RIBS
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
50
DFC for Wing Assembly as a Type 1
FTB
Skin G
a
p PKC
Str4-11
Plus Chord
Alignment PKC
Skin
d
hor
C KC
s
Plu le A
g
An
Plus
Chord
P lu
An s Ch
gle or
AK d
C
Plus Chord
Alignment PKC
Str1-2
Skin
figs for designing assemblies
767 Case Study
Skin Gap AKC
S
K
IN
GAP PKC
RIBS
Two KCs still
in conflict Fwd
Splice
Str3
Aft
Skin
KC
P
p
Ga
FTE
11/30/2004
© Daniel E Whitney
51
“Impossible” as a Type 2
FTB
d
hor PKC
t
sC
Plu nmen
Alig
ap PK
C
Str4-11
Plus
Chord
Plus
Alig Chord
nme
nt P
KC
RIBS
KCs do not conflict
Fwd
Skin
Skin Subassembly
Skin Gap AKC
SKIN GAP PKC
FIXTURE
Skin G
Str1-2
pP
a
G
Skin
KC
Splice
Str3
Aft
Skin
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
52
“Impossible” as a Type 1
FTB
p PKC
KCs do not conflict
Str4-11
Skin Subassembly
Plus
Chord
Plus
Alig Chord
nme
nt P
KC
Fwd
Skin
Skin Gap AKC
d
hor PKC
t
sC
Plu nmen
Alig
SK
N
I
GAP PKC
RIBS
Skin G
a
Str1-2
PKC
p
a
G
Skin
Splice
Str3
Aft
Skin
FTE
767 Case Study
11/30/2004
© Daniel E Whitney
53
Cost Analysis -1
• The basis for analysis was the KC-driven
Precision Assembly (PA) process for the 767
horizontal upper skin assy.
• PA time and cost were estimated for the 767 skin
• The 767 cost/time analysis was scaled for the
remaining 747 & 767 assemblies Vought makes
for Boeing.
• PA assumed to be accomplished in three distinct cells: Tack, CNC Auto-Rivet, Final Assembly
• These cells all require new investment
767 Case Study
11/30/2004
© Daniel E Whitney
54
Cost Analysis - 2
• Baseline times for each step were taken from
Vought’s estimates for its process.
• Required cell time for MIT’s processes was
estimated based on Vought’s times and a
distribution of realization factors applied to obtain
an assembly time for each cell.
• A computer simulation was conducted to
determine the necessary capital equipment.
767 Case Study
11/30/2004
© Daniel E Whitney
55
Simulation Scenarios
• Three PA processes were developed and
analyzed.
• The 3 processes are “Vought,” “MIT 1,” and “MIT 2”
• “Vought” is Vought’s proposed PA process
• “MIT 1” uses holes and slots. It was derived from “Vought” by applying the KC flowdown
method. “MIT 2” uses NC tack cell
• Three scenarios were studied:
- All Boeing assemblies, all programs
- Four representative assemblies
- Introduction of a new assembly
- New assembly would require new fixed tool but not new PA equipment
• One and Two shift operations
767 Case Study
11/30/2004
© Daniel E Whitney
56
Results - 1
• PA estimated to reduce process time by
approximately 50%. At current demand this
results in approximately XX hours saved
annually.* Value of flexibility, “image,” and
freed-up floor space not included.
• Annual savings = $X Million (assumes all
assemblies converted to PA at a rate of
$XX/hour.)
– VOUGHT TO BE = 54% OF AS IS TIME
– MIT 1 = 43% OF AS IS
– MIT 2 = 42% OF AS IS
– *ACTUAL NUMBERS ARE PROPRIETARY
767 Case Study
11/30/2004
© Daniel E Whitney
57
Results - 2
• Estimated equipment investment to implement PA
(example for MIT 1)
All Parts 4 Parts
One Shift
$21.4M
$14.1
Two Shifts
$14.1M
$7.3
(assumes cost per cell is Tack $2M, A-R $4.8M, Final Assembly $0.5M)
767 Case Study
11/30/2004
© Daniel E Whitney
58
Results - 3
• Current economics did not justify the new process
• The new process becomes economical if Vought gains new business for which it can use the new cells, thus saving the cost of new hard fixtures
• Training and cultural issues remain to be evaluated
– Adjusting by hand becomes adjusting via computer
– Ad hoc process becomes a preplanned and designed
one requiring more manufacturing knowledge during
design
– More communication between fab and assembly shops
needed
767 Case Study
11/30/2004
© Daniel E Whitney
59