-------------- --

------------------,

Applying Process
Control to Gear
-Manufacturing
~- ,-. -f- , -I' · 'I '. , ' "

Roben L. Sebetic
Roc1kwelllntern.a,ti,ona,l. Nlewark. OH

Introduction:
A common goal of gear manufacturers is to 2. Engineering standards for accuracy and
produce gearing that is competitively priced, that tooth contacrlocation have been established
meets all quality requirements with the minimum according to design and testing requirements.
amount of costin a timely manner, and that 3, Acceptable quality control. inspeetion
satisfies customers' expectations. methods are in place and GR&R studies (Gage
hi order to optimize this goal, the gear manu- Repeatability and Reproducibility) have been
facturer must thoroughly understand each manu- made, and the measured gage error has been
facturing process specified. the performance ca- deemed acceptable for all inspection measuring
pability of that process, and the effect of that equipment and gages.
particular process as it relates to the quality ofthe What Determines Gear
manufacturedgear. If I,he wrong series of pro- Manufacturing Quality?
cesses has been selected or a specific selected The quality level of a gear or gear set is
processis not capable of producing a qual ity part, determined during its manufacture by the spe-
manufacturing costs are greatly increased, cific sequence of production operations followed
The manufacturing of a desired quality level and the capability of each process .. The process
of gearing is a function of many factors includ- sequence selected for the manufacture of a
ing, but certainly not limited to, the gear design, specifi.c gear is determined by the final gear
the manufacturing processes, the machine capa- accuracies specified for that part.
bility, the gear material, Ilie machine operator, Two typical automotive/truck. axle gear pro-
and the quality control methods employed, In cess flow diagrams are shown in this article. The
this article we will make ome basic assumptions target quality level of this type of gearing is
about the gear design, engineering specifica- generally set at AGMA 8. Example I (pg, 26)
tions, and the quality control methods employed, shows a general flow diagram of bevel gearing.
and concentrate mainly on the manufacturing Example 2 (pg. 27) is a general flow diagram of
processes, their control, and how they affect the spur and helical gearing.
gear quality produced. Note that it is possible, by the addition of a
Assumptions hard profile finishing operation after the heat
In order to concentrate primarily on the gear treating operations, to increase the gear quality
manufacturing processes selected, their control, to AGMA 11 01' 12. Along with thi added
and how they affect gear quality, we have made operation, it may be necessary to tighten up
several basic assumptions: some of the current manufacturing tolerances
I. The gear designs are good, tooth contact and to specify different workholding equip-
analysis programs have been run, and motion ment This would be primarily for bearing and
curves and displacement values are within de- bore diameters and could very easily affect [he
sired limits. Product testing and evaluation have process capability of several different manu-
been completed and found acceptable. facturing operations.
22 GEAR TECHNOLOGY
 What is Process Control?
Many gear manufacturers use process con-
trol techniques as a means of attaining the gear ueL
6
quality specified. AGMA defines process con- CHART
OFTHE
5
trol as a method by which gear accuracy is
A VERA:GES 4
achieved and maintained through control of 3
(X)
manufacturing equipment, methods, and pro- 2
ici,
cesses. without resorting to the inspection of
--+-- .....
- _1__........
-._+· ......
'......__.. -t_ ...,_t_,_iii ..... iiii~+ ... ,iiiiiiii +-,
individualelements of every gear produced. I 2 3 4 5 6 7 8 9 10 11 12
SUBGROUPS
Process control. technique analyze the manu-
facturing processes
(METHODS),
and quality control plans
the gear steels used (MA TE-
RANGE
CHART : I~----- UeL
I -

RIAt.), the machine capability (MACHlNE). ~ I ---- ------- --- ---.- ----.--- --.--- --- R

and the operator (MAN). When these tech- I __ ~--~--~---~.-_~--~---~.--~ •• _~ •• -~---~ ~-

niques are properly applied to a specific process

1 2 3 4 5 6 7 s 9 10 II 12
SUBGROUPS
and that proces i capable, then the gears manu-
Fig. 1.
factured will be of uniform quality,
How is the Process Monitor,ed?' variations within one lot of material and varia-
Data is collected on a specific characteri tic tions caused by one operator may be ignificant;
for a specific process and then grouped in a however. such variations are basic to the process
histogram to get lUI idea of the distribution sample. and not easilyeliminated, If there isno practical
With collected data, we can make some statisti- way of eliminating their influence 011 the process
cal calculations for the mean and the standard performance, these variations must be consid-
deviation. What we find is that the disrributions ered "common" variations.
will vary in shape, spread. and location relative "Assignable" variations may come from many
to the tolerance. or any combination of'the three. . ources. A change in either operator from shift
The mean is defined as the sum of a group of to shift or a material heat code lo; change during
numbers di vided by the total number of elements the run are examples that can cause variation
within the group. The standard deviation is which is external to the basic process.
defined as the measure of di per ion ( cauer or A process is said to be "in control" when
spread) of a set of data around its mean. the assignable causes have been eliminated.
Statistical process control (SPC) techniques A process in statistical control wHl be evidenced
are excellent tools for evaluating a machine on a control chart by the absence of points
process. The results generated from an SPC beyond the control limits and by the absence of
evaluation give a "snap hot" picture of how the non-random panern or trends within the control
process is performing now and can be used to limits (LCt., uci,» See Fig. I.
predict how it wiIl perform in the future. These What is Process Capability?
results can be u edto compare the performance When the process is brought under control by
of other machines or processes and can be used eliminating the assignable causes, we can then
to determine if a machine or process is capable of assess its capability. It is important to note that
performing a specific task. a process "in control" mayor may not be ca-
When is a Process "In Control,"? pable. Process capability is defined by two
There are two causes of variation in any pro- terms: Cp and Cpk,
cess. They are "common" and "assignable" Cp is the Process Capability Ratio. The Cp
causes. Common causes are random occur- ratio is defined and calculated as follows:
rences that are inherent tothe process. They Cp= Specificarion Tolerance Spread/(6 x
Robert L S:ebetic
cannot be removed without changing the pro- Sigma), where Sigma is the standard deviation is 1111,' lIIallager of lilt! Au-
cess. Generally, they are responsible for 85 to of the process being examined, tomotive Operations Gear
90% of process variations. Assignable causes Basically. the Cp index is a comparison of Lab c1I Rockwell lnterna-
tional, Newark. OH. He
are non-random or patterned occurrences that the 6 Sigma spread of the distribution to the
has presented papers at
can be identified and eliminated. specification tolerance. Ideally, the 6 Sigma bot]: AGMA and SME
When conducting a capability study. material spread will fall within the specification toler- gearing conferences.

MARCH I APRIL 199223

 anee along with room to pare on each ide of the M~AN = Process Average
specification limits. The Cp ratio does not ac- Sigma = Standard Deviation of Process
count for centeri ng the process relationship 10 the 'Generally. file minimum accepted values for
. pecification, See Fig. 2. both Cp and Cpk are 1.33, which is equal to 75%
Cpkis the Capabihty Index .. It gives an indi- of the tolerance for a 6 Sigma analysis. The e
cation of the location of the di tribution relative indices give am indication of how well the pro-
to the specification limit. Cpl accounts for the ce . is making the product according to de ign
proce ss centering. or manufacturing specifications. Also, note that
For bilateral tolerance , when both upper and when the mean of the process is equal to the
lower pecification limits are given, Cpk is cal- target value of the tolerance specification, then
culated by the following formula, Cp = Cpk. The CpkcalcuJation from data inthe
The minimum value of Cpk = previous example is shown in Fig .. 3.
(USL - MEAN)/(3 X Sigma) or It is also important to remember that the
(MEAN - LCL)I(3 X Sigma) capability indices Cp and Cpk are based onrwo
For unilateral tolerances when only a mini- elements: the de ign tolerance and the standard
mum or maximum specification limit is given, deviation oflhe process, You can see, given the
Cpk i calculated by the following formula: same data, we can change the tolerance to make
Cpk = (SL - MEAN)/3 Sigma the capability look better or wor e for a given
Where: USL = Upper Specification Limit distribution. But the important element ill the
l.SL = Lower Specification Limit capability equation is the tandard deviation
SL = Specification Limit (Sigma) of the measured distribution, and we
call not change til at Uiii less we chan ge at least one
GIVE PROCESS DATA: of the preces element .
I\>\EA . = 17.5 SPECIF1· Al'lON lARGEl = 16.5 ApplyingProce s Control -
STANDARD DEVIATIO, = 1.3 l'OLERANCE = +/- 5.5
(SIGMA) What Should .BeDone ? i

Cp CALCULA'JIONS: In a word or two, precess control requires

PROCESS SPREAD = 6 x S.lGMA = 6 x 1.3 = 7.8
TOLERANCE SPBCIFICA nON SPREAD = UPPER SPEC. . LOWER SPEC. that the entire manufacturing process be ex-
22.0 11.0 11.0
amJned from the forging coming in the door
J10LERANCESPECI'fICATIQN SPREAD_ ill - 1.41 to the tlinal shlipment of the gear. Where
Cp = PROCESS SPREAD - 7.8 -
process capability uidies are taken on each
individual manufacturing operation to establish
SP CIFI AnO capability, control charts and plan mill I be
TARGET ----
maintained on 8. continuous basis to monitor
performance capability .. AI 0, do not forget the
assumption . made in the beginning about the
1
10- -1213 14-'5u7Isi920
I--- PROCESS SPREAD ----I
2d'-:!3 '
design, engineering
control method
specifications,
employed.
and quality

10 1.I
L TOLERAN
----,-----,---,------
12 13 14 15 I~6
E SPECI~ICATIQ
17 18 19 20
SPREAD
:! I 22 23
AppJying Procecs 'Control. -
Where Do We Start?
Fig. 2
One of the most meaningful and eye-opening
Cpk calculation from datu in .fig. 2: activities a gear manufacturer C8_n pursue is the
Cpk - UPPER SPECIFICATION LIM.iT - MEAN
audit of everal lots of finished gearing. Basi-
3 x.S.IGMA cally, it is a comprehensive self-asse ment of
= 22.0 • .17.5 = 1.15 your gear manufacturing capability. Select. a
3 It 1.3 variety of gearing ba ed on pitch diameter, dia-
SpecificatiQIl------ ,
Target rnetral pitch.and quality class required. Use lot
sizes of at least. twenty-five pieces and inspect

/.0·1 --------~-23
I all. criticalgear-related characteristic . Analyze
the result. using statistical methods, Now an-
f-PROCESS SPREAD---j
swer tile following question:
I. Are allthe characteristies inspected within
prim specification?
,24 (lEAR TECH/oIOLOQY
 2. After statistically analyzing the dataare facturing processes that can be used to produce
the 6 Sigma values calculated less than tile an acceptable gear. Manufacturing engineering
tolerance specification for that characteristic? has the latitude to select the process to be used,
3. Are the data distributions centered with depending on the capability of the equipment
respect to the specification limits? Remember, available and the available open capacity.
Cp and Cpk values of 1.33 are generally the The results of capability studies show what
minimum accepted numerical value. is causing the "assignable" variation and what
If the answer to Questions 1, 2, and 3 are must be done to get it out ofthe process. Once
"yes." and the calculated Cp and Cpk values are this variation is eliminated, true process capa-
greater than 1.33, one should feel very confident bility results.
about the quality of the overall gear manufactur- In most processes, a capability study will
ing operation. Probably SPC techniques to moni- show a large amount of initial variation. If
tor all manufacturing operations are already in capability studies have not been made on the
use and the exact performance capability of each proces previou ly, the "assignable" variation is
and every process in the plant is known. likely to be the greater part of the total varia-
If the answer to anyone of the questions tion .. Itis most likely the variation that gets the
above is "no," there is work to be done. If the hop into the most trouble, The process capa-
discrepancies are few and the manufacturing bility study will work to detect and reduce or
operation has a formal process control. system eliminate the "assignable" variation.
in place, corrective action may not be much
more than trouble-shooting a specific process
Table 1
operation. If the discrepancies are many. are
critical characteristics. and the manufacturing I. MAN
operations are weak in the areas of SPC and · Training on equipment and procedures
process controls, a great deal of work will. · Work day fatigue/awarene s
need to be done, · Setup skill & operating skill
· Operator and operator changes
Applymg Precess Control -
2. MACHINE
Perform Capability Studies · Quality capable
Knowledge of the performance capability of · Machine alignment
a process is essential to the overall concept of · Spindles
producing a quality gear. If the equipment · Draw
selected and used is sufficiently accurate to meet · Speeds, feeds, and thermal growth
· Rigidi.ty
quality requirements, an acceptable gear can be
· Balance
expected. When the quality is marginal orunsar- · Machine maintenance & lubrication
isfactory, or when the processing equipment 3. MATERIAL
cannot meet the quality requieements, then the ..Material hardenability
additional costs of scrap. rework, decreased pro- · Material chemistry
· Microstructure
ductivity, and warranty result.
· Hardness
A process capability studyis a technique for
· Machinability
measuring that which a process is capable of · Material cleanliness
producing under normal, in-control conditions. ..Dimensional characteristic
In a capability study, measurements of gears (diameters, lengths, parallelism, runout)
produced in a run are analyzed to determine · Geometrical considerations
(rims, webs, thickness, position)
whether or not the process is capable of produc-
4. METHODS
ing, to specifications, a given characteristic on
· Workholding equipment type & condition
succes ive part under production conditions, · Workholding rigidity
Process capability is a measurement of the inher- · Coolant. type and volume
ent precision of a manufacturing process. · Cutting tool quality. new & resharpened
Applying Process Control - · Cutting method
(hob, shape, bear speed. mill, broach)
Identifying the Process
· Material handling system
Elements era CapabilUy Study
There are numerous combinations of manu-
MARCH I APRIL 1992 2:5
 EXAMPLE I
HYPOm/SPIRAL BEVEL GEAR. AND PINION SET PROCESSING

PINION

II' FORGING

PRE· TREATMENT NORMALIZE

·· - PROFILE TURNING
K.W.LIU.:1~IN...l,;G
... BLANKrNG - PROFI~E TURNING
- BROACHING - SPLJNJNG
- HOLE DRILLING - GREEN GRJNDfNG
- IDENTIFICATION - THREADING
- mENTIHCA TION

TOOTH CUTTING - MACHINE SETUP TOOTH CUTTING - MACHINE SETUP

GREEN TEST - CUITING SETUP - CUITING SETUP

APPROVAL APPROVAL
- SIZE AND CONTACT - SIZE A -0. CONTACT
COMP~RISON TO COMPARISON TO
PRODUCTI.ON "REF'. PROD CTlON "REF"
- INSPECTION -INSPECHON

BURRING. CHAMFERING BURRING, CHAMFERING

HEAT TREAT - CARBURlZE AND ~2A~~.AT - CARBURIZE AND

QUENCH QUENCH
- BORE SIZE - INDUCTION ANNEAL
- GEAR GEOMETRY - STRAIGHTENING

HARD GRINn - BORE I.D. HARn GRINn - BEARING JOURNALS

MATCH AND LAP - REdNE TOOTH SURFACES OR

'·,R ACCEPTABLE TOOTH CONTACTS
HARDFJN~SH AND NOISE LEV L

- MONITOR Fm[SHlNG OPERA TJON

, PROTECT - PHOSPHATE COA TlNG FOR BREAK

IN, RUST PROOFING" AND
mENTIF]CAT~ON

- ASSEMB! Y LINE OR CUSTOMER

26 G E "FI T E.C H N 0 lOG ...

 To help identify "common" and "assignable" EXAMPLE 2
variations within a process, we would suggest SPURJHELICAL GEAR AND PINION PROCESSING
that every process or operation that is analyzed
be divided into four disti net categories, as shown GIEAR OR PINION
in Table I, and each category evaluated through
the analysis of proces elements, The four FORGING
categories are I) Man, 2) Machine, 3) Mate-
rial. and 4) Methods, PRE- TREATMENT NORMALIZE
Can We Sum This Up?
The gear quality level specified by the de- BLANKING
sign dictates the proces es used in your gear - PROFILE TURNING
manufacturing to achieve that specified gear - HOLE DRILLING, BROACHl G, SPU ING
quality. Each process selected must be perfor- - KEYWAY, THREADING
mance-capable over the long term to assure that - IDENTIFICATION
the qualityi maintained from one operation to - "GREEN" 'GRINDING
the next. Audits of finished product must be
conducted on a regular basis. TOOTH CUTUNG
If quality defect occur, use process capa- -HOBBING,SHAPERCUlTING,SHEARSPEED,
bility techniques to identify and fully under- MILLING, ROLLING
stand the root causes of the problems and the - SHAVING
pay backs as ociated with fixing them. After
identifying the problems, use the Pareto prin- GEAR INSPECTION
ciple to determine the areas of the highest - CUTT1NG SETUP APPROV AL
payback. Developagear quality group, a game - PIN DIMENSION
plan, and start small. After several successes - LEAD, CROWN, AND INVOLUTE FORM
are achieved, the quality group can grow ag- - PRECISION INDEX
gressively and take on more qualiryproblems.
Once process control techniques are em- CHAMFER.lNG. BURRING" ETC.
ployed, a gear manufacturer generally realizes
tremendous benefits. One of the spin-offs of HEAT TREATMENT
this type of analysisis that it will not take long - CARBURIZE & QUENCH, NITRIDE,
before most gear manufacturers will want to INDUCTION HARDENING, ETC.
make some fundamental changes in the manu- - INDUCTION ANNEAL
facturing process to improve quality, improve - BORE SIZE, SPLINE SIZE, ETC.
producti vity, or to implement some cost red uc- - TOOTH SIZE AND TAPER
tion projects. This comes from the thorough - RUNOUT, STRAIGHTENING, ETC.
understanding of the process and its present
capability. HARD GRINDING
There are many day we all struggle with - BORE AND FACE LOCATIONS
gear quality issues. They consume a great - BEARING DIAMETER, JOURNALS, ETC.
deal of our time and effort. There can be a
great deal of frustration; but in today' s com- SllEED OR HONE
petitive environment, a formal process con- - MONITOR MANUFACTURE
trol system in place is essential no matter - NICK AND KNOT R MOV AL
what product is manufactured, •
HARD PROFILE FINISHING

Presented at the AGMA 1991 Spring Gear I CUSTOMER

Manufacturing Symposium and the SME 1991' - ASSEMBLY
Advanced Gear Processing Clinic. Reprinted - OUTSIDE CUSTOMER
with permission.
MAR CHI APR IL I 9 9 2 21