Professional Documents
Culture Documents
Overview
The aim of this training material is to provide engineers with a working knowledge of tolerances stack-
ups, what they are, when they should be performed and how to complete them.
The scope of this training material is Global Powertrain Manufacturing Engineering, Prismatic Machining.
The Fixture Locating Principles for Prismatic Components is a recommended pre-requisite for this
training.
Knowledge of the ASME GD&T standard is a pre-requisite to understanding the content of this material,
as these elements are used but not explained this material.
Contents
Overview ...................................................................................................................................................... 2
Introduction ................................................................................................................................................. 4
What a Stack-Up Is .......................................................................................................................... 4
Why a Stack-Up Should Be Performed............................................................................................ 4
When a Stack-Up Should Be Performed ......................................................................................... 4
Stack-up Methodologies............................................................................................................................. 5
Arithmetical Stack-ups ..................................................................................................................... 6
Statistical Stack-ups ......................................................................................................................... 7
Statistical Method with a Safety Factor ............................................................................................ 9
Comparison of Methodologies ....................................................................................................... 11
Which Methodology to Apply ......................................................................................................... 11
Constructing Stack-Ups ........................................................................................................................... 12
The Loop Method ........................................................................................................................... 12
Stack-up for Part Print Compliance – Hole Depth ......................................................................... 14
Stack-up for Part Print Compliance – Hole Position ...................................................................... 15
Stack-up for Part Print Compliance – Depth .................................................................................. 16
Stack-up for Stock Variation – Faces ............................................................................................. 17
Stack-up for Stock Variation – Diameters ...................................................................................... 18
Introduction
What a Stack-Up Is
A Stack-up is a mathematical study of tolerance accumulation from the geometric dimensioning and
tolerancing of features.
The primary reasons, in Global Powertrain Manufacturing Engineering (GPME) Machining, for performing
stack-ups are;
• To determine that finish part tolerances will be met, when made in two or more process
steps.
• To determine the size of in-process features, to ensure the finish features always have
appropriate stock.
• To determine the optimum number of process steps.
• To optimize the process sequence.
• To optimize in-process tolerances.
There are many different reasons for performing stack-ups within Product Development (PD) and GPME
Assembly, these are outside the scope of this training material, however the basic stack-up methodology
is the same.
Stack-ups should be completed before engineering approval of the Process Illustrations, Tool Layouts
and Gauge Drawings.
A tolerance stack-up shall be completed, to determine or verify the correct size or tolerance, for every
characteristic* in both the following circumstances;
* A single stack-up may be shared across multiple characteristics with common tolerances, process steps
and datums.
Stack-up Methodologies
There are two primary stack-up methodologies, arithmetic and statistical. These two methods provide
pessimistic and optimistic results, respectively.
The basis of the arithmetic stack-up methodology is the assumption that the part print dimensions can
have any value within the tolerance range and the arithmetically stacked tolerances include the full range
of possible tolerances.
The basis of the statistical stack-up methodology is the assumption that the part print dimensions vary
randomly according to a normal distribution centered at the midpoint of the tolerance range with the +/- 3
Sigma spread covering the tolerance range. As shown in Figure 1
Figure 1
+/- 1 Sigma = 68% of components
Simply stated 95% of components fall within 66.67% (+/- 2 Sigma) of the part print tolerance, so to
assume worst case for every tolerance concurrently is very pessimistic. The mathematics of the statistical
calculation makes allowance for this and the fact that the probability of all tolerances concurrently being at
their extreme reduces as the sample size increases. Thus for given part print tolerances the statistical
analysis leads to much more optimistic outcome than the arithmetic stack-up.
However, real world experience shows that the results are not usually as good as these assumptions
require. To enable some of the benefit of the statistical methodology to be realised, with reduced risk, the
outcome of the statistical stack-up may be multiplied by a safety factor.
Arithmetical Stack-ups
An arithmetical stack up is a methodology that employs a simple mathematical addition of the respective
tolerances. The results obtained by using this formula are the worst case conditions accounting for 100%
of all individual variations.
Arithmetical Formula
VA = T1 + T2 + T3… + Tn
Tn = Tolerances
VA = Variation - Arithmetic
Figure 2 shows four plates stacked together. The arithmetic variation of the stack is represented by the
equation
VA = T1 + T2 + T3 +.T4
Figure 3 shows the four plates with a tolerance of ± 0.25 on each plate, giving an arithmetic variation of
VA = ± 1mm.
± T1 ± 0.25
± T3 ± 0.25
± T4 ± 0.25
± T5 ± 0.25
T1 0.2
+ T2 + 0.2
+ T3 + 0.2
+ T4 + 0.2
= =
± VA ± 1.00
Figure 2 Figure 3
Whilst this is a simple and robust approach the calculated variation soon gets unnecessarily large, as all
tolerances are taken into account at their worst case, however, the probability of all the process variables
being at their worst case simultaneously is unlikely.
Statistical Stack-ups
The basis of the statistical stack-up methodology is the assumption that the part print dimensions vary
randomly according to a normal distribution centered at the midpoint of the tolerance range with the +/- 3
sigma spread covering the tolerance range. Based on this reduced variation for the majority of the parts,
the probability of all the tolerances being at their extremes is unlikely. For given part print tolerances the
statistical analysis leads to much reduced variability than the arithmetic stack-up.
Statistical Formula
VS = eT1 + T2 + T3 … +Tn
2 2 2 2
2
Tn = Tolerances Squared
VS = Variation – Statistical
Figure 4 shows four plates stacked together. The statistical variation of the stack is represented by the
equation
VS = eT1 + T2 + T3 + T4
2 2 2 2
Figure 5 shows the four plates with a tolerance of ± 0.25 on each plate, giving a statistical variation of
VS = ±0. 5mm.
± T1 ± 0.25
± T2 ± 0.25
± T3 ± 0.25
± T4 ± 0.25
T12 (0.25*0.25)
+ T22 + (0.25*0.25)
+ T32 + (0.25*0.25)
+ T42 + (0.25*0.25)
SQRT SQRT
= =
± VS + 0.50
Figure 4 Figure 5
In comparison to the arithmetic variation of ±1.0mm the statistical variation of the same four tolerances is
just ±0.5mm.
This is a significant reduction in variation that could ‘in theory’ be applied to either the reducing product
tolerances or increasing the in-process tolerances. However, there are several assumptions that need to
be satisfied in order for statistical results to be accurate.
Using the statistical stack-up carries a higher risk of defects because in reality,
For these reasons 100% statistical stack-ups alone MUST NOT BE USED in GPME Machining.
2
Tn = Tolerances Squared
Figure 6 shows four plates stacked together. The statistical variation of the stack, with a 1.5 safety factor
is represented by the equation
Figure 7 shows the four plates with a tolerance of ± 0.25 on each plate, giving a statistical variation with
safety factor, of
VSF = ±0.75mm.
± T1 ± 0.25
± T2 ± 0.25
± T3 ± 0.25
± T4 ± 0.25
T12 (0.25*0.25)
+ T22 + (0.25*0.25)
+ T32 + (0.25*0.25)
+ T42 + (0.25*0.25)
1.5 x SQRT 1.5 SQRT
= =
± VSF + 0.75
Figure 6 Figure 7
In comparison to the statistical variation of ±0.5mm, the 1.5 safety factor has inflated this value to ±0.75.
This is still a worthwhile reduction in variation compared to the arithmetic variation of ±1.0mm, yet has a
greater robustness than a 100% statistical stack-up. However, there are still some assumptions that need
to be satisfied in order for statistical results with a safety factor to be robust.
Comparison of Methodologies
The chart in Figure 8 demonstrates the increasing benefit of the statistical methodology over the
arithmetic methodology as the number of tolerance elements increases, and how the safety factor simply
inflates the statistical result to increase robustness. No statistical results are shown below four tolerance
elements, as below this threshold the arithmetic method must be used within GPME Machining.
The X axis of the graph denotes how many tolerances in the calculation, i.e. T1, T2, T3, etc… the value of
T in each case is 0.25, thus at T=4 the results align to Figures 3, 5 and 7 on the previous pages.
Figure 8
• Arithmetic.
• Statistical x 1.5 safety factor.
Statistical stack-ups without a safety factor SHALL NOT be used within GPME Machining.
The following consideration shall be made when determining which method to use. When these
considerations are met use the Statistical x 1.5 safety factor, otherwise use the Arithmetical method.
Constructing Stack-Ups
The Loop Method
The mathematics of tolerance stack-ups are simple and consistent. However, the complexity is in
interpreting the process so that all of the correct elements are included. For this element of the stack-up
process the ‘Loop Method’ is often employed.
The Loop method is an approach to constructing stack-ups linking all the dimensions from the start point
to the end point in one continuous loop. The links in the loop provide a graphical representation of the
stack path.
• Denote the start point and end point of the stack-up, i.e. the element that the variation is
being calculated for.
• Join these two elements with a line terminating with a circle at each end.
• List all the Process Steps that fall between the start point and the end point.
• Be sure to include all steps, take special care when there are changes in datum
through the process.
• Bear in mind some of the steps may be in different Operations or in some cases,
considering incoming stock conditions, different Production Lines.
• Show the Tolerance, the Datum, the Op Number and the Stack-up Step ID for process
step.
• With the aid of a schematic diagram, construct a loop diagram, linking each process step.
• Place a diamond at the start point of each step arrow head at the end.
• Note that some process steps may have two elements, such as hole position and
radius.
• When all process steps are linked there should be a continuous loop from the start point
to the end point.
• The stack-up path is the shortest continuous chain of dimensions from the start point to
end point of the dimension in question.
The subject of the stack-up (VA) is the variation in radial stock between two passes of the same hole,
indicated by the start and end point of the stack-up.
There are two process steps involved in the stack-up, each of which have two elements. T1 & 2 are
performed in Op 0010 and T3 & 4 are performed in Op 0020. Both step use a common datum, ABC.
The loop has been constructed from start to finish linking all the process steps. The links have been
highlighted in yellow, show that the stack-up path is the shortest continuous chain of dimensions from the
start point to end point of the dimension in question.
Highlight of Loop
Figure 9
Figures 10 and 11 above depict a drill (red) and milled face (blue) both dimensioned to a common
datum. Figure 10 has tolerances that add up (stack-up) to no more than the +/- 0.4mm Part Print hole
depth. Therefore the component at the end of the line will always be OK so long as the individual process
steps are OK. Figure 11 has tolerances that add up (stack-up) to more than the +/- 0.4mm Part Print
tolerance. Therefore there is no certainty that the component at the end of the line will be OK, even if the
process steps are OK.
Figure 12 shows how a stack-up can be used to determine the variation that will enable the correct in-
process tolerance to be applied to avoid this condition.
Highlight of Loop
Figure 12
jn 0.4 PP jn 0.4 PP
Figures 13 and 14 above depict a two related holes, drilled to a common datum, but in different set-ups.
Figure 13 has tolerances that add up (stack-up) to no more than the jn0.4mm Part Print hole position.
Therefore the component at the end of the line will always be OK so long as the individual process steps
are OK. Figure 14 has tolerances that add up (stack-up) to more than the jn0.4mm Part Print hole
position. Therefore there is no certainty that the component at the end of the line will be OK, even if the
process steps are OK.
Figure 15 shows how a stack-up can be used to determine the stock that will enable the correct in-
process tolerance to be applied to avoid this condition.
Highlight of Loop
Figure 15
The same example could be used when a tool is expected to clean up the bottom of a hole; only in that
case, both tools must always overlap at the extremes of tolerance. In this instance it would be considered
and in-process stack-up.
Figure 24 shows how a stack-up can be used to determine the depth variation that will enable the correct
part print depth to be calculated to avoid this condition.
Highlight of Loop
Figure 24
Note – This stack-up is effectively an in-process stack-up, but may be determining a part print dimension.
This is why stack-ups should be completed in conjunction with PD, as quite often the outcome has an
impact on the part print.
Figures 16 and 17 depict a semi-finish milled face (blue) followed by a finish milled face (red). Both
figures have the same part print and in-process tolerances (not shown). Figure 16 the finish milled face
has stock to remove at all tolerance conditions i.e. there is no overlap between the semi-finish and finish
passes. Figure 17 shows that at the extremes of tolerance there is not always stock for the finish mill to
remove; resulting in non-clean-up (NCU), i.e. the semi-finish and finish passes overlap. The only
difference between the two figures is that the nominal stock between the two passes is insufficient in
Figure 17.
Figure 18 shows how a stack-up can be used to determine the stock variation that will enable the correct
in-process size to be calculated to avoid this condition.
Highlight of Loop
Figure 18
Figure 21 shows how a stack-up can be used to determine the stock variation that will enable the correct
in-process size to be calculated to avoid this condition.
Note that is in this instance the number of elements in the stack-up has met the criteria of four or more, so
the result for the statistical stack-up with 1.5 safety factor has been included.
Highlight of Loop
Figure 21
Process Variation
Types of Variation
Whilst it is true that product definition (tolerances) must be based upon the functional requirements of the
product. Variation is a factor of machine selection and process performance, and it is an unavoidable fact
that manufacturing processes always have variation.
Tolerance is the total amount a feature is permitted to vary. It is the difference between the maximum
and minimum limits.
Variation is the actual deviation of a feature from its nominal size, shape, location or orientation.
Tolerances stack-ups as described in this training utilise tolerance limits, not process variation. There are
more complex methodologies that take into account actual process variation to enable a greater amount
of optimism (or if based on data, realism) to be used to reduce the calculated process variation, however
these methodologies are not approved for use with GPME machining.
However, an in-depth understanding the sources of variation is still an essential part of process planning,
as when it comes to applying a ‘Single Fixture Factor’ this information is key.
Size Variation
Form Variation
Macro
Variation
Location Variation
Orientation Variation
Geometric
Variation
Waviness
Roughness
Micro
Variation
Surface Discontinuities
Edge Variations
Figure 25
Macro Variation
Machining stack-ups almost exclusively consider the tolerances associated with Macro Variation. As
shown in Figure 25 they are;
• Size Variation
• Form Variation
• Location Variation
• Orientation Variation
Size Variation is one of the two primary considerations in completing a stack-up. Where size
variation is present, this must be included in the stack-up.
Form Variation is rarely included in a tolerance stack-up as according to GD&T standards form
is contained within a regular feature of size. What this means for example is, the cylindricity of a
hole cannot extend beyond the diameter tolerance, thus if the full diameter range is included in
the stack-up, the maximum form error is already accounted for.
An exception to this is where the form may influence the process outcome, such as honing and
other non-rigid abrasive processes. As the honing tool expands and contracts with the size of the
hole, it will also expand and contract with variations in form. For this reason form must be taken
into account when calculating stock removal for a hone.
Location Variation is the second of the two primary considerations in completing a stack-up. The
variation in the location of a feature, relative to its datum, must be included in a stack-up.
Orientation Variation is rarely included in a tolerance stack-up as according to GD&T standards
orientation is contained within a location variation.
Micro Variation
Tolerances associated with Micro Variation are rarely taken into account in tolerance stack-ups. The only
circumstance, typically associated with abrasive processes, is when calculating the stock removal for a
very fine finish pass. If it is essential to removal all evidence of the prior pass, then it may be necessary to
take into account micro variations such as surface finish, as depth of the groove in a surface finish
measurement extends beyond the macro variation of the surface.
Sources of Variation
In machining the geometric variations of a part originate from clearances in the machine guide-ways and
bearings, deflection in the machine tool or component, location variation in the fixture, variations in the
tool path, as well as tool setting, tool wear and many more factors.
• Machine
• Clearance between moving parts of slide-ways, bearings and lead-screws.
• Geometric errors in slide-ways, bearings and lead-screws.
• Dynamic stiffness.
• Resolution of the positioning system.
• Cutting Tool
• Wear.
• Variation of size and cutting geometry.
• Tool change repeatability.
• Rigidity.
• Thermal stability.
• Fixture
• Variation in location.
• Wear and contamination of locating surfaces.
• Rigidity.
• Thermal stability.
• Component.
• Variation in physical properties.
• Variation in size.
• Rigidity.
• Thermal stability.
• Stress relaxation.
• Incoming temperature.
• Cutting Media.
• Variation of flow.
• Variation of pressure.
• Variation of mix.
• Variation of temperature (coolant).
• Contamination (coolant).
• Degradation (coolant).
• Environmental Conditions.
• Variation in plant temperature.
• Process Variables.
• Changes in process variable such as, feed, speed and depth of cut.
The extensive list of variations can be summarised into two forms of variations as depicted in Figure 26.
• Location Variation
• Machining Variation
Figure - 26
Locating Variation
Variation between the actual position of the datum feature and the true position of the datum locator. After
a component has been located and clamped the locating variation remains constant until the component
is removed from the fixture, i.e. it is consistent throughout one set-up.
Machining Variation
Variation between the actual position of a machined feature and the intended position of a machined
feature. This is apparent on all features and all processes and is the primary reason that features must
have tolerances.
• Feature Variation.
• Component Variation.
Feature Variation applies independently to each feature or feature type. A primary feature variation is
tool variation. For example, the drill that machines the M6 bolt holes has totally independent variation to
the drill that drills the M8 bolt holes. A bad grind on one may cause the position to wander, and this
positional variation will be isolated to the features drill with that one bad tool.
Datum Methodology
Process Principles
Design Datum – used on the component drawing for purposes of dimensioning.
The ideal is to use a common feature as Design Datum, Process Datum, Fixture Datum and Measuring
Datum. Alignment between these datums, will lead to the greatest available tolerance and thus process
robustness. However, this is often not possible. Each time there is a difference between any of the
datums the additional variation must be taken into account in the tolerance stack-up.
The following principles are the basis for robust process planning
• Features with tight tolerance relationships should be machined in the same set-up
whenever possible.
• If the tolerance stack-up between two features is within the tolerance specification, they
can be machined in separate set-ups.
• When features with tight tolerance relationships cannot be machined in the same set-up,
and the tolerance stack-up between two features exceeds the tolerance specification,
they may be machined by using one of the features as the datum of the other, either by
locating on the primary feature in the secondary set-up or by probing the primary feature
in the secondary set-up.
The use of CNC machines has reduced the complexity of the tolerance stack-ups by allowing multiple
features to be cut in a single set-up, resulting in a reduction of the total number of set-ups within the
process. However, the need for tolerance stack-ups is not eliminated by CNC machining.
When completing stack-ups, for simplicity it is customary to ignore the effect of MMC on the stack-up. Not
because it does not exist, but because the reamed dowel tolerances applied are typically so tight that the
effect on the stack-up is insignificant.
However, there are circumstances where including the additional tolerances associated with the dowel
clearance may be used to advantage, as shown in Figure 27 where the two additional values increase
the quantity of values in the stack-up by two, taking the total quantity of values from less than four to more
than four enabling a statistical stack-up with safety factor to be used, over an arithmetic stack-up.
Highlight of Loop
Figure 27
It should be noted, that the benefit of this is minimal and in many cases will result with the statistical
stack-up with safety factor being greater than the arithmetic stack-up.
However, if the location feature diameter tolerance were to be large, a drilled hole for example, it would
be essential to apply this methodology, not to benefit from the statistical stack-up with safety factor, but to
account for what could be a more significant tolerance.
MMC is also applied to features of size. When completing stack-ups for machining, it is not necessary to
consider the effect of MMC on these features as the tolerance stack-up is not looking at ‘fit’. If however
you are performing an assembly stack-up MMC could be a different consideration.
Changing Datums
When completing a stack-up it is essential to include the datum of each location characteristic in the
column titled ‘Datum’, as well as the set-up in the column titled ‘Operation No’. When completing or
evaluating a stack-up the continuity of the datums must be assessed, to ensure that no process step has
been omitted.
Figure 28 shows this correctly applied to a simple stack-up between two passes completed in separate
set-ups to the same datum.
Highlight of Loop
Figure 28
Figure 29 shows this in-correctly applied to a stack-up between two passes completed in separate set-
ups to different datums. It is not allowed to jump from one datum to another without including the
additional step of machining the subsequent datum. It can be seen on the diagram on the left, that the
path of the stack-up is also discontinuous. It is not possible to get directly from one characteristic linkage
point to another via a vertical or horizontal line.
Highlight of Loop
Figure 29
Figure 30 shows this correctly applied to a stack-up between two passes completed in separate set-ups
to different datums. It is the same basic stack-up but the additional step of machining the subsequent
datum has been included. It can be seen on the diagram on the left, that the path of the stack-up is now
continuous. It is possible to get directly from one characteristic linkage point to another via a vertical or
horizontal line.
Highlight of Loop
Figure 30
Whilst this example shows the datum change on a tolerance stack-up for stock variation, the exact same
method would be applied to any stack-up that a datum change was applied to.
Adapter Plates
A number of Prismatic components used adapter plates. The purpose of an adapter plate is to take a
range of different component locations, and present them to a common fixture set-up (and gantry and
automation). This strategy allows the line to accommodate a greater range of different components, with
relatively simple fixtures, than may be the case with free-state component fixtures.
Figure 44 shows how the adapter plate is considered within the tolerance stack-up.
The first step is to show the adapter assembly as a process step indicating change of datum, just like any
other change of datum. Following the rules for continuity of datums as described in the ‘Loop Method”.
However, from here on it is handled differently. The variation associated with the use of an adapter
(usually 0.025mm) is listed in the tolerance column, but it is not transferred over to the stack-up. It is not
actually used in the stack-up at all. Instead the value is deducted from all subsequent position tolerances
related to the datums that the adapter plate is attached to. The reason for this is that the variation of the
adapter plate, is an unknown value to the process, thus if it were not deducted, when a part is measured
off the adapter plate it may be out of specification by an amount equal to the adapter plate variation.
B
Stack-up start / end point
Highlight of Loop
Figure 44
A drill typically has a greater position tolerance that a reamer, though when in a common set-up the
location variation is equal to both tools, thus the difference is down to the machining variation.
Figure 36
For the purpose of this example the assumption is that the location variation is 0.1. Remember, this is
equal to all features.
Applying the location variation of 0.1 to the reamer, the remaining machining variation is 0.1 mm (0.2-0.1).
Applying the same location variation to the drill, the remaining machining variation is 0.3 mm (0.4-0.1).
It is a reasonable assumption that the drill will have a less machining variation at entry than it will at full
depth. For the purpose of this example the assumption is that the entry variation is half the full depth
variation. Bearing in mind that the full depth variation can be no greater that the remaining machining
variation, the machining variation at the entry would be 0.15 (0.3/2).
The single fixture factor for the drill at entry is 37.5% (0.15/0.4*100).
The single fixture factor for the drill at full depth is 75% (0.3/0.4*100).
Note – The values used in this example are purely to demonstrate the theory. When applying this
principle in practice tolerance analysis or surrogate data must be used. This is explained later in this
chapter.
Firstly the single fixture factor may not be applied to the diameter or any other in-tool relationship, as
there is no location variation associate within in-tool relationships, so these are left at 100%.
Looking at the two positional tolerances in Figure 37, a single fixture factor of 50% may be applied to the
reamed hole position, and a single fixture factor of 37.5% may be applied to the drilled hole position. It
would be wrong to ignore this opportunity, as the location variation associated with both the tools is equal,
thus it can be eliminated from any stack-up of the relationship between the two features. The rational for
determining the Single Fixture Factor is covered on the following pages.
Nominal Nominal
GDT
(diametral) (Radial)
Figure 37
Figure 38 shows a typical example where the chamfer is made by the drill, but the finish chamfer depth is
a combination of the drill position and depth, as well as the face profile, the reamer position and the
reamer diameter.
Figure 38
Figure 39 shows the stack up associated with Figure 38 to determine the resultant variation of the
finished chamfer. It can be seen that due to the number of tolerances involved arithmetic variation of the
finish chamfer is almost three times that of the in-process chamfer. Even statistically with a safety factor, it
is more than double. Now if these process steps were carried out in different set-ups, this stack-up would
be valid, however if they are carried out in the same set-up, then the single fixture factor can be applied.
Tolerance Single Fixture Arith. Radial Stat. Radial PP / IP Process Step Operation Stack-up
GDT Datum
(diametral) Factor Stack-up Stack-up / FIP Characteristic Description No Step ID
Figure 39
Note – The loop method has not been applied in this case. This is because variation has been taken into
account in two planes. This is outside the scope of this training material.
Applying the same location variation as in the earlier example, 0.1, the fixture factor can be calculated.
Note the fixture factor is the percentage of total tolerance accounted for by machining variation.
Figure 40 shows how this has been calculated for each of the tolerances from the prior example.
Figure 40
Figure 41 shows the stack up associated with Figure 40, with the single fixture factor applied. It can be
seen that this has had a significant impact on the resultant variation of the finished chamfer.
Tolerance Single Fixture Arith. Radial Stat. Radial PP / IP Process Step Operation Stack-up
GDT Datum
(diametral) Factor Stack-up Stack-up / FIP Characteristic Description No Step ID
Figure 41
Figure 42 shows a typical application of a composite tolerance. The feature control frame on the left says
that each hole has a position tolerance of 0.2 to ABC, and then a secondary tolerance of 0.1 to one
another. See ASME GD&T Standards for a full explanation.
The 0.1 relationship between the two holes has no location variation, thus this is accounted for entirely by
machine variation. The 0.1 tolerance is split evenly between the two holes, this it can be deduced that
each hole has a machining variation of 0.05 (+/- 0.025)
With the machining variation of 0.05, it can be calculated that the location variation is 0.15 (0.2-0.05)
Figure 42
Figure 43 a typical application of a composite tolerance. The feature control frame on the left says that
each hole has a position tolerance of 0.2 to ABC, and then a secondary tolerance of 0.1 to one another.
See ASME GD&T Standards for a full explanation.
Figure 43
• Can be determined by measuring the relationship between either R1 or R2 and the fixture
datums.
• This position data will include both machining variation and location variation.
• The location variation can be isolated by deducting the machining variation established
above.
Note – when applying this method it is likely that short term data may be used. Caution must be applied
when basing tolerance decisions on short term data alone.
When performing stack-ups to determine an in-process nominal, it is important to understand that the
primary purpose of the stack-up is to determine the in-process nominal that is correct for the tolerances,
not the tolerance remaining to suit an in-process nominal.
Tolerances should be determined on the basis of data that supports the process capability. Those
tolerances should be used as inputs to the stack-up. Following this principle ensures robust processes.
The minimum clean-up should be determined based on what minimum stock is required for the tool. If the
stack-up is allowed to trend to zero stock at the extreme, it is unlikely that a quality will be achieved.
The minimum stock allowance, as agreed with the tool manufacturer, should be used as an input to the
in-process size calculation.
Once the in-process size has been determined with proven in-process tolerances and recommended
minimum stock allowance, the maximum stock allowance should be calculated. An assessment must be
made to determine whether the tool is likely to make a quality part at both extremes of tolerance. If the
assessment is that the tool will not make quality, then there are 3 potential course of action.
• Evaluate the opportunity for a process sequence change. The total tolerance involved
may be reduced by ensuring that both passes are machined from the same datum, or
within the same set-up.
• Evaluate the addition of an extra pass. It may be necessary that the first pass has to be in
the cubing / roughing process and the last pass in the finish operation. If under these
circumstances the tolerance range is too great, then the solution may be a semi-finish
pass in the same set-up, or from the same datums, as the finish pass.
• Evaluate the opportunity to increase the finish part print tolerances.
• Evaluate the opportunity to reduce the in-process tolerance. When evaluating this
opportunity it is essential that whatever tolerances are selected the selection is based on
data that ensures robust process capability, not simply what tolerance fits the stack-up.
Whether this is through application of better tooling, a different process, or ‘better’
surrogate data.
The drill is shown at its maximum diameter at the extreme of its position tolerance to the right.
The reamer is shown at its minimum diameter at the extreme of its position to the left.
These extremes of tolerance indicate the worst arithmetic condition, i.e. those most likely to lead to a non-
clean-up (NCU) condition. A minimum radial stock of 0.25 has been included to avoid cutting close to
zero stock.
To aid visualization of the stack-up, the part print and in-process tolerances in Figure 32 are shown in
alignment with the stack-up in Figure 33.
Figure 32
Figure 33 shows the values T1 to T4 input into an arithmetic stack-up. It must be noted that when
calculating a stack-up for the in process size of a hole, the radial element of the part print and in-process
tolerances are used as the stock removal is radial, as depicted in Figure 32.
The maximum arithmetic variation (VA) is equal to the sum of the radial elements of T1, T2, T3 and T4.
Thus the maximum radial variation (VA) is 1mm.
Highlight of Loop
Figure 33
The stock variation alone is not enough, when calculating the in-process size of a hole. The final step is to
deduct the stock variation from the Outgoing Mean (OM), to establish the Incoming Mean (IM). The
Outgoing Mean (OM) is typically the finish part print diameter, though in some circumstances where there
are 3 passes, this could be the semi-finish diameter.
As discussed previously the whilst the purpose of the stack-up is to ensure that there is no Non-Clean-Up
(NCU) in actual fact a reamer that is cutting close to zero stock removal, will not deliver a quality part
either. For this reason it is important to determine the minimum allowable stock removal, and add this into
the equation.
Figure 34 shows the exact same stack-up with the addition of the Outgoing Mean (OM) and minimum
radial stock, enabling the calculation of the Incoming Mean (IM), which is the required in-process drill size.
Highlight of Loop
Nominal Nominal
GDT
(diametral) (Radial)
Figure 34
When completing a stack-up for minimum stock removal, it is important to also consider the opposite
condition, as whilst guarding against NCU is the primary objective, it is also important not to overload the
tool/component with too much overall stock, or too great a variation between stock conditions, say from
one side of the hole to the other, which may induce form error.
Typically Process Engineers are concerned with ensuring that the tool has the minimum amount of metal
to be removed, as this generally ensures better quality and better tool life. Therefore any unnecessary
stock may be detrimental to quality and economics. As the minimum depth of cut is pre-determined by the
requirements of the tool, reducing the maximum depth of cut can only be achieved through reducing the
perceived process variation of the stack-up. Therefore performing a statistical stack-up with safety factor
instead of an arithmetic stack-up may improve both quality and economics, by allowing an increase in the
size of the in-process drill.
Figure 35 shows the same stack-up but with the addition of the statistical stack-up with a 1.5 safety factor
(VSF).
Highlight of Loop
Figure 35
Whilst it is not so easy to visualize the Statistical stack-up in the same way as the arithmetic stack-up, it is
clear to see the scale of reduction in variation, allowing a 20% reduction in maximum radial stock, for the
same minimum radial stock, and the same tolerances which is a worthwhile improvement.
Note – the tolerances used in these examples have been chosen for their mental arithmetic ease and do
not represent realistic part print or process tolerances.
When a nominal stock allowance is applied, a stack-up is used to verify that the amount of variation does
not exceed to the nominal stock allowance.
Figure 31 shows a stack-up to verify the nominal stock allowance. As most rough milled faces are
typically part of the cubing process and finish milled faces part of the final operation, they are mostly
machined to different datums. Thus a datum shift is included in this stack-up.
As the stock allowance is a nominal the +/- 0.4 variation is applied to the nominal. Thus in this case to
ensure that the finish pass cleans up, the stock allowance has to be greater than 0.4. Both the 0.5mm
and the 0.7mm allowance are OK in this case.
Highlight of Loop
Figure 31
If it is necessary to minimise the stock, then the stack-up variation plus a minimum machining allowance
could be applied as the minimum stock.
In this case it is not allowed to apply a statistical stack-up with safety factor, as there are less than 4
elements.
The first issue being, generally neither the finished face, nor the in-process face is the process datum for
the hole. So the stack-up must consider the relationship between the hole and the process datum.
The second issue is that though the part print has the depth back to the finished face, typically the
finished face is not there when the hole is drilled, leading to the hole depth needing to be dimensioned to
an in-process face. So the stack-up must consider the variation of the in-process face.
The purpose of these three diagrams is to show the progressive loss in available hole depth tolerance
due to the use of an in process face, and the effect of this loss on the ability to measure the hole depth.
• The part print hole depth tolerance of ± 0.6, less the finish milled face tolerance of ± 0.1.
• This is the maximum available tolerance given to the hole depth.
Condition 2 – Has reduced the tolerance available to the hole depth by a further ± 0.1, by referencing the
hole depth to the in-process face, that has ± 0.1 variation of its own.
• Due uncertainty of the measurement reference face, when the gauge is set on a face that
has ± 0.1 potential variation, then that potential variation is deducted from the available
tolerance for the hole, ensuring that the component is still within specification when
measured from that face.
• The apparent tolerance for the hole depth is now only ± 0.3, as when the gauge with ±
0.4 tolerance is set on the face with ± 0.1, it has the same ± 0.1 uncertainty, only this time
it may reject OK parts.
• This is clearer to visualize in Condition 3
Condition 3 – Has halved the tolerance available to the hole depth, by referencing the hole depth to the
in-process face with a tolerance of ± 0.25.
• Now when you sit the gauge with a tolerance of ± 0.25, on a face that is varying by ±
0.25, it can be seen that the entire gauge tolerance could be taken up by variation of
milled face and any slight deviation from nominal on the hole depth would be seen as out
of specification by the gauge. Clearly not something that you want.
It can be seen how a generous tolerance for hole depth, can soon be eaten up by process choices, so
this requirement drives the need for comparatively large tolerances on hole depth.
Often as these features do not really matter and designers leave them as open part print. Which
is generally ± 0.25, so it can be seen how that actually becomes a tight tolerance, resulting
manufacturing failing to achieve something the designer didn’t care that much about.
Figures 46 to 48 show how the pictorial explanation in Figure 45 can be translated into stack-ups.
+/- 0.6 PP
+/- 0.1 PP
+/- 0.4
+/- 0.3
+/- 0.25
+/- 0.00
Figure 45
Figure 46 shows a conventional stack-up for this type of feature. The stack-up shows
• The part print tolerance for the hole depth (VA) has been shown as ± 0.6
• The part print tolerance for the milled face (T2) has been shown as ± 0.1 (0.2 Profile)
• Both the milled face and the drilled hole are datumed back to ABC, which is the part print
datum for the milled face and the process datum for the drill hole.
• The resultant in-process tolerance for drill hole depth (T1) is ± 0.5
Highlight of Loop
Figure 46
This approach is perfectly acceptable and as shown in Figure 46 yields the largest in-process tolerance
for hole depth. However, there are many reasons why this approach may not be possible.
• As the hole depth is dimensioned back to the datum, a simple flush pin gauge can’t be
used.
• For a drill hole either a stand gauge or a CMM could be used.
• For a tapped hole, only a stand gauge could be used as a CMM is unable to measure the
depth to the bottom of a thread.
• Stand gauges are none preferred on flexible machining lines.
• For these reasons it is often preferred to measure this type of feature with a flush pin
gauge.
Figure 47 shows how this stack-up is constructed to allow the use of a flush pin gauge. The stack-up still
follows the Loop Method, but is different for the following reasons.
• The part print tolerance for the hole depth (VA) has been shown as ± 0.6
• The part print tolerance for the milled face (T2) has been shown as ± 0.1 (0.2 Profile)
• The in-process tolerance for the semi-finish milled face (T3) has been shown as ± 0.1
(0.2 Profile)
• Both the finish milled face and the semi-finish milled face are datumed back to ABC,
which is the part print datum for the milled face and the process datum for the semi finish
milled face.
• The resultant measurement tolerance (M1) for drill hole depth (M1) is ± 0.4
• The apparent in-process tolerance (T1) for the drilled hole is ± 0.3
Highlight of Loop
Figure 47
When the tolerance T3 exceeds the apparent tolerance T1 the stack-up is deemed unacceptable.
It could be argued that any time a gauge could reject a good part should be deemed unacceptable.
However this gauge method has been successfully used for many years as the discrimination of a flush
pin gauge it not precise.
That said caution must be exercised when applying this approach, as it is very easy to mistakenly create
a gauge that will either accept out of specification parts, or reject good parts.
Figure 48 shows the exact same stack-up but with a semi-finish milled face tolerance (T3) of ± 0.25. The
result of this tolerance is that the measured tolerance (M1) has reduced to ± 0.25, which is appears
reasonable; however the apparent in-process tolerance for the drilled depth is zero, which is
unacceptable.
This is due to the entire available in-process variation being taken up by the semi-finish milled face.
The following are the options that may resolve this situation
• Reduce the tolerance of the semi finish milled face, as in Figure 47.
• Complete a statistical stack-up, by adding the MMC of the location dowel.
• Change the gauging method, so that the full in-process tolerance is available, as in
Figure 46.
• Increase the part print tolerance.
Highlight of Loop
Figure 48
Figure 49 shows the same stack-up as Figure 48, but this time with the addition of datum change. There
are now enough tolerances to complete a statistical stack-up with safety factor so this column has been
completed.
The addition of the datum change has reduced the measured dimension (M1) from ± 0.25 to ± 0.15, and
left an apparent hole depth (T1) of -0.1, which is unacceptable.
When looking at the statistical column of same stack-up the measured dimension (M1) increases from ±
0.15 to ± 0.278 and the apparent hole depth tolerance for the drilled depth (T1) is ± 0.178. This is better
but still not enough tolerance to be deemed acceptable.
Figure 50 shows the same stack-up as Figure 49, but this time the in-process milled face tolerance (T3)
has been reduced from ± 0.25 to ± 0.2. The apparent hole depth tolerance (T1) is now ± 0.2, which is now
greater than (T3), thus acceptable.
Highlight of Loop
Figure 49
Figure 50