Paper ID #25816

Tolerance Stack-up Analysis in Manufacturing-Based Capstone Projects

Dr. Mustafa Shraim, Ohio University
Dr. Mustafa Shraim is an Assistant Professor in the Department of Engineering Technology & Man-
agement at Ohio University in Athens, Ohio. He received both of his B.S. and M.S. degrees from Ohio
University in 1986 and 1989, respectively. He received his Ph.D. in Industrial Engineering from West
Virginia University in 1996.
Dr. Shraim’s research interests are in the area of quality engineering. Specifically, they cover Lean
/ Six Sigma methods and including incorporating experimental design to optimize operations. Other
research interests include the Deming System of Profound Knowledge (SoPK), developing continuous
improvement programs as well as sustainable management systems based on ISO 9001, ISO 14001, and
other international standards.
He has over 20 years of industrial experience in the quality management field as a quality engineer,
corporate quality manager, consultant and trainer. His experience is extensive in quality management
systems as wells as Lean and Six Sigma methods. In addition, he coached and mentored Green & Black
Belts on process improvement projects in the manufacturing and service industries.
Dr. Shraim is a Certified Quality Engineer (CQE) & a Certified Six Sigma Black Belt (CSSBB) by The
American Society for Quality (ASQ). He is also a certified Quality Management Systems (QMS) Lead
Auditor by the International Register of Certificated Auditors (IRCA) in London. He was elected a Fellow
by ASQ in 2007.
Dr. Neil Littell, Ohio University

Dr. W. Neil Littell is an Assistant Professor at Ohio University within the Russ College of Engineering
in the department of Engineering Technology and Management. Dr. Littell earned a Doctorate of Phi-
losophy in Instructional Systems and Workforce Development (2013) from Mississippi State University.
Dr. Littell also received a Masters in Technology from Mississippi State University (2005). Addition-
ally, he holds Bachelor of Science degrees in both Industrial Technology and Trade and Technical Studies
from Mississippi State University (2004). Dr. Littell also has an Associate of Applied Science degree in
Drafting and Design from Holmes Community College (2002).
Dr. Littell is an accomplished manager with more than 10 years of experience providing results-oriented
leadership. His previous positions include the PLM Coordinator at the Center for Advanced Vehicular
Systems at Mississippi State University from 2004-2008. He was Director at Large for COE, the World’s
largest users group of Dassault Systemès PLM products from 2008 to 2012, where he was acknowledged
with the BJ Fries Award of Merit for making balanced contributions to the organization’s activities and
growth. His most recent position was as the Engineering Program Manager and CAD/PLM Administrator
at Viking Range LLC, located in Greenwood, Mississippi from 2008 to May 2014.

American
c Society for Engineering Education, 2019
 Tolerance Stack-up Analysis in Manufacturing-Based Capstone Projects

Abstract
Most manufactured products typically consist of multiple components assembled in specified
configurations. Such components have specifications for important dimensional characteristics to
ensure adequate performance once assembled. The specifications are typically given as
tolerances within which components must be produced. Consequently, parts produced outside
the specifications are rejected and not used in the assembled product. As dimensional variation is
to be expected in produced components, it may accumulate, or stack up, to cause unwanted
variation in the assembled product. When this happens, additional costs in scrap and rework
would result and may extend to warranty charges and customer dissatisfaction should
unacceptable products find their way to the customer.
In this paper, issues resulting from tolerance stack-up in capstone production will be addressed.
The questions this paper will attempt to answer are: (1) Are students aware of the impact of
tolerance stack-up in product development, and (2) What can be done to minimize the potential
effects of tolerance stack-up before product launch.

Introduction
A tolerance is defined as the range of acceptable variance from the nominal value of a
dimensional characteristic while still allowing the part to fulfill its functional requirements. This
concept is relatively straightforward for engineers, technicians, and students to understand, the
phenomenon of tolerance stacking is not as easily understood. Tolerances might be expressed
bilaterally (in both positive and negative directions from the nominal dimension) or unilaterally
(in only one direction from the nominal dimension). In general parts with greater dimensional
tolerances allow companies to manufacture parts less expensively, because the dimensional
requirements can be fulfilled using less precise tooling and equipment.
During the engineering design phase of the new product development process, design engineers
develop the three-dimensional computer aided designs (CAD) for a product. This geometry is
designed at nominal dimensions, that is, the system of parts is digitally designed based on
theoretically perfect parts. However, the implementation of perfect parts is not reasonable
because of myriad reasons and therefore the design engineers must develop a tolerance. This
tolerance is generally determined by two things: 1.) Part functionality and 2.) process capability.
To be successful, design engineers must respect the process capability of their company or
supplier, or they risk specifying parts that are expensive due to tolerances which are too
constraining [1] [2] [3].
Just as parts have standard tolerances, assemblies comprised of those parts have a tolerance as
well. The total tolerance for the assembly is directly driven by the tolerances of the parts which
comprise the assembly. For example, if the assembly were comprised of parts which are all at the
upper end of their total tolerance, the assembly might not function as intended. Therefore, design
engineers must be mindful of the total tolerance with respect to the assembly. This phenomenon
is known as tolerance stacking. However, the implementation of a proactive plan to avoid
production issues caused by assemblies which are out of tolerance due to tolerance stacking is
not as straight forward to understand because the effects of tolerance staking are a result of
interactions between parts. Once production begins, this problem is further compounded because
of interactions across parts with respect to process capability [2].

Perhaps the easiest way for designers to combat the possibility of tolerance stack interactions is
by using the worst-case method of tolerance stack analysis. When completing this type of
analysis, the designer considers the extreme conditions of every part involved in the assembly.
For example, if each part in an assembly is at the upper end of the tolerance; would the assembly
still function with respect to the designed requirements? Alternatively, the designer could
consider the situation where all parts are at the lower end of their tolerance and ask the same
question. While this is perhaps the easiest way to perform tolerance stack analysis, it is not the
most efficient from a perspective of cost savings. The worst-case scenario for tolerance stack
analysis also assumes that if an assembly functions at both ends of the spectrum (largest and
smallest) the middle will take care of itself [1]. In reality, the performance of an assembly of
parts may be significantly degraded as it approaches either the upper or lower end of the
tolerance spectrum. Design engineers must be careful communicating the concept of total
tolerance of an assembly to avoid the perception that product performance is equal across the
tolerance spectrum [4]
As an example, refer to Figure 1. All of the example parts in this illustration have a standard
tolerance of ± .015. This means that the largest vertical dimension of PART A could be 2.015,
while the smallest acceptable vertical dimension of PART A could be 1.985. When this bilateral
tolerance of ± .015 is applied across the system (assembly) of parts, one can see the stacking
effect of the part tolerances against the total tolerance. In this case, the worst case (largest)
tolerance of the assembly is 10.000+.015+.015+.015+.015+.015=10.075, while the worst case
(smallest) tolerance of the assembly is 10.000-.015-.015-.015-.015-.015=9.925, Therefore, the
total tolerance range of this assembly is 9.925 -10.075 or 10.000 ± .075. This is achieved with
each part involved in the assembly being within the part tolerance.

Figure 1: Tolerance Stack Illustration

As companies seek to produce parts of high quality as well as lower costs, the tolerances of parts
become a driver of both cost and quality. Additionally, different configurations of parts may
drive tighter tolerances because of requirements within different applications using the same
part, further compounding production issue. As companies design products with advanced
features and shorter design cycles, the role of tolerance stacking becomes a more urgent matter
for design engineers to address early in the design lifecycle [5].
Researchers have explored tolerance stack-up analysis using different methods or a combination
of methods. In addition to the conservative method of worst case (WC) method, the root sum
squares (RSS) which started in the 1950s continued to be explored for a variety of stack-up
conditions and modifications. M. Spotts suggested adding a safety factor by averaging the RSS
and WC methods [5]. Additional RSS models were suggested by others to account for mean shift
during production [6] [7]. Since the RSS assumes that component data follows normal
distribution, more research was conducted for data that follow other distributions such as, but not
limited to, uniform distribution, triangular distribution [8] [9] [10].
This study seeks to explore student understanding of the role of tolerances as a component of
their educational experience at a Midwestern university within an engineering technology
program. Students in the capstone course are often engaged in designing products that are
assembled of several components. In some cases, components are chained together and could
create some issues for fit and function for the assembled product should the potential of
accumulation of variation be ignored. As a result of this paper, a process for conducting
tolerance stack-up analysis will be developed and tested with students to determine its
effectiveness.

Methodology Development
Only two popular alternatives are introduced in this study; the WC and RSS. Generally,
engineers and designers would rather have tight tolerances to ensure fit, function and better
quality of the assembled product. On the other hand, the aim of manufacturing is to introduce
products so tight tolerances may cause issues when variation exists in the process.
As shown in Figure 1, the five components (A, B, C, D, and E) are assembled together create the
final product. If the worst-case scenario is used, the assembled product would be assumed to
have as large of the tolerance as the sum of the individual tolerances or ±0.075 inches. If the
designer would like to keep the tolerance as ±0.050, then tighter tolerances for components
would be sought. If this is applied to all components equally over the five assembled
components, then the tolerance of each component will have to be set to ± 0.010 inches which
would increase the cost of components, perhaps unnecessarily.
Using the RSS method, the variation of the assembled product is calculated using the squared
values of the individual components. Let ±Ti denotes the individual tolerance, the assembled
tolerance, TA, can be given by Equation 1.

𝑇 = ∑ 𝑇 (1)

To illustrate, consider the example in Figure 1 where the individual tolerances are ±0.015 inches
for each of the five components. The assembly tolerance can be calculated to be ±0.034 inches
using Equation 1 by taking the square root of the sum of squared tolerances. Compare this to the
assembly tolerance obtained by WC method of ±0.075. Assuming the designer’s tolerance of
±0.050, it is within the desired limits.
While WC method is non-statistical in nature, the RSS method is based on the normal
distribution with the variation expressed in units of standard deviations (σ). The normal
distribution spread is based on adding / subtracting three standard deviations on each side of the
mean (μ) or (±3σ) which would include 99.73% of the data (Figure 2).

Figure 2: Normal Distribution

In tolerance stack analysis, a reliable estimate of the standard deviation of each of the chained
components will be needed. This can be accomplished using one of the following methods:
1. Production data: This is the most reliable method as the standard deviation may be
calculated from available data which may be in the form of inspection records or
statistical process control (charts).
2. Similar products using the same process: When a new product is being introduced with
no production data history, it may be necessary to see how similar products performed
over time.
3. Engineering tolerances: This method is commonly used to determine the assembly
tolerance. Two-sided specifications are equated to the process spread to estimate the
variation. For a symmetrical tolerance, one side of the tolerance is equated to three
standard deviations. This also depends on the capability desired for the assembled
product.
It should be noted here that in many cases, some components of the assembly may already be
utilized as components in other assemblies. Therefore, production data for those components
may already be available. Therefore, the use of a combination of the above methods is also
possible.
Whenever available, estimating variation from available data is preferred given that the process
is stable. Absence of process stability (having special causes of variation) may cause problems in
reliably predicting process performance over time, at least in the foreseeable future [11]. When
data from similar products engineering tolerances are utilized, validation of data should occur
using a limited production run. Further validation and update of tolerances should occur when
full production data is available as well.
Once the component tolerances are reliably estimated, the analysis can be conducted to
essentially answer two questions: (1) does the current assembly tolerance meet requirements?
and (2) which components should be targeted to tighten tolerances if needed? To answer the first
question, the analysis of rejection fraction (or parts per million) may be performed using the
standard normal distribution. This should be done with some allowance for process mean shift
during production of chained components. Utilizing the Six Sigma methodology in estimating
the fraction nonconforming with 1.5 standard deviation allowable shift in the long run, the 4σ
capability would result in a rejection rate of 6.7% or 66,807 defects per million opportunities
(DPMO). This can be calculated using the standard normal distribution with Z=2.5 standard
deviations (4.0 -1.5). When the process is operating at the Six Sigma capability, it is actually at
the 4.5σ with the mean shift is accounted for and the rejection rate is 0.00034% or 3.4 DPMO.
Table 1 displays other capability levels.
The second question as to which components should be targeted to reduce variation is not
obviously decided based on the variability contribution of each component to the overall
assembly, but also other variables including costs. Methods for such allocations have been
introduced in the literature as demonstrated by Chase et al [2]

Table 1: Six Sigma Capability Scale

Sigma % Rejects DPMO

Level
3σ 6.7% 66,807
4σ 0.62% 6,210
5σ 0.023% 233
6σ 0.00034% 3.4

When variation is excessive in the components, it is reflected in the resulting assembly and a
fraction of which could be rejected. To improve performance of the process and reduce the
overall assembly variation, it is imperative to identify which components need to be tackled first.
This can be achieved by calculating the contribution each component makes to the overall
variation using the variances. For example, to calculate the contribution of any of component,
Equation 2 can be used:

% 𝐶𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝐶𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 = 𝑥 100% (2)

Once contribution for each component is known, order of improvement can be made based on
size of contribution as well as cost of improvement. This cost is not easy to determine given that
costs of dissatisfied customer may not be known. The plan-do-study-act (PDSA) continuous
improvement cycle can be used to as means for carrying out such improvements.

Stack-up Analysis Process

In the preceding section, details about conducting tolerance stack-up analysis was presented. The
aim here is to develop a usable methodology for students when they perform stack-up analysis. It
begins with identifying components and their chained characteristics on the overall (assembly)
characteristic. The individual (component) means of the characteristics are added to determine
the overall mean. Similarly, using the RSS method, the variances are added to determine the
overall variance then standard deviation.
The overall (assembly) variation can then be compared against the tolerance set by the customer
or the designer to determine process capability. This will depend on the sigma scale (quality
level) of interest. If the process is not capable or the variation does not quite meet established
capability requirements, then process or product design changes may be necessary. If capability
requirements are met, the process ends. Figure 3 presents a flow diagram that students will use to
ensure that possible tolerance stack-up issues are resolved early in the product design phase. As
the model shows, if no production data is available for the components, validation using initial or
limited production run can be performed to ensure that the estimated standard deviations were
similar and process capability requirements are met.

Figure 3: Proposed Stack-up Analysis Process

 Figure 4: Determining Means & Standard Deviations

Model Testing
The aim of these tests is to validate the adequacy of the topic coverage and the provided model
(flow diagrams) in guiding the students through tolerance stack-up analysis. The students were
divided into their capstone teams and asked to conduct tolerance analysis. Using the models in
Figures 4 and 5 above, teams were asked to:
1. Identify tolerance stack-up potential on their current projects. This will be carried out for
each team on the capstone projects
2. Conduct tolerance stack-up analysis for the following cases:
a. Case 1: Assembled components have both tolerances and production data
An assembly is comprised of 3 components. The parts have been produced before
and both production data and tolerances are available as shown in the Table 2.
Perform tolerance stack-up analysis and make recommendations.
Table 2: Case 1 Tolerance Analysis Data
 b. Case 2: A combination of new components with tolerance only, similar
components with data, and in-production components with data
If the assembly, comprised of 3 components, has only component A available and
in production while components B and C are new. For part C, a similar product
that is slightly larger in size is producing a standard deviation of 0.009 inches
(Table 3). Perform tolerance stack-up analysis and make recommendations.
Table 3: Case 2 Tolerance Analysis Data

Each of the five capstone project teams presented their analysis based on the two questions
asked. For the first question regarding potential tolerance stacking for the product being
designed, teams had their discussions and identified potential tolerance stacking then briefly
presented to the rest of the class. As for the second question with the two cases, students
followed the model presented in the flowcharts of Figures 3 and 4:
 All teams conducted the analysis correctly:
o For Case 1, they correctly used the current statistics provided from production
data and not the tolerance.
o For Case 2, they correctly used production data estimates for component A,
tolerance estimates for component B, and similar product estimates for
component C.
 All but one team correctly identified % contribution of variation by each component for
both cases using Equation 2. One team used standard deviations instead of variances to
make the calculations which was incorrect.
 As a result, all teams made recommendations as to which component to target first to
minimize the variation in assembly.

Concluding Remarks
Accounting for tolerance stacking in manufacturing is a critical skill for graduating engineers to
have. As a deliverable item of this study, the senior capstone course will be modified to include a
section dedicated for tolerance stack-up analysis. This section will include coverage of the topic,
including the process through the developed flow diagrams in this study. Their understanding of
the concepts will be verified by working through a variety of cases. They will also identify any
potential tolerance stack-up in their capstone product.
Future work may include introducing Monte Carlo simulation techniques and compare against
the proposed process. Additional work may include using the Taguchi loss function in
determining which components to target based on total cost to society.

References
[1] Sahani, A; Jain, P & Sharma S, “Geometrical Tolerance Stack Up Techniques”, Chapter 52
in DAAAM International Scientific Book, pp. 857-872, 2013. B. Katalinic & Z. Tekic (Eds.),
Published by DAAAM International, ISBN 978-3-901509-94-0, ISSN 1726-9687, Vienna,
Austria, 2013
[2] Kenneth W. Chase, William H. Greenwood, Bruce G. Loosli, Loren F. Hauglund, Least Cost
Tolerance Allocation for Mechanical Assemblies with Automated Process Selection,
Manufacturing Review Vol 3, No 1, pp 49-59, 1990.
[3] He, J. R., and Gibson, P. R., “Computer-Aided Geometrical Dimensioning and Tolerancing
for Process-Operation Planning and Quality Control”, The International Journal of Advanced
Manufacturing Technology, 7:11-20, 1992
[4] Gavankar, P., and Bedworth, D., “Stacked tolerance analysis and allocation using assembly
models”, Journal of Intelligent Manufacturing (1991) 2, 365-377, 1991
[5] Spotts, M.F., "How to Use Wider Tolerances, Safely, in Dimensioning Stacked Assemblies,"
Machine Design, pp. 60-63, April 1978.
[6] Greenwood, W.H. and K.W. Chase, "A New Tolerance Analysis Method for Designers and
Manufacturers," J. of Engineering for Industry, ASME, v 109, pp. 112-116, May 1987
[7] Greenwood, W.H. and K.W. Chase (1988a), "Worst Case Tolerance Analysis with Nonlinear
Problems," J. of Engineering for Industry, ASME, v 110, pp. 232-235, August 1988
[8] Mansoor, E.M., “The application of probability to tolerances used in engineering design”,
Proceedings of the Institution of Mechanical Engineers, 178 (1), 29–51, 1963
[9] Gladman, C.A., “Applying probability in tolerance technology”, Transaction of the
Institution of Australian Mechanical Engineers, ME5 (2), 82–88, 1980
[10] Chase, K., and Parkinson, A, “A Survey of Research in the Application of Tolerance
Analysis to
the Design of Mechanical Assemblies”, Research in Engineering Design, 3:23-37, 1991.
[11] Deming, W. E. The New Economics. 3rd ed., Cambridge, MA: The MIT Press; 2018.