A Study of Mixed Manufacturing Methods in Sand Casting Using 3D Sand Printing and FDM Pattern-Making Based On Cost and Time
A Study of Mixed Manufacturing Methods in Sand Casting Using 3D Sand Printing and FDM Pattern-Making Based On Cost and Time
A Study of Mixed Manufacturing Methods in Sand Casting Using 3D Sand Printing and FDM Pattern-Making Based On Cost and Time
by
Ram A. Gullapalli
Time
By
Ram A. Gullapalli
I hereby release this thesis to the public. I understand that this thesis will be
made available from the OhioLINK ETD Center and the Maag Library
Circulation Desk for public access. I also authorize the University or other
individuals to make copies of this thesis as needed for scholarly research.
Signature:
Approvals:
for metal casting and especially for parts of large dimensions and low
casting process does have its limitations; one of them mainly being the high
cost of tooling to create molds and cores. With the advent of additive
Previous research shows the cost benefits of 3D sand printing molds and
cores when compared to traditional mold and core making methods. The line
the decision-making process as well as the cost-based effects. This will enable
smoothly into their production process without the need for completely re-
engineering the existing production system. A critical part of this thesis is the
thesis outlines three approaches for achieving this goal apart from traditional
mold and core making methods. The first approach integrates 3D Printing at
the pattern making level where the patterns and core-boxes are “printed” on
i
an FDM printer. This eliminates the tooling costs associated with a traditional
sand casting method. The second approach integrates 3D Printing at the core-
printing process for core-making. The third approach, the 3D sand printer is
used to create both the molds and the cores, thereby eliminating the need for
sand printing. It is finally concluded that the initial hypothesis is valid when
part geometries are highly complex and production volumes range between
medium to high. It is also concluded that a decision making tool based on the
ii
Acknowledgements
First off, I would like to thank my advisor, Dr. Brett Conner, without whom I
would never have had the opportunity to attend school at YSU let alone perform
his constant guidance and support has pushed my research further. I would also
like to thank my defense committee: Dr. Darrell Wallace and Dr. Eric MacDonald
I would also like to thank America Makes, Youngstown Business Incubator and
special thanks to Ashley Martof for providing the costing data on the Fortus 900
machine.
I thank my friends who are the closest to family I have here in the United States.
They have seen me through my ups and downs and were always ready to help
Surya Kumar Gullapalli, who live in India, and have shown unflinching support
iii
Table of Contents
Abstract............................................................................................................................... i
Acknowledgements ........................................................................................................iii
1.3.2 Additive manufacturing for mold and core making: Binder Jetting...... 18
1.4 Implementation and Adoption of AM for mold and core making ............... 20
iv
1.4.3 FDM based pattern-making (FDMP) .......................................................... 22
v
2.4.2 3D Sand Printing (3DSP)............................................................................... 44
3.2.1 Cost per Part vs. Complexity Factor at each production volume........... 56
References........................................................................................................................ 74
vi
List of Tables
Table 1: Complexity factor calculation for each case study and their sub-cases .. 37
vii
List of Figures
Figure 12: Example Calculation for Train Air Brake with 8 Cores ......................... 53
Figure 14: Train Air Brake- Cost per Part vs. Complexity at Q=1 .......................... 56
Figure 15: Train Air Brake- Cost per Part vs. Complexity at Q=30 ........................ 57
Figure 16: Train Air Brake- Cost per Part vs. Complexity at Q=100 ...................... 58
Figure 17: Train Air Brake- Cost per Part vs. Complexity at Q=1000 .................... 59
viii
Figure 22: Train Air Brake- Lead Time vs. Complexity at Q=1............................... 64
Figure 23: Train Air Brake- Lead Time vs. Complexity at Q=30............................. 65
Figure 24: Train Air Brake- Lead Time vs. Complexity at Q=100........................... 66
ix
Chapter 1 Introduction
In 2013, a total of 103.2 million metric tons of metal castings were produced
globally, which was a 3.4% increase when compared to 2012.[1] The United
States metal casting industry is made up of 1961 facilities and the industry
capacity is 15.3 million tons.[1] The total industry sales are expected to reach
$30.6 billion in 2018.[1] Also, the U.S. is second in the world in casting shipments
based on tonnage with China in first place and India in third. To increase their
technologies like AM and eradicate the technological barriers that exist within.
imply that the organization/ foundry are aware of the benefits this technology
rapid advancements in the industry and also plays a key role in faster adoption.
between traditional sand casting practices and AM. The motivation for this thesis
is to work within the confines of the current foundry processes and determi ne a
1
The foundation of this thesis depends on having a good fundamental
understanding of the metal casting process and the AM process. Hence, the aim
casting, specifically the sand casting process, and also the adoption of AM in the
casting sector. The final chapter also discusses existing cost models and methods
Metal casting is a production process which involves molten metal being poured
into a mold made of sand, metal or ceramic to form geometrical shapes. The
global production output in 2014 is more than 105 million metric tons. [3]. Metal
casting inherently lends itself to the formation of parts which are intricate, rigid
which makes casting preferable over other shaping process. [5] Some advantages
are:
forging or welding.
2
2. This process is highly adaptable for mass production. An example of such
primary example.
the designer and foundry, stronger castings can be obtained without being
There are a number of casting methods, which differ in the technique and the
equipment used, but all of which require the following systems: [5]
3
A pattern is used to create the mold shape and is the actual shape of the
metal part that needs to be created. A core-box is used for the creation of
rates and solidification for different metals become important for the
molds which, in turn, are made by packing molding sand (in the case of
There are two kinds of patternmaking: one that happens in foundries and
one that happens in pattern shops. Most foundries are concerned with
modifying existing patterns and preparing them for molding while most
pattern shops produce new patterns and cores. The pattern shops operate
Manually building a pattern requires the skill of a pattern maker who can
which makes the process of removal from the sand mold easier. This also
ensures that extra metal is casted intentionally which can be removed via
4
machining. A key factor in pattern designing is that the pattern must be
foundry.
cores required and also the design and build of the core-boxes. Hence, the
shape of the part and then realize how the internal geometry would look
or aircraft engine work, where the complexity of the cast part would be
2. Core-making
Cores are usually made of sand and then placed inside a mold cavity to
form the interior surface of the castings. Thus the void between the core
and the mold-cavity forms the actual shape of the casting. These cores can
be made of metal, plaster or core sand. But to achieve the utmost intricacy
Cores provide the casting process with the ability to make the most
5
would be impossible to machine. In addition, cores serve a number of
other purposes:[5]
Ram-up cores are used for several purposes. These cores are
A core sand mixture consists of sand grains and organic binders which
provide green strength, cured strength and collapsibility. These mixtures are
then molded into cores by using a mixture of manual labor and machines like
3. Molding
of several ways:[5]
molding techniques
6
By free flow of slurry or liquid aggregate around the pattern.
work is carried out as bench molding, machine molding, and floor and pit
molding.
There are many different kinds of furnaces used for melting pig iron and
make it ready for pouring. But the most common ones are cupolas, open
Once the metal has been melted and is ready for pouring, steps have to be
5. Cleaning
7
d. Finishing- final surface cleaning, giving the casting its outward
appearance.
e. Inspection
Tooling is essentially the creation of patterns and core-boxes. The tooling costs
are generally a result of the sand casting tool design and fabrication process. The
Given this initial part design and a general understanding of the cost, lead time,
accuracy and production volume, the tool builder must decide on the best
approach for constructing the tooling. Some of the primary considerations for
8
Functionality of the casting
How to best achieve the specified external and internal shape of the
casting
How molten metal will flow during filling and feed during solidification
of the casting
How the casting will shrink during solidification and cooling to room
temperature
operations
Where and how much extra material should be provided for machining
Rapid Tooling is the process of creating tools quickly and with minimum direct
more modern approach to the creation of tooling. There are two kinds of rapid
tooling: [7]
9
master patterns can be created to fabricate a silicone rubber mold, which
can then be used for making multiple hard patterns from poly eurathene
Direct Tooling – Where additive processes build the actual tooling. This
In this thesis, the tooling costs associated with direct tooling approaches using an
manufacturing using the binder jetting technology to directly 3D print the molds
The tooling cost estimates received from the industry were a singular number
has been made to show the tooling costs in finer detail. The following
components have been determined to have the most effect on the total tooling
cost:
Re-working costs
costs
10
Since the tool design and manufacture is a highly laborious process in the
traditional foundry environment, labor costs have been calculated based on the
Two mechanical engineers and two tool and die makers are assumed to be
working on this project. The tool design labor cost is assumed to be $45.0 per
hour based on the 2015 Median Pay for experienced mechanical engineers. [9]
Similarly, the manufacturing labor cost is assumed to be $30.0 per hour based on
the hourly mean wage for experienced tool and die makers in the aerospace
The majority of the reworking costs occur due to the presence of casting defects
in the sand casting process. [11] Casting defects usually manifest in the foundry
tooling methods these reworking costs can be avoided and in some cases
eliminated. [6]
This thesis incorporates the cost of reworking into the final tooling costs based on
data obtained from Humtown Products. For the conventional mold and core
making processes, the rework costs have been included in the tooling costs. Also,
11
for the methods involving 3D sand printing, Humtown suggests that a mold and
rapid prototyping and what is now popularly known as 3D Printing. The term
representation of parts that would be the basis for further models, eventually
process used to create quick physical prototypes directly from digital CAD data
with less focus on the quality of the prototype and more focus on the visual
aesthetic.
Due to the recent technological advancements made in this field, the term “rapid
prototyping” failed to encapsulate the emerging new processes and the increase
in quality of output. A new term was needed to describe these processes which
which are based on an additive approach. Due to these reasons, the new term
The basic principle behind AM is that a 3D CAD model can be fabricated directly
determine how the individual features can be manufactured and then fit into an
basic understanding of the AM machine and materials needed to create the part.
as the process of joining materials to make objects from 3D model data, usually layer
As estimated in 2015, the total value of the AM industry including all products
and services were $5.165 billion. Also in 2015, Israel is responsible for 41.1% of
total industrial AM systems and the US shares 16.7% of the same total. [2] These
systems are categorized into seven types based on the various processes. [14]
1. Binder Jetting
3. Material Extrusion
4. Material Jetting
6. Sheet Lamination
13
Each of the process has different weakness and strengths which are based on
Build speed
Material properties
Prototyping
The earliest use of additively manufactured parts was for the strict
purposes of prototyping. The primary use cases were either as visual aids
Tooling
14
The application of AM for the creation of direct end-use parts is probably
the fastest growing market since 2010. Whereas rapid prototyping and
process to a single step. This can be termed as the holy grail of AM.
For damaged parts, which have a long lead time, additive manufacturing
has been increasingly seen as a way to repair them at a faster rate. Also
parts made in the additive way provide a metallurgical bond with the
base material, which reduces the heat affected zone in the nearby material.
This feature is useful for parts that have a high sensitivity to heat
distortion. [15]
Although we are aware of sand printing offering great cost and quality
advantages for casting parts, there are significant benefits to creating patterns
using a 3D printer. Until 2011, the rapid prototyping technique for producing
prototypes of models rather than direct-to-use parts was seen as expensive and
of low accuracy. [16] But it was also known that rapid prototyping technique can
produce high complex functional parts directly from CAD data. This simplifies
the process of pattern-making and also reduces redesign costs associated with re-
working of a pattern.
15
As discussed previously, traditional pattern making costs are much more
expensive than the actual costs of pouring in a casting run. Also, the l ead time to
create the tooling for castings is usually in weeks. But using an AM technique
like Fused Deposition Modeling for creating patterns this lead time can be
Printer for creating patterns which can be used in sand casting give us general
Several case studies are available on the websites of Stratasys and 3D Systems
which have machines capable of printing patterns for sand casting. Based on
these case studies, it has been determined that FDM is a great fit when: [19] [20]
The pattern dimensions are less than the build envelope of the FDM
systems
16
Part geometry and gate/runner system can be redesigned to take
advantage of FDM
Make Pattern
Remov e sand
Finish to specification
A Fortus 900mc 3D Printer is used to analyze the cost of 3D printing the patterns
17
1.3.2 Additive manufacturing for mold and core making: Binder Jetting
Specifically the binder jetting process is used to fabricate these molds and cores
which can then be assembled to create the final mold and core assembly. [21]
The binder jetting process prints a binder into a powder bed to fabricate a part.
Hence, only a small portion of the part material is delivered through the print
head and the rest consists of powder in the powder bed. The binder droplets
form spherical agglomerates of binder liquid and powder particles and provide
bonding to the previous layer. Once a layer is printed, the powder bed is lowered
Due to the fact that binder jetting processes can be economically scaled by simply
increasing the number of printer nozzles, it is scalable and also has a high
Materials
18
The commercially available powder from 3D Systems is plaster based and the
binder is water based [24]. Since these printed parts are typically weak, infiltrants
powder and uses a liquid binder that reacts to room temperature. They
recommend that parts remain in the powder bed for several hours so that the
binder is completely cured. [25]. Voxeljet also offers silica sand with inorganic
ExOne supplies a silica sand and two part binder, where one part is coated onto
the layer and the other part is printed onto the layer, which causes a
polymerization reaction binding the sand particles together. [26] ExOne claims to
use only standard foundry materials which ensure easy integration into existing
manufacturing practices.
eliminated, and the molds and cores can be created directly from their design
files. Removal of the pattern-making process can bring the cost of tooling
19
Make sand mold and cores
Remove sand
Finish to specification
create the molds and cores. Hence the costs and build times
particularly in the mold and core creation, this thesis defines two additional
mixed manufacturing methods apart from the traditional mold and core making
method, and the purely additive method of creating molds and cores.
treatment, machining and inspections. Hence, these costs shall not be considered
20
in the cost and time study of this thesis. However, it should be noted that
eliminate flash which tend to add additional finishing and inspection operations.
The following four definitions are inherent to this thesis and describe each
process from the creation of tooling to the final mold and core assembly.
making. This includes the creation of the patterns and core-boxes, creation of
molds and cores using traditional foundry techniques, and finally assembly of
3D sand printing is defined as the process of 3D printing the molds and the cores
using a binder-jetting process. These molds and cores are directly printed from
their respective design files and therefore the creation of patterns becomes
unnecessary. This also eliminates any tooling costs incurred due to the creation
of patterns. The cost benefits of this method have already been discussed in [27]
For the purposes of this thesis, 3D sand printed cores imply the use of traditional
techniques to create the mold and a binder-jetting system to fabricate the cores.
21
At the design stage, complex geometries can be created without worrying
not required.
dependence on binders eliminates the need for core drying and finishing.
Tool maintenance and storage costs, and tool insurance costs are
eliminated.
This is defined the creation of patterns using the Fused Deposition Modeling
technique described in Section 1.3.1 and the traditional method of creating molds
and cores described in 1.2.1 which finally results in a mold and core assembly.
The distinct advantage of FDM pattern-making is the fact that a set of patterns
can be grouped together and printed in a single build. This parallel processing of
parts implies that the lead time for tooling is drastically reduced and the tooling
22
1.5 Complexity Measurement for CAD models
The cost for traditionally manufacturing a part is a function of its complexity and
There is very limited research on this subject, especially since the complexity of a
literature, two estimation methods have been discussed, which have been proven
For this thesis, the complexity of the various case studies described in Section 2.1
detailed in the paper titled “Quantifying the shape complexity of cast parts” by
factor, most of it has been for specific manufacturing processes such as axis -
Also, initial shape complexity factors did not take into account the tooling cost
considered the tooling cost based on geometric features of the part. The logic
23
behind this decision is that the geometric features influence, design of the
After interacting with toolmakers and designers, the researchers concluded that
the tool manufacturing cost depended on number of cores, volume and surface
area of part, core volume, draw distance and variation in section thickness. These
parameters can easily be determined from a given CAD model. Based on these
This is the ratio of volume of part to the volume of bounding box. The bounding
box is given by the maximum length, width, and height of the part geometry.
When the volume of part is close to its bounding box, less material removal is
Here, is the part volume and is the bounding box volume of the part.
This is the ratio of surface area of an equivalent sphere (with the same volume as
that of the part) to the surface area of the part. This ratio is based on the fact that
24
Where, As is the surface area of the imaginary sphere with a volume equal to
that of the part, given by: . Ap and V p are surface area and
Cores are required for hollow portions of the part and regions that hinder
pattern withdrawal during molding. Each core requires a separate tooling; hence
more the number of cored features higher will be the tooling cost. The criterion
features:
Larger cores require larger size and incur higher tooling cost. Hence the ratio of
complexity.
Where, is the volume of the i th core and V b is the volume of the parts bounding
box.
25
This is the ratio of minimum and maximum thickness of the part. A tooling with
thin section will be more complex and is more difficult to machine as compared
Where, Tmin and Tmax are minimum and maximum thickness of the parts
respectively.
The draw distance, which is maximum depth of the tooling, affects the tooling
manufacturing time and hence its cost. The actual draw distance is compared to
the minimum possible draw distance, which is half the minimum dimension of
the part. The criterion designed such that parts with higher depth ratio will
Where, L,W and H are the length, width and height of the part respectively and
Eq. 1
26
This equation has been validated by determining the actual Complexity factor
The thesis, authored by Dr. Martin Baumers, discusses the economic aspects of
The direct link between the complexity of a design and its traditional
this complexity for free, eliminating the need to simplify a particular design and
[32]
However, a weak connection between part geometry and laser scan time has
been demonstrated when the part volume and part height are left unchanged in
27
two-dimensional shapes. They define these 2D shapes as “configurations
degree of convexity of the perimeter shapes. Full convexity is the property that
every point in the perimeter can be connected to every other point without
crossing the perimeter of the shape as opaque walls. Hence, all points are visible
Based on this definition, three measures are defined by Psarra and Grajewski for
shape complexity:
perimeters cells visible from each location. Fully convex shapes have MCV
intersections of the graph of connectivity with the MCV line. Psarra and
28
This forms a relatively simple and pragmatic way to quantify the shape
surfaces. Baumers’ thesis utilizes this concept to calculate the complexity of the
Historically, sand casting has been a very popular method producing the greatest
have only been to improve process parameters through the use of high quality
cast parts.
Hence in order to assist foundries in adopting AM for mold and core making, the
Industry” [35]. Apart from YBI, Youngstown State University, the University of
Northern Iowa, the American Foundry Society, ExOne, Jenney Capital Market,
29
1.6.2 Hypothesis
This thesis is built upon the paper titled “Quantifying the Role of Part Design
Complexity in using 3D Sand Printing for Molds and Cores”. The paper
evaluates when to use the ever evolving AM sector versus the traditional pattern
making. This was done by examining the cost of molds and cores as a function of
part design complexity quantified by a complexity factor for two case studies.
The complexity of each case study was varied systematically by changing the
geometry and the number of cores. The tooling and fabrication costs were
estimated for both 3D sand printing and traditional pattern-making. Once the
breakeven points were identified, it was shown that 3D sand printing is cost-
effective for castings with complexity factor values greater than that of breakeven
points.
30
One particular idea, mentioned in the future work section of the paper discussed
sand printing. In other words, until now, 3D sand printing and traditional
pattern-making were treated as separate use-cases i.e. the analysis was limited to
entirely 3d printing the molds and cores or entirely manufacturing the molds
and cores in the traditional way. But a potential way to integrate AM into the
Mixing different methods can mean either replacing the traditional pattern-
making process with a 3d printing process where the patterns are fabricated
To elaborate further, the first method analyses the effects of integrating an FDM
follows:
31
molds is cost-feasible when compared to traditional manufacturing or 3D sand
printing.”
eliminates the cost of tooling incurred during a traditional sand casting process
and also fabricating cores using a 3D sand printer in the traditional sand casting
process eliminates the need for using core-boxes which reduces the costs further.
the cost per casting in post-fabrication operations including pour, shakeout and
2. Determine the costs for molds and cores for different families of castings
and draw conclusions from the cost per part vs. shape complexity using
32
c. 3D Sand Printing cores and traditionally manufactured molds
3. Determine the lead times taken for each process from pattern creation
conclusions drawn from this thesis and help make decisions on which
manufacturing method will best suit their needs. This application will be
33
Chapter 2 Methodology
This chapter will describe the methodologies and tools used to achieve the stated
research objectives. The first section describes the five case studies utilized for
the analysis and the unique features of each case study. Section 2.1 describes the
case studies used in this thesis. In section 2.2 the shape complexity calculations
used to calculate the shape complexity factor for each case study is shown.
consideration for these various case studies. Finally the last section looks at how
the entire data can be collated to determine the cost per part and an estimation
These case studies were created with an attempt to study the variation of cost
This case study is provided by Humtown Products for the purpose of this thesis
and contains the mold and core design package needed to cast a Train Air Brake.
34
The design package consists of 1 mold design and 8 core designs. Hence a total of
9 sub-cases are created by beginning the first sub-case with one mold and
35
Case Study 2: Turbo Charger
This case study, also provided by Humtown Products consists of a mold and
varying the number of cores. Hence we create 4 sub-cases, where each sub-case
The shape complexities of these five case studies are calculated using the
methodology outlined in Section 1.7.1. Provided below are the shape complexity
factors for the five case studies including their respective sub-cases.
36
Table 1: Complexity factor calculation for each case study and their sub-cases
The tooling cost quotes received from Humtown Products are compared with
tooling cost estimates can vary extensively from foundry to foundry. These
foundries often base their estimates on historical data and experience. Hence a
reliable way of calculating these costs would prove useful in providing more
Only the reworking costs, tool design and manufacturing costs are incorporated
into the data since, the online cost estimator provides the tool manufacture cost
The breakdown of the industry quote is listed in Section 1.2.2. Based on these
quotes, the tool design and manufacture costs are isolated and detailed in the
table below.
37
Table 2: Foundry Tooling Quote for Two Case Studies
Tooling
Tooling creation Quoted
Material,
time: Tooling Tool Tooling:
Tool Design Maintenance
Part Name or Number Rework Cost design+ tooling Manufact Design + Total Quoted Tooling
Labor and Energy,
manufacture ure Labor Manufacture
overhead
(hours) Labor costs
expenses
Com. Cubic core 400 47.10 2967.30 847.80 4215.10 6784.90 11400
Com. Cylinderical Core 500 55.30 3483.90 995.40 4979.30 7020.70 12500
Orginal core 600 61.10 3849.30 1099.80 5549.10 6450.90 12600
The tooling creation time between design and manufacturing is split in the ratio
0.7 to 0.3 assuming that 70% of the time is spent in designing the part and 30% of
The online cost estimator [36] utilizes a feature based calculator to determine the
tooling, material and production cost for a sand casted part. The cost estimator
and the actual costs may vary based on equipment, specific manufacturer and
market conditions.
Hence, comparing these costs with an actual foundry quote would provide a
38
As shown in Figure 9, the production quantity, material of the cast part, the
envelope (or the bounding box) of the part, volume and number of cores are the
inputs required.
39
Table 3: Cost Estimator Results for Both Case Studies
Online
Online Estimated
Estimated
Tooling Cost : Tool
Part Name or Number Tooling Cost :
Manufacture+Tool
Tool
labor
Manufacture
Imported No Core 1380.00 3528.30
Imported 8 1722.00 6957.30
Imported 8,3 2101.00 7336.30
Case Study I
The tool design labor cost estimated in Section 2.3.1 has been incorporated into
This cost estimation model is based on the paper titled “Casting cost estimation in
by the solid model of the part and its attributes and is useful for cost reduction in
This costing model attempts to provide a module to calculate tooling costs based
40
According to this paper, the tooling costs are calculated as follows:
Where,
Cindex = Tooling cost index that varies with manufacturer, currency and time
Q = Order quantity
Again, the resulting tooling cost has been normalized by adding the tool design
Estimated
Integrated Casting
Relative Tooling Tooling Cost
Casting Accuracy Volume of Cast (in Production Actual relative Individual Cost Estimation
Part Name or Number Complexioty Cost Cost Index based on
Volume index (Cac) m3) Volume (Q) tooling cost ($) cost index model including
Crel_tool_cost($) Integrated
tool design labor
model ($)
Imported No Core 209.16 19.66 90 0.0034 1000 1.38 3542.64 0.39 0.36 1274.26 3422.56
Imported 8 162.49 41.35 90 0.0027 1000 1.722 5830.71 0.30 0.36 2097.25 7332.55
Imported 8,3 144.47 50.10 90 0.0024 1000 2.101 7130.35 0.29 0.36 2564.72 7800.02
Case Study I
Imported 8,3,2 131.76 55.41 90 0.0022 1000 2.857 8055.36 0.35 0.36 2897.44 8951.74
Imported 8,3,2,5 123.11 58.93 90 0.0020 1000 3.339 8732.70 0.38 0.36 3141.07 9195.37
Imported 8,3,2,5,4 120.01 60.88 90 0.0020 1000 3.914 9134.43 0.43 0.36 3285.57 10095.87
Imported 8,3,2,5,4,7 113.83 63.14 90 0.0019 1000 4.574 9619.72 0.48 0.36 3460.12 11089.42
Imported 8,3,2,5,4,7,6 105.66 65.60 90 0.0017 1000 5.091 10179.45 0.50 0.36 3661.45 12046.75
Imported 8,3,2,5,4,7,6,1 103.43 66.72 90 0.0017 1000 5.998 10445.67 0.57 0.36 3757.21 12142.51
Comp. NO core 72.41 27.64 90 0.0012 1000 0.978 4250.17 0.23 0.36 1528.75 3677.05
Case Study II
Com. Cubic core 61.23 41.62 90 0.0010 1000 1.326 5860.91 0.23 0.36 2108.11 5075.41
Com. Cylinderical Core 61.73 45.73 90 0.0010 1000 1.76 6441.99 0.27 0.36 2317.13 5801.03
Orginal core 31.39 59.81 90 0.0005 1000 2.24 8904.51 0.25 0.36 3202.87 7052.17
Average: 0.36
41
2.4 Cost Studies
This study estimates the costs for four different methods of manufacturing molds
described in Section 1.4 and cores starting with the creation of the patterns to the
assembly of molds and cores in order to find the cost per part at various
Also, the cost per part for each process has been determined for each sub -case
and also at various production volumes (Q) of 10, 30, 100, and 1000 units. The
notation used for denoting the costs consists of a subscript and a superscript.
The subscripts always refer to either to the various components of the total cost
such as mold cost, core cost, tooling cost etc. For example: CM is the cost of mold-
making (M).
The TM costs are calculated based on the tooling, mold and core cost estimates
received from Humtown Products. The tooling costs provided were isolated to as
much fine detail as possible and incorporated into the Cost per Part (C P)
calculations.
42
Among the several cost factors in sand casting, the two major components are
tooling and fabrication costs which involve a variety of operations to produce the
molds and cores. In previous research, the tooling making costs were generated
using an online estimator for sand casting process. [27] But this estimator
provides just the cost of manufacturing the tooling and does not consider the
rework costs, the tool design or even the tooling material costs which are
essential components of a tooling cost. Hence, estimates for each sub-case were
Tooling Costs:
Since tooling cost is determined for a set production volume, the tooling cost per
Where,
43
Ctd is the cost associated with tool design
Cac is the additional cost of energy, labor, overhead and material procurement
Where,
CMCTM is the total cost of mold and core making for a given production volume Q
Where,
The data for cost of 3D sand printing has been obtained from [27], which utilizes
For both mold and cores, the fabrication costs are estimated using the industry
quotation methods based on the size of casting, number of cores etc. [27]
44
Where,
Cbj is the volumetric cost of binder jetting process including consumables, labor,
For this thesis, the Cbj is taken as $0.17 per cm3 which is the estimated cost used
The tooling costs associated with 3D sand printing is zero since there is no
Therefore the cost per unit production volume (or cost per part) is
This result shows that the 3DSP cost per part is independent of the production
volume.
45
2.4.3 3D Sand Printed Core (3DSPC)
There is no additional cost of tooling for core making, since core -boxes are
Hence, the total cost per part for 3DSPC is calculated as:
Or
The following are the cost components required to calculate the cost per part for
46
Cost of traditional mold making, CMTM
The lead time calculation is done based on industry estimates for traditional
manufacturing techniques and build rates per hour for any AM processes used in
the thesis. The following components are defined to calculate the final lead times
Lead time for tool design (T td) and tool manufacturing (Ttm) to create the
Products
Time to create the molds (TmTM) and cores (TcTM) from setup of tooling to
Build time for molds (Tm3DSP) and cores (Tc3DSP) on an S-max printer,
based on build volume of the machine and also the build rate provided by
Pre (Tpre) and post (Tpost) processing times for S-max printers
47
Utilizing two or more machines would ideally reduce the lead time due to
parallel processing but would also require more capital investment for the
foundry. Hence for the purposes of this thesis, it is assumed that only one 3D
sand printer and only one FDM printer are available for the foundry.
The total lead time to create a mold and core assembly is the sum of the time to
create tooling (from design to manufacturing) and the time to create the mold
The quote from Humtown provides the total time for tooling including the times
for purchase of tooling material, tool design, tool manufacture and the customer
approvals. The tool design and manufacturing time have been isolated from
these numbers by taking into account the fact that it takes 1 day to purchase the
Therefore,
TmTM + TcTM is the time to create the molds traditionally + time to create the cores
traditionally
Ttm + Ttd is the total time to design and manufacture the tooling
48
2.5.2 3D Sand Printing (3DSP)
The build times for molds and cores are calculated based on the total build
height after parts are nested in the machine and the height build rate specified
Where,
rbuild is the height built rate for S-max which is estimated to be around 1 in/hr
For a 3DSPC method, the total lead time (in days) is a combination of time for
The total lead time (in days) for a FDMP method is a combination of the time to
build the patterns in a Fortus 900 and the time to traditionally create the molds
and cores.
49
TFDMP is the build time to the patterns in a Fortus 900 FDM printer. For each case
study, it is assumed that the mold and all cores are printed together in one build.
This build optimization saves time and cost and it’s a more sensible way of
Based on the data accumulated, creation of a tool which would estimate the
complexity factor, build times and associated costs would greatly serve the
Since Microsoft Excel has been used to store the data generated from the
Hence, Visual Basic Applications (VBA) has been used to create a user interface
which can be used to enter the inputs line by line and subsequently generate the
complexity factors.
The following are the features that have been identified as initial inputs for the
tool:
Part Name
Part Volume
Surface Area
Number of Cores
50
Minimum thickness of the part
Length
Width
Height
The user interface is created in VBA using textboxes and data fields. Three action
buttons (“OK”, “Clear” and “Cancel”) are provided for control as shown in the
figure below.
In Figure 11, the highlighted portion is a click button titled “Complexity Calculator”. This button
acts as a macro and upon clicking; it connects to the user interface show in Figure 10.
51
Figure 11: Macro Button for Complexity Calculator
Once the values have been entered and the user presses the OK button (as shown
in figure 12), the complexity number is calculated in the excel sheet and
52
Figure 12: Example Calculation for Train Air Brake with 8 Cores
53
Chapter 3 Results, Analysis and Discussion
Based on the formulas and equations defined in Chapter 2, the various costs and
analyzed for the four manufacturing methods (TM, 3DSP, 3DSPC, and FDMP).
Also, the same cost variations vs. complexity factor are analyzed based on
Integrated
Casting Cost Online Estimated Quoted Tooling:
Estimation Tooling Cost : Tool Design +
Part Name or Number
model including Manufacture+Tool Manufacture
tool design labor Labor costs
labor
Imported No Core 3422.56 3528.3 2962.1
Imported 8 7332.55 6957.3 6931.1
Imported 8,3 7800.02 7336.3 7031.1
Case Study I
54
The tooling costs estimated in Section 2.3 are now compared using a single factor
The null hypothesis states that, there is no difference between any of the tooling
The alternative hypothesis states that, there are significant differences between
SUMMARY
Groups Count Sum Average Variance
Column 1 13 103682.4449 7975.572681 8684518.888
Column 2 13 108929.5 8379.192308 11680656.55
Column 3 13 93225.5 7171.192308 8495398.011
ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 9833202.549 2 4916601.274 0.511071059 0.604139 3.259446
Within Groups 346326881.4 36 9620191.15
Total 356160083.9 38
The p-value obtained is 0.604 which is greater than 0.05. Hence, we fail to reject
the null hypothesis and reach a conclusion that the three tooling costs have no
significant differences between them. Therefore, the thesis is at liberty to use any
of the tooling costs calculated and for the purposes of this thesis use the industry
quote.
55
3.2 Cost Studies
3.2.1 Cost per Part vs. Complexity Factor at each production volume
method. The cost per part vs. complexity is plotted for each case study and every
For a unit production volume, the 3DSP costs remain the lowest ($159.00-
$212.00) since the need for patterns is eliminated (bringing down the tooling cost
to zero). The cost per part for FDMP is in the range of $595.0 – $686.00 which is
higher than 3DSP but significantly lower than TM or 3DSPC for any value of
complexity.
Figure 14: Train Air Brake- Cost per Part vs. Complexity at Q=1
56
For low production volumes (30+), two break-even points can be observed. The
first break-even point is between TM and 3DSP at a complexity of 58.9 and cost
per part of $206.00. Hence, below the breakeven point, the TM method is more
The second breakeven point is between 3DSPC and FDMP methods. Here we
observe a break-even point of 32.5 and $70.8. Below the break-even point it is
feasible to 3D print the patterns and beyond the breakeven point, a core printing
Figure 15: Train Air Brake- Cost per Part vs. Complexity at Q=30
57
For medium production volumes, the 3DSPC costs are the lowest compared to
any other method for any level of complexity. This can be attributed to the low
cost of 3D sand printing of cores. The highest cost per part is for a 3DSP process
Figure 16: Train Air Brake- Cost per Part vs. Complexity at Q=100
For high production volumes (1000+), the 3DSPC method is the most feasible
beyond a complexity of 19.4 and cost of $44.00. The cost curves for TM and
FDMP almost intersect since both have a tooling cost associated with them and
58
Figure 17: Train Air Brake- Cost per Part vs. Complexity at Q=1000
As in case study 1, a similar set of trends can be observed for case study 2.
For a unit production volume, the 3DSP costs remain the lowest ($127.00-
$160.00) since the need for patterns is eliminated (bringing down the tooling cost
to zero). The cost per part for FDMP is in the range of $543.0 – $605.00 which is
higher than 3DSP but significantly lower than TM or 3DSPC for any value of
complexity.
59
Figure 18: Turbocharger- Cost per Part vs. Complexity at Q=1
For low production volumes (30+), again two break-even points can be
observed. The first break-even point is at (55.4, $161.00), above which a 3DSP
The second breakeven point (52, $72.00), is between 3DSPC and FDMP methods.
Below the break-even point it is feasible to 3D print the cores and beyond the
sense.
60
Figure 19: Turbocharger- Cost per Part vs. Complexity at Q=30
For medium production volumes, break even points are observed at [19.7,
$36.60] and [44.6, $48.5]. The first breakeven point suggests that for parts with
complexity below 19.7, the TM process would be cost economical. The second
breakeven point suggests that beyond a complexity value of 44, a 3DSPC method
61
Figure 20: Turbocharger- Cost per Part vs. Complexity at Q=100
For high production volumes (1000+), the TM and FDMP curves intersect
showing that the cost for these two processes are essentially equal at high
FDMP and 3DSPC. Interestingly below the breakeven point, the cost per part for
TM, FDMP and 3DSPC are almost equal. Hence beyond this breakeven point, a
62
Figure 21: Turbocharger- Cost per Part vs. Complexity at Q=1000
The lead time to manufacture the two case studies is defined from the creation of
the pattern and core-box (if required) to the final mold and core-assembly. These
lead times also include the pre and post processing times required for removing
Figure 25 shows that for unit volume production, 3DSP provides the fastest lead
times (less than 2 days) up to a breakeven point of [65.6, 2.37 days] after which
an FDMP based approach would provide faster lead times. The TM method has
63
the longest lead times for any sub-case at any level of complexity. This is entirely
attributed to the time taken to design tooling for the molds and cores.
Figure 22: Train Air Brake- Lead Time vs. Complexity at Q=1
For small production volume (30+), 3DSP takes the longest lead time (15 days
for the lowest complexity and 90 days for the highest complex part). A breakeven
point is observed at [40, 12 days] beyond which the FDMP method provides
64
Figure 23: Train Air Brake- Lead Time vs. Complexity at Q=30
For medium to high production volumes (100+), the TM method has the fastest
lead times since the binder jetting technology have a limited build volume and
speed. It is also observed that FDMP and TM have comparable lead times and as
the complexity increases the gap between their lead times reduces. Also the
breakevenpoint between 3DSPC and FDMP [27, 38.30 days] shows that, beyond a
complexity of 27, FDMP is capable of achieving faster lead times when compared
to 3DSP or 3DSPC.
65
Figure 24: Train Air Brake- Lead Time vs. Complexity at Q=100
For unit production volumes, 3DSP provides the fastest lead times regardless of
complexity. The TM method has the longest lead times beyond the breakeven
66
Figure 25: Turbocharger- Lead Time vs. Complexity at Q=1
For small production volume (30+), the time curves for 3DSP and 3DSPC are
around 10-12 days apart. This means both the methods have lead times that are
close to each other. A breakeven point is observed at [32, 12.5 days], between
FDMP and 3DSP, beyond which the FDMP method provides faster lead times.
The TM method has the fastest lead times beyond the breakeven point [22.5, 6
67
Figure 26: Turbocharger- Lead Time vs. Complexity at Q=30
For medium to high production volumes (100+), again the TM method has the
fastest lead times among all four processes. However unlike the previous case
study, here it is observed that FDMP and TM have a large difference between
their lead times. Also the breakevenpoint between 3DSPC and FDMP [34, 37.60
days] shows that, beyond a complexity of 34, FDMP is capable of achieving faster
68
Figure 27: Turbocharger- Lead Time vs. Complexity at Q=100
69
Chapter 4 Conclusions and future work
The purpose of this thesis was to examine the adoption of a 3D sand printing
system or an FDM system at various stages of the sand casting process on the
costs and times of the two case studies. This was done by adopting an FDM
been adopted at the mold and core-making stage by the use of a 3D sand printer
(ExOne S-max). This is defined as the 3DSP (3D Sand Printing) method. Thirdly,
a 3D sand printer has been adapted only at the core-making stage, which has
been defined as the 3DSPC (3D Sand Printed Cores) method. The cost per part
vs. complexity graphs were generated for each method and compared on the
Among the major cost components presented in this thesis, the most significant
one has been the tooling costs. Foundries have mostly provided tooling cost
This problem has now been overcome by the use of a casting cost estimation
model described in Section 2.3.3 and has been proven to be significantly accurate
70
In both case studies, it is observed that for a unit production volume the 3DSP
method is the most cost feasible option and also provides the fastest lead times
for any level of complexity. In other words, 3D sand printing the molds and
cores is economical both cost-wise and time-wise. The FDMP method is second
to 3DSP for parts with low complexity, but beyond a complexity factor of 66 for
For short production runs of 10 to 30, the FDMP method has the least cost per
part for a large range of complexities. This implies that, printing the patterns on
an FDM printer is economically feasible for low volume production. But for case
study 1 (train air brake), the graphs for lead times show that FDMP is faster for
complexities beyond 40. Hence, if the requirement is a set of 10 train air brake
castings and the complexity factor is above 40, it is recommended to opt for an
FDMP method.
For medium production volumes (100+), 3DSP is clearly neither time nor cost
effective for both case studies. It is also noticed that the TM method is the most
For large production volumes (1000+), both the case studies clearly agree on TM
method providing the fastest lead times. For case study 1, 3DSPC method has the
71
lowest cost per part beyond a complexity factor of 20 but has a larger lead time in
comparison to TM.
based on the complexity of the final part and also its production volume.
generation of tooling cost using the casting cost estimation method and also the
final mold and core making costs for any method of mixed manufacturing.
There are other 3D sand printers available in the market supplied by Voxeljet
and Viridis3D which supposedly offer faster build rates with more versatile
materials. The cost and time economics of using these printers can be a possible
In Section 1.5.2, the idea of calculating complexity for every layer of a spliced 3D
object has been discussed. This algorithm can be potentially applied to study the
It is also known that FDM based parts have a significant distortion which may
alter the dimensions of the sand molds and cores. The effect of this pattern
72
This study also assumes that 3D sand printing provides an equal surface fi nish
testing the physical and mechanical properties could provide additional decision
73
References
Purchasing. Web.
http://www.afsinc.org/multimedia/contentMCDP.cfm?ItemNumber=18
2015
Handbook (1944), p. 3
6. Stoll, H.W, and J.G Conley. "Tool Design and Construction for Sand
DV08PUB3 – 5, www.sme.org
8. Rapid tooling route selection and evaluation for sand and investment
74
9. Mechanical Engineers, Occupational Outlook Handbook. Web.
http://www.bls.gov/ooh/architecture-and-engineering/mechanical-
10. Tool and Die makers, Occupational Employment and Wages, May 2015.
Nov 2016
15. Hedges, Martin, and Neil Calder. 2006. Near Net Shape Rapid
75
17. Harry Jansen. Creating Casting Patterns Utilizing 3D CAD Data, Z-Corp
http://www.stratasys.com/solutions/additive-
20. Case New Holland Uses Duraform and CasForm Materials for Casting.
Web.
https://www.3dsystems.com/sites/www.3dsystems.com/files/case_ne
23. Sachs EM, Cima MJ, Williams P, Brancazio D, Corie J (1992) Three-
2015
76
25. Materials for 3D Printers, VoxelJet. Web.
27. Almaghariz, E.S., Conner, B.P., Lenner, L. et al. Inter Metalcast (2016) 10:
240. doi:10.1007/s40962-016-0027-5
http://www.exone.com/About-ExOne/News/View/ArticleId/52/Core-
29. Josh, Durgesh, and Bhallamudi Ravi. "Quantifying the Shape Complexity
die and mold cost estimation using design features and tooling
Loughborough University
32. Hague, R., Campbell, I., and Dickens, P., 2003. “Implications on design of
77
33. Ruffo, M., Tuck, C., and Hague, R., 2006b. “Cost estimation for rapid
(9), 1417-1427
34. Psarra, S., and Grajewski, T., 2001.” Describing Shape and Shape
volume:1-25 print
http://www.custompartnet.com/estimate/sand-casting/
37. Chougule R.G., Ravi B., Dec 2005. “ Casting cost estimation in an integrated
78