Developing Process Design Methodology for Investment Cast Thin-Walled Structures

(1)

Cast Thin-Walled Structures

Mohsin Raza EL O P IN G P R O C ES S D ES IG N M ET H O D O LO G Y F O R I N V ES TM EN T C A ST T H IN -W A LL ED S TR U C TU R ES 2018 ISBN 978-91-7485-377-3 ISSN 1651-4238 Address: P.O. Box 883, SE-721 23 Västerås. Sweden

Address: P.O. Box 325, SE-631 05 Eskilstuna. Sweden E-mail: info@mdh.se Web: www.mdh.se

reliable casting process for thin-walled components which will facilitate development of light-weight engineering systems with improved efficiency and fuel economy. The work bridges the gap between established physics behind the castability of metals and the industrial practices. Here it is shown that the variations in process conditions originate from equipment as well as the operator practices in the foundry. These variations effect the targeted values of process parameters and hence effect the reliability and repeatability of process. These process variations not only effect the castability of thin-walled components but also effect the validity of simulation which limits the use of casting simulations in foundry processes.

In order to minimize process uncertainties, a simulation based process design approach has been proposed; where methods to improve the process conditions as well as systematic implementation of simulation for designing casting process has been implemented. The approach resulted in significant improvements in terms of quality of castings as well as overall reduction in process development time and cost in foundry.

Mohsin Raza is enrolled as an industrial doctoral student at the research school Innofacture at the School of Innovation, Design and Engineering at Mälardalen University. Mohsin has a background in Materials Science and Metallurgy involving the materials processing techniques. His research is mainly focused on process development for investment cast thin-walled components for gas turbines and aircraft engines.

velopment of light-weight engineering systems with improved efficiency and fuel economy.

The work bridges the gap between established physics behind the castablity of metals and the industrial practices. Here it is shown that the variations in process conditions originate from equipment as well as the operator practices in the foundry. These var-iations effect the targeted values of process parameters and hence effect the reliability and repeatability of process. These process variations not only effect the castability of thin-walled components but also effect the validity of simulation which limits the use of casting simulations in foundry processes.

In order to minimize process uncertainties, a simulation based process design ap-proach has been proposed; where methods to improve the process conditions as well as systematic implementation of simulation for designing casting process has been implemented. The approach resulted in significant improvements in terms of quality of castings as well as overall reduction in process development time and cost in foundry.

Mohsin Raza is enrolled as an industrial doctoral student at the research school Innofacture at the School of Innovation, Design and Engineering at Mälardalen University. Mohsin has a background in Materials Science and Metallurgy in-volving the materials processing techniques. His research is

mainly focused on process development for investment cast thin-walled components for gas turbines and aircraft en-gines.

(2)

Mälardalen University Press Dissertations No. 257

DEVELOPING PROCESS DESIGN METHODOLOGY

FOR INVESTMENT CAST THIN-WALLED STRUCTURES

Mohsin Raza 2018

(3)

ISSN 1651-4238

(4)

Mälardalen University Press Dissertations No. 257

DEVELOPING PROCESS DESIGN METHODOLOGY FOR INVESTMENT CAST THIN-WALLED STRUCTURES

Mohsin Raza

Akademisk avhandling

som för avläggande av teknologie doktorsexamen i innovation och design vid Akademin för innovation, design och teknik kommer att offentligen försvaras fredagen den 6 april 2018, 10.00 i Filharmonin, Mälardalens högskola, Eskilstuna.

Fakultetsopponent: Professor Mark Jolly, Cranfield University

(5)

Abstract

Components for engineering systems, such as gas turbines and jet engines operating at high temperature are usually produced in superalloys. The investment casting process is most widely used for manufacturing these components due to the ability of the process to produce parts with complex geometries to close dimensional tolerances. Other processing routes are less advantageous due to high mechanical strength and hardness of these alloys, which make formability and machining difficult even at high temperature. The global requirements for lower fuel consumption and emissions are increasing the demands to lower the weight of cast components in jet engines. The ability to produce components with lower wall thickness will not only help to reduce the cost of production and resource usage but also help to improve the efficiency of engineering systems resulting in lower fuel consumption and reduced emissions of environmentally hazardous gases. However, casting of thin walled components is challenging due to premature solidification in thin sections and long feeding distances often resulting in incomplete filling, cold shuts and shrinkage porosity.

The castability of thin-sections is dependent upon selection of appropriate values of casting parameters to achieve favorable conditions for the mould filling and solidification. In foundry environment, fluctuation in these targeted values of casting parameters is common due to semi-automated nature of process. The effects of casting parameters on mould filling and defect formation have been widely reported in the literature, however effect of fluctuations in targeted values of casting parameters resulting from typical variation in the foundry is not well documented. Moreover, the origin of process variation and how to manage them in foundries, especially in relation to thin-walled casting has not been well documented.

In this work, the common variations in critical process parameters, originating from foundry practices and equipment are identified. The effect of variations and resulting fluctuation in targeted values of casting parameters on castability of thin-walled castings is evaluated. The casting process is simulated by defining boundary conditions which replicate the foundry conditions and properties of foundry materials in a commercial casting simulation software. The effect of fluctuation of casting parameters on castability of thin-walled castings is established by casting trials as well as simulations and the validity of simulation is evaluated. A methodology to design a casting process is established by proposing methods to minimize the process variation as well as using Design of Experiments (DoE) based simulation work to achieve reliability and repeatability in the process.

It is concluded that the mould temperature, casting temperature and pouring rate are common casting parameters affected by the variation originating from equipment and the casting practices. The variation in these parameters strongly effects the castability of thin-walled sections. The significance of these variations is validated by simulation and it is concluded that the validity of simulation is not only strongly dependent upon the foundry specific material data but also depends upon setting up valid boundary conditions according to the equipment and practices used. It is also concluded that by introducing material data and accurate boundary conditions, simulation can be used as tool to facilitate process development in foundries. A systematic implementation of simulations based on DoE and optimization resulted in significant reduction in process development time.

The result of this work has been further developed into a process design methodology for investment casting foundries working with casting of thin-walled castings for high temperature applications. The term process design in this work is defined as design and evaluation of gating system as well as identifying optimized values of casting parameters to cast components in foundry.

ISBN 978-91-7485-377-3 ISSN 1651-4238

(6)

(7)

(8)

i

ABSTRACT

Components for engineering systems, such as gas turbines and jet engines operating at high temperature are usually produced in superalloys. The in-vestment casting process is most widely used for manufacturing these com-ponents due to the ability of the process to produce parts with complex ge-ometries to close dimensional tolerances. Other processing routes are less advantageous due to high mechanical strength and hardness of these alloys, which make formability and machining difficult even at high temperature. The global requirements for lower fuel consumption and emissions are in-creasing the demands to lower the weight of cast components in jet engines. The ability to produce components with lower wall thickness will not only help to reduce the cost of production and resource usage but also help to improve the efficiency of engineering systems resulting in lower fuel con-sumption and reduced emissions of environmentally hazardous gases. How-ever, casting of thin walled components is challenging due to premature solidification in thin sections and long feeding distances often resulting in incomplete filling, cold shuts and shrinkage porosity.

The castability of thin-sections is dependent upon selection of appropriate values of casting parameters to achieve favorable conditions for the mould filling and solidification. In foundry environment, fluctuation in these target-ed values of casting parameters is common due to semi-automattarget-ed nature of process. The effects of casting parameters on mould filling and defect for-mation have been widely reported in the literature, however effect of fluctua-tions in targeted values of casting parameters resulting from typical variation in the foundry is not well documented. Moreover, the origin of process var-iation and how to manage them in foundries, especially in relation to thin-walled casting has not been well documented.

In this work, the common variations in critical process parameters, originat-ing from foundry practices and equipment are identified. The effect of varia-tions and resulting fluctuation in targeted values of casting parameters on castability of thin-walled castings is evaluated. The casting process is simu-lated by defining boundary conditions which replicate the foundry conditions and properties of foundry materials in a commercial casting simulation soft-ware. The effect of fluctuation of casting parameters on castability of thin-walled castings is established by casting trials as well as simulations and the

(9)

ii

validity of simulation is evaluated. A methodology to design a casting pro-cess is established by proposing methods to minimize the propro-cess variation as well as using Design of Experiments (DoE) based simulation work to achieve reliability and repeatability in the process.

It is concluded that the mould temperature, casting temperature and pouring rate are common casting parameters affected by the variation originating from equipment and the casting practices. The variation in these parameters strongly effects the castability of thin-walled sections. The significance of these variations is validated by simulation and it is concluded that the validi-ty of simulation is not only strongly dependent upon the foundry specific material data but also depends upon setting up valid boundary conditions according to the equipment and practices used. It is also concluded that by introducing material data and accurate boundary conditions, simulation can be used as tool to facilitate process development in foundries. A systematic implementation of simulations based on DoE and optimization resulted in significant reduction in process development time.

The result of this work has been further developed into a process design methodology for investment casting foundries working with casting of thin-walled castings for high temperature applications. The term process design in this work is defined as design and evaluation of gating system as well as identifying optimized values of casting parameters to cast components in foundry.

(10)

iii

ACKNOWLEDGEMENTS

I would like to thank several people who have supported me during the work which lead to this thesis. First, I would like to thank Mark Irwin for his trust in me. Without of his confidence and support this work would not have hap-pened. I thank Hasse Fredriksson for being source of inspiration. I thank John Danzig for encouragement when I really needed it. My sincere grati-tude goes to Björn Fargerström, who was always there to solve anything that could stop the work. I thank Anders E.W. Jarfors who helped me to recog-nize my strengths and weaknesses.

I would also like to thank all my colleagues at TPC Components AB, espe-cially Peter Edman and Pedro Silva, who spent time helping me in solving practical issues in the foundry. Roger Svenningsson at Swerea SWECAST AB, was always there when i needed someone to discuss ideas. My col-leagues at Mälardalen University have been source of motivation for me throughout my program of study. I thank them all.

I am grateful to Mats Jackson and Anders Fundin for their efforts to make research process smooth for us at Innofacture. I thankfully acknowledge the financial support of the Innofacture program at Mälardalen University fund-ed through K.K. Stiftelsen.

I thank my elder brother, Asad, for giving me confidence to face challenges when we were growing up. I thank my mother, who gave me courage to take steps. I miss my father, who would have been very proud today.

I thank my wife, Asifa, for her never-ending support.

Mohsin Raza

(11)

(12)

v

LIST OF PAPERS

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Raza, M., Irwin, M. (2014) The effect of shell thickness on

de-fect formation in investment cast Ni-base alloy. Proceedings of

ICI 61st Technical Conference & Expo 2014, 5-8 October 2014

Covington, KY.

II Raza, M., Irwin, M., Fagerström, B. (2015) The effect of shell

thickness, insulation and casting temperature on defect for-mation during investment casting of Ni-base turbine blades.

Ar-chives of Foundry Engineering, 15(4): 115-124

III Raza, M., Silva, P., Irwin, M., Fagerström, B., Jarfors, A.E.W.

Effects of process related variations on defect formation in in-vestment cast components. Archives of Foundry Engineering, (In Press, 2018)

IV Raza, M., Svenningsson, R., Irwin, M. (2015) Experimental

study of the filling of thin-walled investment castings in 17-4PH stainless steel. Metallurgy and Foundry Engineering, 41: 85-98

V Raza, M., Svenningsson, R., Irwin, M., Fagerström, B., Jarfors,

A.E.W. (2017) Effect of process related variations on fillability simulation of thin-walled IN718 structures. International

Jour-nal of Metalcasting

VI Raza, M., Svenningsson, R., Irwin, M., Fagerström, B., Jarfors,

A.E.W. (2017) Simulation based process design approach for manufacturing of light-weight cast components. LIGHTer

In-ternational Conference 2017, 22-23 November 2017

(13)

vi

Contribution to the appended papers:

Paper I: Raza was the main author. Irwin contributed with advice regarding

the work.

Paper II: Raza was the main author. Irwin contributed with advice

regard-ing the work. Fagerström supported evaluation of the results.

Paper III: Raza was the main author and performed the data analysis and

evaluation of foundry rejection data. Silva collected the data related to pro-cess variation in foundry. Irwin contributed with advice regarding the work. Fagerström and Jarfors supported evaluation of the results.

Paper IV: Raza was the main author. Svenningsson proposed the work and

contributed in analysis of the results. Irwin contributed with advice regarding the work.

Paper V: Raza was the main author. Svenningsson contributed in data

anal-ysis. Irwin contributed with advice regarding the work. Fagerström support-ed evaluation of the results. Jarfors propossupport-ed the work and contributsupport-ed with advice concerning it.

Paper VI: Raza was the main author. Svenningsson contributed in

simula-tion work. Irwin contributed with advice regarding the work. Fagerström contributed in designing experimental plan. Jarfors supported in analysis of the data.

(14)

vii

1 INTRODUCTION

This chapter starts with brief introduction to the investment casting process, the current challenges and the significance of thin-walled light weight cast-ings in power and aero turbines. The state of the art in foundry practice and the recent trends for process development for thin-walled light weight cast-ing are presented. Increascast-ing application of simulation in castcast-ing design pro-cess and its limitations are discussed. The literature review and discussion highlighted the significance of process related variations and their impact on the quality of casting as well as on predictive capabilities of simulation.

1.1 Background

In the power generation and aerospace industries, the overall weight reduc-tion of engineering systems is much sought after, especially for turbines, to increase efficiency, sustainability and reduced environmental impact. Weight reduction of an engineering system can be achieved either by using integrat-ed multifunction components or using lightweight materials. This is resulting in increased demands by the turbine industry to produce complex thin-walled cast components [1]. The ability to produce components in lower wall thickness will not only help to improve the efficiency of engineering systems but also help to reduce the consumption of raw material used to produce the components [2].

Engineering components used in power generation and aerospace industries are made from superalloys. Nickel base heat resistant superalloys are used for manufacturing of hot-gas-path components for aircraft engines since their

invention in early 20th century [3]. The combination of high strength at

ele-vated temperature, toughness and relative ease of manufacturability made these alloys a cost efficient choice for complicated geometries in aircraft engine. Development of more complex heat treatment processes and hot isostatic pressing (HIP) process has resulted in improvement in the alloys mechanical properties. The components are mostly produced using invest-ment casting since other processing methods are usually less advantageous due to the high mechanical strength and hardness of superalloys, resulting in poor formability even at high temperature [4].

The investment casting process has been an important method to produce components for jet engines since 50’s [5]. It offers excellent tolerances and

(17)

2

surface finish with freedom of design for complex geometries. The invest-ment casting process, also called the lost wax process employ a wax model around which a ceramic shell is formed. The wax is then melted out of ce-ramic shell leaving a net shape cavity inside a cece-ramic mould. The molten metal is then poured into cavity and solidified. When the ceramic shell is subsequently removed, a casting has been created which can have complex shapes and designs [6]. In addition, the investment casting process offers more efficient material usage which results in relatively lower manufacturing cost.

The ability to produce thin-walled investment cast components can result in overall weight reduction of components. However, casting of thin-walled structures is challenging due to premature solidification and long feeding distances often resulting in incomplete filling, cold shuts and shrinkage po-rosity [7]. In addition to that, the investment casting process also has disad-vantages due to susceptibility of superalloys to segregation, porosity and grain coarsening during solidification[4]. Although a lot of work has been done to understand melt flow and solidification behavior [7], the effect of foundry practices on the castability of thin-walled castings are often over-looked. A better understanding of the casting process and improved methods for design of gating and feeding system can result in improved castability, especially for thin-walled castings [8].

This work is focused on developing methods to improve castability for thin-walled castings by enhancing the process reliability which is otherwise af-fected by the semi-automated nature of the process in investment casting foundries.

1.2 Literature review

Castability of molten metal depends upon the fluidity of the melt as well as its solidification rate, segregation, interaction with atmospheric gases and solidification contractions [9]. The term castability is defined here as the ability to cast sound castings with minimum defects. Fluidity, as an empiri-cal concept used in the foundry, is the ability of molten metal to ﬂow and fill the details in the mould [10]. Fluidity measurements are not directly recip-rocal of viscosity and are not presented as a unique property of a certain alloy composition but largely depends on the test-piece used to measure the fluidity length [6]. The fluidity of molten metals depends upon number of materials properties. Some of the important properties are listed below [6].

• Temperature • Solidiﬁcation mode • Viscosity of melt • Composition • Rate of ﬂow

(18)

3

• Thermal conductivity • Heat of fusion • Surface tension

The concepts of fluidity can be separated into two more definable aspects, i.e. flowability and fillability [11]. In foundry practice, flowability is a dy-namic criterion and defined as ability of molten metal to flow which usually depends on melt properties and cooling conditions, for example, composition of alloy, viscosity, heat transfer rate etc. Fillability on the other hand is a static criterion and depend on the surface tension between flowing liquid and adjacent mould material [11]. Flowability limits the fluidity when metal solidify prematurely due to the heat and mass flow whereas fillability limits the fluidity when molten metal cannot reach the fine details of mould due to lack of required metallostatic pressure to overcome surface tension [10]. The fluidity length, i.e. flowability, is a function of properties related to both the alloy and mould system. Flowability is also directly proportional to the thickness of the casting and the heat transfer coefficient. Also increasing the heat content of the alloy at the same time as improving the insulation of the mould would increase flowability. Fillability is related to the ability to fill small cross sections or small features and is due to the difference in actual metal pressure and surface tension of the metal. In literature [10] section thickness of 1.0 mm is mentioned as the threshold when fillability becomes critical. A section thickness threshold of 2.5 mm has also been reported [12] as limit below which, surface tension become dominant.

Although there is no exact limit when the influence of surface tension takes over, its importance increases as the section thickness decreases. The selec-tion of a test geometry for fluidity measurements depends upon the aspects of fluidity under consideration i.e. fillability and/or flowability, alloy solidification morphology and the casting process. Different test geometries used to measure fluidity are reported in the literature [10]. Several parame-ters related to both mould system, pouring and alloy affect the ability to fill a thin section.

In the foundry, both fillability and flowability are highly affected by the casting conditions due to the transient nature of the process [5]. Mould and material properties, metallostatic pressure and the fill rate changes as soon as metal is poured into the mould. Melt viscosity changes as the molten metal cools [7]. Any variation in casting conditions results in unpredicted casting results.

The next section describes the effect of variation in casting parameters on the fluidity of metals as reported in the literature.

1.2.1 On the fluidity of metals

As the ﬂuidity length is dependent on heat ﬂux from molten metal to the mould [6], which is a function of the temperature gradient between the metal

(19)

4

and mould, an increase in mould temperature would be beneficial for extend-ing fluid transport [13, 14]. The fluidity length is inversely proportional to the difference between melt temperature and mould temperature. However an increase in mould temperature has only limited effect on fillability [10]. The effect of mould temperature becomes less significant for thin-walled castings, especially when the flow rate is low [13]. This is attributed to the fact that fillability is the controlling mechanism in thin sections and due to the back pressure exerted by surface tension, flow rate is reduced resulting in premature solidification.

The influence on fluidity of a change in pouring temperature is well docu-mented [10]. A linear relation between fluidity and an increase in pouring temperature has been reported [10]. However, above a certain critical tem-perature, an increased superheat of the melt will not improve fillability and interaction between metal head and surface tension will determine fillability [10].

Metallostatic pressure which is defined as product of melt density and the gravitational force applied on the melt as described in equation 1.1, influences both, the flowability and fillability [10].

P_metal=ρgh 1.1

Where

P_metal is the metallostatic pressure (Pa) ρ is the density of melt (kg m-3

) g is the gravity constant (9.8 m s-2)

h is the height of melt column before it enters the cavity (m)

The increased metal head increases the ﬂuidity length and increases the ﬁlla-bility into the thin cross-sections due to the available force by which the melt is pressing the liquid melt into narrow sections counter balancing the effect of surface tension [10]. Assuming non-wetting conditions, the following conditions presented in equation 1.2 need to be met for metal to enter into a thin section [15]. ρgh- Pmould > γ R 1.2 Where

ρ is the density of melt (kg m-3 ) g is the gravity constant (9.8 m s-2₎

h is the height of melt column before it enters the cavity (m)

Pmould is the backpressure of gases inside the mould and surface tension effect

(Pa)

γ is the surface tension (N)

(20)

5

It has been reported that casting under vacuum may increase fluidity because of a reduction in back pressure due to evacuating gasses before pouring [16]. A linear relation between metal head and fluidity is reported [10, 17]. It has been reported [13] that the flowability can be increased by increasing flow rate. A 70% improved fluidity was observed by increasing the flow rate from 100g/s to 300g/s in a pre-heated (970 ̊ C) mould for a blade like

geome-try. By increasing the mould preheat temperature to 1150 ̊ C, a further

in-crease in the flowability is reported. However, the flow rate used in the above-mentioned work [13] is low as compared to the actual casting condi-tions in foundries [13]. In order to avoid premature freezing at the inlet of mould cavity, the flow rate should be high at the mould entrance [10] . Presented in equation 1.3 is an analytical expression relating velocity to metallostatic head at the flow channel entrance [18].

𝑉𝑉0=[ 2g(Z-y) 1+ 𝜙𝜙 ] 1/2 1.3 Where

𝑉𝑉0 is the velocity of metal as it enters the test channel (m s−1)

g is the acceleration due to gravity (9.8 m s-2 ) Z is the height from entrance up to the free surface (m)

y is the decrease in effective metal head due to surface tension (m)

ϕ is a dimensionless number accounting for head loss at test channel entrance

As can be seen that entrance losses due to surface tension can substantially

lower the velocity, 𝑉𝑉₀ and hence fluidity. Also from [18], we have an

ex-pression, as shown in equation 1.4, that relates the decrease in metal head due to the surface tension.

y=[_Rρg4γ ] 1.4

Where

y is the decrease in effective metal head due to surface tension (m) g is the acceleration due to gravity (9.8 m s-2)

ρ is the density of melt (kg m-3 )

R is the radii of curvature of meniscus (m-1) γ is the surface tension (N)

This relation implies that surface tension, γ, is an important factor affecting the velocity. The drawback when increasing the velocity is that turbulence is expected to be higher causing defects in the casting [8]. A useful expression, as presented in equation 1.5 to evaluate the ﬂow behavior inside the mould cavity is Reynolds number [5]. A low Reynolds number means less turbulent ﬂow.

(21)

6 NRe=[ 4R_HVd₁ μ ] 1.5 Where

NRe Reynolds number, a dimensionless number

R_H is the hydraulic radius of the runner (ratio of cross-sectional area to its pe-rimeter) (m)

μ d1

is the kinematic viscosity of the melt i.e. viscosity of a fluid per unit density (m2/s)

V is the velocity of the melt (m s−1)

It has been found experimentally [5] that ﬂow is laminar when the Reynolds number is below 2000 where melt flow is dominated by viscosity. Above that value, melt flow become turbulent due to inertia.

The mode of filling and tendency for surface break up where the flow goes from continuous to intermittent can be estimated by the Weber number as presented in equation 1.6. W_e=ρrv 2 γ 1.6 Where

We is weber number, a dimensionless number

ρ is the density of melt (kg m-3 ) γ is the surface tension (N) v is the velocity of the melt (m s−1)

r is the hydraulic diameter (wetted perimeter) (m)

Turbulent breakup of the surface during filling will occur when We >100

[19].

The transient nature of mould filling during casting process results in non-steady state conditions where mould filling is affected by many variables mentioned above in this section. These transient conditions are important to consider when designing gating and feeding system. Simulation programs, while used to facilitate gating design process, rely on simplified assumptions to treat flow conditions as steady state which results in approximate calcula-tions. The predictive capabilities of simulation are dependent on the accura-cy of basic material properties and the precision in defining boundary condi-tions which replicate the mould filling condicondi-tions.

The following section describes the effects of thermo-physical properties of mould and alloy on the fluidity as well as highlight the uncertainty associat-ed with availability and reliability of such data in literature. The section also covers some of the reported thermal properties of the alloys used in this work.

(22)

7

1.2.2 Thermal properties of alloys and mould materials

In a casting process, fluidity length is a function of thermal properties related to both the alloy and the mould system. An analytical expression for the fluidity length is presented in Equation 1.7 [17].

𝐿𝐿𝑓𝑓 = 𝜌𝜌𝑉𝑉0 _{2ℎ (𝑇𝑇}𝑡𝑡 (𝑘𝑘𝐻𝐻𝑓𝑓+ 𝐶𝐶𝑝𝑝𝛥𝛥𝑇𝑇𝑆𝑆) 𝐶𝐶 − 𝑇𝑇𝑀𝑀) [1 + ℎ 2 � 𝜋𝜋𝛥𝛥𝜋𝜋 𝑉𝑉0𝐶𝐶𝑝𝑝′𝜌𝜌′𝑘𝑘′� 1 2 ] 1.7 where

ρ is the density of liquid (kg m-3 ) 𝑡𝑡 is the thickness of thin flat test strip (m)

𝑉𝑉0 is the velocity of the liquid at entrance to the section (m s−1)

𝑘𝑘 is the fraction of solidified alloy 𝐻𝐻𝑓𝑓 is the enthalpy of fusion (J kg-1)

𝐶𝐶𝑃𝑃 is the specific heat of the liquid (J kg-1 K-1)

𝛥𝛥𝑇𝑇𝑆𝑆 is the superheat defined as ((𝑇𝑇_𝐶𝐶− 𝑇𝑇_𝐿𝐿) (K)

ℎ is the heat transfer coefficient at the metal mould interface (W m-2 K-1). 𝑇𝑇𝑀𝑀 is initial mould temperature (K)

𝑇𝑇𝑐𝑐 is temperature of liquid metal at the time of pour (K)

𝛥𝛥𝜋𝜋 is the length of choking zone in alloy solidification (m) 𝑘𝑘′ is the thermal conductivity of the mould (W m-1

K-1) 𝜌𝜌′_{is the density of the mould (kg m}-3

) 𝐶𝐶𝑝𝑝′ is the specific heat of the mould (J kg-1 K-1)

In addition to the significance of thermo-physical properties of alloy and mould material, the equation also shows the dependency of fluidity length on casting temperature, mould temperature and mould thickness. In order to acquire thermal properties of alloys and mould materials, experimental char-acterization methods can be employed. These methods include thermal anal-ysis techniques as well as reverse modeling using computation program to find best fit of properties based on known values of material behavior. There are well established methods for measuring alloy properties using experimental techniques, however, these measurements during heating and cooling are made in an equilibrium state which is not the case in casting where cooling rate during solidification is much higher. Similarly, for mould material, the measurement of thermo-physical properties is challenging as multiple layers of different materials are used in mould making, and the po-rosity (and inter connected popo-rosity) in the mould results in non-uniform thermal properties [20]. The metastable nature of mould material that under-goes several thermal cycles before final casting also results in unpredicted thermal properties of the mould at the time of melt pouring and solidification [21]. Reported thermo-physical properties of alloys and mould material used in this work are presented in following sections.

1.2.2.1 Thermal properties of Alloy 17-4 PH

Alloy 17-4 PH is a precipitation hardening martensitic stainless steel with Cu and Nb/Cb additions. The alloy is used in applications requiring high

(23)

8

strength and a moderate level of corrosion resistance. Experimentally meas-ured thermal properties of Alloy 17-4 PH were reviewed in this work. Rack [22] reported specific heat, thermal diffusivity, linear thermal expansion and thermal conductivity data measured as function of temperature. Figure 1.1a show the differential scanning calorimetry (DSC) measurements for specific heat at temperature range between 350 K to 925 K. Figure 1.1b shows ther-mal diffusivity measured at 294 K to 1127 K using a laser flash method. Thermal expansion data measured at temperature range between 298 K to 1178 K by using a dilatometer is presented in Figure 1.1c.[22] Thermal con-ductivity can be calculated using specific heat 𝐶𝐶_𝑃𝑃, thermal diffusivity α and the density ρ corrected for temperature changes relative to the room tempera-ture. The calculated density of 17-4 PH was reported to decrease from 7.75 g.cm-3 to 7.3 g.cm-3 at temperature range 20 ̊ C to 1200 ̊ C [23].

(24)

9

b)

c)

Figure 1.1 Thermo-physical data for 17 4 PH stainless steel a) specific heat b) thermal diffusivity c) percentage thermal expansion [22]

1.2.2.2 Thermal properties of Alloy IN 718

IN718 can be referred to as nickel iron alloy [4]. It is a heat and corrosion resistant alloy [24]. Since its development in 1960’s, the alloy has been used

(25)

10

in wide range of high temperature applications in different industries, espe-cially for manufacturing of hot-gas-path components in gas turbines [25].

The presence of body centered tetragonal γ″ phase in IN718 not only provide

great strength but also improves weldability of the alloy [26]. Figure 1.2 shows thermal properties of IN 718 as reported in literature.

a)

(26)

11

c)

Figure 1.2 Thermo-physical data for IN718l, a) specific heat [27, 28], b) thermal diffusivity [28, 29], c) density [30]

1.2.2.3 Thermal properties of Alloy Udimet 500

Udimet 500 is a nickel chromium cobalt alloy commonly used in gas tur-bines in mid-1960’s [31]. The alloy is discontinued and has been replaced with Udimet 520 [32], however, it is still a commonly cast alloy as it is used in maintaining and refurbishing older industrial turbines. The alloy has good ductility and can be formed by conventional forming techniques [33]. The

thermal expansion coefficient for Udimet 500 is reported between 1.28x10-5

K-1 to 1.80x10-5 K-1 for a temperature range between 400 K to 1300 K.

Simi-larly, the reported density of Udimet 500 is 8.053 g.cm-3 [34]. Reported

thermal conductivity of the alloy is 11.3 W.m-1K-1 [35]. The specific heat

capacity of Udimet 500 reported in the literature is to be 586 J.kg-1 K-1 at

100 ̊C [33].

1.2.2.4 Thermal properties of Mould material

Investment casting moulds are made by applying number of separate layers of ceramic slurry followed by coating with course grit, often called stucco. The slurry is usually water based or alcohol-based silica binder with addi-tional additives to improve the rheology of the slurry. Although, aluminosili-cate slurry is also used in investment casting foundries, the silica based slur-ry is reported to be more stable [36].

The material used for stucco, varies depending upon availability and cost. The ceramic materials commonly used as stucco are silica sand, aluminosili-cate and alumina. It has been reported that moulds made using a fused silica stucco have better mechanical properties when used in combination with silica slurries [36]. Fused silica moulds are reported to be lighter in weight

(27)

12

due to lower specific gravity. The thermal conductivity is also lower for moulds with fused silica as backup layers compared to moulds made with aluminosilicate [36].

The thermal conductivity (λ) of the mould material is an important property as it determines the rate of heat flow through the mould away from the melt which defines the solidification time [37]. A higher value of λ shortens the solidification time and increases maximum temperature of the mould. Values of thermal conductivity of silica based mould measured using the hotwired

method are reported to range from 0.55 W.m-1K-1 to 0.75 W.m-1K-1 while

slightly lower values 0.45 W.m-1K-1 to 0.60 W.m-1K-1 are reported for mould material containing zircon flour [20]. However, the anisotropic nature of mould material in directionally layered structures is not considered in the hotwired method where isotropic conditions are assumed in all radial direc-tions [20]. For aluminosilicate based mould material, the value of thermal conductivity is reported between 0.8 W.m-1K-1 to 1.2 W.m-1K-1 at tempera-ture range 573 K to 1173 K. The measurements were made by passing a measurable heat flux through a known thickness of mould sample and meas-uring resulting temperature gradient to calculate thermal conductivity [38].

The specific heat capacity (𝐶𝐶_𝑃𝑃) of mould defines the amount of energy

need-ed to heat up the mould to a certain temperature. The higher the 𝐶𝐶_𝑃𝑃the short-er the solidification time [37]. The specific heat capacity of silica based mould material has been reported between 700 J.kg-1̊ C-1 to 1300 J.kg-1̊ C-1

at temperature range between 200 ̊ C to 1200 ̊ C when measured using an

improved laser flash method [20]. A higher value of 𝐶𝐶_𝑃𝑃, i.e. 700 J.kg-1̊ C-1

to 1800 J.kg-1 ̊ C-1 is reported for aluminosilicate moulds when measured

using improved laser flash method at temperature range between 200 ̊ C to

1200 ̊ C [20]. The lower heat capacity of fused silica compared to

alumino-silicate also makes it a favorable choice for thin-walled castings [36].

Zirconium silicate, or zircon, is used as grit for the primary layer as well as filler material in the slurry [5]. Most aerospace foundries use fine zircon flour (325 mesh) in the slurry for prime layers. However, It has been report-ed that the thermo-physical properties of zircon basreport-ed face coat have a neg-ligible effect on solidification behavior of a casting [39]. For manufacture of thin-walled castings, a coarser prime layer (120 mesh flour) has been report-ed beneficial due to the lower heat transfer coefficient resulting from less available contact surface between mould and melt [36].

1.2.3 Simulation aided design of the casting process

In investment casting foundries, traditionally the gating and feeding design has relied on the skills of casting technicians and engineers based on trial and error experiments to find an optimized gating and feeding design result-ing in high cost and long development lead times. An optimized gatresult-ing and feeding system should fill the mould with a sufficient pressure gradient and

(28)

13

uninterrupted, laminar supply of melt to the mould cavity while compensat-ing for solidification shrinkages [8]. Design parameters that typically vary from one design to another include location of in-gates, position-ing/orientation of the mould cavity with respect to gating, flow pattern varia-tion etc. [40].

The use of simulation has become a vital step to design optimized gating in many foundries. However, the complexity of mathematics required to model the flow and solidification process results in increased simulation time and requires advanced computing systems as well as user’s experience to inter-pret the results [41]. Without a systematic simulation workflow, the simula-tion process may result in unreliable representasimula-tion of casting process as well as more cost and time than actual casting trials in foundry [40]. The validity of simulation is largely dependent upon the accuracy of material properties i.e. mould, alloy, insulation and filters etc., metallurgical models used to represent flow and solidification behavior as well as accuracy in defining boundary conditions [40]. The boundary conditions are influenced by equipment and operation related variations arising from mould handling and melt pouring [41]. In addition to that, every foundry has a unique mould system with specific properties. Alloy composition also varies with in certain range depending upon the source of material. Similarly, properties of insula-tion material, the type and pour size of filters as well as casting environment factors such as humidity are foundry and process specific and can vary sig-nificantly between foundries [42].

The effect of casting parameters such as pouring rate, casting temperature and mould pre-heat temperature on mould filling has been extensively inves-tigated in literature [10]. However, the variations in the targeted values of these parameters in foundries due to the semi-automated nature of operation and equipment wear is not reported and discussed. Similarly, although the physical principles governing fill and solidification are well established, the relative importance of fluctuation in casting parameters on simulation of castability of thin-walled components is not reported as is also concluded in another study [43]. There is also lack of literature on effective use of simula-tion tools for design of casting process and mould feeder system.

This work is focused on evaluating the effect of process variations on casta-bility of thin-walled structures. The results are used to proposes a gating and process design methodology, as presented in section 4.4. The proposed methodology employs simulation based process design framework to devel-op reliable casting process in order to reduce the manufacturing lead time and process development cost as well as minimize foundry rejections.

(29)

(30)

15

2 RESEARCH APPROACH

This chapter describes the methodology employed for the conducted re-search. The motivation for the research and formulated research questions are presented. The validity and quality of research approach used in this work is also briefly discussed.

2.1 Aim and objective of this work

The aim of this work was to support the development of the ability to manu-facture components with lower wall thickness typically below 2mm to re-duce overall weight of engineering systems, such as turbines in order to im-prove fuel consumption and reduce environmental hazardous emissions. The objective was to develop a process design methodology for thin-walled castings, to improve the reliability of manufacturing process and to reduce the lead time for process development in general and in particular for thin-walled components that commonly are difficult to cast with sufficient soundness.

The scope of the work is:

 To identify and evaluate the effects of the process variations on castability of thin-walled castings.

 To evaluate the effect of process variation on predictive capabilities of simulation.

 To evaluate the use of simulation based approach in designing cast-ing process for thin-walled castcast-ings.

 Developing process design methodology to produce thin-walled components efficiently.

2.2 Delimitations

Castability of thin-walled sections is affected by casting conditions during mould filling and solidification as presented in previous chapter. However, this work is focused on the limitations imposed by casting process related variations and resulting challenges in casting of thin-walled sections. While evaluating effects of casting process related variations,

(31)

16

 Only variations in casting parameters were investigated in detail. Variations related to other process parameters, such as parameters related to pattern assembly, mould making and foundry environment were not investigated.

 All experiments and investigations were performed at one foundry and hence process variations are foundry specific.

 Simulation calculation time was a restriction as too many iterations available for process development phase can take several days to perform simulations.

 In simulation based process design framework, the design was opti-mized only for shrinkage and misrun as not all defects can be pre-dicted accurately with currently available casting simulation tools [44].

2.3 Research questions

Although, the physics describing the castability of metals in foundries is well established, the foundry specific variations in critical casting parameters and their impact on casting quality is not well documented. These process varia-tions limit the use of simulation as these variavaria-tions are hard to predict and replicate when defining boundary conditions with simulation software. A systematic approach to minimize these variations as well as framework to implement simulation in design process can result in faster development process in foundries. In order to develop a process design methodology to aid manufacturing process, following aspects are classified and further developed into research questions.

 All foundry processes generate a certain level of rejection that is closely related to the type of casting, the processes used and the equipment available. It is common in foundries that the causes of re-jections remain unknown due to the complexity of the manufactur-ing process and the manual nature of the work. Castmanufactur-ing of thin-sections is challenging due to premature solidification in thin-walled sections and long feeding distances often resulting in incomplete fill-ing, cold shuts and shrinkage porosity. Although relation between casting parameters and castablity has been investigated in detail, the significance of process parameters in relation to common foundry defects has not been well documented.

RQ 1: What are the important process parameters for castability of thin-walled structures in a foundry set-up? (Papers I, II and III)

(32)

17

 By identifying critical variations in the process and their effect on casting parameters, it is possible to optimize the casting conditions to improve castability. Although the effects of casting parameters on castability in general is well established, the effect of process varia-tion and resulting fluctuavaria-tions in the targeted values of casting pa-rameters hence the reliability of casting process is not investigated and reported.

RQ 2: What are the sources of variations in the casting process and how do they affect casting parameters? (Papers III and IV)

 The characterization of material and identification of boundary con-ditions is important for reliable simulations due to the fact that every foundry has unique set-up in terms of its mould system, alloys and the pouring conditions. Although, significance of material data and boundary conditions has been reported in literature, the change in boundary conditions due to process variations and resulting effect on validity of simulation for thin-walled castings is not investigated and reported.

RQ 3: How accurately does simulation predicts the casta-bility of thin-walled castings? (Paper V)

 The complexity of mathematical models required to model flow and solidification processes has resulted in increased simulation time and has required advanced computing systems as well as experienced users to interpret the results. Without a systematic framework, the simulation aided design process may result in more cost and time than development by trial and error approach. There is lack of litera-ture addressing systemic implementation of simulation in process design for thin-walled castings.

RQ 4: How can simulation be implemented in design of the casting process to achieve reliability and repeatability in manufacturing of thin-walled castings? (Paper VI)

2.4 The research process

Through empirical observation and analysis relevant to casting defects in thin-walled castings, defect formation mechanisms in thin-walled castings was defined and evaluated by performing experiments as well as validated using simulations. Methods to avoid the conditions which result in defect formation were proposed and evaluated. The results of the research are

(33)

trans-18

formed into a process design methodology to reduce lead time and to devel-op reliable manufacturing process for thin-walled castings. Figure 2.1 shows the research process where the aim and objective of the work are outlined.

Figure 2.1 Research process and objectives

By performing literature survey and gap analysis, scope of the work was defined, presented as problem statement in Figure 2.1. Based on the problem statement, the scope of work was divided in to four research questions. In order to address the research questions, research work was conducted as described in steps below.

Step 1: Theoretical knowledge to develop cause and effect diagrams

In this cycle the research question 1 was answered by acquiring theoretical knowledge on various aspects related to the fluidity of metals, especially in investment cast moulds as summarized in section 1.2. This provided baseline knowledge on defect formation mechanism for thin-walled castings which led to design further studies. The literature review was conducted according to guidelines proposed by Rumsey [45]. In addition to a literature review, corrective measures taken in the foundry to eliminate defects were investi-gated. Also archived reports on completed projects were reviewed to identify the most significant and commonly occurring casting defects as well as rele-vant casting parameters that influence defect formation, as described for

(34)

19

study 1 in Table 2.1. The cause of variation in defect rate from one produc-tion run to another were evaluated using experimental studies in the next cycle.

Step 2: Experimental work to evaluate the effect of variation in casting

parameters

In this cycle, research question 2 was answered by identifying the process uncertainties originating from the casting operations and equipment as well as their impact on casting parameters was established, as described in Study 2 in Table 2.1. Experiments were needed, as there were no availible data covering the effects of variation in casting parameters. In order to evaluate significance of variation in casting parameters, a range of values of casting parameters were intentionally induced in an experimental plan and tested in the foundry using thin-walled test geometries as described in Study 3a, Table 2.1. This provided knowledge about the boundary conditions that influence the effect of established crucial casting parameters. In next cycle the results were validated using simulations.

Step 3: Simulation work to verify the effect of process variation

In this cycle, the research question 3 was answered. In order to evaluate the predictive capabilities of the commercial simulation tool [46] under best practice conditions as well as to verify the significance of process variations, the foundry conditions were replicated in simulation software and were compared with specially designed test geometries, as described in study 3b, Table 2.1. The results were used in next cycle where a simulation based pro-cess design framework was proposed and evaluated.

Step 4: Concept of simulation based process design framework

In this cycle, research question 4 was answered by developing framework for a reliable casting process using simulation and design of experiments (DoE) approach. The Response surface method was employed in DoE [47]. The Nelder–Mead method was used in identifying optimized values of the casting parameters [48]. The validity of framework was evaluated using a commercial cast component in the foundry and a process window for im-proved casting quality was identified. The response of tested framework, i.e. the optimized process settings was evaluated for validity in relation to estab-lished science describing how casting parameters effect castability. This process is described for study 4a in Table 2.1. The framework was used in the implementation of the process design methodology to support the relia-ble casting process development and reduce lead time for manufacturing of thin-walled components.

(35)

20

As an outcome of this thesis, a process design methodology is proposed which builds on the findings of this research work. The abstract concepts need to be evaluated by testing on various commercial products in the found-ry and should be documented and communicated with experienced industrial engineers and the research community for verification. Study 4b, Table 2.1 describes the process. This learning cycle is iterated until the research objec-tive is realized.

2.5 Quality of the research approach

Due to the complexity of the casting process, many parameters affect the final quality of casting. Without careful experimentation it is usually not possible to identify the direct relation between the casting conditions and the quality of castings. In this work experimentation was used as a method to find cause and effect relations as well as to exclude the effect of confounding factors which are not possible in other research methods. Bottom-gated and top-gated mould filling system were used in experimental plan to ensure that results are not affected by mode of filling. Similarly, test geometries used in this work were specially designed to capture aspects of castablity focused in this work, i.e. fillablity and soundness of castings. However, experiments in an industrial environment do not allow for complete control of all the critical variables. Although this work, was performed according to an experimental plan, based on significant casting parameters, only a limited number of pa-rameters were analyzed, and thus further investigations are required. Similar-ly, the use of a simulation based process design framework has limited ap-plication and does not completely preclude the need of experimental casting trails as not all defects, such as entrainment defects and uneven grain size distribution can be predicted by existing commercial simulation tools.

(36)

D eve lopi ng proc es s de si gn m et h odol og y for Inve stm en t c as t th in -w alle d s tr u ctu re s

21

T ab le 2 .1 Ov er v iew o f r es ear ch cy cl es Stu d y Ob jectiv es M eth o d Ou tco m e RQ Pap er 1 Id en tif y in g cau se o f m ajo r fou ndr y d ef ects an d id en tif y in g r elev an t p ro ces s p ar am eter s L iter atu re s u rv ey , ev alu atio n o f co rr ectiv e m eas u res for ong oi ng pr oj ec ts i n f ou ndr y a nd r evi ew of a rc hi ve d rep o rts o n co m p let ed p ro jects . Is h ik awa d iag ra m s wer e es tab lis h ed to illu str ate th e p ri m ar y m ech an is m an d p ar am ete rs cr itical to cas tin g d ef ect f o rm atio n . 1 I, I I 2 Id en tif y in g th e so u rce o f v ar iatio n s in cas tin g p ro ces s an d ef fect o f v ar iatio n s o n cas tin g y ield E m p ir ic al stu d y b y e v alu atin g th e fiv e cas tin g o p er a-tio n s, i. e. p o u rin g ti m e, l ead p o u rin g ti m e, l ad le id lin g ti m e, s lag g en er atio n ti m e an d s lag r em o v al ti m e pe rf or m ed by f ou r c as ti ng s hi ft s. T he s hi ft be ha vi or an d r es u ltin g v ar iatio n s wer e an al y ze d . Pr o ces s v ar iatio n s r es u lted in s ig n if ican ce f lu ctu atio n s in tar g eted v alu es o f cas tin g p ar am ete rs an d h en ce v ar iatio n in q u ality o f cas tin g s. 2 III 3a E v alu atio n o f th e ef fect o f v ar iatio n in cas tin g p a-ra m eter s o n cas ta b ility o f th in -walled s tr u ctu res E x p er im en tal s tu d y b y cas tin g a tes t g eo m etr y with four di ff er ent thi ckne ss , 0. 7 m m , 1 m m , 1. 5 m m and 2 m m th ick n es s at two in ten tio n ally in d u ced lev els o f cas tin g p ar am ete rs , i. e. m o u ld te m p er atu re an d m elt te m p er atu re. T wo ty p es o f f illin g s y ste m , i .e. to p -g ated and bot to m -g ated wer e ev alu ated . Fillab ility an d cas tab ility wer e s ig n if ican tly af fected b y flu ctu atio n s in cas tin g p ar am eter s. B o tto m -g at ed s y s-te m was les s af fected b y v ar ia tio n in cas tin g p ar am et er s as co m p ar ed to to p -g ated s y ste m . 2 IV 3b E v alu atio n o f th e ef fect o f p ro ces s v ar iatio n o n p red ictiv e cap ab ilities o f si m u latio n Settin g u p a r elia b le si m u latio n to o l u sin g ch ar acte r-ized d ata f o r f o u n d ry m ater ials an d r ep licatin g th e cas tin g p ro ces s in s im u latio n s o ftwar e. Us in g a s p ecia l-ly d es ig n ed tes t g eo m etr y , s im u latio n s an d cas t s am -p les wer e co m p ar ed in ter m s o f f illab ility . Si m u latio n s o ftwar e was c ap ab le o f p red ictin g f illab il-ity b eh av io r if p ro v id ed with r eliab le m ater ial d ata an d b o u n d ar y co n d itio n s, h o wev er , th e ef fect o f s u rf ace ten si o n was n o t ac co u n ted f o r b y co m m er cial s o ftwar e us ed i n t hi s w or k. 3 V 4a T o d ev elo p a Si m u latio n and Do E b as ed p ro ces s de si gn a ppr oa ch Us in g res p o n se s u rfa ce meth o d , D oE was d es ig n ed . Si m u latio n s wer e p er fo rm ed acco rd in g to D o E and reg res sio n an aly sis was p er fo rm ed to fin d s u itab le pr oc es s c ondi ti ons . T h e s im u latio n b as ed p ro ces s d es ig n f ra m ewo rk Ou t-lin e o f p ro ces s d es ig n m eth o d o lo g y f o r r eliab le an d fas ter d ev elo p m en t o f cas tin g p ro ces s. 4 VI 4b Ver if icatio n o f p ro ces s de si gn m et hodol ogy us ing co m m er cial p ro d u cts . C o m p ar is o n o f d ev elo p m en t co st an d lead ti m e t o quot e a nd m anuf ac tur e c o m pone nt be for e i m pl em ent a-ti on a nd a ft er i m pl em ent at ion of m et hodol ogy . R ed u ctio n o f lea d ti m es f o r d ev elo p m en t; im p rove d q u ality o f q u o tes a s well as r eliab ilit y o f m an u factu rin g p ro ces s r es u ltin g in co st s av in g s. T h es is

(37)

D eve lopi ng proc es s de si gn m et h odol og y for Inve stm en t c as t th in -w alle d s tr u ctu re s

22

(38)

23

3 EXPERIMENTAL PROCEDURES

This chapter describes the experimental work and analysis techniques used in this research. Equipment and materials used in this work as well as

exper-imental procedures and conditions are also presented.

3.1 Equipment and materials

The experimental work was performed in a medium size foundry in which

casting in air and casting in vacuum was performed. Production scale

cast-ing conditions were used in all experimental work. Commercial grade alloys used in high temperature applications were used in this work as shown in Table 3.1. Mould making was performed using automated mould making line in controlled environment. Wax patterns were made using wax with 35 % filler content in a hydraulic injection moulding machines. Runners for gating and feeding were made using unfilled wax. Pattern assembly was performed manually by skilled operators.

A standard ceramic mould was used in the work consisting of three different material layers.

• First Prime coat (0.15mm) applied using water based colloidal silica

slurry containing CoAlO3 and ZrSiO4 (325Mesh) as additives and

Al2O3 (90 FEPA Grit) as stucco.

• Second prime coat (0.30mm) applied using of water based colloidal

silica slurry containing ZrSiO4 (200 mesh) and stuccoed with Al2O3

(54 FEPA Grit).

• Backup coat (8mm) applied using water based colloidal silica slurry

containing fused silica, SiO2 (270 mesh), stuccoed with AlSiO3

(16/30 mesh) and fused silica.

The insulation material used was a fiber blanket consisting of 48% Al2O3

and 52% SiO2. The filters used in the casting experiments were 10 PPI mesh

size.

Table 3.1 shows the alloys used in the experimental work and the measure-ment techniques used to characterize their properties.

(39)

24

Table 3.1 Alloys used in this research

Properties 17-4 PH IN718 Udimet 500

Composition Lab. analysis Lab. analysis Lab. analysis

Thermo-physical data Lab. analysis Lab. analysis JMat.Pro

3.2 Common foundry defects and cause of defects

As described for study 1, Table 2.1, a literature survey was performed to find reported causes for formation of different casting defects as well as to de-termine the effect of process parameters on fillability of thin-walled casting. The term process parameters here refers to parameters relevant to moulding and wax injection as well as melting and pouring whereas term casting

pa-rameters refers only to papa-rameters related to the melting and pouring

pro-cess. Additionally, archived reports [49-51] on the qualification process of commercial products in the foundry were studied and analyzed to develop cause and effect diagrams. Experiments were performed to evaluate the ef-fect of thickness of mould and insulation on feeding distance for geometries with varying thickness profile, such as turbine blades. Moulds with two lev-els of thickness were evaluated for effect of mould thickness on feeding dis-tance. Similarly, using various insulation pattern, attempts were made to determine the effect of controlled cooling on shrinkage porosity.

3.3 Process variations in casting process

The experimental work was performed at two casting stations using two different types of furnaces, as described for study 2, Table 2.1. Figure 3.1 shows the casting in air set-up where a pre-heat furnace, intermediate cast-ing ladle and meltcast-ing furnace is shown. The castcast-ing operation includcast-ing mould handling, ladle pouring, mould pouring, and mould placement are manually controlled and depends upon the skills of operators. In Figure 3.2 the casting in vacuum set-up is shown where the casting process is semi-automated. Mould handling requires operator involvement when loading the mould in to the vacuum furnace. The mould is automatically transferred from the loading chamber to the vacuum casting chamber. Similarly, the pouring rate depends upon the skill of operator as well as furnace tilt mecha-nism. Mould hanging, and positioning are mainly equipment related and highly sensitive to the wear and tear.

(40)

25

Figure 3.1 Air cast induction furnace set-up

Figure 3.2 Vacuum induction furnace set-up

The study on process variation was performed both for casting in air and

casting in vacuum. In the casting in air set-up, variations in casting

parame-ters originating from manual casting were evaluated, where as in the casting

in vacuum set-up equipment related variations were evaluated. In the casting in air set-up, casting practices of four groups of operators were evaluated.

Each group consisted of one operator and one support person. In total, 5 operations, i.e. lead pouring time, pouring time, ladle idling time, slag gener-ation time and slag removal time were analyzed. The casting practices of each group were monitored by recording their routine and execution time for 5 casting trees i.e. pattern assemblies in row. Time measurements were per-formed using a stop-watch while the operations were video recorded to veri-fy the recorded time as well as to identiveri-fy abnormalities in operations. The effect of variation in casting practices and their effect on targeted values of casting parameters, i.e. casting temperature, mould temperature and pour rate

(41)

26

was evaluated by in-situ experiments as well as by examining empirical evi-dence. In order to determine the relation between variation in targeted values of casting parameters and the casting defects, rejections reported for each group of operators were obtained from the foundry’s Enterprise resource

planning (ERP) system. Rejection statistics for one production lot from each

group were evaluated. Each production lot consisted of 25 cast trees, where each tree consisted of 20 components.

In order to evaluate the sensitivity to the variation in casting parameters for different thickness of castings, a special test geometry in four different thicknesses 0.7mm, 1mm, 1.5mm and 2mm were designed, as described for study 3a, Table 2.1. Two filling system designs, i.e. bottom-gated and top-gated were used to evaluate the sensitivity of the filling systems to the varia-tion in casting parameters. Casting temperature and mould temperatures were varied at two levels. In total, 8 test set-ups were designed for this study. Figure 3.3 shows the geometries of varying thickness mounted on two types of filling systems. Four casting trees were manufactured for each casting set-up, giving total of 32 casting trees. The geometries with thicknesses of 0.7mm and 1mm were damaged during the mould making process due to impact from rotation in slurry tanks and sand sprinkling and hence are ex-cluded from further analysis. Only 1.5mm and 2mm geometries were evalu-ated. By varying the level of values of casting temperature and mould pre-heat temperature, the effect of variation on castability of different cavity thicknesses was evaluated.

a) b)

Figure 3.3 Test geometry and the gating system designed to evaluate casta-bility of different thickness as affected by variation in casting parameters. Only 1.5mm and 2mm test geometries were analyzed in this work a)

bottom-gated system b) top- bottom-gated system

Using image analysis software, ImageJ™ [52], the filled area in each test blade was measured. X-ray analysis was performed to evaluate the shrinkage

(42)

27

porosity in each casting. Metallographic analysis of the cast samples was performed using Olympus BX51M optical microscope as well as Hitachi S-3700N scanning electron microscope. Liquid penetrant inspection was per-formed on all cast test samples to see the distribution of shrinkage.

In the vacuum furnace, variation in targeted values of casting parameters due to equipment wear and tear was evaluated and was validated with simula-tion, as described for study 3b, Table 2.1. For the simulation accuracy, the software was provided with thermo-physical data for alloy and mould mate-rial as well as boundary conditions replicating the foundry process. In order to specify boundary conditions related to mould temperature and pouring rate for simulation, in-situ experiments were performed. A ‘’K’’ type ther-mocouple was used to determine temperature inside the mould cavity. Pour-ing rate was estimated from the variation in melt stream diameter over time during the tilt pouring sequence by analyzing the video recording and using image analysis techniques using ImageJ™ software [52]. A special test ge-ometry with varying wall thickness (cross-sectional thickness) was designed to achieve certain degree of un-filled volume in order to enable comparison of extent of filled volume and thus have a quantitative and qualitative com-parison between simulation and cast samples. The test geometry is shown in Figure 3.4.

Figure 3.4 Test geometry used in evaluation of simulation predictive capabil-ities [53]

Casting trials were performed where top-gated and bottom-gated moulds were prepared and cast at two different casting temperatures and pour times.