Research on Porosity Defects of Al-Si Alloy Castings

7/23/2019 Research on Porosity Defects of Al-Si Alloy Castings

1/105

Research on Porosity Defects of Al-Si Alloy Castings

Made with Permanent Mold

MINAMI Rin

J uly, 2005


2/105

Contents

Chapter 1 Introduction 1

1.1 M otivations and Background 1

1.2 The C urrent Status of Porosity Prediction 2

1.3 The Purpose of This Research 2

Chapter 2 Literature Review __ Overview of Porosity Formation 4

2.1 D efinition of Porosity D efects for A l-alloys 4

2.2 M echanism of Porosity Form ation of A l-alloys 4

2.3 The Form ation of a G as Pore 4

2.3.1 The critical condition to form a hydrogen gas pore 4

2.3.2 N ucleation sites 5

2.4 The Volum e Shrinkage, Inerdendritic Feeding and Porosity 6

Chapter 3 Literature Review __ Porosity Prediction for Al-alloy Castings 8

3.1 M odulus and Equisolidification Tim e M ethod 8

3.1.1 M odulus m ethod 8

3.1.2 Equisolidification tim e m ethod 9

3.1.3 The deficiency of the m odulus and equisolidification tim e m ethod 10

3.2 C riterion Function M ethod 10

3.2.1 Therm al param eters used for criterion function m ethod 10

3.2.2 Tem perature gradient, G and other existing criteria 10

3.2.3 The N iyam a criterion 12


3/105

3.2.3.1 The popularity of the N iyam a criterion 13

3.2.3.2 To apply the N iyam a criterion to long freeze range (LFR ) A l-alloys 14

3.3 D irect N um erical Sim ulation M ethod 18

3.3.1 M odels considering only the density change due to solidification 18

3.3.2 M odels considering both solidification shrinkage and gas evolution 20

Chapter 4 The Influences of Controlling Parameters in Foundry 29

4.1 M etal C onstitutes 29

4.1.1 G rain refining effect of the three-m inute elem ents, Ti, Zr and V 30

4.1.2 The influences of m acro-structure on porosity 31

4.1.3 The influences of the m inute elem ents on tensile and fatigue strength

at an elevated tem perature 33

4.2 M etal Q uality 33

4.2.1 Purifying 33

4.2.2 D egassing 34

4.2.3 Q uality check and controlling 35

4.3 Si-Refining 36

4.3.1 Eutectic Si refinem ent 36

4.3.2 Prim ary Si refining _ P inoculation 37

4.3.2.1 M echanism of P Inoculation 37

4.3.2.2 Key points on P inoculation 38

4.3.2.3 P inoculation on porosity 42

4.4 C asting Processing 43

4.4.1 M old designing 43

4.4.1.1 The gate ratio 44


4/105

4.4.1.2 The pouring basin 45

4.4.1.3 The sprue 47

4.4.1.4 The runner 51

4.4.1.5 A filter applied in the gate system 52

4.4.1.6 The gate 53

4.4.2 Foundry operation 55

4.4.2.1 Pouring tem perature 55

4.4.2.2 M old cooling 56

4.5 Inserts 57

4.5.1 M etals inserts (cast iron and steel) 58

4.5.2 N on-m etal inserts (salt, fiber-reinforce m aterial, and sand-core) 59

Chapter 5. Preliminary Calculations for Using Computer Simulation

to Predict Porosity 62

5.1 The C riterion to U se __ the N iyam a C riterion, G /R 1/2 62

5.2 Things to Be N oticed W hile U sing the N iyam a C riterion 63

5.2.1 The critical value of the N iyam a criterion 63

5.2.2 The m om ent to calculate the N iyam a criterion 63

55..22..33TThheeccoooolliinnggrraatteeuusseeddffoorrccaallccuullaattiinnggtthheeNN iiyyaamm aaccrriitteerriioonn 6688

5.2.4 The influences of calculation conditions on theN iyam a criterion values 72

5.2.4.1 Elem ent sizes 72

5.2.4.2 The m old/casting interface heat resistance 72

5.2.4.3 The Initial m etal tem perature 75

5.3 C orrelation Betw een the Potential Porosity Location and the N iyam a

C riterion Values 76


5/105

Chapter 6. Reducing Porosity of Aluminum Permanent Mold Castings

in Daily Production Aided by Simulation 79

6.1 Porosity A round a N on - alum inum Insert __ Porosity at the R ing-carrier

A rea of a G ravity A l Piston 79

6.2 Porosity at a T - junction A rea _ Porosity at the Ingate A rea of a G ravity

A l Piston 86

6.2.1 The influences of the T-junctions structure 87

6.2.2 The Influence of the m old tem peratures 90

6.3 C enterline Porosity and Porosity at a H ot Spot A rea_ Porosity at a Final

Solidification A rea of an A l Squeeze C asting 92

Chapter 7 Conclusions 96

7.1 Porosity Prediction 96

7.2 The Influences of C ontrolling Param eters in Foundry 97

7.3 Reduce Porosity of A l-alloy Perm anent C asting A ided by C om puter Sim ulation 97

Acknowledgements 99

List of Publications 100


6/105

Chapter 1 Introduction

1.1 Motivations and Background

The basic principles behind casting processes are straightforw ard. M olten m etal of

sufficiently low viscosity flow s into cavities of shape com plexity, and solidifies upon

cooling. H ow ever, behind this sim ple principle lies m any com plicated reactions and

phase transform ations. If proper care is not taken, m etal castings, in particular the

alum inum alloys (A l-alloys), are prone to defects, such as porosity, one of the chronic

problem s, w hich im pact the quality of the castings and w orse the m echanical

properties, such as tensile strength and fatigue life1), 2).

Porosity form s w hen there is a gas entrapm ent, solidification shrinkage due to failure

of inerdendritic feeding, and/or precipitation of dissolved gas from the m olten m etal.

Inclusions also play an im portant role as they serve as nucleation sites for dissolved

gas and thus facilitate gas pore form ation. The effect of inerdendritic feeding is

m ainly influenced by the solidification pattern, i.e., colum nar or equiaxed grow th,

w hich is decided by alloy constitutes and the casting process param eters.

H ydrogen is the only gas dissolved to a significant extent in the m elt of A l-alloys. It

is, how ever, a constant source of difficulty for foundry-m en because it dissolves

upon reaction of m olten m etal w ith atm ospheric hum idity and the m oisture of

additions, such as constitute elem ent ingots.

The task of a m old designer and foundry engineer is to m ake an optim ized geom etric

casting design and choose proper process param eters that elim inate or m inim ize

defects evolution w hile ensuring the product shape and structure. B ut porosity

form ation is a com plex phenom enon w here the final sizes and the distribution of

porosity voids are determ ined by several strongly interacting process and alloys

1


7/105

variables. As the result, it is usually difficult to elim inate porosity com pletely from

A l-alloy castings, w hile reducing it or m oving it to an unim portant area can be a

choice.

1.2 The Current Status of Porosity Prediction

U ntil recently, the m anufacturing of m ost A l-alloy castings w as based on trial and

error. C asting process param eters or casting design geom etry w ere m odified

accordingly and the trial process w ould be interacted till desired product quality w as

achieved. This process can be tedious and tim e consum ing.

Starting from the m iddle of 1980s, due to the decreasing cost of com puters and

advances in com puting m ethods, com puter sim ulation of foundry process has been

developed and im proved by both academ ic and industry. Studies on porosity have

then stepped forw ard from experim ent-based investigations to com puter sim ulation

aided research. M ost research jobs have been done to explore the m echanism of

porosity form ation and the w ays predict it. There have been, how ever, very few

publications w hose results can be directly applied in m ass production because the

results of the studies have not been confirm ed w ith tests in m anufacturing scale.

1.3 The Purpose of This Research

W ith the purpose to figure out som e useful counterm easures w ith w hich porosity

defects can be reduced in m ass production, a through literature survey on porosity

covering the recent past thirty years has been m ade. In specific, the existing

therm al-param eter based criteria to predict porosity have been review ed. A

sum m ary regarding to the w hole foundry process, starting from m elting, m etal

treatm ent, and casting designing to process param eters controlling, based on the

2


8/105

authors w orking and research experiences, has been m ade. Various sim ulation

calculations perform ed w ith the purpose to reduce porosity defects in daily

m anufacture for som e alum inum engine com ponents have been review ed. M ost

calculation results have been verified w ith confirm ing tests and applied to m ass

production. This thesis sum m arizes the jobs done w ith the expectation that it can

be a useful guidebook to help foundry people w ho have been irritating by porosity

defects w hile m aking A l-alloy castings w ith perm anent m old.

Reference

(1) J. A . Eady and D . M . Sm ith, The Effect of Porosity on the Tensile Properties of

A l-alloy C astings M at. Forum , 9(4), 1986, pp217-223.

(2) M . J. C ouper, Casting D efects and the Fatigue Behavior of an A l-alloy C asting,

Fatigue Fracture Engineering M aterial Structure, 13 (3), 1990, pp213-227.

3


9/105

Chapter 2 Literature Review __ Overview of Porosity Formation

2.1 Definition of Porosity Defects for Al-alloys

U ndesirable voids in a solidified casting are called shrinkage or porosity defects

according to their volum e and the m ethods to detect them . Porosity is used to

express dispersed pores that are in m icro-scale and can only be detected by density

m easurem ent or m icroscopy. This kind of defects is often found in alloys w ith

m ushy solidification pattern, like A l-alloys.

2.2 Mechanism of Porosity Formation of Al-alloys

It is w ell accepted that porosity form s in A l-alloys due to the follow ing reasons:

(1) The rejection of gas, m ainly hydrogen, from the liquid m etal because of the

solubility changes during solidification;

(2) The inability of liquid m etal to feed through the inerdendritic region to

com pensate for the volum e shrinkage associated w ith the solidification.

U sually gas-pores form first, and shrinkage contributes to increase the dim ensions of

the voids. These voids, nam ed porosity, appear m ost frequently in-betw een

dendrite arm s, and in m ost cases at the areas that solidify last.

2.3 The Formation of a Gas Pore

2.3.1 The critical condition to form a hydrogen gas pore

Since hydrogen (H 2) is the only gas dissolved to a significant extent in A l-alloys, the

discussion in below w ill be concentrated on the form ation of a H 2gas pore. The

pressure difference to form a gas-pore is described by follow ing eq.1),

P = 2 /r (2 - 1)

4


10/105

w here is the interfacial energy of H 2 gas and the liquid m etal, r is the radius of the

pore, and P is the pressure difference betw een the exterior and interior of the

pore. This eq. can be w ritten as another form to show the critical radiusof pore

nucleation,

r* = -2 / P (2 - 2)

Since pore grow th is a diffusion-controlled process, the size of a pore is, therefore,

influenced not only by the hydrogen content in the m elt, but also by cooling rate

during solidification. A higher cooling rate reduces the pore size by lim iting the tim e

of the pore grow th.

2.3.2 N ucleation sites

(1) H om ogeneous nucleation

The eq. (2.2) gives the critical size by w hich w hether a nucleated pore w ill survive or

disappear can be judged. A n estim ation using 2 atom ic cross as the critical radius

and surface tension for alum inum liquid m etal gives a value for P as 30000 atm .

This reflects the real difficulty of hom ogeneous nucleation of pores in liquid m etal. In

practice, hom ogeneous nucleation is alm ost im possible because it needs such a big

interior pressure.

(2) H eterogeneous nucleation

The difficulty of nucleation is reduced by the presence of surface-active im purities in

liquid m etal, since absolute pure liquid m etal is im practical. C om paring w ith

hom ogeneous nucleation, heterogeneous nucleation is easier by a factor 2),

P het/ P hom = 1.12 {(2 cos )(1 + cos )2/4}1/2 (2 - 3)

w here is the contact angle of the m etal liquid w ith a solid (any of the im purities).

P hetand P hom are the interior pressure needed for heterogeneous nucleation and

hom ogeneous nucleation of a H 2pore, respectively. This eq. show s that a solid that

5


11/105

is com pletely w etted by the m etal liquid ( = 0, then the above factor is 1.12) is not

a favorable site for a H 2gas pore to nucleate on. O n the other hand, a solid w hich is

totally not w etted by the m etal liquid ( = 180, then the above factor is 0) is a good

site for bubble nucleation. According to John C am pbell, nucleation on solid does

becom e favorable until the contact angle exceeds 60 or 70 degrees (Fig. 2.1)2).

Figure 2.1 The pressure ratio of nucleating a gas pore heterogeneously/

hom ogeneously Vs. the contact angle

From this graph, it is clear that heterogeneous nucleation on the m ost non-w etted

solid know n (m axim um is 160) requires only about one tw entieth of the gas

pressure as required for hom ogeneous nucleation in the bulk liquid.

2.4 Volume Shrinkage, Inerdendritic Feeding and Porosity

M ost A l-alloys becom e denser in the phase change from liquid to solid. W hen the

solidification is unidirectional, feeding of volum e shrinkage is realized by the com ing

dow n of the liquid surface due to the gravitation effect in the earlier stage. Then

m etal m oves through the inerdendritic channels driven by pressure drop due to

6


12/105

solidification. Porosity voids appear at the areas solidifies last w hen none of the

feeding is successful.

For a casting w ith com plex shape, w hen the solidification occurs due to

3-dim entional heat subtraction, several liquid pools, m ost frequently partial liquid and

partial solid areas w ill form during solidification. For these isolated areas,

inerdendritic feeding is the only driving force to m ove m etal around. Theoretically

speaking, shrinkage voids w ill form at these areas, only w ith the location changed

according to the solidification order of the isolated areas them selves.

It has to be m entioned that to nucleate a shrinkage void can be m ore difficult than to

enucleate a H 2gas pore by hom ogeneous nucleation. Because the interfacial

energy betw een liquid m etal and the air has to be overcom e, and the m otive energy

like interior H 2gas pressure does not exist. O n the other hand, nucleating a

shrinkage void on a gas pore can be as easy as nucleating a gas pore

heterogeneously. As the result, shrinkage voids form ed in any of the isolated liquid

pool w ill usually not be just a pure air void, but a com bination of shrinkage voids and

gas pores. In other w ords, porosity form ation in A l-alloys is the result of H 2

gas-pore form ation and the failure of inerdendritic feeding to volum e shrinkage.

Reference

(1) D. R . Poirier, K. Yeum , and A . L. M aples, A Therm odynam ic Prediction for

M acroporosity Form ation in A lum inium -R ich A l-C u A lloys, M et. Trans. A , 18A , 1987,

pp1979-1987.

(2) J. C am pbell, C astings, B utterw orth H einem ann Ltd. O xford, 1991, pp162-173.

7


13/105

Chapter 3 Literature Review __ Porosity Prediction of Al-alloy

Castings

A lthough the phenom enon of porosity form ation has been w ell understood, the tim e

to predict the defect precisely has not yet com e. In the past fifty years, especially in

the recent tw enty years, research efforts have been m ade to predict porosity w ith

the help of com puter sim ulation. The studies m ade can be classified as the

follow ing three approaches:

(1) M odulus and equisolidification tim e m ethod, w hich determ ines the areas that

solidify last.

(2) C riteria function m ethod, w hich calculates or regresses param eters to

characterize resistance to inerdendritic feeding.

(3) C om puter sim ulation m ethod, w hich directly sim ulates the form ation of porosity

by m athem atically m odeling the solidification process.

The im portant results of these studies are review ed in the below .

3.1 Modulus and Equisolidification Time Method

3.1.1 M odulus m ethod

The m odulus m ethod is based on C hvorinovs rule1)that solidification tim e, t of a

casting area is proportional to the square of its volum e to area ratio, V/A , nam ed

m odulus.

t = B (V/A )2 (3 - 1)

B in this eq. is a factor that depends on the therm al properties of the m etal and m old

m aterial. This experim ent-based eq. has been testified by other researchers 2), 3), 4),

and w as incorporated to som e com puter program s w ith w hich the solidification order

8


14/105

of a 2 or 3- dim ensional m odel can be calculated 5), 6)(Fig 3.1).

(a) shrinkage location (b) calculated m odulus values

Fig. 3.1 Shrinkage prediction by M odulus M ethod 5)

3.1.2 Equisolidification tim e m ethod

W ith the introduction of finite elem ent/difference m ethod to foundry field,

equisolidification tim e contours or other isochronal contours could readily be

calculated 7), 8), 9). The principles of the calculations are w ell established, and the

results calculated are in good agreem ent w ith the corresponding experim ental

results in show ing the last solidification area (Fig. 3.2).

(a) porosity location (b) solidification tim e contours

Fig. 3.2 Porosity prediction of an engine block casting of Al-9.6% Si-3.8% C u by

equisolidification tim e m ethod9)

9


15/105

3.1.3 The deficiency of the m odulus and equisolidification tim e m ethod

To date, the determ ination of the areas that solidify last can be successfully carried

on either by the m odulus calculation or equisolidification tim e calculation based on

num erical sim ulation of heat transfer. In estim ating solidification sequence, the

later is m ore accurate than the form er, because m odulus calculation does not take

into account the m old tem perature variation and the m etal m aterial physical

properties. Therefore, the num erical sim ulation of heat transferring represents the

m ost im portant application of com puter sim ulation in foundry industry currently.

B ut both m ethods have their lim itation in predicting dispersed porosity, since they do

not consider such factors, as interdendritc feeding and gas evolution, w hich govern

separately or cooperatively the form ation of dispersed porosity. This approach is,

how ever, reliable in predicting gross shrinkage.

3.2 Criterion Function Method

3.2.1 Therm al param eters used for criterion function m ethod

D ue to the inefficiency of the m odulus and equisolidification tim e m ethod in

predicting centerline and dispersed porosity, the criterion function approach has

received considerable attention in porosity prediction. These criteria reflect the

lim iting conditions of interdendritc feeding. They are associated w ith therm al

param eters, such as local tem perature gradient G , cooling rate R , solidification

velocity V sand local solidification tim e tf. A com bination of these param eters, w hich

can be easily obtained from the num erical solutions of solidification heat transfer, is

often applied.

3.2.2 Tem perature gradient, G and other existing criteria

The im portance of tem perature gradient w as first proposed by B ishop et al10), and

10


16/105

developed by N iyam a et al.11)into a com puter sim ulation m ethod. This criterion

gives inform ation directly related to interdendritc flow . Therefore, it can predict

centerline porosity m ore precisely than the equisolidification tim e m ethod (Fig.3.3).

(a) solidification tim e in m in (b) tem perature gradient G in deg/cm

Fig. 3.3 C om parison of G and equisolidification tim e m ethod in predicting gross

shrinkage and centerline porosity of a steel casting (13C r-5N i)10)

The existing therm al param eter criteria proposed in literature so far, including

tem perature gradient G , are tabulated in Table 3.1.

Table 3.1. Therm al param eters based criteria for porosity prediction

Criterion Submitter Time of publicationG B ishop et al. 1951G/Vs D avies 19751/Vs

n Khan 1980

G/R1/2 Niyama et al. 1982G/Vs Lecom te-Beckers 1988

G0.33/Vs1.67 Lee et al. 1990G0.38/Vs

1.62 S.T.Kao et al. 19941/ts

mVsn F.C hiesa 1998

N om enclature:G : tem perature gradient V s: solidification velocityR : cooling rate ts: local solidification tim e

A ll these criteria can be reduced to the form of G x/V sy(x varies over the range 0~ 2

and y varies over the range of 0.25~ 1), am ong w hich the N iyam a criterion that can be

reduced to G /V sis a representative one.

11


17/105

3.2.3 The N iyam a criterion

In 1982, N iyam a et al.12)found that the critical tem perature gradient w as inversely

proportional to the square root of the solidification tim e (Fig.3.4). Therefore, they

proposed to use G /R 1/2at the end of solidification as a criterion for porosity prediction.

This criterion w as justified by D arcys Law , so that it included the physics behind the

difficulty of providing feed liquid in the last stages of solidification w hen the

interdendritc liquid channels are alm ost closed. The critical value of the criterion

w as proven to be independent of casting size, first by N iyam a et al. (Fig. 3.5), and

later by other researchers (Table 3.2)13).

Table 3.2 Proposed and calculated critical values of several solidification param eters

for centerline porosity prediciton13)

Param eters Proposed

C ritical Values

C alculated C ritical Values for Plate Thickness Listed

50m m 25m m 12.5m m 5m m

G 0.22 - 0.44 1.8 -2.2 3.6-4.4 6.6-8.0 14.6-19.7

G /R 1/2 1.0 0.92-1.1 0.93-1.08 0.83-0.98 0.94-1.07

P 0.25 0.037-0.042 0.072-0.082 0.12-0.14 0.37-0.41

tf (min)

Fig. 3.4 The relation betw een the experim entally determ ined critical tem perature

gradient G and the calculated solidification tim e tf12)

12


18/105

(min)

Fig. 3.5 The relation betw een the experim entally determ ined critical N iyam a

criterion G /R 1/2value and the calculated solidification tim e tf12)

3.2.3.1 The popularity of the N iyam a criterion

This criterion has been w idely integrated into current existing com puter softw are to

relate the output of the num erical heat transferring calculations (tem perature

gradient, solidification tim e, etc.) to em pirical findings on porosity

14 ), 15), 16 ), 17)

. The

reason of its popularity can be attributed to the follow ings:

((11))TThheeccrriitteerriioonniittsseellffiissssiimm pplleeaannddonly requires data obtainable from tem perature

m easurem ents for verification.

((33))GG //RR 11//22== ((GG //VV ss))11//22,,ww hhiilleeGG //VV ssiisstthheemm oossttiimm ppoorrttaannttppaarraamm eetteerrggoovveerrnniinnggtthhee

ccoonnssttiittuuttiioonnaalluunnddeerrccoooolliinngg,,aannddhheenncceeddeecciiddeetthheerraannggeeooffmm uusshhyyzzoonnee,,

ccoolluumm nnaarroorreeqquuiiaaxxeeddggrrooww tthhiinn solidification. Thheeccrriittiiccaallccoonnddiittiioonnooffccoolluumm nnaarr

ggrrooww tthhiiss,,G /V s m c0(1/k-1)/D ), in w hich m is liquidus slope, c0is alloy

com position, k is the equilibrium distribution coefficient, and D is diffusion

coefficient in liquid. Therefore, this criterion has essentially a close relation w ith

the solidification process, and hence porroossiittyyffoorrmm aattiioonn..

(4) TThheeffiinnaallssoolliiddiiffiiccaattiioonnaarreeaassuussuuaallllyyhhaavveeaallooww eerrvvaalluueeooffGG //RR

11//22

,,bbeeccaauusseetthheessee

13


19/105

aarreeaassuussuuaallllyyhhaassaallooww eerr G but higher V s. The form er is caused by the

deteriorated heat transferring condition at a final solidification area, w hile the later

occurs due to the phenom enon nam ed as the acceleration of solidification 18), 19).

(5) The authors have proposed a critical value of 1.0 (deg1/2m in1/2cm -1), and its

effectiveness has been verified w ith steel castings. There then exist different

values for different m aterials since the value is influenced by m aterial properties

as declared by the authors.

3.2.3.2 To apply the N iyam a criterion to long freeze range (LFR ) A l-alloys

The discrim inability of the N iyam a criterion for casting steel has been w ell know n.

The 1990s have seen a renew ed interest in testifying w hether this criterion is

efficient in predicting porosity for LFR alum inum alloys. C ontroversial opinions have

appeared.

(1) O pinions for the application

Laurent and R igaut carried out som e experim ents w ith A 356 alloy (A l-7Si-0.3M g) cast

w ith sand m old, using risers of different sizes, w ith or w ithout an end chill, grain

refinem ent, and m odification under controlled hydrogen content. A fter com paring

the result of their experim ents w ith that of the N iyam a criterion, they found that the

m inim um density (or m axim um porosity) w as located in the sam e position that

exhibited the m inim um value of the N iyam a criterion. They concluded that the

N iyam a criterion w as valuable w hen considering the relative values of the criterion

for a specific casting geom etry and m elt quality20).

H uang and B erry used a statistical program to exam ine the correlation betw een

porosity in an A l-alloy (again A 356, sand cast) and the several criteria functions21).

They found that tem perature gradient, G , and other criteria containing G , correlated

best w ith em pirical data. Therefore, they argued that G is the m ost influential

14


20/105

param eter am ong the criteria under study. They said that the critical tem perature

gradient to produce a porosity-free casting for both short freeze range (SFR ) alloys

and LFR alloys, depended upon the freezing tim e. This w as just the reason w hy

cooling rate w as introduced to form the N iyam a criterion11), 12). Their final

conclusion w as that therm al-param eter based criterion can be used to predict the

relative porosity level for a casting of LFR alloys.

(2) The opinions against the application

Tynelius et al.22)form ed a statistical m odel, w hich discussed quantitatively the

influence of alloy types and processing conditions on porosity, based on experim ents

w ith A 356 cast by both sand and perm anent m old. The local solidification tim e, tf

and solidus velocity, V sw ere said to be the m ost appropriate predictor for dispersed

porosity am ong the param eters studied. M axim um pore size increased w ith an

increase in local solidification tim e, tf. C oncerning the area pore density, a longer tf

w as beneficial for porosity form ation at low er hydrogen content w hile a shorter tf

w as beneficial at higher gas level. Increasing V sm ade the threshold hydrogen

content for porosity form ation low er, i.e., a larger V sis alw ays beneficial for porosity

form ation. The authors of this research argued the suitability of the N iyam a

criterion to LFR alum inum alloys in considering the influence of tfon m axim um pore

size and area pore density at a low er gas level.

Spittle et al.23)claim ed that the N iyam a criterion did not correlate w ith the pattern of

the m icroporosity distribution in a sim ple cylindrical casting of A l7SiM g alloy,

solidified progressively tow ards the feeder. They m ade further experim ents w ith

A l7SiM g plate castings having three different thickness (8, 15, and 25 m m ) 24), it w as

found that the highest values of porosity w ere associated w ith low G , high V s, and

long tf, in the case of the thicker plate (25 m m ). They claim ed that because of the

15


21/105

interdependence of the variables controlling percentage porosity in LFR alloys under

m ultidirectional freezing conditions, none of the variables alone could be used to

predict the locations having the highest porosity values. They suggested that

extensive experim ents under controlled directional freezing and m ultiple regressive

analyses of data to obtain a criterion function for the prediction of m icroporosity in a

given LFR alloy.

(3) O pinions for the application but w ith conditions

O verfelt et al.25)pointed out that the use of criteria functions derived from the

physical description of inerdendritic flow w as only likely to be effective w hen

dissolved gas contents w ere low , and the presence of oxides and other porosity

nucleating agents is m inim ized and w hen the overall solidification pattern from the

casting through to the feeding system is truly progressive. They also em phasized

the im portance of selecting the appropriate therm ophysical properties, particularly

the conductance at the m old-m etal interface in the calculations.

Visw anathan et al. 26), 27) m ade tem perature m easurem ents and experim entally

determ ined porosity distributions in grain refined A l-4.5% C u alloys. They

concluded that the criteria functions w ere dependent on casting conditions and alloy

solidification m ode. C asting processes and alloy types w ere categorized into four

types w ith a different criterion selected for each type (Fig 3.6).

(a) For castings w ith SFR alloys or casting processes characterized by strongly

directional heat rem oval (direct chilling or continuous casting), G /V s (V s w as

expressed w ith R in Fig. 3.6) calculated at the final stage of solidification is suitable.

(b) For castings w ith LFR alloys and high therm al conductivity, solidified in insulating

m olds, the tem perature, T or solid fraction, fsin the riser at the tim e w hen a particular

location is at the final stage of solidification is suitable.

16


22/105

(c) For processes characterized by fine dendrite arm spacing and relatively high

solidification velocities (perm anent castings), the instantaneous cooling rate, R

(expressed as in Fig. 3.6) at the final stage of solidification is suitable.

(d) For processes characterized by high tem perature gradients and low solidification

velocities (directional solidification processes), the suitable criterion is the sam e as

for group (c).

Fig. 3.6 A sum m ery of experim ental data show ing different criterion functions

suitable to predict porosity for various alloys and casting processes26)

A s can be seen all the opinions, either for or against to use the N iyam a criterion to

LFR alum inum alloys, accept the im portance of G /Vs, and it has been found that a

shorter tf(higher R ) is not alw ays good in considering porosity density. W hen the

hydrogen content is high, a shorter tf(higher R ) is beneficial to porosity form ation.

This gives a possibility to apply the N iyam a criterion to LFR alum inum alloys in

industry w here hydrogen content is w ell controlled.

17


23/105

3.3 Direct Numerical Simulation Method

Beginning from the 1980s, there have been attem pts to predict shrinkage and

porosity quantitatively by direct num erical sim ulation. Sam e as in num erical

sim ulation to calculate equisolidification tim e, a continuum is divided into infinite

sm all convenient shapes, triangular or quadrilateral. These infinite sm all shapes are

called elem ents. Efforts in this category can be classified into tw o groups: (a)

considering only the density change due to solidification; (b) considering both

solidification shrinkage and gas evolution.

3.3.1 M odels considering only the density change due to solidification

The solid fraction w as thought to be the key param eter in deciding the feeding m ode

for these m odels.

Im afuku and C hijiw a calculated the shapes of shrinkage cavities based on this

principle 28). They categorized shrinkage defects as m acroscopic and m icroscopic

cavities, nam ed shrinkage and porosity. The m echanism of the form ation, and

hence the predicting m ethods for the tw o classes of defects, w ere said to be

different. The liquid flow induced by gravity w as considered to be the cause of the

form ation of a shrinkage cavity in an isolated sem i-solid region w ith low solid fraction,

nam ely fs< fsc; w here fsand fscrepresent solid fraction and the critical solid fraction,

respectively. Solidification shrinkage in an isolated sem i-solid region w ith high solid

fraction, i.e. fs> fsc, w as considered to be the origin of porosity. They hypothesized

that the low est portion of the shrinkage cavity corresponded to the location of the

disappearance of fsc loop, and the location of the porosity corresponded to the region

of the low solid fraction gradient w hen the fscloop disappears. The m athem atical

m odel they developed to calculate the volum e of shrinkage cavity V v, generated

during a tim e step t at an isolated non-solid region w as,

18


24/105

[ ]dvvfvfVttvttstvtsl

v

sv)1()1(/ *,

*

,

*

,

*

, = (3 - 2)

in w hich v*v,tis the elem ental shrinkage volum e in an infinitesim al region of volum e

dv at tim e step t, f*s,tis the solid fraction in dv at tim e step t,is the solidification

shrinkage ratio, and land s are the density of liquid and solid, respectively. The

applicability of the m athem atical m odel to practical problem s w as verified w ith steel

castings 29). O ne of the exam ples is given in Fig. 3.7.

(a) defect location (b) fscloops (m in) (c) disappeared points (d) fsdistribution at

of fs loops fscdisappear

Fig. 3.7 Shrinkage prediction of a steel sand casting using eq. (3-2)28)

A program to sim ulate shrinkage quantitatively w as developed by N agasaka et al in

198930). In the program , a critical value of the solid fraction, fscw as assum ed to

determ ine the liquid m etal feeding m echanism , that is, liquid and m ass feeding w hen

fs< fscbut inerdendritic feeding w hen fs> fsc. M acro-shrinkage and m icro-shrinkage

w ere predicted in regions w here the solid fraction w as low er and greater than the

critical value, respectively. The solid fraction gradient, defined as fsG c, w as

considered to be the key param eter that controlled the driving force for feeding in a

high solid fraction region (fs> fsc). For steel castings w ith certain carbon content,

19


25/105

there existed a critical value of the solid fraction gradient, fscG c that w as a function of

fsaccording to the authors. For instance, the relationship for steel castings w ith

0.4% C is,

fscG c= 0.36fs2- 0.36fs+ 0.09 (3 - 3)

W hen the m axim um solid fraction gradient of an elem ent is below the critical value,

m icro-shrinkage is expected at the elem ent. O n the other hand, m ass feeding in

the low solid fraction zone occurs easily by the force of gravity. The total shrinkage

volum e in this zone generates a m acro-shrinkage cavity in the elem ents w ith a

m inim um pressure head, such as free surface elem ents. The basic concept w ith

m icro-shrinkage is sim ilar to that of Im afuku et al w ith a different eq. for calculating

of shrinkage volum e.

V s,i= fs,iV i= U ij S ijt (3 - 4)

in w hich V s,iis the shrinkage of elem ent i during t,is the solidification shrinkage

ratio,fs,iis the increm ent of solid fraction for elem ent i, V iis the volum e of elem ent i,

U ijis the velocity betw een elem ent i and its neighboring elem ent j, S ijis the area

betw een elem ent I and j, and t is the tim e step. It w as said that good agreem ent

betw een the calculated shrinkage value and the experim ental results for steel

castings had been obtained. A typical exam ple from their results is show n in Fig.

3.8.

3.3.2 M odels considering both solidification shrinkage and gas evolution

These m odels start to describe the behavior of inerdendritic flow . In 1966, Piw onka

and Flem ings introduced D arcys law , w hich is valid for fluid flow through a

perm eable m aterial, into m icroporosity prediction31). They also evaluated the

efficiency of inerdendritic feeding through the know ledge of local pressure drop.

20


26/105

Follow ing this pioneering study, m any researchers investigated the im portance of

inerdendritic feeding in m icroporosity form ation by direct num erical sim ulation 32), 33),

34), 35), 36), 37). Som e typical w orks are sum m arized in the below .

(a) X-ray result (b) density distribution (c) predicted shrinkage (d) area G < 1 C /cm

Fig. 3.8 C alculated (w ith eq.3-4) and experim ental shrinkage and porosity for a steel

casting30)

O hnaka et al.34)tried to sim ulate the form ation of shrinkage cavities in a com plicated

casting by m odeling the m otion of inerdendritic flow via solving the heat-and

m ass-conservation eq.s, and em ploying Scheil & D arcys law . It w as assum ed that

shrinkage cavities form in free surface elem ents, i.e. elem ents w here the pressure is

below a critical pressure and/or in w hich the pressure head is m inim um . H ow ever,

to use this m odel, accurate know ledge of perm eability of the inerdendritic region,

the boundary pressure for porosity form ation, the critical solid fraction to calculate

the am ount of porosity, etc., are required. This lim its its application. Besides, the

m odel w as tw o-dim ensional at the tim e it w as proposed.

Kubo and Pehlke35)developed a m athem atical m odel based on the continuity eq.,

21


27/105

D arcys law , and the conservation of energy and gas content. Their m odel

suggested that the sim ultaneous occurrence of shrinkage and gas evolution w as a

key m echanism for porosity defect form ation. M easured values of porosity in

A l-4.5% C u plate castings com pared favorably w ith their calculated values (Fig. 3.9).

They recom m ended m inim ization of gas content by degassing and increasing the

m old chilling pow er for the production of sound castings. H ow ever, the

recom m ended gas content and m old chilling pow er depend on casting shape, alloy

com position, and required m echanical properties.

Fig. 3.9 Porosity distribution of a 1.5cm -plate sand casting of A l-4.5% C u35)

Poirier et al.36)proposed a m odel to predict the form ation and the am ount of

m icroporosity form ed betw een prim ary dendritic arm s of an alum inum alloy. Their

m odel w as sim ilar to that of Kubo and Pehlke35), w ith differences in calculating the

perm eability of inerdendritic flow and how the radii of gas bubbles depend on the

sizes of the inerdendritic spaces. Their results indicated that porosity does not form

w hen the gas pressure is below the pressure in the liquid, and porosity volum e is

proportional to prim ary dendrite arm spacing (Fig. 3.10).

22


28/105

Fig. 3.10 Porosity am ount versus prim ary dendrite arm spacing (H 2- 4 10-5% )36)

A gas pore is stable provided that,

P g- P = (1/r1+ 1/r2) (3 - 5)

w here P gis the pressure of hydrogen w ithin the inerdendritic liquid, P is the local

pressure in the m ushy zone, is the surface tension of the liquid, and r1, r2are

the principle radii of curvature. According to eq. (3 - 5), it is easier for gas pores to

form am ong prim ary dendrite arm s than am ong secondary dendrite arm s in a

colum nar m ushy zone, because the spaces (r1and r2) am ong the prim ary arm s are

larger than those am ong the secondary arm s. They proposed the condition for

porosity form ation am ong prim ary dendrites arm s as,

P g- P = 4 /gld1 (3 - 6)

W here glis the local volum e fraction of liquid, and d1is the prim ary dendrite arm

spacing. The calculated value of porosity as a function of initial hydrogen

concentration agreed w ell w ith the em pirical data (Fig. 3.11). This graph also show s

that an increase in tem perature gradient (G ) and solidification rate (V s) results in less

inerdendritic porosity.

23


29/105

VVssVVss

Fig. 3.11 C alculated and em pirical data of porosity volum e versus H 2content (d1-

prim ary dendrite arm space, lines A , B and C w ere presented by Talbot)36)

Shivkum ar et al.37)developed a m athem atical m odel to sim ulate m icrostructure

evolution and m icroporosity form ation during the solidification of equiaxed structures

for alum inum alloys. Param eters such as grain size, secondary dendrite arm space

and eutectic space, together w ith the pore characteristics, such as the am ount of

porosity and pore size, can be obtained quantitatively w ith their m odel. It w as

found that both grain size and eutectic spacing varied inversely w ith cooling rate, a

slow er cooling rate resulted in a coarser secondary dendrite arm spacing because of

the longer local solidification tim e. Their results indicated that as the cooling rate

increased, there w as a reduction in the total am ount of porosity. B ut at cooling

rates greater than 50C /s, the am ount of porosity is determ ined only by the hydrogen

level (Fig. 3.12). The pore size decreased w ith an increase in cooling rate, and

cooling rate has a greater influence on pore size than does the hydrogen content.

24


30/105

Fig. 3.12 The relation betw een porosity am ount and cooling rate, initial H 2 content37)

A m ong the three approaches described above, direct num erical sim ulation gives

insight into the form ation of dispersed porosity. B ut its application is m ainly lim ited

in research field for its com plexity in use.

References

(1) N . C hvorinov, Theory of Solidification of C astings, D ie G iesserei, 27, 1940,

pp17-224.

(2) J. B. C aine, A Theoretical A pproach to the Problem of D im ensioning R isers, A FS

Trans., 56, 1948, pp492-501.

(3) W . S. Pellini, Factors W hich D eterm ine R iser Adequacy and Feeding R ange, A FS

Trans., 61, 1953, pp61-80.

(4) W . S. Pellini, Practical H eat Transfer, A FS Trans., 61, 1953, pp603-622.

(5) S. J. N eises, J. J. U icker and R . W . H eine G eom etric M odeling of D irectional

Solidification Based on Section M odulus, A FS Trans., 61, 1987, pp25-29.

(6) N . Sirilertw orakul, P. D . W ebster and T. A . D ean, Com puter Prediction of Location

25


31/105

of H eat C enters in C astings, M ater. Sci. Tech., 9, 1993, pp923-928.

(7) J. G . H enzel and J. Keverian, JO M , 17, 1965, pp561-568.

(8) A . Jeyarajan and R . D . Pehlke, A FS Trans., 86, 1978, pp457-464.

(9) H . Iw ahori, K. Yonekura, Y. Sugiyam a, Y. Ym am oto and M . N akam ura, Behavior

of Shrinkage Porosity D efects and Lim iting Solid fraction of Feeding on A l-Si

A lloys, A FS Trans., 71, 1985, pp443-451.

(10) H . F. B ishop and W . S. Pellini, The C ontribution of R iser and C asting End Effects

to Soundness of C ast Steel Bars, A FS Trans., 59, 1951, pp171.

(11) E. N iyam a, T. U chida, M . M orikaw a and S. Saito, Predicting Shrinkage in Large

Steel C astings from Tem perature G radient C alculations, A FS Inter. C ast M et. J., 6,

1981, pp16-22.

(12) E. N iyam a, T. U chida, M . M orikaw a and S. Saito, A M ethod of Shrinkage

Prediction and Its A pplication to Steel C asting Practice, Inter. Foundry C ongress

49 in C hicago, paper 10, 1982.

(13) S. M inakaw a, I. V.Sam arasekera and F. W einberg, Centerline Porosity in Plate

C astings, M etal Trans., 16B, 1985, pp823-829.

(14) H icass, U sers M anual, H itachi Research Institute of H itachi M anufacturing,

Ibaraki, Japan, 1998.

(15) AdStefan3D , U sers M anual, V6.1, H itachi Jiaohou system , Japan, 2003.

(16) The A FS Solidification System (3-D ), V2, System D ocum entation, U SA , 1994.

(17) M agm asoft, U sers M anual, V3, M AG M A G ieB erei technologic G m bH A achen,

G erm any, 1995.

(18) E. N iyam a, K. A nzai, Trans. of Japan Foundry-m ens Society, Vol.13, 1994,

pp83-88.

(19) E. N iyam a, K. A nzai, M ater. Trans., JIM , Vol. 36, N o. 1, 1995, pp61-64.

26


32/105

(20) V. Laurent and C . R igaut, Experim ental and N um erical Study of C riteria

Functions for Predicting m icroporosity in C ast A lum inum A lloys, A FS Trans., 100,

1992, pp647-656.

(21) H . H uang and J. T. Berry, Evaluation of C riteria Functions to M inim ize

M icroporosity Form ation in Long-Freezing Range Alloys, A FS Trans., 1993,

pp669-675.

(22) K. Tynelius, J. F. M ajor, D . A pelian, A Param etric Study of m icroporosity in the

A 356 C asting A lloy System , A FS Trans., 1993, pp277-284.

(23) J. A . Spittle, M . A lm ehhedani and S. G . R . B row n, The N iyam a Function and its

Proposed A pplication to M icroporosity Prediction, C ast M etals, 7, 1994, pp51-56.

(24) J. A . Spittle, S. G . R . B row n and J. G . Sullivan, Application of C riteria Functions to

the Prediction of m icroporosity Levels in C astings, Proceedings of the 4th

D ecennial International C onference on Solidification Processing, Sheffield, July

1997, pp251-255.

(25) R . A . O verfelt, R . P. Taylor and J. T. Berry, D ispersed Porosity in Long Freezing

R ange Aerospace A lloys, Proceedings of the 4th D ecennial International

C onference on Solidification Processing, Sheffield, July 1997, pp248-250.

(26) S. Visw anathan, V. K. Sikka, and H . D . B rody, U sing Solidification Param eters to

Predict Porosity D istributions in A lloy C astings, JO M , Sept., 1992, pp37-40.

(27) S. Visw anathan, V. K. Sikka, and H . D . B rody, The A pplication of Q uality C riteria

for the Prediction of Porosity in the D esign of C asting Process, M odeling of

C asting, W elding and A dvanced Solidification Processes VI, 1993, pp285-292.

(28) I.m afuku and K. C hijiw a, A M athem atical M odel for Shrinkage C avity Prediction

in Steel C astings, 91, 1983, pp527-540.

(29) I. Im afuku and K. C hijiw a, Application and C onsideration of the Shrinkage C avity

Prediction M ethod, A FS Trans., 91, 1983, pp463-474.

27


33/105

(30) Y. N agasaka, S. Kiguchi, M . N achi and J. K. B rim acom be, Three-D im ensional

C om puter Sim ulation of C asting Processes, A FS Trans., 117, 1989, pp553-563.

(31) T. S. Piw onka, M . C . Flem ings, Pore Form ation in Solidification, TM SA IM E Trans.,

236, 1966, pp1157-1165.

(32) V. de L. D avies, Feeding R ange D eterm ination by N um erically C om puted H eat

D istribution, A FS C ast Research Journal, 11, 1975, pp33-34.

(33) Y. W . Lee, E. C hang and C . F. C hieu, M odeling of Feeding Behavior of Solidifying

A l-7Si-0.3M g A lloy Plate C asting, M etall. Trans., 21B, 1990, pp715-722.

(34) I. O hnaka, Y. M ori, Y. N agasaka, and T. Fukusako, N um erical A nalysis of

Shrinkage Form ation w ithout Solid Phase M ovem ent, J. of Japan Foundrym ens

Society, 1981, pp673-679.

(35) K. Kubo, R . Phelke, M athem atical M odeling of Porosity Form ation in

Solidification, M et. Trans B, June 16B, 1985, pp359-366.

(36) D . R . Poirier, K. Yeum , and A . L. M aples, A Therm odynam ic Prediction for

M acroporosity Form ation in A lum inum -R ich A l-C u A lloys, M et. Trans. 18A , 1987,

pp1979-1987.

(37) S. Shivkum ar, D . A pelian and J. Zou, M odeling of M icrostructure Evolution and

M icroporosity Form ation in C ast Alum inum A lloys, A FS Trans., 98, 1990, pp

897-904.

28


34/105

Chapter 4 The Influences of Controlling Parameters in Foundry

Foundry is a black art! Foundry is a m iracle! These proverbs reflect the difficulty

of foundry process because so m any param eters need to be controlled at the sam e

tim e and not all of them are controllable, as they are often interactive. A lthough

foundry has a history as long as that of hum an beings, the typical foundry defects

like porosity has never disappeared and w ill continue to trouble foundrym en as long

as foundry operation continues.

In order to reduce foundry cost, foundry people are trying by all m eans to reduce

porosity defects. Precious experiences have been accum ulated in daily foundry

operations. G ood guideline textbooks can som etim es be w ritten based on such

kind of experiences. A im ed at to construct such a textbook for daily foundry of

A l-alloy perm anent m old castings, this chapter sum m arizes the im portant aspects of

gravity perm anent m old casting process based on the authors w orking and research

experiences.

4.1 Metal Constitutes

For m ost com m ercial alloys, the constitutes of the im portant elem ents are give in

ranges, w hile only upper lim its are given to those that are thoughtunim portant. For

exam ple, the constitutes of the com m ercial alum inum alloys AC 8A is defined as 1).

2.0 4.0% C u, 8.5 10.5% Si, 0.5 1.5% M g, 1.0% Fe, 0.2% Ti, etc.

Except for C u, Si and M g, all the other elem ents are only given w ith an upper lim it,

am ong w hich Ti is specified as less than 0.2% . There is no problem w ith such a

specification w hen the other m inute elem ents, such as Zr, V, do not present a

noticing level. W ith the existence of Zr and V, a sm all difference in Ti content, both

29


35/105

under 0.2% , m ay cause a totally different m acrostructure, even if all the other

foundry param eters are the sam e.

4.1.1 G rain refining effect of the three-m inute elem ents, Ti, Zr and V

The grain refining effect of Ti to alum inum alloys w as realized from 1970s 2). The

refining effect is caused by peritectic reaction betw een Ti and A l. Therefore, other

elem ents like Zr, V, etc., that can have sim ilar peritectic reaction w ould also have

grain refining effect. Such an effect of Zr w as reported in 1980s 3). The

m echanism is theoretically a nucleation phenom enon. The m etastable phase Al3Ti,

or A l3Zr, form ed from the peritectic reaction betw een A l and Ti, A l and Zr, respectively,

behave as the nuclei of alpha Al. It needs to be pointed out that once the

interm etallic com pounds becom e stable, they loss the effect of refinem ent 4).

The use of the three elem ents, Ti, Zr and V in com bination, in concentration of 0.1

0.4% , as addition to A l-Si alloys for pistons and cylinder heads, w as proposed by a

French patent 5). The m ain benefit claim ed w as the im provem ent in creep

resistance at elevated tem perature. Industries started to look for the best

com bination for the three elem ents, although the refining m echanism of V w as still

not clear. A range for each of the three elem ents started being given for som e of

industrial self-m ade alloys.

According to the authors experience, a sm all change in the am ount of the three

elem ents can cause a significant change in the m acrostructure of A l-alloy castings.

Fig. 4.1 show s the m acrostructures of an A l-Si alloy piston cast w ith exactly the

sam e perm anent m old and the sam e conditions, but a sm all differences in the three

m inute elem ents as show n below the pictures. The m acrostructure changes from

partial colum nar and partial equiaxed (a) to w hole equiaxed (b) (Fig. 4.1).

30


36/105

(a) Ti: 0.116% Zr: 0.122% V: 0.088 (b) Ti: 0.195% Zr: 0.151% V: 0.101%

Fig. 4.1 The effect of the three-m inute elem ents on m acrostructure

4.1.2 The influences of m acro-structure on porosity

D ifferent m acrostructures are the results of different solidification patterns.

Therefore, it has a correlation w ith porosity form ation. In the authors early research

on colum nar - equiaxed transition w ith cylindrical ingots 6), those ingots that

contained the largest area of equiaxed grains also contained the m ost porosity.

C onversely, those ingots contained none or few equiaxed grains contained the least

porosity as w ell. Both dye-check and m icrostructure exam ination of casting (a) and

(b) show ed that the colum nar + equiaxed structure presented no detectable porosity,

w hile the w hole equiaxed structure presented w ell-distributed fine porosity (Fig. 4.2).

Fig. 4.2 M icrostructure of area A taken from (b) of Fig. 4.1

31


37/105

To confirm the influence of m acrostructure fineness on porosity form ation,

exam inations have been done to another A l-Si piston, w hich has alm ost the w hole

equiaxed structure, but different in m acrostructure fineness due to constitutional

differences. The result w as that the finer m acrostructure contained m ore porosity

voids in various dim ensions than the coarser m acrostructure (Fig. 4.3). A conclusion

then can be m ade that the finer the m acrostructure, the m ore porosity voids w ill

form . W orth to m ention that the data show ing in Fig. 4.3 w ere taken from 5 piston

of a new A l-Si alloy w ith high C u and N i, nam ed M 174. The practical constitutes of

the alloy cannot be opened here for business reason.

0

2

4

6

8

10

12

14

16

174 - fine 174 - coarseAveragenumbersofmicro-po

rosityindifferent

dimensions


38/105

m ade w ith perm anent m old. Since the total am ount of the three m inute elem ents

has a m agic effect in controlling m acrostructure, and hence porosity, a m inor

adjustm ent in one of, or the three m inute am ount elem ents can help solve the

w ell-distributed porosity problem readily som etim es. N ow adays, spectrum analysis

has brought about a great convenience to daily routine chem ical constitute analysis.

It is therefore im portant to do the spectrum analysis regularly in daily production, and

perform extra checks w henever an abnorm al m acrostructure, like that as show n in (b)

of Fig. 4.1, is observed.

4.1.3 The influences of the m inute elem ents on tensile and fatigue strength at an

elevated tem perature

W ith the condition that there is no porosity defects, the m echanical properties of the

casting can expect an im provem ent 5). The laboratory tests in the authors com pany

have also confirm ed the im proving effects on tensile and fatigue strength at about

350 C w ith the addition of proper am ount of the three elem ents. W hether the

reasons of the strengthening effects com e from the grain refinem ent or due to the

w ell - distributed therm ally stable interm etallic com pounds, w hich behave as grain

refiner in their unstable stage, needs further investigations.

4.2 Metal Quality

A s often heard from foundrym en, there w ould be no w ay to get a casting free from

porosity, if the m etal is not clean. G ood m etal quality here m eans appropriate

chem ical com positions and a low gas content. The form er can be checked by

spectrum analysis, w hile the later need extra exam ination.

4.2.1 Purifying

In order to rem ove undesired m etal constitutes, such as K and N a, and the organic

33


39/105

m aterials m ixed from using m achinery chips as the source m aterial, C l-containing

chem icals, such as C 2C l6, is often used as purifying agent. B ut this operation is not

allow ed in the advanced countries like Japan because of the environm ent regulations.

Then, counterm easures, such as double filter system , enough calm ing tim e, etc.

should be applied in m etal transferring before pouring, if alloying is being perform ed

inside. W hen outside ingot providers are used, strict lim its should be given to

im purity constitutes w hen m aking the purchasing specifications.

4.2.2 D egassing

A s m entioned in chapter 2 (see section 2.3.1), H 2is the only gas dissolved to a

significant am ount in the m elt of A l-alloys. H ow ever, it has a constant source

because it is derived upon reaction of m olten m etal w ith atm ospheric hum idity and

m oisture of contained by w hatever the m olten m etal contacts to.

A l + H 2O A l2O 3+ H 2 (4 - 1)

(1) W ay of degassing and the agents used

Electronic B ubble Flow (EB F) w ith inert gas, such as A r or/and N 2are w idely applied

in industry to rem ove H 2off from the m elt. The degassing effect of A r and N 2are

different, even though both of them are inert gases. A proportion decided from

both cost and effect is often applied in industries. If allow ed, a certain proportion of

C l2can be m ixed into the bubbled gas, so that the purification process can be

om itted.

(2) Frequency of degassing

The absorption of H 2occurs continuously. As show n in Fig. 4.4, the hydrogen

content w ill exceed 0.25cc/g after 7 hours even if there is no touch to the m etal.

Therefore, degassing operation should be perform ed w ith a regular interval. The

interval depends on the m axim um lim it set to the m elt and the hum idity of the air,

34


40/105

w hile the later depends on seasons. In the rainy season, shorter intervals, or even

continuous degassing is often applied. It should be noticed that this operation also

causes continuous oxidization of fresh m etal, though there is no w orry for an ove

degassing.

r

Increase of Hydrogen Contenet with the time

in a Holding Furnace

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

1hr 2hr 3hr 4hr 5hr 6hr 7hr 8hr 9hr

H2Content

/100

Fig. 4.4 C hanging of hydrogen content w ith tim e

4.2.3 Q uality check and controlling

The G as content can be easily m easured by the Initial B ubble M ethod operated

under a reduced pressure. Re-degassing is perform ed if the gas content is over the

lim it set beforehand. The cleanliness of the m etal, how ever, can only be checked

through m easuring the density of the m etal, because only density is the

com prehensive reflection of porosity level. A convenient m ethod, nam ed D ensity

Index m ethod, w hich com paring the density of the sam ple solidified under a reduce

pressure, nam ely vac, w ith the density of the sam ple solidified in the air, nam ely air,

w ith the follow ing equation,

D ensity Index = ( air- vac)/ air 100 (4 - 2)

It is obvious that the index value w ill be zero, if the tw o densities are the sam e,

w hich m ean no porosity form under the vacuum condition. The low er the index

value, the better the m etal quality, w hile a upper lim it, 1.2 is often proposed for A l-Si

35


41/105

alloys. The sam ples for index density check are taken from the holding furnace right

before pouring. A nd the values set for different alloys are different in daily

production, depending on w hether the alloy has a larger/sm aller trend to form

porosity.

4.3 Si-Refining

Si is a very bristle phase as being w ell know n. O nly w hen it is w ell distributed in

the alpha m atrix, can Si optim ize the m echanical property of A l as an alloying

elem ent. O therw ise it behaves as a defect to A l alloys just as carban does to

casting iron.

4.3.1 Eutectic Si refinem ent

N ot until N a - m odification technique w as patented in 1921 7), did hypoeutectic A l-Si

alloys com e into com m ercial im portance. Later on, Sr, Sb, etc., has been found to

also have a sim ilar refining effect on eutectic S i8). This refinem ent dram atically

enhances the alloysm achinability and m echanical properties. The m odification

m echanism has been explained w ith various theories, in w hich attention has been

m ainly paid to the change of interfacial energy betw een eutectic alpha A l and Si, and

betw een S i and the m elt. It is found that there is a reduction in the interfacial

energy m entioned above upon the addition of the refining elem ents 9), though a

w idely accepted theory has yet to be established.

The reduction in w ear resistance after eutectic Si m odification, the aggravation of

high-tem perature toughness, the difficulties of this operation in foundry, such as

introducing gases to the m elt and the fading phenom ena of the refining effect, etc.,

have lim ited the application of eutectic S i m odification to the alloys of piston and

cylinder heads. Except for very sm all m otorcycle pistons, w hich solidify at a higher

36


42/105

cooling rate and w ork at a low er tem perature environm ent, eutectic m odification is

no longer being applied in piston m anufacture.

4.3.2 Prim ary Si refining _ P inoculation

Before the finding that P inoculation could control the size and shape of prim ary Si,

hypereutectic A l-Si alloy had not received serious attention because of their

brittleness and lack of m achinability caused by the large and irregular natural shape of

prim ary Si crystals. The P inoculation is, how ever, applied not only to hypereutectic

A l-Si alloys, but also to eutectic A l-Si alloys, because prim ary S i can frequently be

seen for m ost industrial eutectic A l-Si alloys, especially at a thicker area w here the

cooling rate is low . This phenom enon is explained by coupled zone theory of A l-Si

alloys 10). Just as the m icrostructure consisting of only eutectic can be obtained not

only at exact eutectic com position, prim ary Si can also be obtained at the eutectic or

even a hypoeutectic com position. O f course, other alloying elem ents can also have

their contribution in altering the eutectic com position.

4.3.2.1 M echanism of P Inoculation

P reportedly com bines w ith A l in the m elt to form tiny insoluble alum inum phosphide

(A lP) particles that, due to their close crystallographic lattice constant to Si, acts as

suitable nuclei on w hich prim ary Si grow s during solidification. Both A lP and Si have

a diam ond cubic crystal habit and a sim ilar lattice constant (Si, a0= 5.43A ; A lP, a0=

5.45A ). A lthough this theory has never been proven conclusively, a reduction in

prim ary Si size by 90-92% w ith P inoculation has been observed, and the m icroprobe

analysis has show n that the seeds of prim ary Si contain both A l and P 11).

W orthy to m ention an unexpected finding, obtained w hen confirm ing w hether P has

refining effect on eutectic S i or not, that P alters the norm al A l-Si eutectic

com position tow ard a low er Si level, thereby causing prim ary S i crystals in 11.5% Si

37


43/105

m elt12), w hich w ould norm ally contain none of prim ary Si (refer to (a) and (b) of Fig.

4.5). From this picture, it can also be seen that P has no effect on the size or shape

of eutectic Si, although there have been som e argues on this.

(a) no P addition (b) w ith P addition

Fig. 4.5 The influence of P on eutectic com position 12)

4.3.2.2 Key points on P inoculation

H aving know n the m echanism of P inoculation, there seem no hard rules applying to

this operation. H ow ever, in order to obtain the expected effect, there are som e

im portant things need attentions. W ith years of experiences in using P to A l-Si

alloys, the author considers the follow ing aspects are to be understood and

rem em bered.

(1) W hat to add?

Eutectic P-C u containing 7-8% P, instead of 15% P-C u, dissolves readily in A l-Si alloy

m elts and provides consistent and reliable refinem ent. For using in large-scale

production, shot form is suggested because it is m uch cheaper than brazing rod.

(2) W hen to add?

38


44/105

Tw o questions are raised here: at w hat tem perature and how long before pouring P

should be added.

A higher tem perature is preferred to obtain a better dispersion of A lP. B ut excess

superheat causes a higher m ean m old tem perature that slow s dow n the

solidification and hence w orse the refining effect. If the m old can be intensively

cooled, this w ill not becom e a problem . B ut the bad effects, such as hydrogen

pickup, m elt oxidation and M g burnout under a higher m elt tem perature should never

be forgotten. The P addition tem perature should generally be the desired pouring

tem perature decided based on other practical considerations such as feeding in

coping w ith the practical m old tem perature. N o excess superheat is needed, but

never add P at a tem perature low er than 705 C in order to get a good distribution

of AlP.

The refining agent should be added to the m elt after the m elt has been otherw ise

prepared (degassed for exam ple). This is because any agitation of the m elt w ill

cause agglom eration of A lP particles, flux gas bubbles also float A lP particles to

the m elt surface w here they are entrapped w ith dross and rem oved during skim m ing.

If P-C u is added in the alloying stage, a P bearing salt, such as PC l5, is suggested to

be added just before pouring in order to provide active A lP for inoculation. Action

like calm ing to rem ove gas introduced by P inoculation is alw ays suggested.

(3) H ow long the refining effect retains?

The effect of P inoculation is not perm anent. The refining effect is loosing not only

due to the loss of P from the m elt, but m ore due to agglom eration of the A lP

particles. In other w ords, chem ical analysis m ay give the sam e P percent after

com plete loss of refining effect as that during the period of effective refinem ent.

B ut cluster of A lP can be detected by high m agnification inspection 12).

39


45/105

The useful life of a refinem ent treatm ent is dependent on m elts lot-w eight and the

stirring or agitating frequency that the m elt undergoes. U nder otherw ise sim ilar

conditions, refinem ent rem ains effective longer in large m elts than in sm all ones.

For exam ple, m elts in the range of 2268kg-4536kg retained their refinem ent

com pletely for 24 hours, w hile m elts in the range of 91-227kg, refined w ith the sam e

agent, retained refinem ent for only 4 to 5 hours12). Therefore, each com pany should

do som e experim ent in order to find the proper holding tim e for their holding furnace

volum es and operating condition.

(4) H ow m uch to add

The level of P needed in a m elt to provide good refinem ent is quite sm all. It has

been observed that 10 ppm (0.001% ) can give adequate refinem ent, w hile 15ppm is

considered as a safe level. H ow ever, because only 5-15% of the added P rem ains in

the m elt, an addition of at least 200ppm (0.02% ) is generally required to reach the

desired retained m inim um . W hen P-C u of 8% P is used, adding 0.25-0.3% of the

m elt P-C u w ill release 0.02-0.024% P to the m elt, of w hich 0.001-0.0036% P

(10-37.5ppm ) w ill be retained, w hich is adequate for inoculation.

(5) W hat happens if too m uch P is added?

According to the literature, there is no obvious harm regarding to Si size from excess

P. H ow ever, excess AlP can be detrim ental in another w ay w hile A lP behaves as an

inclusion.

The first observation of m assive A lP inclusion the author has experienced w as during

m achining a batch of B 390 (17% Si) alloy cast w ith perm anent m old. W hat first

appeared as darkened areas flush on the m achined casting surface, expanded

overnight to stand w ell proud of the casting surface. U nlike typical oxide-dross, this

inclusion is very hard and has a ceram ic-looking structure (Fig. 4.6). Various

40


46/105

attem pts to identify the inclusion w ere m ade in the beginning, but not conclusive.

EPA analysis has show n the inclusion containing 9.2 20% P, about 30% O , and other

elem ents, except for A l.

Fig. 4.6 M icrostructure of P inclusion ( 200)

A review of the m elt handling and casting process of these parts w ere m ade to see

w hat had been changed. It w as noticed that 15% P-C u w as added to adjust C u

content of the alloy for the trouble batch castings. Inadvertently, P content w as

increased from 97ppm , at w hich there had been no such a problem , to 439ppm . B y

referring to sim ilar inclusion in the literature 13), the author considered that the

problem had originated from A lP and the sw elling of the darkened areas w as som e

kinds of reactions occurring betw een A lP and m oisture in the air or m achining lube

upon exposure during m achining, for instance,

A lP + H 2O A l2O 3+ P 2O 5+ H 2O (4 3)

Then, the P-C u added w as replaced by pure C u ingot for C u-content adjustm ent.

The defect disappeared im m ediately after the replacem ent. The problem has been

adm irably solved. From then, people w orking in the plant are m ore careful about P

41


47/105

am ount in A l-Si alloys. A rule w as m ade that never add m ore than 400ppm P to any

of the A i-Si alloys.

4.3.2.3 P inoculation on porosity

The strengthening effect of P inoculation on the m echanical properties of A l-Si, due

to the refining effects on prim ary Si, has been w ell established. B ut there has so far

no report on the influence of P inoculation to porosity form ation. C onsidering the

im proved condition of interdendritic feeding after the shape of prim ary Si m odified, P

inoculation should benefit porosity reduction. This effect has once been verified

w ith a high-C u (2.5 - 4% ), A l-Si eutectic alloy piston cast w ith perm anent m old. In

the beginning, porosity on the finished m achined surface of the land area of a piston,

w here there is a diam eter change, can be recognized w ith dye-check (Fig. 4.7).

(a) dye-check of the finished surface (b) m icrostructure of the red area ( 100)

Fig. 4.7 Porosity at the land area of a piston

Various tries, such as intensive local m old cooling, have been tried. The situation

becam e better but not conclusive. Finally a sm all am ount of P bearing salt (5-10% P,

10-15% K, the rest C l) w as added before pouring and the porosity problem has been

solved (not detectable w ith dye-check), though the real m echanism of the

42


48/105

im provem ent needs further confirm ation.

4.4 Casting Processing

C asting processing consists of m old designing and casting processing-param eter

specification. C om paring w ith the big progress and great efforts m ade w ith

com puter sim ulation in the past tw enty years, developm ent in casting-designing

rules is so unobvious, as if it has been neglected for m any years. G ood textbooks

containing som e useful casting designing rules, such as the one w ritten by W lodaw er

14), w ere published m any years ago. It is not the case that there has been no

progress in casting-designing rules developm ent, but it is an area left consciously or

unconsciously for the foundrym en to sum m arize them selves. N evertheless, for

objective or/and subjective reasons, there exists a big black in this area. This

section serves to fill som e part of the blank.

4.4.1 M old designing

W hile m aking a m old design for a gravity casting, at least tw o things have to be

taken into consideration. O ne is a desired m old filling to elim inating air and oxide-

film entrapm ents in m elt; the other is unidirectional solidification to avoid serious

shrinkage and porosity in castings. In recent years, m old-cooling technology to

im prove productivity, w hich has long been applied to pressurized die-casting, has

also been introduced to gravity perm anent casting. W ith this introduction, the

solidification order naturally decided by the casting structure has collapsed. The

intensive m old cooling can be a useful w ay to rem ove/reduce porosity, and it can

also cause porosity that m ay not occur if w ithout it.

A bad gate system causes not only the random oxide-film inclusion, but also the

random porosity. The oxide-film inclusions are created by the surface folding during

43


49/105

m old filling. The existence of oxide-film helps porosity form ation, because it usually

contains air, and itself is the good site for pore nucleation. A good gating system is

tolerant of w ide variations in foundry practice. Thus pouring w ill be under the

control of the gating system , not the caster. W hether a gating system is good or

not is decided by every part of the gating system , w hich usually are m ade of a

pouring basin, a sprue, a runner, and a gate/gates. The designing of these parts is

inter-related, because a proper area ratio of them , nam ely gate ratio, is required. Fig.

4.8 gives the im age of a gating system frequently applied to a cylindrical castings

cast w ith perm anent m old.

Ss

Sc

Sg

Fig. 4.8 A gate system w ith vertical runner and tw o vertical gates

4.4.1.1 The gate ratio

G ate ratio is the sectional - area ratio of sprue, runner and the gate. There are

unpressurized gating system and pressurized gating system . The form er is an

expanding system that slow s the m etal flow at each stage prior to entering the m old,

w hile the later is a contracting system that chokes the m etal flow at one place and

44


50/105

causes the gating system to fill back from the choking point. It is a com m on sense

to apply an unpressurized gate system to alum inum alloys. In literatures, there are

som e recom m endations on the gate ratio for the unpressurized gate system : S s: S r:

S g= 1:1.2 2.2:1 41), in w hich S sS rand S grepresent the sectional areas of the

sprue, runner and gate respectively. For a gating system w ith vertical runner and

gates as show n w ith Fig. 4.8, the com m on area of the sprue and the runner, S cin Fig.

4.8 is m ore im portant than the runner sectional area, S rbecause S ris not fully filled

until the last stage of the m old filling. W hile m aking a gating system design for an

alum inum casting, instead of trying to satisfy such a uncertain ratio, the gating

system should be m ade to slow dow n the m etal flow gradually, and to create as

m uch opportunity as possible for the m elt to becom e quiescent before entering the

cavity. For exam ple, even S g= 2 S sin Fig. 4.8, it is not a guarantee of a good gate

system , if the gate is a high narrow slot because m etal w ill then splash into the m old

as a jet, and surface turbulence w ill im pair the quality of the casting. W hen talking

about gating ratio, the fam ous foundry expert John C am pbell said designing a gate

system based on gate ratios is a m istake15). The author also does not suggest

paying m uch attention to the practical value of the gate ratio, but w ould like to

em phasize one thing that, for the gating system w ith a vertical runner and gate/gates,

the gate sectional area should be m ade at least the sam e as or even bigger than the

runner sectional area as show n in Fig. 4.8, so that m etal is slow ed dow n just before

entering the cavity.

4.4.1.2 The pouring basin

Pouring basin is the entrance through w hich m olten m etal is introduced into the

cavity. According to the authors experiences, follow ing aspects are to be

considered w hile designing a pouring basin. (a) The basin itself has to be filled up as

45


51/105

quickly as possible in order to reduce the pouring head. This can be realized by

properly chocking of the sprue bottom , such as applying a filter at the sprue base like

show ing in Fig.4.8. (b) M elt should be poured from the blind end of the basin so

that the fall of the stream is arrested, and bubbles and dross w ill have chance to float

up to the surface. (c) The exit of the basin should be larger than the entrance of the

sprue to avoid air aspiration at the basin/sprue connecting location. (d) The ideal

profile of a pouring basin is the one that w ill introduce the m elt to flow along the

sprue w all, rather than m ake it flow tow ards or collide onto the w all, w hich causes

collapse and entrapm ent of surface oxide film . Besides, a pouring basin w ith a

round bottom w ill cause the tendency of the m etal to rotate, form ing a vortex and

hence aspirating air. Fig. 4.9 show s tw o types of pouring basins used in our

production. The type show ing w ith (a) gave a higher rejection of random inclusions

and porosity than that given by the type show ing w ith (b), w hen all the other foundry

conditions w ere sam e.

(a) the old type (b) the optim ized type

Fig. 4.9 Tw o types of pouring basin

4.4.1.3 The sprue

46


52/105

Sprue is the dow n runner that introduces m elt to the runner and provides driving

force for m old filling in gravity casting. To design a sprue w ith right size is the key

point in gating system design. A n oversized sprue is, in particular, a liability because

it leads to oxide built-up and air entrapm ent. H ow then to design a sprue w ith the

right size?

(1) The height of the sprue

Theoretically, if a stream of liquid is allow ed to fall freely from a starting velocity of

zero, after falling a distance of h, it w ill reach its m axim um velocity given by

Bernoullis relation,

V m ax= (2gh)1/2 (4 4)

in w hich is the friction coefficient of the sprue-w all. In a gating system , the

m axim um velocity, V m axusually appears at the sprue exit. To keep lam inar flow , the

m axim um velocity should below a certain value theoretically decided by the W eber

num ber15).

W e= V m ax2 r/ (4 5)

W here is the density of the m etal, is the surface tension and r is the radius of

the surface curvature. Surface turbulence w ill occur if the W enum ber exceeds its

critical value. Jhone C am bell15)has proposed 20% loss of friction, i.e., = 0.8; and

a lim it range of 0.2-0.8 for W e to avoid surface turbulence. Taking the upper lim it of

W e, for a thickness of 10m m filled w ith pure alum inum liquid, the V m axto keep the

m etal free from surface turbulence as calculated by eq. 4-5 w ill be,

V m ax= (W e / r )1/2= (0.8*0.9/2500*0.005)1/2= 0.24 (m /s) = 24 cm /s

A nd the corresponding head lim it, hm axas calculated by eq. 4-4 w ill be,

hm ax= V m ax2/2g 2= 24 24/2 980 0.82= 0.46 (cm ) = 4.6m m

H ow ever, a sprue of 4.6m m high is never im practical. This calculation tells us that

47


53/105

surface turbulence is actually occurring at the bottom of all the practical sprue in use.

W hat w e can do w hile m aking a gate system designing is trying to m ake the head, h

as low as possible. This is particularly im portant w hen deciding the location of a

vertical gate. The low er end of the gate should not be 5m m higher than the cavity

bottom line, because there is no any brake, a filter for exam ple, to slow dow n the

m etal there afterw ards.

For gravity casting, a sprue should, at least, have a height not low er than the casting

height in order to m ake filling driven by gravity possible. In practice, the sprue

height, hs, for a gate system as show n in Fig. 4.8 is decided according to the

follow ing relation,

hs = hc + hr - hp (4 6)

In w hich hc, hr, and hp are the height of the casting, riser, and the pouring basin,

respectively. The riser height, hr is decided according to the riser am ount needed

after its horizontal dim ension decided based on the feeding-distance principle. The

height of a pouring basin, hp, should alw ays be taken into consideration together

w ith the sprue height, hs.

(2) The sectional area of the sprue _ the average flow rate of m old filling

The casting w eight filled per unit tim e is called average flow rate. W hile a casting is

m ade, the average flow rate of the casting can be easily obtained by sim ply dividing

the casting w eight, W casting, w ith the m easured filling tim e, t.

Q ave= W casting/ t (4 7)

For an unpressurized gate system as show n in Fig. 4.8, it is the exit area of the sprue,

S s, that decides the average flow rate, because this area is the narrow est area in the

gate system and it is supposed to be fully filled during pouring. This is w hy the flow

rate topic is discussed here together w ith the sprue designing. W hen m aking a

48


54/105

gate system designing, the target average flow rate can be estim ated by the

follow ing eq.,

Q ave= V m ax A Sexit (4 8)

In w hich is a coefficient reflect the effect of the filter, and friction of m old- w alls,

and w hatever other param eters that give barrier to the m etal. is the density of

the m etal, V m ax is the velocity of the m etal calculated by eq. 4-4, and A Sexitis the sprue

exit area.

A n extrem ely big flow rate is not good w hether it is due to a high velocity or a large

sprue exit area, because the form er causes severe surface turbulence and the later

offers difficulty for itself being fully filled, and hence accom panying w ith air aspiration

and m etal oxidization. O n the other hand, if the flow rate is too sm all, som e part of

the casting w ill solidify, and cold-lap defects, such as cold-shut, w ill occur. The

principle is that the filling tim e, decided by the flow rate, should be not longer than

the tim e needed to solidify the thinnest section of the casting, even if som e

solidification be allow ed during m old filling. H ow ever, it is w orthy to note there is

actually no low er lim it, i. e., no not shorter than for the m old filling tim e.

For a specific casting, an average flow rate that seem s reasonable according to

long-tim e experiences can be used in the first-tim e gate system designing, and can

alw ays be m odified in subsequent trials. The designed flow rate can be checked

w ith a stopw atch w hen the m old is first poured.

A ssum ing the average flow rate, Q ave has been know n from the long-tim e

experiences, how can w e

Research on Porosity Defects of Al-Si Alloy Castings

Documents