Paper 08-008(05).pdf, Page of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA A Coputer-Aided Syste for Melt Quality and Shrinkage Propensity Evaluation Based on the Solidification Process of Ductile Iron P. Larrañaga, J. M. Gutiérrez, A. Loizaga, J. Sertucha, R. Suárez Engineering and Foundry Dept., Azterlan Metallurgical Center, Durango (Bizkaia), Spain. Copyright 2008 Aerican Foundry Society ABSTRACT The present work develops a new ethod for pre-evaluating elt quality of ductile irons and for predicting their etallurgical behavior during the solidification step. Based on the characterization of the liquid-solid transition using the cooling curves and database techniques, this study provides a new effective approach to predict the nucleation ability of the etal and its shrinkage foration tendency. Both factors are considered as critical paraeters in order to guarantee the effective anufacturing of sound castings and increase the productivity. Two specific paraeters have been defined for this purpose: N (nodule count prediction factor) and k (shrinkage foration tendency factor). A theral analysis database has been set up by analyzing castings of an extended variety of ductile irons anufactured under fixed production conditions. This database contains ore than 600 records obtained fro experiental cooling curves, the results of etallographic studies (graphite orphology characterization), the cheical coposition of the alloy and X-ray inspections for evaluating the appearance of contraction defects in specific saples. It is found that necessary feeding requireents related to riser design and the contraction defects appearance are strongly correlated to the shape of the cooling curves and the corresponding N and k factors. Therefore, it is possible to predict the etallurgical behavior of alloys by analyzing their solidification before casting. The ai of the present work is to introduce the procedure for evaluating the N and k factors and to describe their effectiveness when copared to the inforation fro the database. INTRODUCTION Although any iproveents have been ade regarding ductile iron anufacturing processes, unsound castings are still produced under noral production conditions. It is known that different etallurgical behaviors, such as nodule count, shrinkage appearance and carbide foration, can be obtained using siilar cheical alloy copositions, inoculation procedures and sand olding processes. These experiental variations have been generally attributed to iportant differences concerning solidification processes of the liquid iron that originate as a consequence of any processing factors such as the etallic charge aterials, the elting and the Mg treatent ethodologies -3. Due to lack of knowledge as to how all those factors influence SG iron solidification processes, several papers have been published containing different ethods for this purpose. 2, 4-8 In this context, theral analysis concerning the cooling curves have been one of the ost used techniques for investigating the solidification processes of ductile irons. Although this technique has been used for cheical deterinations (carbon, silicon and carbon equivalent contents), 9-0 its principal advantage consists in the fact it provides a way to analyse the different steps of the liquid-solid transforation and to establish relationships between the cooling curve paraeters and the etallurgical characteristics of the alloy -4. According to this, an extensive nuber of technical works concerning the eutectic and eutectoid transforations in graphitic irons have been published and, as a result of the, several atheatical odels and data bases have been developed including final 2,4-5,5-2 prediction systes. Thus, theral analysis has been eployed for obtaining inforation regarding the inoculation efficacy 6,22-23, echanical properties and hardness predictions 7,2,24, graphite orphology 2,5,5,8, latent heat calculations 5,25-26, shrinkage tendency 6,8,,27-28 and the graphite nucleation potential during the solidification process 2,29-33. The resulting inforation can be stored into a data base which ay be used to control the real anufacturing process at foundries and avoid the appearance of etallurgical defects in cast parts. Research results show that variations in processing factors also influence the shape and characteristics of the cooling curves. This fact has served as an incentive to develop new studies and systes in order to extend the theral characterization to the subsequent steps of ductile iron anufacturing processes 34-36. However, such results should be considered as exclusive paraeters in each foundry, as they can only be safely used in well-controlled processes 37. This work provides a new solidification odel based on cooling curves for evaluating elt quality as a function of the graphite foration, nodule count and the contraction-expansion balance in iron saples analyzed in plant before old pouring. An additional novelty lies in the fact that both austenite and graphite foration are characterized and related to 547
Paper 08-008(05).pdf, Page 2 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA different cooling curves and to the etallurgical behavior of castings. Consequently, it has been necessary to develop new theral analysis tests in order to deterine carbon, silicon and active agnesiu dissolved in Mg-treated irons and copare these results with inforation obtained fro ethods norally used in foundries 38. In this way, a shrinkage foration prediction through cooling curves analysis can be considered essential to iniize engineers odifications based on their experience and trial and error tests 39. The resulting correlations will be discussed along this article. THEORY OF SOLIDIFICATION BASED ON COOLING CURVES As illustrated in Figure, the liquid-solid transforation of graphitic irons observed on cooling curves presents characteristic features. The solidification process occurs in three successive stages: While the cooling rate is considered to be constant until the liquidus teperature is reached, a slope change is observed when the first austenite crystals and/or graphite nodules are fored. The second stage corresponds to the bulk liquid-solid transforation, during which both austenite and graphite crystals for and progressively grow. Austenite can appear as etallic shells surrounding the graphite nodules and/or as free dendrites, this is the so-called eutectic plateau. During the final solidification step, grain boundaries are generated fro residual liquid into the solid network. Teperature (ºC) Teperature (ºC) 20 20 90 90 70 70 Liquid range 50 50 30 30 0 0 090 090 Te ax T liq st derivative curve Te in T sol 070 070 Solidification 2 nd derivative curve 050 050 00:00 00:00 00:7 00:7 00:34 00:34 00:5 00:5 0:08 0:08 0:25 0:25 0:42 0:42 0:59 02:6 02:33 02:50 Tie Tie (in:s) (in:s) Solid range 0 - - -2-2 -3-4 -4 Figure. Solidification curve and theral paraeters used in the present work. Figure shows the critical teperatures defined in order to characterize the solidification curves: liquidus teperature (T liq ), iniu eutectic teperature (Te in ), axiu eutectic teperature (Te ax ) and solidus teperature (T sol ). These paraeters are calculated in all cases using the first and second derivative curves. In addition, the recalescence Rc is given as Te ax Te in. HEAT BALANCE DURING PHASE TRANSFORMATION The evolution of the ass of austenite and of graphite during solidification will be evaluated following a siple procedure using the cooling curve recorded by standard theral analysis (TA). The overall heat flux out off a TA standard saple ay be written as a first approxiation as the su of the loss of energy of the aterial due to the decrease in teperature and the latent heat release due to austenite ( ) and graphite ( G ) precipitation: dtsaple d d G Saple Saple c p G Equation dt dt dt 548
Paper 08-008(05).pdf, Page 3 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA where Saple is the total heat flux extracted fro the saple, Saple c P dt Saple dt is the ter due to cooling of the saple, d d is the latent heat release due to austenite solidification and G G is latent heat release related to graphite dt dt precipitation. The range of teperatures where the liquid-solid transforation takes place in ductile iron is narrow enough in order to consider the specific heat (c p ) as a constant, and it will be assued equal for liquid and solid. In order to estiate the energy associated to solidification, the experiental cooling curves are copared to a reference cooling curve obtained assuing that the saple does not undergo any phase transforation (Figure 2). In this reference case, the overall heat flux ( Reference) out off the saple cup is written as follows: dtre ference Re ference Saple c p Equation 2 dt Teperature T i Experiental cooling curve Theoretical cooling curve (no transforation) t i Tie Figure 2. Coparison between an experiental and the reference (with no transforation) cooling curves. In Figure 2 is defined the teperature difference T i = (T Saplei -T Referencei ) between the easured saple teperature and the calculated reference teperature. The subscript i indicates that the data are discretized in tie, with index i referring to tie step i. The change in T i is associated to the instantaneous latent heat release only. Cobining equations and 2 gives: c T Equation 3 Saple p i G i G i where i and Gi are the instantaneous change in ass of austenite and graphite respectively. Integrating equation 3 thus relates the experiental cooling curves to the total ass of austenite and graphite depositing during solidification. On the other hand, the final values of austenite and graphite fored during solidification of the saple ( and G ) depend on the cheical coposition of the elt. The total carbon and silicon contents of the elt, respectively %C and %Si expressed in wt. %, can be obtained using theral analysis experients 9-0. Further, considering that the aount of carbon dissolved in austenite is conveniently evaluated as 2.03-0.20 %Si, the total aount of graphite fored at the end of solidification is then calculated as follows 37 : Saple Saple G %CTotal %C %C Total 2.03 0.20 %Si Equation 4 00 00 549
Paper 08-008(05).pdf, Page 4 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA in which Saple is the ass of the standard saple that can be obtained fro the tie of cooling of the etal through a defined range of teperatures. In our case, the 20-050ºC range has been selected because the solidification of the saple is guaranteed into it. The total aount of austenite can then be calculated as the difference between the total ass of the analyzed saple and that corresponding to the graphite precipitated ( = Saple G ). Because the nearly isotheral eutectic plateau is known to relate to solidification kinetics that is nearly linear in tie, it is proposed to write that such a siple evolution applies to austenite precipitation. The instantaneous discretized change in austenite ass is thus assued to be given as: o t i Equation 5 t t solidus liquidus where o is thus the aount of austenite that should precipitate during tie step i in order to obtain the total ass of austenite in the saple ( ) at the end of an ideal solidification process with linear solidification kinetics. Analysis of the experiental cooling curves shows however that there are tie steps where the actual heat flux available for the phase change (given by equation 3) is lower than the necessary one for crystallizing the quantity of austenite expected fro equation 5. In those cases, all the available latent heat release is assued to lead to austenite deposition. In all practicality, successive corrections are needed in the theoretical value of austenite for each of the tie periods when T i is lower than necessary. These corrections are successively added so that after and n tie steps one has: o o Saple t o c P T Equation 6 n o n i Saple i i t c P T i Equation 7 where t i is the length of tie on which the deficit of latent heat release is dispatched. After scanning of the whole cooling curve to deterine the successive values of i and n, the deficit in latent heat release is progressively redistributed in a unifor way as described by Figure 3. Heat flux G t Gi n n o t o Tie Figure 3. Heat flux evolution in tie during the solidification range (austenite and graphite foration areas). 550
Paper 08-008(05).pdf, Page 5 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA Fro this, the latent heat release possibly available for graphite precipitation is finally deduced according to equation 8. Saple c p G T i n G Equation 8 i CONTRACTION-EXPANSION BALANCE In the solidification process an iportant change of volue due to the variation of density between the liquid and the solid phases can occur. This phenoenon is the result of graphite (expansion) and austenite (contraction) crystallization fro the liquid and their subsequent growth fro the elt. Thus, as the aount of austenite and graphite fored at each period of tie ( n and Gi ) can be deterined using the experiental cooling curves, it is possible to obtain a contraction-expansion balance. The change of volue between two successive tie steps due to the foration of austenite and graphite can be calculated as follows: n Gi Liquidi Vi Vi Vi Equation 9 G Liquid where, G and Liquid are the densities of austenite, graphite and liquid, respectively. Also, a ass balance gives: Equation 0 Liquidi n Gi Equation 9 ay thus be written as: V i n Liquid Gi G Liquid Equation Taking into account that the density of austenite is higher than that of liquid and that both are uch higher than the density of graphite, the resulting variation of volue ay be either negative or positive. Accordingly, the calculation of the volue changes due to the graphite and austenite fored in different steps of the solidification leads to deterine the contractionexpansion balance at each tie. NODULE COUNT The nodule count (expressed as N) plays an iportant role in the echanical properties of ductile irons. Therefore, the estiation of this paraeter using a standard saple is considered as very relevant in order to obtain a good indicator for evaluating the elt quality in this type of irons. In the present work, an iportant dependence between several additional paraeters obtained directly fro the cooling curve and the final nodule count has been found. The evaluation of these paraeters can be consequently used for estiating the nucleation ability of the iron during the solidification range. Bearing this in ind, a wide range of nodule count values has been obtained via etallographic analysis fro all experiental tests using inoculated and non inoculated saples. Then, they have been related to the ain characteristics of the experiental cooling curves expressed with the paraeters evaluated fro the first and second derivative curves (Figure ). In addition, a new paraeter (G R ) has been defined according to the equation 2 in order to estiate the graphite foration rate during the solidification process. G R i Gi Equation 2 t i Under those conditions, a atheatical expression to calculate the nodule count has been developed using four experiental paraeters according to equation 3. dt N A Equation 3 Tein A2 Rc A3 GR A4 A5 dt Solidus 55
Paper 08-008(05).pdf, Page 6 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA In this expression, the dt ter corresponds to the axiu cooling rate at the very end of the eutectic dt Solidus transforation (near to the teperature T sol ). MICROSHRINKAGE TENDENCY Although they are not the unique ain causes, etallurgical behavior during solidification and the evolution of the contraction-expansion balance are critical aspects as to shrinkage defects appearance in ductile iron castings. Therefore, a useful indicator to easure the contraction tendency of the elt ust be included into the developed coputer-aided syste for effective elt quality evaluations. According to the odel calculated for the contraction-expansion balance and the foration of graphite and austenite at different oents of solidification, the experiental cooling curves show an initial contraction period close to the liquidus teperature (Figure 4). This sall area is a consequence of the foration of priary austenite. When the transforation goes on, an iportant expansion period occurs due to graphite precipitation and its subsequent growth. Finally, a contraction range is found at the end of the solidification process. In this last area, graphite expansion decreases and a final net contraction occurs. Microshrinkages are norally fored in this period of tie due to the ipossibility of liquid feeding through the existing solid network. The obtained contraction-expansion balance and especially the final contraction period strongly depend on the shape of the experiental cooling curve. In order to evaluate this behavior and consequently the tendency of the elt to for icroshrinkages, a new paraeter has been defined as: t Expansion k Equation 4 t t Expansion Final contraction where t Expansion and t Final-contraction are the total periods of tie for expansion and for final contraction during the solidification process. In this way, the tendency to icroshrinkages foration in a specific ductile iron can be turned into a nuber only fro of etallurgical considerations. Thus, the risk of defect foration decreases when the final contraction periods becoe saller and, therefore, the resulting value of the k factor approaches. Teperature (ºC) 20 90 70 50 30 0 090 070 050 Liquid range t Expansion Contraction-expansion balance Cooling curve Expansion Initial area Final contraction area contraction area Solidification t Final-contraction 00:00 00:25 00:50 0:5 0:40 02:05 02:30 02:55 03:20 03:45 Tie (in:s) Solid range 7 6 5 4 3 2 0 - -2-3 -4-5 Relative variation of volue Figure 4. Contraction-expansion balance obtained fro an experiental cooling curve. 552
Paper 08-008(05).pdf, Page 7 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA EXPERIMENTAL A database has been established including 620 records obtained fro different fields: the eutectic transforation paraeters of cooling curves, the cheical coposition of the alloys, the nodule count estiated using an iage analysis software and the tendency to contraction defects appearance calculated fro the solidification patterns. An extensive nuber of solidification experients were perfored on different ductile iron qualities ranging fro EN GJS 400-5U to EN GJS 700-2U. All etal saples used in this work were picked up fro the basin area close to the rod stopper in vertically parted autoatic olding lines equipped with autoatic pouring furnaces. Saples were also collected for later analysis by eans of cheical and spectroscopic techniques in all experiental tests in order to copare the with the corresponding theral analysis results. Table shows the obtained ranges for the ost iportant eleents analyzed by cheical techniques and spectroscopy. Table. Cheical Coposition Ranges for Melts Used in the Present Work. C Si Mn P S Cu Mg Ti 3.50-3.85 2.25-2.80 0.2-0.25 0.020-0.045 0.00-0.006 0.02-.05 0.025-0.045 0.008-0.026 All liquid etal saples were elted in ediu frequency induction furnaces (250 Hz), 7-0 t capacity and 4000-5000 kw power. Two different etallic charges were used in elting area: charge was coposed of 40% autootive steel scrap, 0% pig iron and 50% return scrap; charge 2 contained 50% pig iron, 20% autootive steel scrap and 30% return scrap. A coercial recarburizer product and a FeSi alloy were added together with the etallic charges introduced into the furnaces for achieving the desired carbon and silicon contents. After elting, 0.0% of SiC conditioning product was added. Then, carbon and silicon contents were checked by spectroscopic analysis and adjusted to the specified values. Finally, the elt teperature was increased up to 500ºC (2732F) and its surface skied. The liquid etal obtained fro furnaces was treated in.5-2.0 t capacity ladles with a FeSiMg alloy (5-6% of agnesiu content) by eans of the tundish cover ethod. All agnesiu treatents were carried out in the 470-490ºC (2678-274F) range. The resulting treated batches were then transferred to the corresponding pressurized pouring devices (8-0 t of capacity). Theral analysis tests were recorded in the range of 20-050ºC (220-922F) in order to characterize the liquid-solid transition in all saples and to deterine the critical paraeters using a specific theral analysis software developed in Azterlan (Figure 5). A correct evaluation of theral paraeters requires that the following conditions are fulfilled: As the axiu capacity of the pouring cups is 340-350 g, a iniu weight of 320 g ust be poured into cups for representative results. The weight of saples should be as constant as possible. The coponents fro the sapling cups should not react with the elt. Initial teperature of liquid saples has to be high enough for adequate recording of solidification curves. A correct inoculation of the elt ust be guaranteed. Stable conditions at the sapling cups are needed during all data collection process. Metal saple Data collection Coputer Data Logger (T vs tie) Cooling Curves Measuring Cup Theral Analysis software RESULTS Figure 5. Schee of the syste used for theral characterizations. 553
Paper 08-008(05).pdf, Page 8 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA According to these requireents, a considerable aount of etal saples were poured into standard coercial cups with a K-type therocouple located at the center of the cup. Two different cups were used for coplete theral characterization: plain cups and cups with telluriu and sulfur. In addition to this, inoculated and non inoculated tests were conducted using the plain cups in order to obtain different nodule counts in the elt. For this purpose, approxiately 0.20% of a coercial inoculant product (cheical coposition: Si = 75-78%, Ca =.8-2.0%, Zr =.3-.4%, Al =.0-.2% and 0.2-0.5 of grain size) was introduced in the corresponding cups before pouring. After solidification, all etal saples contained in cups were cut and prepared for etallographic inspection. Optical icroscopy observations were carried out in the central areas of the saples (where K-type therocouples are located) in order to estiate nodule counts and then copare the with those predicted using the new coputer-aided syste based on cooling curves. All the etallographic nodule count values were obtained using a coercial iage analysis software. Apart fro theral studies, experiental tests for evaluating the contraction tendency of alloys were perfored using specific cross shaped castings. Cheical bonded sand olds (Figure 6a) were anufactured for this purpose and 0.20% of the coercial inoculant entioned above was added into the before etal pouring. Resulting cross shaped castings (Figure 6b) had a total volue of 79 c 3 and were subitted to x-ray inspections in order to detect the shrinkage defects fored into the. The area affected by shrinkage was evaluated fro two perpendicular x- ray iages analysis, giving values S and S 2 in square centieters. Fro these, the volue affected by shrinkage (V s ) was calculated using equation 5 and then copared to the corresponding k factors calculated fro the experiental cooling curves. V (c ) s 3 S S 2 2 S S 2 Equation 5 (a) (b) Figure 6. (a) Cheical bonded old used for shrinkage easuring tests; (b) obtained cross shaped casting. RESULTS AND DISCUSSION NODULE COUNT PREDICTION A nodule count prediction odel based on equation 3 has been developed setting up a database which contains all theral paraeters obtained fro the experiental cooling curves and the results of the etallographic inspections. In each case, etallographic nodule count estiations were carried out as an average value of three different zones analyzed in the central area (where the therocouple is located). Carbon, silicon and agnesiu contents have been considered as additional fields into the database in order to guarantee the correct foration of graphite nodules. Two different possibilities have been established for deterining the cheical coposition: the results obtained fro spectroetric analysis and its calculation fro cooling curves where the etal saples have been treated with telluriu and sulfur. In both cases, siilar nodule count predictions are obtained. A free agnesiu content of 0.023% has been adopted as the critical liit for obtaining the predicted nodule count values. The effective control of this active eleent through theral analysis becoes a very useful tool to iniize the treatent costs and to reduce the influence of high agnesiu contents on the shrinkage foration process. If spectroetric Mg content was used for nodule count prediction, a critical value of 0.025% has been adopted for guaranteeing spheroidal graphite foration. 554
Paper 08-008(05).pdf, Page 9 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA Figure 7 shows a coparative analysis between the nodule count values obtained using equation 3 and the results of etallographic inspections ade on all standard saples. A good agreeent has been found in all the nodule count range achieved. 500 Predicted nodule -2 450 400 350 300 250 200 50 00 50 R 2 = 0.922 Inoculated saples Non inoculated saples 0 0 50 00 50 200 250 300 350 400 450 500 Metallographic nodule -2 Figure 7. Correlation between predicted N values and etallographic nodule counts for standard TA saples. Cooling curves recorded fro inoculated saples give rise to highest nodule count values (200-450 nodules -2 ). The lower range was generally achieved in the case of non inoculated etal saples. In this last case, the inoculation effect does not exist and the original nucleation potential of the elts becoes a critical aspect anaging the graphite crystallization in the liquid-solid transition. In order to confir the validity of the developed predictive odel, a coparative study has been ade with respect to an extensive aount of real castings, analyzing their nodule count via etallographic inspections. Taking into account that the cooling rate (expressed as the geoetric odulus) widely influences the obtained nodule counts, specific areas with siilar odulus to the standard saples (0.60-0.65 c) have been selected in all castings. The cooling curves were siultaneously recorded fro poured standard cups in order to obtain the N values in each case. The etal saples used for these theral characterizations were picked up fro the pouring furnaces during each casting anufacturing process. The results obtained fro this study have been plotted in Figure 8. In this figure, data were nubered in the way of a gradual increase in the etallographic nodule count values for a better coprehension. A good correlation can be found between the predicted N values and those obtained fro castings inspections. An average error of 3.5% has been obtained for predicted N factor, being the axiu difference of 29 nodules -2 for all cases. Although all the castings have been anufactured using inoculated irons, the nucleation capabilities of the etal saples are not siilar and nodule counts within the range of 200-30 nodules -2 are obtained (Figure 8). This behavior is an effect of different qualities of the elt, which odify the solidification process and the resulting cooling curve shape. As a consequence of this, different N values are finally obtained which agree satisfactorily with the experiental nodule counts. 555
Paper 08-008(05).pdf, Page 0 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA 320 300 Nodule count in real castings Predicted N factor Nodule -2 280 260 240 220 200 80 60 23 83 246 Test No. Figure 8. Coparative analysis between the predicted N values and etallographic nodule count values in real castings. This prediction odel only applies to the standard saples used for theral analysis and those areas with siilar geoetric oduli in castings. However, it can be used as a useful standard tool in order to quantify the nucleation potential of the elt before pouring, so it can be possible to correct the processing paraeters if necessary. Additionally, the prediction of the N factor can be used as an alternative ethod to etallographic inspections for guaranteeing the correct nodule distribution in anufacturing processes. Further developents are necessary to predict the N factor in different sections of castings taking into account their geoetric oduli. This last paraeter ust be included as a new field into the database and added in expression 3. These studies will be the subject of future works. MICROSHRINKAGE PROPENSITY ESTIMATION Shrinkage foration tendency has been evaluated using the k factor directly obtained fro the experiental cooling curves. In order to correlate it to the incidence of contraction defects, a new test casting has been designed including a central area prone to this kind of defects. This critical area has a higher geoetric odulus ( c) than those of surrounding zones and originates a cross shaped distribution (Figure 9). Thus, an isolated ushy area is created during the last steps of the solidification where shrinkage can occur. Siulation studies have been carried out in order to confir this fact and to obtain the best geoetric characteristics for the cross castings. In addition, a standard riser (3.5 c of diaeter, 2.5 c of height and 0.65 c of odulus) has been included in the top of the castings to avoid the foration of priary shrinkage defects fored in the early steps of solidification. This feeding syste ust also proote the appearance of defects only in the central areas (Figure 0). A total of 20 cross castings (40 inoculated and 70 non inoculated) have been poured in all experiental tests used for setting up the developed database. The obtained cross castings were classified, the defect volues easured and assigned to the corresponding cooling curves in order to establish a final correlation. Figure shows the results of this study. The highest k factor values are related to the sallest shrinkage volues (V s ) in the cross castings. Although a progressive reduction of this paraeter brings about an iportant increase of V s, this tendency is not linear and a axiu volue of.5 c 3 is achieved when k factor is lower than 0.50. This behavior can be considered as a consequence of the axiu contraction capability of ductile irons poured into rigid olds (in this case up to 2% in volue). 556
Paper 08-008(05).pdf, Page of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA (a) (b) Figure 9. X ray iages for shrinkage evaluation: (a) front view and (b) perpendicular view. Figure 0. Geoetric characteristics of cross castings and shrinkage prediction based on siulation studies. Although the curves recorded fro non inoculated saples usually lead to the lowest k values (0.45-0.75), a few nuber of these saples with higher eutectic teperatures and elt qualities also give rise to higher k factors (0.80-0.82). In general, inoculation process increases the k values which are then in the range 0.75-0.92. When the solidification of saples with high nucleation potential occurs, etallic contraction due to austenite crystallization can be copensated by the foration of graphite nodules. Thus, saller contraction periods are obtained in the final steps of the expansion-contraction balance and higher k values are expected. In those cases (that relate to inoculated and several non inoculated saples) the incidence of icroshrinkages becoes lower as shown in Figure..6.4.2 Vs (c 3 ).0 0.8 0.6 0.4 0.2 Inoculated saples Non inoculated saples 0.0 0.40 0.50 0.60 0.70 0.80 0.90.00 k factor Figure. Correlation between the volue of icroshrinkage and the k factor. Metal saples where the nucleation potential is not large show a different solidification process. Under these conditions a decrease of the Te in and an increase of Rc is detected. Coparatively, the crystallization of nodules starts late and is not as favored as in the saples with high nucleation ability. Thus, the resulting austenite-graphite foration balance is not copensated and an iportant contraction period is found in the final step of the solidification. In this period, dendritic coherency point has been widely exceeded, so the feeding systes are not useful and icroshrinkages occur. In spite of the tendency shown in Figure, different icroshrinkage volues have been obtained in cross castings using saples with siilar k values. This fact has been assigned to slight differences in the pouring teperature during the cross casting anufacture. In order to evaluate the incidence of the icroshrinkage distribution for each k range, all experiental 557
Paper 08-008(05).pdf, Page 2 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA data have been plotted in Figure 2, expressed as percentage. The analysis of these distributions confirs the correlation observed between the k factor and the size of contraction defects. Shrinkage distribution (%) 00 80 60 40 20 Volue of defect (c 3 ) >.40.20-.39.00-.9 0.80-.00 0.60-0.79 0.40-0.59 0.20-0.39 0.00-0.9 0 0.9-0.95 0.86-0.90 0.8-0.85 0.76-0.80 0.7-0.75 0.66-0.70 0.6-0.65 0.56-0.60 0.5-0.55 <0.50 k factor Figure 2. Distribution of icroshrinkage size for different k values. Several tests using real castings have been perfored for checking the k factor utility in industrial conditions. The anufacturing process of a specific reference designed for the autootive industry was onitored using theral analysis and deterining the Te in, N and k paraeters fro the obtained cooling curves. The resulting castings were cut to detect the presence of icroshrinkages so as to correlate the to the previous paraeters. In order to obtain different elt qualities, green sand olds were poured using inoculated etal saples with siilar cheical coposition (%C = 3.75-3.80, %Si = 2.40-2.45 and %Mg = 0.030-0.035) and different solidification processes [alloy A: N = 266, k = 0.85, Te in = 48ºC (2098F) and alloy B: N = 03, k = 0.7, Te in = 32ºC (2070F)]. Sound castings have only been obtained when alloy A was used. All castings anufactured using Alloy B showed contraction defects as icroporosities in the last solidification areas (Figure 3). This results are in good agreeent with the correlation found between the k factor and the contraction capability of the elt. Although a correlation between the k factor and the appearance of icroshrinkage in specific real castings has been established up to now, a statistical study for the industrial use of this paraeter will be ade in future works. Alloy A Alloy B Figure 3. Influence of different elt qualities on the contraction defect appearance. 558
Paper 08-008(05).pdf, Page 3 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA MELT QUALITY ESTIMATION The results of the present work indicate that both N and k paraeters can be considered as useful indicators for elt quality evaluation in ductile iron anufacturing process. Although an increase of the k factor is often linked to higher predicted nodule counts, the evolution of the k paraeter ust be only explained on the basis of the contraction-expansion balance in the final step of the solidification. Therefore, it is possible to analyze etal saples with a high nucleation potential and k factors lower than expected. This fact is norally due to the existence of an insufficient graphite expansion at the end of solidification even if the nucleation ability in the liquid state becoes large. Miniu eutectic teperatures also provide inforation on the solidification patterns and the nucleation potential. Though this paraeter is clearly related to the predicted N values, it is not sufficient for quantifying the influence of the subsequent steps of the solidification on the etallurgical behavior of the elt. N and k paraeters have been designed for this purpose, showing an appropriate evolution when copared to the experiental behaviors of real castings. The possibility of gaining adequate knowledge on the different solidification paths in a short tie before the pouring process is considered as an iportant developent in order to iniize the appearance of defects in castings. Therefore, elt quality evaluation can be used as a crucial tool to correct or validate in real tie the processing paraeters. Significant differences aong several foundries have been found when adequate elt quality paraeters are defined. This is a consequence of different behaviors of etal solidification for each anufacturing process and their own characteristics (pattern plates design, elting ethodologies, Mg treatent paraeters, inoculation process, olding conditions, etc). These specific factors odify their influence in final etallurgical properties. So it is necessary to carry out a previous study in order to define the ore convenient critical values of N and k paraeters for a certain foundry. CONCLUSIONS According to previous works published in the literature, it has been found that the cheical characteristics of the pouring etal are not the only cause for the different etallurgical behaviors shown in castings. The new syste developed in the present work becoes an effective control tool for anufacturing processes in order to guarantee the correct graphite orphology in the solid state and to iniize the shrinkage appearance. Thus, a new solidification odel based on experiental cooling curves has been developed for this purpose focusing on two different paraeters: predicted nodule count (N) and contraction tendency paraeter (k). An extensive nuber of experiental tests have been ade to calculate these paraeters and to relate the to the results of etallographic inspections and to the shrinkage incidence in cross shaped castings. On the basis of this solidification odel, a good agreeent has been found between the predicted and experiental nodule count values in inoculated and non inoculated standard saples. A further investigation perfored with different inoculated quality irons confirs the validity of the obtained correlations. On the other hand, decrease in contraction tendency factors has been related to larger shrinkage volues in poured cross castings. These results have also been copared and then confired by analyzing the contraction defects detected in several real castings anufactured using the sae pouring etal in each case. A relationship between both N and k paraeters and the nucleation potential of the liquid etal has been found. Selected paraeters becoe useful for a quick evaluation of the elt quality in industrial conditions before the pouring step. As a consequence of this, it is possible to odify and anage the processing paraeters in order to optiize the anufacturing processes. The obtained coputer-aided cooling curve analysis can be considered as an inexpensive, siple and fast procedure that finds any applications in the etallurgical and foundry industries. In this case, an iportant issue is the possibility of cobining different studies coing fro the elt quality evaluation, the feeding requireents in castings and the resulting echanical properties. ACKNOWLEDGMENTS This paper is based on work supported by the Industry Departent of the Spanish Governent (PROFIT FIT-020600-2005- 6). The authors would like to thank Betsaide, S.A.L., Fytasa, S.A., Garbi, S.A., TS Fundiciones, S.A., Furesa, S. Coop. and Fuchosa, S.A. foundries for all the collaborating efforts ade in order to obtain the etallic saples analyzed in the present work. Especial thanks go to A. Bergara and J. Lacaze for their help in the atheatical calculations. 559
Paper 08-008(05).pdf, Page 4 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA REFERENCES. Li, X., Li, Y., Evaluation of Melt Quality and Graphite Degradation Prediction in Heavy Section Ductile Iron, Metallurgical and Materials Transactions, vol 36A, pp 2455-2460 (2005). 2. Frost, J. M., Stefanescu, D. M., Melt Quality Assessent of SG Iron Through Coputer-Aided Cooling Curve Analysis, AFS Transactions, vol 00, pp 89-99 (992). 3. Sertucha, J., Suárez, R., Legazpi, J., Gacetabeitia, P., Influence of oulding conditions and ould characteristics on the contraction defects appearance in ductile iron castings, Revista de Metalurgia. Madrid, vol 43, pp 88-95 (2007). 4. Li, Y. X., Wang, Q., Intelligent evaluation of Melt Iron Quality by Pattern Recognition of Theral Analysis Cooling Curves, Journal of Materials Processing Technology, vol 6, pp 430-434 (2005). 5. Li, Y. X., Xu, X. R., Wang, Q., Liu, B. C., Evaluating Melt Iron Features by Coputer Aided Recognition of Theral Analysis Cooling Curves, International Journal of Cast Metal Research, vol 6, pp 4-45 (2003). 6. Skaland, T., Ductile Iron Shrinkage Control through Graphite Nucleation and Growth, International Journal of Cast Metal Research, vol 6, pp -6 (2003). 7. Rivera, G. L., Boeri, R. E., Sikora, J. A., Research Advances in Ductile Iron Solidification, AFS Transactions, vol, paper 03-59 (2003). 8. Bradley, F. J., Fung, C. A., Theral Analysis for Shrinkage Prediction in Coercial Ductile Iron Castings, Canadian Metallurgical Quarterly, vol 30, pp 25-260 (99). 9. Heine, R. W., Austenite Liquidus, Carbide Eutectic and Undercooling in Process Control of Ductile Base Iron, AFS Transactions, vol 03, pp 99-206 (995). 0. Heine, R. W., The Fe-C-Si Solidification Diagra for Cast Irons, AFS Transactions, vol 94, pp 39-402 (986).. Labrecque, C., Gagné, M., Interpretation of Cooling Curves of Casting Irons: A Literature Review, AFS Transactions, vol 06, pp 83-90 (998). 2. Sparkan, D., Understanding Theral Analysis of Iron, AFS Transactions, vol 02, pp 229-233 (994). 3. Udroiu, A., The Use of Theral Analysis for Process Control of Ductile Iron, Proceedings of the Seinariu NovaCast, Caposapiero, Italy (2002). 4. Sillén, R., ATAS A Practical Guide for Ductile Iron, NovaCast, internal work. 5. Chen, I-G., Stefanescu, D. M., Coputer-Aided Differential Theral Analysis of Spheroidal and Copacted Graphite Cast Irons, AFS Transactions, vol 92, pp 947-964 (984). 6. Sertucha, J., Suárez, R., Izaga, J., Hurtado, L. A., Legazpi, J., Prediction of Solid State Structure Based on Eutectic and Eutectoid Transforation Paraeters in Spheroidal Graphite Irons, International Journal of Cast Metal Research, vol 9, pp 35-322 (2006). 7. Chang, S., Shangguan, D., Stefanescu, D. M., Prediction of Microstructural Evolution in SG Cast Iron fro Solidification to Roo Teperature, AFS Transactions, vol 99, pp 53-54 (99). 8. Barlow, J. O., Stefanescu, D. M., Coputer-Aided Cooling Curve Analysis Revisited, AFS Transactions, vol 05, pp 349-354 (997). 9. Maijer, D., Cockcroft, S. L., Patt, W., Matheatical Modeling of Microstructural Developent in Hypoeutectic Cast Iron, Metallurgical and Materials Transactions, vol 30A, pp 247-258 (999). 20. Guo, X., Stefanescu, D. M., Solid Phase Transforation in Ductile Iron. A Benchark for Coputational Siulation of Microstructure, vol 05, AFS Transactions, pp 533-543 (997). 2. Ribeiro, C., A., S., Santos, D., Baugart, W., Vilela, F., Henke, C., New Approach to the Use of Theral Analysis to Predict Microstructure and Mechanical Properties of As-Cast Ductile Iron, International Journal of Cast Metal Research, vol 6, pp 47-52 (2003). 22. Strong, G., R., Theral Analysis as a Ductile Iron Molten Metal Processing Evaluation Tool, AFS Transactions, vol 9, pp 5-55 (983). 23. Godsell, B. C., Preconditioning of Ductile Iron, AFS Transactions, vol 78, pp 273-276 (970). 24. Louvo, A., Pellikka, E. Alhainen, Eklund, P., Criterion Functions Based on Alloying and Cooling Rate for Siulating the Microstructure and Mechanical Properties of SG Iron Casting, AFS Transactions, vol 99, pp 237-244 (99). 25. Tinoco, J., Delvasto, P., Quintero, O., Fredriksson, H., Theral Analysis of Nodular and Laellar Eutectic Cast Iron under Different Cooling Rates, International Journal of Cast Metal Research, vol 6, pp 53-58 (2003). 26. Ekpoo, U., Heine, R. W., Theral Analysis by Differential Heat Analysis (DHA) of Cast Iron, AFS Transactions, vol 89, pp 27-38 (98). 27. Ohnaka, I., Iwane, J., Yasuda, H. Zhu, J., Prediction of Porosity Defect in Spheroidal Graphite Iron Castings, International Journal of Cast Metal Research, vol 6, pp 293-299 (2003). 28. Sillén, R., Production of Ductile Iron Castings in Green Sand Moulds without Feeders, Proceedings of 67 th World Foundry Congress, Harrogate, UK, paper 27 (2006). 29. Larrañaga, P., Sertucha, J., Suárez, R., Analysis of the Solidification Process of Ductile Iron, Revista de Metalurgia. Madrid, vol 42, pp 244-255 (2006). 560
Paper 08-008(05).pdf, Page 5 of 5 AFS Transactions 2008 Aerican Foundry Society, Schauburg, IL USA 30. Onsoien, M., I., Grong, O, Skaland, T., Jorgensen, K., Mechanis of Graphite Foration in Ductile Iron Containing Rare Earth Metals, Materials Science and Technology, vol 5, pp 253-259 (999). 3. Fras, E., Wiencek, K., Górny, M. and López, H. F., Nodule Count in Ductile Iron: Theoretical Model Based on Weibull Statistics, International Journal of Cast Metal Research, vol 8, pp 56-62 (2005). 32. Zhao, H., Liu, B., Siulation of Microstructure Foration of Spheroidal Graphite Iron Casting, International Journal of Cast Metal Research, vol 6, pp 28-286 (2003). 33. Dioszegi, A., Svensson, I. L., Inverse Kinetic Analysis Method to Study Eutectic Growth, International Journal of Cast Metal Research, vol 8, pp 4-45 (2005). 34. Eadi, D., Whiting, L. V., Nafisi, S., Ghoashchi, R., Applications of Theral Analysis in Quality Control of Solidification Processes, Journal of Theral Analysis and Calorietry, vol 8, pp 235-24 (2005). 35. Li, Y. X., Hu, X., Xu, X. R., Pattern Recognition of Theral Analysis Cooling Curves and Quality Evaluation of Melt Cast Iron, Journal of Materials Science & Technology, vol 7, pp 73-74 (200). 36. Stefanescu, D. M., Qiu, H. Q., Chen, C. H. Effects of Selected Metal and Mold Variables on Dispersed Shrinkage in SG Cast Iron, AFS Transactions, vol 57, pp 89-97 (995). 37. Suárez, R., Calidad etalúrgica del hierro. Análisis térico, Proceedings of st Syposiu: Azterlan and the Foundry Sector, Bilbao, Spain (2002). 38. Gutiérrez, J. M., Therolan, The Metallurgical Prediction Tool, st ed., Azterlan, Durango (2005). 39. Sabau, A. S., Prediction of Alloy Shrinkage Factors for the Investent Casting Process, AFS Transactions, vol 4, paper 06-004 (2006). 56