Abstract
As follow-up to a similar comparison of recarburizing materials reported on in 1999 AFS Transactions, a second, more extensive trial has been completed at a commercial ductile iron foundry. The results from 58 production days have been accumulated and are reported on in the paper, including carbon recovery, complete metallographic evaluation, and effects on mechanical properties. Over 1200 ductile iron heats were accomplished during this time frame, and the results mirror those reported on in the earlier paper. Once again there was no advantage seen in the use of a crystalline recarburizer, as compared to the use of a low sulfur, non-crystalline petroleum coke.

Introduction
When reviewing the literature on the subject of recarburizers to be used for ductile iron production, one is hard pressed to find any recommendation except that to use crystalline graphite, such as crushed electrode scrap. This is very logical, since that was the only type of material totally suitable for quite some time. However, about 12 years ago, a low sulfur, non-crystalline petroleum coke became available and it had an immediate impact on the marketplace, primarily due to the favorable economics associated with this new product. Some controversy evolved as to the ability of this material to produce high quality grades of ductile iron, in spite of the fact that dozens of foundries had been using it successfully for a period of years. With the cooperation of a high production ductile iron foundry, data from a five-day production test was reported on in a previous paper published in the 1999 AFS Transactions. The conclusion of that paper was that equivalent, or even slightly superior results were produced when the foundry was utilizing the non-crystalline product, as compared to irons produced when using a crystalline graphite product. That conclusion was based almost entirely on metallographic evaluations. Several months later, when another foundry became interested in the subject and wanted to complete a similar, but considerably longer trial, to include carbon recovery and mechanical properties comparisons, the wheels were set in motion to begin the trial and start collecting data, all of which is summarized in this paper.

Also included in this most recent report is similar data generated with the use of a partially graphitized, refined petroleum coke, which will be referred to hereafter as "PG". Data was gathered from 58 production days, which included 1275 ductile iron treatments. The data is divided into four chronological segments; initially, an 8 day production run using "PG", followed by 16 days with graphitic material, then 18 days using non-graphitic recarburizer, and followed lastly by a second 16 days use of graphitic material. Carbon recovery, microstructural results, effects on mechanical properties and base chill wedges will all be reported on.

Discussion
The foundry in which the testing was completed utilizes as primary melters, two 5.5 ton medium frequency coreless induction furnaces. The base charge typically contains about 50% local return scrap, 43% steel scrap, 5% low phosphorous pig iron, with the balance being recarburizer plus small amounts of 90% SiC grain and 75% ferrosilicon for trim purposes. The base metal is duplexed through a 25 ton vertical channel furnace. Nominal chemistry at this point is 3.8% carbon, 1.6% silicon, <0.30% manganese and <0.10% copper. Base sulfur levels will be discussed later in this report. Nodulization is in a 1500 lb. Tundish ladle, with a 1.6% addition of a 6% magnesium, 0.3% cerium alloy. Post-inoculation is accomplished by using 0.6% calcium bearing 75% ferrosilicon, plus a small amount of barium containing alloy, added to the metal stream as the treated ductile iron is transferred to a 3500 lb. unheated, bottom stopper auto-pour that supplies metal to the single vertically parted molding machine in the plant. In spite of the fact that the castings produced here are quite small, varying from 0.4 – 15 lb. in weight, there is no mold or stream inoculation employed. Final silicon and magnesium levels vary between 2.65 – 1.80%, and 0.038 – 0.044%, respectively.

Carbon Recovery
With the medium frequency furnaces as melters, a chemistry sample is taken when the metal gets to temperature and prior to that entire heat being tapped and laundered directly to the holding furnace. Because the foundry keeps accurate records of the amounts of metallics in each furnace charge, it is thus very convenient to calculate carbon recovery for each 11,000 lb. furnace charge, which in fact was done for the entire 58 days of data collection. As can be seen in Table 1, the recovery numbers are unusually high, probably due to some error in the assumptions made about the carbon level of the various charge materials; however, in any case they are relative, since the same assumptions were used throughout the entire test period, and irregardless of the recarburizer in use at the time.

Base Chill Wedges
This foundry does not normally pour chill wedges, but since some cores were available, it was decided to pour a few random samples on different days for informational purposes. The wedge dimensions are 1-3/8" high, by ½" wide at the base. Results are shown in Figure 1. 

One can draw his own conclusions as to these results. It would not be appropriate to present an average chill depth, as the author’s opinion is that this would be misleading. First of all, not many samples were poured. Second, the wedge poured on June 8 had so much less chill that it appears possible it received "inoculation" from a random particle of graphite. The simplest conclusion is that wedges poured with graphitic recarburizer has lower chill values than those poured with non-graphitic recarburizers, as would be expected. In any case, the chill wedge values had no relevance to the final irons produced, as will be shown by later data.

Metallographic Results
During the entire 58 days of testing, not one suspect nodularity heat was produced. Nodularity checks are taken from a coupon cast into the mold, with the sample taken to correlate to the last metal from each tundish ladle treatment. Due to the light nature of the castings being produced, a casting is removed from the molding line once each hour to check for carbides. If the amount of carbide found is in excess of that permissible for the particular job being run, those castings are contained for further examination. This foundry had a history of carbide problems, which at least partially prompted the testing of the alternative recarburizers, as the incumbent product had been "PG".

The summary of the 58 days of hourly carbide checks is shown in Table 2.

"PG" production days were from a period of time thought to be representative of normal production, and comparable to the successive test days of the other materials, but actually took place a couple of months prior to the balance of the testing. To be fair, the data in Table 2 does not tell the whole story as related to the propensity for carbides in this foundry. There is certainly nothing wrong with the "PG" product. Furnace sulfur levels when using "PG" averaged 0.0042%, normal at that time, and this foundry was encountering a lot of carbides due to the low base sulfur of the iron being melted, as has been experienced at other foundries, as well. It was this scenario that led to the testing of the alternative recarburizers, both of which were slightly higher in sulfur content that "PG". Average base sulfur levels when using the non-graphitic recarburizer was 0.0106%, and nearly identical with the graphitic material. As soon as the base sulfur level increased, the frequency of positive carbide checks was reduced dramatically. If anything, the sulfur level of "PG" was simply too low.

Effect on Mechanical Properties
According to the foundry quality program, it is required to pour a test bar daily. For certain customers, a test bar is required for each production run. Thus, there are typically anywhere from one to four bars poured daily. Both 65-45-12 and higher tensile grades are poured in the foundry, but only the data form the 65-45-12 iron was considered for purpose of this report. Mechanical property data collected is shown in Table 4.

Once again, there is little to choose from between the graphitic and non-graphitic recarburized irons. Note the standard deviations associated with all the properties of the irons produced with "PG" recarburizer. The easy conclusion is that this is an inferior product, but such is not the case. The reason for the greater variability of these irons again rests with the very low sulfur contents of those base irons and resultant carbides that occurred on a very frequent basis.

Conclusions

1. Graphitic and non-graphitic recarburizers both produced irons with less carbides and reduced property variability, as compared to when a partially graphitized refined petroleum coke was in use. The main reason for this was in the different base iron sulfur levels.
2. No significant differences could be seen in metallographic or mechanical property results, whether a graphitic, or non-graphitic recarburizer was in use.
3. Carbon recoveries of either type of product are comparable. In fact, over the 50 production days studied, there was a slight advantage to use of the non-graphitic product.
4. Based on limited information, there was no relevance between base wedge chills and final iron characteristics.
5. There results confirm the findings reported in an earlier paper on the same subject published in the 1999 AFS Transactions.

Acknowledgement
Many thanks to Fred Fudge for volunteering to wade through all of the furnace charge data and calculate the carbon recoveries.

References
W.A. Henning "Comparing Crystalline and Non-Crystalline Recarburizers in Ductile Iron Production", Presented at the 1999 Casting Congress and to be published in the 1999 AFS Transactions.