|
||||
|
||||
|
|
ATAS Practical Usage at Torrance Casting, Inc.
John Torrance ABSTRACT Torrance compared to other ATAS foundry users is relatively small in its melting capacity, roughly 40 tons per day, but has experienced big savings through using ATAS in the first year alone. We have saved over $60,000 through optimizing our inoculant and FeSiMg additions, and have reduced our overall metallurgical scrap. Besides using ATAS for initial cost savings, I believe it is most valuable when it’s used at the furnace. Recorded ACEL (Active Carbon Equivalent = C% + ¼ Si% + ½ P%) and silicon values, along with knowing the bath’s equilibrium temperature versus the bath’s holding temperature is my key drivers for quality control. |
|||
| INTRODUCTION ATAS, created by NovaCast Technologies in Ronneby, Sweden, is an advanced thermal analysis system that has established itself as a valuable tool used in the ferrous metal casting industry. Technical papers within the past few years have cited examples on ATAS optimizing each foundries melting, treating, and pouring process. ATAS’s method is making relevant parameter data available for analysis, verifying and correcting each process using pattern recognition learning for quality assurance, and testing and selecting optimal alloying materials and inoculants. As a result foundries have experienced and reported an overall reduction in metallurgical scrap, improvement on casting yield, less variations in mechanical properties, and a cost-effective benefit by reducing the amounts of Mg-alloys and inoculants used. I was first introduced to ATAS by Rudy Sillén (founder of NovaCast) while pursuing my M.S. degree in Metallurgical Engineering at the University of Alabama under Dr. Doru Stefanescu. Through Rudy’s mentoring while at Alabama and our continual friendship, my knowledge of solidification and thermal analysis extended beyond my University experience. When I returned to Torrance we purchased and installed ATAS. Rudy’s immediate recommendations were to work on lowering our inoculant levels. At that time Rudy made a blanket statement by saying, “most foundries over inoculate their iron without even knowing it.” During that conversation I mentioned that prior to my return we were having issues with shrink on some of our ductile jobs. The Company was advised to keep their current post inoculant addition rate of 0.7% FeSi75 and also add a proprietary in-mold tab of roughly 0.35% FeSi75. The new process resulted in more shrinkage scrap which put the Company in even a worse situation. Once I told this to Rudy, he again reaffirmed me to use ATAS for inoculant optimization for all our alloys poured. Before using ATAS I read, and highly recommend reading, “The Use of Thermal Analysis for Process Control of Ductile Iron”, by Adrian Udroiu who presented this paper at the NovaCast 2002 Seminar. The paper represents a great overview on ATAS and reviews topics such as;
TESTING AND OPTIMIZATION As for our ductile iron procedure the previous method was to add 1.6% FeSiMg using steel chips as cover material. Through ATAS optimization the addition rate dropped to 1.16% FeSiMg, a cost savings of $12,500 during the first year, with 0.4% foundry grade FeSi75 as a cover. Using 0.4% FeSi75 as cover material was a recommendation from Rudy and Metallurgical Engineer Dr. Torbjorn Skaland. Taken from an email discussion with Dr. Skaland he wrote, “In my opinion, the Mg treatment itself is the most important preconditioning of ductile iron, and I like to say that the basis for a subsequent effective inoculation is made during the Mg treatment. In fact, I believe that very often the Mg treatment is a more important part of inoculation (nucleation) than the post-inoculant addition itself. Just imagine the huge difference in nucleation behavior between a pure Mg metal converter process on the one extreme and an in-mold Mg treatment on the other extreme. The pure Mg treatment will be completely white iron if left un-inoculated. The in-mold iron cannot be inoculated and cannot produce carbides even if you try to (I have tried high carbide promoting elements and copper chills without success in forming carbides!) Hence, my belief is that you can best get around issues with micro-shrinkage by designing the MgFeSi alloy and the inoculant alloy correctly. Other alloying is secondary to this issue. The key design criteria for MgFeSi are Mg and Ca contents, as well as rare earth type and content.” The optimized ductile process improved the thermal properties of the iron (shown through ATAS results), allowed the furnace holding temperature to be reduced by 50 °F (which reduces overall nucleation fading), and also improved the cleanliness of the ladle iron by eliminating the rusty cover steel. The tundish pocket design was even modified for better ATAS results, as shown in Figure 1. We changed the half moon dam design with a circular pocket design as shown in the figure. The new cylindrical pocket is designed to contain the calculated FeSiMg and FeSi75 cover material, and also suppress the magnesium reaction thereby improving the magnesium yield and reducing dross formation. Figure 1 Tundish ladle pocket modification.
TORRANCE TESTING PROCEDURES Figure 2 Standard Electro-nite 0.65 cm modulus cup. However, what if ATAS is used at an automotive foundry that pours thin-walled castings (high cooling rates) that require a high ACEL value (hypereutectic)? Or what if ATAS is used at a wind turbine foundry that pours extremely heavy sectioned castings (very slow cooling rates) that require a low ACEL target (hypoeutectic) to avoid carbon flotation and shrink? Would this 0.65 cm modulus cup represent those scenarios well? To date I believe only two different size Electro-nite pouring cups have been tested with ATAS, the standard 0.65 cm, cup, see Figure 2 and the smaller 12 mm cup. I made an in-house trial where I wanted to see what would happen if I poured two cups side by side with the same iron where the second cup’s modulus was much higher. To modify the second cup I wrapped a ceramic fiber blanket around the sides and covered the cup once the cup was poured as shown in Figure 3. The modulus of this modified cup was not calculated. Figure 3 Top view looking down on Electro-nite’s 0.65 cm modulus cup wrapped in ceramic fiber blanket. After pouring, the cup is covered by the blanket. Two trials were poured with base gray furnace iron. The top curves in Figure 4 represent the first trial and the bottom curves represent the second trial. For both trials the curves on the left were poured into the regular 0.65 cm modulus cup while the curves on the right were poured into the modified cups. The pouring temperatures between the regular and modified cups were roughly the same. The modified cups took much longer to solidify because of the slower cooling rate. As expected, different ACEL and Active Carbon, “C”, values (active carbon = dissolved carbon that has crystallized as graphite. Due to the evolved latent heat it will influence the cooling curve from TeLow to TS) were recorded between the regular and modified cups. It is my belief that seeing ATAS was designed off the 0.65cm modulus cup, these results confirm that ATAS recalibration is needed when changing over to a different size cup (change in modulus). It was also interesting to see that the regular cups (higher cooling rate) had a higher liquidus temperature (TL) than the modified cups (slower cooling rate). This was not to be expected seeing a higher cooling rate would promote undercooling, due to lack of nucleation sites, which would lower the liquidus temperature (TL). Figure 4 Regular (0.65 cm modulus) cups (left side) vs. modified (insulated) cups (right side). INITIAL ATAS FURNACE USAGE 1. Inaccurate filling (possibly diluting tellurium effect thereby altering results) 2. Carbon equivalent (CE) results vary when pouring two cups side by side using a tellurium system. (Tighter ACEL target ranges have been maintained using ATAS) 3. The tellurium system only records “total carbon” (active plus inactive carbon) where ATAS grey cups supposedly predict “active carbon”. (To be discussed in depth later) After running the ATAS system side by side with our tellurium system for more than 6 months without having any issues I decided to drop out the tellurium system. The initial observations were;
I needed to find a way around having to wait this extra amount of time (possibly reducing our pouring capacity) for ATAS’s curve to complete and report the predicted “active carbon”. My solution was remembering how I implemented a Dr. Stefanescu test question in our research foundry at Alabama. The test question involves using the Ellingham diagram to explain how a “system” always wants to strive towards its lowest free energy as illustrated in Figure 5. Based on this principle if one knows the carbon/silicon ratio of their furnace bath, the equilibrium temperature can be calculated and compared to the bath’s temperature which will then assist in predicting whether a bath of iron is losing carbon or silicon over time. Therefore if the bath temperature is below the equilibrium temperature silicon slag (oxides) will form on top of the melt surface and silicon units will be lost as an example. On the other hand if the bath temperature is above the equilibrium temperature, carbon monoxide (CO) gas will form and carbon units will be lost. It is my understanding that this is not a linear relationship but behaves exponentially. So, if you are superheating your bath you will lose more carbon than if you were just holding the bath above the equilibrium temperature. Nevertheless, this knowledge allowed me to educate the furnace operators and explain to them the thermodynamic basics behind their daily work. Figure 5 Metallurgical test question explaining Gibbs free energy,
I created a Microsoft Excel mass balance spreadsheet for the furnace operators that act as their “guide” for calculating trim additions and figuring out the equilibrium temperature for each furnace, as shown in Figure 6. Currently ATAS has a system built into the program that also assists the user for making trim additions, but there are two issues I had when trying to use this feature. The first problem was based on ACEL and carbon results. The program only advised on using either carbon or FeSi75 for making the trim additions, not both. I designed my Excel program to recommend adding both materials at once if needed, as shown under Furnace 1 in Figure 6. The second problem was that I had to wait over three minutes for the curve to complete before I could view the “active carbon” and also review ATAS’s recommendations. As I mentioned before, this could potentially slow production down. Figure 6 Excel cheat sheet used by our furnace operators.
ACTIVE CARBON Total Carbon =Active Carbon + Inactive Carbon. Carbon analyzed with a spectrometer or Leco shows ALL carbon in a sample. However it does not tell us in which form the carbon is present. It can be as carbon dissolved in the ferrite, as iron carbide, as other carbides, as graphite precipitated from carbon dissolved in the liquid during solidification, and as graphite particles that were not dissolved but suspended in the liquid. For the metallurgist the interesting part is of course to know how much carbon there is that can result in graphite precipitation, known as “active carbon”. Active Carbon =dissolved carbon that has crystallized as graphite. Due to evolved latent heat it will influence the cooling curve from TeLow to TS; therefore it is measured by calculating the area under the curve from TeLow to TS. See figure Figure 7 Inactive Carbon =what is tied up as primary carbides, and /or micro particles of graphite that has not been dissolved. Figure 7 Example of cooling curve which shows area between TeLow and TS. During ATAS installation NovaCast recommends using the carbon correction function (see Figure 8) to offset the carbon differences found between ATAS and Leco carbon combustion, or to use a tellurium thermal analysis system that predicts total carbon. The problem I see is how can we be confident that the “active carbon” from ATAS equals the “total carbon” measured by other means? To look into this further we have to mention what factors can alter “active carbon” metallurgically.
Figure 8 ATAS screen showing carbon correction function. “Active carbon” can be influenced metallurgically by the amount of oxygen in the melt. The oxygen level in the bath is erratic and hard to control. High oxygen content reduces the ACEL (higher liquidus temperature) which makes the iron solidify as is the carbon content were lower than revealed by chemical analysis (“total carbon”). Factors that increase the oxygen content in iron include high amounts of steel scrap (worse if rusty), open furnace lid while melting and pouring (oxidizing atmosphere), and holding above the equilibrium temperature for a long time. Factors that decrease the oxygen content in the iron are addition of deoxidants such as FeSi, SiC, aluminum, or magnesium. Theoretically a small amount of deoxidant could be added to the pouring cup or spoon in order to drive out as much oxygen in the iron as possible when trying to compare ATAS “active carbon” results with Leco “total carbon” results (assuming there is no graphite in the Leco iron sample). Another variable that can alter the amount of “active carbon” precipitation is the solidification rate of the iron poured as shown in Figure 9. This figure is the top portion of Figure 4 where the curve on the right had an insulated ceramic fiber blanket (see Figure 3) wrapped around the cup which slowed the solidification rate. Here you can see the “active carbon” on the right is higher because more time was allowed for graphite precipitation. Another way of altering the solidification is through different pouring temperatures. It is highly recommended to pour the cups at the same temperature all the time. The red arrows in Figure 9 point to the top of the curve that represents 2462 °F. Pouring temperatures above this may break the quartz tubes and result in failed curves. Figure 9 Top portion of Figure 4. Curve on left represents normal cooling curve, while curve on right was from a cup that was insulated to alter the cooling rate. Care must be taken to make sure the cooling curves are correct before fully trusting the carbon reading. A “broken” curve, illustrated in Figure 10, is when the quartz tube fractures and alters the “active carbon” reading because the area between TeLow and TS is falsely smaller. So far I have come across two scenarios where the quartz tube has failed. The first situation shown on the bottom of Figure 10 is when there is so much graphite precipitation at once, reflected by a high recalescence value; the quartz tube cannot withstand the pressure from the precipitation of graphite and fractures usually around TeHigh. The second situation shown on the upper portion of Figure 10 is when there has already been pressure exerted on the quartz tube and there was a mechanical vibration such as dropping charge materials on the floor accidentally which then fractures the tube. Figure 10 Example of "broken curves" that alter carbon readout. MODIFIED AND CURRENT ATAS FURNACE USAGE Periodically I would burn a furnace Leco wafer and compare the recorded silicon versus the back calculated ATAS silicon value to make sure the correct carbon/silicon ratio was kept in line. On one occasion I noticed the ratio was way off, over 30 points! After that occurrence it was my belief that ATAS was inaccurately calculating “active carbon”. Therefore I ran some experiments where I would purposely add a bunch of carbon to the melt or allow the carbon to fade away to alter the carbon points. Besides tracking the “active carbon”, ACEL results were also recorded as shown in Figure 11. Samples were taken throughout four different heats. Notice on heat #2 at the beginning of the furnace the recorded ATAS Verifier (non tellurium cup) ACEL was 3.97 while at the end of the furnace it was 3.91. Carbon was being lost throughout the furnace but ATAS Verifier was reporting a carbon drop of only two points. Remember from thermodynamics at this point carbon should be fading due to being above the equilibrium temperature. I purposely added FeSi75 at the end of this furnace to try and “save” the ACEL. ATAS White (tellurium cup), DataCast (tellurium cup), and Leco all showed that the “total carbon” was down and the silicon was up as expected. As we know “active carbon” cannot be greater than “total carbon” as it was shown here. Heat #3 was a furnace where I intentionally added FeSi75 throughout emptying. Notice how ACEL started at 3.93 and ended up at 3.94, a one point change while constantly adding FeSi75. “Active carbon” actually went up a couple of points while the ATAS Verifier calculated silicon value, from using the Excel spreadsheet, stayed roughly the same. However the tellurium cups and Leco results showed that silicon increased, as expected, and the carbon dropped. Heat #4 I intentionally added a lot of carbon to see if the “active carbon” value would increase. The results stayed the same as not to be expected. We know the carbon got into solution however because the ACEL drastically jumped up accordingly. Figure 11 ATAS Verifier (non tellurium cup), ATAS White (tellurium cup), DataCast (tellurium cup), and Leco data comparisons. The goal of using ATAS Verifier (non tellurium) cups at the furnace is for getting stable ACEL readings, which is by far the most important parameter, and to trust the “active carbon” (carbon/silicon ratio). As you can see I had an issue with the “active carbon” reporting, which throws off the carbon/silicon ratio. So, I developed a new thermal analysis testing method that is still being used today, without having any issues. The new and current method is to pour an ATAS non tellurium cup at the furnace to record the ACEL, which is read 30 seconds into the curve. A tellurium cup (ATAS White or DataCast for example) is also poured at the beginning of every new furnace to record the silicon value, usually out of the same spoon of iron. If you are in a hurry to get a silicon result, I would recommend pouring a DataCast sample over an ATAS White sample because the DataCast curve completes a lot sooner. Besides that, either system would work. Also, I had to modify the Excel mass balance program so that it would back calculate my carbon value instead of silicon. Warning: When the type K thermocouple wire is exposed to temperatures over 2480 °F significant variation of the predicted silicon measurement can occur. To maintain the best accuracy of the carbon and silicon prediction the cup manufacturer recommends pouring the molten iron into the cup at temperatures less then 2480 °F. Independent test results show that if the temperature is poured below this value, the reported silicon value may offset the actual target/chemical value plus or minus 4 points, resulting in an eight point spread. When the pouring temperature is above 2480 °F, the reported value may offset the target value plus or minus 8 points, or a 16 point spread. Instead of pouring a tellurium cup to analyze the silicon content, a spectrometer could also be used. This by far is the most accurate way to measure silicon, but I decided to not go this route because not only was it a slower method, more manpower would be needed to run the spectrometer. Besides, if our tellurium cups were poured below 2480 °F our silicon value should be at the most 4 points off target. That is equivalent to only one ACEL point, and obviously better than being off 30 silicon points! Currently our target ACEL range for gray and ductile iron is 3 points total for each. So looking at it from a different perspective as long as we have our initial ACEL and silicon values and knowing our holding temperature is above the equilibrium temperature, a change in ACEL throughout the furnace is a change in the carbon (loss if ACEL drops). Again, the rest of the cups only need to record ACEL for the operator to use the Excel mass balance spreadsheet. Double checking silicon by pouring an extra tellurium cup is always an option too. Conclusion has impOnce optimization testing has been carried through, I believe using ATAS at the furnace is where is it is needed most. Reducing process variation upstream has improved our overall quality control downstream. |
||||
|
||||
The improved tundish iron quality along with ATAS utilization allowed us to drop our current 0.7% ductile iron inoculant down to a proprietary FeSi75 inoculant addition rate of 0.18%. This was a cost savings of $12, 240 during the first year.
By knowing the bath’s equilibrium temperature, I changed our sampling procedure so that on a new furnace, the furnace operators would pour an ATAS grey cup and allow the curve to go to completion. Allowing the curve to go to completion is done solely to obtain the “active carbon” reading. The ACEL reading is still viewed at the liquidus temperature, roughly 30 seconds into the curve. The furnace operators would then input the recorded ACEL and carbon reading into the Excel mass balance program and make any recommended trim additions. Once the furnace operators know the ACEL
and “active carbon” (C) reading, and the fact that we hold our pouring temperature above the equilibrium temperature, all they need to do for the remainder of the furnace is to pour a cup and obtain the ACEL (which takes 30 seconds not 3 minutes to view). Any variation between the newly poured ACEL and the previously poured ACEL is relative to the furnace carbon activity, which is based on the furnace temperature versus equilibrium temperature, in theory. Example, if the furnace temperature is above the calculated equilibrium temperature and the ACEL dropped by three points, only carbon would be used to trim the furnace.
