Less than five years ago, the mining of olivine sand ceased in the U.S., leaving North American metalcasters looking for alternative sources. Some chose to import olivine sand from Norway, though the increased shipping costs when compared to a domestic supplier made this an expensive option.
For voestalpine Nortak’s metalcasting facility in Decatur, Illinois, importing its olivine sand more than doubled its cost, from $140 per ton in 2011 to $280 in 2012. The facility’s flaskless molding system for manganese steel trackwork castings relied on new olivine facing sand for surface quality and reclaimed olivine/silica backing sand to minimize costs. Plant expansion plans were developed to increase production from 35 to 55 castings per week short-term and 75 castings per week long-term.
voestalpine Nortak contacted an engineering firm that had installed a turnkey casting facility in 2010 for a company site in France. Company officials hoped the project for the Decatur site could be partially paid for with cost savings realized from an improved sand molding system.
Choosing a New System
The molding system was assigned top priority in the phased expansion. Sand system scenarios compared were:
1. ECP, olivine/silica system, mechanical reclamation (the existing system at the time of study).
2. ECP, 100% olivine system, mechanical reclamation.
3. ECP, 100% olivine system, mechanical and thermal reclamation.
4. ECP, chromite/silica system, mechanical reclamation.
5. Furan, chromite/silica system, with mechanical reclamation and magnetic separation.
6. ECP, chromite/silica, mechanical reclamation, magnetic separation and chromite thermal reclamation.
Figure 1 shows the sand, treatment costs and investment requirements for each scenario. The furan bonded chromite/silica system offered the best payback. Figure 2 contains a schematic showing the sand flow diagram for this system.
voestalpine Nortrak still had concerns, even though two sister casting facilities in Europe were successfully running this system, including:
• Would the furan bonded chromite make rangy, manganese steel trackwork castings more crack prone?
• Would we reliably separate the silica from the chromite to avoid manganese silicate formation that produces unacceptable surface quality?
Initial trials with furan bonded chromite revealed some unexpected cure time issues. These issues were resolved and reduced concerns about cracking associated with the furan binder system. voestalpine Nortrak also made several trackwork casting molds with segments faced with 200-lb. batches of furan bonded chromite that produced good surface quality.
The goal was to be making furan bonded chromite/silica molds with reclamation by April 2014, while continuing full production. The new molding lines required relocating the plastic injection equipment, pattern shop and manganese casting processing from cleaning through shipping. voestalpine Nortrak arranged for the new core mixer (120 lbs./min.) to be delivered by August 2013 to increase the scale of the molding trials and be able to run small trackwork molds with furan bonded new chromite facing and furan bonded new silica backing.
The project timeline for the sand system changeover was initially estimated to take roughly seven months from project approval, though it ended up requiring a full calendar year. The project delays were mostly due to unexpected plastic production requirements, which delayed relocating the plastic injection equipment.
Because the new olivine and reclaimed olivine/silica silos were planned to be used for reclaimed chromite and reclaimed silica storage, the old sand system had to be shut down before the new sand
system could operate. A two week shutdown was scheduled for March 2014.
The 1,200 lb./min. mixer was installed just before the Christmas shutdown. Day bins were being filled by crane from supersacks to enable equipment debugging. Chromite/silica shakeout sand from these test molds was being segregated from the ECP-bonded olivine/silica production molds until the shutdown, when it could be processed through the new cooler/classifier (Fig. 2) for transport to the 35-ton silo located above the new magnetic separator (Fig. 3). Additionally, the magnetic separator (Fig. 4) is the crucial piece of equipment for the chromite/silica reclamation system.
The shutdown activity was critical to success. During the planned two week shutdown, the following needed to happen:
1. Empty and relocate two large sand silos from the west side of the plant to the south side. The foundation and elevated platform were ready.
2. Commission the magnetic separator and process the stored shakeout sand from trials through the magnetic separator.
3. Debug the heater/coolers and pneumatic transport system.
After the sand silos were empty, the ineffective inverted, partial cone anti-segregation device was replaced with the inverted, full cone version. The silo relocation and magnetic separator commissioning proceeded smoothly, meeting the April 12 goal.
Assessing the Installation
After the project was complete, voestalpine Nortak asked itself some very straightforward questions: How is the new system working? How is it being controlled?
Since start-up, production goals have been achieved and the project has exceeded budgeted cost savings. The goal is to keep sand usage at or below the business plan with a new sand addition rate of 12%, including new chromite cores. Currently, patterns are being faced with a blend of 9.6% new and 90.4% reclaimed chromite and backed with a blend of 11% new and 89% reclaimed silica.
Current stable binder addition rate is 1% for both facing and backing sand. Acid levels vary between 25-45%, depending on resin and blending technique.
Revised Wash and Application Practices: Facility engineers noticed sporadic surface quality issues with the olivine/magnesite wash on the 90/10 reclaimed chromite/new chromite facing sand. The olivine/magnesite coating had been used successfully on the new olivine sand facing in the old sand system. voestalpine Nortrak tried a magnesia (MgO) coating, which improved surface quality as measured by substantially reduced cleaning times. In addition the following controls were implemented:
– Baume control was shelved in favor of weight per volume measurement.
– Results from flowcoating are more consistent using the weight per volume method. If temperature biasing is added, engineers expect variability will be reduced again.
– Mold wash penetration is on the order of 0.100-0.125 in. from flowcoating with a surface layer of 0.010-0.015 in.
– Mold wash pooling can occur if care is not taken to inspect every mold and brush or blow down drips and puddles
– Because of the change to the MgO wash, the pH of the heat treat quench water has gone up to 10pH or more. As a result, castings have a bluish color coming out of the quench with little surface rusting. It is not known yet if there are long term implications of the high pH quench water.
Lenoir Magnetic Separation System Monitoring: Other than the routine daily XRF checks, the separator is inspected daily for an even flow of sand from the distributor to the upper vibratory pan feeder. If the flow is obstructed by debris from the attrition mill, the blocks are cleared and the attrition mill is inspected for holes in the screens and plates.
A layer of fine dust and metallic particles builds up on the discharge lip of the lower vibratory pan feeder that can alter the presentation of sand to the phased array roll. The lower pan is inspected weekly and cleaned as needed.
The chromite/silica separation plate was set during start-up in March 2014 and has not been changed. The sand throughput was increased by increasing the vibration parameters of the two pan feeders and slightly opening the distributor gate. By using a mouse-hole type distributor gate, as shown in Figure 5, engineers can tolerate more debris from the attrition mill without having to check for uniform flow multiple times per shift.
Mechanical Reclamation Control (Attrition Mill): The controls of the attrition mill have been upgraded to monitor sand availability and process time by adding three ammeters, which are used to measure the current in circuits. One meter is set to a high threshold, one is set to a low threshold and one verifies the vibrator motor is running. The mill starts and stops the shaker/conveyor and feed belt, and sleeps/wakes on a fixed time interval to check for sand. These upgrades mean operators no longer rely on optical switches for sand presence verification, and the mill does not shake itself empty for long intervals.
Troubleshooting: The new system has not totally been without problems. About six months after conversion, the magnetic separator could be mistakenly switched to manual mode where sand level probes were deactivated. The magnetic separator would keep running when there was no room in the reclaimed silica sand silo to accept sand transports. The result was that the reclaimed silica bin could overflow into the reclaimed chromite bin, which could then be transported into the reclaimed chromite silo. Layers of reclaimed chromite contaminated with high levels of reclaimed silica could be used as facing sand. The end result was about 15 castings scrapped for severe metal reaction/penetration (as seen in Fig. 6).
Narrow flangeway areas of trackwork castings required new chromite facing to avoid metal penetration. To achieve the cost savings of facing with a blend of 90/10 reclaimed/new chromite, engineers changed from an olivine/magnesite flow coating to a pure MgO flow coating, as noted previously. Figure 7 shows flangeway penetration with the 90/10 reclaimed/new chromite and the olivine/magnesite wash. Figure 8 shows the improved surface achieved with the magnesia wash.