Sand Casting: the process of pouring molten metal into a sand mold to make a metal casting.
This project’s value lies not only in its end product, a working hydropower station for a rural village in Rwanda, but also in its ability to be replicated on-site. Sand casting requires only basic materials: bricks for a furnace, fuel to burn, aluminum to melt, sand to mold, and wood for frames. By using such readily accessible materials, DHE hopes to instruct the local residents to be able to fix their own turbines and start new hydro projects on their own.
Basic Sand Casting
Minimally, sand casting requires two parts: a furnace and a mold. Stock aluminum is collected in a steel crucible, which is then placed inside the furnace and heated to roughly eight-hundred degrees. During this melting process – which can take fifteen minutes or much longer, depending on the fuel and furnace – a mold of the object is made out of sand. At the beginning of our sand casting practices, we used a simple one-part mold which worked by impression, similar to footsteps on a beach. The melted aluminum is then poured into this impression, and after a few minutes of cooling and quenching, the object is finished.
This type of casting – i.e., implementing a one-part mold – resulted in turbine buckets with vertical sides: well-defined buckets, but too heavy for real use. As casting progressed, we moved to the two-part mold, a trickier but more accurate version that replicated full 3-D objects. Just like the name suggests, a two-part mold involves two parts which result in a cavity between two layers of sand: three air holes leading to the cavity allow room for aluminum to be poured and air to escape. Our resulting buckets have been lighter, smoother, and much more suitable for turbine use. This more complicated procedure, however, has presented new challenges that the DHE hydro team continues to meet and resolve.
When the Hydro Team started out this term, we had minimal experience with sand casting. Previously, DHE had made hydropower turbines in Engineering Sciences 89-90 classes or at a professional foundry, and we only had books and internet tutorials on which to rely. We spent one of the first meetings exploring sand casting in front of computers, googling “How to sandcast aluminum;” and for the freshman, our first “homework” was to choose a specific component of sand casting and research it before the weekend.
When we started, we knew two things: that we would be using a sand clay mixture to make a mold in the box, and that we needed to melt aluminum. This turned out to be more difficult than we’d anticipated.
We were proud of our first attempt, but looking back, the bucket that we ended up with looks almost hazardous. In making our sand clay mixture, we thought that in order for the mold to retain its shape, it would need a high water and clay content. We were wrong. As we poured the molten aluminum into the one-part mold that we’d crafted, the excess water started evaporating upon contact, making bubbles in the aluminum. But molten aluminum solidifies rapidly at room temperature—and even faster in snowy Hanover. The water vapor didn’t bubble through the aluminum so much as push it to the side, where the aluminum solidified. As you can see, there are holes in the bottom of our first bucket, and when flipped over, the aluminum bits are sharp and barely stuck together.
It took us a few more attempts before we realized youtube’s potential as a resource. In one trial, the mold-making took so long that the group working at the furnace to melt the aluminum (melting point ~660°C) accidentally melted the steel crucible (melting point ~1370°C) containing the aluminum. For this furnace, we’d used a steel bucket insulated with concrete, leaving a cylindrical hole in which to place fuel and the crucible. Toward the bottom of the bucket, a pipe circulated air – supplied by a hairdryer on the other end – into the furnace, giving us greater heating efficiency. Of course, aluminum came seeping out from the melting crucible and clogged the hairdryer-pipe combination that we used to feed air into the furnace. Lopsided casting boxes also did not help our progress.
A few instructional videos later, we learned that the sand clay mixture should have about a 1:9 ratio of clay to sand as opposed to the 1:1 ratio that we’d previously been practicing. As for water, we only needed enough for the sand to be able to maintain a “hot dog” shape after squeezing it with our hands. When broken in half, this ideal hot dog would break cleanly without crumbling. Our previous mixtures had the consistency of snowballs; but this new mixture would still flow smoothly through your fingers if you scooped some up.
Our sand casting improved greatly after this revelation, and our casting became fairly consistent.
We then decided to move on to the two-part mold.
This first two-part mold was surprisingly successful. The aluminum was poured in through the main pouring hole and flowed into the mold up until the air holes—but no farther. The highest point on the bucket is on its back, and we put the air holes at the farthest end of this back; however, the lip of the bucket continues on. The aluminum was flowing through the mold slowly and solidifying quickly, so that once the aluminum reached the air holes, it couldn’t move into the lips since there was no place for the air there to escape.
Several of our members worked over break to replace our hair dryer as our furnace’s air circulator, instead using wooden billows to manually pump air through the system. Manual billows relinquish our dependence upon electricity for a hair dryer and can also be made on-site in Rwanda.
Additionally, two winter projects tested the viability of a brick furnace. Both successfully melted aluminum in a brick furnace, so the practice has carried over into our work for this term.
One DHE member with ceramic training constructed several clay crucibles, as our steel crucible sometimes proved difficult to pour, and we thought we would give it a try. On returning for winter, we tested one larger clay crucible, but removed it from the furnace due to increased melting time.
Lastly, as our two-part molds had little success before break, one project focused solely upon the two-part mold, pouring wax instead of aluminum for ease of process. This has also translated into the greater success of our two-part molds this term.
13W; Winter term
Our original steel-bucket furnace has now been replaced by a brick furnace, which forms a square around the crucible and will hopefully increase efficiency and reduce melting time.
The one-part mold has been abandoned: future casting will be exclusively two-part, as the two-part mold shows promise for excellent turbine buckets following more practice. On Jan. 9th, three days after returning to campus for the winter term, we headed to the woodshop to make new boxes to replace the parellelogram-esque boxes that we’d used in the fall term. Two days later, we did our first cast. Unfortunately, the aluminum had not melted enough when we did our pour and filled only half of the mold. We filled the rest of the mold from its air holes a few minutes later.
This morning, we conducted two more trials, and on the second, successfully made our first full bucket. On the first bucket, we encountered the same problem we had before: the aluminum did not reach the extremities of the mold. The surface of the bucket was rough and felt similar to some of the buckets that we’d made earlier in the casting process. We believe that this is because there was higher water content in the sand clay mixture that we used, and this should be an easy enough issue to avoid for the next round.
For this cast, we melted almost double the amount of aluminum necessary to do a single cast. The aluminum was able to maintain a higher temperature throughout the duration of the pour, and the aluminum flowed smoothly into the mold. We believe that this is why we were successful in this attempt.
More tests will be done using ceramic crucibles, including three different sizes.
A separating agent may be employed within the two-part mold, which would result in cleaner separation of the bucket from the sand and give a finer finish to our end bucket.
For future tests, more documentation will be made concerning several variables which might be affecting our molds: e.g., pouring temperature, aluminum amount, and exact water-sand ratio.
Cecilia Robinson ’16, Shinri Kamei ’16