6 Steps to an Optimized Aerospace Casting
The scenario is a familiar one: an OEM has determined it wants to slash the weight of its equipment and typically this means a change in process or material. But an emerging design approach—topology optimization—shows promise for drastically reducing weight while still meeting existing quality standards using the same material.
During a presentation at last year’s AFS Metalcasting Congress, Tom Mueller used a real case history to illustrate how a military aircraft component could be redesigned using topology optimization to save the OEM thousands—perhaps even millions—of dollars.
While Mueller could not show the designs for the actual part or name the specific aircraft it was used on due to military confidentiality, he used a different shape with a similar geometry to explain the case (Figure 1).
The housing had been produced by an investment casting company in Texas for many years, and while it was not high volume, it was steady work. But the aircraft was undergoing a major lightweighting effort, and the casting buyer was considering replacing the cast housing with a sheet metal fabrication or injection molded part to cut some weight.
With the hope of saving the job, the investment caster explored how his company might use topology optimization to reduce the weight and asked Mueller to form a team to tackle the problem.
Mueller enlisted help from Altair, which makes topology optimization software called Inspire; Ultimaker a 3D printer manufacturer; Polymaker, a manufacturer of a filament designed for investment casting; and the investment caster’s engineering team.
The group outlined a six-step process to win back (or retain) the job.
1. Identify the loads and stresses.
2. Optimize the design.
3. Verify the manufacturability.
4. Prototype the casting.
5. Estimate the value.
6. Make a proposal to the customer.
1. Loads and Stresses
The team did not have access to specific design loads or stress limits the customer wanted to impose on the design, so it had to estimate them (Fig. 2). The team knew the weight of the instruments and assumed the maximum G loading and the abuse load. The location of the part in the cockpit made it convenient to be used as a footrest. The material was aluminum E357 that achieved the following properties: E=72.4 Gpa; Rho=2.67 kg/cu.mm; Y=241 MPa.
Once the loads were determined, the team created a finite element model, applied the loads and constraints, and solved for the stresses (Fig. 3).
2. Optimize Design
After defining the constraints taken from the FEA model, the team defined the fixed locations for the bolt holes, identified the surfaces that can’t move, and determined the minimum wall thickness.
Using Altair’s Inspire software for topology optimization, the team calculated the stresses in the part. Where the stress was high, material was added. Where the stress was low, material was removed. Through recalculation, new iterations were computed. After dozens to hundreds of iterations, a final optimized design was chosen (Fig. 4). Rough edges were smoothed for better aesthetics.
The casting engineering team added a strut to avoid potential warping of the printed pattern and to help casting solidification. The strut would be removed after casting to return to the minimum part weight.
The original casting was 4.72 lbs., while the new optimized design weight came to 1.66 lbs. (Fig. 6). This is a 3.06-lb. savings and a 65% weight reduction, which was far more than what the customer required.
3. Verify Manufacturability
Next, the team had to determine if the design could be cast. The geometry posed a couple of big issues. It was very complex and not easily moldable. To mold the wax pattern, the tooling would be prohibitively expensive and require dozens of small inserts to form the undercuts in the geometry. In addition, assembling the mold would take a long time, and then more time would be required when removing the inserts to retrieve the pattern. This would lead to long cycle times as well as a high tooling/molding cost. The only real option was to use printed patterns.
Filling the casting was also a challenge. The thin struts had the potential to be problematic and result in filling problems or shrink voids.
Complicating all this were quality requirements. The housing was a noncritical structural component in a flying aircraft, so it had to meet Class 4 Grade D per AMS 2175.
The casting engineering team created the gating design based on their decades of experience. Mueller said they were reasonably confident it would result in a good casting, and solidification software confirmed expectations.
4. Prototype Casting
The process to produce the prototype casting included:
• Print the patterns.
• Assemble the patterns to the sprue.
• Build the shell.
• Preheat the shell.
• Pour the molten metal.
• Remove the shell.
The first estimate to build the printed pattern was based on printing the whole pattern at once and required a significant amount of support material. The predicted build time was 66 hours, using $48 worth of material.
The next idea was to cut the pattern in half and build it with very little support material. These two halves together took 26 hours to build using $20 worth of material. Post processing just required assembling the two halves together (Fig. 6).
The printed pattern is shown in Figure 7, and the prototype casting is in Figure 8.
In terms of casting quality, the buyer required the part to meet a minimum of Class 4 Grade D per AMS 2175, but after radiographic analysis, the part was found to be a couple levels above the minimum.
The casting supplier now had the confidence to cast the housing at the quantity needed.
5. Estimate the Value and 6. Make a Proposal
To convince the buyer to use the newly optimized design, the foundry had to answer a number of questions and estimate the value of the change. What is the cost to implement the optimized geometry? What are the benefits of adopting the new design? Do the benefits justify the cost?
The cost of the new casting design was calculated by adding the pattern and casting cost and subtracting the metal savings. The pattern cost for the original was $22.79. The cost for the printed pattern of the optimized design was $120.85. The casting cost for both was about the same. Finally, the optimized design, which was 3 lbs. lighter than the original, saved $5.36 per part. The total cost of the new design was $115.49 compared to $22.79 of the original—an additional $92.70.
But a lighter part will help reduce fuel savings, as well. This is harder to determine. In 2014, American Airlines claimed it saved $1.2 million per year by replacing 35 lbs. of paper manuals with an iPad. That comes out to $32,285 in fuel savings per 1 lb. of weight reduction per year across the entire fleet of 864 planes. Thus, 1 lb. of weight reduction is estimated to reduce the cost by $39.68 per plane per year. The team estimated the military aircraft for this particular application flew half as often as the commercial airplane, leading to a $19.84 fuel savings per lb. per plane per year. A 3.06 lb. weight savings was estimated to save $60.71 per year.
In summary, the cost to implement the change was $92.70, and the annual fuel savings was estimated to be $60.71. This would mean a payback period of 1.53 years. With an average lifespan of 20 years, the aircraft’s total savings would be $1,121.50.
Total life savings for a hypothetical fleet of 1,000 aircraft would be a little bit over $1 million.
The supplier submitted the report of its findings to the customer for review by its engineering group.
Mueller said the example shows investment casting could achieve significant market expansion through reducing weight in parts through topological optimization. This instrument housing, he said, is one of dozens of small noncritical structural components in that plane. If 20 components were identified as good candidates for redesign using topology optimization, and the conversion removed 3 lbs. out of all of them, ultimately it could save $22 million over the lifetimes of the full fleet of aircraft. CS
Click here to see the article in the March/April 2022.