3-D Solutions, Delcam’s Power Solution Helps Push Iron Foundry’s State of the Art for Turbochargers

When Cummins Engine Co.'s new Signature 600 turbocharged diesel engines start powering the biggest semi-trucks, their mileage and exhaust emissions compliance will be a function of the work of a small but technically savvy foundry pattern shop in Waukesha, Wis., called 3-D Solutions.

On those 600 hp engines will be a new turbocharger designed by Holset Engineering Co. It will be assembled with as-cast rather than machined components, a decision made to hold down costs as well as to take advantage of steady improvements in foundry quality assurance.

Holset engineers make no bones about their ambitions. "We are trying to get ‘green sand’ foundries to raise their standards for the accuracy of their surfaces and surface smoothness as close as possible to those of investment castings," said Holset’s Keith Dewhirst. He is Holset service engineering manager for North America. A wholly owned subsidiary of Cummins, Holset is based in Huddersfield, England.

Turbochargers boost engine power by pressurizing the air going into the cylinders; the pressure is generated by palm-sized turbine wheels spinning at up to 130,000 rpm. (An engine without a turbocharger is 'naturally aspirated.' Air is sucked into its cylinders by the downstroke of its pistons.)

What makes the Signature 600 turbocharger different is its twin-volute design and integral ‘wastegate,’ or bypass. The bypass opens up automatically at high torque levels to prevent the turbocharger’s turbine wheels from revving beyond their speed and torque design limits. The bypass also limits the turbocharger’s maximum ‘boost’ to the engine while keeping engine in compliance with emissions laws.

Wastegates are not new. Holset has built them into its small and medium-sized turbos since 1991. Automobile turbochargers have always had wastegates. They are new only on large turbos such as this for the Signature 600.

 Every Surface, Every Cross-Section Critical

The speed, pressure and temperature of the air blown into the engine by the turbocharger is determined partially by the device’s spiral-shaped volutes (a scroll-like passage). To build pressure exactly as specified, the cross-sections of the volute must taper precisely. Any inconsistency in the volutes’ surfaces will affect the volume and speed of the airflow. And any excessive surface roughness will create turbulence which further reduces airflow and the efficiency of the turbocharger.

Worse still are larger errors in the 3-dimensional geometry of the spiral volutes. These cause the high-speed gas flows into the turbine wheel to be uneven; they are seen by each blade as a pressure pulse as it goes through each revolution. This can have a huge impact on the turbine wheels, causing them to run ‘out of the box,’ that is, outside their envelope of tolerances. If this lasts long enough, or happens often enough, the turbocharger fails--sometimes catastrophically.

What makes this turbocharger difficult to manufacture is its highly integrated design. Into the main-housing casing goes the volute, a fitting to mount an actuator bar, an inlet flange that looks like a Monarch butterfly, and a gas outlet called a flared Marmon flange; previous designs used a straight outlet tube.

The housing has bosses, ribs and some unusually thin sections to meet mandated weight reductions while adding strength. The housing--a single 30-pound ductile iron casting--must withstand high thermal cycling without developing stress cracks and it must contain all fragments in case of a turbine wheel burst.

"Aircraft also use turbochargers like this," Dewhirst pointed out. "But in aerospace, they often use fabricated or machined castings. We can’t afford to. We have to go with the as-cast surfaces." Moreover aircraft turbochargers undergo rigorous periodic inspections. In trucks, this is not always the case--even though truckers expect them to perform flawlessly for a million miles or more.

Truck engines are the core of Cummins’ business and it is a very competitive market. Even small gains in performance are eagerly sought after and enthusiastically marketed. With a displacement of 15 litres or 915 cubic inches, the new Signature is a bit larger than the biggest of Cummins’s current high-volume diesels for the heaviest of the over-the- road trucks, Class 7s and Class 8s. (Cummins also builds diesels with up to 50 L displacements for generator sets, marine use, and auxiliary power.)

Dewhirst and Glenn Baker, Holset technical advisor on turbochargers for heavy duty trucks, explained that the volute presented the greatest design challenges: every surface and every cross-section is critical. These as-cast surfaces control the amount of cyclic strain undergone by the turbine wheel’s blades.

"No two pattern makers will interpret the geometry model in exactly the same way. There are always subtle variations," said Baker. "So each set of core tooling [from the pattern shop] has to be qualified and approved for both boost and blade-strain levels. This is under the control of no one other than the pattern maker."

Some New CAD/CAM Technology

Technology has lent a hand. The turbocharger was designed as a solid model which was imported directly into the pattern shop’s CAD system, PowerSHAPE from Delcam. Previously Holset relied on 2D drawings from a wireframe and surfacing CAD package. "We gave the pattern maker a series of station drawings which are detailed cross-section slices taken at different spots around the volute. Every point on the curves in between the stations was interpreted by the pattern maker." Baker and Dewhirst made it clear they regard such methods as The Bad Old Days.

3-D Solutions was selected for the work by Cummins’s casting supplier, Grede Foundries Co., Reedsburg, Wis. Grede selects pattern shops according to skills and track records. The selections drive many vital but subtle tooling considerations, Holset notes; this is why each set of tooling for other turbochargers is qualified to a specific range of strains. Thus Grede rather than Holset is 3-D Solutions’ customer. 3-D Solutions is also making patterns for the compressor covers for Ross Aluminum Foundries, Sidney, Ohio, a division of Eagle-Picher Industries.

PowerSHAPE also gives 3-D Solutions the capability to handle geometry from other CAD systems easily. No trouble was encountered importing Holset’s controlling geometry, a 3D solid model created in another system. PowerSHAPE also simplified 2D / 3D conversions.

"The main advantages of PowerMILL for 3-D Solutions is that complex 3D shapes can be machined much easier than before," said Mike Marcott, 3-D Solutions’ design engineer and Arrowood’s business partner. "The toolpaths and chips hit the shop floor fifty percent faster and, because it’s so easy to use, we can train new machinists very quickly.

"PowerMILL also allows us to maximise the accuracy of our machine tools. Surface finishes are better so there is less hand work. That also helps make more accurate parts," Marcott said. "All this is of great benefit to our company and our customers because we can get more work through the shop and hit targeted delivery dates. "

"One of PowerMILL’s great features is optimizing tool hangout," he continued. "If a machinist uses too long a tool it can potentially destroy the job. PowerMILL optimizes the toolpath for a given tool length by recognizing the desired length of the tool. All segments of the toolpath that would need a longer tool are then automatically deleted. Without this feature, all the software in the world is no good."

Arrowood believes the job would never have been accomplished without some other new equipment and technology in management, materials, and machining.

Thanks to modifications in the way 3-D Solutions’s shop is managed, the first patterns were done in nine weeks. In addition to the CAD/CAM software, 3-D Solutions also substituted a new tooling material, a polyurethane called PP-1052 Pattern Plank, for the traditional pattern woods, and thoroughly overhauled its main metal cutting machine tool, a Mazak VTC-30C a traveling-column type vertical machining center. Arrowood and Marcott also found economical ways to incorporate changes in the production tooling which were made after the prototype tools were built. "Otherwise this would have taken six months," Arrowood said.

"Someone’s Idea Of What A Casting Should Be"

This job like so many others started out as "someone’s idea of what a casting should be," said Arrowood. "One of the challenges of jobs like these is that casting designers don’t always have an in-depth understanding of the foundry’s tooling needs for something this complex." To make an acceptable casting, he and Marcott go through three basic steps on every job:

"First," Arrowood said, "we figure out how to pull the casting apart so that it can be modeled for tooling and so that we can design tooling which will maintain the integrity of the foundry’s process."

"Second, we eliminate all the bad molding conditions such as areas where green sand will not pack tightly. During molding the metal immediately expands into any such mushy area," Arrowood explained. "This causes a bulge in the part which must be machined away." The process of eliminating all the bad molding conditions also helps 3-D Solutions decide what shapes and surfaces are to be cored and which ones will be formed with green sand. "One big help we get from PowerSHAPE is its Split command," he continued. "It quickly identifies all surfaces which will be in the cope, i.e., above the parting line, and all those below it which will be in the drag."

"Third," said Arrowood, "we design the tooling to operate with green-sand molding machines with minimal parting line flash and core snags."

Arrowood and Marcott had seen many jobs like this before but this turned out be more complex than anyone anticipated. "We expected to do this job with two cores and those just for the volute," Arrowood said. "Many jobs that seem like they are one mold with two halves turn out to be anything but. This one ended up with four core boxes each having two halves plus a cope and drag: 12 tooling elements in all. "We had to make 12 separate models, get them approved, and machine 12 separate shapes for the green sand molds and the cores," he added.

Previous Holset turbochargers had a single volute and separate wastegate cavities; this simpler design made everything much easier.

 Part Building The Tool

Among the challenges 3-D Solutions resolved with the moldmaker’s functionality in PowerSHAPE:

• As features were added to the casing, numerous back-drafted situations and some re-entrant shapes were created. Modifying these made generating the parting line more difficult.

• Grede didn’t like the way the parting lines were done. Ultimately, the pattern and core were split in two different directions along two different parting lines.

• The volute shape had to be modified so the tooling would work well in the foundry. Surfaces were tweaked in PowerSHAPE until they worked and again matched Holset’s surfaces. The outside of the housing’s scroll area was modified so it would pull properly during molding.

• A core was added at the volute’s Monarch flange because this area was too small for green sand to reliably pack. The core actually wrapped around the parting line -- helically.

• New parting banks, the flat faces where the mold halves close together, were required. Where these had to be cored surfaces, PowerSHAPE was used to specify natural, or zero, draft angles, create the parting banks in 3D, and then split the core in half. It got trickier. Closer to the pattern, the parting bank consisted of complex curved surfaces which became flat about an inch out.

• A cleat face was redesigned with a molded-in slot to help avoid heat-stress cracks in service. With PowerSHAPE and an international conference call, the slot was designed in 2-1/2 hours even though the tolerances were very difficult to hold.

In the crotch of the casting where the inlet value with its Monarch flange meets the housing, 3-D Solutions had to use a core fillet split by the parting line, something pattern makers try to avoid. Explained Arrowood: "With PowerSHAPE displaying the fillet and identifying which surfaces were on which sides of the parting line, it was straight-forward. This could not have been done without PowerSHAPE’s interactive multi-surface filleting. Other systems treat fillets as a long string of individual, separate patches," Arrowood noted, "but PowerSHAPE combines these into a single surface."

Even after all the geometry was created, verified and approved, 3-D Solutions still had to build the tool. The main steps were:

• Pasting the two volute core halves together including making sure they would interlock tightly in the mold; even space for glue had to be calculated.
• Pasting in the core for the wastegate passage
• Making the drag pattern for the mold impression in the green sand.
• Setting the butterfly flange and crotch core into the drag at the waste gate.
• Adding the other cores to the drag.
• Adding a large core cover for the Marmon body flange in the cope side of the mold.

"A pattern maker cannot be successful with work this complex unless he embraces the new materials, new methods, and new software," Arrowood said. But they aren’t enough: good communication throughout the project, from beginning to end, is essential.

Because Holset is an English company, there were some language difficulties. Accents, technical terms, and even subtle differences in the meanings of everyday words caused problems, Arrowood noted, especially when people were hurrying. As some wit said, "The English and the Americans are one people divided by a common language."

The customer, of course, always has the last word. Martin Holzschuh, foundry engineer at Grede in Reedsburg, also noted that the central issue in the Signature turbocharger job was communications. "We have traditionally tried to communicate very complex 3D geometry with 2D drawings. Pattern makers have to do a lot of interpreting and that’s when the pattern may not represent the intent of the designer. Now we can communicate a 3D solid directly to the pattern maker and get rid of problems of interpreting the design intent from 2D drawings."

"The files were communicated smoothly from another sustem," Holzschuh added.