6 September 1944
Allison Engine Company
Indianapolis, Indian, USA
The bosses at Allison had kept Tom Kaczmarczyk busy over the past few months. His recommendation to build a Turbo-Compound version of the V-1710 for the Convair XP-81 had landed on the right ears and the Government had awarded Allison a development contract for just such an engine. While several of the most talented and brightest engineers were working with G-E on the I-40 Jet Turbine engine for the new Lockheed P-80; Tom was given a rag-tag team of the “leftover” engineers.
In truth, he could not have picked a better group.
The men working with him on the Turbo-Compound project were some of the most experienced engine designers at Allison. They were the men who first laid the V-1710 down on paper a decade earlier and prior to that had worked to improve the old Liberty engine. Some of these men were with James Allison from the beginning, working on “hot-rodding” cars for racing and pleasure. They were passed over on the Jet teams because of their lack of knowledge on the new Gas Turbine systems, but their ability to tweak, adjust, and improve the piston engine was unparalleled.
With full War Department backing, the Allison team was joined by a few NACA scientists who had worked on the Turbo-Compound concept the previous year. Chief among these was Leland Desmon, who was co-authoring the final NACA report on the findings of their investigation with Dick Turner. Dick Turner was the engineer who had developed the methods already being used to design proper exhaust stacks to produce the greatest benefit of exhaust thrust with minimal backpressure.
Taken together, the combination of those men at the forefront of exhaust turbine systems and some of the best piston engine builders in the country left Tom with a project team that had the exact set of skills and experience needed to make the Turbo-Compound V-1710 happen.
Work had started almost immediately after Tom made the proposal. The first phase of development was to get as much technical data from NACA as they could at the time and use that little bit of information to build a proof-of-concept engine prototype. For this, they had grabbed a surplus E-Series engine and cobbled together a power-recovery turbine by scavenging the exhaust turbine from a G-E C-Series turbo-supercharger. They combined all of the exhaust using standard P-38 exhaust collectors and fed the turbine through its standard single torus-type nozzle box.
Those first tests were limited and plagued with failures due to the rough nature of the construction but on the few full runs they were able to record they proved a gain of as much as 10% Horsepower in the engine system.
After the official acceptance of the R&D Proposal came down, work began in earnest on a production level prototype and NACA became officially involved in a development partnership.
Mr. Desmond had worked out, based on the available F32R engine data at 3000 RPM, that the most efficient use of exhaust blowdown power would be to use six, two-to-one exhaust pipes, each terminating in a 2.83 square inch nozzle. The arrangement of the pipes were such that only cylinders in a single bank with non-overlapping exhaust cycles would share a nozzle. They had tried finding an arrangement that would allow the impulses to “cross” the turbine wheel from each bank to better balance the forces on it but any such arrangement would result in overly complicated exhaust layouts. Instead they settled on a “mirrored” exhaust setup which—while not perfectly balancing the forces on the turbine—served to simplify the layout and allow adjustment in pipe length to improve the impulse timing. This arrangement provided impulse to the turbine, given the standard V-1710 firing order (1L-2R-5L-4R-3L-1R-6L-5R-2L-3R-4L-6R), at 30°-270°-90°-210°-150°-330°-30°-270°-90°-210°-150°-330°.
The total nozzle area with this arrangement was about 17 square inches but because two cylinders each share a single nozzle the effective area was nearly 34 square inches. The area was calculated to be sufficient to nullify any additional exhaust back-pressure at the engine due to restriction at the jet nozzle at all Exhaust Pressure to Manifold Pressure ratios down to 0.7. At lower pressure ratios—when at high altitude or when running a Manifold Absolute Pressure above 48”Hg. at Sea Level—the restriction should be minimal and have only a small impact on the engine Brake Horsepower due to additional pumping losses.
This is where Leland’s insight from his work with NACA really made a difference.
“We honestly don’t need to consider any additional effects of back pressure for the engine system beyond that imposed by the jet stacks themselves,” he explained. “The F32 engine was designed to exhaust to a collector to stabilize the pressure for the Turbo-Supercharger, which means the engine power data already factors in the pumping losses from the increase of back-pressure at the given manifold pressures and altitudes.”
This prompted Tom to ask the obvious follow-on question, “But won’t the Blowdown turbine create additional backpressure on the engine?”
“No,” Leland continued, “the exhaust pressure critical to the engine is the net total pressure of the system at the turbine exit. If we continue to feed a turbo after the Blowdown turbine, then the turbo waste-gate will manage the pressure in the system as normal. All we need to do is collect the exhaust pulses after their work in the blowdown turbine to a collector ring and feed that right back into the standard turbo.”
“Oh-kay, but doesn’t the blowdown turbine then reduce the pressure available to the turbo?”
“In theory, it shouldn’t. Dick Turner, Ron Doyle, and I talked about it after our work on the R-1340. Since the blowdown turbine relies on an impulse jet and the exhaust turbine of a Turbo-Supercharger is a steady-flow system, we
think that so long as the exhaust pressure at the blowdown outlet remains sufficient to maintain steady system pressure to the turbo, there should be zero loss.”
“And you’re supposing the turbo waste-gate will ensure that happens in fact.”
“Correct.”
To build the turbine they once again used the wheel and bearings from a C-Series Turbo-Supercharger, this time they were able to get a new C-23 exhaust turbine wheel directly from G-E which had a re-rated speed limit. They disassembled the torus nozzle box and removed the upper case of the housing. In its place they built new upper with the six nozzles integrated directly into it and reconfigured the oil supply to properly accommodate the new design. Within each nozzle was a series of vanes to deflect the incoming exhaust jets to the appropriate angle to achieve optimum impact on the turbine blades.
Since the C-Series exhaust turbine was already designed for toroidal gas flow, the modifications to the turbine itself were minimal. Over a series of “dry” tests, using compressed air to test the new nozzle box and blade-to-jet angle, they were forced to make small adjustments to the turbine blade design. Leland indicated that they would probably want to contract G-E or another turbine developer with building a new turbine specifically for their use, but that this arrangement should be sufficient for the initial development.
With the modified turbine and new nozzle boxes complete the next step was to fabricate a custom turbine exhaust manifold which would exit directly to a collector ring. The collector ring was built to match the 29 inch diameter turbine housing and was fed from a one-to-three split-pipe from the inside of the ring. The inner pipe diameter of the ring decreased slightly toward the exit until it was the same as the standard P-38 exhaust pipe.
For complete tests, they ran this pipe directly to the B-33 Turbo-Supercharger used with the F32 engine. In this way, they would have a complete system for thorough testing which would include both exhaust recovery stages—the new blowdown turbine and the existing turbo—as well as both induction compressors—the turbo and the engine driven supercharger.
The tests were carried out on the bench with the engine propeller shaft connected to a 2500 horsepower brake originally built when they were developing the F29/F32 engine. Since the blowdown turbine speed was dependent on the mean exhaust jet velocity and temperature, rather than on engine speed, they could not directly gear it directly to the engine. Instead they would need to build a custom gearbox utilizing a fluid coupling for power transportation but the design parameters of such a gearbox would be dependent on the speed and power differential between the turbine and engine crank. For the bench tests they instead connected the output shaft of the turbine to a 750 horsepower brake for direct measurement of the power recovery.
Induction, engine, and oil cooling were handled using standard P-38 radiators in a remote setup with high speed fans blowing across them to achieve the required airflow. They discussed testing under fully manual control but decided instead to use a P-38J Unit Engine Control system so they can make direct comparisons and estimations based on well-known performance characteristics at each setting.
The PT-13E9 carburetor was replaced with the PT-13E12, re-configured to use the newly available 115/145 octane fuel which allowed slightly leaner fuel-air-ratios. In all other ways the core of power unit was an unmodified F32R (V-1710-123).
For the past two months now they have been running the engine is a series of tests. The time was full of starts and stops initially as they worked out the kinks with the new turbine and adjusted the fuel and water-methanol Anti-Detonation Injection metering.
Although NACA’s estimation of exhaust pressure control was largely borne out, since the system was designed to always allow some flow into the turbo even with the waste-gate fully open, the minimum exhaust pressure at the blowdown turbine exit they were able to achieve was 33.2”Hg. When running at 1600 RPM and throttled down to 28”Hg. MAP this produced a pressure coefficient on the blowdown turbine of 1.2 and limited the power to a theoretical maximum of only 16.6 HP. Still, this was a free gain in the entire system and no additional pumping losses were observed at the low boost settings.
As they increased the pressure differential, however, they soon ran into a few limitations of the turbine, specifically design over-speeds at maximum settings and—more critically—high temperatures due to the direct impingement of the exhaust jets on the blades when running an Maximum LEAN.
These tests also uncovered an unexpected level of pumping horsepower loss at high boost, RICH fuel settings. Examination of the data revealed that the 2.83” nozzles were too small for 3200 RPM operation when pushing high exhaust masses, causing an initial restriction to the gas flow
prior to entry to the blowdown turbine. Leland recalculated the design size and after some discussion they landed on 3.01 square inch nozzle size as a sufficient balance between increase jet velocity and minimal pressure loading. The result was that the net nozzle-box area was increased to 18.1 square inches with an effective area of 36.2 square inches. This had the additional benefit of permitting a slight increase in the turbine diameter to 12 3/16”, thereby reducing the blade-tip speed and RPM—although it was still insufficient to bring it below the design speed of the C-23 turbine wheel it did allow high power operation within the turbine over-speed limits for short periods.
With the re-built blowdown turbine, they continued the tests. A series of successful runs were completed at progressive engine power settings available through the UEC. They were unable to run all the way up to the theoretical MAP limit of 100”Hg. with the 145 octane fuel and ADI due to mass flow restrictions in the carburetor. At their tested atmospheric pressure of 29.17”Hg the best they were able to achieve before hitting the mass restriction was 78.8”Hg MAP. A few quick “back of the napkin” calculations and they determined that to achieve 100”Hg at Sea Level the Carburetor would need to be able to flow nearly 18,500 pounds per hour, 3,500 over the design limits of the PT-13E12.
Even so, the maximum power they hit on the bench at 3200 RPM running up to 78.8”Hg. MAP under War Emergency Power settings using ADI was 2,284.1 Horsepower on the engine brake and another 567.1 horsepower on the blowdown turbine brake. After assuming a constant propeller efficiency of 0.85—admittedly high but an easy number to work with for these tests—they calculated an Engine Thrust Horsepower of 1941.5. Adding in the Net Turbine Horsepower, less an assumed 0.85 gear-box efficiency, and then once more adjusting for a standardized propeller efficiency, the end result was 2,351.3 Net Thrust Horsepower with the Power Recovery Turbine in place. This represented a 21% improvement in usable power at the tested altitude pressure.
Similar tests were performed at each significant UEC setting and the entire system performance calculated with the same assumptions. Power gains ranged from 4% at 1600 RPM, 28”Hg. MAP, and a fuel-air ratio of 0.063, through 12% at Rated Power of 2600 RPM, 42.5”Hg. MAP, with a near Stoichiometric FAR of 0.069, and on to the high of 21%.
The power gains were impressive but just as important for the design requirements of the intended aircraft role of Very Long Range Escort was the reduction in Net Thrust Specific Fuel Consumption. These ranged from 4% at the low end on up to about 17% at the high end, with a calculated gain of about 8% at Maximum Cruise.
Tom knew the numbers were for “ideal” conditions and that the practical performance would likely be reduced due to the dynamic nature of the physics involved and lower than expected efficiencies but they were still sufficiently in line with the available NACA data for him to sign off of the final report to his R&D bosses for full dissemination to the affected parties.
With the static altitude tests complete, the Turbine Group started working out the formulas needed to properly predict the system performance under varying conditions and altitudes. Without full lift/drag and weight information for the airplane, they had to settle for base-line numbers assuming no Ram effect. Also, without the full propeller performance curves they stuck with the 0.85 maximum efficiency they assumed on static tests.
In the end, they were able to use the available test data to provide three performance charts: one representative of their bench tests; and one each showing the Performance at Altitude up to 45,000 feet under Normal Rated Power and the calculated Maximum Lean Cruise.
Power Runs, 29.17”Hg. Atmospheric pressure
When they looked deeper into the data another interesting possibly became apparent. At all of the tested power settings the blowdown turbine, before assumed gearbox losses, output more power than the engine driven supercharger required. This could allow them to decouple the supercharger from the engine and power it from the blowdown output. This would allow them to reduce the load on the engine by the amount used by the supercharger plus the amount lost through the supercharger gearbox. Assuming they could find a way to continue to route any excess power from the turbine to the engine they calculated a net gain of 1%-4% in available brake horsepower.
The Engine Group of the project team were meanwhile finding ways to massage more power and efficiency from the core power unit in what was already internally referred to as the V-1710-G.
Their first task, based on the input from the Turbine Group, was to increase the Mass Air Flow of the induction system and improve the fuel metering to match. They had considered going to one of the new four-barrel Bendix-Stromberg pressure carburetors but Bendix came back with a memorandum announcing to Allison their intent to develop new Speed-Density Fuel-Injection systems.
Moving forward with the assumption that they would use a newly developed Fuel-Injection system built specifically for the requirements of the G-Series engine they were now focusing on 4 main tasks:
- Integrate the Unit Engine Control directly onto the engine
- Develop ideal power settings for the engine
- Build specific engine exhaust parameters for more precise calculation of the exhaust nozzles
- Investigate removal of the engine-driven Supercharger and replacement with a blowdown driven supercharger
Together, the entire group were also exploring ways of improving the cooling for the turbines and the induction charge. Convair had set specific drag area and packaging benchmarks that Tom’s team at Allison were hard pressed to meet at the current stage of development.
There was a lot work ahead of them, possibly even several years, but Tom felt they had a good start and already they had drawn up an initial installation package for modification of an F-32 with the added blowdown turbine.