In 1937, the Curtis-Wright R-3350 radial engine made its first test run. This was the largest double row, air cooled radial engine designed to date. Later versions of this engine produced over 3000 hp.
The R-3350 was used exclusively during World War II in the B-29 Super Fortress, as well as used in early wartime production versions of the Lockheed Constellation.
The engine had many problems; one of the more significant was its nasty habit of catching fire. This was due to it destroying exhaust valves, which was caused by poor cooling design. At least one of the causes of the poor cooling design was the positioning of the exhaust manifold of the forward row of cylinders, which faced forward, routing heat from the exhaust stacks and exhaust collector ring right back into both rows of cylinders. See figure 1.
Figure 1 – static display of early R-3350 with exhaust ports (red) facing forward.
I have often wondered why this was done. At first blush, this may have been a financial decision in order to reduce manufacturing costs. The forward-facing exhaust stacks are easy to form, requiring no twists or turns-just a short, straight section of circular tubing leading into the collector ring, which, of course, sat in front of both rows of cylinders! In the earlier Curtis-Wright R-2600 14-cylinder engine design, the exhaust ports for the front row of cylinders faced rearward. Anyone who has witnessed the running of these aircraft engines at night will recall seeing blue flames exiting the exhaust stacks, the temperature of these flames are in the neighborhood of 3000°F. Placing all of this heat in front of a row of air-cooled cylinders will obviously cause cooling problems.
Was this design forced upon the engineering staff? No one in the right mind would have allowed such a design to proceed through to quantity manufacturing, yet it did. It was a root cause of many B-29 crashes, including the crash that killed test pilot Eddie Allen during prototype testing. The aircraft was a total loss along with all of its crew members and several people in a Seattle office building that it crashed into.
Crash Coverage Link:
http://www.historylink.org/index.cfm?DisplayPage=output.cfm&file_id=2874
Engine fires were exacerbated by igniting engine components manufactured from a magnesium alloy. Once the magnesium caught fire-burning at a temperature of about 5000°F-fires quickly burned through aluminum sheet metal and structural components and melted the wing spar, with obvious catastrophic consequences.
During the early bombing raids over the Japanese home islands by B-29s, reliability was extremely poor prior to very aggressive engine maintenance and engine change procedures implemented by Curtis Lemay in order to reduce aircraft losses due to engine fires and engine failures. If my information sources are correct, the engines were changed out after approximately 25 hours of operation, little more than 2 missions.
Postwar developments of this engine-which were used on the Lockheed Constellation, among other aircraft-had the exhaust ports for the forward row of cylinders facing aft. See figure 2. In more recent developments, the only airworthy B-29 in existence (Fifi) went through an engine change several years back that incorporated post-war production engines having the aft-facing exhaust stacks. A crew member can be quoted as stating that just before the decision was made to change out the wartime engines, a newly installed engine began “making metal” after less than seven hours of runtime! A freshly overhauled R3350 goes for around $200,000!!
The engine design had many other problems as well, not the least of which were cast cylinder heads, which did not have the strength that forged cylinder heads have. Forged cylinder heads were used on the Pratt & Whitney R-2800, a very successful engine design used on many aircraft during World War II.
Figure 2-Lockheed Constellation R-3350 engine installation showing rear-facing exhaust manifold for front cylinders.
Engine Run Video:
https://www.youtube.com/watch?v=8Dp9zBL2Ttk
The Point of All This
Take a very critical look at the design you are proposing to manufacture. If necessary, make the painful decisions early on to implement changes before entering quantity production. History demonstrates to us the consequences of not following this practice.
Modern design work can incorporate the myriad tools we have available to us that manufacturers did not have 70 years ago. Sophisticated software can simulate conditions to minimize the number of design iterations necessary. Nothing beats real tests, but we can get a lot closer to a working design now versus then.
Analysis Approach
Given that engines of this design-double row radial air-cooled engines-were being manufactured and used with good reliability-an incremental analysis approach could be implemented.
In the case of Curtis-Wright, the R-2600 engine was the predecessor to the R-3350. Test data from the R-2600 engine could be used in calibrating a simulation of the earlier design’s hypothetical modification using a forward-facing exhaust port for the front row of cylinders.
The first step would be setting up a simulation to duplicate the current successful design. Following this, changes would be made to the 3-D model that would incorporate the forward-facing exhaust stacks and exhaust manifold.
Simulation Methods
CFD-Classical Linear Analysis
The simulation method could take several forms. One of the simplest would incorporate a combination of FEA and manual calculations. Heat conduction, convection and radiation effects would be applied to the various surfaces, combined with steady-state heat sources from combustion and friction, all this blended into a thermal analysis.
This could be used as a first-cut analysis, and it would be a good one, and has the advantage of being a linear-type simulation, as opposed to nonlinear, which all CFD simulations are.
Non-Linear CFD Approach
The next step would be a computational fluid dynamic (CFD) simulation using one of the many analysis packages available. My choice would be ANSYS CFX. The use of test data would be very important to the simulation’s success. Test data is compared to program-generated output and the input variables adjusted to so that the simulation would closely match the test data.
This simulation method would be more expensive than the linear method, but considering the consequences of being wrong, I believe it would be justified.
Engine Test
Lastly, a rebuild of an R-2600 engine to the forward-facing exhaust manifold configuration may be desired or even necessary, considering the capital investment required in committing a new design that may be prone to one of the worst failure modes aircraft can have-fire!
All of these steps could be completed before committing any resources to a new design. The new design could move forward with the knowledge and confidence that a fundamental design concern was thoroughly investigated and put to rest.
Norman T. Neher, P.E.
Analytical Engineering Services, Inc.
Elko New Market, MN
www.aesmn.org