U-Bend Replacement Test

Posted by Alex Darmos on


For quite some time, there have been many claims and conjectures about the performance impact from removing the factory U-Bend in the exhaust system on the GM 3800 Liter V6 (Naturally Aspirated L36 and Supercharged L67). For those who are not familiar with the U-Bend, it is a bend that GM designed into the exhaust pipe exiting the catalytic converter. (See Above Photo) The purpose of the bend it to permit the O2 sensor downstream of the catalytic converter to be mounted in a position that would provide the maximum protection from possible road hazard damage. The downside of this configuration is that the bend possess a moderate restriction to exhaust flow. A popular modification is to replace the U-Bend with a straight pipe that contains a fitting for the O2 sensor. The sensor is typically mounted in a horizontal position to provide accessibility and keep it out of harm’s way as much as possible.

To help resolve these claims and conjectures, I obtained a test vehicle (Pontiac Grand Prix GT, L36 engine) that had a complete factory exhaust system. This factory exhaust system included cast iron exhaust manifolds, catalytic converter in the down pipe, U-Bend, resonator and dual rear mufflers. The testing would involve replacement of the U-Bend with a straight pipe that would include a port for mounting the post-cat O2 sensor.

Prior to the installation, I collected performance data to baseline the car’s performance. After this data was collected,

I pulled down the down pipe with the U-Bend and replaced the bend with a straight pipe.
Once this modification was competed, I repeated the performance tests. The weather conditions for the day were warm and humid. During the testing with the U-Bend, the temperature was 83ºF, Humidity of 78% and a barometer of 30.02 “of Hg. The engine was thoroughly heat soaked and running 192ºF by the PCM’s Engine Coolant Temperature values. A total of 12 passes were used to collect all the performance data, with 4 replications for any one parameter. The results for replicated runs of each parameter were averaged together in the final analysis. The test conditions with the U-Bend removed were nearly identical, the temperature held at 83ºF, Humidity of 83% and a barometer of 29.95 “of Hg. The GT was running on 93 octanes, has 8,000 miles on the engine.

As can be seen in the charts below, the gains in peak HP and Torque were ~5.0 HP and 2.9 ft-lbs respectively. Also noteworthy is the torque and resulting HP curves were benefited throughout the entire RPM range. This data does not support many beliefs that removing the U-Bend will improve the upper RPM power & torque and sacrifice low end power & torque. A 1% increase in volumetric efficiency and total volumetric flow rate was also seen, which corroborates the increase in HP and Torque. Also interesting is the slight increase in overall spark advance through reduction in Knock Retard. About 2º of advance was picked up. This increase in spark advance and the slight improvement in volumetric air flow is what produce the small but significant performance gains. My conclusion is this modification has benefits all round.


While these results were obtained on a relatively stock GT, they are representative of the gains to be made on more significantly modified L36 engines as well as L67 engines. The more modified the L36, the more restrictive the U-Bend becomes and the more it constricts power increases. I would estimate proportional increase would be observed on an L67 supercharged engine. If this is the case, the U-Bend replacement on an L67 engine would extrapolate to a gain of 6 Peak HP and 3.5 ft-lbs Peak Torque. While these gains are not huge, they are significant and provide an edge over cars still running with the U-Bend in place.

– Article contributed by Jim Wierzbicki on August 15, 2002… Thanks Jim!