goatofrafin

TLDR: A Rotrex C38-91 supercharger was selected for going north of 650bhp. Engine produced ~280bhp when N/A. Spinning the supercharger to ~91% (~10900 pulley rpm, 100mm pulley) of its max capability, produced only 350bhp / ~0.3bar boost instead of ~580bhp / 1.2bar boost. After a long process of eliminating any problems (e.g.: boost leakage, faulty supercharger, restricted intake, etc.), spinning the supercharger to max (or a bit more than max) pulley speed was decided. A spin of ~107% (~12900 pulley rpm, 85mm pulley) of Rotrex's max capability, produced 500bhp / ~0.9bar boost, instead of ~690bhp / ~1.6bar boost.
Two things seem extraordinady here; a) the fact that a ~18% increase in pulley rotation produced a ~43% increase in horsepower with tuning parameters being optimal in both cases and b) the percentage of the actual and estimated horsepower was 60% for the 100mm pulley and 72% for the 85mm pulley. The theory proposed is the fact that the traction drive of the Rotrex has variable efficiency which depends on both spinning speed and flow resistance (boost).

Full: Wanting to make a high horsepower 350Z without the hassle of turbocharging and the associated torque bursts associated with turbocharging, a supercharged solution was selected. Living in Europe, the Rotrex solution seemed the most reasonable one so using the process that Rotrex describes in its manual, an estimation of the horsepower and associated boost was made. Here, it is very important to describe the operation of the Rotrex with respect to how the impeller is being spun. In contrast to Roots-type superchargers where the pulley-to-rotor spinning is 1:1 (i.e.: the rotors spin as fast as the supercharger pulley), Rotrex superchargers have a different pulley-to-rotor spinning ratio. In the C38-91 supercharger, this ratio is 1:7.5 and is being handled by a traction drive that Rotrex is really proud of.
Two things are worth mentioning here: a) the supercharger efficiency map is in terms of impeller speed and NOT pulley speed and b) the traction drive features an efficiency metric where Rotrex says "MAX drive efficiency: 97%". The word MAX is INTENTIONALLY capitalized because it can be inferred that since there is a MAX, there must also be a MIN, an AVERAGE, etc. In other words, drive efficiency is variable and unfortunately, Rotrex does not state (correct me if I am wrong, plz), either the variance or the factors that influence this efficiency. Instead, in the supercharger model selection process that Rotrex depicts in the manuals provided, it is suggested to multiply the pulley-to-impeller ratio with the pulley rpm in order to calculate the impeller speed, without any reference to the drive efficiency. In my case, this detail caused me a lot of headache and delays because the actual numbers differed GREATLY compared to the estimated ones, causing me to suspect underlying problems related to the whole supercharging system (e.g.: boost leakage, faulty supercharger, restricted intake, etc.).



VQ35DE RevUp, naturally aspirated with 8.8:1 compression ratio, longtube headers, true dual 70mm exhaust with H-pipe

In particular, after building a forged internals engine with 8.8:1 compression, it was dyno-tested and produced ~280bhp, a result that was in-line with the power loss associated to the compression drop. Going for ~600bhp, a 100mm pulley was selected according to the Rotrex calculation process and efficiency map which would produce ~1.2 bars of boost. Dyno testing proved to be VERY disappointing, producing ~350bhp and ~0.3 bars of boost.


VQ35DE RevUp with Rotrex C38-91 (~11k pulley rpm), 8.8:1 compression, longtube headers, 70mm true dual exhaust with H-pipe

After LOTS of tests and making contact with various people with Rotrex applications, I had either no insight or complaints that Rotrex maps are not to be trusted. Some of them, also suggested to install a smaller pulley in order to rectify my problem. Although this suggestion partially rectified my problem as you will see later on, it did not explain the actual mechanism of how and why this could make the engine produce >650bhp when now it produced only 350bhp.
The decision to make the change to a 85mm pulley was made after talking to a guy with a TTS performance Rotrex C38-91 kit for the Honda S2000. He told me that he tracks the car a lot, he has a rev limiter which uses fuel cut (which means that revs are bouncing during limiting which is a BIG no-no for the Rotrex superchargers), he overspins the Rotrex by about 5% and that he has no problem whatsoever. The most interesting thing though was the fact that the aforementioned S2000 produced 450bhp at ~1bar of boost. Following the Rotrex calculation process and efficiency map for this car (1998cc, 9200rpm), it was estimated that for this application ~550bhp should be produced. This discrepancy led me to use the Rotrex efficiency map reversely and from there, calculate how fast the impeller should spin in order to have 1bar of boost and the associated air flow. The impeller speed for these numbers was 65k-70k rpm. Making the assumption that I won't go above 1 bar and that the impeller would spin at about 65k, I estimated that using the 85mm pulley, I would get about 470bhp and 0.7bars of boost.


Rotrex C38-91 map with estimations of airflow for 350Z and S2000

Gladly, my estimations were correct and a bit pessimistic as the engine got to 0.8bars of boost outputting 500bhp.


VQ35DE RevUp with Rotrex C38-91 (~13k pulley rpm), 8.8:1 compression, longtube headers, 70mm true dual exhaust with H-pipe

It is very interesting though to observe that for a ~18% increase of pulley speed, a ~43% increase in power was achieved, indicating the difference in traction drive efficiency, possibly related to both pulley rotation speed and boost.


Rotrex number crunching for the VQ35DE


Rotrex C38-91 map, comparison between estimated and real data for ~11k pulley rpm


Rotrex C38-91 map, comparison between estimated and real data for ~13k pulley rpm

Conclusion: A process for accurately estimating the hp produced by a Rotrex supercharger seems to be missing and overestimation of power seems to be very frequent when being based to the process that Rotrex specifies. My findings suggest that the sweet spot for the Rotrex is near 1 bar of boost and, depending on the volumetric characteristics of the engine, one could theoretically get from 250-650bhp. I personally have not found any indications of installations where the C38-91 exceeds 1bar of boost and this fact could explain why this supercharger seems to perform better at larger displacement and higher revving engines, going towards the 650bhp value.

I hope that anyone involved with the Rotrex superchargers might find this handy and save him a lot of time.