TWIN ROTREX ENGINE MODELLING
The GT Suite model seems to be extremely accurate in predicting the K series engine performance. This took man years to develop the model as each part of the motor's piping, combustion chamber, everywhere the inlet air goes, exhaust system, etc has to accurately be CAD for valid results. Accuratcy was surprisingly predicting torque within a few %. The model was created partly to understand the compound SC, and partly for camshaft developement to suit our Rotrex installations. We realized the need for the second intercooler doing the model but time has meant that we never built this in the model although it's on the car. Our model showed boost will climb until the 1st Rotrex starts to choke as it reaches it's 1100cfm flow limit which we believe will be around 750hp. (Rotrex were kind enough to supply comprehensive flow test data.) With our engine combination simulations predicts around 8500rpm. This effect will be similar to running an air restrictor on the air inlet the way many racing series are forced to, causing the BHP to stay fairly constant while torque gradually drops off. But as we are not revving past 8500rpm we will never see this.

This is a copy of the Poster from the final presentation, I thought you guys might be interested. The K25 engine simulation specification is from several years ago as we are now at high 600's WHP but otherwise the information is fairly current.

Rotrex have been extremely helpful and supplied these comments when we spoke via email about our twin series compond concept.

"When utilizing a twin charge setup where the air supply mechanisms (in this case Rotrex superchargers) are connected in serial, the first boosting device must have a larger capacity than the second unit to create boost pressure. In order to build up pressure over the first boosting device, you must be able to supply more air mass to the second boosting device, than what the second boosting device can supply itself. In your case the two boosting devices are identical which basically means that whatever boost the first unit can supply is somewhat eliminated by the second unit, because it can supply an equal amount of air mass as supplied by the first unit.
For instance, we correct our compressor maps to accommodate an inlet temperature at 15°C (SAE standards) and similar calculations are very important to do when you are designing parallel setups so you can calculate an equivalent air mass at a different temperature
I took my time doing some basic calculations on your system:
Air supply 1 unit: C38-91
Estimated inlet temperature: 25°C
Air mass flow (estimated from your power output): 0,5kg/s
Impeller speed: 90.000 rpm
Measured ∆t at the above specifications: 142°C
Outlet temperature: 25+142 = 167°C
Air supply 2 unit: C38-91
Calculated inlet temperature (as per above): 167°C (440°K)
Corrected impeller speed: 74,400rpm
So in a serial setup it is extremely important to cool down the air between the two boosting devices as the hot air will have an excessive volume dramatically reducing the potential performance. Rotrex believes that it would be much more ideal to use a twin charge setup with two similar chargers in parallel, because you would increase the total air mass supply (which generates hp) and increase the boost pressure (because you can supply more air mass than the motor can supply itself) without the need to deal with high temperatures. Parallel twin charging is the most common setup in high output supercharger installations (for instance Novitech, TTS R8, etc.). "

Our 4 cylinder relatively small motors for optimum performance need more midrange boost than a single Rotrex or parellel Rotrex can produce. Bring on the "Jaydee Twins".


JAYDEE TWIN ROTREX SERIES SUPERCHARGING

Can you please explain the difference in serial and parrallel setups?"[/I]

This is a boost log overlay of a single C38-91 at 594WHP compared to the series twins at 650WHP, however the twins are in constant TC and part of our TC is to control boost via the WG so boost is controlled at 21psi. Peak boost is pretty similar to the single but look at the midrange boost.....
The bottom graph is the single, the one above is the series twins.

I tried the parallel twins concept about 8 months ago out of curiosity. With parallel SC's each SC can flow more air than a single so you get about 1 or 2psi more boost at the higher rpm with the pressure curve following the single SC boost curve you have in this datalog. Overall acceleration split times were about 25% slower than the twins due to the lost midrange even though peak power was greater than the twins. It's the area under the torque curve that accelerates your car.

Our series twins is completely different. The first SC has a pressure ratio, say x1.5 at 5500rpm. This takes air pressure 101KPa and gives you 150KPa which goes into the second rotrex. so = (150 x 1.5) = 225kpa absolute or about 18psi instead of the 7psi you get with a single or 8psi you get with twins in parallel. By 6500rpm the twin C38-91 in series with twin intercoolers are making 24psi while boost continues to build as the rpm's climb.

At 5500rpm the car is accelerating 22% faster even though it's in traction control.(0.9G up to 1.1G)

It took almost 1hour to figure out how to do that overlay......

 

As a general rule air to air intercoolers are better than air to water apart from in boats (unlimited cold water) or drag racing when you can use ice. Air to water IC's have 2x temperature differences, one for each heat exchanger so inlet temps will be worse than a good air to air system and they are heavier. My intercooler gets hot as it's sucking out so much heat energy. 150degC to 44degC and 1000CFM is a massive amount of energy.


Bypass or Blow off Valves only dump the boost when the throttle closes as the inlet manifold pressure get below around 70kpa. Otherwise the sc air would have nowhere to go and the SC compressor would surge.
I fit and advocate using a WG on the inlet air pipes. This dumps boost if the inlet manifold boost is higher than the preset boost level. This enables boost reduction for traction control as most ECU's have WG control software built-in for turbocharger applications. Also a large SC with maximum gearing will create more top end boost than wanted in some applications, however maximum bottom/mid range is achieved by using an oversized compressor. Care must be made to ensure you do not surge the compressor.

You cannot run a Rotrex SC with a TB in front of the compressor wheel creating less than 50kpa pressure so a draw thru installation is not possible. This is due to the compressor seal which would leak the traction fluid if the absolute pressure was below 1/2 atmosphere. (I was advised of this by Rotrex's Lead engineer)

On my new car, the Jaydee Quantum I am running 2 TB. The driver controller cable operated TB is a 74mm S2 on the ported S2 Ultra inlet manifold. I have placed a Gen 3 DBW 90mm TB before the first Rotrex as this is a compound twin Rotrex installation. I measure the inlet pressure to the Rotrex and I have written special software in the ADL3 dash which keeps this pressure above 50Kpa. So on part throttle or overrun this first DBW TB runs the Rotrex in a partial vacuum.

My reasons to add this complexity were many.
1. Firstly this concept mostly eliminates the lost efficiency of driving the SC on part throttle, so we should get better fuel economy. During the emissions test the car averaged 8.7L/100km. This is an engine capable of 700WHP. Honda has produced some amazing castings for this motor!.
2. It completely eliminates the noise going thru the bypass valve as I have balanced the flow thru the SC with the engine air demand using 3D tables. Stealth supercharging!
3. Running the twin C38-91's in series I needed a 3D method to control SC surge.
4. With a conventional installation part throttle emissions are hard to control due to the bypass valve's stiction. As the inlet manifold pressure changes it opens at different points depending on whether the engine inlet manifold pressure is climbing or falling. This alters the engine's VE for the same load/rpm making light load transient tuning less than ideal. Our emissions here are tough so we needed a better solution.
5. Mapping the DBW TB to a knob on the dash enables the engine power level to be adjusted from a throttled NA engine up to max boost.

I have also integrated this into our existing TC software by providing open loop power control. Much like the boost by gear concept, it is much easier to closed loop TC if the engine power output is only slightly above the available grip.

Here is a photo showing both TB's.

You must mount the oil canister lower than the Rotrex so the oil gravity does not feed past the seal. The seal is designed to seal the pressure from the compressor side and by having the canister too high the oil weeps past into the housing. I have spoken with Rotrex about this and the seals have been upgraded for longer life. The newer seal is designed for low friction so the design will not be changed. This new seal made this effect more of an issue so they updated their mounting instructions. I had the same happen with a new -91 and lowered my canister and all was well after this was done. It's not a fault, just an installation placement limitation.

I have found the Rotrex 100% reliable with all my abuse over 7 years, however many people have had issues. I believe it is essential to follow all their installation requirements.

Below are explanations about the usual Rotrex supporting modifications.




MANUAL TENSIONER
I use Dayco belts as they are good quality and easy to buy in Australia. Rotrex specify 75-120N pr. rib and we run a 7 rib belt. I have found from experience 580N is needed to stop a C38-91 from slipping and use 620N which works well and although this is on the upper limit I have not had alternator or bearing issues.

If a longer belt is fitted with the auto- tensioner is in a lower tension positon to the right then the belt tension is between 310-350N.
A shorter belt brings the tensioner geometry into a better position to the LHS and the belt tension then rises to 440-480N. In any case this is well below your minimum tension specification. If you datalog pulley speed or Rotrex outlet temperature you will see the slip straight away. A boost trace is not very visible as the slip progressively increases. You must run a manual tensioner override for the Rotrex drive to be successful on any K. This is why my lower bracket has the maximum belt wrap with an extra idle pulley.

These are old photos before we made the new manual tensioners. The new tensioner retains the rubber cushion of the prototype I used for 7 years.

Have you got any more pictures of this manual tensioner mate?

This is a photo of the mark2 tensioner. The photo was taken with belt tension but the 2 lock nuts are undone as I was about to remove the belt. The rubber is a KTM dirt bike exhaust mount and has a 8mm bolt on one end and a female 8mm nut on the other. The bock is in compression, and as you increase belt tension it slightly distorts giving a visual perception of the belt tension. Over the last few years I have found this usefull as a "quick look check". As it's rubber it provides some shock isolation during shifts and cold starts as I noticed the original solid adjuster would chatter some times on the dyno. (On the road there was too much general noise to tell).

BYPASS, BLOW OFF VALVES
On the Atom twin rotrex engine I used a HKS blow off valve and a 38mm Tail WG. The WG works very well, but the HKS is slow to open and boost spikes at the SC on throttle closing into overrun over the 30psi maximum of the 3bar sensor. So I fitted an additional turbosmart bypass valve. With a centrifugal spring fitted it blew open so I mounted it backwards. This created internal stiction so it did not fix the issue as the turbo smart bypass was slow as well.

On the Jaydee Quantum twin rotrex engine I started with a synapse bypass valve and a synapse WG believing I was upgrading these components. The bypass works ok. But after 2 months wasted attempting to solve low speed tuning for emissions I realised that the bypass mechanical stiction creates a possible open or closed situation depending on your vacuum increasing or decreasing at the time.(hysteresis) So you can never tune this setup 100%. At the time I was battling the Skunk2 ultra manifold pressure signal to the ECU being crazy (their placement of their pressure ports causes crazy transient readings), ID1000's on time errors at hot idle and a fuel regulator plumbing error so I was really struggling to solve the combination of things for emissions tuning.


Just to frustrate me more the synapse WG uses pressure ports. I could not get the boost control to work. Even on the test bench the control range was poor.

I then reverted back to the Tail 38mm WG and it worked 100% straight away. To solve the bypass I fitted what all the eaton or twin screw SC use. It's a throttle butterfly bypass driven by a diaphram. No hysteresis so that tuning problem was solved instantly.

http://www.kennebell.net/KBWebsite/A...passvalves.htm

WASTEGATE
I use a 38mm Tail with a MAC 4 way valve. With a 4psi spring fitted this gives boost adjusment from 7psi to 26psi. You don't need a big WG as you are bleeding excess air that the engine is not using. A smaller WG gives the ECU a better opportunity to control the pressure evenly.

If you have a WG or bypass valve mount them backwards as they will be blowing open for sure. They blow open at 1/2 their rated pressure so a 15psi WG will start to bleed boost at 7psi unless you mount it backwards. By the way you don't need a big WG. 38mm is fine for twin rotrex. A 4psi spring mounted backwards yields about 7psi.


Make sure you setup the 4way MAC valve in the normal way.

Check the boost pressure gets to your WG with the wiring disconnected to the valve. The WG should then open at the default spring pressure plus a bit due to the back of the WG valve pressure blowing it shut. Usually a 4psi spring will give 6 to 7psia as minimum boost. Now connect your wiring and setup boost by gear. Sometimes you might need to create an RPM offset table between the boost aim and the boost you get. With this arrangement any issues with the ECU, MAC valve or wiring makes the boost the WG spring pressure.



EWP

Jeff, could you share if you've seen any differences with using a factory mechanical water pump and electric water pump? Have you found that an electric pump set up is as reliable on a street car, or would you shy away from something like that?

I could not stop overheating issues with the factory water pump at elevated power levels as a street car going from 50hp to 600hp. Partly this is due to my RWD layout and the time delay of the cold water arriving, and partly due to the time delay of the thermostat opening. During the several seconds for things to stabilize the water would boil in the head. I tried removing the thermostat, drilling wholes in it, changing the gearing on the crank pulley to no avail. I log the water temperature at the honda sensor, and at the head outlet pipe, and the water pressure. I would see the pressure spike and a short time later this extra hot water would arrive at the head outlet. Sometimes cylinder 4 would get hot and read on that sensor, sometimes not. This seemed to depended on the size of the heater water pipe and where it was routed but was hard to isolate.

The EWP solved all the on full throttle power issues but created other problems. As I had deleted honda's internal warming-up water loop this meant that the EWP spins very slowly when the motor's cold so there is a temperature gradient across the motor. A non driven warmup cycle is very important and this thermal gradient across the motor also effects emissions as the pistons are colder. I have added another sensor in the bottom radiator tank and 3D mapped the EWP speed with full throttle overrides (done in the Motec ADL3 dash). The head operating temperature is 88 to 92degs C under pretty much all conditions even when the water radiator outlet temperature can be 35deg C to 90degs. To make this system 100% I need to add an electric thermostat/bypass valve and add back Honda's internal warmup. Then it would be perfect. I have not found a suitable valve yet.

As far as reliability goes it's been 100%, at the track and on the dyno. Much of my testing has been done as a street car.



DRIVETRAIN
IMO the stock gearbox will break. It's just when. 400WHP is ok as the single Rotrex torque builds so smoothly but once you get into the 500's the fine teeth on the stock gears strip. Using GearX helicial gears have worked for me and are OEM quiet with the TC helping them last. Straight cut are a better option if you can stand the noise and there are many companies offering good gearsets. The clutch needs to be repaced to suit your power goal. I have used the FX700 for many years but it's grabby. The carbon clutch is like OEM but very expensive. A K20 with a single Rotrex the 4.7 FD is fine but for those k24's 4.0 to 4.3 is a better option.



CAMSHAFTS
I suggest you use K20 ITR or Z1, Z2, or Z3 cams. I use Z1 cams. The existing aftermarket NA. cams have too much overlap and the boost you loose will wipe out any power gain the cams might have created. The aftermarket SC cams we have modelled on GT Suite have not worked as well as stock Honda Z1 cams. Period. Don't get sucked into thinking they have done back to back testing on a Rotrex K series motor without before after dyno proof.

lifted from my Rotrex thread...

So when should we expect these cams of yours to be ready Jaydee? Do we have to add any supporting mods? How much power would you expect to gain a a c38-91 k20 at 14-15psi? Also, what are you expecting the characteristics of this cam to be like? More up top? More torque? More everywhere? And how high would I want to rev my k20 to make them work in they're 'sweet spot'

Thanks Jaydee, I'm sure the answers will be of much interest of lots of people

So when should we expect these cams of yours to be ready Jaydee? Although we have gone to extreme lengths to ensure smooth acceleration loadings on the ramps we need to do comprehensive testing both on the dyno and in cars. This will take some time. If everything goes to plan the first batch of 10 will be ready late 2014.


Do we have to add any supporting mods?
Below I have attached the poster which contains the Hertzian stress on the lobes(red and blue circles), and it's similar to DC 2.2's as a reference you would know about. This is a dynamic loading validation created from our valve train model. We will be using a TODA tensioner and type R tensioner for the testing as I consider this the minimum. The acceleration loads are lower and similar to stage1 cams like DIC, but with 12.9mm valve lift you will need to ensure your valve springs do not coil bind. So we will recommend aftermarket valve springs. We have built a CAD parametric model already and are working towards trying to create a cam chain loading simulation. This is very difficult and might not be possible with our current resources. We have one guy on this part time for the whole year but it's a really tough job.

How much power would you expect to gain a a c38-91 k20 at 14-15psi?
I have driven a single Rotrex K series since 2007 so the cams are designed with my driving and tuning back ground. With any SC K motor the mild cams are very restrictive and we are limited to a 3000rpm vTec due to the oil pump. So with this in mind we have totally reworked the small lobes to give performance gains while still maintaining a 900-1000rpm idle. Driveability will be massively better. The big lobes are a new concept for a SC K and we believe will gain power everywhere but mostly midrange. Any non built ITR or Z series K20 should be limited to 8500rpm IMO so we aimed for 5500-8800rpm. The model shows power still climbing at 9500rpm but it does not have the Rotrex drive speed limit or rod bolts that let go....

Clearly gains will be installation dependent. For the same boost the model is gaining 50BHP+ at 8500rpm which is huge. We also modeled smaller SC's and if the airflow hits the SC limit and boost drops say 3psi due to the extra flow then the power gain is 20BHP+. Let's wait for the dyno. Too many camshafts are on the market that don't live up too their claims.

As with our brackets, we are doing this development because we are passionate about supercharging the K motor with the best results possible. If they work as well as the GT Suite model and we end up selling some then then I will be pleased to become a vendor on K20a.org. If they are duds then I will be very grumpy! and wasted years on the project. Hat's off to Honda for the K motor.


All of this information I am providing with the aim to help forum members. If you don't agree with me then that's fine. I hope you enjoy your supercharged K series as much as I have no matter what form.
Cheers Jaydee.

 

Thanks for making this new thread, thanks for all of the information over all these years, thanks for being so kind and willing to actually share all of this information, thanks for answering questions, thanks for being AWESOME!!

I have your bracket, I am getting very close to actually putting it to use. There are people out there as passionate as you Jeff but very few as brilliant.

I am looming forward to more results, info and product development as always.

What are your thoughts on the 06-08 USDM TSX cams ?

 

Excellent thread... From a roots/twin-screw supporter

 

Awesome thread and so easy to read. Thanks mate

 

GREAT! Going to have to get with you soon on a bracket and some other things. I am actually looking for your upper bracket/tensioner setup because I have to keep AC here in houston! Also, do you NEED to delete the water pump if so? Thanks

 

Deleting your water pump depends on you installation and power level. For RWD and road cars at high power levels over 450WHP it's mandatory IMO. We can do a custom top bracket based on you SC size, inlet manifold and panel height clearance. It's more messing around but then you have exactly what you need.