Calibrating the Reference Engine Model

EngMod2T requires a large number of input values to define an engine. Many of these values were measured directly from the reference engine, while others were taken from nominal dimensions listed in the manufacturer's homologation report. Most of the basic engine geometry is understood to be reasonably accurate. This includes values such as bore, stroke, port timing, and general engine layout. Some values still have lower confidence, like the squish band ratio, crankcase compression ratio, and some internal exhaust system dimensions.

One of the biggest modeling challenges so far has been the exhaust pipe. After using the physical exhaust pipe dimensions in EngMod2T, the simulation repeatedly produced an unrealistic power curve. The main issue was a strange hump around 9,500 RPM. After reviewing the output data, it became clear that the delivery ratio was rising quickly in that RPM range while trapping efficiency stayed mostly flat.

1. Delivery ratio is a measure of how much fresh charge is supplied to the engine compared to the reference mass that would fill the swept volume.
2. Trapping efficiency is a measure of how much of the supplied fresh charge remains trapped in the cylinder by the time the exhaust port closes.

Because trapping efficiency was not rising along with the delivery ratio, I initially assumed the problem might be in the intake model. My initial impression was that the engine appeared to be inhaling more mixture, but not necessarily trapping more of it. This led me to spend a lot of time adjusting the intake system by running various parametric sweeps.

None of those changes removed the hump. The major change happened when I shifted focus to the exhaust pipe. I originally planned to alter the exhaust duct because the peak power was too low, and I thought the issue may have been related to the stinger or outlet restriction in the pipe. This turned out to be partly true, but the real discovery was that the exhaust pipe had a much stronger effect on the entire power curve than expected. After creating a quick and simplified tuned pipe, the model not only gained power, but the strange 9,500 RPM hump disappeared completely, and the trend looks much more realistic.

This was not the result I originally expected, but it clearly showed how sensitive 2-stroke engines are to the exhaust system. The current exhaust model does not include the same filter, baffling, or silencing geometries as the real physical pipe or the original modeled pipe. Because of this, the output performance data may be more optimistic than reality but is a good start for now.

After solving the major trend issue, the next step was tuning the model across the rest of the program. This included adjusting variables to bring the power and torque trend closer to the available reference curve. A major reminder from this work is that the goal should not be to blindly chase peak power values, especially with 2-stroke engines. These engines are already known for having a peaky power curve, and a narrow or hard-to-manage power band is not the goal of this project. A method I will be using in the next phase of this project when designing the REV system is to use the concept developed by Frits Overmars called the power trend, which is defined as the highest rpm of a power curve divided by its lowest rpm. When compared to several other engine types and different uses, this can give a better estimate of the quality of the powerband. The current input and output data looks to be good enough to move on to the next step. When real dyno data is available, some variables will be tightened up more.