Data Analysis: PV diagram, IMEP, and Experimental Results
 

What am I doing wrong, or, what am I not thinking of correctly?

I have simulated one cylinder of an engine that I have experimental data for. I know the engines actual power output is 1,925 lb-ft of torque per cylinder and with an efficiency of 30% gives 6,416 lb-ft of indicated torque per cylinder.

When I use Converge's "Calculate Statistics" function, I get the readout showing that the area under the curve (Y area, the work per cycle) is 20,231 Nm (14,922 ft-lbs) for the experimental data and 19,902 Nm (14,679) for the simulated data.

I also have IMEP data per cylinder - using the IMEP data I calculate 14,973 ft-lbs of torque which aligns nicely with the PV diagram area and the simulation.

Now for the problem, why is my power output so drastically high compared to actual output? Converge and the IMEP data is saying each cylinder is making around 850 hp per cylinder but I also the hp per cylinder output is about 110 which doesn't correspond to the 30% efficiency.

So, again, what am I doing wrong, or, what am I not thinking of?

Hi Greg, a few notes on your issue -

1: How many cycles are you running before you perform your analysis? We recommend no fewer than two engine cycles to wash out the initial conditions that may not be quite as accurate as you'd like.

2: Have you checked the fuel and air masses in the cylinder just before ignition? Do these seem reasonable? You may also want to check the equivalence ratio at this time as well.

3: If you have pressure traces from experiment have you compared the min/max/mean pressure traces from experiment with your simulation pressure trace?

This is not an exhaustive list but it may help us narrow down what's going on.

Thanks!

- John

Hey John - Thanks for replying.


I have the pressure trace for the experiment - I tuned the simulation model with it to within reasonable limits. The experimental PV diagram gives 20,231 Nm and the simulation gives 19,902 Nm - pretty close to actual. My confusion is that these numbers don't correspond to the known actual horsepower and torque of the engine unless the efficiency is dropped drastically. I know the efficiency of the engine is about 30% but the above values give around 18% efficiency (if I remember correctly).


The experimental data comes from 1000 cycles and the simulation was tuned using the 1000 cycle averages. The simulations, however, are only running 1 cycle. Because of the physical size of the engine (14" bore and 14.75" stroke) I am only running 1 cycle so that I'm not waiting literal weeks for a simulation to finish.



Fuel masses and equivalence ratios are right where they should be - in line with the experimental data.


Pressure traces line up nicely - min, max, mean are all within a couple percent or less.


The only thing different between the models is the spark timing. The sim SOI is retarded by a few degrees to account for the faster HRR of the chemical mechanism I am using (GRI 3.0).