4.5. Troubleshooting

The following are some of the common causes for engine simulation errors.

  • Check the validity of all the boundary conditions. Sometimes, the boundary conditions that are meant to be inlets or outlets may be set to the wall type by mistake.

  • The Default Initialization is by default the primary cylinder (or reference cylinder) with region ID 1. In a multi-cylinder case, sometimes one more region called cyl 1 may have been added under Initial Conditions for the primary cylinder. The result is that this cyl 1 region will be ignored. In the run folder, you'll see .csv files reported for different cylinder regions, for example, thermo_1.csv, thermo_2.csv, thermo_3.csv, where thermo_2.csv is for the ignored second region and it contains no content. This issue is benign, but it is better to delete the redundant region cyl 1.

Additionally, troubleshooting instructions for the following common issues are covered in this section:

 

4.5.1. No Combustion or Misfire in Engine

If there is no combustion happening in the engine simulation, or that the engine misfires:

  1. Check if chemistry is activated or not.

  2. Check if the working fluid contains oxygen.

  3. Check if fuel is introduced into the combustion chamber in a correct manner. Specifically:

    • If this is a Direct-Injection (DI) engine, check the amount of liquid fuel injected in the Spray Models ("Total Injected Mass").

    • If this is a Spark-Ignition (SI) engine with premixed fuel-air mixture from the intake, check the Composition of the inlet mixture.

    • If gaseous fuel is injected using an inlet boundary condition, check the Accumulated Mass Flow on that boundary in open_boundary_flow.csv to verify the amount of injected fuel.

    In any case, it is always good practice to check the averaged equivalence ratio in the cylinder prior to the combustion event. This parameter is reported as In-cylinder (or Region) Averaged Progress Equivalence Ratio in thermo.csv. Combustion won't occur if the equivalence ratio is beyond the ignitable range.

  4. For an SI engine, check if the settings of spark, flame propagation model, and mesh controls for the spark and flame propagation region are based on the best practice. These are explained in Spark Model, Flame Speed Model for SI Engine Simulations, and Mesh Refinement.

  5. For an SI engine, if the flame completely quenches after the spark ignition, it could also lead to misfire. Refer to Flame Completely Quenches After the Switch from Ignition Kernel to G-Equation to handle such situations.

 

4.5.2. Mass in Cylinder Not Constant When Valves Are Closed

  1. Check if the valves are tightly seated in the surface mesh.

  2. Check if the valve gap region is not sufficiently refined such that the port regions and cylinder are connected through large cells.

 

4.5.3. Mismatch of Compression Pressure

Below are a few tips on diagnosing mismatch in pressure during the compression stroke:

  1. Make sure the pressure agrees reasonably well with experiment at IVC. A small discrepancy at this point can be amplified through the compression process. If the engine geometry contains a large intake manifold, the initial condition inside it may affect the IVC pressure too. To eliminate the uncertainty of initial condition, try running multiple engine cycles and see whether the results in the intake stroke change.

  2. Incorrect compression ratio is often the main reason for mismatch of compression pressure. The compression ratio can be affected by the squish height, stroke, and whether the crevice volume is included in the geometry.

  3. Since the compression stroke can be typically approximated as an isentropic compression process, the pressure at the end of compression is mainly affected by the compression ratio:

    in which and are the IVC pressure and volume, and are the end of compression pressure and volume, and is the ratio . The trapped mass and IVC temperature only have secondary impact on the pressure curve.

 

4.5.4. Mismatch of Expansion Pressure

When there is reasonable agreement in compression pressure between simulation and experiment and a mismatch in expansion pressure after combustion, tweaking the combustion model may appear to be a logical attempt to improve the result. However, a large mismatch in expansion pressure is often caused by incorrect total gas mixture mass or fuel mass trapped inside the cylinder at IVC. Adjusting the combustion model typically has very limited effect on expansion pressure. Try the following steps to troubleshoot such a problem:

  1. A mismatch in expansion pressure typically indicates that either the total trapped mass or the trapped fuel mass is wrong. Larger mass value leads to overprediction of expansion pressure.

  2. The trapped mass can be affected by a few factors:

    1. Check the boundary pressure at the inlet. If the boundary pressure is too high or too low, it is typical to see a mismatch in the pressure trace during the intake stroke.

    2. The internal residual of combusted gas from the previous engine cycle can affect the temperature of the trapped mixture. The residual gas typically has a higher temperature than the fresh charge, and therefore a larger residual fraction can increase the intake charge temperature and reduce the trapped mass.

    3. Check the valve lift profiles for correctness. Incorrect valve lift profiles may cause a large amount of blow-by flow – fresh fuel and air flowing into exhaust manifold without being combusted.

  3. If the experiment can provide measured mass flow rate of air and fuel (typically with units of kg/hr), these data can be used to estimate the trapped fresh charge mass per engine cycle and doublecheck the air/fuel ratio. For example, if the measured air mass flow rate is M kg/hr at a given rotation speed n [RPM] (per cylinder for a multi-cylinder engine), the trapped air per engine cycle can be estimated as kg for four-stroke engines and kg for two-stroke engines.

 

4.5.5. Flame Completely Quenches After the Switch from Ignition Kernel to G-Equation

In certain spark-ignition cases, the flame completely quenches right after the switch from ignition kernel to G-equation. This is indicated by a message in Forte.log that "Flame has quenched completely in cylinder 1", occuring in the time step after the switch from ignition kernel to G-equation occurs.

When the flame suddenly quenches immediately after the ignition kernel to G-equation switch, it typically indicates that the cylinder region is connected to the intake and/or exhaust port regions through the "seated" valves. In general, the flame propagation (G-equation) model is disabled when an engine cylinder is not a closed system.

Note that even if the valves are supposed to be seated, and "ALL PORTS ARE JUST CLOSED" is reported in Forte.log, the valve gap might still cause the cylinder to be connected to the intake and/or exhaust regions. This undesired inter-region connection is due to large valve gaps in the original surface mesh (when the valves are at a closed location). And the connection could happen on both the intake port side and exhaust port side.

To check if the valves are seated properly or not:

  1. At the moment when the valve is closed, use cut planes to check the actual gap size between the valve surface and its seat.

  2. Use the Valve Lift utility to reduce large gaps at the valve-closed location by translating the valves.

  3. Check the surface mesh on the valve surface. If the surface mesh resolution is too coarse, it might cause an uneven surface and affect the completeness of the sealing. In this case, refining surface mesh resolution is recommended.

  4. One can also check whether the orientations of the valves are properly aligned with their seats.

Although it is possible to fix the flame quenching problem by modifying the option that controls when to turn off the G-equation model, it is better to fix the root cause, such as the large valve gaps. This is because sometimes the large gaps can also cause region ID marking problems in meshing.

Other possible causes include:

  • The combination of very large energy deposition rate (for example, 400 J/s) and very small initial kernel radius (for example, 0.1 mm) results in an expansion speed that is too large. In fact, the plasma velocity is proportional to the deposition rate and inversely proportional to the kernel radius, hence the fast expansion.

  • Refinement level: Make sure that when the transition to the G-equation happens at the specified kernel radius, there are ~3 layers of cells within that region.

 

4.5.6. Error: "Spark plug # 1 must be put inside a Cylinder/Primary region or a Sub-Chamber region"

Assuming the location of the spark has been double-checked, this error must be caused by incorrect region IDs in the mesh. What typically happens is that the cylinder/sub-chamber region containing the spark plug is not given the correct region ID and initial conditions due to some topology issues in the surface mesh. Intake and/or exhaust valves not being tightly seated is the most frequently encountered cause for such an error. To confirm, you can do a test run by turning off the spark ignition model and selecting region ID as a spatially resolved output variable. Turning off the spark ignition model will allow this error message to be bypassed and hence allow the initial spatially resolved solution to be saved. You can then check the region IDs and initial conditions in the initial solution using EnSight. If the valves indeed contain non-trivial gaps in the surface mesh, you can use the Check Valves utility within the Utility ribbon group to seat the valves to a specified tolerance.