FRM modelling using FRM converter approach and FRM builder approach in GT suite
Objective 1: To build Fast running engine model (FRM) using FRM Converter approach
Fast running model(FRM):
Fast Running Engine Models are dynamic, fully-physical engine models that are designed specifically to run fast. While high-fidelity engine models are commonplace in the engine performance department, they are often too slow running to incorporate into system level models where long transient events may be simulated, or where the simulation model must respond faster than real-time, such as for HiL (Hardware-in-the-Loop) simulations. When simulation speed is of priority, high-fidelity GT-POWER engine models can be simplified into FRMs using a standard conversion process. It is up to the user how far to simplify the model, depending on the accuracy-vs.-runtime requirement.
FRMs are able to achieve such fast run times by two means:
(1) increasing the simulation time step size, and
(2) decreasing the number of calculations per time step.
These two means are often accomplished at the same time by lumping various flow volumes together which will reduce the total number of subvolumes and also allow for a larger time step size by increasing the effective subvolume length (which is responsible for the time step size). In addition to simplifying and combining flow volumes, there are additional solver options that can reduce the number of calculations per time step and, thus, further decrease the run time without changing the time step size.
Model SI_4Cyl_GDI_turbo:
This model is a 4-cylinder 2.0 liter turbocharged Gasoline Direct Injection (GDI) engine. The turbine wastegate is being dynamically controlled in order to target BMEP. A full-load speed sweep is being simulated, from 5000 RPM down to 1000 RPM. A semi-predictive charge air cooler is included such that the outlet temperature is predicted as a function of inlet gas temperature and mass flow rate, coolant temperature, and charge air cooler effectiveness.
In order to convert high fidelity model to FRM model, first minimum courant time step has to be determined in all parts like shown below,
This will give an overview of all parts and the associated courant number. The lower the value, the more likely the part will limit the timestep. This enables the user to determine multiple parts that will limit the timestep.FRM tags are used to parts to their respective sub systems.
STEP1: Exhaust manifold
It is now time to start simplifying the first subsystem of the model. It is most efficient to start by simplifying the subsystem that is currently restricting the time step. For high-fidelity models, this is almost always the exhaust manifold where the highest gas velocities occur, which in turn restricts the time step.
If accuracy is of higher importance than run-time, choose the "Simplify for Accuracy" option. This will combine the Exhaust Ports/Runners together, and left separate from the manifold volumes. If a real-time capable model is required choose "Simplify for Speed". In this case the Ports, Runners, and Manifold volumes will be combined together at once.
The Discretization Length in the pipe parts will be overwritten with an automatically created parameter called [dxe_FRM]. Its value will be set to 300mm. Port Heat Transfer dominates vs. Runner Heat Transfer, therefore a parameter for the heat transfer multiplier called [HTM_Exhaust_Manifold_1_Step_1] will be automatically created in all port/runner pipes Thermal tab. This parameter will be used to calibrate heat transfer in the combined volume to assure the Turbine Inlet Temperature is maintained.
For calibration, Target is set as massflow averaged inlet temperature and calibration is done for case 1.The Target Value for the calibration is pre-defined as a parameter. The values for this parameter are taken automatically from the results of the baseline, high-fidelity model. This will assure that the calibration will try to match the results from baseline model.
After calibration, model is rerun for all cases.
STEP 2: Exhaust pipes
For the second step of model simplification, the exhaust pipes downstream of the turbine will be simplified.
Set the value of [DIA_Exhaust_Pipes_1_Step_2] to 40 mm. This is only an initial guess, and will be used to calibrate the Turbine Outlet Pressure. Calibration is done wrt to Massflow averaged outlet pressure.
As shown above the upper limit and lower limit of exhaust pipe diameter is set to 45 and 35 mm respectively. This variable will depend on the turbine outlet pressure.
The above graphs are extracted after step 2 conversion. run time vs real time plots can be seen above. We can observe that for high fidelity model, the factor of run time vs real time is around 20 but after step 2 conversion the factor has been reduced to around 10 for same rpm.We can also observe that average time step size has also increased than original model.
STEP 3: INTAKE MANIFOLD
Now the boundary data for part IM1_plenum1-1 as shown below,
The characteristic length is defined as the distance that the flow entering the boundary will travel before crossing another boundary or impacting a surface. Boundary #1 represents the boundary between the flowsplit and IM1_Pipe1-1. The fluid entering the flowsplit in this direction can travel all the way across the plenum before impacting a surface, so the characteristic length is equal to the length of the plenum. The other boundaries are adjacent to the runners, where backflow from the cylinder would travel across the diameter of the plenum before impacting the plenum wall, which is why they are given a characteristic length of 70mm.
The expansion diameter defines the maximum diameter that the flow may expand to after entering a boundary (or contract from when leaving a boundary). For Boundary #1, it is simply the diameter of the plenum because the fluid expands from IM1_Pipe1-1 into the plenum in a straight direction. The expansion diameters for the other boundaries are equal to the plenum diameter multiplied by 1.4. This is because as the fluid backflows into the plenum, it encounters a wider cross section than the plenum diameter to expand into.
For step 3, calibration is done wrt to compressor massflow averaged outlet pressure.
From above figure, we can determine the percentage of error in bmep of engine of FRM model wrt to original model. We can observe the percentage of error is very less.
from above graphs we can observe , further decrease in factor of real time and also increasing in average timestep size.
STEP 4: Compressor outlet pipes
Reconnect the Control links, sensing Upstream Temperature from "BP1_Pipe1-1" Pipe and Upstream Mass Flow Rate from the "6” OrificeConn upstream of the "BP1_CAC1-1" Flowsplit, and sending the actuator output to both "BP1_CAC1-1" Flowsplit (actuating Wall Temperature) and to the HeatExchangerConn (actuating Imposed Fluid Temperature).
Set the value of [DIA_Boost_Pipes_1_Step_4] to 35 mm as an initial guess and calibration is done wrt to Compressore massflow average outlet pressure. Since the wastegate controller is maintaining engine BMEP, engine air flow is also effectively maintained since the Air-to-Fuel ratio is fixed. Therefore, as the effective diameter of the Intercooler changes, the resulting compressor outlet pressure will change so that engine air flow is maintained (via the wastegate controller). Therefore, the Intercooler Diameter will be adjusted in order to maintain compressor outlet pressure. Set the lower and upper values of the range for DIA_Boost_Pipes_1_Step_4 at 30 and 40 mm, respectively.
from above graphs we can observe , further decrease in factor of real time and also increasing in average timestep size.
STEP 5: Intake pipes
Set the value of [DIA_Intake_Pipes_1_Step_5] to 45 mm (this is just an initial guess).Since every subsystem has now been simplified, we will assure that there are no remaining pipes with short discretization lengths.View the Discretization Length column to assure that only [DXI_FRM] and [DXE_FRM] are used.
Calibration of diameter of intake pipes is done wrt to compressor massflow avergaed inlet pressure.Set the lower and upper values of the range for DIA_Intake_Pipes_1_Step_5 at 40 and 50 mm, respectively.
From above table we can observe that part EM1_Collector1-1 is restricting the time step 96% of the time. Therefore, the Exhaust Manifold will be simplified further in order to achieve larger time steps and, thus, a faster runtime.
STEP 6: Additional changes.
In this stage the exhaust manifold is further simplified for speed.
In this step, we can observe an abrupt or rapid decrease in factor of real time wrt to original model and also averge timestep size is higher than original model.
STEP 7: Cylinder slaving:
Cylinder slaving is process in which the equations and wave dynamics is solved for one master cylinder and other cylinders results are interpreted from master instead of solving separately. This process will significantly help in reducing the solution time in case of multi cylinder models. Therfore one cylinder is defined as master cylinder and others are defined as slave cylinders.
For the full cylinder slaving to function, each Valve must be specified as either an Intake or Exhaust valve. The intake ones start with "IM1_PortRun” and the exhaust ones start with "EM1_PostRun".
after step 7, we can observe rapid decrease in factor of real time varible and can be obsrved close to 1 which is desired. Significant increase in average timestep size can be observed.
OBJECTIVE 2: Build FRM model with following configurations using FRM builder approach in GT suite
Configurations:
6 Cylinder CI engine
Bore x Stroke: 102 x 115
Compression ratio 17
Twin scroll turbine
Model:
Turbo:
FRM_Diesel:
Results:
From above results, we can observe tha factor of real time value is around 5, which is acceptable considering the accuracy of the solution.