Fast Running Model in GT-Power

INTRODUCTION TO FAST RUNNING MODEL (FRM):

Fast Running Engine Models are dynamic, fully-physical engine models that are designed specifically to run fast. In typical industrial applications where multiple cases have to be run, solving a detailed model will cost a lot of resources and time. While performing calibrations and trials, many a times, more than accuracy, the overall trend and the range of estimated values for results are more important. When simulation speed is of priority, GT-POWER engine models can be simplified into FRMs using a standard conversion process.

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.

In GT-Power FRMs can be created through two methods. One is using a FRM Convertor and the second is using a FRM Builder. In this project, both the methods are showcased. Both these tools can be found under Tools tab in GT-Power.

Creating a FRM Model using FRM Convertor [Tutorial 9 of Engine Performance in GT-Power]:

This is the initial model which we will convert to a fast running model. 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.

Before launching the FRM convertor, we need to run the initial model for results. After that, with the obtained results as reference, we proceed with the conversion.

During the conversion to the FRM, the solver will reduce the timestep. But, in order to maintain stability, we need to satisfy the CFL conditon. To check which parts limit the courant number, open the results in GT-Post and select RLT Contour Map -> Pipes/FlowSplits -> Flow Control -> "Minimum Courant Time Step (Explicit Circuits Only)". 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. The image of the same is attached below. It is clearly seen that the exhaust manifold has the least values of timestep and hence will restrict the timestep the most in the conversion process.

Now, we can head back to GT-ISE and launch the FRM Convertor tool. Either the baseline model can be specified for new conversion or an already existing FRM file can be loaded for editing.

During conversion, there will be a decrease in the performance and other parameters in the reduced model. There is an option to set the accuracy at the start of the FRM conversion that is needed to be maintained. Optional Tolerances may be specified for which additional comparison plots will be generated, and the user will be warned if a particular RLT becomes outside of a given tolerance.

STEP 1 : Exhaust Manifold

Once the above parameters are input, the Converter should now be at the Model Reduction page, as shown below. It is most efficient to start by simplifying the subsystem that is currently restricting the time step. Hence we start with the Exhaust manifold.

The exhaust manifold can be clicked on this window. Once this is done, depending on the required tradeoff between accuracy and speed, the suitable option can be selected- Simplify for accuracy (or) Simplify for speed. Here 'Simplify for accuracy' is clicked as the exhaust system needs to be modelled with accuracy to maintain CFL condition. The 'Apply Smart Calibration Settings'. This setting prefills the attributes for the design optimizer used later for model calibration steps.

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. Relatively, the port heat transfer is more significant then 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. These parameters will automatically be appended to the case setup as shown below.

The reduced model can be calibrated if necesary. The next window that pops up is shown below. For the cases for calibration, only the case with highest speed is done.

Next, the calibration is configured. In calibration, the response RLT is selected as turbine Inlet averaged mass flow temperature. 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. Go into the "Factors" tab to define the factor along with the Lower and Upper values for the Range, and finally the Resolution. The [HTM_Exhaust_Manifold_1_Step_1] parameter is automatically placed into the first column. A different parameter can be selected if desired using the value selector. This image is shown below:

After all this setup, the calibration is run for the step 1 reduction and can be compared for accuracy deviation. After this we proceed to step 2 reduction.

STEP 2 : Exhaust Pipe

Next, the exhaust pipes connecting the ports have to be reduced. Select Exhaust Pipes 1 from the Tagged Subsystems list. Click on "Simplify for Accuracy" and click "Yes" for the "Delete Part" dialog. Since the pipe at the outlet of the Turbine is short (70mm), this pipe will be combined with the Catalyst volumes into a single FlowSplit part. Further the resulting flowsplit and the flowsplit jsut after the turbine can be combined. Select the flowsplits and right click > Select combine flow volumes wizard. Once inside the Combine Flow Volumes wizard, keep "Combine flow volumes to Flowsplit" selected and click "Next" through the wizard until the last page, "Object Attribute Definition and Object Naming". Select "EP1_PDrop1-1" at the top of the dialog, then set the object name to "CAT". Now, we have obtained the reduced model as shown below:

Finally, the Case Setup has to be opened and the value of the reduced exhaust pipe has to be set [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. The in the FRM Convertor tool, we can proceed for calibration. Under configure calibration, we set the followign parameters. The calibration range for the exhaust pipe dia is set from 35mm to 45mm in factors.

Next, the calibration is run and if the obtained resuts are satisfactory, we can proceed to the next step.

Step 3: Intake Manifold

Select Intake Manifold 1 from the Subsystems list. Click on "Simplify for Accuracy". Click "Yes" on the Delete Part dialog. In the resulting part IM1_Plenum1-1, define the Boundary Data folder as shown below. Such values are given because, in boundary #2 to #5, backflow can occur in these boundaries.

To resolve the "Delete Part" warning, double-click on the EngineCrankTrain, go to the "RLT Norms" tab, and change the value of the attribute "Part Name for Manifold Volumetric Eff. Reference" to "IM1_Plenum1-1". Then, go into "Run Setup" -> "ConvergenceRLT" folder and change the first RLT Variable to "pav:IM1_Plenum1-1".

Now, the reduced model is ready for Calibration. Finally, open Case Setup and check [DXI_FRM] is 200 mm, and [HTM_Intake_Manifold_1_Step_3] is 1. Under design optimization of configure calibration, the range of factors for optimization is set as 0 - 1.1 as shown below.

Next, the calibration is run and if the obtained resuts are satisfactory, we can proceed to the next step.

Step 4: Compressor Outlet Pipes

Since the intercooler core (previously a bundle of pipes) has been replaced with a FlowSplit and "HeatExchangerConn" part, the "BP1_CAC1-2 HeatExchangerConn" part will be used to assure the intercooler outlet temperature is maintained. Now, Double-click on the "BP1_CAC1-2 HeatExchangerConn" part and set the Imposed Fluid Temperature to 300. Open the "BP1_CAC1-1" Flowsplit part, go to the Thermal tab and select Imposed Wall Temperature. Set the temperature to 300. Finally the missing control links have to be reconnected as shown.

Open Case Setup and set the value of [DIA_Boost_Pipes_1_Step_4] to 35 mm as an initial guess. The wastegate controller is maintaining engine BMEP, engine air flow is also effectively maintained since the Air-to-Fuel ratio is fixed. 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. Now, the reduced model is ready for calibration. Next, the calibration is run and if the obtained resuts are satisfactory, we can proceed to the next step.

Step 5: Intake Pipes

Select Intake Pipes 1 from the Subsystems list. Click on "Simplify for Accuracy". Go into Case Setup and set the value of [DIA_Intake_Pipes_1_Step_5] to 45 mm. With this, all the sub systems have been reduced. To avoid any errors related to model, go to the Data tab in the GT-ISE toolbar and click "Delete Unused Objects". The reduced model is shown below:

Now, calibration can be started. In the Calibration page, select Case 1 for calibration. Under "Configure calibration", change the Response RLT to the Compressor Massflow-Averaged Inlet Pressure. Set the lower and upper values of the range for DIA_Intake_Pipes_1_Step_5 at 40 and 50 mm, respectively. Next, the calibration is run.

At the moment, we have reduced the model to an extent where we can expect a Factor of Real-Time of around 5. Finally, we can do one more reduction in the cylinders of the model. This is called Cylinder Slaving.

Step 6: Cylinder Slaving

Right-click on any of the EngCylinder parts on the map and choose "Edit Parent Object". Go to the "Advanced" tab and for "Cylinder Slaving Option", select "slave-RT-full-v2017", then click OK. Now double-click on Cylinder-01, go to the "Advanced" tab, and set the Cylinder-01 override value of "Cylinder Slaving Option" to "master".

Next, for the full cylinder slaving to function, each Valve must be specified as either an Intake or Exhaust valve. This can be done by using the "Edit In Spreadsheet" option in the template ValveCamConn. Go to the Output folder and set "Valve Type for RLT Variable Calculations" to "intake", respectively "exhaust" as shown below.

Next, the calibration is run. Calibration will not be performed for this step, so simply click Next to get to the Results page to check the performance of the entire FRM. Shown below is the final FRM.

PERFORMANCE AND SIMULATION TIME COMPARISON:

The following tables show the performance and simulation times for the original model and the fast running model which is run at an engine speed of 5000 RPM. The performance of the original model is shown on the left and the FRM's on the right.

Inference:

• The obtained FRM has a deviation of ~0.19% in brake power, ~0.13% in Brake Torque and ~1.04% in BSFC. Performance wise, these values look good. The simulation time has to be examined to arrive at the conclusion.
• For both the model, the steady state period number is different. Hence to compare the two, we find which model has a better ratio of CPU time and Simulation time. Let us call this the time ratio.
• The original model has a time ratio of 14.51/0.768 = 18.89 and the FRM has a time ratio of 2.17/1.272 = 1.70.
• Hence we have mainatined the accuracy in the FRM as well as CPU time is lowered by the reduced model.

Next, FRM is studied using the FRM Builder which is present in the tools tab.

PART II: Studying FRM with FRM Builder:

In this section, a FRM circuit is built using FRM builder. The following are the specifications of the engine.

• Bore: 102 mm, Stroke: 115 mm, Compression Ratio: 17
• 6 cylinder CI engine
• Twin Scroll Turbine
• GT Controller

When the FRM builder is clicked upon, it asks for step by step inputs. The images are shown below. The specifications of the required engine model is input in these steps.

After all the parameters are input, the model of the engine is automatically generated by GT-Power. The image of the final model is shown below:

The model consists of different modules: FRM Intake, FRM Diesel Engine, Turbocharger module, EGR module, and FRM Exhaust. The GT contoller is present just below the engine module. It consists of Diesel injection controller, EGR Controller and Turbocharger wastegate controller. These modules are shown below:

1) Intake:

2) Engine:

3) Turbocharger:

This is a Twin Scroll turbine with Waste gate controller.

4) EGR Module

5) Exhaust

6) Controller:

For simplicity, the GT controller has clubbed all the signals of different controllers in two modules namely: Send signal and Receive Signal. Shown below is the list of all the signals which are sent from the respective devices.

Signals received in the Controller:

h

Signals Sent from controller:

CASE SETUP:

By default, GT creates 25 cases, where the RPM is varied gradually from 4000 RPM to 2000 RPM, BMEP is varied from 4 bar to 17 bar and target boost pressure is varied from 2 bar to 2.8 bar. A sample of the case setup is shown:

RESULTS AND PERFORMANCE:

The performance at the Rated speed (3500 RPM) and Intermediate speed (2500 RPM) at full load conditions are shown below. The target BMEP that is set as a parameter in the wastegate controller is 22 bar and the target boost pressure to be maintained in the compressor is 2.8 bar:

The engine gives a max power of ~330kW at 3500 RPM (Bottom Left) and ~990 Nm of torque at 2500 RPM (Bottom Right) full load condition. The other cases have part loads which

TIME CALCULATIONS:

The time taken for simulation at the rated speed and intermediate speed are shown below:

• The time ratio at the rated speed is 23.39/13.27 = 1.76 and at the intermediate speed is 3.41/1.776 = 1.92
• These values are very similar to that of the FRM Convertor case.

CONCLUSION:

GT provides an approach of creating a Fast Running Model (FRM) which are designed specifically to run fast. In this project, Fast Running Model was built using FRM Convertor and FRM Builder. Both the methods have a time ratio of ~2. FRM not only maintains accuracy, but also accelerates the simulation which is suitable for typical engine calibration purposes in the industry level.