Showing results for 
Search instead for 
Did you mean: 

How to build a real-time compatible model of an automotive cooling system with Simcenter Amesim 17

Siemens Enthusiast Siemens Enthusiast
Siemens Enthusiast

The cooling circuit is a key subsystem in your car since it primarily ensures the reliability of the engine by limiting and regulating components’ temperature. The coolant temperature also greatly impacts the fuel consumption, the car performance as well as the passenger comfort.

That's why it is crucial to size it right the first time while considering all the subsystems interactions.


In a very early design stage, the characteristics and performance of each individual component are however unknown. A realistic behavior of the system needs to be represented with only a limited set of parameters.


Moreover, when the system design is almost frozen, it needs to be validated together with the associated controller, which requires for the simulation to comply with real-time constraints.


To tackle both challenges, in Simcenter Amesim 17 we have introduced functional cooling system components that will be presented hereafter.


System schematic


The following figure represents a typical automotive cooling system:Cooling system schematic.png


The coolant flows in the circuit thanks to a pump which is entrained by the engine.

  • After the pump, the coolant goes through the oil cooler before entering the engine to cool down different parts of the block and the cylinder head.
  • The thermostat then regulates the coolant temperature by splitting the flow into the radiator and by-pass branches.
  • Note that the heater core is not represented in the example.


Simcenter Amesim model


 Let’s now see how traditionally we represent this system in Simcenter Amesim.


Simcenter Amesim cooling system model.pngSimcenter Amesim cooling system model

 The model consists of 3 main subsystems:

  • The cooling loop
  • The oil circuit
  • The engine thermal masses

This model enables us to evaluate the oil, engine and coolant temperature evolution during a transient run. As an example, we simulate a warm-up during a NEDC cycle starting from 25°C. The evolution of the different temperatures is the following:


Temperature evolutions throughout a NEDC cycle.pngTemperature evolutions throughout a NEDC cycle

In terms of performance, this reference model is quick to simulate since the CPU time using a variable-step solver is about 14 seconds for 1170 seconds simulated, thus giving a simulation time 80 times faster than real time.


However, the succession of capacitive and resistive elements for the hydraulics generates very small time constants that requires a time step of around 1 nanosecond to ensure the fixed-step solver stability, thus preventing more or less a usage of the model for real-time purposes.

 Cooling system functional components.png

To ensure this real-time compatibility we can work in 2 different ways: we can either apply different reduction techniques to the original model or rely on cooling system functional components.


Let’s take a look at the second option and see how the new cooling loop, oil circuit and oil cooler functional components help to address real-time applications.



Results and comparison


The different subsystems of the reference model are now replaced with these components while keeping the same boundary conditions.


The new model with 3 functional components from the Cooling System library.pngThe new model with 3 functional components from the Cooling System library

  How the subsystems' performance is now defined?


  • From a thermal point of view, the heat exchanges need the information on flow rates and inlet temperatures for a proper calculation, the pressure being of second order and not used consequently
  • The cooling subsystem is hence defined by 2 curves that can be easily generated from the detailed model:
    • Flow rate vs. pump rotary speed for thermostat closed
    • Flow rate vs. pump rotary speed for thermostat open

      Generation of cooling subsystem performance curve.pngGeneration of the cooling subsystem performance curve

From the detailed model, we have, on one hand, the coolant pump characteristics for different rotary speeds, on the other hand, we can generate the Q vs. dP curve characteristic of the coolant circuit for the 2 configurations: the closed and open thermostat.


The intersections at different rpms of pump and circuit characteristic curves enable us to build the 2 curves needed for the functional cooling subsystem component. Note that the interpolation between these 2 curves is done automatically for intermediate thermostat positions.


Regarding the oil circuit, the principle is quite similar: the 2 curves needed to be generated here for 2 extreme oil viscosities.


For the rest, fluid volumes, initial temperatures and radiator performance data are essentially the only required data, thus limiting the number of input parameters:

  • 13 parameters for the coolant
  • 9 for the oil

 The oil cooler is using the same data as the original one.


Using the same scenario as the one used for the reference model, we can compare both results:


Temperature comparison.pngTemperature comparison

Coolant comparison.pngCoolant comparison

Temperatures are very similar as well as coolant flow rates.

In terms of CPU time, the model using the functional components is about 10 times faster.


CPU comparison.pngCPU time comparison

Moreover, this model is now real-time compliant. Switching to a fixed-step solver with a default step of 10ms, we obtain the same results:


Variable-step and fixed step comparison.pngVariable-step / fixed-step comparison

This real-time compliance is illustrated by a plot of the ratio between the simulation elapsed time over the simulated time. The maximum value of 1e-3 shown below (i.e. simulation is 1000 times faster than real time) confirms this model capability.


elapsed vs simulated time.pngElapsed time / simulated time


To sum up, Simcenter Amesim enables you to study and optimize the vehicle cooling system at different design stages  - from predesign up to controls validation. From a detailed model, you will easily set up and run with a very limited number of parameters a functional, realistic and real-time-ready model.


 Want to see how it works? Watch this video: