Engine designers are being pressured to design more efficient engines to decrease emissions and fuel consumption. To achieve these design goals, the trend is pointing toward engine downsizing and having higher compression ratios in order to increase power output. Increasing the power output, in turn, creates higher demand on thermal heat management. The high thermal loads will generate thermal stresses which could lead to shorter engine life, or failure. In addition to meeting new, challenging government regulations, engineers are working on shorter development cycles to more quickly introduce superior products to the market.
Combustion engines generate heat by fuel combustion and by friction of moving engine parts in relative motion. A key phase of engine and vehicle development is designing the various technologies to manage and dissipate this heat, collectively termed heat rejection. Well-considered engineering of heat rejection is essential for peak engine performance at high operating temperatures, and also for optimizing engine behavior during warm-up. Efficiency of combustion engines is significantly lower at cold start than at steady-state operating temperatures, so an important objective of engine development is to reduce energy losses by ensuring that systems and components reach their intended operating temperature range as rapidly as possible.
Conventionally, predicting heat rejection is done using physical engine prototypes and special-purpose heat rejection test benches. The problem is that once a physical prototype has been built, it is expensive to make changes and optimize the engine design. Ideally, issues with heat rejection need to be identified early in the design cycle, before first hardware has been built.
A new method to predict heat rejection using software-based virtual testing, backed up by data from physical tests, has been developed by InDesA (Integrated Design Analysis). InDesA is a consulting and engineering services firm specializing in simulation and analysis of complex fluid flow and heat transfer systems based north of Munich, Germany.
InDesA’s approach is built around a detailed STAR-CCM+® engine model embedded in a virtual underhood environment. Combustion and exhaust temperatures are derived from 1D engine process simulation using Gamma Technologies’ GT-SUITE software, while friction heat is measured in physical testing.
Figure 1: An early-development-stage physical engine prototype
Image courtesy of BMW
Shortcomings of physical prototype testing for heat rejection
The conventional approach to measuring heat rejection has been to use an early-development-stage physical engine prototype, as shown in Figure 1. To begin, the engine is instrumented with pressure indicators in the cylinder head. This gives a measure called Indicated Mean Effective Pressure (IMEP) and another called Brake Mean Effective Pressure (BMEP) representing the torque at the flywheel. The two measures are used to derive the friction for the complete engine, called Friction Mean Effective Pressure (FMEP). Adding to the complexity and cost is that a specially cast and equipped cylinder head must be fitted to the engine to obtain these measurements.
The prototype engine is also instrumented with temperature sensors (thermocouples) to monitor engine temperatures. This is important in order not to damage the engine as it is exercised on the test bench. A related complication is the special test bench required: Heat rejection testing requires the bench to have conditioning appliances for engine oil and coolant, which are not present on ordinary engine test benches.
Early-stage prototype engines typically have built-in performance limits to safeguard the engine during testing. These can include restrictions on speed and torque, and an enrichment of combustion mixall designed to protect the engine by keeping temperatures lower than that of a series engine, with significant impact on heat rejection. The dilemma is that heat rejection needs to be understood early in development, but the engine’s combustion and exhaust characteristics are often not mature enough at this stage for accurate evaluation of heat rejection based on physical testing.
Supplementing physical with virtual testing
To remedy these shortcomings, InDesA developed a new approach that uses standard physical test procedures to calibrate simulation models based on 1D and 3D representation of fluid flow and heat transfer, outlined in Figure 2. The simulation is then used to obtain complementary information that overcomes the uncertainties and lack of accuracy caused by the immaturity of early-stage physical engine builds
Figure 2: InDesA’s approach to engine and vehicle underhood thermal simulation
Physical testing provides comprehensive information about the engine, e.g. combustion pressure, temperatures, friction (tear-down measurements) and fuel consumption, whereas thermal maps of integrated heat exchangers should be tested on separate virtual or physical test benches. These measurements are used to populate and calibrate various 1D engine models in GT-SUITE. The resulting 1D simulations yield predictions for fuel consumption, basic engine operating parameters, mass flow rates, pressure and temperature in the air induction, and exhaust systems and in the coolant and lubrication circuit. This 1D simulation output then provides the boundary conditions for a STAR-CCM+ model of the engine as well as the underhood and full-vehicle environment. This is used to calculate heat exchange inside the engine, through the exhaust and cooling systems, and finally heat rejection to the ambient environment.
Replacing physical test cells with virtual underhood environments
Engine heat rejection is controlled through thermal management technologies embedded in the engine thermal design, exhaust, cooling and lubrication system design, and underhood environment. To virtualize all of these, InDesA used STAR-CCM+ to model a virtual engine. The model was designed to demonstrate thermal simulation techniques with options for different thermal management technologies: split cooling, water-cooled exhaust manifold, engine oil cooler, and thermal encapsulation. The virtual engine is “brought to life” with 1D simulation models from GT-SUITE for engine performance, combustion with air intake and exhaust, heat transfer to the engine structure, lubrication and coolant circuit (shown in Figure 3).
Figure 3: The virtual engine is “brought to life” with data from GT-SUITE
Then, to improve on the traditional physical test cell, InDesA developed a virtual car with underhood and full-vehicle environment using STAR-CCM+ that brings together CFD and Conjugate Heat Transfer (CHT) models. InDesA’s virtual concept car is named “Pandora” with the intention that no automotive OEM would take this name for a production car model given the negative connotations of Pandora’s box. On the other hand, it is noted that in Greek mythology Pandora was a most beautiful woman created by gods, and hence the name was considered to be suitable.
Figure 4: Pandora is a virtual concept car with underhood and full-vehicle environment modeled with STAR-CCM+ to simulate engine thermal performance with heat transfer to the ambient environment.
The Pandora virtual car model is used to simulate engine thermal performance with heat transfer to the engine compartment and ambient environment, demonstrated in Figure 4. The Pandora model includes the engine and a simplified engine compartment, with air induction, exhaust and coolant systems and front-end heat exchanger module. The goal is to provide greater fidelity together with a wider range of operating conditions than a physical test cell operated typically at decent test cell temperatures, see Figure 5. Air flow through the engine compartment is modeled in accordance with vehicle speed and cooling fan performance. Air temperature is also modeled in accordance with heat release from the radiator module in front of the engine. The 1D vehicle model, which includes a drive train, road model and ambient environment, is used for transient simulation in GT-SUITE. Hence boundary conditions for any driving cycle from warm-up to race-track operation can be provided.
Figure 5: Engine installation of InDesA’s virtual concept car “Pandora”
Figure 6: Heat flux vectors in engine structure
The engine model is detailed down to a level suitable for thermal stress analysis (see Figure 6), with heat flux from combustion into the liner, piston, flame deck and exhaust ports. In addition, dissipated frictional heat is added to the engine liner. This allows calculation of internal heat flux, i.e. the heat exchange between the engine structure, coolant and engine oil.
Based on these heat exchanges and 1D simulation models, a unified vehicle model of between 100 to 150 million cells is created in STAR-CCM+ , which is used to calculate the heat flux from the engine to the coolant and lubrication system from where it is transported to the heat exchanger pack in the front-end (see Figures 7 and 8). There the heat is released to the cooling air passing through the heat exchanger and cooling fan, shown in Figure 9. Therefore, the engine sees the correct underhood air temperature and flow conditions, which is a significant difference to conventional test cell testing.
Figure 7: Breakdown of heat sources and where the heat is released to in the Pandora model
Figure 8: Heat flux from engine surface to ambinet environment at 240kph and 135kW
Figure 9: Heat rejection of the engine combined with the cooling system at 240kph and 135kW
Figure 10: Streamlines through engine compartment for 240kph at 135kW
This integrated STAR simulation model allows the various heat sources to be quantified precisely; the breakdown of each is shown in Figure 10. In the example for a vehicle speed of 240 kph with 135kW engine brake power, 51% of the total combustion heat comes from the combustion chamber, 37% from the water cooled exhaust manifold, and engine friction accounts for 12%. It must be noted that the latter quantity is basically an input from physical testing. The model also reveals heat release to the coolant (79.6%), to the engine oil (14%) and to the ambient environment through the engine surface (5.8%). InDesA notes that due to internal heat fluxes and redistribution, the values revealed by simulation differ from what might be expected based on engineering intuition. The simulation also gives temperatures for coolant, oil in the oil gallery, and temperature of air after the heat exchanger.
Used in combination with bench testing of physical prototypes, InDesA’s virtual approach can predict heat rejection early in engine development with higher fidelity and confidence than bench testing alone. With further development, InDesA believes this methodology has the potential to replace physical prototype heat rejection testing altogether.