New Underhood Module Simulation Methodology for Vehicle Thermal Management

Thomas Morel / Nicholas Tobin / Christian Armbruster – Gamma Technologies and Siddharth Jain -- Modine Abstract Vehicle thermal management is a topic that covers a multitude of design tasks asso-ciated with the development of engines and vehicles. An important part of thermal management design and analysis is the simulation of the underhood cooling module (UHM). The most commonly used methodology is based on 1D streamtubes, ex-tended in the main air direction along the 3D UHM space. However, this method has its difficulties: first, it has to be calibrated by time-consuming and expensive tests or by CFD calculations, and second, it has very limited ability to predict the effects of design alternatives and thus system optimization. Recently, a new approach to simulation of UHMs has been developed, based on Navier-Stokes flow solution, which uses quasi-3D flow elements to achieve rapid 3D UHM flow solution. Among the advantages of this approach is that it is much more predictive than the 1D streamtube methods, needs very little calibration, and is ideally suited for system op-timization under steady-state and transient conditions. This 3D methodology is inte-grated into a system-level design tool, which can model the entire vehicle+engine thermal management in one simulation platform. With it, one can accurately analyze transient behavior of the relevant vehicle systems as influenced by the UHM design. It includes all of the essential components: engine, vehicle, driveline, heat exchang-ers, air fan, water pump, thermostats, A/C, etc. Consequently, it can be applied to the optimization of complete engine & vehicle system to analyze fuel efficiency pre-diction and overall energy and CO2 management. Kurzfassung Das Wärmemanagement von Kraftfahrzeugen ist ein Bereich, der vielfältige Design-aufgaben umfasst, die eng mit der Entwicklung von Motoren und Fahrzeugen in Zu-sammenhang stehen. Ein wichtiger Teil bei der Auslegung des Wärmemanage-ments ist die Simulation der Luftströmung und des Wärmeübergangs im Motorraum (Underhood cooling module, UHM). Die am meisten verwendete Auslegungsmetho-de für UHM basiert auf der Theorie von 1D Stromröhren, die in der Hauptströmungs-richtung durch das Frontend und den Motorraum verlaufen. Diese Methode hat jedoch folgende Nachteile: zuerst sie muss durch aufwändige und teuere Versuchs-reihen oder durch CFD Berechnungen kalibriert werden, und zweitens sie hat begrenzte Fähigkeiten, was die Bewertung von Designalternativen und folglich, was die Systemoptimierung betrifft. Deshalb wurde ein neuer Ansatz zur Simulation von Luftströmungen im Motorraum entwickelt, der ähnlich, wie für die Lösung der Navier-Stokes-Gleichungen 3D finite Volumina mit entsprechend schneller Konver-genz verwendet. Zu den Vorteilen der neuen Methodik zählen die höhere Lösungs-genauigkeit im Vergleich zum Stromröhren-Ansatz, der geringere Kalibrieraufwand,

und die bessere Eignung für Systemoptimierungen von stationären und dynami-schen Betriebsstrategien. Die Methodik ist zudem in eine Tool-Umgebung integriert, die das gesamte thermische System Fahrzeug/Antrieb abbildet. Dies ermöglicht die genaue Analyse verschiedener UHM-Auslegungen für dynamische Fahrzyklen in Bezug auf das Wärmemanagement des Gesamtfahrzeugs. Dabei finden im Wesentlichen die folgenden Komponenten Berücksichtigung: Motor, Fahrzeug, Antriebsstrang, Wärmetauscher, Lüfter, Kühlmittelpumpe, Thermostat, A/C, etc. Die Methodik eignet sich daher auch für die Optimierung des kompletten Fahrzeug-Antrieb-Verbunds hinsichtlich Verbrauch, CO2 und Energiemanagement. 1. Introduction The cooling systems of today must be precisely designed, requiring the use of ad-vanced CAE tools. They must support the overall vehicle objectives, which impact the engine thermal design as well as the vehicle installation issues. This includes cooling and heating of sub-systems such as transmission, oil circuit, intercooler, heater and A/C system, and also the selection of a variety of components such as radiator, CAC, EGR cooler, thermostat, fan and pump. Among the design objectives are: engine durability, fuel consumption, emissions, fast warm-up, passenger cabin heating and air conditioning, component sizing and specification, and system control. 1.1 Underhood Cooling Module Modeling An important part of vehicle thermal management is the design of the underhood cooling module (UHM). This module assembles, in a very confined space, several heat exchangers, stacked in parallel and in series. It is here that converge the cool-ing lines from the engine coolant, oil, CAC, EGR, transmission, A/C, etc. The UHM has to provide the needed cooling to all of these “customers” even at extreme condi-tions (idling, high altitudes, long climbs, max speed), all exacerbated by hot weather. It also is called upon to provide fast cabin warmup, and contribute to the goals of lowering fuel consumption and emissions. Because of all this complexity, the design of the UHM relies extensively on the use of simulations. The most commonly used methodology for analysis of UHMs is based on 1D stream-tubes, which extend in the main air flow direction along the 3D UHM space [1]. The main benefit of the 1D streamtube method is its simplicity, but this method has its difficulties. Since it does not allow flow in the two cross-directions, it cannot by itself predict the inherently 3D flow through the module. Therefore, it always must first be calibrated by time-consuming and expensive wind-tunnel experiments or by 3D CFD calculations. Furthermore, any change in geometry, unless it is very small, requires a re-calibration or acceptance of degradation in accuracy. For this reason it has very limited ability to predict the effects of design alternatives, and it is not well suited for system optimization. Another method in use today is the full 3D CFD, modeling the air flow around the whole vehicle. The weakness of this method is that it requires CPU-time on the order of days for one steady state operating point. Another limita-tion is that it is not a system level simulation (it does not include the rest of the cool-ing system), and thus it cannot be used for analysis of drive-cycle transients.

More recently a new approach to simulation of UHMs, described here, has been de-veloped, based on a modified Navier-Stokes flow solution methodology. This meth-odology has been previously established and proven for solving 3D flow in engine airboxes and mufflers [2]. It uses quasi-3D flow elements to achieve a rapid 3D flow solution. One advantage of this new approach is that it is much more predictive than the 1D streamtube method, and needs very little calibration. Another advantage is the integration into the entire vehicle thermal management model to accurately ana-lyze transient behavior of all relevant vehicle systems as influenced by the UHM de-sign. As a result, it is well suited for system optimization under steady-state and transient conditions. 1.2 Overall-System Modeling Vehicle thermal management is a broad area, whose tasks are distributed over a number of departments within an engine/vehicle design and development activity. To deal with these tasks, engineers have turned to a wide range of analytical tools to help them design and optimize heat management systems (e.g. KULI, Flowmaster, AMESim, Cruise, GT-POWER and Simulink). To study a complete vehicle thermal management system requires that these tools interact and work together. This is not a small task, and in fact, the requirement for so many tools to interact is a roadblock to progress. For one, these tools vary from department to department, and this in-hibits sharing of relevant data across the various disciplines involved in heat man-agement. Another problem is that it is very difficult to synchronize these separate tools, to assure an efficient and reliable flow of the needed information between the various sub-systems. To solve the difficulties arising from the use of disparate analytical tools, an inte-grated software GT-SUITE has been developed, which combines in one place all of the relevant capabilities needed to carry out an analysis of engine/vehicle thermal management. The flow modeling is based on one-dimensional fluid dynamics, rep-resenting the flow and heat transfer in the piping and in the other components of a cooling system. Several parallel fluid circuits can be modeled simultaneously, each containing a different fluid (coolant, oil, transmission fluid, air, combustion products, A/C fluids, etc.). These circuits interact through heat exchangers, transferring heat from one circuit to another, which allows the calculation of the overall heat balance in the system. In addition to the fluid flow and heat transfer capabilities, it contains all of the specialized models required for engine/vehicle system analysis, including built-in engine+vehicle simulation for calculation of thermal loads under any driving cycle. It is thus well suited for integration of all heat management activities arising in engine and vehicle development [3,4]. What is also important is that GT-SUITE is compre-hensive, and so it provides all the flexibility needed to model various advanced con-cepts. Finally, there is another large advantage of an integrated model: much lower cost of licensing and maintenance of interfaces due to the use of a single tool. This paper presents an overview of the solution methodology, and application exam-ples from one OEM and from one automotive supplier. Examples are also shown of applications to analyze a complete engine & vehicle system for fuel efficiency predic-tion.

2. New Underhood Module Modeling Methodology The method described here utilizes a CAD-based tool (COOL-3D) to describe the 3D underhood module and its space. In this tool one first defines the overall 3D dimen-sions delimiting the UHM space. Into this space are inserted all of the components of the module, consisting of any number of crossflow multi-pass heat exchangers and fans (complete with shrouds), plus various obstacles, such as the grille, trans-verse bars, and the engine. The open UHM space that remains is discretized into a large number of cubes in a fashion similar to CFD (Fig 1). The space within the heat exchangers is discretized into “molecules”, which channel the air as well as the transversely flowing fluids (coolant, oil, etc.). The fan is also discretized into smaller sub-elements. COOL-3D automatically meshes and connects all of the components and exports the model to the flow solver. One key methodology is the solution of the flow in the cubes. These cubes have six faces, which in general are all open, although some of them may be closed (e.g those at the UHM walls, at obstacles and at the fan shroud). The flow solution is based on the Navier-Stokes equations. The flow solver is implicit and it operates on a staggered mesh, with the scalar equations (mass and energy) solved at the center of the cubes, while the vector quantities (velocity) are solved at the cube faces. The numerical method used to solve the equations in 3D had been originally developed for solution of flow inside mufflers and air cleaners, where discretization in 3D is nec-essary. It has been extensively used and experimentally validated in those applica-tions [2]. Another key element is the handling of the heat exchangers. These are discretized into many “molecules”. Each molecule has two channels: one for the cooling air and the other is transversely oriented and carries the cooled fluids (coolant, oil, charge-air, etc.). These two channels are connected by a thermal mass that represents the heat exchanger core. The fan is discretized into a number of smaller fans, all con-nected to the same drive shaft.

Figure 1: Assembled UHM components (left), and discretized mode (right) The exported subassembly, which represents an entire UHM, has a number of “ports”. These ports are ready to be connected to the rest of the system, which re-sides in the main assembly file (much like an actual UHM being installed under the hood). There, all inlet and outlet pipes (ports) of the individual heat exchangers are connected to their respective fluid circuits, while fan shafts are connected to electric motor or engine cranktrain, grille inlet and engine compartment outlet boundaries to

vehicle velocity (Fig 2). These main assembly connections can be set up to allow automated UHM design iterations.

Figure 2: Subassembly inserted into a system model 3. Some Important Features of the New Methodology The new methodology brings key new features to underhood module modeling:

- No 1D streamtubes, but full 3D flow mesh - Flow can travel in any direction, including reverse flow - Flow resistance is computed from obstacle geometry and location - Includes flow through the grille, and around the engine - Built-in resistance is not required (resistance is calculated)

The 3D flow solution capability allows the calculation of flow around obstacles (Fig 3); this is absolutely essential, as without it the calculation cannot be predictive.

(a) (b) (c) (d) Figure 3: Unlike the 1-D streamtubes (a), the mesh allows 3D flow (b), and thus ef-

fect of an obstacle (c), is clearly seen in the flow through the radiator (d)

A very important benefit is the capability to model the fan shroud, with one or more fans (Fig 4, left). Including the shroud in the flow calculation allows the prediction of various design features. For example the shroud may contain pressure-activated flaps (Fig 4, right). These are used to relieve the fan resistance at high vehicle speeds, and these effects need to be modeled (Fig 5). Similarly, one needs to be able to assess the effect of the distance between the fan and the radiator. Typically, the shroud and the fan are located very close behind the radiator. As a result, the flow through the radiator is very non-uniform, which leads to a reduced radiator effec-tiveness (Fig 6). The only way to assess such basic effects is by a 3D flow solution.

Figure 4: Shroud with multiple fans (left), and with high-speed flaps (right)

Figure 5: Flap operation: closed at low speeds (left), open at high speeds (right)

Figure 6: Effect of fan-radiator distance, showing the performance degradation

with a fan that is too close (left)

Capturing recirculating flow is another important capability. An open area in the module that allows the flow to bypass the fan and shroud can lead to an undesirable recirculating flow (Fig 7) at low vehicle speeds (part of the hot flow that passed through a heat exchanger returns upstream and then passes through that heat ex-changer again).

Figure 7: Open shroud area can produce undesirable recirculating flow

at low vehicle speeds 4. Validation by Comparing to CFD and Experiments Validation of COOL-3D was performed by comparisons to data from one OEM manu-facturer and to one supplier. In the first instance, a comparison was made to CFD calculations made by BMW for an UHM of a passenger car. The UHM consisted of a two radiators, a CAC, a fan, and a grille (Fig 8). A comparison was made of the two calculations, and the results were found to be in close agreement, with no model ad-justment or calibration (Fig 9).

Figure 8: The BMW CAD model (left) and the COOL-3D equivalent (right)

Figure 9: Comparison of COOL-3D to StarCD calculations made by BMW. Another comparison was made to CFD calculations (validated by experiments) pro-vided by Modine. In this case the UHM was for a truck application, and it consisted of a radiator, CAC and an AC condenser (Fig 10).

Figure 10: The Modine CAD model (left) and the COOL3D equivalent (right) The results of the calculation, made as a “blind” test, were found to agree well with the measurements as shown in Fig 11.

Figure 11: Comparison of COOL3D to CFD calculations made by Modine

The validation tests showed that the new methodology is fundamentally predictive, requiring at the most only moderate adjustments to match CFD calculations or tests. These adjustments would be typically applied to the fan flow rate (by a multiplier), or to the grille (adjusting the effective area of the openings). The computational time is about 30 sec for a model of moderate resolution (12 x 9 air grid). 5. Application of the UHM Model to Parametric Studies One of the main uses of the new UHM methodology is for parametric studies and design iterations. In this section we will describe the predicted effects of changes of several basic parameters, for a cooling module of a small truck which is shown in Fig 12 in its baseline configuration (modified version of the UHM of Fig 10).

Figure 12: Baseline configuration of a truck UHM

The module contains a radiator, CAC and an A/C condenser. For demonstration purposes this model was run for vehicle speeds up to 250 km/h (to force openings of shroud vents) with all other flow parameters constant (fan speed, coolant and CAC flow rates and inflow temperatures). The baseline results are shown in Fig 13. They show that, as one would expect the flow rate increases with increasing vehicle speed, and so do the temperature drops in the radiator and the CAC. The flow rate split between the components is fairly constant.

Figure 13: Baseline mass flow rates (left) and flow fractions (right)

Figure 13 Baseline temperature drop in the radiator (left) and in the CAC (right)

Moving the fan in its plane: the fan was moved upwards by 350 mm from its base-line position and produced large changes in the temperature drop through the radia-tor and the CAC (Fig 14). The overall flow rate increased by about 8%, while the fraction of air passing through the CAC more than doubled.

Figure 14: Effect of fan relocation upward on temperature drop

in the radiator (left) and in the CAC (right) Moving the fan back: the fan was moved 50 mm further downstream from the radia-tor, and this had a positive influence. It increased the overall flow rate through the UHM by about 6% (primarily helping out the CAC), and increased the temperature drop of the CAC significantly (Fig 15, left). This is because of the better utilization of the heat exchanger area, as shown already in Fig 6. Moving the A/C condenser in its plane: the condenser was moved vertically up and down from its baseline position (which was in front of the radiator and CAC) by a to-tal of 400 mm. When moved upward, i.e. in front of the CAC (which had the higher flow resistance), the main effect was to reduce the flow through the condenser (Fig 15, right).

Figure 15: Effect of fan-radiator distance on CAC temperature drop (left) and the effect of condenser vertical position on condenser flow fraction (right)

Presence of a leak path: an important factor affecting the UHM flow distribution are leak paths. In this case, allowing an opening above the CAC showed an important negative effect on the CAC temperature drop (Fig 16); this effect is particularly large at low vehicle speeds, where it produces undesirable reverse (recirculating) flow, re-turning the heated air back in front of the CAC.

Figure 16

Figure 16: Effect of a leak path on CAC temperature drop (left) and on the total

mass flow (right) Flaps installed in the shroud: the final design change studied here was the effect of shroud flaps. These flaps are designed to open automatically when the pressure in front of the shroud exceeds the pressure behind it, in order to relieve the pressure loss in the system. In this case the flaps opened at about 100 km/hr. They were located on the upper portion of the shroud, and therefore they affected the CAC the most, as seen by the increased temperature drop (Fig 17).

Figure 17: Effect of pressure-activated flaps on the CAC temperature drop

It should be noted that none of the parametric changes in this section could be pre-dicted by the usual 1D streamtube methodology. This is because all of those design changes require the calculation of the resulting flow redistribution in a 3D fashion; it is this requirement which is met by the present methodology, and which led to its de-velopment. 6. System Modeling with a Mean Value Engine Model

Figure 18: Integrated model: vehicle + mean-value-engine + cooling

The UHM model can also be used within complete system simulations. To illustrate such use an analysis was carried out of a vehicle operating on the FTP-72 City Driv-ing Cycle. The vehicle was a RWD car with a 3.1 L 4-cyl turbocharged diesel engine,

with a computer controlling the variable geometry turbo and the EGR valve. A fairly detailed model was used for the vehicle and its driveline. The engine was repre-sented by a mean value model calibrated by a detailed GT-POWER model. The en-gine heat load (coolant and oil separately) was modeled by neural networks, cali-brated by the same detailed engine model. The cooling system consisted of the coolant flow, oil circuit and water-cooled EGR, all of which, together with an A/C heat load, were channeled through the UHM module. The UHM was discretized in the transverse plane into a 12x9 grid, giving it a good flow resolution. The system model is shown in Fig 18 above. The whole system was run in a fully transient mode, with the time step set to 0.10 sec for the cooling circuits, and 0.05 sec for the mean value engine. Despite the level of detail contained in this model, the computational time is about 4.5 times the real time; this definitely makes it practical for system design and optimization.

Environmental Conditions: Once a baseline model has been constructed and vali-dated, it becomes available for a great variety of possible design studies. One ex-ample is a study of the effect of the environmental conditions (Fig 19). A comparison was made of the effect of high altitude (2700 m) compared to the baseline run at 1 bar and 300 K. As Fig 19 shows, the coolant temperature increases during the war-mup phase until the thermostat opens (top broken line is the high-altitude case). What can be seen is that in the high-altitude case the thermostat opens fully (10 mm), and it no longer controls the coolant temperature, and that an additional control is needed.

Figure 19: Comparison of the effect of high altitude (2700 m) compared to the baseline. It shows the evolution of the coolant temperature entering the block (left) and the thermostat lift (right), during the FTP City Cycle

Electrically driven fan and coolant pump: An investigation was made of a remedy that would solve the inadequate cooling at the high altitude. Since the baseline case was run with fan and the coolant pump driven directly mechanically from the engine, the possibility was explored of introducing electrically driven fan and coolant pump, with controls linked to coolant temperature, allowing the fan and the pump to be op-erated at full capacity whenever needed, even at idle and low speeds. The results of this approach applied during the FTP city cycle are shown in Fig 20. With the elec-

trically driven pump and fan the cooling objective is met, stabilizing the engine tem-perature at the desired level. The plot of the fan power in Fig 20 illustrates the usage of the fan during the cycle, running at low power during the initial cool period, and smoothing over the cycle variations after the engine has warmed up.

Figure 20: Improved engine cooling by use of electrically powered fan and coolant pump, coupled with their electronic control, showing the coolant temperature entering the block (left) and the fan power (right) during the FTP City Cycle

Effect of coolant mass: Another parameter investigated was the effect of the coolant mass on the coolant warmup. In this study, the volume of the coolant in the cooling circuit (excluding the radiator) was approximately doubled, compared to the baseline. The main result of this change was that the initial warmup was substantially slower, while at the later portion of the cycle the coolant temperature became about the same for both coolant masses.

Figure 21: Effect of coolant mass on coolant warm up

Summary A new approach to simulation of UHMs has been developed, based on Navier-Stokes flow solution methodology. It uses quasi-3D flow elements to achieve rapid 3D UHM flow solution. Among the advantages of this new approach are that it is much more predictive than the usually used 1D streamtube methods, it needs very little calibration, and it is ideally suited for system optimization under steady-state and transient conditions. The new flow solution has been validated by comparison to CFD and experiments. Being 3D based, the flow solution predicts the effects of various design changes, such as in positioning of heat exchangers, fan location, fan shroud design with flaps, leak paths, grille, etc., none of which are possible to predict with the 1D streamtube methodology. This 3D methodology is integrated into a system-level design tool, which can model the entire vehicle+engine thermal management in one simulation platform. With it, one can accurately analyze transient behavior of the relevant vehicle systems as in-fluenced by the UHM design. It includes all of the important components: engine, vehicle, driveline, heat exchangers, air fan, water pump, thermostats, A/C, etc. Con-sequently, it can be applied to the optimization of complete engine & vehicle system to analyze fuel efficiency prediction and overall energy and CO2 management. Acknowledgment The authors would like to acknowledge the kind assistance of Dr. Gerald Seider of BMW, who provided the CFD and test data used for the model validation. References [1] C. Stroh, Magna Steyr, “KULI Air Side”, KULI User Conference, 2006. [2] T. Morel & J. Silvestri, Gamma Technologies, K-A. Goerg, BMW AG and

R. Jebasinski, J. Eberspaecher, “Modeling of Engine Exhaust Acoustics”, SAE Sound and Vibration Conference, Traverse City, MI, May 1999

[3] J. Dohmen, R. Barthel & S. Klopstein, FEV Motorentechnik, “Virtuelle Kuehlsystementwicklung“, MTZ Edition No.: 2006-12

[4] G. Seider, BMW AG, ”Design of Automotive Cooling Sustems with GT-COOL“, 11th GT-SUITE Conference, October 8, 2007, Frankfurt, Germany.