falloutboy's Fundgrube und Texte

Platzierung/Dimensionierung Motorkühler

[falloutboy]

Cooler by design

Formula 1 teams have had to go on something of a journey of discovery when it comes to cooling. Cranfield University student Pierre Salmon looks at how computational fluid dynamics can help the teams to extract the most from their cooling systems

Cranfield University’s Advanced Motorsport Engineering MSc allows students to get closer to Formula 1 through interaction with professionals in the industry, detailed academic courses, a group design project and an individual thesis. One of the 2014 theses looked at evaluating the required air mass flow rate through the sidepods needed to reject the waste heat produced by the F1 powertrains to comply with the new 2014 regulations. Computation fluid dynamics was used to investigate the potential cooling and aerodynamic benefits of five different

configurations and any effects they might have on the performance of the engine.

The 2014 F1 technical regulations concerning powertrains have provided considerable challenges to the engineers regarding the packaging of the cooling system. The addition of a turbocharger has resulted in more required heat rejection to the air flowing through the sidepods. The oil and water cooling requirement for the engine remains relatively the same for the 2014 V6 engine compared with the 2013 V8 engine, but the addition of the charge air cooler results in a much higher requirement of cooling air mass flow. There is also an increase in electronics

cooling requirement from the higher power outputs of the MGU-K and more complex energy management electronics. A strong need therefore exists to find a solution that could possibly reduce the cooling requirements or find a cooling configuration that allows for the highest heat rejection rates.

Designing the most efficient cooler configuration has ample benefits as it affects the three performance differentiators of the 2014 season – power, aerodynamics and reliability. Firstly, effective cooling of the charged air reduces the density of the air going into the cylinder. This allows more molecules of oxygen per unit volume to be reacted and means more fuel can be combusted per cycle, allowing for a higher IMEP. Secondly, the e ective cooling of the air, water and oil will reduce the average operating temperature of the engine and so extend the life of the engine. The reliability and life of all the components in the engine is crucial to the successful operation and racing of the car since only ve powertrains were allowed per season per driver in 2014.

The enlarged sidepods of the 2014 cars provide a major reduction in aerodynamic efficiency of the vehicle as they slow down more air and reduce the clean ow of air to the rear of the vehicle. It would therefore be of benefit to any team to be able to increase the heat transfer abilities of the cooling systems and reduce the cooling air mass flow rate.

figure1

Transferring heat

An increased efficiency of the engine fluid cooling will also free up cooling capacity for the electronics and turbocharger. Throughout the 2014 pre-season testing it was found that efficient cooling of the cars’ systems resulted in major gains in reliability and performance, and the first half of the 2014 season saw numerous iterations of bodywork to optimise the cooling of the vehicles. In this project, the cooling from

an air to air charge air cooler (CAC), a water radiator (WRAD) and an oil radiator (ORAD) are considered. The typical mass flow rates of air

through the sidepods would be between 1.2 and 2.1kg/s, depending on vehicle velocity and sidepod inlet area. The 2014 engine needs to reject around 113kW of heat from the water radiator, around 41kW from the CAC and in the region of 58kW from the oil radiator.

The five different configurations see the CAC, WRAD and ORAD placed in a way as to maximise the heat transfer or reduce the external pressure drop across them. Extreme Temperature Difference (ETD) is the main driver behind heat rejection from coolers and designers typically ‘sweat off’ the heat as much as they can by staggering the hotter surfaced heat exchangers behind the colder ones as in

Configuration 4. Another strategy is to have all heat exchangers exposed to the cool inlet air and size them relative to their heat rejection requirement, as per Configuration 2.

graph1

The geometry of each heat exchanger is optimised to provide similar performance while maintaining a specified volume. It is often the greatest challenge for aerothermodynamicists to package these relatively bulky devices around the engine and in the sidepod. Dense

cooling cores are used to maximise the surface area and fin efficiency, with fin density being as high as 21 fins per inch. The fin efficiency

of each core is also crucial as this allows for greater heat transfer to the cooling air. It is also important to understand how the heat exchangers affect the performance of other heat exchangers downstream. If you imagine the heat exchangers are packed very close together inside the sidepod, then once the air has passed through the front heat exchanger, it has warmed up and is traveling slower and is more laminar, all of which negatively affects heat transfer. Traditionally louvres are used to break down the aerodynamic and thermal boundary layers along the fins to re-establish a high temperature gradient and turbulent flow near the heated surface.

graph2

Front wheel tyre wake is also an issue as this highly turbulent air, which is moving at a slower relative velocity, will inhibit heat rejection performance of the cooling system in the sidepod if it were to enter it directly. Aerodynamicists now cleverly use the Y-250 vortex to help divert the wheel wake around the lower side of the sidepod. Not only does this help with reducing drag, but it also allows for faster cleaner moving air to enter the sidepod and effectively, and predictably, extract heat from the cores. See Figure 1.

The CFD process used a generic sidepod geometry generated in Catia V5, with the five different cooling cores input as finite tubes of

varying temperature. There were in excess of 25 million cells in the meshes and the simulations took four days each to run on the Cranfield

High Performance Computing Cluster. The macroscopic heat exchanger model was not used because obtaining the heat rejection to mass flow rate correlation curves from an F1 team are like trying to find hen’s teeth. In addition, the separation events and effects of tube geometry was interesting to see, so the full detailed CFD was done in order to gain as much information as possible.

What is interesting to note is that Configuration 4 yields the highest heat rejection rate at the car’s maximum velocity of 90m/s. This is due to it having the cooler CAC in front of the warmer radiator, so the absorption of heat into the cooling air is maximised. The pressure drop across the configurations are also of interest because it is no good having a cooling system that rejects the heat but costs a significant internal aerodynamics drag penalty. Now, what is of further importance is to observe what the air is doing after it has left the cooling cores and how this affects the rear of the vehicle. The air velocity uniformity at the sidepod outlet is of concern because it affects the aerodynamics at the rear of the car. The sidepod inlet shape determines the air mass flow distribution and thus the outlet flow condition. The ideal situation would be to have uniform air mass flow distributions through the cooling cores, and no severe pressure gradients at the outlet. Observing the rear of the sidepod (so looking forward towards the outlet) we can see a comparison of the outlet velocity streamlines of each configuration.

Rotational flow

The flow structure behind Configuration 3 might appear more chaotic, but of importance is the strength of the rotational flow present

behind all the cores. Configuration 1 shows a double counter-rotational structure, with the primary stronger vortex on the right-hand side, which develops initially due to the non-uniform pressure distribution aft the cores. This induces the left-hand side counter-clockwise vortex due to viscous sheer forces between particle layers.

figure2

Configurations 4 and 5 show similar trends in a strong rotational ow structure aft the cores, where the region of low mass flow rate results in lower particles per volume, which directly relates to a lower pressure, which accelerate high pressure particles region towards it. The high pressure particles already have momentum towards the sidepod exit, and their gradual acceleration towards the low pressure region results in the rotational flow structure being established.

figure3

Observing the Q-Criterion 0.018, vortex laments show how the rotational flow structures develop in the sidepod after the cores. Configuration 3 shows smaller and fewer vortices than Configuration 1. The counter-rotating vortices of Configuration 1 are shown, and the two larger vortices labelled near the top of Configuration 3 dissipate due to longitudinal acceleration as the cross sectional area decreases near the outlet.

The vortex exiting on the left (1) is produced from, initially, the flow over the last row of tubes at the bottom of the water radiator

rolling over and creating a standing vortex. As the air ow from higher layers exits the cores it follows the ow and encourages the rotational

ow laterally (around the axis indicated by the dashed arrow), which wraps around and then rotates around an axis perpendicular to the page as the vortex nears the exit, due to the lower pressure aft the core from the non-uniform pressure distribution mentioned before. The difference in strength between the vortices of Configuration 4 and 5 is that with Configuration 4, the flow encounters the longer radiator tubes last, and so the flow is disturbed less than with Configuration 5 with the shorter CAC tubes at the rear. The less disturbed flow of Configuration 4 has a higher velocity (boundary layer build up and effective flow area) and the higher the exit velocity the higher the tangential velocity of the vortex. The higher tangential velocity, due to Newton’s first law, results in a larger radius, but this also increases the volume of the vortex and a lower core pressure is experienced, increasing centripetal acceleration, Newton’s Second Law. The equilibrium state between the two forces results in a faster spinning stronger vortex, observed by Configuration 4.

figure4

Hotspots

The right-hand side vortex is also rotating clockwise, but is slightly weaker, and in the case of Configuration 5, is dissipated before the outlet because the flow accelerates in the longitudinal direction as the cross-sectional area decreases. The core regions of these vortices (inside the laments) are relatively calm, where viscous dissipative forces damp out any large scale turbulent fluctuations. Generally this helps preserve the core region of the vortex and helps extend the life of the vortex. The resulting pressure contours at the outlet is shown below, where Configuration 3 has the lower pressure gradients due to more uniform flow, and Configuration 4 and 5 have the strong vortex’s low pressure in the filament core. As a result of the sidepod outlet flow conditions, it could be speculated that in the future teams might want to make use of this flow in their aero packages.

The thermal energy exchange between the heated surfaces and the cooling air can be gauged by the temperature differences of the air after the cores. The following images show the temperature rise inside the sidepod. Configuration 3 clearly shows a hotter region just behind the oil radiator as the air is being heated by three stacks of cores. A noticeable difference between Configuration 4 and the other configurations is the uniformity of the air temperature after passing through the cores.

In contrast, Configuration 1 and 3 have regions of higher temperature and regions of lower temperature, showing variation of the

mass ow rates through the cores. Ideally, equal mass flow rates of air through all the rows of the cores should be achieved with optimisation of the sidepod inlet shape, plan view shape and exit duct shape. This image clearly shows the temperature gradients and streamline flow into

and out of the cooling cores.

figure5

The affect of these CFD results were then applied to an engine simulation of a 2014 regulation F1 engine. The CAC was of particular interest as it determines the temperature leading into the combustion chamber, and this has a major impact of volumetric effciency of the cylinders and the rate of evaporation of the fuel, amongst others which influence the overall combustion performance.

pic1 pic2

The advantage of having a CAC which cools the air down to a lower temperature is that this air has a higher density, and so more molecules of air can be induced by the engine per cycle allowing for improved engine breathing. Graph 3 shows that, in order to achieve the same output power, for a poorer performing CAC, a higher pressure ratio over the compressor would be required. This higher pressure needed to pump this relatively hotter air would require more power from the turbine wheel, and so result in higher back pressure in the exhaust manifold – bad for scavenging, and less energy available for the MGU-H. The overall effective design of the cooling system has a holistic benefit on the F1 package, in so far as aerodynamics and power are concerned.

figure6

Low speed cooling issues

It is also important to consider the performance of the CAC at low vehicle velocities, as the cooling and pressure drop is highly dependant

of air velocity through the sidepods. A cooling map can be generated, which is a combination of the CFD and engine simulation work, which

shows that at lower vehicle speeds the car might not be cooling exactly as much as is needed. These maps can then be used to perform an energy audit of the total watts of heat rejected over the course of a lap and this will provide engineers feedback on the cars performance and that of the cooling system. Figure 6 shows a map for cooling Configuration 4 and Figure 7 is the heat rejection trace of around one lap of Circuit de Spa-Francorchamps.

table1

figure7