The reason for this unusual choice was that data obtained during track test had indicated the possibility that normalised downforce levels fell when the vehicle was decelerating. In particular, the team thought it was possible that the normalised downforce was declining because of the development of a boundary layer at the rear wing as the wake from the rear wing overtook the wing section as the car decelerated. The team’s then current simulation methods could not be relied on to accurately predict such subtle transient flows, so they decided to use a full-sized rear wing in a water towing tank. This would allow analysis of the occurrence, or non-occurrence, of flow separation and the changes in loadings as the vehicle was decelerated.

The most important factor in such a test was to account for the differing densities of air and water. The key benefit of such a test is the fact that water’s higher density effectively allows flows to be simulated in slow motion. Roughly speaking, for the same dynamic pressure across the wing, the speed of water flow over the wing element need only be 1/30th of that if it were in air. Provided the differing factors such as viscosity are properly accounted for – the acceleration rate needed to generate representative forces were about 1/1000th of those in air – it is therefore feasible to collect data in a water tank that can be directly related to operation in air.

Given this, it is possible to conduct slow-motion tests in water, and assess transient aerodynamic loads during deceleration – which are challenging to measure when the vehicle is actually running on a race track – to be measured with a high degree of accuracy. During Honda’s specific rear wing assessment programme, deceleration tests from 0.005-0.05 g and fixed-speed tests from 1.02-2.94 m/s (corresponding to 106-307 kph in air) were conducted.

The tank used for the test was 200 m long, 10 m wide and 5 m deep. The rear wing assembly was placed in the tank, inverted and attached to a six-component load cell, 1 m below the surface. To prevent waves forming on the surface of the water that could potentially skew the results, a flat acrylic plate was placed near the water surface.

Honda conducted tests with this system to assess the rear wing flows under deceleration, and was able to conclude that the flow was not reversing over the wing and that the wing was operating as it should at all times. The tests also showed that such methods were a useful additional testing tool, particularly for physically assessing hard-to-measure behaviours such as aero elasticity.

Whether other teams have undertaken such tests is not known, but it raises an interesting question – would such methods fall outside the CFD and wind tunnel limits set by the FIA, therefore providing a valuable extra source of unrestricted aerodynamic data?

Written by Lawrence Butcher

A lot can be gleaned simply by looking at the cars as they sit in the pit lane, for example McLaren’s ‘mushroom’ rear wishbones in the 2014 season. But obtaining quantitative data on the competition can be trickier. In the world of production cars, a manufacturer will often buy a competitor’s vehicle to assess the technology it contains, but this is clearly not feasible in Formula One. Teams therefore need to be a little more creative. 

For example, before on-track data was recorded and displayed by the governing body for TV audiences, teams would use audio analysis techniques to ascertain the revs the cars were reaching at various points on the track. This could then be combined with speed data gathered by someone trackside using a speed gun, and factors such as likely gear ratios calculated. 

The widespread availability of handheld thermal imaging equipment, and even the images generated by thermal cameras as part of the current TV coverage, can provide an insight into the way a particular car is using its tyres. For example, by recording the tyre surface temperatures using a camera at a specific part of the track, comparisons can be made between cars. Also, by looking at variations in surface temperature distributions, teams can combine that with other analysis and data gathered from their own on-car sensors to begin to see how the opposition may be working their tyres. 

Image analysis can also provide a considerable volume of information. Using still images, known reference points on the car can be taken, and from these the orientation of other components can be studied. For example, if a known vertical plane can be established at the rear of a car, then the camber of the rears wheels can be estimated, as can the roll attitude of the body. Fig. 1 below shows how this is achieved, though it should be noted that this is a quick Photoshop mock-up, albeit based on a real team’s analysis imagery. 

Taking this a step further, capturing stereoscopic images can help to understand the relationship of one part of  car to another, particularly when it comes to suspension or aerodynamic components. This information can be fed into a CAD package, and from it 3D approximations of components created, revealing useful information such as suspension geometry and relative component sizes. Although it is not known whether any teams have used such devices, compact handheld 3D laser scanners capable of millimetre-accurate measurement are commercially available, and if someone could devise a means of using one in the pit lane without being noticed, no doubt the data gathered would be very revealing. 

Video footage, either captured by the a team trackside or from the readily available TV feeds, is another useful analysis aid. By comparing footage of competitors’ cars taken from a fixed position to their own (for which they have real-world data), factors such as corner entry speed, differing racing lines and braking points can all be ascertained. Teams have even used high-speed cameras trackside to deduce factors such as wheel slip rates, by comparing the difference in rotational angles between front and rear wheels. 

There a undoubtedly many other cunning methods used by teams to get the measure of their opposition, all of which will be fed back into the overall car development programme for assessment. While understanding how a competitor’s car works will not win a team races, it is important intelligence when combined with a thorough and verified understanding of their own car. As Sun Tzu, the 5th century Chinese military strategist, said in his seminal work The Art of War, “If you know the enemy and know yourself, you need not fear the result of a hundred battles. If you know yourself but not the enemy, for every victory gained you will also suffer a defeat. If you know neither the enemy nor yourself, you will succumb in every battle.” 

Fig. 1 - Basic trackside photography can be used to calculate factors such as approximate camber and roll angles (Composite image: Lawrence Butcher)

Written by Lawrence Butcher

Since that first round of testing, the crash tests a chassis must pass before it is accepted for competition have grown considerably in number and complexity. Meeting the requirements of these tests, while also building a car that is as light and tightly packaged as possible, is a stern test for teams’ engineers. Cars must now pass two frontal impact tests, one that assesses the frontal crash structure and another that covers the integrity of the main chassis monocoque. The monocoque must also undergo various rear and side impact tests, as well as impact tests on the roll structure, and various static load tests on areas such as the monocoque sides and cockpit surround. The standards for the frontal tests give an insight into the severity of the tests as a whole.

For the first test, the frontal crash structure is fitted to the monocoque, which must be fitted with a fuel tank filled with water and a 75 kg test dummy. This is then attached to a sled, which has an all-up weight (including the monocoque) of 780 kg. The sled is fired at a solid wall at a velocity of at least 15 m/s (33.6 mph). It is worth noting here that the standard safety rating tests for roadcars feature a similar velocity, but use a deformable barrier (not to mention the fact that roadcars have a much larger frontal volume available for energy absorption). As the monocoque structure impacts the wall, accelerometers record the level of deceleration using a high-speed data acquisition system, and the test is only passed if the levels of deceleration fall within the FIA requirements.

The requirements are as follows:

a) The peak deceleration over the first 150 mm of deformation does not exceed 10 g.

b) The peak deceleration over the first 60 kJ energy absorption does not exceed 20 g.

c) The average deceleration of the sled does not exceed 40 g.

d) The peak deceleration in the chest of the dummy does not exceed 60 g for more than a cumulative 3 ms, this being the resultant of data from three axes.

Or:

a) The peak force over the first 150 mm of deformation does not exceed 75 kN.

b) The peak force over the first 60 kJ energy absorption does not exceed 150 kN.

c) The average deceleration of the sled does not exceed 40 g.

d) The peak deceleration in the chest of the dummy does not exceed 60 g for more than a cumulative 3 ms, this being the resultant of data from three axes.

During this test, the condition of the monocoque, particularly areas such as the safety belt mountings, is also monitored to ensure they do not suffer damage or deformation.

The second frontal test is to account for the fact that it is perfectly feasible that a car will suffer multiple impacts during a crash. It is therefore important that the driver is still protected, even if the front crash structure is no longer present.

For the second test, the monocoque without its front crash structure is fitted to the sled, with the crash structure replaced by an aluminium plate 50 mm thick. This time though the sled is fired at a deformable impact wall, which must resist a minimum load of 60 kN at the same 15 m/s. To obtain a pass, the monocoque must be able to resist the resulting impact with no damage to the survival cell, with particular attention again given to the safety belt mountings.

The requirement for the multitude of other tests that are undertaken are no less stringent. It is therefore no surprise that drivers are now usually able to walk away from accidents that 20 years ago would have held much grimmer consequences. Passing the tests may present its fair share of headaches to teams’ engineers, but the fact that there have been no driver fatalities in a GP since Senna and Ratzenberger in 1994 shows that they are more than worth the effort. 

Written by Lawrence Butcher

By and large, modern racecars still rely on water-to-air coolers to remove heat from the engine, and oil-to-air coolers for the lubricating fluids, although oil cooling of electric motor/generators and batteries is also becoming more common. Invariably, these coolers are mounted in the sidepods of a car, and the primary concern when specifying these systems is ensuring that sufficient cooling capacity is available to keep the powertrain operating comfortably under race conditions.

Often teams will build radiator cores in-house to their own specification. The internal pipework and external fining of these units is optimised using CFD, to ensure the most efficient fluid flow through the system while providing the maximum possible surface area for cooling. In Formula One the cores need to be able to withstand in excess of 3.75 bar of pressure, the maximum allowed by the FIA regulations; at this pressure the boiling point of the coolant rises to around 120 C, allowing for smaller radiators to be run. This in turn allows for smaller inlets, reducing the drag penalty on the overall aero package, with the difference between the maximum and minimum cooling package accounting for up to 5% of the total downforce. However, there is a constant trade-off between aerodynamic drag and engine performance, with a 5 C increase in engine temperature reducing the power output by about 1 hp.

Before track testing and wind tunnel time in Formula One was limited, assessment of cooling systems could be completed either in wind tunnels or trackside. The reduction in time available to test cars though has seen overall aero package testing take centre stage, to the detriment of other areas of development (such as cooling optimisation) when it comes to assigning track or wind tunnel testing time. As one Formula One engineer has put it, “Generally we have very limited (or no) time within our ‘aerodynamic’ wind tunnels for coolant system development.” This has meant teams have had to outsource some of the testing for cooling components, using companies who provide thermal and aerodynamic testing of water, oil and charge coolers.

To test coolers, the units are usually mounted on either a suitable template or enclosed in a sidepod before being attached to a wind tunnel test section. One such commercially available facility is equipped with a 300 kW boiler to provide high-temperature coolant to the radiator cores. A boiler of such capacity is needed owing to the high levels of heat rejection in a modern race engine, with Formula One motors transferring anywhere up to 230 kW of heat into the coolant.

The key tools for assessing the cooling efficiency of a core are a number of temperature sensors placed at the inlet and outlet of the radiator, and pitot tubes to measure the inlet and outlet airflow. As temperature-controlled airflow is passed over the outer ‘finned’ surface of the radiator, a temperature-controlled flow of fluid is passed through the internal tube passages. The system is then run through a pre-determined test matrix of differing load and airspeed conditions, with the level of heat dissipation being derived for each condition.

If cores are tested in isolation, without being mounted in sections of the vehicle, a closed-loop wind tunnel can be used. The closed-loop system allows for the level of heat entering the airstream to be calculated and a heat balance obtained, in order to double-check the data gained from simulation.

Beyond the basic measurements of inlet and outlet coolant temperatures and airflow velocities at the cooler face, subcontractors have also started to use laser doppler anemometry (LDA) and thermal imagery to provide visualisation of core performance. The use of thermal imaging cameras gives a very clear idea of surface temperatures across the core, allowing for hot and cold spots or blockages to be identified quickly, while LDA allows for detailed analysis of air velocity over the cooler.

Given the likely complex nature of the cooling packages that will be present in 2014, facilities providing such services will no doubt be in greater demand than ever.

Written by Lawrence Butcher

A close look at the floor of cars such as the 2013 Red Bull or Ferrari bears out this approach, revealing a plethora of tiny flicks and winglets, each deemed to have produced some gain in performance in the wind tunnel. However, in order to ascertain whether such subtle changes are effective, the correlation between results obtained with a 50 or 60% scale model, CFD simulation and the car on track must be precise. This means that as many factors that dictate aerodynamic performance on the full-scale car must be replicated on the scale model, including the tyres.

Producing scale-model tyres that behave in the same way as their full-sized counterpart is not easy. The high sidewall tyres used in Formula One present some interesting challenges in this area due to the way they deform under load, changing their aerodynamic characteristics. Accurately modelling these changes in the wind tunnel can be beneficial to improving overall aerodynamic performance. In order to achieve this accuracy, teams and tyre suppliers have gone to great lengths to produce scale wind tunnel tyres that can replicate these changes and methods of testing that best replicate on-track performance.

In the past, scale models were run with solid composite or aluminium ‘tyres’ that produced repeatable results but did not accurately mimic the aero effect of a real tyre. It is only relatively recently that tyre manufacturers have begun to produce scale-model tyres that represent their full-scale counterparts.

Producing these tyres is not simply a case of scaling down the construction of a full-scale item, as the forces a tyre is subject to in the real world differ considerably from those found in the wind tunnel. For example, if the construction of a full-scale tyre were to be reproduced in scale form, the resulting tyre carcass would be far too stiff and would not create the required deformation. Also, whereas a ‘real’ tyre is optimised to provide maximum grip, this is not desirable in a wind tunnel where the model may be required to yaw from side to side across the tunnel’s rolling road. Thus the scale-model tyre needs to be as low grip as possible.

In order to generate the required vertical loads to get the tyre to deform, teams can use various systems of actuators in the scale model’s suspension to load the tyre against the rolling road floor. However, these loads cannot be too high as this would be detrimental to the life of the tunnel’s rolling road belt, which is normally made from either steel or fabric. To counter this, model tyres will be run at very low inflation pressures to reduce the loads needed to cause sufficient deformation.

While it is one thing to make the tyre deform, it is another to ensure it is deforming in the right way. This means the deformation needs to be measured, and one way this can be achieved is by using stereoscopic cameras. The images taken are processed using software that identifies the location of specific pixels from frame to frame, and the level of deformation at different loadings can be measured and compared to data provided by the manufacturer for the full-scale tyre.

While all of this effort may seem excessive given the relatively small difference accounting for changes in a tyre’s shape has on the overall aerodynamic package performance, in a sport where performance between rivals can sometimes be measured in thousandths of seconds, it is an expenditure in time and resources that teams feel is justified.

Written by Lawrence Butcher

With the current ban on full-scale wind tunnel testing, and the potential for even tighter restrictions in 2014, teams have been forced to gather more and more aerodynamic data from track testing in order to validate their scale model and CFD programmes. The bulk of this data is obtained by two methods – measuring air pressure at points on the bodywork, and measuring the aerodynamic loads on the aerodynamic surfaces. However, teams will also use non-data based analysis in the form of flow visualisation, more on which later.

The most common method for collecting pressure data is through the use of pitot tubes, sometimes arranged in arrays to analyse flow in a specific area. The latest generation of sensors feature built-in processors to provide individual pressure readings, usually to gauge airspeed, and can often be seen on cars in test configuration mounted high up above the airbox in clean airflow. The processor built into the base of the pitot compares the dynamic and static pressures to provide an accurate airspeed reading.

Arrays of pitot tubes without built-in processors are used to measure pressure differentials across an area. Often, and particularly when a team is struggling to correlate simulated and real-world performance, cars can be seen sporting these complex pressure sensing arrays next to key aerodynamic appendages. These are linked to a differential pressure sensor that processes the pressure provided by each tube. The data is then transmitted through the car’s CANbus network, greatly reducing the complexity of the wiring loom. Teams have used these very dense arrays of pitots in a host of placements, looking at everything from wheel wakes to diffuser flow. 

In addition to pressure measurements, bespoke load cells are often incorporated into highly stressed downforce-generating components in order to ascertain load distribution across the surfaces. For example, sensors can be incorporated into the front and rear wing supports, encased inside the composite structure. The sensors use load cells to gauge the forces on the wing, and each cell produces outputs for lift, drag and pitching moments, and from these outputs the centre of pressure for the wing can be calculated.

While the data gathered using both of these methods can be invaluable, sometimes it is useful to be able to visualise the flow around a component. This is one of the oldest methods of analysing the aerodynamic performance of a car. In the past, wool tufts would be attached to the bodywork, and when the vehicle moved they indicated the direction of airflow over surfaces. The modern approach is to use ‘flow-vis’, a high-visibility fluid sprayed onto the surfaces of the car. As the car is driven, the fluid is pushed over the surface, leaving trace lines indicating the direction of flow. The traces can show up factors such as stalled flow or areas where the flow has reversed or is not heading in the desired direction.  

Using a combination of these methods, engineers will gather data with which they can improve their off-track simulation models. It is this model validation that is the most valuable product of on-track testing. While it is useful to see if a new component has delivered tangible benefits on-track, it remains a fact that, for the foreseeable future, most development will occur in the virtual domain. Therefore, the more real-world data that can be fed to the simulators, the more effective developments will be when they hit the track.

Written by Lawrence Butcher

One such development in the sphere of aerodynamics testing has been the appearance of PIV (particle image velocimetry), a process that allows engineers to gain an accurate visualisation of flow characteristics and velocities over a vehicle or component. In the past, the only way to physically visualise the flow over a part was to use either a smoke wand or coat the component in a dye that would highlight flow lines, but neither of these provided a particularly detailed picture of the flow interactions at the all-important surface-air interface. Flow can also be visualised in a CFD (computational fluid dynamics) simulation, but it has been difficult to ascertain whether the results are accurate. PIV is therefore a very useful tool, allowing for simulated results to be validated and new insights gained into flow characteristics.

The process involves filling a wind tunnel with a mist of tracer particles that have essentially the same density as air, meaning that the flow conditions will be as close as possible to those found on track. When the tunnel is running, the particles flow with the air stream through the test section. For a PIV test, engineers position a camera at 90º to the plane of the flow field they want to study, and the part to be studied is illuminated with a high-power laser, creating a 2D plane of light.

A series of two-set photos is then taken at extremely rapid intervals – generally 10-20 µs – which are then processed using powerful software to track the position change of each particle in the cross-section, allowing for the rate and direction of flow to be calculated. The resulting image looks very similar to those generated in CFD, but instead of being created from simulated data it is a representation of the actual flow velocity over a part.

The result is that PIV now provides engineers with the ability to carry out accurate checks on CFD predictions on flow velocity in the real world, allowing for the CFD models to be refined for greater accuracy. An example of this is the Toyota Formula One team’s efforts during the 2009 season to understand the flow generated by a car’s wheels. The team had run CFD simulations of the flow but PIV allowed them to test the accuracy of these results. The use of the new technique was very revealing, and showed that the CFD simulated results were considerably underestimating the size of the wake generated by the wheels. Once this was discovered, the engineers were able to adjust the CFD parameters to negate the discrepancy, giving them greater confidence in subsequent CFD test results.

The other major benefit of PIV is that it does not interfere with other wind tunnel testing procedures. With wind tunnel use regulated, this represents a considerable benefit, increasing the data yield from a particular test run.

Fig. 1 - A comparison of PIV and CFD plots showing the wake off a front wheel. The results on the right show the increase in correlation between results generated in the wind tunnel and CFD following optimisation of the CFD programming (Courtesy of Toyota Motorsport)

Written by Lawrence Butcher

Ever-tightening track testing restrictions has meant that teams have had to rely increasingly on off-track testing methods in order to develop their cars both before and during the season. This has seen considerable investment in facilities such as wind tunnels, with many teams running tunnels continuously to test different component iterations.

However, recognising that this was leading to spiralling costs, the FIA began to place limits on the volume of off-track testing that could be undertaken, specifically targeting wind tunnels. Computer simulation packages also came to the fore, combining data collected on track, on test rigs and from CFD and wind tunnel testing to simulate the impact of set-up changes or new components. With the increase in complexity of these simulations, ever more powerful computing facilities were required to process the vast quantities of calculations required. But the constantly falling cost and rising power of mainstream computers have made it possible for even small teams to have processing power which even the top teams could only dream of a decade ago.

Track testing restrictions not only impact car development programmes but also have implications for drivers, both experienced and inexperienced. New drivers no longer have the opportunity to gain experience in Formula One machinery, while current drivers are left with only a couple of free practice sessions in which to acclimatise to new car developments. To counter this, driver simulators have seen widespread adoption to mitigate some of these problems. While simulators have long been in use in the aviation industry, machines capable of realistically representing the forces experienced in a modern Formula One machine have only recently become widely available. Tricking the human brain into believing that the body is experiencing lateral, horizontal and vertical accelerations without moving it any great distance is no mean feat.

While there have been simulators that work on rails, along which the whole driver capsule is accelerated, these require a lot of space and are limited to one or two acceleration events in a row, making them unsuitable for replicating the rapid accelerations and decelerations encountered in a racecar. Instead, most simulators (at least those used by Formula One teams) are now of the ‘six degrees of freedom’ variety. These place the driver – often in a replicated vehicle cockpit – on a platform supported by hydraulic actuators that provide forward, backward and lateral movement, as well as rotation around the x, y and z axes.

These simulators do not recreate the precise movements of a racecar; instead they trick the driver’s brain into thinking they are experiencing greater directional changes than are actually occurring. The key is to provide these cues in a subtle fashion. For example, if the platform tilts too much while simulating a cornering event, the driver will recognise that it is tilting, rather than associate the forces as those experienced during a high-speed turn. It is a fine line for simulator developers to tread, but when the correct level of cueing is achieved, time spent in a simulator will invariably correlate directly with performance on the track.

Beyond simple driver training, the accuracy of the latest generation of simulators means they can be used reliably as a development tool. The physics engines that determine car behaviour – dictating everything from steering feedback to grip levels and aerodynamic downforce – are fed real-world data, including aero maps and tyre models. The result is that the impact on performance of car set-up changes can be assessed without the car every hitting the track. As mentioned earlier, it is possible to do this using regular simulation tools, but the factor these tools lack is the impact of the driver on car performance.

The best example of why this could be a problem is where a set-up derived from a simulation should theoretically reduce lap times, yet on track the changes present a driver with handling traits that prevent them from exploiting the theoretical performance. Given the limited testing time teams are allowed, this equates to a substantial waste of resources. If, however, the team and driver can test changes in a simulator, eliminating those which will definitely not work, then time spent trackside can be dedicated to developments with a greater chance of improving performance.

To this end, it is not unusual for teams to have a dedicated ‘sim’ driver, who will be testing set-up changes as a Grand Prix weekend progresses. The caveat here though is that, as with any ‘virtual’ testing tool, the data generated from simulators is only as accurate as the data fed in. Therefore a careful eye must be kept on the correlation between simulated and on-track results. As long as this condition is taken into account, simulators assume a similar role to that of CFD, acting as a filter for developments that will not work and reducing the amount of resources wasted on blind alleys in either the wind tunnel or at the track.

Fig. 1 - Most Formula One teams now rely on simulators for both driver and car development (Courtesy of Red Bull Racing)

Written by Lawrence Butcher