A couple of weeks ago, I wrote a blog post about creating a simple multibody system model of an active ballast system, found on the Mercedes-Benz DTM/ITC car from the mid-1990s. It was not the only highly innovative technical development found on the DTM/ITC machines of that era. Constant technical competition meant that as one marque broke new ground, so the others had to respond in kind. For the 1996 season, Opel introduced an active aerodynamic system on the front of its Calibra race car. Modelling of the system is not an immediately straightforward task but provides some interesting insights into efficient Dymola modelling of multibody systems. Especially as the Opel active aero system features primary and secondary elements.
Figure 1: Bearing it’s teeth! A 1996 specification Opel Calibra 4×4 rounds a bend with the lower active grille in the closed position. Image: Moritz21 (Wikimedia creative commons).
A system ahead of its time
In the present day, we are quite familiar with active aerodynamics. In fact, opening grilles are now found on many new road cars. Rear spoilers which automatically raise at speed for stability have been found on road cars for over a decade, with F1 using the Drag Reduction System (DRS) since 2011. But back in 1996, active aerodynamics were still at the cutting edge of technology. Not least because moveable aerodynamic devices had been outlawed in F1 since the late 60s; those devices were much cruder, manually moved wings.
The concept of Opel’s system was very simple. At the front, there were two air intakes – the precise geometry varied over the season – with a series of shutters, which would close above a certain speed to reduce drag. Once the car was below a certain speed threshold, where the drag penalty of the open intake was lessened, they would open to ingest air. Helpfully, the Opel teams would paint sponsor logos or designs on the shutters, meaning their action was easily identifiable from both trackside and TV. Likely this was to ensure they were able to see the system was working correctly, otherwise they risked damage to the car from overheating! Figure 1 provides a good example of the type of graphics used to indicate the grille was in the closed position.
Building the model: Inlet and shutters
As with many popular motorsport series, a devoted community has continued to maintain these cars, keeping them in working condition, appearing in demonstration events or in historic competitions. Austrian driver Alex Wurz’s 1996 Opel Calibra car is one such example, with a fascinating video taken of it recently running on a test day posted to YouTube. In the video, we get to not only see the system working as the car is at speed on the track at 3:48, but also as the car “undressed” (without bodywork) in the pits.
Without the front bodywork, the system is revealed to be a simple yet effective one; two Bowden cables emanating from a single actuator (most likely hydraulic judging by the braided fluid lines going to and from it) actuate a crank on each side air intake. Connected to the top of one of the shutters (master), the crank has a pinion gear which actuates a rack. As the rack moves, the other shutters (slaves) in the intake unit are actuated as they are mounted to the rack with pinion gears at the top, supported with a bearing at the bottom. Simple yet elegant, such a system is easily packaged with the inlet housing, compared to if it were say controlled with a 4-bar linkage with added levers for each of the shutters. Using Bowden cables enabled the hydraulic actuator to be located elsewhere in the engine bay at a more appropriate location.
Image 2: A still taken from the video embedded above. Both the Bowden cables are connected to a single actuator, not shown on this image.
First, a rudimentary vehicle model is required. The RoadsterSport example featured in the first part of this blog post was used again, but this time with a 4WD system to match that of the racing Calibra.
At the heart of the system, is the rotational action of a shutter, or flap. Translational rack movement from the rack input flange simulates the action of the rack moving, with the slaveGear component converting the translational movement to a rotational one, which serves to spin the shutter on its axis. Being a multibody model with no compliance, only the one support (bearing mounting flange) is required to hold the model in space (relative to whatever it is connected to) versus the two (top and bottom) in reality. Otherwise, the two supports will in effect be supporting the same degree of freedom, creating a kinematic loop.
Both the shaft models enable the gyroscopic forces of the shutter supports (top pinion and bottom bearing) to be included. A small detail but required to understand the precise control problem. A revolute joint connected to the flange of the shaft enables the bodyBox, which is the actual mass of the shutter itself, to be constrained to only be able to spin on its axis as do the central shaft models. Two versions of this model are made: a master and a slave.
The master shutter acts as the shutter connected to the Bowden cable by the shutter, with the slave positions driven off the position of the rack (in response to the rotational crank input) in the master shutter model.
Figure 3: Both the Master shutter and the Slave shutter are very similar, but the master drives the slave. Notice the side parameter in the Master revolute joint; this is to enable the Master shutter to rotate in the correct direction (towards the centre) despite what side it is deployed on.
Using component arrays enables the inlet mechanism of 6 separate shutters (1 master and 5 slaves) to be formed from only two models; reusing the slave shutter means only the two models need maintenance rather than 6. To further reduce model maintenance, a common flap template minus the rack and pinion actuation method (rotational crank versus translational rack) could have been created.
Each flap is connected to the single housing frame individually. Reviewing the text layer shows how, with loops of individual connections specified in the connect statement section. Included in this is the fixed translation which dictated the position of the shutters relative to the housing frame they reside in. As there are 6 fixed translations by virtue of the component array, the position vector can be supplied as an array of individual positions. A modifier linked to a real parameter at the top layer of the inlet model is used to enable it to become handed, by mirroring the y vector position; this is the same principle as the leftSide Boolean in Suspensions library linkages. Again, this promotes component reusage and lower model maintenance overheads, as it enables the single model to be used for handed applications.
Figure 4: A single inlet model comprises 1 Master shutter and 5 Slave shutters. A component array of 5 Slave shutters is created, with connect statements in for loops enabling those 5 plus the Master shutter to connect to the mounting translation and thus frame. Note, that for each Slave shutter to be actuated in turn, they must all connect to the translational rack position output of the Master.
Moving on the actual grille model itself, two inlet models are deployed to form the left and the right. A simplified actuator model is used to determine the rotational input on the crank. A 1D lookup table determines the rotation of the crank given the longitudinal velocity of the vehicle. This rotation demand then passed through a filter. Any non-linear system formation is broken (the shutters effect the speed of the vehicle through drag, with the speed of the vehicle then effecting the angle of the shutters in turn), as well as provides a rudimentary actuator delay. Like the real car, the single actuator provides a rotational input into the two inlets. Both inlet models report the position of the crank separately onto the control bus, emulating physical rotational sensors on the inlet cranks.
Figure 5: A common actuator provides a rotational input, just as the Bowden cables connected to the common actuator provided a rotational input onto the cranks.
Incorporating system level aerodynamic effects
With the rotational position of the cranks available on the control bus, the effect of the shutter position on vehicle performance can be modelled. For this example, a very simplified calculation reduces the drag force acting on the car body based on the shutter position.
Figure 6: Rudimentary estimation of the drag change due to the grille position. No change on downforce is incorporated, although it would be present in real life.
Enough for demonstration purposes, a wind tunnel or CFD data map of the vehicle aero forces (drag and downforce) would be used to apply the correct aerodynamic effect to the vehicle system. Precise tuning of the vehicle chassis setup could then be undertaken, as in some fast corners the shutter will be closed, but in slower ones it would be open. Tuning which otherwise would have to be conducted in a trial-and-error method on the racetrack. For this example, a simple acceleration and braking test was performed, with the grille set to open above 20m/s (around 45mph).
Figure 7: Comparison of the drag force as the car accelerates, then brakes. You can see where the drag produces diverges and then converges.
Complex systems can be broken down into small elements, with the concepts of model reuse being applied to reduce the modelling time and effort used to create a system model. It is remarkable that with modern tools, systems which were at the cutting edge of technology just a couple of decades ago can be quickly and easily recreated in the virtual world. Imagine the time that would have been saved on design and development of such a system if these tools were available back in 1996, and what a competitive advantage it would have been!
Written by: Theodor Ensbury – Project Engineer
Please get in touch if you have any questions or have got a topic in mind that you would like us to write about. You can submit your questions / topics via: Tech Blog Questions / Topic Suggestion.