27 Oct 2016

The tricks of testing without testing

Firstly, thanks for stopping in as always, I’m sorry the website has taken a bit of a back seat this year but anyone that follows my work will know I’ve taken up residence over at Motorsport.com working alongside the one and only Giorgio Piola, although I’ve maintained the tech galleries after each race as always.  

As brilliant as the blog has been over the years it does take a considerable effort to stay on top of everything I do, which also includes work for PlanetF1.com and a day job thrown in the mix for good.  Both of these put some extra sheckles in my pocket, whereas the blog doesn’t, bar a six monthly payment from google ads for your kind clicks.  And that’s ok, as I never anticipated that I’d end up where I am now when I started this back in 2012.  However, I have been considering using Patreon as a method of rewarding yourselves with content in exchange for some of your cold hard cash. Perhaps pop some feedback in the comments section below if you think it’s something you’d like to see.

Anyway, onward with the good stuff in a piece that has been requested by fellow panelist on the Missed Apex podcast - Anil Parmer, who wants to know more about the rumoured Red Bull VTT (Virtual Test Track).  However, I can’t get straight into that and want to exorcise some associated demons first….

We all know that the on-track testing ban that came into force in 2009 has been a massive pain in the ass for the established big guns of the sport, with Ferrari and McLaren failing to secure titles in the years following.  However, whilst those at the head of the sport believed costs would be reduced if teams stopped pounding around their dedicated test venues, they couldn't have been further from the truth.  

Formula One is frankly an industry like no other, the more you try to cut budgets the more the teams will try and find ways to spend that same amount of money to gain the same if not more performance elsewhere.  The whimsical notion that cutting real world testing would reduce spend has been utterly fruitless and more money than ever has been spent on the simulation of these conditions at the factory instead, leaving us to question is one better than the other?

Of course, you can argue that this can be seen as a big carbon offsetting schtick and claim that the sport is being more responsible to the environment as it no longer carts a spare car off to a test location but, the expansion of the race calendar from 18 races, in 2008, to 21 races in 2016 slightly belittles that argument, especially as the calendar has expanded ever outward from the euro-centric home of the team's  (Ten European races, in 2008 vs Nine, in 2016, and that’s counting Baku as Europe even though it’s tenuous as to whether it is or not).

Ok, ok, I hear those in the back shouting that the teams are testing in-season once again too and whilst that is true it is not the same level of testing that was done nearly a decade ago.  These post race test sessions have been lambasted by many as a waste of time, money and resources and only attended because if they don’t it gives the other teams testing mileage they don’t get.  I’d argue they are relative and are actually the future direction that the sport needs to take in terms of marketing the sport outside the walls of current convention, attracting fans they might ordinarily lose, but I digress onto a topic I’d have to spend further time explaining.

Red Bull add kiel probe arrays behind the front wheels on the RB12 to collect data on how the front wing, nose, wheels etc interact with the airflow and will use the data to correlate their findings with the simulations conducted at the factory
So, factory based work that is being conducted primarily revolves around the simulation of conditions seen at the circuit but because it’s not a real-world exercise we often see teams having to apportion some of their Free Practice time on a Friday to correlation work.  This is when you’ll see teams run kiel probe arrays, flo-viz painted on the cars. infra-red or slo-motion camera’s mounted on the cars to establish deflection and of course more ride height sensors and pressure taps.  The information gathered during these runs is used as a baseline in order to improve the accuracy of simulation techniques such as CFD, wind tunnel models, mechanical simulation rigs and the driver simulator.  All well and good but still costly, in terms of creating these solutions, mounting them and the time spent on track dialing them in.
Scale model in Sauber's wind tunnel in Hinwil
One of the things that struck me as a cost misnomer back in the approach to the 2009 regulation shift was the change in scale for wind tunnel use.  Whilst not every team had gone to the expense of creating and running a 100% scale tunnel, some of the lead teams had made that investment, cripes BMW even invested in a space at Hinwil that could have two models running one behind the other, albeit scaled.  You can argue that the cost to run the a 100% scale tunnel has now exceeded the reinvestment needed to rescale to the 60% model limit set out in the regulations.  However, what cannot be made up for is the results that can be achieved for said scaling difference.  Furthermore, there is the initial outlay required to retool and create these models, although you’ll hear the argument that over a sustained period the costs have been recouped.  I firmly believe this is why BMW, Honda and Toyota decided to step away from the sport.  These three teams alone had sunk a monumental amount of money into the sport at this point, especially in terms of the investment in wind tunnel facilities, only for it to be pissed on and the rules changed so that they’d need to spend more.  This can perhaps be seen as a bridge too far in terms of getting it passed a board of directors and shareholders, when the global financial system is already in meltdown…

Anyway I’ve got off topic again, perhaps that’s what these blog posts can be about, a more irreverent look at the sport, rather than the melodic factoid style pieces I have to do elsewhere..

Ok, so point proven about costs? It’s a never ending chicken vs the egg scenario where overlap in terms of investment is crucial to exploiting regulatory change and is certainly something we’ve seen Red Bull and Mercedes exploit at the last two major turning points, in 2009 and 2014.  With another change literally on the doorstep we are undoubtedly going to see another investment in infrastructure come to the fore and likely be key in the performance of the adopting teams results.

The main purpose of this article could be just that, with rumours suggesting that Red Bull have invested in what has been dubbed as a VTT, which is what I’ll abbreviate it as for the remainder of the article but stands for Virtual Test Track.  This is a story that was run some time ago by the German publication AMuS and even though I’ve asked the team to comment several times I’ve yet to have a concrete yes or no on whether it exists.  Hardly surprising…
Martin Brundle stands infront of a huge scalextric track which consists some of the best and well known sections of track on the Formula One calendar
Now you might be thinking, what the hell does a VTT mean? Or have images of a huge Scalextric track in an underground bunker somewhere but, it’s a little more comprehensive than that when we consider the other simulation techniques in play.  Furthermore, I think it’s important to know that Red Bull’s VTT builds upon the foundation of technology already in use amongst the other teams and has been earmarked as part of the initial success enjoyed by Mercedes since 2014.

Chassis Dynamometer

Those interested in the engine side of Formula One, or any motorsport or automotive endeavour may well already understand the importance of a dynamometer but, for those that don’t it’s imperative that an engine be tested under strict conditions before it even be mounted in a car.  It’s essentially a test laboratory for the engine, with a huge amount of parameters run and problems diagnosed in advance of a situation arising that would impact life or performance.

In terms of Mercedes, they understood that such was the complexity of the powerunit, with its various energy recovery systems, reliant on evolving track situations made it imperative that the powerunit and chassis divisions worked together to create a unified front.  Chassis dynamometers pre-date the relatively recent adoption in Formula One, with car makers using it as a way of improving the dynamic relationship of their cars and is perhaps something that Stuttgart instigated.

This original image is no longer available on their website but AVL Racing, who've been at the forefront of this chassis dynamomter technology are worth a visit if you really want to scrape beneath the surface of the VTT's I'm talking about.
The marriage of chassis and powerunit means that more accurate data points can be created from which the team then build their understanding of the race weekend from, even before they’ve got to the venue.  Software simulates the dynamic conditions the car would encounter at the track with the entire chassis mounted on a moving platform and 7 post rig (the latter much like the one used to test components in the factory and all-in-all very similar to the hexapod setup we’d associate with a simulator).  These simulation of these dynamic loads are what it's all about, understanding what happens throughout a corner, a sequence of corners, an entire lap and even a grand prix distance allows the team to plan in advance, optimising strategies and energy recovery and deployment, be it electrical or petrochemical, with engine modes refined to extrapolate more power or less energy consumption based on dynamic and transient conditions that the entire car is subjected to, rather than just the powerunit in isolation.  Think about it, the forces encountered during braking and cornering can be used to manage, recover and dispense energy.  Understanding the nuances of this can lead to improvements that make up the difference between winning and being a few tenths off the pace. 

Energy recovery background

The energy recovery systems used in the current crop of powerunits, coupled with the fuel flow restriction and fuel weight limit mean that the fastest way around a lap might not be the same as it was in the V8 era.  These energy and fuel figures combine to create an energy matrix for the lap, however, how that energy is created and deployed isn’t a fixed metric, with conditions on track and variability deployed by the driver creates a host of scenario’s.  That’s partly why the MGU’s energy deployment is left unlimited, with the 4Mj per lap electrical storage in the energy storage acting like a buffer, that can be topped up in combination with the energy being released and deployed between the two MGU’s (MGU-K and MGU-H).

The dynamic information of how the car accelerates, brakes and moves does mean that the energy models can be tweaked, something you’ll often here referred to as the ‘engine map’, these are pre-set modes selected by the driver from the steering wheel and provide different levels of fuel, energy deployment and energy harvesting.  A driver may choose for instance to use a more aggressive map to attack an opponent, which results in more fuel being used and energy being deployed throughout the throttle's application rather than at set increments.  Furthermore, in this high energy mode harvesting is reduced, as this may ordinarily hold the car back during or at corner exit.  However, running at this deployment level for long periods or at the start of a Grand Prix can deplete the battery store and cause what we’ve come to know as a ‘derate’, thanks to the circumstances surrounding Rosberg and Hamilton’s collision in Spain.

Understanding when and how to use these tools is all part of the strategic battle that goes on between the driver and the pitwall and is often overlooked as it’s not seen as overwhelmingly interesting for the casual viewer, whilst the graphic overlays used by FOM on the world feed are also quite poor at showing how it works.

Back to the Dyno...


The more the chassis dynamometer is used as a driving and setup tool as opposed to just being used to validate the powerunits life cycle and performance the more that can be got from it, with more complex energy schemes used to improve on-track performance.  This is where Red Bull seem to have stolen a march, with their fresh install of a chassis dynamometer reportedly paired to a new simulator that they had installed at the same time.  

In their case, rather than simply feed the chassis dyno with a set of logic based parameters they have their drivers ‘in-loop’ and driving the circuits in only ways they know how.  For this it may help to read an article I did after visiting Ansible Motion in Norfolk, as Kia Cammaerts, the company's owner, made a statement to me that has stuck ever since - “Drivers often do strange things, things that a computer can’t account for but are often better off for it.”  Of course he’s talking about driver nuances and styles, the things that us as humans do for no reason other than it is a trait that we’ve picked up and is how we often generalise the difference between F1 drivers too.  He was also referring to the non-linear way in which one driver may apply the brakes, throttle or steering input when compared with say the expected rationale or another driver's baseline.

As such, I can see why Red Bull wanted to include the drivers in their use of this going forward, as why follow Mercedes (and Ferrari who now have a chassis dynamometer too) when you can make progress with the technology.  As Kia rightly pointed out, having the driver as part of the feedback loop you get a direct correlation with what is happening with the dyno in real time rather than being reliant on data fed into the software simulation, even if this is extrapolated from data previously gathered in simulation sessions and real world data.  I’d assume it’s a two way experience too, as conditions and feedback ordinarily fed into the simulator by software, can be fed from the chassis dyno, improving the drivers correlation with the sim and real world environment and their understanding of how to maximise the setup of the car.

Has this had an impact yet?..

What stands out to me this season in terms of Red Bull is not only how they’ve managed to leapfrog over Ferrari toward Mercedes but the fact that they’ve done it with the minimum amount of aero work.  Ordinarily Red Bull are one of the busiest teams on the grid in terms of aero, with updates coming thick and fast.  However, this year has been quite the opposite, which of course you could argue they’ve opted to spend their limited CFD and Wind Tunnel time apportioned to next year's car rather than this but I think it comes down to more than just that.

Red Bull’s aerodynamic design philosophy is ingrained in their last 7 challengers (RB5 through RB12) and have all the hallmarks of Adrian Newey designed machines but, the last three cars have been heavily compromised, as they’ve had to marginalise their chassis to maximise the relatively poor performance of Renault’s powerunit.  Like Mercedes they’ve been working on areas of the car that run into the grey, with tyres and kinematics a prime example of where their respective cars outshine the rest of the field and again allowing them to close the perceived gaps.  However, since the use of the VTT I think where they’ve excelled is being able to define their limitations, rather than just thinking about setup, mechanical or aerodynamic, in a raw way.  For example, there is no point being able to achieve 201mph on the pit straight for one lap if that means you’ve destroyed your energy matrix for the next five and can only achieve 195mph.  You have to be smarter than that, pick your fights and know when the car, tyres and strategy allow you to deploy maximum attack.  I believe the VTT is a big part of that jigsaw puzzle, as it not only hones these setup choices before getting to the circuits but also gives the drivers an insight into what they can achieve and why.

The thing that really stands out to me about this method of testing is that whilst it seems like an extravagance and as usual there is a substantial outlay to purchase the equipment, it may be better than real world testing in some respects.  That is the cost is shelved at the purchase and upkeep of the equipment, whilst real world testing involves the build of new parts (imagine the cost of a front wing, that can be simulated for load just like in the simulator for predictive results), the crew to assemble and run the car at the given circuit and the costs to get it all there.  Furthermore, the team may be testing at a circuit that only covers certain conditions and not representative to an array of circuits that the VTT can accomodate (think of the straightline speed example I offered earlier, which is reliant on understanding energy deployment, something that wasn't such a moveable 'component' in the V8 era).  However, what's still needed is a comprehensive understanding of how to maximise this when at the circuit and adapting on site if necessary.

Conclusion

Have Red Bull stolen a march on their competitors by being first to integrate their chassis dyno and simulator? Most probably, as they're on the ground floor understanding how to use the data it creates from the get-go, understanding what's not right when they get to the track and resolving these issues as they unfold.  Is it a silver bullet? Who knows, the depth of technical expertise that goes into building and running an F1 challenger is mind boggling but we can't deny that most of the pace they've gained over the course of this season has come from a better understanding of setup and the powerunit, rather than their relentless aero programme.  Perhaps this'll change their approach to aero going forward too, although you'd argue that their wind tunnel has probably had the 2017 model in it for almost all of this season in any case.

3 comments:

  1. Every F1 engine manufacturer was using a chassis VTT well before red bull dreamed of using one.

    ReplyDelete