Showing posts with label Technical. Show all posts
Showing posts with label Technical. Show all posts

Monday, August 12, 2013

Race Strategy explained

Hi everyone, we are back on track with another exciting and exclusive installment of F1 Framework related topics. Judging from the stats, you have really enjoyed the insight on Engine Maps brought to you by Maurizio Bollini - he worked as engine engineer for Michael Schumacher when he was the Ferrari F1 driver, during 1996 to 2006 period.
Today, he goes strategic to explain what are the main considerations behind Race Strategy - an imminent element of every single start in Formula 1.


When explaining the basic concepts of the F1 Race Strategy, let’s consider two of the factors contributing to race pace: the tyre behavior (and degradation) and the weight effect of the car on the lap time. We are going through a very simplified demonstration.

Tyre behavior

A given set of tyres is used in a race stint. During the utilization in the stint, the tyres go through four phase of life:
  • warm-up phase (the firsts one or two laps)
  • performance phase (depending on the compound, from 10 to 20-30 laps), 
  • degradation phase (depending on several factors, it can be very few laps), 
  • Give-up (in very short time).
During the warm-up phase the tyre reach the optimal working temperature in its working range. The working range depends on the compound and other factors. Pirelli tyres for F1 2013 season have the following working ranges:

Low Working Range

  • Super Soft, 85-110 deg
  • Medium, 90-115 deg
  • Hard New, 90-115 deg (from Bahrain 2013 onward in the season)

High Working Range

  • Soft, 105-125 deg 
  • Hard, 110-135 deg
The performance phase is a window whose duration depends on the compound, the track characteristics and the car. For the softer tyres it could be somewhat like 10 laps, for harder tyres it could be in the range 20-30 laps. In this phase the tyres gives the best performance and so the lowest lap time.
The degradation phase occurs for wear, abrasion, graining and blistering and the lap time start to increase. It is time to make a pit-stop to change tyres.
The give-up occurs when the cyclic stress reached the maximum acceptable level for the compound and construction. The combination of stress and heating generate mechanical and chemical changes in the rubber causing the lap time increasing sharply. Usually the car is called at the pit stop before this phase because it can cause tenths of retard on the pace.

Weight effect (on lap time)

The weight-effect is a value that expresses how faster the car run while the weight decrease because of the fuel burned. It is measured as seconds/lap/10 Kg.

It can be estimated at the Lap Time Simulator (LapSim), an application running on a PC, with experiments at different level of fuel or just by fitting real data public available from FIA.
The LapSim gives, for Australia 2013, a value for Weight Effect of 0.22 sec/lap/10kg.

As alternative, the Weight Effect can be easily estimated by fitting real data. Let’s consider the real lap times from Australia 2013 F1 race from three drivers.


For comparison, if we take the best of the fitting cases above we get a Weight Effect of 0.0722 sec/lap. Knowing that the fuel consumption in Australia is somewhat as 2.5 kg/lap, the fitted Weight Effect will be 0.29 sec/lap/10kg, not so far from what we get from the more accurate simulation and not so bad considering the “noise” affecting the second measure.
Going through a normalized value per 10 Kg is not really needed for this purpose but it is important to compare the weight effect of different race circuits.

Race Strategy example

As a race strategy example, we want to compare two tyre management options for the Australia race, called Option-A and Option-B. Both options use the same sequence of tyres, SuperSoft/Medium/Medium.

Option-A do pit at lap 13 and lap 35. Option-B do pit at lap 17 and lap 38. The pit-stop time is the same in both cases (20 seconds) and the tyre model, very simple and just for the purpose of this demonstration, is based on the Weibull degradation formula.

We calculate the lap time lap-per-lap for both options on the race distance including the weight reduction effect and the tyre behavior effect.


Then we sum the lap times in order to have a cumulative curve for Option-A and Option-B and we make the difference, lap by lap, of the last two cumulative curves. Results are plotted on the next chart.


By reading this chart, the conclusion is that at the end of the race, Option-A is slower of about 3.61 seconds respect to Option-B.
Using this procedure, it is possible to play with different scenarios. During the race, simulated lap times are updated, lap by lap, with the real lap time and the scenario update in accordance.
This is, in a very simplified way, what is behind a dynamic race strategy application.
This tool is used trackside by the Performance Engineers with the support of the Strategy Engineer at remote garage, the facility located at the Team’s Headquarter. Such example is Neil Martin from Scuderia Ferrari, who has been known to update the team in terms of strategy via remote link from Italy. 

Making the right strategy call is, of course, very important. The recent example was Mark Webber in Hungary, 2013 - he had to start P10, running for a long period on Medium compound and eventually making it to P4 with very late usage of the Softs. 


Maurizio Bollini is the owner of MET Milano (www.met.it), a consultancy firm involved in motorsport. In the past he worked as engine engineer for Michael Schumacher when he was the Ferrari F1 driver, during 1996 to 2006. He can be contacted at maurizio.bollini@met.it.



Thursday, December 13, 2012

Inside wind tunnels

In Formula 1 world there are three types of testing validation for new parts: CFD, on the track (free practices or tests) and Wind Tunnel installations.
Lot has been said about the latter, so I'm not going into details about rolling floors or boundary layer suction, but instead I'm going to focus on what and how specifically is being measured in a Wind tunnel and why is that of paramount importance for F1 teams.

One of them has recently been all over the specialized news feeds, namely Ferrari, with their Wind Tunnel struggles, failing to produce correct correlation between tested parts and what's actually being measured on the track as results.
Various sources have been suggesting that Ferrari F1 Team will shut the tunnel down and make re-calibration, system checks, improvements, even calling external consultants, presumably from aerospace industry to check what's wrong. Just how much of all these is true is unknown to the general public, but talks about new wind tunnel construction have also appeared. It is perhaps worth saying that this is large investment and its rough estimated cost is about 40 to 50 million US dollars, besides it takes a lot of time for pure construction and realization.
UPDATE [20.Dec.2012]: Stefano Domenicali confirmed that Ferrari's wind tunnel will be undergoing major overhaul, will remain closed until August and Ferrari's 2013 car will be developed exclusively in TMG facility - see details below.

Apart from the economic outlook, teams are allowed to test with scaled model, up to 60% of the real size, and use speed of up to 50 m/s, which is 180 km/h max or 111.8 mp/h if you prefer imperial units.

Since the model parts do not represent the real world car size, Reynolds number is, for example, used to make coefficient correlation between the 60% model parts and 100% model. However, it is important what the source data shows.

Below you may find a sample spreadsheet with totally bogus and unrelated data, which is there to simply demonstrate the various parameters being measured in a wind tunnel session. Time spent inside the installation is also precious, as teams are having certain time frames allowed.



 Let's break down the data sheet into details:
  1. Cell B holds the value for Lift, in our case there is downforce generated, hence the negative sign. It's usually being measured in Nm, but it may vary in different wind tunnels implementation (Kilograms are often seen, too). 
  2. Next we have Drag, which is pretty obvious.
  3. L:D is lift-to-drag ratio - one of the most used measures for aerodynamic efficiency - sure, you can add more downforce via Angle of attack settings, but this comes at the expense of added drag, too, so balance is important. 
  4. Drag power - this is the consumed horse power, the power required to overcome aerodynamic drag.
    Note: this is a velocity-sensitive parameter, and the formula is simple:
    DRAG HP = ( Drag coefficient (0.9 or 1 for F1 car) * Frontal Area in square meters (assume 1.35) * current velocity (200 km/h = 55 m/s) cubed ) / air density (1225 kg/m3) = ~ 124 HP to overcome aero drag. The same formula gives about 380 HP at 250 km/h, etc.
  5. The next three columns, FY, FX and FZ are moments, measured as distribution of aerodynamic forces around Center of gravity, aligned with the three axes:
    Copyright: F1 Framework
  6. Next, in columns I and J we have vertical loads on the front and the rear, which is very important parameter measured from suspension point of view. While measures here differ, these are usually Newton meters, in this sample sheet the values are in kilograms.
These are some of the most common parameters that teams are measuring in a wind tunnel. Apart from those numbers, more can be crunched as a mathematical representation of the measurements taken, namely calculating coefficients. All of those numbers and their values are carrying certain amount of knowledge in regards to how your car will supposedly behave and that's why it is very important that the output is correct, as this decides what configuration and parts to put on the car.

TOYOTA MOTORSPORT 

The next chunk of details goes directly into the heart of Toyota Motorsport GmbH wind tunnel facility in Cologne. F1 fans have heard a lot about it, especially when it comes to Ferrari, but first, I, with "no wax", like to thank their staff for being gracious enough to release those images and data for the public and being patient enough to explain everything. Below you may find some exclusive images from the tunnel.

Outside look of the tunnel's diffuser section
The instrumentation of this state-of-art facility consists of two wind tunnels : one full size and one 60% scales.
Both wind tunnels use a continuous steel belt rolling road which has a maximum speed of 70 m/s. What can really describe those installations is accuracy and high quality - 512 pressure measurement channels are available. The tunnels are also equipped with Particle Image Velocimetry system (PIV) - an optical method of flow visualization used to obtain local flow velocities. The PIV system is usually used to validate CFD predictions and thus better calibrate the setup.
One of the advantages of such PIV system is that its preparation time is almost zero, due to the seeding technology used. That means that the local air is filled with tracer particles and there's no cleanup needed after that. During the process of measurement, cameras are used to obtain images and samples, which are then processed by computer software.
A laser is used to illuminate the flow field around the car’s front wheel to take particle image velocimetry (PIV) measurements (see below).

One of the main advantages of this method, just as well as the PSP method, is that it's non-intrusive, as opposed to the traditional pressure taps.
"Many of today’s motorsports cars are based on existing, commercially-available cars." Frank Michaux, a researcher at TMG, reveals.
"If researchers can identify a way to reduce drag on a motorsports car, it’s reasonable to assume that this information also may apply to future versions of a normal road car." Frank continues: "We need to deliver data quickly. If you see that you are not capturing the flow correctly, then you need to adjust your CFD methodology until you get it right".

After gathering the raw data from the PIV measurements of the “separation point” on the front tires, engineers post-process the data using Tecplot software, which allowes them to see and measure the exact position of separation. Each of Toyota’s PIV measurements consist of 300 datasets, with each dataset containing two images taken 10–20 microseconds apart.

The Tecplot software package itself is widely recognized and used throughout the industry.

INTEGRATION POINT - PIV & CFD

During the 2009 season, F1 engineers wanted to carefully study the flow wake behind the front wheels. Since this is critical part of overall car performance, the goal was to push this flow as outboard as possible. Engineers have realized that they need precise back-to-back correlation, and here's a direct comparison between CFG and PIV images: 


In order to get to those precise points, the CFD methodology has to be optimized, and here's a real case:




The next thing about Toyota Motorsport GmbH is its Continuous Motion System.

In CMS mode, a user-defined programme of ride height, yaw, roll, steer and individual pre-load changes provides continuous motion on a predefined trajectory while the HSDA system is continuously acquiring data at high frequency.
This allows realistic road or track analysis and increasing the amount of useful data from each individual test compared to standard motion and acquisition systems.
The typical "Wind-On" time is reduced by 70 to 100%  with CMS, which allows more time to be dedicated to, for example, changing parts. This Wind-On time is extremely precious for F1 teams, which we all know run under certain restrictions when it comes to wind tunnel time. 

Below you may find some exclusive images from inside the tunnel (full sizes available upon click): 
Very good representation of scaled model compared to a man

Toyota TS030 early prototype version

The control and monitoring room


Testing preparation - Toyota TF110
Front wing flap angle at 27 degrees
Surveillance and monitoring  
Glad you have reached this point - I assume you were very interested or your scroll button was stuck :)
As usual, comments and questions are welcome.



Tuesday, December 4, 2012

The evolution of pitstops

The following article is originally posted by Marc Priestley (ex-Mclaren mechanic) at his blog and I have his written permission to repost it here, which I'm very grateful for. His, Marc's website is here - F1 Elvis.
Also, you can follow his Twitter account.



Fangio, British GP, 1950
If you look back at footage of pitstops in Formula One from the early days of the championship in the 1950′s, they’re an all together different spectacle than those of 2012. They appeared leisurely, disorganized and, in the case of refueling, downright terrifying.

Drivers got out of cars while tyres were changed using large copper hammers and fuel was poured into, and often over, the back of the vehicle. Well over a minute later he’d hop back into the moving car as it was pushed away by mechanics and rejoin the race. All this took place at the edge of the track, circuits had no such thing as a separate pitlanes and there were certainly no speed limits.
Now I’ve stood with a front jack in pitlane as a modern F1 car brakes from 100kph to a stop at my feet and it raises the heart rate I can assure you. Mechanics of the 50′s and 60′s faced a series of dangers that would put today’s health and safety officers into an uncontrollable frenzy and have the sport shut down for good. Thankfully Formula One’s moved on in every area and pitstops are no exception. Today’s world record breaking stops to change four wheels in under 2.5 seconds are impressive to say the least, but they’re the culmination of years of development, learning, and improvements in technology. Here’s how they have done it.

For many years we’ve had pitstops becoming faster and faster as the sport woke up to the reality that the pitstop or stops can play a real part in a Grand Prix outcome. Teams initially threw more and more people at the process and gradually, over time, made improvements to their equipment used for the job too.
Just as competitors will always find innovative ways to make their cars faster on track, teams from every era and every formula have found ingenious ways to shave time from their pitstops. Some have done it legally, some questionably and some downright illegally, but the fact that it’s an area where time can be gained over rivals means it’s also an area where they’ll use resources and expertise to minimize the time it takes to service their cars.

1983 saw a number of fires as teams tried all means possible to shove as much fuel into their cars as quickly as possible at a pitstop and back then it was often a form of highly volatile rocket fuel, making it perilously dangerous. Refueling was banned the following year and this brought the focus away from pitstops and into the ability to conserve fuel and tyres during a race.
1983, Keke Rosberg is on fire

In an effort to spice up the show after a ten year break, 1994 saw the reintroduction of mid-race refueling and inevitably teams very quickly began to look at ways of speeding up their stops once more.
Perhaps the most infamous example was the Benetton pitstop fire of the same season, which allegedly uncovered that the team had removed a filter in the standard fuel rig, designed to restrict the flow rate of fuel from rig to car. The increased flow rate would mean the required amount of fuel could be pumped into the car faster than their rivals and of course reduce the time spent in pitlane.
During my early years at McLaren we developed a system to shave tenths of a second off the refueler’s reaction time once the fuel rig had delivered it’s required quantity.
Where normally the ‘nozzle man’ would react to a standard set of lights on the rig telling him to pull off the hose when the fuel was in, we used a stethoscope type device to listen for the valve closing inside the nozzle itself.

Just before the required amount of fuel was delivered, the motorized valve would begin to ‘whir’ and close, a process which took a couple of tenths, and once shut, the lights on the nozzle would illuminate to alert the fuel man he could begin detaching from the car. With the operator listening through an earpiece for the ‘whirring’ of the valve starting to close, he could start to react and be pulling the nozzle off the car by the time the lights switched on, therefore saving valuable time.
In those days of course, refueling was almost always the limiting factor in a pitstop. Time was saved or lost in connecting and disconnecting the rig from the car and the reactions of the lollipop man and the driver in getting away from the box. Although everyone had to use the same refueling rig, pumping at the same rate of 12 litres per second, the same hose, nozzle and valve in the side of the car, teams tried simple ways to make the operation smoother, more reliable and ultimately faster. Extra handles were welded onto the nozzle for someone to assist in pulling it off, fluorescent strips stuck on the nozzle end for the lollipop man to spot when it was completely detached from the car and the same hi visibility vinyl stuck onto the car valve to aid the refueler in lining himself up.
The sport’s always pushing the envelope in terms of technology in all areas, but whereas groundbreaking developments of ten years ago in the ‘real world’ may now seem a little dated, Grand Prix racing moves at a considerably faster pace. Technology in Formula One supersedes itself often before new parts even make it onto a car, such is the rate of R & D, but until now pitstop equipment has been something which got attention only when time permitted within the busy schedule.

Today the focus is different. With refueling gone once more, pitstops are all about replacing four wheels and tyres as fast as physically possible. For the first time in years, F1 teams have a relatively blank canvass with which to create equipment and systems to complete the operation in the shortest time. Wheel changing technology was almost an untapped market if you like, there was no desperate need to change them in 3 seconds when refueling would always take at least another 4 or so. With FIA rules being comparatively open when it comes to pit equipment and procedures, it’s been a case of micro-analyzing every element of the stop to see what can be done to save time. In the past that might have meant finding the fastest way to work with the car and equipment they had, but today there’s considerable resource allocated to the project and so the car and any new equipment can be designed from the ground up around the necessity for the ultimate stop.
Many years ago, frustrated at the difficulty of working on one of Adrian Newey’s complexly designed McLarens, I asked him one day how much of the design brief is about making the car practical and easy to work on? His answer was simply “None”. For the first time in modern F1 designers have had to think, not just about making the car light and fast, but about how they can improve the way wheels are changed during the race, with the realisation that it now has a direct impact on the outcome of a Grand Prix.
Stub axles on the car have as little as three threads of engagement for the wheel nut, to minimise the number of revolutions it has to do in tightening or releasing the wheel and the axle tip’s rounded to guide the wheel into place. Wheel nuts themselves are retained within the rims to reduce the opportunity for cross threading and take away the possibility of one falling out of the socket. In days gone by it wasn’t an uncommon occurrence to see wheel nuts spinning off down pitlane during pitstops or pitstop practice as they popped out of the gun as the wheel was removed. Of course it was hardly ever an issue as a spare nut could be picked up and fitted long before the fuel man had finished his part of the job.
Like the stub axles, drive pegs on the car’s uprights and the wheels themselves are designed to guide them seamlessly together and avoid ‘pegging’, where the two butt up against each other and fail to engage.
The jacking points, particularly at the rear of the car, are specifically made to make it easy to slot the jack into place. Whereas it used to be quite easy to ‘miss’ the old, relatively small, lifting hook with the jack when doing at speed, now things are designed differently. Most don’t even use a ‘lifting hook’ as such and a large carbon ‘splash’ on the jack can be thrown in anywhere under the rear crash structure to avoid the need for two bites at positioning it.

When it comes to the equipment and tools used in today’s pitstops, the advances are somewhat staggering.
The days of four ‘standard’ wheel guns, a couple of compressed air bottles, two basic steel jacks and a lollipop seem like a distant memory in 2012. In car terms, that level of technology would equate to the F1 cars of perhaps twenty years ago, but that basic layout of the pitstop area was still in use by almost every team just three or four seasons back. If you walk down pitlane now, it’s all very different.
The gantries carrying air lines overhead to the outside of the box are complex and beautiful. Huge moulded carbon structures house lightweight, super flexible airlines, ‘traffic light’ systems to release the driver, cameras and electronic cabling connecting almost everyone involved in the stop. As each operation’s completed, the ‘system’s’ notified through either mechanics pushing a button, or an automatic switch. Once all four wheels are on tight and the car’s on the ground, the chief mechanic gets a light and once he’s happy, he pushes a button to give the green light to the driver to pull away.



Wheel gun men now use heavily modified guns from the originals. ‘High flow’ gun backs allow greater air flow through the gun itself and effectively spin the socket faster, around 9000 rpm. Until 2012, when it was banned on environmental grounds, teams used compressed helium instead of air to power the guns. It’s extremely low density again allowed guns to spin faster and therefore remove and replace nuts quicker. The guns of today have bespoke sockets to match the team’s own nut designs, lights to indicate to the user when the nut’s done up to the correct torque and buttons to signify to the jack man that the operation’s complete. Before the recent focus on pitstop speed, gun men would signify they were done by raising a hand in the air and when the jack men saw the two hands at their end of the car, it would be dropped back onto the floor. When the chief mechanic saw four hands and both jacks out of the way, he’d release it. Now the action of raising a hand 50 cm in the air is considered to take too long and so the flick of a switch or button on the side of the gun shaves off valuable hundredths, that’s how thoroughly things are analysed.
One of the very latest developments in wheel gun technology is an automatic direction change. Up to now mechanics slide across a ‘shuttle’ on the back of the gun once the nut’s undone to change it’s direction of rotation, before doing the new wheel back up again. But Rhodri Griffiths of Palindrome Sports, the man credited with most advances in F1 wheel gun technology, has devised a new system which switches direction automatically as the gun’s removed from the first wheel. Whilst it may not directly save time in a standard stop, it’s a device which removes the need for a human operation and therefore one less thing to think about during the two and a half seconds or so that the car’s waiting in the box. Teams carefully study human performance in pitstops as well as that of the equipment, so anything which can simplify an action as well as speed it up, is worth looking at.

There’re teams who have lasers mounted in the overhead gantry which direct two beams incredibly accurately towards each wheel position. The beams cross each other at the exact height from the floor and for and aft position of the wheel nut when the car’s stopped on it’s marks. It enables the gun man to hold the gun ready at the exact height, avoiding any vertical adjustment to his position as the car comes in and hopefully the driver stops right on his marks to avoid having to move left or right.
Front and rear jacks have had a lot of thought put into them recently too. Often now made from carbon fibre to make them light and quick to move into position, they all have a quick release mechanism operated by a lever on the handle. It’s quicker than the motion of raising the jack handle to drop the car on the floor. When the lever’s pulled, the lifting arm of the jack simply breaks away, dropping the car instantly and it’s ‘reset’ ready for the next use. As someone who’s sat in a car for pitstop practice before, I can tell you it’s not a comfortable experience being dropped off the jacks with no comfy cushioned chair and very little in the way of suspension, but ultimately its faster and that’s what counts.
The jack men are looking for two lights from each of their respective gun men in order to dump the car on the floor and teams house those lights in various positions on the gantry, on the jack, or in some cases now, in ‘heads up’ displays inside the jack man’s crash helmet visor. Another example of trying to minimise human reaction times and remove opportunity for error wherever possible.
When the jacks drop they trigger another light for the chief mechanic and if pitlane is clear, he can illuminate the final light for the driver to pull away. Some teams now use a separate spotter to keep an eye on the pitlane, so that the chief mechanic can simply watch his own car and only when the system sees the ‘pitstop complete’ light AND the ‘pitlane clear’ light, does it release the driver.

When the whole procedure’s written down it seems a vast operation and although the bottom line is that it’s just a case of changing four wheels, when you look at it more closely, it is indeed complex. Each of the individual elements are studied using video analysis, onboard car data and of course the good old fashioned stop watch, although these days most teams use their own computer based systems for timing and so even the stopwatch has been replaced to a degree.

One interesting development in the pipeline is an automated jack release system. McLaren have been evaluating for some time, but are yet to race, a jack which drops the car upon receiving an electronic signal from both wheel guns, removing yet another human reaction delay and perhaps saving another tenth or two.

Crews practice over and over and over to the point where, whatever your role in the stop, it becomes literally second nature. Drivers don’t get the chance to rehearse so much and we do still see car’s over shooting, coming in too slowly or too far off center  but mechanics practice for those situations too and normally deal with it without most people noticing.



Throughout F1′s pitstop history they have always played a direct role in the outcome of races and as such, the speed with which they’re completed is crucial. For the ultimate stop, whether it’s tyre changes or refueling, each consecutive action has to preempt the smooth completion of the previous one and that’s where things normally come undone. We’ve all seen lollipops lifted as the fuel nozzle begins to detach from the car, only for it to stick slightly and end in disaster. Today, with stops lasting less than two and a half seconds, there is no room for error or glitch. Everything happens so fast that there isn’t often time to react to anything out of the ordinary, and just as through practice everything’s second nature, you’re conditioned for perfect stops and perfect stops only. Add to that, the pressure that I know only too well and that each of those guys feel when the spotlight’s on them and there’s an incredibly fine line between the perfect stop where all four guns operate perfectly in unison, and complete disaster. Talk of automated jacks, guns and lights is fascinating in terms of the technological advance and perhaps even the ‘show’, but it’ll undoubtedly have many squirming with discomfort at the prospect of the pitstop becoming too fast for humans to react to a problem and the potential safety issues that may cause. At the moment FIA regs are fairly open in this area, but just as with the cars, surely it’s only a matter of time before they step in to restrict spending, close up the field and safeguard everyone involved.


Tuesday, September 25, 2012

PSP in F1

Hello and thanks for tuning in on this frequency again.

The following article comes exclusively with the help of the kind people from ISSI - Innovative Scientific Solutions Incorporated. Their commercial site is located here.

We are not going to talk about PlayStation Portable, as the headline hints, but for Pressure Sensitive Paint (henceforth abbreviated PSP) - a method which stands between traditional F1 development paths - wind tunnels and CFD (but not a replacement for any of them). This is something that has been in use for quite some time, especially when it comes to NASA and airplanes.

WHAT IS PRESSURE SENSITIVE PAINT

In short, this paint-like coating fluoresces under a specific illumination wavelength of incident light and the fluorescent response is a function of the external air pressure being applied locally to its surface.
A typical PSP is consists of luminescent molecule and a polymer binder which must be permeable to oxygen.

There has been some use and some interest by certain F1 teams in using PSP in their own tunnel testing. 30m/s is typically the lower limit of PSP due to smaller pressure gradients below that speed. Most use PSP to validate CFD results or vice versa.

Imagine that in the end you could receive the same pressure distribution picture as you normally get from a CFD simulation run:
Image credit:  http://www.psp-tsp.com/

WHY IS PSP ADVANTAGEOUS 

A key advantage to traditional experimental techniques like pressure taps and transducers:
  • cost savings 
  • not limited by model geometry
  • provides much higher spatial resolution than traditional methods
Essentially you'll have a "pressure tap" at every pixel of your camera. So if you are using a 1-megapixel camera, that's like having 1 million pressure taps on the surface. Once the experiment is set up, many runs at various conditions can be made rather quickly by comparison to CFD and data turnaround is much quicker as it can be processed on site with some knowledge of the test conditions and local pressure taps on the model if available.
Paint formulations have also been developed recently which allow for unsteady measurements of pressure using a high speed camera. Measurements can be made upwards of 10 kHz on the surface.

THE PROCESS

The typical process could be described with the following simple steps:
  1. Painting: Whatever the testing object might be, the usual paint gun or airbrush could be used. 
  2. Excitation: The molecules inside the paint have to be excited, so there is a light illumination source applied to the painted surface.
  3. Data gathering: The CCD camera kicks in collecting the fluorescent response from the illuminated surface
  4. Data visualization: Different software packages could be used to visualize what's already being recorded and thus used for analysis of pressure gradients. 
Two immediate questions have arisen, fortunately Steve Palluconi, a research engineer, was available for this short interview: 

Question: Due to its (PSP's) spraying technique - would that be too much disturbance of boundary layer or not at all? Negligible, maybe?
Answer: Generally the layer thickness is 20-30 microns and is very smooth. Negligible for these types of tests. Pressure taps are more invasive as we’ve actually seen flow separation caused by them.

Question: Are there any estimates on cost? For example, would it be too expensive to implement or improper, for example, due to model scales (in F1 - up to 60% of the real car size)
Answer: Costs depend on the scope of the test and size of the model. Large models like the one referenced would usually be imaged with a multi-camera system. Smaller models can be imaged with a single camera. We've done testing on some large aircraft models.

Should you have more in-depth and scientific questions you can always contact Steve through their web site.


Monday, June 4, 2012

F1 factories

The following article is aimed at describing the 'average' Formula 1 factory and the processes happening behind the walls of these buildings. I'm even going to talk about specific vendors and introduce you the solutions that they offer to F1 teams.

When we discuss F1 teams and the way they engineer/produce parts, we should note that some choose to outsource the car's development to a different extent. For example, the brakes for Mclaren MP4-27 are provided by Akebono (since 2007).
Image credit: akebono-brake.com

Certainly, Mclaren don't have to invest time in creating that component, since there's already a good product on the market. That's the case with Brembo as well, which is provider to 6 of the teams. Nevertheless, that doesn't mean each team gets the same product - instead, there are custom solutions, "tailor-made", as they say.
Also, we are not going to talk in depth about the case where the respective team manufacturers its own engine, gearbox and KERS inside the factory, since it's only Ferrari (and former Toyota) who choose to build entire car (100%) at their own factory.

What's inside a typical F1 facility

Before going into details, we should note that different factories will vary in what type of machinery is inside, but most are having design offices, fabrication and assembly area, wind tunnel, CFD center, autoclaves (Red Bull have two, for example, the bigger one is from USI) and supporting areas.
Often you will find that some parts are produced with high precision in isolated areas with no external access whatsoever (3D printers), exhausts are being welded behind closed walls (single pipe takes about 40 hours, one exhaust set lasts around 1000 race kilometers, material is Inconnel), but sometimes you will just find screwdrivers hanging around together with boxes of bolts close to adjustable jigs - purely human environment.
Enstone Virtual tour, Renault F1 engine

Rapid prototyping / stereolithography

3D Rapid prototyping is very intriguing process where scaled-down parts are produced in shorter periods of time than the usual carbon fiber components being 'baked' in autoclaves.
Those type of machines allow the teams to produce testing parts (for the wind tunnel) directly from the CAD design, i.e. you draw your advanced geometry feature on the new front wing for the next race, send it to the printer where the new front wing is produced in just a few hours, as opposed to the real one, which can take days. Brake ducts, winglets, full-scale wind tunnel models - literally anything can be created. The pure magic happens inside that 3D printer where the object is created by laying down successive layers of material (epoxy resin) via ultraviolet laser. 
Here's a short video from NASA's tool lab, printing a wrench:


Essentially, this is a technology that is becoming more and more popular in the recent years, even though, depending on the size and the purpose, a complex of such machines can cost several million US dollars.


Another interesting F1 example would be creating a wheel rim with rapid prototyping methods:

Wind tunnels

A lot can be said about the aim of the wind tunnels and their leading role in aerodynamic analysis, but most of the talk will be self-explanatory, so let's just watch this video, courtesy of Sauber F1 team (HD):


Let's recall that no wind tunnel testing may be conducted using a scale model which is greater than 60% of full size, as the rules say, just as well as: No wind tunnel testing may be carried out at a speed exceeding 50 metres/second, which is 180 km/h or respectively 111 mph for those who prefer Imperial units.

A company named "Wind Shear inc" is citing the following interesting facts about wind tunnel anatomy here:

  • The air in this wind tunnel design flows from the fan to the vehicle, then is collected and returned to the fan in a closed circuit
  • The circuit will cover an area of 160,000 square feet
  • It will take an estimated 20,000 tons of steel and 2,000 cubic yards of concrete to construct the circuit
  • The main fan has a diameter of 22 feet, and is rated at 5,100 hp
  • The fan is capable of producing a maximum air speed of 180 mph
  • At maximum air speed, the fan produces an air flow volume of 2.85 million cubic feet per minute, and its total power consumption is 7 megawatts (one megawatt is equal to one million watts)
Finally, a word about costs - an average wind tunnel installment can be estimated to tens of thousands of US dollars (40 to 50).

CFD / Computing power

Generally, F1 teams will use the Computational Fluid Dynamic (CFD) as primary step for a new design, which would allow more flexibility in the early testing process. Certainly, most teams are using CFD in conjunction with wind tunnels to correlate the produced data, but some have tried CFD only approach, such  as Virgin with Nick Wirth. However, it turned out that CFD is not a sole replacement for wind tunnels, simply because it relies on approximations and assumptions by solving Navier-Stokes equations. Therefore, exhaust gases and turbulence are hard to model with CFD. Still, as soon as you have the results (having created the mesh prior to running the simulation, often a time consuming task) you can visualize the flow.
In order to get those results, you need a huge computational power, if you want them as fast as possible. Let's drill down into some technical details.

A research made by students from Brigham Young University a while ago is stating:
The CFD analysis was performed with the help of the BYU Fulton Supercomputing Laboratory (FSL).
The BYU FSL contains 9592 core processors and a total operating memory of 27.1 TB. A simulation took approximately 22 hours and 30 minutes to reach the set.
More on that subject follows, again from Sauber and the vendor that supplies their super computer, DALCO, the information is from 2009, but the idea is to grasp the ballpark figures:
The system, based on Intel Technology with a total weight of 21 Tons, was already one of the most powerful supercomputers in Formula 1 when it was launched. Albert 2 featured 256 compute nodes, each with two Intel Xeon 5160 dualcore processors, which gave a total of 1024 processor cores. The capacity of the main memory was 2048 GBytes and the maximum compute power was 12,28 TFlops (12.288 GFlops). An extension of 32 more compute nodes to a total of 288 nodes or 1.152 processor cores was added soon afterwards.
Now, BMW Sauber F1 Team has launched the next step by extending the existing system. A further 384 nodes, equipped with Intel Xeon E5472 quadcore processors (four cores per processor) and related Intel technology where added to the existing system so that the new supercomputer, Albert3, now has 4224 cores. The main memory grew to 8448 GBytes and the peak compute power is now at 57,7 TFlops, that's 50,700,000,000,000 arithmetic operations per second.
This is Albert 3, the super computer that runs the CFD software from ANSYS - the same vendor that Red Bull is working with, too. Pat Symonds says that the governing body sets limits on what can be used with CFD technologies, and the max peak is 40 TFlops.


Here we can add the numbers for Toyota Motorsport facility and services in Cologne:
  • Up to 80 million hexahedral cells making up a complete vehicle model
  • 600 CPUs and 1,200 cores cluster
  • With a typical full-car model of 60 million hexahedral cells, calculation, including automatic generation of post-processing movies, can be performed within 24 hours and three cases of this size can run simultaneously.

Lotus, for example, are using CD-ADAPCO and a system provided by Boeing Research and Technologies. Still, let's not forget that even though you have supercomputer at your disposal, the price and the number of calculations are limited in order to fit into the Resource Restriction Agreement (RRA).

On a related note, it is my guess that the teams are also using pressure sensitive paint in the wind tunnels in order to have a backup for CFD results, but at this point I have no solid evidence about that.

Design office

In general, design offices in F1 factories are the usual places with many people in open area sitting close to each other behind at least two large computer monitors, working on CAD designs.

For example, Red Bull Racing's solution for Product Lifecycle Management comes from Siemens.
They say it's essential to have good tools when you want to assemble about 4,000 pieces and make a race car out of them. 
A sneak peek inside their design office (over engineer's shoulder):
And a shot from the assembly area (race bay):
2010, RB6

Suspension, stress and other rigs

Some factories have suspension rigs where they simulate the loads coming from the track and forces acting on a car - downforce, pitch, roll, etc. Usually, these hydraulic rigs transfer collected data from previous races and then it's being replayed in order to fine tune the different components. The vibrations and the pressure that chassis and individual components take is enormous (simulating real bumps with high speeds), but this is not unusual for a stress test. The results are collected from sensors and subsequently displayed for analyzing potential weak points.
The potential cost of some of these machines is in the 100 - 200,000 US Dollars range.

Autoclaves 

Unlike the usual sterilizers that are used in the medicine, the typical automotive autoclave acts like a giant pressure cooker where the carbon fiber parts are being 'baked' with temperatures about 140C. Prior to that, the carbon fiber comes at the form of a sticky cloth, which is then laid over the respective part. The whole process could take about 7 to 12 hours.  
In order to get a good grasp of the sizes, here's a picture from the autoclave in Mclaren Technology Center: 
Photo credit: theroom.ru
These large ovens (7-8 m2) are one of the least expensive parts in F1 factory - around 1 million US dollars considering that the teams that build their own engines are investing lot more.

Here's another shot, this time from Red Bull Racing factory at Milton Keynes:

Simulators

Simulators are another type of extremely sophisticated machines that you may find in F1 factory. Usually they are built to recreate to a maximum extent the driver experience while driving the car, sitting in a real racing tube.
More on the subject comes from Ferrari: 
Photo credit: blog.axisofoversteer.com
This is a picture of their simulator, a giant spider, created by the company "Moog". You can read the entire press release here involving words from Marco Fainello, Head of the Car Performance Department, who says:
The dynamic driving simulator completely meets our specifications and expectations for a system that can test car designs as well as train drivers. Working closely with Moog during the two years of development on this system has helped us realize the maximum benefit from high performance simulated motion control.”
The simulator cost can really vary, but let's say that a number of 3 million US dollars would not be unusual.
In the meantime I was reminded by my Polish friends from F1 Talks that there's a video footage of the simulator in action:


Clearly you can see the red prancing horse tube inside the spider's cabinet.
Further explanation from Moog's engineers:
The test pilot is seated in front of a screen providing a viewing angle of more than 180°. Ten multiprocessor computers control the system with a total 60 GB of RAM producing around 5 GB of data per day. It features a 3,500 Watt Dolby Surround 7.1 sound system

Other stuff inside a typical F1 factory is paint shop (more further down the article), large CNC N-axis milling machines for chassis creation or engine parts (example is HURCO, used by Williams), metal engineering machines like these ones by Mazak:

Photo credit: http://theengineer.co.uk

Another interesting example is the partnership between Mitsubishi and Sauber. In the picure below we can see a wire-cut EDM machine, which is being used for rapid prototyping parts:


or a sneak look around in Milton Keynes, RBR factory
Photo credit: CNN

  • Painting
    The paint on an F1 car is around two kilograms, and we already know that each kilogram costs around 0.05secs per lap (depending on the track). Apparently, the paint department can't afford too much paint while the rest are trying to engineer low-weight components. 
    When it comes to time frames, painting the chassis, for example, is a very lengthy process - it can take about  three full days! You can imagine how much time does it takes to build/assemble a completely new car from scratch. 

    One of the most prominent partnerships in this paint department is between Mclaren and AkzoNobel - a move mostly driven by Ron Dennis in 2008. Back then, he decides to pursue a unique silver/chrome appearance for Mclaren, and this turned out to be a great challenge for painting company. Not only they managed to reduce the weight of the paint, but the base coating was no longer required and the chroming process took only four and a half hours instead of six. 
    The next speed performance gain was achieved in the drying process - a normal road car bonnet is cured in about 40 minutes (at 60C), while the high-tech ultra-violet LED gun by AkzoNobel does the same in just 7 minutes. 
    While this may seem fairly marginal interval for us, the mere mortals, in F1 speed matters a lot - both on the track and in the factory.


    I'm glad that you made it all the way down here. The sources of that article are purely available to the public, though some a bit more cryptic, so you just have to know where and how to search.
    Some special people have contributed for the accuracy of that article, and I'd like to take the chance to thank them, although they prefer to stay comfortably hidden.

    All multimedia and facts are used under "Fair use" doctrine for purely educational purposes and best efforts are made to credit the original owner. 





    Monday, April 9, 2012

    F1 car setup

    Hello and thanks for tuning in.

    The goal of the following article is to give an overview, as well as two interesting pictures, to the reader in regards to Formula 1 car setup - and a rough idea of the complexity that every team/driver has to cope with.
    Note: Some of you may be more advanced in F1's technical matters, so to some extent you may be familiar with the information below. But I'm sure you'll like the pictures.

    Probably quite often you hear from a driver after the race: "For some reason, we couldn't get the balance right with the current setup and I was struggling on every fast|slow corner".
    Conventional racing wisdom says that the car setup is a balance between different things, but sometimes could be a trade-off, too, for example, if you want to overcome certain shortcomings of your design, e.g. running "more wing" to compensate for lack of downforce.

    Let's dive into the details about on-track settings. 
    • Tires - A bit aside from tire optimal working temperatures, depending on the compound (90 - 125C), a word about the pressure. Usually, they (the tires) are filled with a special, nitrogen-rich air mixture, designed to minimise variations in tyre pressure with temperature. The mixture also retains the pressure longer than normal air would.
      The tire manufacturer would provide the teams with a limits for variation, so it will be up to the race engineers to decide, because incorrect values may simply ruin the car's performance.
      A typical average pressure number would be 19 psi.
    • Suspension - There are couple of settings and terms I'd like to highlight:
      Camber, caster, toe, rear and front ride height and rake.

      - Camber - That's the angle between the vertical axis of the wheels and the vertical axis of the vehicle when viewed from the front. In a simple picture, that looks like this (negative camber):
      Ferrari F150 - Negative Camber Tire setting
      (click for larger image)
      Photo credit: http://ferrari.com/

      Negative camber merely means that both wheels are inclined inwards at the top, as seen above on Ferrari F150. If you are looking for a typical ballpark number, that would be for example 3 degrees, i.e. the wheels are inclined inwards by 3 degrees compared to the center line.

      - Caster - That's the angle to which the steering pivot axis is tilted forward or rearward from vertical, as viewed from the side.
      Just have a look at any shopping cart wheel to understand what this is.
      As it can be seen from the final two images below, teams can use figures from 9 to 12 degrees, for example.

      - Toe - If you look a car from above, a pair of wheel can have their leading edges pointed to each other - this is Toe-In, whereas opposite - edges away from each other, that's Toe-out. Again, let's use Mclaren 's 2012 competitor, MP4-27, to demonstrate this:
      Mclaren MP4-27 (click for larger image)
      Photo credit: http://www.mclaren.com/mp4-27

      On the left, that's the axis showing Toe-In, only in case where both wheels are aligned the same direction (V like shape).
      Similarly, the opposite is true for the right axis, which is Toe-Out. (Like Lambda ( λ ), from the Greek alphabet).
      More details on what Toe setting can bring in the comment section of the post.

      - Rake, ride height - Both are related, but put simply, rake is the car’s attitude from front to rear. Such setup should, in theory, increase the diffuser exit area, and thus increase rear downforce, for example, but it's not that simple, because there are number of settings to take into account, like suspension geometry, overall aero setup, induced oversteer from stiff rear suspension, etc.
      Example or rake here (image link @  http://scarbsf1.wordpress.com Mclaren MP4-27).
      In both pictures below you can see real numbers for rear and front ride height (in mm.)
    • Engine maps/modes
      - Map is rather a high-level term describing variations of fuel strategies, ignition timing, torque settings and so on, whereas mode (or mix) is more like a setting for being inline with race demands, like fuel saving and proper amount of power.

      Examples of team communicating those are:
      Red Bull telling Vettel: " torque map 5 "is available", while Mercedes say to MSC:  "torque mode 3"

      It won't be unusual to hear race engineer on the radio: "Engine 2, mix 5" on the start - they use those values to communicate with the driver the proper settings throughout the course of the race.
      Example of engine software modification was the infamous "Off-throttle blown diffuser" employed last year mainly by Renault-powered engine teams.
      Small hint about how the things could be done here - software / automotive engineers will know the answer to that riddle.

      - Gear ratios - While real numbers can be seen in Arrows A22 picture below, we should note that this setting can really make a difference. For example, back in 2011 season it was quite evident that Sebastian Vettel had shorter 7th gear choice for Monza - a move which had its merit, namely for better acceleration out of the corners. It was a bit of a gamble, too, because Vettel was assuming that he would lead right from the start and won't need to overtake anyone. His top speed was affected, too, but it played well for him in the end.

    • Differential - Among one of the most important things that driver can control from the cockpit. An example of its importance is the end of the race, where the fuel levels are going down, and car starts to behave differently, so alterations are needed.
      On average of 5 laps there is at least one adjustment, though sometimes drivers are doing it from one corner to another.
      The settings are usually a numbers, just as Rosberg was advised to try "mid-corner diff 7 setting" in Valencia FP2. 
    • Angle of attack (AoA) - That's something pretty self-explanatory - it's related to the more or less extreme angles of a wing, for example, aimed to achieve respectively downforce or high-speed.
      The efficiency of a wing is its downforce/drag ratio - more downforce (or lift) typically comes at the cost of more drag and lower top-speed. The greater the angle of attack, the more downforce and drag.

      A while ago (2010), a mechanism for changing the angle of attack of the front wing was available for drivers, but in 2011 this was no longer allowed.

      Illustrated nicely with the following image, F2012 front wing:
    • More on aerodynamics is coming here, in the blog, as well as other places, in the next few months - not general aerospace, but Formula 1 related.
    • Brake balance / bias - Often adjusted by the driver from the steering wheel. During the race, the brakes can worn to some extent, for example, and in order to avoid instability during braking, the pilot can switch the bias from rear to front or vice-versa. In some cases the brakes can experience extreme overheat (thermal runaway), being loaded with values between 400C and up to 1000C.
    • Ballast - Usually plates from high density metal like tungsten steel. Often planted on places where balance and in particular weight in small available space is needed.
      One piece of it usually fits on your palm, and could weight about 11 kg.
    And finally to the promised pictures. I found the first one in Ebay a while ago, screenshot it, but the original URL is no longer present.
    It's a setup sheet of Arrows A22 - a car that competed back in 2001.

    "Arrows A22" car setup 2001

    And here's a more recent one. Prior to joining back the F1 madness in 2012, Kimi R. had to complete a "smoke-test" with an old Renault car, namely R30.

    Renault R30 - Kimi in pre-2012 private test
    Photo credit: http://f1news.cz

    These are real numbers from the notebook of his race engineer and that's part of the amount of data that goes into F1 car setup.

    Finally, there's a picture available with a team race weekend program, Toro Rosso were generous enough not to hide the sheet from curious eyes. While it's not exactly a car setup, it still gives a good overview of how many of the described permutations are being tried on Friday alone. The picture is from Monza, 2012:

    Larger version available, image courtesy of http://f1.f-e-n.net


    Thanks for reading all the way down. These are some of the important pieces that take place in a Formula 1 car setup. I'd love to hear more from you, in case I missed something significant, or just a general feedback.
    Cheers.