F1 Testing data loggers

[turbof1]

Does anyone have an idea on how much data they actually generate over a lap? In terms of Gigabytes, or if needed, Terrabytes.

[zoro_f1]

I guess that the data is not very heavy... measuring in terabytes... i guess it is in bytes or megabytes only because they have software on their computers to read all the data... the hardware they are using on the cars only collect the data without any other software.

someone suggest that that hardware could be flash memory or ssd's... it is logical just to write that data but not to read without any software which is capable of dealing with all informations.

this is just my opinion... im not sure if it is this way... it is just my logical explanation...

[boyracer94]

Does anyone have an idea on how much data they actually generate over a lap? In terms of Gigabytes, or if needed, Terrabytes.

I'm not sure about a lap, but apparently about 27 terabytes is transferred over a race weekend, maybe more now

Source: http://goo.gl/61TcCB

[turbof1]

Does anyone have an idea on how much data they actually generate over a lap? In terms of Gigabytes, or if needed, Terrabytes.

I'm not sure about a lap, but apparently about 27 terabytes is transferred over a race weekend, maybe more now

Source: http://goo.gl/61TcCB

Nice. Thanks a lot!

[WillerZ]

Does anyone have an idea on how much data they actually generate over a lap? In terms of Gigabytes, or if needed, Terrabytes.

The majority of the car data is logged in the standard ECU which has a capacity of 8GB.

The extra thing taped on the wing could be something like this logger (32GB capacity) so that they can read from USB devices such as cameras. Or it might just be another car logger because they don't have enough spare capacity on the ECU logging inputs or it would be too difficult to cable-up. In any case the logger capacities give you a way to approximate the upper limit on how much data they can capture.

[Skett]

Well I'd imagine that when running additional sensors, as they have been, its probably simpler to set up a separate recording device than the all the standard car logging. So at a guess I'd say its just that

[Kiril Varbanov]

Does anyone have an idea on how much data they actually generate over a lap? In terms of Gigabytes, or if needed, Terrabytes.

Peak data transfer: 2MB/s in race mode. Data source: Dell.

For comparison sake: each engine of a jet on a flight from London to New York generates 10TB of data every 30 minutes. Source: Pratt and Whitney.

[acosmichippo]

For comparison sake: each engine of a jet on a flight from London to New York generates 10TB of data every 30 minutes. Source: Pratt and Whitney.

why specifically from London to New York? wouldn't the data generation rate be pretty much the same anywhere?

And I thought passenger aircraft jet engines were fairly old tech, so how would they have kept dozens of TB's of data from every flight? That seems barely doable even by today's standards.

[tommylommykins]

I have worked on commercial airliner jet engine software.

That 10 terabytes per 30 seconds number is fishy. That comes out to 5.8 gigabytes per second. There aren't many things that can record data to long term storage that fast, especially suitable for attaching to a jet engine. As a guess, Pratt has summed all the raw data rates from all the sensors built into the engine. I'd even seriously doubt that the Engine Controller would even be able to process that much raw data. CPUs that get put in planes are ruggedized and slow.

Plus, nobody has the money to sit and watch what an aeroplane engine is doing in flight, so there is no realtime data transmission at all. In fact, aero engines only bother saving data when something has gone wrong: It'd typically record an errorcode and save a snapshot of some other vital parameters for later analysis -- very similar to the sort of data logging you might get with a normal car engine.

[PhillipM]

For those mentioning the terabytes of data and transmissions in the gigabit/s ranges, how much of that transmission is onboard camera footage and how much is actual datalogging?
That seems a very large amount of info even for the amount of sensors and the sample rates on F1 cars if it's pure data.

[Pilatus]

I have worked on commercial airliner jet engine software.

That 10 terabytes per 30 seconds number is fishy. That comes out to 5.8 gigabytes per second. There aren't many things that can record data to long term storage that fast, especially suitable for attaching to a jet engine. As a guess, Pratt has summed all the raw data rates from all the sensors built into the engine. I'd even seriously doubt that the Engine Controller would even be able to process that much raw data. CPUs that get put in planes are ruggedized and slow.

Plus, nobody has the money to sit and watch what an aeroplane engine is doing in flight, so there is no realtime data transmission at all. In fact, aero engines only bother saving data when something has gone wrong: It'd typically record an errorcode and save a snapshot of some other vital parameters for later analysis -- very similar to the sort of data logging you might get with a normal car engine.

Two examples:
Flight Data Recorder for military medium tactical transport aircraft (around 30 tonnes MTOW) has 24 MB flash memory.
And for engine monitoring unit for turboprop engine (PT6), one 1,44 floppy is enough to download all stored data.

So, that P&W numbers are exaggerated. At least.

[beelsebob]

I have worked on commercial airliner jet engine software.

That 10 terabytes per 30 seconds number is fishy. That comes out to 5.8 gigabytes per second. There aren't many things that can record data to long term storage that fast, especially suitable for attaching to a jet engine. As a guess, Pratt has summed all the raw data rates from all the sensors built into the engine. I'd even seriously doubt that the Engine Controller would even be able to process that much raw data. CPUs that get put in planes are ruggedized and slow.

Plus, nobody has the money to sit and watch what an aeroplane engine is doing in flight, so there is no realtime data transmission at all. In fact, aero engines only bother saving data when something has gone wrong: It'd typically record an errorcode and save a snapshot of some other vital parameters for later analysis -- very similar to the sort of data logging you might get with a normal car engine.

Two examples:
Flight Data Recorder for military medium tactical transport aircraft (around 30 tonnes MTOW) has 24 MB flash memory.
And for engine monitoring unit for turboprop engine (PT6), one 1,44 floppy is enough to download all stored data.

So, that P&W numbers are exaggeretd. At least.

Most likely they're talking about how much data the ECU is actually processing, not how much is stored.

[Per]

I have worked on commercial airliner jet engine software.

That 10 terabytes per 30 seconds number is fishy. That comes out to 5.8 gigabytes per second. There aren't many things that can record data to long term storage that fast, especially suitable for attaching to a jet engine. As a guess, Pratt has summed all the raw data rates from all the sensors built into the engine. I'd even seriously doubt that the Engine Controller would even be able to process that much raw data. CPUs that get put in planes are ruggedized and slow.

Plus, nobody has the money to sit and watch what an aeroplane engine is doing in flight, so there is no realtime data transmission at all. In fact, aero engines only bother saving data when something has gone wrong: It'd typically record an errorcode and save a snapshot of some other vital parameters for later analysis -- very similar to the sort of data logging you might get with a normal car engine.

This is going way off-topic but Rolls-Royce is actually monitoring all of its jet engines world wide during each flight. All of the data analysis is automated obviously, the system only passes anomalies on to some engineers in Bristol. And like you said, the data rate doesn't come close to 5.8 GB/s (they only use 25 sensors and take just a few snapshots during each flight). Rolls Royce has some info here.

I wouldn't be surprised if P&W generate that amount of data during engine testing&development but obviously not during a normal flight.

As for F1 I find it hard to believe the above mentioned figure of 27 TB for a race weekend. If that's for two cars, and each car spends 4 hours on track each weekend, that's still almost 1 GB for every second of driving. The same article claims 200 sensors are on the car (seems a bit high to me, does anyone have other info?) so that's still 5 MB/s for every sensor. Conclusion is that those 27 TB surely are not only data logged on the car, or the figure is simply wrong.

As for the unit used on the Williams, the rake installed in front of the rear wing contains about 80 pitot tubes. So no surprise that they need extra storage capacity (it adds probably 50% of the amount of sensors they are normally using). An extra benefit would be they can go out and collect their data, come back to the garage, take the drive out and start transferring the data back to England immediately while the car can go out again for some other test. I can imagine the happy faces back at the factory if they get their precious data quickly.

[acosmichippo]

if we're talking storage of hundreds of gigabytes, it might be quicker just to fedex overnight a copied drive.

[Lycoming]

If you assume it takes 12 hours to fedex a terabyte drive overnight, thats 194 mbps. How many of you have internets that fast?