A car design odyssey: Part two

A couple of weeks back in part one of this tale, I claimed it typically takes four years to develop a car from scratch, trying in the process to clearly define the start and finish lines of this exercise. But just why does it take so long? After all, Apple can pump out a miracle of multi-discipline engineering like an iPhone in a couple of years. A television is superseded by a superior model every 18 months or so. Why is the car industry so tediously slow? How does the time evaporate, sometimes creating rather stressful situations like the one described above?

Let’s start at the beginning. Even the most ‘conceptual’ phase of car creation takes more time than you might think. To take the car from a designer’s sketch to a signed-off, production-feasible full-scale master model takes many months, and probably more than a year. We’ve talked before on Ti about the critical role of the packaging engineers – their work with the design team is, by nature, iterative: a technical game of tennis where the ball is constantly shuttling back and forth between the engineers and the industrial designers, getting more and more refined every time it sails over the net.

Next step – tooling. Let’s say we’ve signed off a perfect 1:1 scale clay model of our car. We’ve now got to turn it into physical reality, and that means launching a whole battalion of tools to produce the various parts. The largest, and one of the most important of these components is what’s known as the ‘body-side outer’. It’s that enormous one-piece panel that forms the inner front wing (or fender), the sill, the roof-rail, the pillars, and the rear wing on most modern cars.

In production, these panels are hammered out of rolls of high-tensile steel (or aluminium) every minute or so by gargantuan hydraulic presses. The press dies – the ‘heads’ of these gigantic hammers – often have the longest lead-times in the entire fabrication process.

These dies are themselves stupendous pieces of heavy engineering. For a typical family hatchback, the die will probably be the size of your living room: two huge chunks of steel that will be slammed together with the ground-shaking force of a 5000-tonne press. These press dies are not made of any old material. They are cast as giant blocks of ‘tool-grade’ steel – a very pure, extremely hard form of the metal that will not wear out, even after stamping out millions of panels.

These blocks then need to be to be milled to the correct shape, micron-precise, to form the panel exactly as the body engineers designed it. But the tool steel is so hard that even diamond-tipped, liquid-cooled milling heads that create the die can advance only very slowly, nibbling the material away, sliver by tiny sliver. Just the milling stage will take probably four months – and that’s with the computer-controlled mills running 20 hours a day, seven days a week. After it’s milled, it still needs to be fitted with sliders, cooling ducts, ejection pins and a whole mess of hydraulic veins and arteries, before being polished and chrome plated. Making just this one tool will take probably five to six months of non-stop work.

As well as the time required to physically produce the tools themselves, we must also factor in global logistics. Most press dies are now produced in Asia – few ‘Western’ companies still even try to compete with the specialists in China, South Korea and elsewhere in the Asia-Pacific region. If you are building your car in Europe or the US, you will therefore need to ship the enormous, finished die to your stamping plant – hopefully by sea and not via eye-wateringly expensive Ukrainian charter flights.

Onwards. We’ve now got parts and we’ve duly assembled them into prototype cars. Once again, we run into highly inconvenient laws of physics – and geography. All modern cars have ABS/ESP systems. Turns out that these are devilishly tricky systems to test on high-friction surfaces like tarmac. The only safe way to tune them is on large, low-friction surfaces – and convenient stretches of those are created only temporarily, when lakes freeze over in places like Sweden, Minnesota, Hokkaido and other chilly parts of our planet.

They are so tricky they require at least two season-long test-calibration sessions out on the frozen lakes. Therefore, this one test alone spreads out over two winters – at least one calendar year. Yes, you can try to speed things up (if you’ve not blown all your cash on hiring the Antonov) by carrying out some testing in the Austral Winter down in New Zealand or Chile, but take my word for it – most projects end up taking two calendar winters anyway.

Other, more mundane tests also take frustratingly long. Red paint. Not exactly a new invention, granted. But each new formulation needs to be tested for UV exposure. Red is notoriously hard to get right – next time you are in southern Italy or Spain, check out the paint fading on red taxis. Exterior paint must be tested by exposing it to the sun’s rays and there is no way around it: not just a few weeks, but months on end, either in a real UV-intense environment like the Arizona desert, or in test chambers set up to simulate such places.

Chemical processes like this cannot be accelerated – testing them properly takes the time it takes. Another example is EV battery degradation. A battery is not like a suspension joint. It does not care how many ‘miles’ it does or mechanical ‘damage’ it sees. You cannot just speed testing up by doing more miles, or by increasing the stroke or speed of the physical inputs. Batteries ‘care’ about charge-discharge cycles and heat – to properly test an EV battery, you simply have to put it through many thousands of such cycles, at various temperature conditions. There are no short cuts or time machines to rent here, alas.

Even if we’ve steered clear of the long-drawn-out testing of EV batteries by sticking with a good ol’ ICE powertrain, we hit another snag. Increasingly, governments want proof that the emissions from ICE vehicles are what the manufacturers claims them to be – not just when they leave the dealer, but for the lifetime of the vehicle. In the US, for example, this translates into a mandatory 150,000-mile emission test. The car maker must run a 100 per cent production-representative car (and no, you can’t cheat by running 10 cars in parallel for 15,000 miles each) to prove that the exhaust emissions are within declared tolerances at the end of the test. Try as you might, with round-the-clock relay teams of test drivers, you will not be able to do this test in fewer than four months – more realistically, six.

In the interests of space, I’ll surf over the myriad other tests – crash, chassis and body durability, EMC testing (see the lead image), extreme weather and high-altitude – which inexorably absorb the months and, eventually, years. Somewhat ironically, we’re increasingly seeing that it’s not these ‘traditional’ tests at all that drive the actual launch dates of new cars, the time required to get vehicle software right: witness the delayed launches of the VW ID.3 and the Mustang Mach-E…

To conclude – could all this be done faster? Sure, it could. The time required to develop a new car cannot be fully justified by the kind of long-duration physical phenomena we’ve listed above. Some time is doubtless lost to the natural conservatism and risk-adversity of most car companies, as well as the inefficiencies to be expected of any large organisation. But there is a physical asymptote – and it’s probably somewhere around about 24 months. It’s simply physically impossible to design, manufacture and adequately test a car – at least any car being built in reasonably high numbers and to any decent quality standards – in less time than that. So, please treat any claims to have broken development speed barriers with a dose of healthy scepticism.

Oh, and if you happen to be a desperate chief vehicle engineer who’s deep into the red zone on your timeline, do feel free to message me for the number to call to charter that wonderful Antonov. I’ve still got it here somewhere…