Uncheatable real world emissions testing was torpedoed by EPA and industry

RapidScreen Picture courtesy aircarecolorado.com

Vehicle emissions regulation is one of the greatest success stories ever to come out of activities involving engineers and governments: the air is vastly cleaner than it was in the bad old days, and it keeps getting cleaner even though many more people are driving many more cars much longer distances. Dieselgate has brought up some seldom-discussed topics such as how—or whether it’s even possible—to adequately and simultaneously control locally-relevant emissions (HC, CO, NOx, particulates) and fuel consumption and its resultant globally-relevant CO2 emissions. That overspanning discussion, though, is hobbled by a lack of good data on exactly what’s coming out of all those tailpipes. It’s not that the data would be hard to come by. For decades, there has been technology to produce mountains of it. The tech has been buried by the EPA and big industry.

Oh, sure, we have decades’ worth of new-vehicle emissions certification (US/Canada) and type-approval (Europe/rest-of-civilised-world) tests, but it’s now clear those readings don’t represent reality. Even if we ignore the out-and-out cheating lately come to light, automakers (all of them) have been “teaching to the test”, calibrating and tuning their vehicles for optimally clean emissions under the specific conditions tested, for as long as such tests have existed. Some (probably most) of them have sometimes (probably often) released cars calibrated for optimally high performance and/or fuel economy—and emissions be damned—under conditions not tested. At what point does that stop being legitimate compliance with the rules as they’re written and start being deliberate malfeasance? That’s a blurry, grey demarcation, but it’s the backbone of a conversation about how well or poorly the emissions standards serve their nominal purpose; how representative the test protocols are of real-world operation.

As more detail emerges about the defeat devices VW (et al?) put in their cars, it increasingly looks like the very idea of a test protocol simulating certain driving conditions is a cat-and-mouse game bound to collapse under the weight of increasing layers of scrutiny warranted by increasingly-devious cheating tactics. It’s unwinnable, an endless goose chase with the enforcers always at least a step behind, wasting money and time looking for outmoded cheats—think of U.S. flyers made to remove their shoes at airport security.

Technology exists to measure vehicle emissions with far less opportunity for cheating, and it’s not new. Dr. Don Stedman is the world’s expert on remote measurement of vehicle emissions—he should be; he invented it in the mid-late 1980s. This 1990 article is illustrative, and Stedman’s 1995 lecture on the science and politics of it all is very well worth a read. Most of a quarter-century ago he was spearheading a pilot deployment of a roadside emissions scanning system capable of either clean-screen—identify cars with clean exhaust, then exempt them from their annual emissions test—or dirty-screen, in which cars with dirty exhaust are identified, then you haul ’em in and have your way with ’em. The sniffer was set up alongside a busy I-25 ramp. Having driven past the variable-message sign that told all passing drivers about their cars’ contributions to Denver’s notorious brown cloud, I arranged in 1993 to spend a day in Stedman’s laboratory at the University of Denver. My curiosity was well rewarded; the technology was fascinating to watch at work. There was a computer with a rolling readout in damn-near-real time of the cars passing the sniffer: plate number, make/model/year, emissions readings. Part of the pilot study was that owners of cars caught out as dirty were invited to bring their cars in for free repairs and detailed before/after emission tests (then release ’em back into the wild and watch what happens the next time they pass the scanner).

They found and fixed some very interesting gross polluters. One was a ’79 Impala, whose elderly owner was the car’s one and original. It had a 350 cubic inch V8 engine, one of the many Chev smallblocks of ’76-’79 with unhardened camshaft lobes. There’d been a recall over it, but this guy had never taken the car in. When it came in for its free repairs, it barely had enough power to drive into the repair bay—to keep up with road traffic, the driver had the accelerator on the floor most of the time, with the wide-open carburetor dumping tons of fuel into the gasping engine. The camshaft (which was on Stedman’s shelf as an artifact) had almost no lobes left at all. Another belchmobile was an ’80-or-so Cadillac. Turns out it had originally been equipped with GM’s thoroughly lousy dieselised Olds 350 V8 engine. When that failed for the nth time, someone had swapped in a gasoline 350…and kept the diesel’s no-catalyst exhaust system. Stuff like that, plus a big many more boring cases of car simply neglected or elsewise in poor repair.

That pilot program, and others like it, proved Stedman’s remote-sensing much more effective and cost-effective than periodic drive-in-and-pay emission tests, not to mention a whole hell of a lot more convenient for most drivers. It should’ve long ago become standard practice, but the well-entrenched periodic-inspection industry squawked about it loudly enough that EPA got onside with them and jiggered the rules to ensure states didn’t get any funny ideas about shutting off the money flow to the private emission-testing contractors. This appears to be changing recently (finally).

Fine and dandy, but emission test results—whether the annual bring-in-and-pay type or any other—are intended to find cars rendered gross polluters by neglect or tampering so they can be fixed. That’s nothing to do with new-vehicle emissions certification or type-approval tests intended to see whether an automaker’s product complies with the design, construction, and performance requirements imposed by regulation. Nobody’s ever suggested a VW might fail a periodic (or roadside-sniffing) emissions test because of its factory-installed cheat box, because that won’t happen. It’s two different kinds of test for two different sets of questions.

Nevertheless, there’s a large mountain of evidence that scrutinising emissions by a roadside sniffer on a real road creates dependable data and separates the pollution from the perfume accurately, cost-effectively, and conveniently. Data from years of Stedman-type remote sensing nails VW bang to rights on their 2.0-litre cheatbag diesels. And the nudge-nudge/wink-wink loopholes in today’s emissions type-approval and certification protocols have been elucidated in these pages and others. Perhaps if the test moves from a treadmill to a real roadway, with roadside apparatus sniffing emissions produced by cars instrumented so as to verify they’re being driven normally, then “teaching to the test” will mean calibrating cars to produce acceptable emissions in the real world.