大众汽车公司“排放门”事件曝光后，美国环保署修改了相关的测试项目，而该事件给汽车制造商带来的影响远远不止这些。

专家指出，传动系统的控制策略，包括利用售后市场设备来调整发动机的思路，都将可能进行修改。他们还希望欧美政府加强现有法规的执行、严整排放测试协议、缩小实验室测试与路上测试之间的差距，并在新车型上市后抽样检查排放超标的情况。

总而言之，在经历这场“排放门”之后，汽车行业的工程师们将在测试和验证方面受到更为严苛的监督。

美国环保署表示大众在功率计校验测试中使用违法的发动机控制软件来减少尾气管柴油排放之后不久，就宣布即将实施道路测试。这也是美国环保署对大众CEO辞职、调查启动、车主诉讼蜂拥而至、公司股价大跌等一系列连锁事件做出的迅速回应。

大众事件近日甚嚣尘上，但这不是历史上第一起同类事件。1998年，美国环保署曾对7家重型柴油发动机生产商开出了11亿美元的罚单，这是史上因违反环境法而作出的最严重处罚，因为他们在发动机控制器中了使用一种“致胜装置”。当年，这种软件在130万辆卡车上更改燃油喷射正时装置并提高燃油效率，同时非法地规避了排放控制设备的使用。

一些业内人士透露，这种专为测试而开发的控制图，曾经一度流行于柴油车和汽油车市场。包括福特、通用、本田在内的整车厂都曾因作弊而受罚。

实况测试与实验室测试的比较：氮氧化物含量高出4倍

在测试与实际驾驶中使用不同的发动机控制软件的做法被称作“双图法”。在排放测试中，发动机运行而方向盘不动，专为测试而开发的标准图正是在这个时候使用的。

“90年代估计很多公司都干过类似的事，但大多数在90年代后期就停了，”一家不愿透露姓名的顶尖半导体公司的技术专家匿名评论道。与许多行业高层一样，他非常犹豫是否在《汽车工程杂志》上公开评论四面楚歌的大众公司。

这位专家称，目前的作弊主要集中于柴油机，“因为能控制的参数更多了”。“柴油是一阵接一阵燃烧的，这将决定你怎样管理机内的元素，比如未燃烧的柴油在汽缸内翻滚的状态。管理不当，你的柴油颗粒数量就会飙升。”

法规制定者发现，要对排放进行综合且持续的监控实属不易。美国在几年前改善了测试，好让实验室的测试条件接近于实况测试，但相对来说，欧洲的测试要温和不少。英国的道路排放测试公司EmissionsAnalytics的专家表示，欧洲柴油车的公路测试结果至今都与实验室结果相距甚远。

“在欧洲700辆车的实况测试中，我们发现氮氧化物的水平比政府测试标准高出4倍，二氧化碳高了31%左右，而燃油经济性低了24%，” Emissions Analytics的首席执行官Nick Molden表示。“但在欧洲，这些数据可能都算合法，这说明欧洲测试还不够到位。”

他指出，目前的欧洲测试没有坡道、急加速或冷启动等美国测试中使用的内容。许多观察家认为，环保署宣布即将开展的高速公路测试仅仅是大众“排放门”事件之后将要改变的众多内容之一。

“这可能对售后市场和安全领域造成巨大影响，”这位半导体专家说。“车主在购买车辆后可以轻易地使用现成工具来调整发动机，而汽车制造商并没有加以阻止。但是未来他们可能不想看到这样的事发生，因为惹怒监管人员的代价是非常巨大的。”

功率计测试与实况测试之争，以及对政府标准协议的批评促使人们思考各种解决方案，以缩小二者间的差距。详见下文http://articles.sae.org/12610/ 和 http://articles.sae.org/7094/。

发动机控制策略有望得到修改

专家指出，汽车制造商还有可能修改发动机控制策略。如果新软件可以集合当今微型控制器的所有优势，那么开发控制系统的任务就会变得更简单。当然，这个系统必须在确保性能的同时也满足排放与燃油经济性法规要求。这是密歇根州安娜堡市Simu Quest有限公司首席执行官John Mills的观点，然而这一愿景恐怕不会立刻实现。

“公司可以使用能够更真实地呈现物理状态的软件，并推行更多的参数化发展（即基于模型的发动机控制系统优化），”他表示。“但许多公司都害怕改变传统策略，因为他们知道该策略是有用的。”

法规要求新发动机里程必须超过15万英里，这是汽车公司不愿舍弃久经考验的软件的原因之一。另外一个原因则是控制一台发动机需要使用大量的软件。

“代码的数量因公司而异。某些OEM对所有发动机都使用一套代码库，但每辆车只使用其中的一段，” Mills表示。“另一些则为不同发动机开发不同策略。某些甚至有几十万行代码。其中一大部分都是冗余，但仍有上万条异常关键。”

微控制器的性能不断提升，是发动机开发人员能够为排放测试与普通运行分别开发程序的原因之一，专家解释道。目前的发动机控制程序只需要做一件事——调节发动机参数，因此它的处理速度和记忆要求远低于手机、平板电脑等消费者电子产品。

“低端发动机控制单元的处理器的最低频率目前为100兆赫兹左右，未来将上升至300兆赫兹左右，”这位半导体技术专家表示。“处理器将升至6MB。不久的将来将升至7或8，升至16MB。”

发动机控制系统的复杂性，及其保证汽车性能与燃油经济性的重要性，意味着软件必然受到密切关注。有些观察家相信，正因为这样，大众汽车的工程管理团队（也许还有高层主管）很可能认为监管机构对使用额外软件误导测试机构的做法早已心照不宣。

“这就说不清了，”一名一线工程师表示。“有人说这是一小部分人干的，但这件事可不是整个团队都会看漏的小事，而且整个集团完全没有注意到这些行为，也是不可能的。”

作者：Terry Costlow

来源：SAE《汽车工程杂志》

翻译：SAE 上海办公室

VW emissions scandal will impact future engine controls, testing

A change in the U.S. Environmental Protection Agency’s testing programs won’t be the only consequence for automakers in the wake of the Volkswagen emissions-cheating scandal.

Powertrain control strategies, including the ability to adjust engines with aftermarket devices, may also be revised, experts note. They also expect government regulators in the U.S. and Europe to more strictly enforce existing rules and tighten emission-testing protocols, close gaps between laboratory and real-world testing, and perform more spot checks that can catch abuse after new vehicle models are approved for sale.

Overall, industry engineers should expect greater oversight over automotive testing and validation in the aftermath of what some are calling “TDI-gate.”

Shortly after the EPA said VW was using illegal engine-control software that reduced diesel tailpipe emissions during dynamometer calibration testing, the agency announced it would begin performing on-road tests. That quick response came as VW’s CEO resigned, investigations began, lawsuits by vehicle owners flooded in, and the company’s market value plummeted.

While VW’s deception may be the largest such event, it’s far from the first. In 1998 EPA levied a $1.1 billion fine on seven heavy-duty diesel engine makers—the largest civil penalty ever imposed for violations of an environmental law—for using "defeat devices" in the engines’ controllers. This software altered injection timing and boosted fuel efficiency while illegally bypassing the emission control equipment on 1.3 million trucks.

Some industry insiders, speaking on background only, said that such strategies using a control map created just for testing, were once fairly common for both diesels and gasoline engines. Other OEMs, including Ford, GM, and Honda, have also been fined for cheating.

Real-world vs. lab: 4x higher NOx levels

The practice of using different engine-control software for testing and actual driving, is known as “dual mapping.” The control map written solely for emissions tests usually kicks in when the engine’s running but the steering wheel isn’t moving.

“A lot of companies probably did similar things with gasoline engines in the 1990s, but most of them stopped in the late ‘90s,” said a technologist at a leading semiconductor company who requested anonymity for this article. Like many industry executives, he was reluctant to talk with Automotive Engineering publicly about the embattled OEM.

He said the trickery is now conducted primarily in diesels “because you’ve got more parameters to control,” he continued. “Diesel fuel burns in bursts, depending how you manage things like tumbling as unburned fuel moves in the cylinder; you can end up with way more particulates.”

Regulators have found it difficult to monitor emissions comprehensively and consistently. The U.S. upgraded its tests a few years ago to make lab tests closer to real-world driving conditions, but in Europe tests are comparably benign. European diesel vehicles’ highway performance is nowhere near the levels found in regulatory labs, according to experts at Emissions Analytics, a U.K. company that performs on-road emissions tests.

“In real world tests of about 700 cars in Europe, we found NOx levels four times above the government test levels, CO2 was 31% above official levels, and fuel economy was 24% worse,” said Nick Molden, Emissions Analytics' CEO. “The twist is that in Europe, that’s probably all legal, it’s more a factor of inadequate testing in Europe.”

He noted that current European tests don’t have hills, rapid acceleration or tests used in the U.S. including cold starts. Many observers feel that the EPA’s announcement that it would begin running on-highway tests is only one of the changes that will occur in response to the VW scandal.

“This may have a big impact on the aftermarket and security,” the semiconductor expert said. “Car buyers can fairly easily use off-the-shelf tools to adjust the engine; carmakers haven’t gone out of their way to prevent this. Going forward they may not want to let these things happen. This could also impact who’s able to reflash the controller.”

The ongoing controversy over dynamometer vs. real-world testing, and criticisms of government standards protocols, have spawned various solutions aimed at closing the gaps. See http://articles.sae.org/12610/ and http://articles.sae.org/7094/.

Control strategy revisions expected

Experts noted that it’s also possible that automakers will revise their engine control strategies. If new software took advantage of everything that today’s microcontrollers can do, it would be simpler to create controls that meet emissions and fuel economy mandates while still delivering performance, according to John Mills, CEO at SimuQuest Inc., an Ann Arbor, MI, software development company. However, that may not happen soon.

“Companies could move to use a better representation of physics and start adopting more parameterization [model-based optimization of engine controls],” he said. “But a lot of companies have legacy strategies that they’re afraid to change because they know it works.”

Demands that new engines must reliably exceed 150,000 miles is one reason for this reluctance to change from software that’s been proven on the road. Another factor is the large volume of software needed to control an engine.

“The amount of code varies widely; some OEMs have one code base used for every engine, using only a segment of it for each vehicle,” Mills said. “Other OEMs have separate strategies for different engines. Some have hundreds of thousands of lines of code. A lot of that is fairly verbose, but there are tens of thousands of lines of code that are critically important.”

The rising capability of microcontrollers is among the reasons that engine developers can create separate programs for emissions tests and normal operations, experts explained. Engine control programs only do one thing, adjusting engine parameters, so the processing speeds and memory requirements of engine controllers are small compared to consumer devices like phones and laptops.

“The low-end engine control units have processors that run at about 100 MHz at the low end, going up to around 300 MHz,” the semiconductor technologist said. “The processors have up to six Mbytes. In the fairly near future, that will go up to seven or eight, maybe even 16 Mbytes.”

The complexity of engine controls, along with their importance in providing reliable vehicles with good performance and fuel economy, means that software gets close attention. Some observers believe that because of this, VW’s engineering management, and perhaps higher level executives, likely knew that regulators knew that additional software was being used to mislead those who tested the vehicles.

“This is quite troubling,” observed a Tier 1 engineer. “Some people say that this might be the work of a few people. But these aren’t the type of metrics the full team would miss—nor are they activities that the group would not observe.”