Thermogram of a car engine

Diesel: a model of efficiency

Car makers look to electronics, electrification and sophisticated modelling for the next wave of efficiency gains in diesel engines.

In the wake of the Volkswagen emissions scandal, diesel has suffered a setback in terms of public perception. But that overshadows intense work on R&D that is steadily bringing both carbon dioxide and other emissions down as combustion efficiency goes up. Seeing this, manufacturers such as Mercedes are sticking with the technology.

Mercedes placed a strong vote of confidence in the diesel engine in February 2016 when it unveiled the redesigned E-Class car. Other projects aim to bring the CO2 emissions for smaller cars down to 70g/km in the near future. The firm’s average today is 125g/km for its passenger fleet.

The vehicle maker intends the engine to be the first of a family that will define the company’s diesel strategy for the next 10 years. As part of the vote of confidence in diesel, Mercedes has allocated close to €3bn over the next three years to develop a series of future diesel engines.

The car maker is not alone. Suppliers such as Bosch and Ricardo are working on designs that aim to eke more out of every millilitre of fuel and slash the emissions of gases such as nitric oxide to keep up with ever more stringent targets.

One direction in design is to have more electrification within the car, says Simon Edwards, global director of technology at Ricardo: “The segmentation is blurring much more as we add more levels of electrification but without necessarily going straight from non to full hybrid.”

Bosch takes the position that the combination of combustion engine and electric motor in the powertrain will give its design team more options in its optimisation programme.

The near future for a number of manufacturers, particularly the German companies, is micro-hybridisation. Matthew Viele, vice president of engineering at National Instruments’ Drivven subsidiary, says that instead of installing a large electric motor for traction, micro-hybridisation sees smaller motors being used to drive devices such as pumps or the cam phasing that controls valves on the exhaust ports of the engine cylinders.

Edwards adds: “It is a hybridisation that gives you some performance increase or gives the ability to run improved-performance electric ancillaries.”

Model-based control

The second route to efficiency is through much greater use of electronic computation that started with a big step up in the use of electronics in powertrain in the 1990s. Since then a growing number of systems have come under electronic control, and that trend will continue with projects such as Ricardo’s Advanced Diesel Electric Powertrain (ADEPT).

“Ten years ago a thermal system would not necessarily have been electronically controlled at all. These days there might be one electrical component, such as a coolant control thermostat,” Edwards explains. “With ADEPT there are several parts of the thermal circuit that are electronically controlled. But the basic control principles in the near-term are the same.

“There is a transition that we’re making in the industry towards model-based control. This is where we are controlling parts of the behaviour of the engine based on a real-time simulation of the engine,” Edwards adds. Sensors provide feedback to the model that then determines when to fire ignition pulses or alter cam timing.

Modelling in general has become crucial to engine design and optimisation. Fluid dynamics and structural analyses have guided the R&D that drives better understanding of combustion processes and, with them, improved control.

The key to emissions control in diesel, says Viele, is “getting the fuel injected properly. The fuel system is the most important part of making the engine clean and effective.”

The current direction is towards generating more injection pulses within a cycle, moving to higher pressures and deploying more complex electronics to deliver smart pulse strategies, where multiple ignition pulses are fired in sequence within the same cycle. It means a move away from simple open-loop control strategies into more complex closed-loop control.

Viele explains: “There are a lot of interactions between the injections. You have intended interactions and unintended interactions with the fuel rail. They are on the order of hundreds of microseconds. We need a lot of very fast processing to be able to do that.”

The models that drive real-time controller development come from a collection of simulation tools and techniques that look at different elements of the physics of engines, such as fluid and gas flow, temperature distribution and the stresses these put on the mechanical systems.

“You can model the physics using lumped parameters but when you want accuracy you can’t rely on lumped parameters. You need to go more deeply into the physics,” says Valerio Marra, marketing manager at Comsol. Simulation tools that perform computational fluid dynamics let designers explore the boundary conditions and test parameters that can be fed into other parts of the simulation. “Then they can use that data to build a new curve and feed those back,” he adds.

Arwel Davies, technical services engineer at Ansys, says designers have incorporated greater levels of detail into the modelling. “Rather than using two or three thermal values, they map very complex thermal distributions across the entire cylinder, and then, in simulation, allow that temperature to spread through the engine.”

Ansys senior engineer Samir Patel adds: “We are seeing more design-exploration type analysis being done now. You are not going to just look at a single point of operations but the whole design space and see how these engine under various conditions.

“It used to be the case going five or six years back where you were just passing fluid data between models. Now you see how your thermal and pressure data leads to structural changes – how parts will deform – and then pass that data on deformation back for fluid analysis. It’s all unified. You get much closer to modelling the reality,” Patel adds.

The timescales of different types of simulation means different types of model need to be switched in and out. Viele explains: “The timescales of computation go everywhere from big megawatt supercomputers calculating one cylinder event over hours or days down to one-dimensional CFD tools that can be optimised to run in real-time.”

Noah Reding, principal product marketing manager at National Instruments, says this diversity creates a major need among customers for interoperability between modelling systems. “You may have one tool that can perform computations of one type. And maybe next time bring in a different simulation model to get better performance. You are creating a model-agnostic, tool-agnostic environment.”

The ability to perform modelling in real-time is vital to ensure that the simulations marry up with reality. Although the algorithms for model-driven control will be baked into silicon when the technology is ready for production, today’s programmable embedded computers are not able to keep pace with the microsecond demands of ignition timing.

Companies such as Dspace and NI have used programmable logic to fill the gap. Programmed through LabView in the case of NI or Matlab/Simulink, the field-programmable gate arrays (FPGAs) can implement the algorithms as logic circuits. The result is much faster processing than is possible with software running on a microprocessor.

Viele adds: “We see people building advanced research test cell type engines to explore these different combustion techniques. FPGA-based systems dominate this advanced research because they are very flexible. It is easy to add channels and there are novel things that you can do with the FPGA. They provide very tight synchronisation so you can capture injection events with high precision and couple those with accurate physics models.”

Results captured from the hardware-in-the-loop systems can be compared with the supercomputer simulations to see where the simplified real-time model diverges from reality and fed back into revisions that gradually improve the accuracy of the control loop.

Viele says these advanced control techniques have led to “phenomenal increases in power and reductions in emissions. So you can do things such as downsize the engine”.

Integration

Through system-level modelling and integration, vehicle makers can begin to find ways to recover energy that the drivetrain cannot use immediately. Increased hybridisation will see integrated cycle engines become more common. “You can put steam turbines on the engine to gain electric power as well as well augment the turbocharger with an electric system,” Viele says.

As the in-vehicle computer models improve, some of the electronics will start to disappear. Edwards says: “There will be sensor deletion. Some of the sensors that we have, for example, for temperature measurement or constituents in the exhaust will not be needed.” Instead, their values will be inferred from other sensors and the real-time model.

Further integration means the vehicle will have greater understanding of driving conditions and how they will change along its route. Edwards explains: “There will be a further step-change towards model-predictive control. The system can make predictions of what conditions will mean for emissions. When you use some of the other information available in the world – weather, traffic information, geography – your control strategy will have another level of complexity on top of it.”

The advantage of model-predictive control is that the engine-management system can make decisions that maximise efficiency or which are calculated to control emissions to a lower level at the expense of performance. If the vehicle is running within a city, it can favour battery power being supplied to the electric motors. The diesel engine may kick in once the car has left the low-emissions zone or battery levels run low.

On the open road, the engine can use driving conditions and geography to determine the best trade-off between diesel and electric power. Volvo has already deployed some of these ideas with the cruise-control software on trucks, determining when to apply power to achieve greater momentum and when to perform braking. Future systems will use energy-recovery braking to make the most of the kinetic energy available.

Those changes will provide scope to cut emissions. Edwards says: “The target is 70g/km of CO2 in a C-class car with evolutionary control.”

Driving the diesel engine using models and in closer cooperation with electric motors and batteries will bring the bar down even further.

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them

Close