Behind electric cars, there is a sustainability issue that has not yet been solved: large vehicles, such as trucks and long-distance planes, are still very difficult to electrify.
Alternative fuels can be a viable solution, but it’s not as simple as filling existing engines with the latest biofuel. Engines are designed with a particular fuel in mind.
Carrie Hall, associate professor at the Illinois Institute of Technology, presented a simple and cost-effective solution to ease the transition to large vehicle electrification. He has developed a new computer model that can help diesel engines run on different alternative fuels with a simple software update.
Since we are focusing on a software update, someone can install it in your vehicle without incurring many additional costs. In reality, they will not have to change the hardware of their vehicle.
One of the big obstacles to running a diesel engine over gasoline is the difference in responsiveness. Gasoline injected into an engine cylinder normally does not burn until the engine provides a spark to ignite the fire, creating an explosion that will spread away from the spark evenly throughout the engine cylinder.
Diesel fuel, on the other hand, tends to burn spontaneously after being compressed in the cylinder. When trying to run gasoline in a traditional diesel engine, the cylinder explosion may be unpredictable or not burn at all. For this reason, synchronization is essential, as engine efficiency depends on multiple cylinders working in harmony.
If the fuel is burned too soon or too late, it is not used to its full potential and its efficiency is reduced. To get the most out of gasoline combustion, diesel engines need real-time information about when the fuel has ignited.
The things that happen inside the cylinder of the engine are very difficult to measure economically. So what we’re trying to do is take the information we get from simpler, cheaper sensors that are outside of the actual cylinder of the engine where combustion occurs, and from there, to diagnose what is happening inside the engine.
All of this has to happen in a split second, every time. Some motor control designers update their model using machine learning techniques or storing large data tables to avoid model calculations, but Hall took a different approach.
We’ve tried to create models based on the underlying physics and chemistry, even when we have these very complicated processes. Recently, interest has been shown in the use of neural networks to model combustion. The problem is then it’s just a black box, and you don’t really understand what’s going on underneath, which is a challenge for control because if you’re wrong you can have something bad .
Hall started with the complicated version of the calculations and explored ways to simplify them until he found a way to describe science with equations that were faster to solve while still meeting industry accuracy standards for control models.
We tried to capture all of the underlying effects, but in more detail than we know for real-time control, and let that be our benchmark.
We then simplify it strategically using things like neural networks, but we keep that overall structure to understand what each element means and what it actually does in there.
The result is a simpler and more adaptable model.
Whereas a pure machine learning approach needs to completely retrain for a new fuel, Hall can simply update a few parameters that correspond to the measurable properties of the fuel.
We are working with these companies to help them understand the underlying combustion processes, but also to create tools that they can integrate into their own software and enable their next generation of engines to use these fuels and do them well. utilize.
Going through www.iit.edu