The whole package: Dr. Hussein Dourra on controls
On 4/13/2016, FCA US' Senior Manager for Powertrain Controls Sciences, Hussein Dourra, met with retired Chrysler engine lab technician Marc Rozman at the Society of Automotive Engineers Congress in Cobo Hall, Detroit.
I'm the senior Technical Fellow for the advanced work, we call it decoupled projects. We work a lot with universities, and we deal with very advanced research work in the powertrain controls arena.
Let me start by saying the challenge these days is the fuel economy requirements, the CAFE [corporate average fuel economy] requirements, the CO2 (greenhouse gas), forcing the industry - not just us, everybody - to become very innovative, come up with ideas and new approaches and apply it.
My area is controls. Everything you do, at the end of the day, we need to control it. The control becomes a lot more complex when we add so many features, so many technologies. Unless we use controls in a different way and move to an advanced control approach, the solution will not be ideal. Therefore, we have to change the way we do things.
There is a lot more complexity in the system. To be more efficient, we're adding many actuators, many technologies; and integrating all these technologies, between the transmission, engine, and driveline, is challenging, but at the end of the day, we are delivering a correct solution that the customer wants.
Right. We're challenged by having the governmental regulations for emissions and meeting fuel economy standards, but we need to make the vehicle reliable and perform well, and we have to make sure the customer is satisfied with what they're driving.
Hussein Dourra has 40 patents; he was named an SAE Fellow in 2015. Mr. Dourra joined Chrysler Corporation in 1982, and was named the first Chrysler Technical Fellow in 2008. He shared the Walter P. Chrysler Technology Award in 2008, for work on an algorithm to precisely control multiple shift elements, and later won the SAE's Edward Cole Award (2013) and Forest R. McFarland Award (2014).
That's forcing the guys in the industry to start thinking differently. Take my work in controls, for example.
In the old days we used to apply some basic classical control like PID, Proportional Integrated Differentiator. Now we are going into much more advanced techniques in solving these control issues, such as a physics-based approach. A physics-based approach can capture a lot of the dynamics of the system, and as we add more technologies to the vehicle we already have, because we have the physics in place, we can react more easily. We don't have to deal with the global issues, we only deal with the new, added features. It's a different approach to controls.
We have the technology now where we can look at the vehicles can even adapt to our driving habits. They can actually learn, to a certain point, where we're going next.
Adaptation is one of the main items we have in the vehicles these days because as you know aging, temperature conditions, operating conditions, it all changes with time. Therefore, we need algorithms and software to adapt to all these changes and keep giving owners the same quality they had when they bought the car.
So I would say adaptation's one thing, as we mentioned. Optimization is another thing. For example, in the old days we used to have a shift schedule. Now, we have to ask, what's the demand at the wheel? What kind of torque do you need at the wheel? What's the best state to put the vehicle in to give you the best fuel economy, best drivability and so on?
The approach is different; as you see, we start thinking globally, of the entire system. It's not a component in the vehicle; it's the vehicle as a whole. Again, the physics approach is a solution. And when I say physics approach it's not simple - you have to really have the basic understandings of what's going on. In adding an EGR to an engine, what does that do? The whole air path changes, the flow changes, the control changes. How are we going to monitor all this to come up with a good control system? A good model of the system is needed to come up with a good control system.
Or, for example, when you go from a four-speed transmission to a nine and ten or eleven speed. I know GM and Ford are working on a ten-speed transmission. Going from four gears to ten gears brings new challenges, what's going to happen to the shift busy-ness and drivability? How are you going to make sure all the desired control items are in place and are putting the vehicle in the right operating point?
The controls should model all these inputs and reacts accordingly.
Even maintaining manufacturing and cost controls is a difficult balance between getting what you need to have done.
Exactly. And you brought a good point when you said manufacturing. You have to design your system nowadays to be modular. When adding another technology, I don't want to go and redesign from scratch. You want the system to be adaptable when we add new technologies, and be able to use what we had before. So all that challenge is in place as we design algorithms or come up with ideas for how to solve problems.
Then you start interjecting the hybrid-type platforms, all different . . .
The 55 miles per gallon rule, 54.5 miles per gallon in 2025, is going to require hybrid. We're pushing the IC [internal combustion, or traditional gasoline/diesel] engine to its limit, transmission efficiency to its limit, but at the end of the day to achieve that you need some kind of electrification. In 2025, we must have one. So as we design new technologies, we try to have in mind that hybrid is coming, and how are we going to integrate that with our current system?
It is coming. You may not like it at times. There's a cost factor involved.
Exactly. That's where the business case comes in. Gas price fluctuation makes it hard to make those calls, really.
Right. There's a challenge. And the products are looking good. Now do we have hybrids in production right now? The Pacifica is coming out shortly, but we don't have another.
There are some coming, but to be honest with you I don't work in hybrids. There's another area that deals with hybrids. I can say that hybrids will add a lot of challenge to controls. When we start talking about multiple sources of energy [e.g. gasoline engine and electric motors], optimizing the system becomes very important.
Just imagine having multiple sources of power, many actuators, and new technologies, and now I'm going to add GPS information coming to the vehicle. What do you do? How do you change controls to take advantage of all this?
Optimization, physics-based solutions, and adaptations are needed.
How do you manage that?
Exactly. How do you manage and control that power that's coming to the vehicle? You've got multiple sources. Especially when you're talking hybrid, right? What level of electrification do you want in the vehicle? That becomes a big question and it demands a physics-based approach.
Just imagine the complexity of having such systems, and it's coming and we have to be aware of these things. Safety is a big issue, obviously.
You can have the best product design, but if you don't operate it in the most efficient state, you're not really getting the best out of that product.
FCA's Eric Mayne: Let me jump in here for a second because Hussein has a good story. You mentioned optimization a minute ago and you mentioned electrification. I've noticed in the media sort of a knee-jerk if you will suggestion that electrification is the automatic gold standard when it comes to efficiency. But do the gasoline engines get enough credit for being more efficient [than they had been]?
We are pushing the internal combustion engine, to its limit. We're applying technologies like variable lift, cooled EGR, and cylinder deactivation. The whole thing is really expanding, and we are pushing the IC engine to its limit. Thermal management is becoming a big thing in the vehicle.
It's all being controlled.
Yes, it's all in controls. How you operate the component is a big thing. And again, it's not one component alone. It's the system as a whole. It's the whole package.
So, for example, if I'm running a transmission, is lockup in third gear or is fourth gear unlock more efficient? Which is better? Which is most efficient when the engine is operating at this condition? They're all connected.
At the end of the day you want the torque at the wheel. How you achieve that torque is an optimization issue, and you need to know how the system works. You can achieve that torque in the most optimum way, and guess what? The customer, the driver, doesn't know what's happening. He's getting that torque he wants. These are the kinds of tricks and innovation that we have to apply.
Again, from a control perspective, the old PID classical approach is going to diminish. We have to start doing a lot of simulation, running a lot of models in the vehicle, use physics as we apply those algorithms, and therefore we can end up with a product that can survive any operating condition and still provide the required quality. Again, you can add technology to it without tearing it apart from ground-up anymore.
I would like to add one more thing which is very important, speed to market. The way we do things these days is in a virtual environment; a lot of simulation work is done at the beginning of the project.
Nowadays, we do most of the work before getting the tangible product. We can live in the simulation world and evaluate a lot of things even before we get any actual hardware.
You know, FCA was probably ahead of others in this one. I remember a product that I worked on a few years ago, where we did most of the work in the virtual environment. When we got the real hardware, it took only four hours - honestly four hours - to get that running in the dynamometer. Four hours when I get the tangible product because all the work was done before in the simulation.
Now, I'm not saying it was ready for production at that time, but it was in operation in the dyno in four hours. My point is, speed to market becomes very challenging, and with the technologies we have today, with the microprocessors and the electronics, you can do a lot of things in the virtual environment before applying the solution on real hardware. We are optimizing this approach a lot at FCA.
You mentioned modeling. You have a number of different products out there, different configurations, the V6, the V8 and four-cylinder. There is a lot of worked involved in them in the modeling and the calibration and the packaging of that. It's a complex thing to do, and to get it right, I think we've done a good job so far with Chrysler.
As you said, modelling is very important for multiple reasons. When you say modeling, people might understand modeling differently. So when I look at modeling from a control perspective it means one thing. When you look at it from hardware design prospective it means different thing.
The fidelity of the models means a lot. It depends on the application of a model. I cannot get a very detailed model, for example, and put it in a microprocessor for control. That doesn't apply. That kind of detailed modeling is good for hardware design, right? But when you want to apply modeling to controls then you have to come up with a mathematical model that can operate in the ECU [electronic control unit, or "powertrain computer"].
So modeling is also a whole area that needs innovation. It's how to apply modeling - unless you do it right, unless you apply it in the right place, you don't get the advantage of modeling.
In controls, we do a lot of modeling in the loop (MIL), software in the loop (SIL), testing before we get the real hardware. All the "control disturbances" that you can imagine, are applied in this environment, so you know you have a robust solution before applying the controls in the real product. The process of going from an idea to a model in the loop to software in the loop to hardware in the loop and finally in a dyno in a vehicle is very important, and should be adopted to increase speed to market.
Other interviews • SAE Congress • Transmissions • Electronic Design at Chrysler • Alternative Energy
Chrysler Heritage • History by Year • Chrysler People and Bios • Corporate Facts and History
Home •
Engines •
Reviews •
Chrysler 1904-2018 •
Upcoming •
Trucks •
Cars
Spread the word via <!--Tweet or--> Facebook!
We make no guarantees regarding validity or accuracy of information, predictions, or advice - .
Copyright © VerticalScope Inc. All rights reserved. Dodge, Jeep, Chrysler, Ram, and Mopar are trademarks of Fiat Chrysler Automobiles.
I'm the senior Technical Fellow for the advanced work, we call it decoupled projects. We work a lot with universities, and we deal with very advanced research work in the powertrain controls arena.
Let me start by saying the challenge these days is the fuel economy requirements, the CAFE [corporate average fuel economy] requirements, the CO2 (greenhouse gas), forcing the industry - not just us, everybody - to become very innovative, come up with ideas and new approaches and apply it.
My area is controls. Everything you do, at the end of the day, we need to control it. The control becomes a lot more complex when we add so many features, so many technologies. Unless we use controls in a different way and move to an advanced control approach, the solution will not be ideal. Therefore, we have to change the way we do things.
There is a lot more complexity in the system. To be more efficient, we're adding many actuators, many technologies; and integrating all these technologies, between the transmission, engine, and driveline, is challenging, but at the end of the day, we are delivering a correct solution that the customer wants.
Right. We're challenged by having the governmental regulations for emissions and meeting fuel economy standards, but we need to make the vehicle reliable and perform well, and we have to make sure the customer is satisfied with what they're driving.
Hussein Dourra has 40 patents; he was named an SAE Fellow in 2015. Mr. Dourra joined Chrysler Corporation in 1982, and was named the first Chrysler Technical Fellow in 2008. He shared the Walter P. Chrysler Technology Award in 2008, for work on an algorithm to precisely control multiple shift elements, and later won the SAE's Edward Cole Award (2013) and Forest R. McFarland Award (2014).
That's forcing the guys in the industry to start thinking differently. Take my work in controls, for example.
In the old days we used to apply some basic classical control like PID, Proportional Integrated Differentiator. Now we are going into much more advanced techniques in solving these control issues, such as a physics-based approach. A physics-based approach can capture a lot of the dynamics of the system, and as we add more technologies to the vehicle we already have, because we have the physics in place, we can react more easily. We don't have to deal with the global issues, we only deal with the new, added features. It's a different approach to controls.
We have the technology now where we can look at the vehicles can even adapt to our driving habits. They can actually learn, to a certain point, where we're going next.
Adaptation is one of the main items we have in the vehicles these days because as you know aging, temperature conditions, operating conditions, it all changes with time. Therefore, we need algorithms and software to adapt to all these changes and keep giving owners the same quality they had when they bought the car.
The approach is different; as you see, we start thinking globally, of the entire system. It's not a component in the vehicle; it's the vehicle as a whole. Again, the physics approach is a solution. And when I say physics approach it's not simple - you have to really have the basic understandings of what's going on. In adding an EGR to an engine, what does that do? The whole air path changes, the flow changes, the control changes. How are we going to monitor all this to come up with a good control system? A good model of the system is needed to come up with a good control system.
Or, for example, when you go from a four-speed transmission to a nine and ten or eleven speed. I know GM and Ford are working on a ten-speed transmission. Going from four gears to ten gears brings new challenges, what's going to happen to the shift busy-ness and drivability? How are you going to make sure all the desired control items are in place and are putting the vehicle in the right operating point?
The controls should model all these inputs and reacts accordingly.
Even maintaining manufacturing and cost controls is a difficult balance between getting what you need to have done.
Exactly. And you brought a good point when you said manufacturing. You have to design your system nowadays to be modular. When adding another technology, I don't want to go and redesign from scratch. You want the system to be adaptable when we add new technologies, and be able to use what we had before. So all that challenge is in place as we design algorithms or come up with ideas for how to solve problems.
Then you start interjecting the hybrid-type platforms, all different . . .
The 55 miles per gallon rule, 54.5 miles per gallon in 2025, is going to require hybrid. We're pushing the IC [internal combustion, or traditional gasoline/diesel] engine to its limit, transmission efficiency to its limit, but at the end of the day to achieve that you need some kind of electrification. In 2025, we must have one. So as we design new technologies, we try to have in mind that hybrid is coming, and how are we going to integrate that with our current system?
It is coming. You may not like it at times. There's a cost factor involved.
Exactly. That's where the business case comes in. Gas price fluctuation makes it hard to make those calls, really.
Right. There's a challenge. And the products are looking good. Now do we have hybrids in production right now? The Pacifica is coming out shortly, but we don't have another.
There are some coming, but to be honest with you I don't work in hybrids. There's another area that deals with hybrids. I can say that hybrids will add a lot of challenge to controls. When we start talking about multiple sources of energy [e.g. gasoline engine and electric motors], optimizing the system becomes very important.
Just imagine having multiple sources of power, many actuators, and new technologies, and now I'm going to add GPS information coming to the vehicle. What do you do? How do you change controls to take advantage of all this?
Optimization, physics-based solutions, and adaptations are needed.
How do you manage that?
Just imagine the complexity of having such systems, and it's coming and we have to be aware of these things. Safety is a big issue, obviously.
You can have the best product design, but if you don't operate it in the most efficient state, you're not really getting the best out of that product.
FCA's Eric Mayne: Let me jump in here for a second because Hussein has a good story. You mentioned optimization a minute ago and you mentioned electrification. I've noticed in the media sort of a knee-jerk if you will suggestion that electrification is the automatic gold standard when it comes to efficiency. But do the gasoline engines get enough credit for being more efficient [than they had been]?
We are pushing the internal combustion engine, to its limit. We're applying technologies like variable lift, cooled EGR, and cylinder deactivation. The whole thing is really expanding, and we are pushing the IC engine to its limit. Thermal management is becoming a big thing in the vehicle.
It's all being controlled.
Yes, it's all in controls. How you operate the component is a big thing. And again, it's not one component alone. It's the system as a whole. It's the whole package.
So, for example, if I'm running a transmission, is lockup in third gear or is fourth gear unlock more efficient? Which is better? Which is most efficient when the engine is operating at this condition? They're all connected.
At the end of the day you want the torque at the wheel. How you achieve that torque is an optimization issue, and you need to know how the system works. You can achieve that torque in the most optimum way, and guess what? The customer, the driver, doesn't know what's happening. He's getting that torque he wants. These are the kinds of tricks and innovation that we have to apply.
Again, from a control perspective, the old PID classical approach is going to diminish. We have to start doing a lot of simulation, running a lot of models in the vehicle, use physics as we apply those algorithms, and therefore we can end up with a product that can survive any operating condition and still provide the required quality. Again, you can add technology to it without tearing it apart from ground-up anymore.
I would like to add one more thing which is very important, speed to market. The way we do things these days is in a virtual environment; a lot of simulation work is done at the beginning of the project.
Nowadays, we do most of the work before getting the tangible product. We can live in the simulation world and evaluate a lot of things even before we get any actual hardware.
You know, FCA was probably ahead of others in this one. I remember a product that I worked on a few years ago, where we did most of the work in the virtual environment. When we got the real hardware, it took only four hours - honestly four hours - to get that running in the dynamometer. Four hours when I get the tangible product because all the work was done before in the simulation.
Now, I'm not saying it was ready for production at that time, but it was in operation in the dyno in four hours. My point is, speed to market becomes very challenging, and with the technologies we have today, with the microprocessors and the electronics, you can do a lot of things in the virtual environment before applying the solution on real hardware. We are optimizing this approach a lot at FCA.
You mentioned modeling. You have a number of different products out there, different configurations, the V6, the V8 and four-cylinder. There is a lot of worked involved in them in the modeling and the calibration and the packaging of that. It's a complex thing to do, and to get it right, I think we've done a good job so far with Chrysler.
As you said, modelling is very important for multiple reasons. When you say modeling, people might understand modeling differently. So when I look at modeling from a control perspective it means one thing. When you look at it from hardware design prospective it means different thing.
The fidelity of the models means a lot. It depends on the application of a model. I cannot get a very detailed model, for example, and put it in a microprocessor for control. That doesn't apply. That kind of detailed modeling is good for hardware design, right? But when you want to apply modeling to controls then you have to come up with a mathematical model that can operate in the ECU [electronic control unit, or "powertrain computer"].
So modeling is also a whole area that needs innovation. It's how to apply modeling - unless you do it right, unless you apply it in the right place, you don't get the advantage of modeling.
In controls, we do a lot of modeling in the loop (MIL), software in the loop (SIL), testing before we get the real hardware. All the "control disturbances" that you can imagine, are applied in this environment, so you know you have a robust solution before applying the controls in the real product. The process of going from an idea to a model in the loop to software in the loop to hardware in the loop and finally in a dyno in a vehicle is very important, and should be adopted to increase speed to market.
Other interviews • SAE Congress • Transmissions • Electronic Design at Chrysler • Alternative Energy
Chrysler Heritage • History by Year • Chrysler People and Bios • Corporate Facts and History
Home •
Engines •
Reviews •
Chrysler 1904-2018 •
Upcoming •
Trucks •
Cars
Spread the word via <!--Tweet or--> Facebook!
We make no guarantees regarding validity or accuracy of information, predictions, or advice - .
Copyright © VerticalScope Inc. All rights reserved. Dodge, Jeep, Chrysler, Ram, and Mopar are trademarks of Fiat Chrysler Automobiles.