6
Sustainable engineering triggers advancements in engineering modeling The push for greener, sustainable design is universal. Whether it’s in the cars we drive or where we get our energy for our houses, green design has left the periphery of modern engineering and is now a dominant influence in contemporary engineering, often even the most important factor. In essence, sustainable design is designing to do more with less. The green road, however, is turning out to be winding and hazardous, and is presenting some monumental challenges for engineers. We want products that consume fewer resources during production, require less energy during operation, but still have the performance and pizzazz that we all love, and at competitive prices, too. In many ways, these demands are contradictory, but the engineering community is responding with some creative thinking and new technologies. Of course, one of the most important techniques that engineers have responded with is computerized control. In auto, aerospace, power, robotics, instrumentation, and other fields, today`s control techniques have literally launched a revolution. Simultaneously, the increased deployment of control systems has necessitated the development of robust and rich modeling and simulation methods. With so many critical systems now being electronically controlled through intricate computer logic, new techniques are needed to test for all possible cases. Not doing so can hide bugs or improper logic. At best, this introduces inefficiencies, and at worst, it can threaten consumer safety. The entire field of Model-Based Design (MBD) is now an entrenched dimension in the design process in many industries. Advanced System-Level Modeling

Sustainable engineering triggers advancements in ... · PDF fileSustainable engineering triggers advancements in engineering modeling The push for greener, sustainable design is universal

Embed Size (px)

Citation preview

Sustainable engineering triggers advancements in engineering modeling

The push for greener, sustainable design is universal. Whether it’s in the cars we drive or where we get our energy for our houses, green design has left the periphery of modern engineering and is now a dominant influence in contemporary engineering, often even the most important factor. In essence, sustainable design is designing to do more with less. The green road, however, is turning out to be winding and hazardous, and is presenting some monumental challenges for engineers. We want products that consume fewer resources during production, require less energy during operation, but still have the performance and pizzazz that

we all love, and at competitive prices, too. In many ways, these demands are contradictory, but the engineering community is responding with some creative thinking and new technologies.

Of course, one of the most important techniques that engineers have responded with is computerized control. In auto, aerospace, power, robotics, instrumentation, and other fields, today`s control techniques have literally launched a revolution. Simultaneously, the increased deployment of control systems has necessitated the development of robust and rich modeling and simulation methods. With so many

critical systems now being electronically controlled through intricate computer logic, new techniques are needed to test for all possible cases. Not doing so can hide bugs or improper logic. At best, this introduces inefficiencies, and at worst, it can threaten consumer safety. The entire field of Model-Based Design (MBD) is now an entrenched dimension in the design process in many industries.

Advanced System-Level Modeling

An element essential to the success of a Model-Based Design initiative is, of course, the quality of the models. With the increasing complexity of green applications, the effectiveness of conventional modeling approaches is starting to come under scrutiny. The example often cited by experts is that of the hybrid electric vehicle (HEV). Aside from the obvious complexity of such a vehicle, the real impact within an engineering organization is more subtle.

By definition, HEV powertrains are multi-domain systems whereas most of us in the engineering community were trained as specialists. When cars only had one type of power source, we were able to organize the workflow by function:

i.e. engine, driveline, chassis, etc., and each would have reasonable correlations to how we were trained. In the HEV case, the powertrain team now requires expertise from the electrical motor side. Additionally, system-level expertise is also required so that the respective subsystems can be analyzed as a collective whole. Compounding this issue is the fact that the software toolchains in place were, for the most part, developed by specialist vendors whose products reflect a more conventional organizational dynamic. The tools themselves, as powerful as they are, often struggle to deal with multi-domain systems.

Modeling challenges for green design

Toyota, in collaboration with the Waterloo Centre for Automotive Research (WatCAR) at the University of Waterloo in Canada, Maplesoft, and various government agencies, launched a comprehensive research initiative to modernize modeling techniques with a special emphasis on resolving the challenges posed by HEV and EV applications. Dr. John McPhee of WatCAR was appointed the NSERC/Toyota/Maplesoft Industrial Research Chair, to lead this initiative.

One highlight of the overall project was the development of modern, high-fidelity models of automotive batteries for HEV and EV applications. A decade ago, the car battery had a relatively simpler life and advanced modeling was not particularly important. With EVs and HEVs, the battery behavior is of paramount importance and automotive engineers are scrambling to squeeze in sufficient-fidelity models into their analysis. Even with this one particular

component, you can see the basic challenges. It is multi-domain in nature with electrical, thermal, and chemical dimensions. Some of these domains require expertise in subfields that are not part of the core engineering competence of teams that have historically been dominated by mechanical engineering specialists. At the system level, to virtually

test candidate models, you need to connect the battery model to proper load scenarios. The test results also improve in quality when you have greater fidelity in the models associated with the load scenarios.

Part of McPhee’s goal is to increase the level of mathematical rigor embodied in the models. Although math has always

Modeling green batteries

Figure 1: McPhee’s math-based chemistry model of a Lithium battery in MapleSim

Sustainable engineering triggers advancements in engineering modeling

been a part of engineering modeling, the actual equations have typically been buried deep inside compiled code with no way for users to modify or enhance the structure. In light of emerging challenges, engineers are now paying keen attention to new tools that embody a “symbolic” (as opposed to numeric) approach to model definition. The word symbolic is synonymous with “open” in the sense that the software maintains the full algebraic structure of the equations so that you can easily see and work with the equations. So if you need to include non-linear effects, or true dynamics as opposed to static or statistical approximations, the task is greatly simplified through the direct manipulation of the model and component equation definitions.

McPhee’s team applied the MapleSim system, which is natively symbolic, to model key battery types such as NiMh and Lithium. He was able to take into account electrical and chemical effects in the models. Other models also capture temperature effects. In the case of a chemistry-based model of a lithium battery, the math problem is particularly tricky as the chemical reactions are typically modeled by partial differential equations (PDEs) whereas virtually all system-level modeling tools are based on ordinary differential equations (ODEs).

McPhee’s team found, however, that within the flexible framework of a symbolic math system, they were able to mathematically transform the PDE problem to a more workable set of equations which was then solvable

by a well-known technique called Galerkin’s method. The key point here is not so much that one can implement Galerkin’s method but that, within a modern engineering tool like MapleSim, you have a direct connection into the mathematical formulations without paying the traditional penalty of requiring armies of highly specialized staff.

In terms of the overall vehicle system, the various battery models were integrated into a high-fidelity full vehicle model that includes a mean-value engine model and a complete suspension (including tires). As evidence of the potential of increased symbolic manipulation of model mathematics, this particular HEV model, without pre-processing, would have required 605 differential algebraic equations (DAEs). With pre-processing, McPhee’s team was able to reduce the computation load to 41 DAEs. In cases where the model ultimately runs in realtime for hardare-in-the-loop (HIL) testing, this reduction in complexity becomes significant and can sometimes be the difference between feasibility and infeasibility for HIL.

In the long term, the battery research feeds into the larger goal of optimizing the control of HEV powertrains. The HIL simulation would have already been used to test new controllers based on advanced formulations including fuzzy logic and Pontryagin’s Minimum Principle.

Figure 2: HEV with mean-value ICE model, electrical subsystem, and suspension in MapleSim

Sustainable engineering triggers advancements in engineering modeling

Sustainable engineering triggers advancements in engineering modeling

Another interesting initiative was launched by General Motors, also in collaboration with WatCAR. This initiative is focused on EV applications in particular, but the larger context is a vision of a truly green and intelligent transportation infrastructure that integrates the vehicle, the roads, and the electrical network required for practical implementation of the vision. The research leader is the WatCAR Executive Director, Dr. Amir Khajepour.

At the vehicle and component modeling level, there are some strong parallels between the Toyota/McPhee initiative and this project. A very significant aspect of the project deals with intelligent power management tools – that is, the development of algorithms that will identify optimal engineering parameter values that will ensure minimal fuel consumption. The requirement for a good foundation of models is clear and Khajepour has also chosen the symbolic

approach to accelerate the modeling phase. But for the task of smart algorithms for predicting system-level fuel consumption, the answers came, literally, from Mars.

Earlier, Khajepour collaborated with the Canadian Space Agency (CSA), best known for their work in designing the various large robotic manipulators on the International Space Station and Space

Shuttle missions, to develop a terrestrial simulation platform for optimal power management of space rovers – i.e. land vehicles deployed on Mars, the Moon, etc. The key question of how to go from point A to point B on an alien landscape must be answered intelligently to ensure the success of a mission. Should the rover take a geometrically direct route straight up a hill, or a winding route with shallower slopes? What about terrain and environmental influences? Ultimately, this becomes a complex optimization problem.

The research concept was a rich system of models for major rover subsystems, including terrain and environment, a terrestrial HIL test platform that emulates the dynamics of the rover, solar conditions, and genetic algorithms for the task of parameter optimization. For this project, as well, symbolic model formulation techniques became a key factor. For the fundamental physical models of systems, as expected, the direct treatment of the mathematical detail of

Figure 3: Concept for a green transportation system of the futureSource: www.greenits.ca, used with permission

Green engineering from the red planet

Figure 4: Main simulation model for rover dynamics developed in MapleSim and converted to an S-function

Sustainable engineering triggers advancements in engineering modeling

the model equations allowed fairly rapid development of the core rover dynamics (the estimate was approximately 2 person-weeks). Through code generation tools, the team was able to deploy the main model in downstream tools such as Simulink®, LabVIEW™, and the custom genetic algorithm code under development.

In this particular case study, the innovation is not so much in the individual elements but the overall acceleration of the workflow through the increased use of mathematics via symbolic computation. The automation of conventionally manual steps such as equation derivation and coding is one very significant benefit, but the ability to quickly deploy the code in more advanced algorithms required for optimization is also facilitated by access to the equations.

These two case studies illustrate the successful attainment of challenging engineering goals that arose through the push towards sustainable design. But in the end, there is no magic in designing for sustainability. In fact,

just the opposite seems to hold. Better green design comes from increasing the number of design factors that you consider, rigorously testing in multiple ways, from pure virtual to HIL to full prototypes, and finding efficient ways to systematically sort through your design space to identify optimal configurations. These goals are all very intuitive to the engineering community, and in some sense, they are at the heart of all engineering projects. The main difference is that the range of tools that

are emerging is starting to open up and provide greater access to the underlying mathematics to facilitate the necessary information and knowledge flow among the different phases. These tools are offering ways to manage the internal mathematical complexity and make these steps more accessible to a wider range of engineers.

Figure 5: Sample optimization results

www.maplesoft.com | [email protected] • Toll-free: (US & Canada) 1-800-267-6583 | Direct:1-519-747-2373 MapleSim is a trademark of Waterloo Maple Inc. Simulink is a registered trademark of The MathWorks, Inc. LabVIEW is a trademark of National Instruments. All other trademarks are the property of their respective owners.

www.maplesoft.com

A C y b e r n e t G r o u p C o m p a n y