James T Allison - Optimal System Design

Our world is replete with complex systems; transportation, agricultural, energy, and other systems are interlaced together, and the dynamic interactions between technical systems, society, and the environment have important and lasting impacts on humanity. I approach systems as a designer, seeking to understand the effects of system design decisions, and to develop methods that enahnce our ability to make better design decisions. I'm particularly interested in exploring system design with an eye toward sustainability. My current position provides me with remarkable opportunities to survey design practice across a variety of industries, and to identify areas for improvement. Presently I am focusing on the following three topics:

Design for Energy Efficiency
Decomposition-based Methods for System Design
Industry Adoption of Advanced Design Techniques

Security and sustainability of energy systems is a critical issue of increasing importance. Reducing energy consumption, through energy efficiency measures and economic incentives, complements efforts to develop and deploy renewable energy systems. Recent studies indicate that energy efficiency improvements stand to yield rapid and substantial progress toward energy sustainability.

Advanced design techniques, such as design optimization, can aid engineers working to create systems that minimize energy consumption, while maintaining competative system performance and cost. Enhanced design processes can help on three fronts:

Refine existing systems: Designers can apply optimization techniques to reduce mass, improving durability, allow lower-impact manufacturing processes, and design systems that operate more efficiently.
Switch to more efficient technology: New technologies, such as electric drives for automobiles, offer significant energy efficiency improvements, but can be a challenge to implement. Fundamental design changes can propagate in unexpected ways. Emerging design tools can help expedite successful transition to new technology.
Accelerate deployment: The impact of new, more efficient, designs is tempered by the time required to them into production. Advanced system design techniques help move critical discoveries and decisions to earlier stages of the product development process, reducing time required to reach production.
Facilitate switch to new solutions: Better design may turn previously infeasible solutions into real alternatives. For example, current bicycle designs and current transportation infrastructure and culture impede adoption of human-powered transportation. More functional bicycle designs, as well as enhanced infrastructure and transportation policy, could make human-powered travel a mainstream solution.

Some relevant projects include electric, hybrid electric, and solar vehicle design, PHEV life cycle analysis, bicycle design, and passenger aircraft design.

While advanced design techniques can help develop energy efficient systems, deploying them on a scale large enough to make meaningful changes requires the right incentives. The technological solution is coupled with economic, cultural, and policy solutions.

Systems are composed of multiple interacting components, making the analysis and design of systems a complex task. Progress in energy efficiency requires effective strategies for managing system design. A common approach is to partition a system design problem into smaller, more manageable parts. Some design organizations decompose a system according to physical subsystem (e.g., powertrain, structure, chassis, etc. in automotive design), while others might partition according to discipline (e.g., thermal, structural, aerodynamic, and ergonomic analysis).

While a decomposition-based approach to system design can make design activites practicable, proper coordination of subsystem design tasks is essential to developing successful system designs. Tradeoffs and interactions must be understood and managed well. Inadequate coordination may lead to subsystems optimized individually without adequate consideration of overall system objectives, as well as detrimental effects from unexpected interactions.

Several formal coordination strategies have been developed to address these issues. Many are besed on optimization techniques, and enable the use of parallel processing in analysis and design activities. Dr. Allison's current research in the area of decomposition-based design includes the following topics:

Automated partitioning and coordination decisions: Applying decomposition-based design requires determining appropriate system partitions and coordination strategies. Dr. Allison showed that partitioning and coordination decisions are coupled in his doctoral work, and has developed graph-theoretical methods and evolutionary algorithms for automating these decisions. Dr. Allison is working to include additional factors in partitioning and coordination decisions that will enhance computational results and enable broader application of these methods.
Decomposition-based design of dynamic systems: Partitioning of dynamic systems is especially challenging due to interactions quantified by function-valued time histories. Some strategies include decoupling via diagonalization of state equations, temporal partitions, and approximation of function-valued quantities using low-dimension representations. The latter approach was employed in a research project at GM involving the design of hybrid electric vehicles (relevant article here). Dr. Allison is developing methods for automating the partitioning of dynamic systems.
Combined control and plant design: Many engineered systems operate using embedded control systems. In current design practice, one group of engineers designs the physical system, and then the control system designers work to create a control system that meets system design objectives, without opportunity to modify the physical design to enhance performance of the control system. This sequential approach does not account for coupling between physical system design and control design decisions. Simultaneous approaches, or "Co-Design", exploit this coupling to achieve superior results. Dr. Allison is investigating Co-Design methods intended to aid design organizations to transition to a Co-Design process.

Many engineered systems have been developed over the course of decades; much has been invested, and mature design organizations exhibit profound inertia. This impedes adoption of emerging design techniques that could accelerate progress toward energy sustainability. Why the resistance? How can we stimulate change? Potential solutions may include:

Adapt and align: Adapt advanced design methods to align with current organizational structure and dynamics. For example, some system optimization methods, such as analytical target cascading and collaborative optimization, are based on design processes in the automotive and aerospace industries, respectively.
Appropriate tools: Many advanced design methods are not available in commercial software applications. Designers must develop their own software implementations from scratch. Availability of software tools that mesh well with design engineers current workflow, while providing convenient access to advanced design methods, may enhance adoption.
Enhanced education and training: Many practicing engineers are unaware of emerging design methods, or are reluctant to use them due to a variety of reasons. Better penetration of these topics into engineering curriculum, particularly at the introductory level for undergraduates, would enhance awareness for engineers entering the workforce, while graduate courses and industry training programs could help experienced engineers develop expertise in advanced system design.
Incentive: Even if training is available, engineers and managers need to understand the potential impact of advanced design techniques. Motivation exists to maintain the status quo and continue to use the same formula for design; we need to generate motivation to move to the next level of design practice.