Overview of Multiscale Systems Engineering
Fundamental advances in physical and biological sciences and the development of new measurement and characterization tools have made it possible to understand spatial and temporal phenomena on the atomic, molecular, microscopic, and macroscopic scales. Microelectronics has led this revolution through the development of integrated circuits with shrinking scales, increasing density, and faster speeds. Knowledge of biological systems is following a similar course as understanding focuses on molecular and cellular processes. Recent progress in nanotechnology has extended the envelope of scales, making it possible to design materials and devices designed with nanoscale building blocks. New processes, ranging from self-assembly to directed molecular evolution, are changing synthesis and manufacturing.
The ability to translate these advances into the engineering of new products and industries will require a transformation in the methodologies of engineering modeling, simulation, and design. Interactions at all scales affects the ultimate behavior of the complete system, and engineers must learn to model and design across this range of scales. While many physical and biological principles are specific to their domains, the challenge of representation across scales, automatic synthesis of reliable simulations, optimization of design decisions, propagation of uncertainty across scales, and validation of multiscale methods are core issues that map across applications, and can best be addressed by a systems approach to engineering design.
Figure 1. Components.
Multiscale systems engineering research is concerned with the new technologies and design paradigms required to optimally account for multiscale interactions in materials, devices, and systems. The application of multiscale systems engineering will lead to the development of optimized products and processes (Figure 1). Engineers using these tools will be supported by (i) a hierarchy of models that provide a consistent description of multiscale phenomena, (ii) adaptive simulation methods that account for scale interactions, (iii) efficient computational analysis, optimization and control methods and (iv) the representation of uncertainty and its propagation.
As we look forward to the creation of powerful multiscale systems engineering technologies, it is easy to envision the pervasive influence they will have on product and process development. Consider the development process for a cardiovascular stent or a tissue-engineered replacement artery. The designer's goal is to provide a product that restores vascular function, remains compatible with the rest of the vasculature, and minimizes sites for clotting or cell proliferation that would result in a failure of the therapeutic procedure. There currently are no effective modeling and simulation tools to support the optimum design of the artery repair or replacement for long term performance by simultaneously considering the vascular system, local flow features, vessel wall response, and cell response. Today's tools do not allow the designer to model the interactions between the device, blood flow, and vessel wall on system and cellular levels. Furthermore, there are no methods to determine what nano-scale surface coatings might inhibit cell proliferation or clot formation.
Figure 2. Multiscale systems engineering to accelerate solid-state lighting development roadmap.
Solid-state lighting provides another example of the important role that multiscale systems engineering will play in the development of future products. Currently, white light can be produced using blue LEDs through phosphor conversion at a lumen-for-lumen cost that is about 100 times that of incandescent lamps. However, it is projected that in 20 years, cost-effective white LEDs will be 16 times as efficient and will last 100 times longer (Figure 2), making them the main source of white light. Multiscale systems engineering techniques should be able to cut this predicted 20 year timeframe in half as indicated at the bottom of Figure 2. Recent research shows it is possible to mix different color LEDs to produce white light, which is more efficient because of the absence of Stokes losses. Also, improvements in efficiency can be achieved by eliminating dislocations in the LEDs. Current research to produce higher quality blue LEDs is almost exclusively done by trial and error, varying the composition, substrate material, and growth geometry. Multiscale systems engineering will make it possible to much more quickly determine the optimal composition, substrate material, growth geometry, and wavelengths for optimum spectral power distribution.
Similar issues hold true when considering such diverse applications as arrays of microfluidic devices for investigating enzymes and metabolic pathways in drug discovery or design of heterogeneous materials that optimally satisfy a set of performance requirements.
A recent project involving SCOREC faculty illustrates the importance of multiscale modeling on product and process design. These faculty members collaborated with an industrial sponsor to apply coupled multiscale mechanical and processing simulations that revealed the processing method being used caused local material buckling. Examination of unsatisfactory parts and experimental runs confirmed the predictions. Based on the knowledge gained from multiscale simulation, the processing procedures were successfully modified.
The challenge of multiscale systems lies in the development of design and optimization tools that fully span the spectrum of scales. This paradigm will carry physical principles across many scales and propagate functional characteristics to adjust parameters of the system. The optimal selection of such variables is critical to effective design and will be integral to system and product development approaches. This integration of modeling and simulation methods with design optimization across multiple scales is fundamental to the challenge and promise of multiscale systems engineering.
Research Program Development
SCOREC is developing a strategic research and education plan for the development of the field of multiscale systems engineering. The powerpoint presentations linked to below were presented at a planning meeting of industrial and government partners held in October 2002. The five technological thrusts defined in this plan are:
- Modeling Methods (Powerpoint, 3.9 MB) addresses the modeling methods and aspects of the uncertainty and interface to experiments barriers.
- Adaptive Multiscale Methods (Powerpoint, 3.7 MB) addresses the adaptive multiscale methods barrier.
- Design, Optimization and Control (Powerpoint, 1.9 MB) addresses the system optimization and control, and multiscale systems design barriers.
- Adaptive Simulation Methods (Powerpoint, 7 MB) addresses the adaptive simulation methods barrier.
- Multiscale Systems Testbeds and Validation (Powerpoint, 4.2 MB) addresses the multiscale system validation and aspects of the model uncertainty and interface to experiments barriers.
Key to the development of multiscale systems engineering is ensuring the ability of technologies developed to meet the need of critical applications. As part of this research program a large number of applications are being developed in conjunction with industry. This set of applications includes four featured applications in which there is a direct coupling with both industry and a university research center or laboratory focused on the application area that includes both modeling and experimental validation programs. The four featured applications (Figure 3) and the associated research center or laboratory are:
- Nanocomposites (Powerpoint, 2.7 MB) with the Nanotechnology Center at Rensselaer
- Optoelectronics Devices (Powerpoint, 3.5 MB) with Rensselaer Center for Integrated Electronics
- Biomolecular/Materials Interface (Powerpoint, 1.8 MB) with the Laboratory for Enzyme Technology and the Nanotechnology Center at Rensselaer
- Cardiovascular Devices (Powerpoint, 6.2 MB) with the Cardiovascular Biomechanics Laboratory at Stanford University