Olivier L. de Weck

A COMPARISON OF PARTICLE SWARM OPTIMIZATION AND THE GENETIC ALGORITHM

Particle Swarm Optimization (PSO) is a relatively recent heuristic search method whose mechanics are inspired by the swarming or collaborative behavior of biological populations. PSO is similar to the Genetic Algorithm (GA) in the sense that these two evolutionary heuristics are population-based search methods. In other words, PSO and the GA move from a set of points (population) to another set of points in a single iteration with likely improvement using a combination of deterministic and probabilistic rules. The GA and its many versions have been popular in academia and the industry mainly because of its intuitiveness, ease of implementation, and the ability to effectively solve highly nonlinear, mixed integer optimization problems that are typical of complex engineering systems. The drawback of the GA is its expensive computational cost. This paper attempts to examine the claim that PSO has the same effectiveness (finding the true global optimal solution) as the GA but with significantly better computational efficiency (less function evaluations) by implementing statistical analysis and formal hypothesis testing. The performance comparison of the GA and PSO is implemented using a set of benchmark test problems as well as two space systems design optimization problems, namely, telescope array configuration and spacecraft reliability-based design.

Staged Deployment of Communications Satellite Constellations in Low Earth Orbit

The “traditional” way of designing constellations of communications satellites in low Earth orbit is to optimize the design for a specified global capacity. This approach is based on a forecast of the expected number of users and their activity level, both of which are highly uncertain. This can lead to economic failure if the actual demand is significantly smaller than the one predicted. This paper presents an alternative flexible approach. The idea is to deploy the constellation progressively, starting with a smaller, more affordable capacity that can be increased in stages as necessary by launching additional satellites and reconfiguring the existing constellation in orbit. It is shown how to find the best reconfigurable constellations within a given design space. The approach, in effect, provides system designers and managers with real options that enable them to match the system evolution path to the actual unfolding demand scenario. A case study demonstrates significant economic benefits of the proposed approach, when applied to Low Earth Orbit (LEO) constellations of communications satellites. In the process, life cycle cost and capacity are traded against each other for a given fixed per-channel performance requirement. The benefits of the staged approach demonstrably increase, with greater levels of demand uncertainty. A generalized framework is proposed for large capacity systems facing high demand uncertainty.

Degree of Modularity in Engineering Systems and Products with Technical and Business Constraints

There is consensus that modularity has many benefits from cost savings due to increased commonality to enabling a higher variety of products. Full modularity is, however, not always achievable. How engineering systems and products whose design is heavily influenced by technical constraints, such as weight or size limitations, tend to exhibit rather integral architectures is shown in this study. For this, two metrics are defined on the basis of a binary design structure matrix (DSM) representation of a system or product. The non-zero fraction (NZF) captures the sparsity of the interrelationships between components between zero and one, while the singular value modularity index (SMI) captures the degree of internal coupling, also between zero and one. These metrics are first developed using idealized canonical architectures and are then applied to two different product pairs that are functionally equivalent, but different in terms of technical constraints. Empirical evidence is presented that the lightweight variant of the same product tends to be more integral, presumably to achieve higher mass efficiency. These observations are strengthened by comparing the results to another, previously published, modularity metric as well as by comparing sparsity and modularity of a set of 15 products against a control population of randomly generated architectures of equivalent size and density. The results suggest that, indeed, some products are inherently less modular than others due to technological factors. The main advantage of SMI is that it enables analysis of the degree of modularity of any architecture independent of subjective module choices.

PRODUCT FAMILY AND PLATFORM PORTFOLIO OPTIMIZATION

In this paper, a methodology is presented to determine the optimum number of product platforms to maximize overall product family profit with simplifying assumptions. This methodology is attempting to aid various manufacturing industries who are seeking ways to reduce product family manufacturing costs and development times through implementation of platform strategies. The methodology is based on a target market segment analysis, market leader’s performance vs. price position, and a two-level optimization approach for platform and variant designs. The proposed methodology is demonstrated for a hypothetical automotive vehicle family that attempts to serve seven different vehicle market segments. It is found that the use of three distinct platforms maximizes overall profit by pursuing primarily a horizontal leveraging strategy.Copyright

Flexible and reconfigurable systems: Nomenclature and review

The demands on today’s products have become increasingly complex as customers expect enhanced performance across a variety of diverse and changing system operating conditions. Reconfigurable systems are capable of undergoing changes in order to meet new objectives, function effectively in varying operating environments, and deliver value in dynamic market conditions. Research in the design of such responsive and changeable systems, however, currently faces impediments in effective and clear discourse due to ambiguity in terminology. Definitions of the terms flexibility and reconfigurability, two related concepts in reconfigurable system design, are explored based on their original lexical meanings and current understanding in design literature. Design techniques that incorporate flexibility both in the design (form) and performance (function) space are presented. Based upon this literature survey, a classification scheme for flexibility is proposed, and its application to reconfigurable system design is explored. This paper also presents recent methodologies for reconfigurable system design and poses important research questions that remain to be investigated.

Assessing risks and opportunities of technology infusion in system design

Most new technologies only deliver value once they are infused into a parent system. While the literature on innovation itself is abundant, there is a lack of understanding and methodology in terms of evaluating both the risks and opportunities of new technologies not in isolation, but in terms of their integration into a parent system in a wider regulatory and competitive context. This paper presents a technology infusion assessment methodology to quantify the potential performance benefits of new technologies using multiobjective Pareto analysis. Moreover, the costs of infusing new technologies are also considered using the concept of architectural invasiveness relative to a baseline system. The degree of invasiveness of different system architectures is related to the amount of design change required to accommodate the new technology. This is quantified with a component-based change Design Structure Matrix (DSM). Risks and opportunities are quantified by evaluating the utility of future benefits and costs of a new technology against uncertain exogenous variables such as gains made by competing technologies and potential future regulatory actions. The technology infusion methodology is demonstrated for a hydrogen-enhanced combustion engine, where the effects of integrating a plasma fuel reformer are quantified and discussed in terms of fuel economy, NOx emissions, and add-on vehicle costs. The methodology is generally applicable to support quantitative analysis of technology infusion problems in system design.

Modeling Methods and Conceptual Design Principles for Reconfigurable Systems

Reconfigurable systems can attain different configurations at different times thereby altering their functional abilities. Such systems are particularly suitable for specific classes of applications in which their ability to undergo changes easily can be exploited to fulfill new demands, allow for evolution, and improve survivability. This paper identifies the main factors that drive the need for reconfigurability and proposes methods for modeling reconfigurable systems. A survey of 33 different reconfigurable systems is also presented to provide broader insights and general design guidelines for reconfigurable systems.

PLATFORM-BASED DESIGN AND DEVELOPMENT: CURRENT TRENDS AND NEEDS IN INDUSTRY

Many companies constantly struggle to find cost-effective solutions to satisfy the diverse demands of their customers. In this paper, we report on two recent industry-focused conferences that emphasized platform design, development, and deployment as a means to increase variety, shorten lead-times, and reduce development and production costs. The first conference, Platform Management for Continued Growth, was held November-December 2004 in Atlanta, Georgia, and the second, 2005 Innovations in Product Development Conference - Product Families and Platforms: From Strategic Innovation to Implementation, was held in November 2005 in Cambridge, Massachusetts. The two conferences featured presentations from academia and more than 20 companies who shared their successes and frustrations with platform design and deployment, platform-based product development, and product family planning. Our intent is to provide a summary of the common themes that we observed in these two conferences. Based on this discussion, we extrapolate upon industry’s needs in platform design, development, and deployment to stimulate and catalyze future work in this important area of research.

Multi-objective genetic algorithm for the automated planning of a wireless sensor network to monitor a critical facility

This paper examines the optimal placement of nodes for a Wireless Sensor Network (WSN) designed to monitor a critical facility in a hostile region. The sensors are dropped from an aircraft, and they must be connected (directly or via hops) to a High Energy Communication Node (HECN), which serves as a relay from the ground to a satellite or a high-altitude aircraft. The sensors are assumed to have fixed communication and sensing ranges. The facility is modeled as circular and served by two roads. This simple model is used to benchmark the performance of the optimizer (a Multi-Objective Genetic Algorithm, or MOGA) in creating WSN designs that provide clear assessments of movements in and out of the facility, while minimizing both the likelihood of sensors being discovered and the number of sensors to be dropped. The algorithm is also tested on two other scenarios; in the first one the WSN must detect movements in and out of a circular area, and in the second one it must cover uniformly a square region. The MOGA is shown again to perform well on those scenarios, which shows its flexibility and possible application to more complex mission scenarios with multiple and diverse targets of observation.

Flexible platform component design under uncertainty

Incorporating flexibility into product platforms allows manufacturers to respond to changing market needs with a minimal increase in product family complexity and investment cost. To successfully design a flexible product platform, proper design of flexible platform components is critical. These components can be described as “cousin” parts as they are neither completely unique nor completely common among variants. In this paper, a multidisciplinary process for designing flexible product platform components is introduced, assuming the platform component is decided a priori. The design process starts with identification of uncertainties and generation of multiple design alternatives for embedding flexibility into the component. Design alternatives are then optimized for minimum cost, while satisfying the component performance requirements. The flexible designs are then evaluated for economic profitability under identified uncertainty, using Monte Carlo simulation. At the end, the most profitable flexible component design is selected. The proposed design process is demonstrated through a case study, in which different flexible designs are generated and optimized for an automotive floor pan, an essential element of most vehicle product platforms. Results suggest that the way in which the flexibility is incorporated in the component, production volume trends, and the degree of built-in flexibility are important factors to consider when designing flexible product platforms.