Academics

Papakonstantinou to use NSF CAREER award to optimize structural life decisions

UNIVERSITY PARK, Pa. — Kostas Papakonstantinou, assistant professor of civil engineering at Penn State, will advance innovation in optimal structural engineering decision-making, thanks to a National Science Foundation (NSF) Early Career (CAREER) award.

“At the core of every engineering problem lies a decision-making quest, either directly or indirectly,” Papakonstantinou said. “Despite this fact, current engineering decision-making methodologies are largely solely dependent on the useful, but static, traditional cost-benefit analysis framework.”

In Papakonstantinou’s project, titled “Optimal Engineering Decision-making Under Uncertainties for Enhanced Structural Life-cycle,” the problem will be approached from a stochastic optimal control perspective, embracing and fully integrating predictive physics-based stochastic models and uncertain life-cycle observations and data.

The idea came to Papakonstantinou after reading a paper on autonomous robotic exploration.

“A robot wants to navigate and perform a certain task on a terrain based on partial and uncertain information from its sensors and with actions that may or may not be perfectly executed,” Papakonstantinou said. “It also wants to do it without using much of its power for these actions. This is a sequential decision problem with a long-term objective since the robot may locally observe its surroundings and then it decides on an action and movement according to its updated information. The process then repeats at its new location. In the end, the goal is to successfully complete the exploration task with minimal energy consumption and expense.”

The same problem applies to structural systems.

“A structural system wants to satisfy its objectives — safety, use, reliability— during its life-cycle with minimum cost,” Papakonstantinou said. “This is again a sequential decision problem with a long-term objective.”

In this scenario, the decision-maker may choose to observe the structural states at various times, based on whether the value of the information is greater than the cost of observation, and then the decision-maker decides how to proceed at each time based on the probabilistic belief about the system states.

By approaching decision-making from a stochastic optimal control and reinforcement learning perspective, this data-supported process naturally leads to condition-based estimation rules for structural systems, an enhanced approach to performance-based analysis, and a reconsideration of the traditional approaches to structural design, analysis, retrofit and maintenance. This turns a conventional static optimization problem into an integrated lifelong controlling process.

“In developed countries, the majority of infrastructure systems have already been established, are aging and are even approaching the end of their designated lifetime,” Papakonstantinou said. “Completely rebuilding the infrastructure, however, is not an option and methods for optimum allocation of resources are needed.”

The $500,000 CAREER award will allow Papakonstantinou to recruit graduate students, support a post-doc and participate in educational activities in order to advance this research.

Papakonstantinou joined Penn State in 2015. He received a five-year diploma and a master’s degree in civil engineering from the National Technical University of Athens and another master’s degree and doctorate in civil engineering from the University of California, Irvine.

The NSF CAREER award program seeks to support early-career faculty who have the potential to serve as academic role models in research and education and to lead advances in the mission of their department or organization. Activities pursued by early-career faculty should build a firm foundation for a lifetime of leadership in integrating education and research.

Kostas Papakonstantinou, assistant professor of civil engineering at Penn State, recently received a National Science Foundation Early Career (CAREER) award. Credit: Penn StateCreative Commons

Last Updated May 15, 2018

Contact