Abstract
After conflict and disaster, social stability is a high priority strategic goal for stakeholders. Reconstruction and infrastructure development close capacity gaps, gain popular support for governments and institutions, and stave off illegitimate authority. Development allows the population to resume their daily lives and the government to demonstrate its reach and capabilities. It is a means to undermine support for insurgents and illegal activity while fomenting order. Infrastructure portfolios with carefully determined characteristics can be explicitly selected with this in mind, constituting a system. An optimal infrastructure portfolio for such a nebulous environment should include robust design features. It well satisfies design criteria and demonstrates resistance to exogenous factors. A systems approach using agent-based modeling, response surface methodology, robust parameter design, and a local optima filter provides leaders with statistically distinct, locally optimal choices better informing infrastructure decisions in a complex environment by using noneconomic measures to recommend settings in support of population stability.
The meta-model robust design process is introduced as a systems engineering methodology to address infrastructure decisions in complex, adaptive environments with exogenous factors. The process is comprised of a several sub-components that trade accuracy (bias) for robustness. The robust features of the methodology include robustness in regression and robustness in parameter design. The local optima filter is able to differentiate between control variable recommendations when response confidence intervals and associated statistical tests fail to do so.
The meta-model is applied using known functions to demonstrate its properties. It is also applied to a post-combat infrastructure selection scenario in the city of Jalalabad, Nangarhar Province, Afghanistan. Finally, it is applied to infrastructure policy selection in Tijuana City, Baja California, Mexico. Though application, the meta-model robust design process provides stakeholders with recommendations that might not be otherwise selected due to the myriad permutations, the curse of dimensionality, heteroscedasticity, the desire for robustness, and the use of non-economic assessment measures.
