Analysts, Data Science/Mining, Decision Analysis/Science, Mathematics, Operations Research Professionals, Programming, Statistics, Systems Analysis, Other
FHWA’s Office of Transportation Policy Studies based in Washington, DC is seeking a talented Research Associate to support enhancements to the Highway Economic Requirements System (HERS). The HERS model provides national projections of highway condition and performance under alternative potential levels of investment over a 20-year horizon and yields estimates of the investment needs relative to alternative targets. The model is currently being re-coded from Fortran into Python, which will facilitate the incorporation of advanced optimization algorithms. The current algorithms are dated and unable to optimize the investment program over more than a single period at a time without adequately considering investment timing across the time horizon. The investment program consists of improvement options to more than 100,000 sections of highway in the model database. The computation run time is ` expensive because of the vast sample size, the existence of constraints on the total dollar amount of investment, the need to select among alternative potential improvements (i.e. resurface or reconstruct, add lanes, shoulder improvements) to a given highway section, along with inefficiency of the modeling codes. In addition, the quality of the data input from the Highway Performance Monitoring System (HPMS) impinges on the model and is in some need of improvement.
The Research Associate will assume a high level of responsibility for contributing to the next generation version of HERS by working both independently or collaboratively on a variety of tasks including setting the program of model development. Key among these tasks will be advising on the model re-code into PYTHON, which is being written by another FHWA office, and developing the new optimization framework. The latter involves remodeling the problem into a more efficient constrained model by reviewing all constraints and objectives in order to improve the mathematical model and remove redundancies and inefficiencies; and by developing an efficient solution algorithm and heuristic for the model as needed to improve runtime.
In addition to this responsibility, the Research Associate will contribute to other tasks including:
Assisting in reviewing and improving various modules within HERS, such as the travel demand forecasts;
Exploring possibilities for integrated computer applications of HERS and other models (FHWA is using outputs from HERS as inputs to a national economic model, USAGE; the models are run separately, so planning for appropriate formats for input/outputs of the models and development of an integrated application could improve model connectivity);
Ascertaining the minimum sample size needed to generate reliable model outputs (the current sample size is geared toward obtaining reliable descriptive statistics on the highway system at sub-State levels of geography; a smaller sample may suffice for the HERS model where the focus is national);
Reviewing the quality of the HPMS sample data, devising fixes or errors and omissions, and recommending changes to the data collection instruments;
Reviewing HERS to identify elements that could possibly be eliminated or simplified without any significant loss, but with benefits in faster run times, increased model transparency or integrity, or reduced costs for model documentation and updates, and recommend specific changes; and
Contributing to a range of modeling enhancements that involve engineering, economic, or statistical analysis (e.g., ways to factor into monetized benefits the increase in rider comfort from better pavement condition, to model the effects of traffic congestion on crash rates by severity level, and to fold into HERS safety-focused improvements that are currently outside the model scope).
*NRC reviews candidates quarterly; online applications will be open from March 1st through May 1st for March NRC review.
The selected applicant is expected to be an independent thinker and a self-starter. Talented mathematical modelers, preferably with a background in transportation engineering and economics, that possess innovative ideas and expertise in developing efficient optimization models and solution algorithms, analyzing transportation data sets, and statistical modeling are encouraged to apply. Additionally, applicants should have proficiency in algorithm development using optimization software (e.g. CPLEX, GAMS, AIMMS, and FICO Express), statistical programs (e.g., SAS, R), and programming languages (e.g., Python). Applicants should be proficient in using callable functions for the optimization software within API.