Author Type

Graduate Student

Date of Award

Summer 6-26-2026

Document Type

Dissertation

Publication Status

Version of Record

Submission Date

July 2026

Department

Civil, Environmental and Geomatics Engineering

College Granting Degree

College of Engineering and Computer Science

Department Granting Degree

Civil, Environmental and Geomatics Engineering

Degree Name

Doctor of Philosophy (PhD)

Thesis/Dissertation Advisor [Chair]

Evangelos I. Kaisar

Abstract

Freeway Service Patrol (FSP) programs are central to Traffic Incident Management (TIM), delivering rapid response, clearance, and motorist assistance on high-volume corridors. The prevailing continuous roaming patrol model provides broad coverage but generates substantial non-productive vehicle-miles traveled (VMT), increases responders’ exposure to risk, and treats Service Level Agreement (SLA) compliance as a statistical outcome rather than a fixed operational constraint. Existing literature lacks a framework that simultaneously optimizes corridor segmentation, jointly deploys patrol and staged units across space, time, and direction, and enforces response-time SLA as a binding constraint.

This dissertation introduces the Segmental-Spatio-Temporal Hybrid Service Model (SSTHSM), a two-tier optimization framework. Tier 1 uses a dynamic program to partition the corridor into feasible service segments based on demand, speed, and SLA constraints. Tier 2 applies a Mixed-Integer Linear Program (MILP) to allocate patrol and staged units across each cell (segment × time-period × direction), minimizing daily operating cost while enforcing a strict 15-minute per-cell SLA. The formulation incorporates industry standard operational policies, including interchange U-turn penalty for cross-direction dispatches and constraints aligned with most DOT FSP practices. A scale-relaxation heuristic ensures convergence between segmentation and allocation.

The model is calibrated using five years (2020–2024) of FDOT D4 SunGuide data for a 20.685-mile section of I-95 in Palm Beach County Florida. From 8,705 countywide lane-blocking events, 4,076 valid incidents feed the model, with nearly equal directional distribution and a predominance of crash events. The baseline operation consists of continuous patrol with 11 vehicle-shifts per day (4+4+3), generating 4,360 VMT/day at a cost of $5,157/day.

The optimized MILP solution deploys three vehicles (1 patrol, 2 staged) strategically located along the corridor for 24 hours. This configuration reduces fleet size to 9 vehicle-shifts (−18.2%) and peak concurrent vehicles to 3 (−25%), while achieving an 85.2% reduction in VMT (to 644 mi/day) and a 20.9% cost reduction (to $4,080/day), yielding approximately $393,000 in annual savings. Crucially, it maintains 100% SLA compliance across all cells, with a maximum response time of 12.97 minutes, well within the 15-minute threshold.

These operational improvements also produce policy-relevant environmental and safety screening estimates. Applying MOVES-based running-emission factors to the 3,716 mi/day reduction in responder VMT yields potential avoided running emissions of approximately 1,743 metric tons of CO₂, 2.44 metric tons of NOₓ, and 57 kg of PM₂.₅ annually. These values are reported as VMT-based screening estimates and do not represent a complete net emissions inventory because staged-unit idling, auxiliary loads, and engine start/stop activity are not explicitly modeled. Separately, the bounded response-time envelope and reduced responder exposure are estimated to prevent approximately 12–18 secondary crashes per year.

The dissertation contributes by (1) introducing a hybrid optimization framework integrating segmentation and deployment under strict SLA constraints, (2) formalizing segmentation feasibility as a prerequisite to allocation, (3) providing empirical validation using multi-year operational data, and (4) quantifying environmental and safety co-benefits alongside cost and performance. The findings demonstrate that data-driven, policy-aware optimization can significantly improve FSP efficiency while maintaining reliability and enhancing broader system outcomes.

Download

Share

COinS