Welcome
Interactive Map
Analysis
Purpose: This prototype presents a strategic facility-location analysis for North Carolina's interstate truck parking network. The objective is to identify where new rest areas and truck stops should be added along the major corridors (I-26, I-40, I-77, I-85, I-95) to reduce unserved peak-hour parking demand within an investment budget.
The dashboard combines an interactive map of corridors and facilities with a set of analytical views summarizing current system performance, the trade-off between cost and unmet demand, and the optimization model's preferred facility mix.
The Interactive Map shows corridor segments colored by base-case unmet demand and toggleable layers for existing facilities, candidate sites, and new facilities selected by the model at p = 10, 20, 30, 40, 50. The Analysis tab provides a system snapshot, demand patterns by interstate, capacity-utilization histograms, the detour-threshold sensitivity, a capacity-expansion alternative, and the optimal facility mix by p.
Demand Estimation. Peak-hour truck parking demand is estimated for each corridor segment using the FHWA Jason's Law method (FHWA 2002), scaled by AADTT (from NCDOT Traffic Volume Maps), peak seasonal factor (1.15), peak-hour factor, and class-specific facility-preference factors. Demand is decomposed across four driver classes: short-haul vs long-haul, crossed with rest-area vs truck-stop preference.
Candidate Sites. 123 NCDOT-owned parcels were shortlisted from 520 state-owned parcels in NC OneMap. Selection criteria: lying within a 10-mile buffer of I-26, I-77, I-85, or I-95, with capacity ≥ 10 spaces.
Demand Assignment. A two-stage system-optimal assignment is solved: stage 1 maximizes the total number of trucks parked subject to capacity and detour-distance constraints; stage 2, among all assignments achieving that maximum, minimizes total detour distance. Two detour thresholds are reported: 5-mile (stricter) and 10-mile (more permissive).
Facility Location Optimization. A Mixed-Integer Linear Program (MILP) parameterized by p (number of new facilities). For each p from 1 to 50, the model chooses which candidates to open and as which type (rest area or truck stop), subject to demand-balance, capacity, detour, and type-compatibility constraints (rest-area classes can only be served by rest areas; truck-stop classes only by truck stops). Solver: CPLEX.
Cost Specifications. Unit costs are taken from the NCDOT Phase II Truck Parking Study (Cambridge Systematics, Table 3.9):
| Description | Unit | Cost |
|---|---|---|
| Concrete Pavement with Curbs | Per Space | $75,000 |
| Asphalt Pavement, no Curbs | Per Space | $48,000 |
| Gravel Surface | Per Space | $37,000 |
| Vault Toilets | Per Site | $60,000 |
| Lighting | Per Space | $1,400 |
| Fencing (60" chain link) | Per Space | $2,500 |
The per-facility cost is built up as a fixed site cost plus a capacity-dependent pavement cost. For a new truck stop, the fixed component is $60,000 (vault toilets) + $1,400 (lighting) + $2,500 (fencing) = $63,900, and the capacity-dependent component is $75,000 × capacity (concrete pavement with curbs), giving a total of $63,900 + ($75,000 × capacity). For a new rest area, the fixed component is $1,400 (lighting) and the capacity-dependent component is $37,000 × capacity (gravel surface), giving a total of $1,400 + ($37,000 × capacity).
Lighting and fencing are modeled as fixed per-facility line items rather than scaled per space, on the assumption of a single shared installation per site. Existing-facility expansion is costed at the added capacity multiplied by the relevant per-space pavement rate.
Further reference: FHWA Truck Parking Development Handbook (HOP-22-027).
For more details, please read our full report here.
This research was supported by the Center for Rural and Regional Connected Communities (CR2C2), a Regional University Transportation Center funded by the United States Department of Transportation (USDOT). The primary design for the tool was developed by Komal Gulati and Venktesh Pandey, with refinement from the project 3-1 team.
Disclaimer: The views and accuracy of the information presented belong to the authors alone. The United States Department of Transportation assumes no liability for the contents or use thereof.
Komal Gulati (PhD Student, Electrical & Computer Engineering, NC A&T), Dr. Venktesh Pandey (Assistant Professor, Civil, Architectural, & Environmental Engineering, NC A&T).
A high-level view of the network: how big it is, how stressed it is, and what changes when the model adds 20 new facilities.
Peak-hour parking demand decomposed across the four FHWA driver classes for each of the five major NC interstates. Demand is computed using the FHWA Jason's Law method, scaled by AADTT, peak-hour factor, and class-specific facility-preference factors.
I-40, I-85 and I-95 together carry approximately 83% of statewide peak-hour truck parking demand. Long-haul truck-stop demand (Class 4) is the dominant component on every corridor, consistent with NC's role as a long-haul freight through-state.
The class-level breakdown directly drives the facility-type decisions downstream: rest-area classes (1, 3) can only be served by rest areas, and truck-stop classes (2, 4) can only be served by truck stops. Class composition is therefore the upstream cause of the truck-stop-heavy mix that the model chooses at every p.
Distribution of existing facilities across 5%-utilization bins under the base-case demand assignment (no expansion, no new facilities). Two thresholds are reported: 5-mile and 10-mile detour radius.
At the 10-mile threshold, 74% of corridor-adjacent facilities (72 of 97) operate at or above 95% utilization. Mean utilization is 85.5% and median is 100%. Even with the more permissive radius, the network is at saturation. At the 5-mile threshold, fewer facilities are reachable (94 in buffer) and 50% are at near-full capacity.
The right-most bin (95-100%) has roughly an order of magnitude more facilities than any other bin. A linear axis crushes the rest of the distribution; the square-root transform keeps single-digit bins legible while preserving the visual prominence of the saturation bar.
Residual unmet demand (left axis) and total cost (right axis) as a function of p, the number of new facilities added. Two curves per axis correspond to the 5-mile and 10-mile detour thresholds.
The first ~10 facilities cut unmet demand sharply; the next 10 less so; beyond p≈30 the curve flattens. Each additional facility serves smaller residual pockets of demand, while construction cost continues to rise approximately linearly.
The 5-mile curve sits above the 10-mile curve at every value of p. The gap is the demand that exists within 10 miles of a facility but outside 5 miles, i.e. the demand whose service depends on accepting a longer detour. This gap is itself a useful planning quantity.
The alternative to adding new facilities: expand existing facilities by a capacity multiplier f (1.0 = current, 4.0 = quadrupled). The chart plots residual unmet demand against f, with expansion cost on the secondary axis.
Even at f=4.0 (~$1.2B investment), unmet demand only drops to about 37%. This is because the existing network has spatial coverage gaps. Some demand segments lie beyond the detour radius of any existing facility, and no amount of capacity expansion at the existing sites can serve that demand.
This chart is the empirical justification for the facility-location problem itself: capacity is necessary but not sufficient, and spatial coverage is the binding constraint. The optimization in the next panel (p=20) achieves better coverage at less than half the expansion-only cost.
For each value of p from 1 to 50, the count of new rest areas vs new truck stops the MILP chooses to open. At p=20 the model selects 4 rest areas and 16 truck stops.
Long-haul demand dominates, and Class 4 (long-haul + truck stop) is the largest single class because of the FHWA facility preference factors (P_facility = 0.77 for truck stops, 0.23 for rest areas). The optimization therefore consistently prefers truck stops to maximize served demand per facility.
New rest areas and truck stops have different cost structures: a new RA costs $1,400 + ($37,000 × capacity); a new TS costs $63,900 + ($75,000 × capacity). Truck stops are roughly twice as expensive per space, but serve more demand, so the model's preference for them reflects that demand-weighted cost-per-served-truck still favors truck stops.