⚠️ Work in progress — data and figures are preliminary.

Solvent Streets — Pavement Management Viewer

Pavement condition analysis and cost forecasting for US cities.

What is Solvent Streets?

Solvent Streets is a planning-grade pavement estimator. It aggregates OpenStreetMap pavement geometry into a hex grid, applies an exponential Pavement Condition Index (PCI) decay model per road classification, projects treatment costs from PCI-banded cost tiers, and simulates multi-year maintenance budget scenarios.

It is not a substitute for field-measured pavement surveys or a formal asset-management system — its outputs should be read as order-of-magnitude planning inputs rather than engineering specifications.

Examples

Each example uses a different configuration. Select one to explore an interactive dashboard with maps, forecasts, and cost projections.

Bay Area, CA

Alameda, CA, Albany, CA, Berkeley, CA, Dublin, CA, Emeryville, CA, Fremont, CA, Hayward, CA, Livermore, CA, Newark, CA, Oakland, CA, Piedmont, CA, Pleasanton, CA, San Leandro, CA, Union City, CA, Antioch, CA, Brentwood, CA, Clayton, CA, Concord, CA, El Cerrito, CA, Hercules, CA, Lafayette, CA, Martinez, CA, Moraga, CA, Oakley, CA, Orinda, CA, Pinole, CA, Pittsburg, CA, Pleasant Hill, CA, Richmond, CA, San Pablo, CA, San Ramon, CA, Walnut Creek, CA, Belvedere, CA, Corte Madera, CA, Fairfax, CA, Larkspur, CA, Mill Valley, CA, Novato, CA, Ross, CA, San Anselmo, CA, San Rafael, CA, Sausalito, CA, Tiburon, CA, American Canyon, CA, Calistoga, CA, Napa, CA, St. Helena, CA, Yountville, CA, San Francisco, CA, Belmont, CA, Brisbane, CA, Burlingame, CA, Colma, CA, Daly City, CA, East Palo Alto, CA, Foster City, CA, Half Moon Bay, CA, Hillsborough, CA, Menlo Park, CA, Millbrae, CA, Pacifica, CA, Portola Valley, CA, Redwood City, CA, San Bruno, CA, San Carlos, CA, San Mateo, CA, South San Francisco, CA, Woodside, CA, Campbell, CA, Cupertino, CA, Gilroy, CA, Los Altos, CA, Los Altos Hills, CA, Los Gatos, CA, Milpitas, CA, Monte Sereno, CA, Morgan Hill, CA, Mountain View, CA, Palo Alto, CA, San Jose, CA, Santa Clara, CA, Saratoga, CA, Sunnyvale, CA, Benicia, CA, Dixon, CA, Fairfield, CA, Rio Vista, CA, Suisun City, CA, Vacaville, CA, Vallejo, CA, Cloverdale, CA, Cotati, CA, Healdsburg, CA, Petaluma, CA, Rohnert Park, CA, Sebastopol, CA, Sonoma, CA, Windsor, CA
98 cities hex 100m imperial

Denver Metro, CO

Denver, CO, Aurora, CO, Lakewood, CO, Arvada, CO, Westminster, CO, Centennial, CO, Boulder, CO, Thornton, CO
8 cities hex 100m imperial

Greater Boston, MA

Boston, MA, Cambridge, MA, Somerville, MA, Quincy, MA, Newton, MA, Medford, MA, Malden, MA, Waltham, MA
8 cities hex 100m imperial

Livermore, CA

Livermore, CA
1 city hex 100m imperial

Los Angeles, CA

Los Angeles, CA, Santa Monica, CA, Pasadena, CA, Long Beach, CA, Glendale, CA, Burbank, CA, Inglewood, CA, Torrance, CA
8 cities hex 125m imperial

Portland Metro, OR

Portland, OR, Beaverton, OR, Gresham, OR, Hillsboro, OR, Tigard, OR, Lake Oswego, OR, Milwaukie, OR
7 cities hex 80m metric

Top 50 Cities

New York, NY, Los Angeles, CA, Chicago, IL, Houston, TX, Phoenix, AZ, Philadelphia, PA, San Antonio, TX, San Diego, CA, Dallas, TX, Fort Worth, TX, Jacksonville, FL, Austin, TX, San Jose, CA, Charlotte, NC, Columbus, OH, Indianapolis, IN, San Francisco, CA, Seattle, WA, Denver, CO, Nashville, TN, Oklahoma City, OK, Washington, DC, El Paso, TX, Las Vegas, NV, Boston, MA, Detroit, MI, Louisville, KY, Portland, OR, Memphis, TN, Baltimore, MD, Milwaukee, WI, Albuquerque, NM, Fresno, CA, Tucson, AZ, Sacramento, CA, Atlanta, GA, Kansas City, MO, Mesa, AZ, Raleigh, NC, Colorado Springs, CO, Miami, FL, Omaha, NE, Virginia Beach, VA, Long Beach, CA, Oakland, CA, Minneapolis, MN, Bakersfield, CA, Tulsa, OK, Tampa, FL, Aurora, CO
50 cities hex 150m imperial

Methodology

This section describes the data sources, models, and assumptions behind the analysis presented in each dashboard.

Data sources

The exact sources and endpoints used for a given example are listed in that example's Config tab.

Decay model

Each road classification decays independently via

PCI(t) = PCI₀ · exp(−k · t)

where k is an annual decay constant that depends on the road classification. Higher-class roads (motorway, trunk, primary) decay more slowly than lower-class roads (residential, service) because they are built to thicker, more rigorous design standards and typically receive more frequent maintenance. Default values are derived from LTPP data reported in FHWA-RD-01-156, Long-Term Pavement Performance and ship as part of the forecast package; they are continental-US averages and do not account for local climate, traffic, or construction quality. A config may set a per-city decay_rate to tune for local conditions (e.g. freeze/thaw or road salt); that override is applied as the rate for a typical road and scales every road class proportionally, so the per-class ordering (higher classes decay slower) is preserved rather than flattened. Sidewalks decay on a separate, slower track and are not treated as a highway class.

Cost model

Treatment costs are banded by PCI: each band has a representative $/sq m value, and costs between bands are linearly interpolated at the tier midpoints, so the cost-versus-PCI curve is smooth rather than step-shaped. Above the highest anchor (the midpoint of the preventive tier) and below the lowest anchor (the midpoint of the reconstruction tier), the cost is clamped to that anchor's value rather than extrapolated. Default cost tiers are expressed in $/sq m and sourced from FHWA treatment-selection guidance; they are calibration inputs, not measurements, and local bid prices will differ. Roads and sidewalks use independent cost tiers because the treatment economics differ substantially.

Scenario comparisons

PVMT ships with three comparison runs driven by annual funding level, all using the worst-first allocation strategy (budget is spent on the lowest-PCI segments first):

A do-nothing baseline (no spend, uncontrolled decay) is shown alongside the funded runs for comparison.

The forecast library also implements a preventive-first strategy (prioritize highest-PCI segments that are still in the preservation window), but the default UI comparisons do not exercise it. Preventive vs. worst-first allocation is governed by per-strategy efficiency multipliers; those multipliers are illustrative calibration constants chosen to reflect the direction and sign of the effect reported in FHWA-HIF-12-042, Pavement Preservation: Preserving our Investment — that $1 of preventive maintenance is reported to avoid $6–$10 of future reconstruction — not to reproduce that benefit-cost ratio as a single-year spending efficiency.

Area growth

Optional compound annual growth applies to pavement area each year:

Area(y) = Area₀ · (1 + g)^y

where g is configured per city (default zero). This lets an example model a city that is still expanding its street network; it does not model demolition or removal.

Solvency metrics (streets/roads only)

The dashboard's Financials headline and the cross-city leaderboard report three solvency figures. They are computed on the roads/streets cohort only — the aggregate scenarios blend roads, parking, and sidewalks but cost the blend at road tiers, which would mis-price sidewalks, so an absolute dollar claim must be roads-only. They are derived from a worst-first run at the city's configured annual budget.

Three caveats apply to these figures specifically:

Assumptions and limitations

References