FIU Florida International University
Moss School of Construction
Worlds Ahead — College of Engineering & Computing
Ph.D. Dissertation Research  ·  Working Analytical Tool

Interpretable Network Metrics for Equity-Aware Prioritization of Bridge Rehabilitation and Recovery

Five-State Bridge Data & Formula Analysis Tool

Prepared byMostafa Sameer, PhD Candidate & Graduate Research Assistant
InstitutionFlorida International University (FIU)
DepartmentMoss School of Construction
Faculty AdvisorDr. Wallied Orabi, Associate Professor

Interactive five-state bridge-data analysis using the supplied Excel/CSV records and the documented state-specific formulas and decision rules in the research workbook.

New York · California · Florida · Pennsylvania · Texas  |  Reference year: 2020  |  133,554 supplied bridge records

Evidence BaseDecision ArchitectureState SystemsData DictionaryResearch GatesPrimary Sources
133,554bridge records across five supplied state datasets
14common variables in the current analytical schema
5core states: FL · TX · PA · CA · NY
2020reference year used for age calculations in this study build

Research objective — rehabilitation prioritization under limited resources

Decision question: if a DOT has a finite rehabilitation/replacement budget and many candidate bridges, what defensible evidence should determine which bridges receive priority?

This platform separates condition, eligibility, risk, network consequence, treatment recommendation, programming constraints, and post-score engineering review. It does not assume that one score or one state's formula is transferable to another state.

Data integrity rule. All bridge-level values displayed in this tool come from the five supplied datasets. The current files contain 14 common fields: structure number, owner, NHS code, overall condition, deck/superstructure/substructure/culvert condition ratings, structural evaluation appraisal, inventory rating, year built, AADT, and deck area. Material, lane count, bridge type, truck percentage, detour, scour, fracture-critical status, flood status, postings, and equity are not present in these supplied files and are therefore not fabricated.

What the evidence establishes

Evidence discipline

AVAILABLE = in supplied filesDOCUMENTED = verified policy/sourceGAP = required but absentDO NOT INVENT = no proxy substitution

Public priority-list fields are not automatically weights. A factor can be present in a dataset yet have no documented role in the official selection system. Conversely, a policy can require a factor that is absent from the current analytical extract.

Decision architecture used throughout the research

1 · Candidate universeDefine bridge population and data vintage
2 · EligibilityIdentify bridges that qualify for a program/treatment
3 · PriorityScore, rank, tier, or order eligible candidates
4 · ConsequenceAccount for traffic, detour, risk, reliability, etc.
5 · TreatmentRehab, replacement, maintenance, countermeasure
6 · ProgrammingBudget, timing, bundling, engineering review

Key methodological rule: a priority rank, a bridge condition score, a treatment recommendation, program eligibility, and a funded project are different outcomes. They must not be collapsed into one dependent variable without justification.

Five-state evidence map — current role, not invented universal weights

State / systemDecision mechanismDocumented factors / rulesUse in this platform
Florida
FDOT bridge program / BMS
Condition classification + management/programming; current statewide composite rehab score not established here.Current Performance Rating: 8–9 Excellent; 6–7 Good; 5 Fair; ≤4 Poor. Historical SD/FO/Health Index/Sufficiency Rating are not current FDOT calculations.Condition analysis only from supplied data; no invented Florida score.
Texas
HBP / Category 6
Eligibility + ordered priority until funds are exhausted.Poor first; selected Fair cases include narrow deck, scour criticality, debris/scour history, and steel/timber piles. Safety programs add other thresholds.Poor population is assessable; missing width/scour/history/pile variables remain gaps.
Texas
TEBSS
0–100 eligible bridge selection score.Current TEBSS existence/use verified; historical published weighting structure must not be presented as today's equation without current verification.System acknowledged; no fabricated current weights.
Pennsylvania
Bridge Risk Score
Multiplicative risk model used for state-owned bridge risk/ranking and programming support.(√Deck Area × AADT) × F_scour × F_fc × F_det × F_aadtt × F_flood; condition is assessed separately.Formula documented; cannot calculate fully from current 14 fields.
New York
BRIDGE NY LBPI
Weighted 0–100 index + qualitative/engineering review.Capital Need 47.5; Facility Importance 27.5; Restrictions 5; Risk/Resiliency 20. Includes general recommendation, structural deficiency, model recommendation, AADT, truck %, detour, postings, fracture critical, material/design, hydraulic vulnerability.Framework documented; not calculable from current extract.
California
SHOPP / HBP
Performance-driven, fiscally constrained programming and priority lists.Current SHOPP emphasizes condition, seismic/scour vulnerability and operational limitations; HBP lists eligible scopes and priority-list fields. Historical scoring models are kept separate.Current process described without inventing a current universal bridge formula.

Analytical data dictionary — supplied five-state schema

FieldRole in analysisAvailable?Methodological note
State NameState stratificationYESFive-state comparison variable.
Structure NumberBridge identifierYESBridge-level traceability.
Owner AgencyOwnership/program contextYESUseful for agency segmentation; not a universal priority weight.
Inventory Route NHS CodeNetwork classificationYESUsed descriptively/exploratorily; not assumed equivalent to criticality.
Bridge ConditionPrimary condition classificationYESGood/Fair/Poor as supplied; state-specific interpretation must be retained.
Deck / Superstructure / Substructure / Culvert ratingsComponent conditionYESWorst applicable component is a diagnostic measure; do not silently replace source overall condition.
Structural Evaluation AppraisalStructural screening variableYESPreserved as supplied; interpretation requires source documentation before modeling.
Inventory Rating (US Tons)Capacity-related descriptive variableYESPotential engineering screening variable; not treated as a universal priority weight.
Year BuiltAgeYESValidated before age calculation; suspicious PA records require source verification rather than deletion.
Average Daily TrafficTraffic exposure / consequence proxyYESLog transform may support exploratory modeling; no universal official weight is claimed.
Deck Area (sq ft)Bridge size / exposureYESDirectly used in PA's documented risk formula; elsewhere treated as a descriptive/exploratory variable.

Paper 1 — comparative evidence stage

Critically compare actual state systems: eligibility, factors, formulas, thresholds, weights, decision stage, treatment type, engineering review, and programming constraints. Identify which factors are common, state-specific, or absent.

Output: defensible cross-state decision architecture and factor taxonomy.

Paper 2 — resource-constrained optimization

Only after the target decision, benefit definition, treatment type, cost basis, and constraints are established should optimization be formulated.

Output: constrained selection of projects under a finite budget or other resource limit, grounded in Paper 1 rather than arbitrary weights.

Primary policy / technical sources used for the evidence layer

AgencySourcePurpose
TxDOTHighway Bridge ProgramCurrent HBP eligibility and priority order.
TxDOTBridge System Safety ProgramHigher-risk bridge and safety criteria.
TxDOTAdvanced PlanningRehab vs replacement considerations.
PennDOT2023 Transportation Asset Management PlanBridge Risk Score and risk-factor definitions.
PennDOTPublication 55Risk assessment and programming context.
NYSDOTBRIDGE NY 2023 Planning Process GuidanceLBPI weights and decision stages.
NYSDOTBRIDGE NYProgram purpose and current framework.
FDOTBridge InformationCurrent Florida bridge condition classification and historical metrics note.
Caltrans2026 SHOPPCurrent performance-driven and fiscally constrained programming context.
CaltransHighway Bridge ProgramCurrent HBP scope and priority-list context.

Historical formulas are explicitly labeled historical in the platform. They are not silently substituted for current policy.

Five-State Dataset Overview

Overall condition distribution
Poor-condition rate by state
LIVE DATA ANALYSIS

Interactive Five-State Bridge Analytics

Change the filters and the charts, bridge population, condition rates, and exploratory priority distribution update immediately from the embedded NBI-based records.

—bridges in current
analysis population

How to use this page

  • Select a state, condition, and NHS status.
  • Use the age and AADT controls to narrow the population.
  • Watch the histogram and state comparison update live.
  • Open Scoring & Prediction to tune factor weights.
Real supplied records No fabricated missing factors 2020 reference year Live recalculation in browser
Condition Distribution by State
Counts update with the current filters.
Poor-Condition Rate by State
Percentage of the filtered bridges classified as Poor.
Age Distribution
Valid Year Built records only.
Age vs AADT — Live Population
A sampled scatter plot for visual exploration; the underlying calculations use all filtered records.

Live Scoring & Prediction Lab

Research safeguard: this is an exploratory, tunable research score built only from fields available in the supplied five-state files. It is not an official DOT score and does not replace the documented NY/PA/TX/CA/FL decision systems. Missing state-specific factors are not invented.
Tune factor importance
Condition severity0.35
Good → Fair → Poor severity from the supplied condition field.
Bridge age0.25
Age normalized from validated Year Built.
AADT0.15
Log-transformed traffic exposure.
Deck area0.10
Log-transformed bridge size proxy.
Component severity0.10
Worst applicable deck/super/sub/culvert condition.
NHS exposure0.05
Exploratory network-classification factor.
What-if bridge scenario
0.000
Exploratory Priority Score · 0–1
MODERATE
Factor Weight Spider Chart
Live Score Histogram
Score distribution for the current analysis population.
Selected Bridge Score
Select a real bridge below to compare it with the what-if scenario.
RankStateStructureConditionAgeAADTComponent MinScore

Bridge-Level Data — Actual Supplied Records

The table is calculated directly from the supplied records. Select a state/condition and search to inspect individual bridges. No synthetic factor values are used.
StateStructureOwnerConditionDeckSuperSubCulvertYearAgeAADTDeck Area

Structural Components — No Aggregation Assumption

Component-level calculations use the NBI condition ratings supplied in fields 58, 59, 60, and 62. Numeric ratings are used; “N” is treated as not applicable/not rated. The component minimum is a descriptive worst-applicable-component metric, not a new weighted score.
StateBridgesDeck poor ≤4Superstructure poor ≤4Substructure poor ≤4Culvert poor ≤4Any component poorKnown component min
Poor component counts
Overall vs. component classification

Progressive Bridge Population Filtering

This is the advisor-requested research workflow: reduce the large population transparently before detailed calibration. Filtering is descriptive; it does not imply a rehabilitation decision.
StateAll bridgesValid Year BuiltKnown conditionKnown componentFair/PoorPoor
Age groups (2020 reference year)
Suspicious Year Built records

Formula Engine — Workbook-Based, Evidence-Constrained

Important: formulas are only calculated where all required inputs exist. Missing inputs are reported as unavailable rather than replaced with assumptions.

1. Bridge age

Age = 2020 − Year Built

Applied to every record with a valid Year Built.

2. Component minimum

Component Minimum = min(Deck, Superstructure, Substructure, Culvert), ignoring non-numeric “N” entries

Descriptive structural-condition metric. No arbitrary component weights.

3. Component poor indicator

Component Poor = 1 if any applicable component rating ≤ 4; otherwise 0

Used only to check the supplied overall condition field against component data.

4. Pennsylvania Bridge Risk Score

Risk = (sqrt(Deck Area) × AADT) × F_scour × F_fc × F_det × F_aadtt × F_flood

Workbook formula. Not calculated because the supplied files do not contain scour, fracture-critical, detour, truck percentage, or flood status.

5. New York BRIDGE NY LBPI

LBPI = Capital Need 47.5 + Facility Importance 27.5 + Restrictions 5 + Risk/Resiliency 20

Workbook subfactors include General Recommendation, Structural Deficiency, Model Recommendation, AADT, Truck %, Detour, Postings, Fracture Critical, Material Type, Design Type, Hydraulic Vulnerability. Not calculated from the supplied five-state files because those required inputs are incomplete.

6. Texas current HBP

Priority order: Poor → selected Fair cases (narrow deck, scour critical, debris/scour history, steel/timber piles)

The supplied data allow the Poor condition portion to be identified, but not the remaining criteria because width/scour/history/pile fields are absent.

7. Data-derived traffic exposure

log(AADT + 1)

Displayed as a descriptive transform only. It is not treated as a universal prioritization score or weight.

State-Specific Decision Systems & Factor Roles

The same variable is not automatically required in every state. This table follows the workbook's distinction between a factor being available, being used, and having a particular decision role.
StateSystemDecision roleDocumented formula/ruleCan calculate now?Missing / limitation
New YorkBRIDGE NY LBPIWeighted score47.5 Capital Need + 27.5 Facility Importance + 5 Restrictions + 20 Risk/ResiliencyNot calculable from supplied five-state filesRequires General Recommendation, Structural Deficiency, Model Recommendation, AADT, Truck %, Detour, Postings, Fracture Critical, Material Type, Design Type, Hydraulic Vulnerability.
PennsylvaniaBridge Risk ScoreMultiplicative risk formula(sqrt(Deck Area) × AADT) × F_scour × F_fc × F_det × F_aadtt × F_floodNot calculable from supplied five-state filesScour, fracture-critical, detour, truck traffic, and flood status are not in the supplied 14-field datasets.
TexasHBP / Category 6Eligibility + priority orderPoor; then selected Fair cases: narrow <26 ft, scour critical, short spans over water with debris/scour history, steel/timber pilesPartially assessableThe supplied files have condition but not deck width, scour, span/debris history, or pile material.
TexasTEBSS / Priority 10–100 selection scoreCurrent existence/use verified; full current attribute-weight equation not suppliedNot calculableDo not substitute historical TEBSS weights.
California2026 SHOPPPortfolio/programmingPerformance-driven and fiscally constrained; no universal current bridge-level weighted formula in the supplied workbookNot calculable as a single scoreCurrent public process is not represented by a single bridge-level formula in the workbook.
CaliforniaHBPEligibility + priority listReplacement/rehab/scour/preventive-maintenance eligibility and constrained programmingNot calculable as a single scorePublic priority-list fields should not automatically be treated as weights.
FloridaCurrent bridge conditionCondition classification8–9 Excellent; 6–7 Good; 5 Fair; ≤4 PoorCondition calculable from supplied CAT10; component cross-check calculableA current statewide composite rehabilitation score was not verified in the workbook.

Advisor-Requested Data Gaps

Factor / VariableCurrent five-state filesActionReason
Deck / Superstructure / Substructure / CulvertAVAILABLEUse directlySupports component-level variation and condition checks.
Material / structure typeNOT AVAILABLEJoin full NBI fieldsNeeded to segment concrete/steel/timber/type populations.
Lane countNOT AVAILABLEJoin NBI/state fieldsNeeded for 1-, 2-, 3-, 4+ lane filtering.
Truck percentageNOT AVAILABLEState-specific sourceUsed in PA/NY/TX contexts but not present here.
Detour lengthNOT AVAILABLEExternal/later modelAdvisor said keep it external for now; develop later if required.
Scour / floodNOT AVAILABLEVerify state role firstDo not force flooding/scour into every state.
Fracture criticalityNOT AVAILABLEState/NBI joinCandidate risk modifier; role differs by state.
Postings / load restrictionsNOT AVAILABLEState/NBI joinEngineering/eligibility role must be preserved.
Equity / community impactNOT AVAILABLELater research layerNo current state formula in the workbook uses a published equity weight.
Year BuiltAVAILABLEValidate outliersSix PA records have Year Built before 1800 and require source verification.
Next modeling gate: obtain the missing bridge-type/material/lane and state-specific decision variables first. Only after that should a state-specific rehabilitation model or optimization formulation be built.