Application of AI in Railway Earthwork Project Verification | Presentation
© 2025 Autodesk. All rights reserved.
Rail Summit 2025 Design & Make the Future of Rail
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
05/11/2025
\
David Gurion Tiago Engineering Project Leader, Vale
Speakers
Civil Engineer
Master in Civil Construction from UFMG
BIM (Building Information Modeling) Specialist in Infrastructure Management and GIS
MBA in Business Management
Civil Engineer
Specialist in Geotechnical Engineering and Concrete Structures
MBA in Construction Management and Building Technology
Priscilla Cordeiro Digital Engineering Leader
\ Organizational Structure
Technical VP
Urban Projects Directory
Infrastructure Projects Engineering Management
Quality Management
Planning ManagementEngineering Management
Innovation and Digital Transformation Hub
Our presence in theworld To meet global demand for ore, our operations, projects, and offices are located in 18 countries.
United Kingdom
United States
Switzerland
Netherlands
Japan
Office or terminal
Iron Ore Solutions
Metals for Energy Transition
Legenda:
BrazilPeru
China
United Arab Emirates
Oman India
Singapore
Australia
Chile Argentina
Canada
Indonesia
Malaysia
Our presence in Brazil
Carajás PA
Ponta da Madeira Terminal MA
Tubarão Terminal ES
MG
Northern System
Southern System
Southeast System
q Current Scenario – Challenges / Objectives / Proposed Situation
q Operational Research – Example 1 e 2
q Case – Toos Used
q Simulation with AI (Video 1 e 2)
q Final Considerations
07
11
29
35
39
A G E N D A:
q Template Document – Command Prompt 33
8
Mass haul calculation/Earthwork distribution
Manual process
Introduction
Application of AI in Railway earthwork project verification
Current Scenario – Challenges
Excessive time
Low precision
Frequent rework
Limited capacity
3–4 business days for a 10 km railway section
Frequent calculation errors and inconsistencies
Multiple revisions required
to simulate alternatives
Excel/Similar (volume calculation and
masses distribution)
AutoCAD Civil 3D (data extraction)
C3D XLX IA WEB BROWSER
Solution Objectives
ü Automate the distribution calculation process
ü Significantly reduce analysis time
ü Minimize material transport costs
ü Increase result accuracy
ü Enable rapid simulation of multiple scenarios
9
Current Situation
Application of AI in Railway earthwork project verification
EXCAVATION
TRANSPORT
COMPACTION
A
REM/SUB CI R SLa
C
CV C RB REP SLa
ESC COMP
Compaction operations
Earthwork operations
Natural field level
Natural field level
CP-B-501 (VALE)
10
q Volume Calculation Refinement
q Volume Calculation Summary
q Bruckner Ordinate
Current Situation
Application of AI in Railway earthwork project verification
Bruckner Diagram (1860s)
A1
C1
A2
C2
A3
C3
q Mass Distribution
q Mass Distribution Summary
q Distribution Drawings
q Typical Sections
q Templated Cross Sections
q Volume Calculation - Does not automate distributions
- Inflexible to changes
- Inadequate for complex geometries - Ignores alternative routes
- Oversimplification (distance only) - Ignores technical material constraints
- Limited simulation capacity
Limitations:
2,5 – 3,5km/day
11
Proposed Situation
Application of AI in Railway earthwork project verification
SPREADSHEET WEB BROWSER
AI
OPERATIONAL RESEARCH
Data parameterization
§ Finds lowest-cost paths
§ Considers weights (distances) between nodes § Considers technical constraints (CBR, expansion)
§ Fully automates material movement optimization
Using algorithms in AI prompts:
Command prompt
Data analysis
Contextualization
Initial rules
Volume calculation (CV)
Results CV
Results segmentation
Mass distribution
OPERATIONAL RESEARCH EXAMPLE 1
C1
C2
C3
C4
A1
A3
A2
CUTS (C) LANDFILLS (A)
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
STEP 1
STEP 2
STEP 3
STEP 4
STEP 5
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
EXAMPLE 1 – DIJKSTRA ALGORITHM
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2
CUTS (C) LANDFILLS (A)
R$3/m³
STEP 1
STEP 2
STEP 3
STEP 4
STEP 5
EXAMPLE 1 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2
CUTS (C) LANDFILLS (A)
R$3/m³
STEP 1
STEP 2
STEP 3
STEP 4
STEP 5
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
EXAMPLE 1 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2
CUTS (C) LANDFILLS (A)
R$3/m³
STEP 1
STEP 2
STEP 3
STEP 4
STEP 5
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
EXAMPLE 1 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2
CUTS (C) LANDFILLS (A)
R$3/m³
STEP 1
STEP 2
STEP 3
STEP 4
STEP 5
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
EXAMPLE 1 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2
CUTS (C) LANDFILLS (A)
R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
Custo Total: R$ 97.000
EXAMPLE 1 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
► Earthwork Volumes
► Transportation Costs
► Distribution Results
STEP 1 STEP 2 STEP 3
STEP 4 STEP 5
EXAMPLE 1 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
OPERATIONAL RESEARCH EXAMPLE 2 – WITH COMPATIBILITY CONSTRAINTS
C1
C2
C3
C4
A1
A3
A2
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
QUALITY OF MATERIALSSTEP 4
COMPATIBILITY RESTRICTIONSSTEP 5
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
STEP 4
STEP 5
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
QUALITY OF MATERIALS
COMPATIBILITY RESTRICTIONS
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
STEP 4
STEP 5
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
QUALITY OF MATERIALS
COMPATIBILITY RESTRICTIONS
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
STEP 4
STEP 5
CBR:20% EXP: 0,3%
CBR:12% EXP: 1,2%
CBR:6% EXP: 1,5%
CBR:10% EXP: 2,0%
CBR≥10% EXP≤ 1,0%
CBR≥8,0% EXP≤ 1,5%
CBR≥2,0% EXP≤ 2,0%
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
QUALITY OF MATERIALS
COMPATIBILITY RESTRICTIONS
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
STEP 4
STEP 5
CBR:20% EXP: 0,3%
CBR:12% EXP: 1,2%
CBR:6% EXP: 1,5%
CBR:10% EXP: 2,0%
CBR≥10% EXP≤ 1,0%
CBR≥8,0% EXP≤ 1,5%
CBR≥2,0% EXP≤ 2,0%
Exp. alta
CBR baixo EXP alta
CBR baixo
Exp. alta
Exp. alta
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
QUALITY OF MATERIALS
COMPATIBILITY RESTRICTIONS
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
STEP 4
STEP 5
CBR:20% EXP: 0,3%
CBR:12% EXP: 1,2%
CBR:6% EXP: 1,5%
CBR:10% EXP: 2,0%
CBR≥10% EXP≤ 1,0%
CBR≥8,0% EXP≤ 1,5%
CBR≥2,0% EXP≤ 2,0%
Exp. alta
CBR baixo EXP alta
CBR baixo
Exp. alta
Exp. alta
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
QUALITY OF MATERIALS
COMPATIBILITY RESTRICTIONS
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
STEP 1
STEP 2
STEP 3
STEP 6
STEP 7
STEP 4
STEP 5
CBR:20% EXP: 0,3%
CBR:12% EXP: 1,2%
CBR:6% EXP: 1,5%
CBR:10% EXP: 2,0%
CBR≥10% EXP≤ 1,0%
CBR≥8,0% EXP≤ 1,5%
CBR≥2,0% EXP≤ 2,0%
Exp. alta
CBR baixo EXP alta
CBR baixo
Exp. alta
Exp. alta
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
QUALITY OF MATERIALS
COMPATIBILITY RESTRICTIONS
DEFINE THE PROBLEM
IDENTIFY COSTS
QUANTIFY VOLUMES
APPLY ALGORITHM
OPTIMAL DISTRIBUITION
R$5/m³
R$7/m³
R$4/m³
R$6/m³
R$3/m³ R$8/m³
R$9/m³
R$5/m³
R$8/m³
R$6/m³
R$4/m³
C1
C2
C3
C4
A1
A3
A2R$3/m³
8.000m³
5.000m³
6.000m³
4.000m³
7.000m³
7.000m³
9.000m³
CBR:20% EXP: 0,3%
CBR:12% EXP: 1,2%
CBR:6% EXP: 1,5%
CBR:10% EXP: 2,0%
CBR≥10% EXP≤ 1,0%
CBR≥8,0% EXP≤ 1,5%
CBR≥2,0% EXP≤ 2,0%
Exp. alta
CBR baixo EXP alta
CBR baixo
Exp. alta
Exp. alta
► Earthwork Volumes
► Transportation Costs
► Distribution Results
EXAMPLE 2 – DIJKSTRA ALGORITHM
Total Cost: R$ 102.000
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CUTS (C) LANDFILLS (A)
STEP 1 STEP 2 STEP 3 STEP 4
STEP 5 STEP 6 STEP 7
EXAMPLE 2 – DIJKSTRA ALGORITHM
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
CASE TOOLS USED
Case – Tools Used
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
LEGENDA
Federal and State Highway
Feeder road/servisse road to the site
EFC – Segment LOC 45-46C
Various local roads
Railway and road-rail bridge
. Over the Jacundá (RJ) river
. Over the Mãe Maria (RMM) river
. Over the Flecheiras (RF) river
. Over the Tocantins (PRF TOC) river
EFC – Segment PÁTIO 46
EFC – Segment LOC 46/47C
Highway connecting Yard 48 (VALE)/BR-150
Highways and acess roads
Railway
Special Structures
LOC 45/46 C
6,28km
LOC 46/47 C
9,03Km13,01Km
PÁTIO 46
3,66Km
ACESSO CONTIGUO A FERROVIA
[PF RMM]
[PF RF] [PF RJ]
PROTECTED AREA
^ N
712+790 703+760 700+100 693+820
Rio Tocantins
16,4km14,0Km
Rio Tocantins
BR 222
BR 150
21,0km
5,5km
Jazida Lourão
9,6km
7,2km
[PRF TOC]
Marabá
BR 230
BR 150
PÁTIO 48
4,6Km
17,8km
São Luís-MA à
ß Carajás-P
A
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
BI M
S EG
U IM
EN TO
4 5/
46 • Encontro 2 - Rio Jacundá:
BI M
P ÁT
IO 4
6
• Caixa de empréstimo C3 – Estr. de Manutenção: • Cx. de empréstimo C5 – Estr. de Manutenção: • Cx. de empréstimo C5 – Estr. de Manutenção:
BI M
S EG
U IM
EN TO
4 6/
47
• Encontro 1 - Rio Mãe Maria:• Ponte Ferroviária sobre o Rio Mãe Maria: • Encontro 2 – Rio Mãe Maria:
• Ponte Ferroviária sobre o Rio Jacundá: • Encontro 1 - Rio Jacundá:
Case – Tools Used
OTIMIZAÇÃO DA DISTRIBUIÇÃO DE TERRAPLENAGEM COM INTELIGÊNCIA ARTIFICIAL
§ Railway heavy haul
§ 40t/axle
§ Gauge 1,60m
§ 15km of double-track railway BIM Model
TEMPLATE DOCUMENT COMMAND PROMPT
Template Document – Command Prompt
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
VIDEO AI Simulation
Video - AI Simulation
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
VIDEO – APP VISUALIZATION ENVIRONMENT
Video APP – Visualization Environment
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
FINAL CONSIDERATIONS
Final Considerations
APPLICATION OF AI IN RAILWAY EARTHWORK PROJECT VERIFICATION
75% reduction in analysis time
More consistent calculations
Average 5% reduction in transport effort
Quick adaptation to changes
Consistent reports
Time
Accuracy
Optmization
Flexibility
Standardzation
Thank you!