3. Industrial Requirements – Operating At A Large Scale
• Privatization of container rail operations in India, has created an overall freight
market (size of ~3bn tonnes) by trying to shift volumes from road to rail
• With 500 rakes expected to be operational in FY12/13, around 16 players are
offering integrated, value-added logistics solutions with last mile connectivity. This
includes few key players like Concor, Arshiya, Gateway Distriparks, etc.
Source of above: IDFC SSKI Sector Report, December 2009
• Given the capital investment in this segment is high, logistics players need to
operate with minimum resources yielding maximum capacity utilization
• Routes will overlap with each other, and under dynamic loading situation at each
nodes, forms a classic “Time and Capacity Constrained Routing Problem”
• Minimizing the empty run will be a tough nut to crack
• Practical conditions such as – customized containerization, maintenance cycle adds
to the planning difficulties
• Traditionally, railroad planning is being done using more than one methodologies
– MILP, Heuristics, NP-Hard. This requires solid conceptual understanding & real-
time tested algorithm
Private & Confidential 3
4. Rail Planning Optimization Solution Framework
INPUT STATIC DATA INPUT DYNAMIC DATA
NODE ROUTE • Monthly forecasted loading data at each node
Location Route Id • Material freight rate on each route
Type (ICD/Port etc.) Source • Starting Rake & container availability data at
Max Rakes per Month Destination
each node
NODE ACTIVITIES Intermediate Nodes
• Capacity change for the next month
Activity (dwelling, Distance • Previous month transaction to update certain
customs etc.) time/cost parameters
Travel Time
Cost
Max Load Allowed
Time + Business Rules
Double Stacking Allowed
RAKE (such as maintenance
Rake Id
Max Rake Length run of 6,000 km) Monthly Route-Rake
Type ROUTE FREIGHT allocation
Length Route Id Number of rakes
Capacity (MT) Material required at each node
No of Rakes Freight Amount Transportation volume,
CONTAINER From Date cost & time estimates
Container Id End Date
Optimization Capacity utilization
Type Engine
Capacity deficit
Material Specific
Capacity (MT)
No of Container
Private & Confidential 4
5. Understanding Flow Network
•In this example, nodes (could be port,
IDC locations etc.) are shown
3
x32 •Each node has interconnections with the
other nodes
x23
2 •All Nodes have demand-supply estimates
x13
x24 x42
x31
x21 x12 •Once demand and supply or flow in/out is
x45 achieved for all nodes, we try to calculate
5 4 the number of movements between each
x51 x54 nodes, in the given set A of origin-
x15 destination (O-D) pairs
1
• In the equation below, no of movements
Figure: Network flow illustration
is represented by m and x is the flow
volume between i-j (O-D). This basically
becomes part of the optimization
objective function
Private & Confidential 5