Port sequences, vessel classes, speed profiles and cargo flows — all designed from scratch.
Suez closure? Panama drought? Alliance under review? Network Design returns a complete, profitable network in ~60 minutes for a problem with 14M binary variables. The final word before you publish a schedule or sign an alliance.
Engine: Hexaly · Local search · LSFRP
Why this matters right now
Four disruptions are forcing every carrier to redraw its service network.
The slow way takes months. The cost of getting it wrong runs into tens of millions.
Houthi attacks (2024-present) force Cape of Good Hope rerouting. Asia-Europe transit time blows out by 10–14 days per voyage.
Vessel transit restrictions reducing throughput across one of the world's most critical shipping corridors.
Asia-Europe trade lanes severely disrupted. Network designs that worked in 2023 don't work in 2026.
Tensions across multiple corridors. Carriers need to redraw the network in days, not the months traditional planning requires.
Strategic, not operational
Short-term schedules are operational.
Network design is strategic — where corridors get re-drawn, alliances get stress-tested, and vessel classes get reassigned across an entire trade.
The stakes
$50M+
Annual cost from a single bad network decision.
A wrongly-sized vessel class on the wrong corridor compounds across every voyage, every week of the year.
What this means for your network P&L
Network re-evaluation that used to take a quarter of manual work returns a complete plan in a single working day.
Local search on list variables replaces brittle column generation. The problem fits in a single optimization run.
Bad network choices compound across thousands of voyages. A single corrected vessel-class assignment can return $50M+ a year.
Strategic network design feeds straight into operational planners (OptiFleet, OptiBox). One source of truth across both horizons.
What Network Design tackles
Port sequences, vessel classes, speed profiles and cargo flows — all designed in a single solver run, at the scale of the world's largest container alliances.
Solver architecture
Traditional Mixed-Integer Programming (Gurobi / CPLEX) on 14M binary variables takes days to converge.
Local search on list variables returns a high-quality solution in minutes — and scales near-linearly.
Variables
MIP solver (Gurobi): Binary matrix x[port][service][position]
Hexaly: list(113) per service · ordered subset
Assignment
MIP solver (Gurobi): Thousands of binary constraints
Hexaly: Built-in partition / disjoint operators
Search method
MIP solver (Gurobi): Branch-and-bound on LP relaxation
Hexaly: Intelligent local search (insert / swap / reverse)
Time to solution
MIP solver (Gurobi): Hours / Days
Hexaly: Minutes ✓
Optimality
MIP solver (Gurobi): Guaranteed (with enough time)
Hexaly: Heuristic — very good in practice
Scalability
MIP solver (Gurobi): Degrades exponentially
Hexaly: Near-linear with problem size ✓
Key insight: list variables replace column generation. route[k] ← list(113 ports) · vc[k] ← int(0, 16) · speed[k] ← int(0, 4)
How it works
Warm-start seeding, then network design, then cargo routing — with a feedback loop back to Phase 1 that re-balances oversupplied and undersupplied legs.
Converges in 3–5 iterations.
Feedback loop: Oversupplied legs dampen the signal; undersupplied legs boost it. Converges in 3–5 iterations.
The objective function
Network Design doesn't blend objectives into a single weighted score — it ranks them.
Feasibility before coverage. Coverage before structure. Structure before profit. Profit before utilization. The order is the design.
Cannot exceed vessels owned per class or 400 total. Hard constraint.
#1 commercial priority — route as much cargo through the network as possible.
Min 10 services, 8 mainline (≥10K TEU), 20 feeders (≤6K TEU), corridor coverage.
Revenue at $500 / TEU-leg minus fuel, charter, port costs and transshipment penalties.
Tiebreaker — drives toward 90% headhaul, 60% backhaul utilization targets.
What-if scenarios
Six scenario types covered out of the box.
Change one input, re-solve in 5–60 minutes, compare the new network to the baseline.
Remove PortSaid, re-solve → Cape routing across the affected services.
Reduce Panama-compatible port max TEU. Network adapts via alternative corridors.
Add port to the network with coordinates. Re-solve picks up the new hub or spoke.
Modify vessel counts per class. Network rebalances around the new fleet mix.
Scale O-D demands (e.g. +20% Asia-Europe). The plan reflects the new commercial reality.
Modify fuel price input. Vessel-class assignment and speed profiles re-optimize.
Solver budget tiers
~5 min
Quick test
1 iteration
~30 min
Standard
5 iter × 5 min
~1 hr
Production
5 iter × 10 min
Output of a single solver run
Quick-test result on a global-alliance-scale network.
Standard runs (5 iterations) deliver tighter optimality bounds in ~30 minutes. The cargo all moves; the question is just how.
82
Services deployed
85%
Of fleet utilized
0%
Cargo rejected
~30 min
Solver time
Cargo movement breakdown
Network capacity: 632,000 TEU / week · avg slot cost ~$700 / TEU
Pick your next step
Custom session with your network, alliance proposal or disruption case. Watch the re-solve happen live.
Port sequences, vessel classes, speed profiles and cargo flows — all designed from scratch. Suez closure? Panama drought? Alliance under review? Network Design returns a complete, profitable network in ~60 minutes for a problem with 14M binary variables. The final word before you publish a schedule or sign an alliance.
~60 min
Full network re-solve
14M+
Binary variables
Hours
Not months
$50M+
Per bad decision
Why this matters right now
Four disruptions are forcing every carrier to redraw its service network. The slow way takes months. The cost of getting it wrong runs into tens of millions.
+10–14 days
Houthi attacks (2024-present) force Cape of Good Hope rerouting. Asia-Europe transit time blows out by 10–14 days per voyage.
Capacity cut
Vessel transit restrictions reducing throughput across one of the world's most critical shipping corridors.
Lanes disrupted
Asia-Europe trade lanes severely disrupted. Network designs that worked in 2023 don't work in 2026.
Rapid response
Tensions across multiple corridors. Carriers need to redraw the network in days, not the months traditional planning requires.
Strategic, not operational
Short-term schedules are operational. Network design is strategic — where corridors get re-drawn, alliances get stress-tested, and vessel classes get reassigned across an entire trade.
The final word before you publish a schedule or sign an alliance.
The stakes
$50M+
Annual cost from a single bad network decision. A wrongly-sized vessel class on the wrong corridor compounds across every voyage, every week of the year.
What this means for your network P&L
Network re-evaluation that used to take a quarter of manual work returns a complete plan in a single working day.
Local search on list variables replaces brittle column generation. The problem fits in a single optimization run.
Bad network choices compound across thousands of voyages. A single corrected vessel-class assignment can return $50M+ a year.
Strategic network design feeds straight into operational planners (OptiFleet, OptiBox). One source of truth across both horizons.
What Network Design tackles
Port sequences, vessel classes, speed profiles and cargo flows — all designed in a single solver run, at the scale of the world's largest container alliances.
400+
Vessels
15+ vessel classes (1K–24K TEU)
100+
Ports
Hub + spoke topology
~600K
TEU / week
10,000+ origin-destination pairs
14
Regions
East Asia to Latin America
100+
Services
Concurrent weekly services
35–40
Trade corridors
Distinct trade lanes to cover
Solver architecture
Traditional Mixed-Integer Programming (Gurobi / CPLEX) on 14M binary variables takes days to converge. Local search on list variables returns a high-quality solution in minutes — and scales near-linearly.
Key insight: list variables replace column generation. route[k] ← list(113 ports) · vc[k] ← int(0, 16) · speed[k] ← int(0, 4)
How it works
Warm-start seeding, then network design, then cargo routing — with a feedback loop back to Phase 1 that re-balances oversupplied and undersupplied legs. Converges in 3–5 iterations.
PHASE 0
PHASE 1
PHASE 2
Feedback loop: Oversupplied legs dampen the signal; undersupplied legs boost it. Converges in 3–5 iterations.
The objective function
Network Design doesn't blend objectives into a single weighted score — it ranks them. Feasibility before coverage. Coverage before structure. Structure before profit. Profit before utilization. The order is the design.
TIER 1
MIN
Cannot exceed vessels owned per class or 400 total. Hard constraint.
TIER 2
MAX
#1 commercial priority — route as much cargo through the network as possible.
TIER 3
MIN
Min 10 services, 8 mainline (≥10K TEU), 20 feeders (≤6K TEU), corridor coverage.
TIER 4
MAX
Revenue at $500 / TEU-leg minus fuel, charter, port costs and transshipment penalties.
TIER 5
MIN
Tiebreaker — drives toward 90% headhaul, 60% backhaul utilization targets.
What-if scenarios
Six scenario types covered out of the box. Change one input, re-solve in 5–60 minutes, compare the new network to the baseline.
Remove PortSaid, re-solve → Cape routing across the affected services.
Reduce Panama-compatible port max TEU. Network adapts via alternative corridors.
Add port to the network with coordinates. Re-solve picks up the new hub or spoke.
Modify vessel counts per class. Network rebalances around the new fleet mix.
Scale O-D demands (e.g. +20% Asia-Europe). The plan reflects the new commercial reality.
Modify fuel price input. Vessel-class assignment and speed profiles re-optimize.
Solver budget tiers
~5 min
Quick test
1 iteration
~30 min
Standard
5 iter × 5 min
~1 hr
Production
5 iter × 10 min
Output of a single solver run
Quick-test result on a global-alliance-scale network. Standard runs (5 iterations) deliver tighter optimality bounds in ~30 minutes. The cargo all moves; the question is just how.
82
Services deployed
85%
Of fleet utilized
0%
Cargo rejected
~30 min
Solver time
Cargo movement breakdown
Services by tier
Network capacity: 632,000 TEU / week · avg slot cost ~$700 / TEU
What Network Design puts on the table
~60 min
Network re-solve
14M+
Binary variables
5-tier
Lexicographic objective
6
Scenario types
Pick your next step
Three ways to start — pick the one that fits where you are. Or drop us a line directly.
Just looking
Strategic overview of Network Design — what it does, what it needs, where it fits in your planning stack.
Request brief →Evaluating
Custom session with your network, alliance proposal or disruption case. Watch the re-solve happen live.
Schedule session →Ready to deploy
45 minutes with a Network Design architect. Integration plan, timeline, and strategic-ROI projection.
Book a call →