Linear Programming in GAMS for Production, Sourcing, Distribution, Tariff & Sustainability Decisions
A supply chain optimization model for a global electric scooter network that determines where to produce, which suppliers to source from, how much to ship, and how to respond to tariffs, disruptions, service-level requirements, and CO₂ constraints.
The model is built in GAMS using Linear Programming and evaluates a 12-month supply chain connecting manufacturing plants in India and China with East and West U.S. markets.
Global manufacturers must balance several competing priorities:
- Production and transportation cost
- Supplier availability
- Plant capacity
- Import tariffs
- Customer service levels
- Inventory and shortages
- Environmental constraints
- Supply chain disruptions
This project models these trade-offs for NovaRide, a hypothetical electric scooter manufacturer, and identifies the lowest-cost feasible production and distribution strategy.
flowchart LR
S1[Supplier 1] --> IND[India Plant]
S2[Supplier 2] --> IND
S1 --> CHN[China Plant]
S2 --> CHN
IND --> EAST[East U.S. Market]
IND --> WEST[West U.S. Market]
CHN --> EAST
CHN --> WEST
C[Battery • Motor • PCB • Chassis] -. Component Requirements .-> IND
C -. Component Requirements .-> CHN
| Dimension | Model Scope |
|---|---|
| Manufacturing Plants | India, China |
| U.S. Markets | East, West |
| Suppliers | 2 |
| Components | Battery, Motor, PCB, Chassis |
| Planning Horizon | 12 months |
| Optimization Method | Linear Programming |
| Modeling Platform | GAMS |
The optimization model determines:
Production
- How many scooters should each plant manufacture each month?
Supplier Allocation
- Which supplier should provide each component to each plant?
Distribution
- How many finished scooters should each plant ship to each U.S. market?
Inventory
- How much finished-goods inventory should be carried between periods?
Shortages
- When demand cannot be fully satisfied, how much unmet demand should be allowed?
The model evaluates these decisions simultaneously rather than optimizing each part of the supply chain separately.
The objective is to minimize total supply chain cost while satisfying operational and policy constraints.
Production Cost
+
Component Procurement Cost
+
Transportation Cost
+
Inventory Holding Cost
+
Shortage Penalties
+
Component Tariffs
+
Finished-Goods Tariffs
+
CO₂ Emission Cost
The model ensures that:
- Market demand is satisfied subject to allowed shortages
- Monthly production stays within plant capacity
- Required components are available for every unit produced
- Supplier purchases stay within supplier capacity
- Inventory follows material-balance requirements
- Total CO₂ emissions remain within the specified environmental limit
- All production, shipment, inventory, sourcing, and shortage decisions remain non-negative
The optimized base-case network produced the following outcome:
| KPI | Optimized Result |
|---|---|
| Total Annual Cost | $7.07M |
| India Production | 4,800 units |
| China Production | 2,280 units |
| East Market Shipments | 4,000 units |
| West Market Shipments | 3,080 units |
| Shortages | 0 units |
| Service Level | 100% |
| CO₂ Emissions | ≈ 1,415 tons |
| Ending Inventory Strategy | Make-to-order / no inventory |
India becomes the primary cost-efficient production source, operating at approximately 400 units per month and serving the full East market.
China provides flexible production capacity, particularly for demand in the West market.
The optimized network also follows a make-to-order strategy, avoiding unnecessary inventory and holding costs.
The project tests how the optimal supply chain changes under different business and policy conditions.
Optimizes the supply chain under normal production capacity, supplier availability, tariffs, demand, and emission constraints.
Result: 100% demand fulfillment with no shortages and a total annual cost of approximately $7.07M.
Battery tariffs were increased to test the network's sensitivity to component trade policy.
| Metric | Base Case | Battery Tariff Scenario |
|---|---|---|
| Total Cost | $7.075M | $7.123M |
| Cost Change | — | +0.7% |
| Production Pattern | Baseline | Unchanged |
| Shortages | 0 | 0 |
The tariff increase raises cost but does not change the optimal production or routing structure, indicating that the network is operationally stable under this uniform tariff increase.
PCB tariffs were increased by 20%.
| Metric | Base Case | PCB Tariff Scenario |
|---|---|---|
| Total Cost | $7.075M | $7.154M |
| Cost Change | — | +1.1% |
| Production Pattern | Baseline | Unchanged |
| Shortages | 0 | 0 |
The network is more financially sensitive to PCB tariffs than battery tariffs, although the optimal production and shipment pattern remains unchanged.
Different combinations of India and China finished-goods tariffs were tested.
The analysis evaluates how changes in trade policy influence:
- Plant competitiveness
- Production allocation
- Routing decisions
- Total supply chain cost
As tariffs rise, the relative cost advantage between India and China changes, causing the optimizer to reconsider where production should occur.
A temporary component supply disruption was introduced to evaluate network resilience.
The disruption resulted in:
- Significant production reduction during affected months
- Use of previously accumulated inventory
- Temporary shortages in the East market
- Production recovery after component availability returned
255 units of unmet demand occurred during the disruption scenario, demonstrating the importance of supplier diversification and inventory buffers.
The model compares 95% and 99% service-level requirements.
| Metric | 95% Service | 99% Service |
|---|---|---|
| Total Cost | $6.717M | $7.003M |
| India Production | 4,800 | 4,800 |
| China Production | 1,831 | 2,110 |
| West Shipments | 2,606 | 2,885 |
Increasing service reliability requires additional production and shipment capacity.
Higher service levels → fewer shortages → higher operating cost and emissions.
This illustrates a core supply chain trade-off between cost efficiency and customer-service reliability.
India provides the primary low-cost production capacity in the optimized network.
China provides additional capacity when demand exceeds India's available production capability, especially for the West market.
Component disruptions can create shortages even when the overall production network has sufficient manufacturing capacity.
Moderate uniform component tariff increases raise total cost without immediately changing production or sourcing decisions.
Moving from 95% to 99% fulfillment improves reliability but requires additional production and operating cost.
CO₂ emissions are incorporated directly into the decision model through emission limits and carbon-related costs.
GAMS
- Linear programming model development
- Constraint formulation
- Optimization
- Scenario experimentation
Excel
- Scenario tracking
- Result comparison
- Sensitivity analysis
- Visualization
Operations Research
- Network optimization
- Capacity planning
- Supplier allocation
- Production planning
- Cost minimization
| Variable | Decision |
|---|---|
q[p,t] |
Production quantity by plant and month |
x[p,m,t] |
Shipment quantity from plant to market |
y[c,s,p,t] |
Component purchases by supplier and plant |
I[p,t] |
Ending inventory |
B[m,t] |
Unmet demand / backlog |
Demand Satisfaction
↓
Inventory Balance
↓
Plant Capacity
↓
Component Requirements
↓
Supplier Availability
↓
CO₂ Emission Limit
↓
Non-Negativity
For a stronger portfolio presentation, the repository can include an /assets folder with these visuals:
Show:
Suppliers → India / China → East / West U.S.
This gives recruiters an immediate understanding of the optimization network.
Recommended chart:
Base Case $7.075M
Battery Tariff +10% $7.123M
PCB Tariff +20% $7.154M
Supplier Disruption $8.440M
Suggested file:
assets/scenario_cost_comparison.png
Plot monthly India and China production across the 12-month horizon.
This clearly shows the production reduction during the disruption and subsequent recovery.
Suggested file:
assets/disruption_production.png
Show total supply chain cost across different combinations of:
China Finished-Goods Tariff
×
India Finished-Goods Tariff
This is one of the strongest visuals for demonstrating sensitivity analysis and policy-driven decision making.
Suggested file:
assets/tariff_heatmap.png
Compare:
95% Service Level
vs.
99% Service Level
using:
- Total cost
- Shortages
- China production
- Emissions
This makes the cost-versus-reliability trade-off immediately visible.
Multi-Plant-Supply-Chain-Optimization/
│
├── FINALPROJECTCODE.gms
│ └── Main GAMS optimization model
│
├── FINALPROJECTCODE.gms.log
│ └── GAMS execution log
│
├── FINALPROJECTCODE.lst
│ └── Solver output and model results
│
├── FINALPROJECTCODE.lxi
│ └── GAMS solution-related file
│
├── EXPERIMENTATION.xlsx
│ └── Scenario experiments and result tracking
│
├── Multi-Plant Production and Supply Chain Optimization.pdf
│ └── Complete project report and methodology
│
├── assets/
│ ├── supply_chain_network.png
│ ├── scenario_cost_comparison.png
│ ├── disruption_production.png
│ ├── tariff_heatmap.png
│ └── service_level_tradeoff.png
│
└── README.md
Open:
FINALPROJECTCODE.gms
in GAMS.
Execute the model using the LP solver configured in GAMS.
Check:
FINALPROJECTCODE.lst
FINALPROJECTCODE.gms.log
Open:
EXPERIMENTATION.xlsx
to compare scenario results and sensitivity experiments.
Course: ISE 501 – Introduction to Operations Research Project: Multi-Plant Production and Supply Chain Optimization Under Tariff and Environmental Policies
This project demonstrates how linear programming and operations research can support real-world supply chain decisions involving production, sourcing, logistics, trade policy, resilience, service levels, and sustainability.
Future versions of the model could explore:
- Multi-objective optimization for cost vs. emissions
- Demand uncertainty and stochastic optimization
- Supplier risk scoring
- Additional plants and markets
- Safety-stock optimization
- Lead-time uncertainty
- Transportation capacity constraints
- Dynamic tariff scenarios
- Supplier diversification strategies
- Interactive scenario dashboards
This project demonstrates how mathematical optimization can transform a complex supply chain problem into actionable decisions by answering:
Where should we produce, where should we source, how should we distribute, and how should the network respond when business conditions change?
The model connects operations research with real-world supply chain strategy, allowing production, sourcing, cost, resilience, customer service, and sustainability decisions to be evaluated within a single optimization framework.