Optimized AppliChem’s global supply chain strategy using Mixed-Integer Linear Programming (MILP) in Python with Gurobi. Modeled 6-year production plans across 6 plants, accounting for costs, tariffs, exchange rates, and CO₂ emissions. Delivered insights on plant selection, cost reduction, and sustainable long-term operations.
📘 Course: IE 7200 – Supply Chain Engineering
👨🏫 Professor: Paul Pei
👩💻 Team Members: Sakshi Agarwal · Shivie Saksenaa · Neha Patil
Applichem, a global chemical manufacturer, faces rising costs and environmental pressures in its plant network.
We built a Mixed-Integer Linear Programming (MILP) model in Python (Gurobi solver) to optimize global production of “Release-ease” over a 6-year horizon (1982–1987).
Key features of our model:
- ✅ Considers inflation, exchange rates, tariffs, transport costs
- ✅ Accounts for startup/shutdown penalties
- ✅ Integrates CO₂ emission constraints
- ✅ Provides strategic plant selection & production allocation
We engineered data from case study files, including:
- Local production costs (per lb, 1982 baseline)
- Country-specific inflation & FX adjustments
- Tariff schedules by plant/year
- Transportation cost per lb
- Plant capacity limits (M lbs)
- Global demand forecasts (1982–1987)
- Plant-specific CO₂ emission factors
Assumptions:
- Costs converted to USD
- Startup penalties discourage short-term opportunism
- CO₂ costs applied as sustainability incentives
[ \text{Minimize: } \sum (C_{pt} \cdot x_{pt} + \lambda \cdot CO2_p \cdot x_{pt} + \alpha \cdot y_{pt}) ]
[
- \sum \beta \cdot startup_p ]
Constraints:
- Meet annual global demand
- Respect plant capacity
- At least 2 plants active per year
- At least 1 plant active in the Americas each year
- Startup triggered if plant becomes active
- Python 3.10
- Gurobi 12.0.1 (MILP solver)
- Jupyter Notebook (experiments & reporting)
- Plotly + Pandas (visualizations)
- Custom dashboard for scenario exploration
Baseline (No Startup Cost):
- Total Cost: $780.88M
- Plants: Frankfurt, Gary, Mexico active; Canada used only in 1987
- Excluded: Venezuela & Sunchem (too costly & high emissions)
With Startup Cost:
- Total Cost: $792.88M
- Same plants as baseline
- Greater stability & realistic long-term strategy
- Frankfurt = Core hub (low cost, low emissions)
- Gary (USA) = Strong supplementary site (low transport, tariff-free)
- Mexico = Growing competitiveness (FX + tariff benefits)
- Canada = Backup for peak demand
- Venezuela & Sunchem = Inefficient → excluded
Conclusion: Adding startup costs slightly increases total cost but leads to sustainable, stable operations.
- Add global CO₂ emission caps
- Model plant expansion or new facilities
- Extend to warehousing & multi-echelon transport
Extra data needed:
- Freight rates & lead times
- Plant downtime schedules
- Tax & incentive policies
Our optimization framework demonstrates how MILP + real-world economics can:
- Balance cost, environment, and policy
- Provide data-driven, actionable insights
- Support long-term strategic planning in global supply chains