Inaccurate Forecasting: The existing approach failed to account for distinct demand patterns across products (e.g., seasonality vs. volatility), leading to inventory misalignment and inability to project future requirements
. Workforce Instability: The lack of a cohesive aggregate plan meant the company was over-relying on hiring and firing to meet fluctuating demand. Specifically, firing costs were $28,000 per worker, creating massive administrative waste
. Capacity Constraints: Downstream stations were being flooded without flow control. Station 7 was identified as a "primary bottleneck" operating at nearly 99% utilization, causing severe congestion and increasing Cycle Time
.
3. Methodology & Technical Approach We utilized a multi-stage Industrial Engineering approach to rebuild the production logic:
Advanced Forecasting: Applied tailored statistical models (Holt’s Method, Multiplicative Seasonal, Linear Regression) to five distinct products based on historical data. Additionally, I implemented Support Vector Regression (SVR) using Python to test machine learning capabilities against traditional methods
. Aggregate Planning Analysis: Conducted a cost-benefit analysis comparing "Chase" (matching production to demand) vs. "Level" (constant production) strategies over a 5-week horizon to determine the optimal workforce plan
. MRP & Capacity Planning: Exploded the Bill of Materials (BOM) into a time-phased Material Requirements Plan (MRP) and calculated machine loads to identify necessary capital investments
. Discrete-Event Simulation: Built a FlexSim 2025 model to simulate the Week 21 schedule, applying Shortest Processing Time (SPT) dispatching to minimize queues and validate feasibility
.
4. Key Findings The data analysis revealed that the primary cost driver was not production materials, but workforce volatility:
Strategy Superiority: The "Level Strategy" was vastly superior to the "Chase Strategy." Maintaining a stable workforce avoided $182,123 in hiring/firing costs, despite incurring minor inventory holding costs
. Forecasting Accuracy: Traditional methods (Exponential Smoothing with Trend) proved highly accurate for short-term planning, achieving Mean Absolute Deviation (MAD) scores well below the tolerance of 8 for all products
. The Hidden Bottleneck: While upstream stations had slack (Stations 1-3), Station 7 was the critical constraint. The simulation showed it operated at 98.7% utilization, dictating the throughput of the entire system
.
5. The Solution (Recommendations) We delivered a three-tiered improvement strategy:
A. Strategic Workforce Stabilization
Concept: Eliminate the "hire-fire" cycle.
Implementation: Adopt the Level Production Strategy immediately. Maintain a constant weekly production rate (e.g., 146 units/week for Product 1) to stabilize labor costs and retain skilled workers
.
B. Capacity Expansion
Concept: Targeted capital investment to break bottlenecks.
Implementation: Purchase additional machines specifically for Work Centers 1, 4, and 6 to handle peak loads in Week 23, ensuring capacity exceeds the 2,400-minute weekly limit
.
C. Dynamic Scheduling (SPT Rule)
Concept: Minimize flow time for high-volume orders.
Implementation: Implement Shortest Processing Time (SPT) dispatching at all stations. The simulation confirmed this rule successfully cleared all 1,841 units within the work week without requiring overtime
.
6. Impact
$182,123 Cost Reduction: By shifting to a Level Strategy, the division can eliminate excessive turnover costs and restore profitability
. 100% Order Fulfillment: The FlexSim simulation validated that the new schedule allows all Week 21 orders to be completed in 2,312 minutes (38.5 hours), meeting the deadline with time to spare
. Forecast Reliability: All five products achieved forecast accuracy with MAD < 8, ensuring that the Master Production Schedule is built on reliable demand data
.