Demand Forecasting, Inventory, and Lean Management
Demand Forecasting
Demand Forecasting Defined
A demand forecast estimates future demand over a specified time. Forecasts are crucial inputs for various organizational decisions. In operations management, demand forecasting is vital for:
- System design
- Medium-term system planning
- Short-term system scheduling
Characteristics of a Good Forecast
Effective forecasts are:
- Timely
- Accurate
- Reliable
- Meaningful
- Easy to understand
- Cost-effective
Forecasting Process
- Define the forecast’s purpose.
- Establish a forecasting horizon.
- Gather and analyze relevant historical data.
- Choose a forecasting technique.
- Prepare the forecast.
- Monitor the forecast.
Forecasting Approaches
Forecasting approaches include judgmental methods (like executive opinions) and quantitative methods.
Time Series Analysis
A time series is a chronologically ordered sequence of observations taken at regular intervals. Key time series components include:
- Level: The average or horizontal pattern.
- Trend: A consistent upward or downward movement.
- Seasonality: Regular, repeating variations often linked to calendar, weather, or events.
- Cycles: Wavelike patterns lasting over a year, often tied to economic or political factors.
- Irregular Variations: Data points caused by unusual, one-time explainable events (e.g., strikes).
- Random Variations: Unpredictable variations from numerous factors.
Naïve Forecasting
The naïve forecast is simple and inexpensive but often less accurate. For stable series, the last data point predicts the next period. With seasonal variations, the last season’s value predicts the next. For data with a trend, the naïve forecast adds or subtracts the difference between the last two data points to the last value.
Moving Average Methods
The moving average method averages recent actual data values to create a forecast. A weighted moving average assigns different weights to different periods.
Exponential Smoothing
Exponential smoothing is a sophisticated weighted averaging method where each period’s forecast is based on the previous period’s forecast plus a percentage of the forecast error.
Inventory Management
Inventory Defined
Inventory refers to goods held for future use or sale. Inventory management involves planning and controlling these stocks.
Effective Inventory Management
Effective inventory management requires:
- A system for tracking inventory on hand and on order.
- Reliable demand forecasts with error indications.
- Knowledge of lead times and their variability.
- Estimates of holding, ordering, and shortage costs.
- An inventory item classification system.
Inventory Systems
A periodic inventory system (fixed-time period) involves regular physical counts to determine order quantities. A perpetual inventory system (continuous review, fixed-order quantity) continuously tracks inventory levels.
Inventory Tracking Tools
Tracking tools include barcodes and Radio Frequency Identification (RFID) technology for identifying objects in supply chains. Other tools include Point of Sale (POS) systems for electronically recording sales.
Inventory Metrics
Poor inventory management leads to stockouts and overstocking. The fill rate measures fulfilled orders from available inventory. Unit fill rate is units delivered divided by units ordered. Order fill rate is complete orders divided by total orders. Inventory turnover measures inventory efficiency (unit sales/average inventory or COGS/inventory value).
Lean Management
Just-in-Time (JIT) Production
Just-in-time (JIT) production processes and moves parts only as needed, usually in small batches. Key principles include:
- Identifying customer value
- Focusing on value-creating processes
- Eliminating waste to create flow
- Producing according to customer demand
- Striving for perfection
Lean Principles and Tools
Muda represents waste and inefficiency. Kanban is a manual system controlling parts and materials movement based on need signals. Heijunka addresses production volume variations. Kaizen promotes continuous improvement. Jidoka emphasizes quality at the source.
Lean Product Design and Production
Lean product design features standard parts, modular design, and highly capable production systems. Concurrent engineering and Single-Minute Exchange of Die (SMED) reduce changeover time. A cell is a specialized production center. Statistical Process Control (SPC) and Six Sigma reduce variability. Poka-yoke mechanisms prevent errors. Automation, particularly intelligent automation, stops production if abnormalities occur.
Jidoka and Andon
Jidoka (quality at the source) prevents defective products from moving to the next workstation. Andon systems use lights to signal problems or slowdowns.
Takt Time and Planning
Takt time is calculated by dividing net available time by daily demand. Planning and control methods include level loading (heijunka), pull systems (with kanban), close supplier relationships, preventive maintenance, and 5S housekeeping. A pull system moves work as needed, while a push system pushes completed output to the next station. The Kanban formula is: N = DT(1+x)/c
Example: Exponential Smoothing
A boutique owner forecasts October demand using exponential smoothing (α = 0.2) and August’s forecast of 145. Given September sales of 160, the October forecast is: 0.2 * 160 + (1 – 0.2) * 140 = 144
” alt=”Image 1″>
” alt=”Image 2″>