Professional Documents
Culture Documents
Forecasting
• Actual demand when available
• Forecasts
– When lead time is insufficient
– On push side of push-pull frontier
• Logistics rarely forecasts except input on
constraints
• Term (shorter terms are more reliable)
– Long-term (3+ years)
– Medium-term (1 year or so)
– Short-term (less than 1 year)
Nature of Forecasting
• Future is uncertain
• “Forecasts are always wrong”
(track accuracy data)
• Bias
• Benefit-cost
• Aggregate more reliable than item forecast
• Check data sources and simplify models
• Use demand history instead of sales history
• Consumer demand not intermediate customer
Bullwhip Effect
• 1960’s “beer game”: Lack of communication root
cause
• Distribution center order treated as demand
(safety stock, overstock)
Supplier Factory
Distributor Retailer
Customer
Small demand uncertainty becomes more and
more distorted.
Factors Affecting Demand
Trend line (linear)
Seasonal
peaks
Actual demand line
Random variation
Step 2:
Step 7:
Determine aggregation and Step 10:
Test against historical data
units Periodically review and
improve models for accuracy
Step 3: Step 6:
Determine time horizon Prepare data
Step 4: Step 5:
Visualize data Choose method or model
Forecasting Methods
Qualitative Quantitative
Data for
next two
slides
Deseasonalized values
closer to year average
Simple and Weighted Moving Average
Smooths out random spikes or dips:
(M1 + M2 + M3 )
3-Month Moving Average =
3
(15 + 12.96 + 14 )
January 2016 forecast = = 13.99 units
3
(1 M1) + ( 2 M2 ) + ( 3 M3 )
3-Month Weighted Moving Average =
6
(1 15 ) + ( 2 12.96 ) + ( 3 14 )
January 2016 forecast = = 13.82 units
6
Sum of weights
Exponential Smoothing and Re-seasonalizing
• Smoothing constant (alpha, α): 0–1
(percentage)
• Reseasonalize
– 15.54 units × 2.214 (January 2015 seasonal index)
= 34.4 units, the forecast for January 2016
Quantitative: Associative (Causal)
Simple Regression
• Independent (predictor) and dependent variable (predicted)
y = + x
Roofing Sales
USD 1,600,000
USD 1,400,000
USD 1,200,000
USD 1,000,000
USD 800,000
β (slope, here USD per housing start)
USD 600,000
USD 400,000
α (intercept)
USD 200,000
USD 0
0 20 40 60 80 100 120 140 160
Housing Starts
Coefficient of Correlation (r)
• A statistical measure of the degree to which
changes to the value of one variable predict
change to the value of another
• Range of values between -1.0 and +1.0
r = 1.0 is perfect positive correlation
MSE =
Number of Forecast Periods
Average of errors squared then summed
• Mean absolute percentage error (MAPE):
A −F
Average of summed forecast error A %
MAPE =
percentages n
Accuracy (Error Rates)
• Tracking signal:
Used to indicate the existence of any
positive or negative bias in a forecast
– Algebraic sum of forecast errors, divided by
the mean absolute deviation value
• Exceptions:
Outliers, could be errors or not
Manufacturing/Distribution Balancing
$
External Balancing Internal Balancing
• Price • Production flexibility
• Lead time • Inventory holding
• Both promote substitutes – Safety stock helps to prevent
(yours or others) stockouts, but makes
oversupply issues worse
• Marketing methods
– Sales incentives
– Trade discounts
– Consumer promotions
Logistics Balancing
$
External Balancing Internal Balancing
• Switch from manual to EDI • Reconfigure transportation
ordering routes or modes
• Offer substitutes located • Adjust stock levels at
closer distribution centers
• Offer free shipping • Drop ship from alternative
• Leaving trailers at distribution sites or channels
customers to minimize back • Tracking devices for more
haul detailed and accurate
tracking
Distribution Channels
Production
Retail
Business-to-
Business-to- Consumer (B2C)
Business (B2B)
Ecommerce
Supplier to
Customer
Operations Planning and Control
Strategic Planning
Production Rate =
(Ending Inventory - Beginning Inventory ) + Forecast
Number of Periods
=
(12,000-10,000 ) + 100,000
= 8,500 units per month
12
• Resource planning
• Inventory planning
• Distribution requirements planning (DRP)
• Performance metrics and targets
Resource Planning
Resource Profile Units Capacity (Monthly)
Condenser Shop Hours 35,000
Final Assembly m3 8,000
Resource Units Family Family Family Total Load Capacity Load vs.
Plan A B C Capacity
Jan. Plan Units 5,000 2,000 1,500 8,500 n/a n/a
Condenser
Hours 15,000 10,000 7,500 32,500 35,000 92.9%
Shop
Final
m3 3,000 2,400 2,100 7,500 8,000 93.8%
Assembly
Source: Adapted from David F. Ross, Distribution Planning and Control—Managing in the Era of Supply Chain Management, Third Edition.
Fulfillment Center Capacity Bar Chart
Fulfillment Center Capacity
10000 Exceeds Capacity
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Released
ReleasedLoad
Load Planned
Planned Load
Load
Demand Management Process
Demand Plan Dashboard----Units
Demand for Product Family XYZ in Units
15,000
10,000
5,000
0
Units
–3 –2 –1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Historical Demand Demand Plan for 18 Months in Future
Supply/Demand Supply/Demand
Match Mismatch
Period 1 2 3 4 5 6 7 8 9 10
Forecast 20 22 21 25 24 23 21 21 25 25
Customer orders 19 17 15 11 9 5 2 1 0 0
PAB 50 31 14 49 24 0 27 6 35 10 35
ATP 14 15 43 49
MPS 50 50 50 50
Planning
factors Material Requirements Planning Inventory
status
• Gross-to-net calculations
• Planned order recommendations On hand?
Bills of • Exception action messages On order?
Lead time?
material
Safety stock?
Source: Adapted from David F. Ross, Distribution Planning and Control – Managing in the Era of Supply Chain Management, 3rd edition.
Production Rate =
(Ending Inventory - Beginning Inventory ) + Forecast
Number of Periods
=
(1,000 - 1,500 ) + (5,200 + 5,400 + 4,900 ) = 5,000 units per month in Q1
3
Logistics Capacity Check
Capacity
Tactical-level check
• Warehouse capacity – aggregate by storage type
• Labor and equipment capacity and load – set standard hours
by product family
– Two categories: receiving and put-away, picking and shipping
Where selling?