Stochastic Models
Materials Management
• Involves Short Time Horizon Decisions:
– Supplies
– Inventories
– Production Levels
– Staffing Patterns
– Schedules
– Distribution
} Logistics or
Supply Chain Management
• Tactical not Strategic ⇒ Detailed not Aggregate
• Intimate relation between production and
inventory control
Stochastic Models
Inventory
• Definition: stock of materials held to satisfy future
demand
• Why Inventory?
(1) Time – Quickly meet demand
(2) Uncertainty
– Demand
– Raw Materials availability
– Production process
(3) Economy
(4) Decoupling of operations
Stochastic Models
Inventory — Categories
HIGH
• Raw Materials
– Stocked at plant or supplier FLEXIBILITY
– Lowest cost
• Work-in-Process Inventory (WIP)
– Raw materials transformed by one or more
operations
– As many types as operations
– Need just enough
• Finished Goods
– WIP after final operation
HIGH – Stocked at plant, DC, retail outlet
COST
Stochastic Models
Inventory Costs
• Amount of inventory to carry depends on its costs
• Holding Costs (denoted h)
– Based on capital invested in inventory
– Expressed in dollars or percentage of purchase
price per time period (e.g., $/unit/yr)
• Fixed Order Costs (denoted K)
– Cost of placing an order
– Not dependent on order size
– Equivalent to setup cost in production
• Shortage Costs
– Hardest to quantify
– Derived from desired service level
• Cost of purchased items (quantity discounts)
Stochastic Models
Inventory Theory
• Scientifically answers the questions:
(1) How many should we order? (Lot Sizing)
(2) How often should we order?
• Lot Size (denoted q) varies inversely with order
frequency
• Possible to find a policy that is optimal wrt some
criteria (usually cost)
• Similar analysis used for production batch sizing
Stochastic Models
Components of Inventory
• Cycle Inventory - inventory that varies with lot size
• Safety Stock - inventory that protects against
uncertainty
• Pipeline Inventory - inventory in transit
• Anticipation Inventory - inventory to meet varying
demand
O/H Qty
q
Reorder
Point
Safety
Stock
Lead Time Time
Stochastic Models
Basic Economic Order Quantity (EOQ)
• Assumptions
(1) 8 planning horizon
(2) Continuous deterministic demand with constant
rate
(3) Zero lead time (Instantaneous replenishment)
(4) No shortages (no backorders or stockouts)
• Data
D = demand rate (units/period)
K = order cost ($/order)
C = unit cost ($/unit)
h = holding cost ($/unit-time period)
(cost of holding I units for T periods is ITh)
p = unit purchase cost ($/unit)
Stochastic Models
Economic Order Quantity
300
250 Total Cost
200
$ 150 Holding Cost
100
50
q opt Ordering Cost
0
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
K = $10/order
h = $0.20/unit/year
D = 10,000 units/year
Stochastic Models
Sensitivity of the EOQ
• Cycle Inventory vs Demand
– Cycle Inventory increases with SqRt of Demand
– Not sensitive to small changes
• Lot size vs Setup Cost
– Reducing CO reduces optimal lot size
– Allows small lots
• Lot size vs Interest rates
- Interest part of holding costs
– Lot size decreases as holding costs increase
– Low holding cost allows larger lots
• Square Root function implies low sensitivity to
small errors in estimating K , D, h
Stochastic Models
Variability in Demand – Safety Stock
• Relax assumption that demand is constant and
deterministic
• Assume we know mean and variance of Lead time
demand
• Use Continuous Review (Q,R) Inventory system
• Definition: Inventory Position – measure of ability
to meet future demand
IP = (On Hand) + (On-Order) – (Back Orders)
• Policy: When Inventory Position is less than or
equal to the Reorder Point, R, order Q units.
Stochastic Models
Stochastic Demand Case
O/H Qty
Reorder q
Point
Safety
Stock
Lead Time Time
Stochastic Models
Safety Stock and Service Level
• Safety Stock level depends on:
– Variability of lead time demand
– Desired service level
• Service level
– Overcomes measurement problem of Shortage Cost
– Tolerated shortage is specified
– High service level means high safety stock
• Service level policies
– Cycle-Service Level, Pr[stockout in a cycle] «
– Fill-Rate, % of demand filled off the self
– Number of stockouts
– % of days in the period without stockouts
Stochastic Models
Other Inventory Schemes
• Periodic Review (P-System)
– Check inventory daily, weekly, monthly, etc.
– Place order each period for (T - IP)
where T = target inventory position level
– A.K.A. "order-up-to" systems, fixed interval
systems
• Optional Replenishment System (Hybrid)
– Inventory checked periodically (like P-system)
– Order up to S if IP is below s
– Good if both review and ordering costs are high
– A.K.A s,S system or min- max system
Stochastic Models
Continuous vs Periodic Review
• Advantages of Periodic Review
– Easier administration
– Order at convenient times
– Fixed pickup and delivery times
– Multiple items can be ordered together
– Only have to know IP at review time
• Advantages of Continuous Review
– Individualized for each item cheaper
– Fixed lot sizes are easier
– Less safety stock is needed