Supply Chain (SC) Definition and Goals Implied Uncertainty Spectrum: (i) New Product (high implied demand

Spectrum: (i) New Product (high implied demand - Allows supply chain to exploit Economies of Scale, fully utilizing fixed cost Carried to satisfy demand that exceeds forecasted qty or supply arrives later than (vii) Obsolete Inventory: Measure the fraction of inventory older than a
Definition uncertainty) (iv) Function Product (i.e. salt) (low implied demand uncertainty) with transport and order thus lower cost expected: (i) Raise level of safety inventory increase product availability and specified obsolescence date.
- Movement of products from suppliers to customers Step2: Understanding the SC capabilities - Average amount of inventory in the SC used to satisfy demand between margin is recovered from customer (ii) Raise level of safety inventory increases (viii) Inventory Turns / Turnover: Measure the no. of times inventory turns over
- Information, funds and products in both direction (Flows) Cost-Responsiveness Efficient Frontier supplier shipments and is usually due to production or purchases in lot sizes that holding costs in a year. Ratio of average inventory to either COGs or sales. Comparing
- Customer is the only source of revenue Curve shows the lowest possible cost for a are larger than demanded by customer. Function of lot sizes 3 Key questions for Safety Inventory: (i) What is the appropriate level of performance of inventory and is measured in $. Indicates how fast a SKU moves
Parties and Stages given level of responsiveness Lot / Batch Size: Qty produce/purchase at a stage of SC product availability? (ii) How much safety inventory is needed for the level of and the higher the turns the better. Shows cash flow and good benchmarking
Suppliers->Manufacturer->Distributors->Wholesalers-> *Improve by pushing the curve outwards [**Ignore Demand Variability and Demand is stable] product availability (iii) What actions can be taken to reduce safety inventory Inventory Turns = Annual Sales / Average Inventory $
Retailers->Customers* **More responsive more costly Q: Qty in a lot or batch size D: Demand per unit time without hurting product availability? Functional Classification of Inventories
Supply Chain Objective: Responsiveness is the ability to: (i) Respond Cycle Inventory = Q / 2 Factors Affecting Level of Safety Inventory: (i) The desired level of product (i) Cycle Stock: Refer above
Max: SC Surplus = Customer Value (Top) – SC Cost (Bottom) to wide range of demand (ii) Meet short lead Ave flow time = ave inventory / ave flow rate | T = I / D availability (ii) Uncertainty of dd and supply (iii) Inventory replenishment (ii) Safety Stock: Not needed if no uncertainty in demand and supply.
3 Supply Chain Decision Phases time (iii) Handle large variety of products Ave flow time from cycle inventory = Q / (2D) policy Determined by desired level of service
Strategy (Long-term, exp to reverse & constraints to planning) (iv) Build highly innovative products (v) Lower level of cycle inventory is desirable, reduce lag time between product is Measuring Product Availability (iii) Congestion Stock: Items sharing resources with limited capacity, Work in
Decisions on (1) Supply Chain Configuration (2) Resource allocation Meet high service level (vi) Handle supply produced and sold (i) Short lead time allows firm better to prepare when (i) Product fill rate (fr): Fraction of product demand satisfied from product in progress (WIP), inventory in queues. Inventories build up because the
(3) Processes each stage will perform: (i) Outsource/inhouse functions uncertainty demand changes in marketplace (ii) Reduce working capital requirements (iii) inventory (1 product) equipment is not available. Not purposeful and occurs uncontrollably.
(ii) Location and capacities of facilities (iii) Products to be made and inventory SC Efficiency is the inverse to Reduce inventory holding costs | Reduce cycle inventory improves supply chain (ii) Order Fill Rate: Fraction of orders filled from available inventory (Orders (iv) Anticipation Inventories: To deal with seasonal demand/supply. Accumulate
(iv) mode of transport (iv) information systems the cost of making and deliveringability to match supply with demand can have multiple products) stock in advance to account for spikes. Purposeful and allow stable product rate
**Account for market uncertainty the product to the customer (i.e.Cycle inventory is held to take advantage of economies of scale by reducing (iii) Cycle Service Level (CSL): Fraction of replenishment cycles that end with instead of overtime during peak period.
Planning (1 year, constraints to OPs and max SC Surplus) 7-11 Japan: highly responsive cost in the supply chain (refer below) all customer demand being met (v) Pipeline (WIP) Inventories: In transit between SC stage (en route)
Decisions on (1) Demand Forecast (2) Markets supplied from locations (3) SC) Average price paid per unit purchased (C): Key cost in lot sizing decision. *Replenishment Cycle: Interval between 2 successive replenishment deliveries Supply Chain Coordination & Impact on Performance
Planned inventory buildup (4) Subcontracting (5) Inventory Policies (6) Timing Step3: Achieving strategic fit Material cost decrease in greater lot sizes Measuring Demand Uncertainty Supply chain coordination all stages of the chain take actions that are aligned
and sizing of market promotions (i) Ensure that degree of SC Average price per unit = $C/unit D = Ave DD per period | σD = Std Dev of dd per period and increase total supply chain surplus, requires each stage to share information
*Objective to max Supply Chain Surplus responsiveness is consistent withFixed ordering cost (S): All costs that do vary with the size but are incurred each Lead Time (L) is the gap between when an order is placed and when it is and account for the effects of its actions on the other stages
**Account DD uncertainty, exchange rates, competition the implied uncertainty time an order is placed received Lack of coordination results when: (i) Objectives of different stages conflict (ii)
Operations (Day or Week, constrained by Strategy and Planning) (ii) Assign roles and to different
Fixed ordering cost per lot = $S/lot DL = dd during lead time | DL = D X L Information moving between stages are delayed or distorted
Decisions on customer orders: (1) Allocate orders to inventory or production (2) stages of supply chain that ensure
Ex of Ordering Cost: (i) Buyer time (ii) Transportation cost (iii) Receiving cost σL = Std Dev of dd during L |
the appropriate level of Bullwhip Effect: Fluctuations increase as they move up the SC from retailers-
Order due date (3) Picklist (4) Delivery schedule (5) Replenishment (iv) others
L wholesalers-manufacturers-suppliers (i) Distorts demand information within the

√∑
* Objective is to handle individual customers’ orders effectively responsiveness Holding cost (H): Cost of carrying 1-unit inventory for a time period =
supply chain (ii) Results from a loss of supply chain coordination
(iii) Ensure functions strategies 2
**Less uncertain due to short time horizon
2 Supply Chain Process Views
Comparison Efficient Supply Chain
support competitive strategy
Responsive Supply Chain
$H/unit/year = hC | h is a fraction of C
Eg of Holding Cost: (i) Cost of Capital (ii) Expired or obsolescence cost (iii)
Handling cost (iv) Occupancy cost (v) Miscellaneous (theft, security, tax,
σ = L σ +2 ∑ ρ σ σ
i ij i j
*Move all parties in the SC away from the efficient frontier, results in drop in
customer satisfaction and profitability
(1) Cycle: Processes in SC are
divided into a series of 4 cycles Goal Supply lowest cost Quick to demand insure) i=1 i> j Lack of Coordination increases variability and hurts Supply Chain surplus
Impact on the following cost
and performed at the interface Product Max Performance Modularity to allow Primary Role of Cycle Inventory: Allow different stages to purchase product in Assume norm dist, Demand Between the Various Period Independent:
Design Low Cost postponement lot sizes that minimize sum of material, ordering and holding costs and should σL = sqrt (L σD2) = σD sqrt(L) Performance Measure Lack of Coordination
between 2 successive stages
Pricing Low margin High margin consider costs across the entire SC. Increase cycle inventory increase SC cost. Coefficient of variation = cv = σ/μ Manufacturing Cost Increase
**Cycle view (Flow) is useful
Manufacturing Low cost – High utilization Capacity flexibility to allow Economies of Scale exploited in replenishment decisions in 3 typical case: COV measures the size of uncertainty relative to demand Inventory Cost Increase
when considering operational
decisions because it specifies the uncertainty (i) Fixed cost in incurred each time an order in placed or produced (ii) Supplier Measuring Supply Uncertainty (Assume Norm Dist) Replenishment Lead Time Increase
roles and responsibilities of each Inventory Minimize to cut cost Maintain buffer offers price discounts based on the qty purchased per lot (iii) Supplier offers L = Ave lead time | σL = Std Dev of lead time Transportation Cost Increase
short-term price discounts or trade promotions Shipping and Receiving Cost Increase
member of the supply chain Lead-Time Reduce but not at the Reduce even if cost is 2 2 2-- s = std dev of L
(i) Procurement Cycle: Supplier -
Manufacturer
expense of cost
Supplier selection Cost and quality
significant
Flexibility, speed, quality
Economies of Scale to Exploit Fixed Costs
Lot sizing for a single product (EOQ) – Minimizing all 3 costs
DL = D X L |
σ =√ Lσ + D s
L D L
L Product Availability
Profitability
Decrease
Decrease
(ii) Manufacturing Cycle: Manufacturer - Distributor and reliability [**Assume demand is steady at D units per unit time, No shortages allowed and *Increase in supply uncertainty significantly increase ss required. L-uncertainty Obstacles to Coordination in SC
(iii) Replenishment Cycle: Distributor - Retailer demand supplied from stock, Replenishment lead time is fixed (initially assume has more impact than L and reduction in L-uncertainty can dramatically reduce (i) Incentive Obstacles: Occur when incentives offered to different stages or
Product Life Cycle required ss without affecting product availability
(iv) Customer Order Cycle: Retailer – Customer to be 0)] participants in a SC lead to actions that increase variability and reduce SC
Beginning Stages: Demand is uncertain & supply unpredictable. Margins are
(2) Push/Pull: Processes in dependent on timing of execution relative to Annual material cost = CD Replenishment Policies profits (misaligned) (1) Local optimization within functions or stages of SC (2)
high & market share critical. Product availability is crucial, and cost is
customer dd (i) Pull: initiated by customer order (REACTIVE) (ii) Push: in Number of orders per year = D / Q (i) Continuous Review: Inventory is continuously tracked, Order for a lot size Q Sales force incentives to maximize sales at providing price discount to retailers.
secondary
expectation of order (SPECULATE) Annual ordering cost = (D / Q) S is placed when the inventory declines to the reorder point (ROP) Artificially inflate demand
Later Stages: Demand is more certain & supply predictable. Increase
*Push/Pull boundary separates push from pull Annual holding cost = (Q / 2) H = (Q / 2) hC (ii) Periodic Review: Inventory status is checked at regular periodic intervals (ii) Information Processing Obstacles: When demand information is distorted as
competition lowers margin. Price is a significant factor in customer choice
Supply Chain Macro Processes Total annual cost = CD + (D / Q) S + (Q / 2) hC and Order is placed to raise the inventory level to a specific threshold. it moves between stages of the SC, leading to increasing variability of orders in
5 Supply Chain Levers (Deal with Supply Chain Uncertainty) Economic Order Quantity (EOQ) = Q* = sqrt [(2DS) / (hC)]
1. Customer Relationship Management (CRM): Between firm and customer (i) Determining the Appropriate Level of Safety Inventory SC (1) Forecasting based on orders and not customer demand (2) Lack of
(1) Capacity: Excess and flexible capacity Optimal Order Freq = n* = D / Q* = sqrt [(DhC) / (2S)]
Source (ii) Negotiate (iii) Procure (iv) Design and Supply Collaboration Evaluating Safety Inventory given ROP information sharing
(2) Inventory: Common lever used to deal with uncertainty * Total ordering and holding cost are relatively stable around the EOQ. Firm is Expected Demand during Lead Time = D X L
2. Internal Supply Chain Management (ISCM): Internal to the firm (i) Strategic, (iii) Operational Obstacles: Occur when placing and filling orders lead to an
(3) Time: Supply timeliness & customer willingness to wait better served ordering lot size close to EOQ rather than the precise EOQ
Demand, Supply Planning (ii) Fulfilment and Field Service Safety Inventory = ss = ROP – D X L increase in variability (1) Ordering in large lots (2) Large replenishment lead
(4) Information: More information to reduce SC uncertainty ** Demand increase by k factor, optimal lot size and order frequency ↑ by sqrt Cycle Service Level = P (Demand during lead time < ROP)
3. Supplier Relationship Management (SRM): Between firm and supplier (i) time (3) Rationing and shortage gaming (inflate orders to secure products that
(5) Price: Varying prices of products and services (k) and flowtime ↓ by sqrt (k)
Market, Price Sell (ii) Call Centre and Order Management [*Assume norm dist, CSL = NormDist (ROP, DL, σL,1)] have shortages will artificially inflate order)
**Investing more in one lever generally allows the Supply Chain to invest less *** Fixed order cost reduced by k2, EOQ is reduced by k,
Competitive and Supply Chain Strategies Given CSL = P (DD during lead time < D L +ss) (iv) Pricing Obstacles: When pricing policies for a product lead to an increase in
in other levers. Strategic fit, SC must find the right balance between investments
Competitive Strategy: Defines the set of customers needs a company offering Production Lot Sizing (Production requires time) ss=Norminv (CSL, DL, σL) – D X L = Normsinv (CSL) X σL variability of orders placed (1) Lot-size based qty decisions (2) Price
in the levers to effectively serve the target customer segment(s).
must satisfy [**Assume Rate of Production (P) > Demand (D)] Inventory Builds Up = P – D Evaluating Fill Rate Given Reorder Point fluctuations (promotions that encourage forward buying)
Framework for Supply Chain Decisions Economic Production Qty (EPQ) (v) Behavioral Obstacles: Problems in learning within organizations that
The Value Chain: New Product Development > Marketing & Sales > Operations Expected Shortage per Replenishment Cycle (ESC): is the ave units of demand
Competitive Strategy -> Supply Chain Strategy -> Supply Chain Structure QP = sqrt [(2DS) / [(1 – D / P)(hC)]] contribute to information distortion (1) Each stage of SC views it actions locally
> Distribution > Service that are not satisfied from the inventory in stock per replenishment cycle
(Efficiency vs Responsiveness) -> Drivers Annual setup cost = (D / QP) S and unable to see impact on other stages (2) Different stages of the SC react to
(i) Product Development (Functional) Strategy: Specifies the products that the ESC = -ss [1 - normdist (ss / σL,0,1,1)] + σL normdist (ss / σL,0,1,0)
company will try to develop 6 Drivers of SC – Logistical (L) / Cross-Functional (C) Annual holding cost = (1 – D/P)(QP / 2) hC Product fill rate = fr = 1 – ESC / Q = (Q-ESC) / Q the current local situation rather than trying to identify the root cause (3)
(ii) Marketing & Sales (Functional) Strategy: Specifies how the market will be Structure Drivers to meet required performance Efficiency (E) and Lot Sizing with Capacity Constraint * Desired product availability (fr) goes up, required safety inventory increases Different stages of the SC blame other functions for the fluctuations (4) No
segmented, and product positioned, priced, and promoted. Responsiveness (R) and thus strategic fit. Interactions affect SC performance If order size is constrained to K units from limited Capacity (i.e. fixed cost like ** Goal is to reduce the level of safety inventory and not affect product stage of the SC learns from its action over time (5) Lack of trust among SC
(iii) Supply Chain (Functional) Strategy: Determines the nature of material Facilities (L): Physical location in the network where product is stored, transport) and Q* > K: (i) Compare cost of ordering K units and Q* (ii) availability: (i) Reduce the supplier lead time, L (ii) Reduce uncertainty of DD causes them to be opportunistic at the expense of overall SC performance
procurement, transportation of materials, manufacture of product or creation of assembled or fabricated (Capacity); More facilities more R, centralized more E Optimal order is min (Q*,K) (σD) and supply uncertainty (sL) Managerial Levers to Achieve Coordination
service, distribution of product, follow-up service, whether processes will be in- Inventory (L): All raw materials, work in process, and finished goods within a Demand Lumpiness: Demand comes in large order adds variability to demand (i) Aligning Goals and Incentives: So that every party in SC activities work to
Lot Sizing with Multiple Products or Customers
house or outsourced supply chain network. High inventory more R, Low inventory more E and cos inventory to drop below ROP. Solution to raise ROP by half the maximize SC surplus (1) Align goals across the SC (2) Align incentives across
Strategic Fit (Competitive and supply chain strategies have aligned goals)
Functional Strategies support one another and competitive strategy. Functions
Transportation (L): Moving inventory from point to point in the supply chain.
Faster transport more R, slower modes might more E
D = Annual Demand for Product i
i average size of order
Seasonal Demand DL & σL can change w season. Fixed ROP can lead to
functions (3) Pricing for coordination (4) Alter sales force incentives from sell-
in (to retailer) to sell-through (by retailer)
Information (C): Data and analysis concerning facilities, inventory,
must structure their processes and resources to execute these strategies.
Step1: Understanding the customer and SC uncertainty
transportation, costs, prices, and customers throughout the supply chain
Sourcing (C): Entity to perform the supply chain activity (e.g. is = Additional order cost incurred for Product i stockout. Evaluate ROP in terms of days of demand to account for varying D L &
σL .
(ii) Improving Information Visibility and Accuracy: (1) Sharing customer
demand data (2) Implementing collaborative forecasting and planning (3)
(i) Quantity of product needed in each lot (ii) Response time customers are Inventory Related Metrics (Inventory Decisions) Designing single-stage control of replenishment (a) Continuous replenishment
¿
willing to tolerate (iii) Variety of products (iv) Service level required (v) Price
of product (vi) Desired rate of innovation
Inhouse/Outsource manufacture/transport, Supplier)
Pricing (C): Determines how much to charge for offerings. Differential pricing
can be used to attract customer who value more R or E
S =S+ ∑ S i
(i) Cash to Cash Cycle Time (C2C): High level metrics that includes
inventories, AP and AR. Duration between purchase of inventory to AR of
programs (CRP) (b) Vendor managed inventory (VMI)
(iii) Improving Operational Performance: (1) Reduce replenishment lead time
Demand uncertainty – uncertainty of customer demand inventory sales. Combines 3 SC ratios (1) Days of Inventory (DOI) (2) Days of (2) Reduce lot size (3) Rationing based on past sales and sharing info to limit
Implied demand uncertainty – resulting uncertainty of the portion the demand Inventory (Logistical Driver in Supply Chain) k
Payables (DOP) (3) Days of Receivable (DOR) gaming

√
that the supply chain plans to satisfy based on the attributes the customer desires Role: (i) mismatch between supply and demand (ii) exploit economies of scale C2C Cycle = DOI + DOR – DOP (iv) Designing Pricing Strategies to Stabilize Orders: Encouraging retailers to
Increased implied demand uncertainty leads to increased difficulty in matching
supply to demand and to stockout or oversupply
(iii) reduce costs (iv) improve product availability (v) affects assets, costs,
responsiveness and material flow time
Overall Trade-off: (i) increase inventory generally makes SC more responsive
∑ D hC
(ii) Average Inventory: Measure the average inventory amount carried,
measured inidays of demand,
i $ value, units
order in smaller lots and reduce forward buying (1) Moving from lot-size based
to volume-based quantity discount (2) Stabilizing price (EDLP)
(v) Building Strategic Partnerships and Trust: Easier to use levers to achieve
Customer Need Implied Demand Uncertainty ¿ i(iii)
=1Products with more than a specified number of days of inventory: Identify
More Range of Qty Increase (greater dd variance) but hurts efficiency (ii) higher inventory may facilitate reduction in production,
purchasing and transport cost because of improved economies of scale (iii) Optimal Order Frequency=n = products which the firm carrying high level of inventory, identify oversupply
¿ reason to justify for high inventory, discounts / slow-mover
coordination if trust and strategic partnerships are built (1) Sharing accurate info
(2) Lower transaction costs between stages (3) All parties must believe that the
Decrease lead time
Increase variety
Increase (less lead time avail)
Increase (dd per product less predict)
inventory holding costs increase
Material Flow Time: time between point at which the material enters the SC to
2S
products, identify
(iv) Seasonal Inventory: Measures the amt by which the inflow of product benefits of improved coordination are being equally shared.
* Limit forward buying and stabilize orders
¿ exceeds sales (beyond cycle and safety inventory), solely to deal with
Increase service lvl
Increase innovation
Increase (unusual surge in demand)
Increase (new product unpredictable)
the point at which it exits the SC
Throughput: The rate at which sales occur Annual Order Cost=S ×n anticipated spikes in dd
(v) Average Replenishment Batch Size: Measure the ave amt for each
Improving Coordination in Practice: (i) Get top management commitment for
coordination (ii) Devote resource to coordination (iii) Focus on communication
More channels Increase (less predict per channel) k
Low Price Increase (more demand on SC)
Little’s Law: Inventory(I) = Throughput(D) X Flowtime(T)
Components of Inventory Decisions D hC i
replenishment order. Batch size should be measured by Stock Keeping Units
i (SKUs) in terms of both units and days of demands. Estimated by averaging
with other stages
Practical Approaches to Improve SC Coordination
Implied Uncertainty Low High (i) Cycle Inventory: (Refer Below) Annual Holding Cost =∑ over time the difference between maximum and minimum inventory (measured (i) Continuous Replenishment and Vendor-managed Inventories: (1) Single
Product Margin
Ave Forecast Error
Low
10%
High
40% - 100%
(ii) Safety Inventory: (Refer Below)
(iii) Seasonal Inventory: Inventory built up to counter predictable variability in 2n i=1
in each replenishment cycle)
(vi) Average Safety Inventory: Measure the ave amt of inventory on hand when
point of replenishment (2) CRP-wholesaler or manufacturer replenishes based
on POS data (3) VMI-manufacturer or supplier is responsible for all decisions
Ave Stockout Rate 1% - 2% 10% - 40% demand. Cost of carrying additional inventory versus cost of flexible production replenishment order arrives. Measured by SKU in units and days of demand. regarding inventory (4) Substitutes
Ave Forced Markdown 0% 10% - 25% (iv) Level of Product Availability: Fraction of demand that is served on time D hC i i
Estimated by averaging over time the min inventory (ii) Collaborative Planning, Forecasting and Replenishment (CPFR): (1)
Supply Source Capability: Frequent breakdowns/ Unpredictable and low yields/ from product held in inventory. Trade-off between customer service and cost
(v) Inventory-related Metrics: (Refer Below) Cycle Inventory for i= (v) Fill rate: Fraction of order / demand met on time from inventory. Should not
be averaged over time but by unit of demand (Quantity)
Strategy and planning (2) Demand & supply management (3) Execution (4)
Analysis
2n
Poor quality/ Limited supply capacity/ Inflexible supply capacity/ Evolving
production process => Cycle Inventory (Inventory Decisions) (vi) Fraction of time out of stock: Measure the fraction of time that a SKU had 0 CPFR Scenario Where in SC Industries
inventory. Use to estimate lost sales during the stockout period Retail Event Highly promoted All except EDLP
Safety Inventory (Inventory Decisions)
Collaboration
DC Replenishment
channels/category
Retail DC or Distributor Drugstore, hardware, n=number of potential plant locations/capacity
Information Centralization - Allow customers or stores to locate stock: Improves
product availability without adding inventories and reduces the amount of safety
Gravity Model

Step 1: For each supply source or market n, evaluate d

c ej =cost ¿ ship 1unit ¿ warehouse e ¿ m
Collaboration grocery inventory
n
Store Replenishment Direct or retail DC store Mass merchants, club
Specialization (Aggregation) m=number of markets∨demand points Step 2: Obtain a new location (x’,y’) for the facility where
Decision Variables
Collaboration delivery stores
Aggregation is dependent on required level of safety inventory and considers
k y i=1 if plant i isopen ,0 otherwise
Collaborative Assortment Apparel and seasonal
Planning goods
Department stores,
coefficient of variation of demand: (1) Low demand, slow moving items,
specialty retails D j=annual demand ¿ market j xn D n Fn
K i= potential capacity of plant i ' ∑
typically have a high coefficient of variation -> Centralized (2) High demand,
*Organizational and technology requirements for successful CPFR
fast moving items, typically have a low coefficient of variation -> Decentralized y e =1 if warehouse e is open , 0 otherw
* DC replenishment collaboration is often the easiest to implement because it
Product Substitution on Safety Inventory (Aggregation) n=1 dn
requires aggregate level data.
(1) Manufacturer-Driven Substitution (i) Allows aggregation of demand (ii)
* Risk and Hurdle store replenishment collaboration requires a higher level of x= k x hi =quantity shipped ¿ supplier h ¿ pl
Reduce safety inventories (ii)Influenced by cost differential, correlation of dd
investment in technology and data sharing to be successful. f i=annualized ¿ cost of keeping plant iopen Dn F n
Manufacturer Organization (iii) Component substitution
Retailer Organization
∑¿ plant x ie=quantity shipped ¿ plant i ¿ wareh
Demand Planning
Sales
(2) Customer-Drive Substitution (i) Allows aggregation of safety inventory (ii)
Merchandise Planning
Buying Substitue product that is similar for the customer c ij =cost of producing∧shipping one unit d i¿ market j
Customer Service / Logistics Replenishment
Component Commonality (Aggregation)
(1) Unique Components: (i) Uncertainty of demand for a component is the same Cost includes production , inventory ,transportation∧tariffs
n=1
k
n
x ej=quantity shipped ¿ warehouse e ¿
Impact of Aggregation on Safety Inventory
y D F Objective Function
Di=Mean Periodic Demand ∈Region (2) Commoni, Components:
i=1 , … , k for components is an aggregation of the Decision Variables
as final product (ii) Results in high levels of safety inventory
(i) Demand
∑ d n n n
n t
demand for the final product (ii) Demand is more predictable (iii) Reduced
y i=1 if plant i isopen ,0 otherwise ' n=1 MinCost F y Fe y e + ∑
σ i =Std Dev of Periodic Demand ∈Postponement Region(Aggregation)
levels of inventories
i , i=1 , … , k y= k n =Min ∑ i i + ∑
ρij =Correlation of Periodic Demand for x =quantity shipped ¿ plant i¿ market j D F i=1 e=1 h
chainRegionsi
for the push phase (ii),Product
j, 1differentiation
≤i ≠ j≤ closekto the pull
Delayed product differentiation or customization: (i) Common components in
ij n n
Assumption that demand follows normal distribution
the supply
phase of the supply chain (iii) Inventories in the supply chain hence can be Objective Function: ∑ dn
Constraint
n
mostly aggregated n n m n=1
C k ∑ x hi ≤ Sh for h=1 , … , l
Mean Demand for all Regions=D Cσ= D canD
∑ be applied ¿ all methods of Aggregation MinCost =Min ∑ f i y i + ∑ ∑ c ij x(x,y)ij = (x’,y’) and go to step 1.
Step 3: If the location (x’,y’) is almost the same as (x,y) stop. Otherwise, set

i i=1
Complex Capacitated Plant Location (Single Sourcing)
Managerial Levers to Reduce Safety Inventory i=1 i=1 j=1 l
i=1 of Supply Uncertainty (i) Sharing Information (ii) Coordinated
(1) Reduction Constraint
y i=1 if plant i isopen ,0 otherwise
Variance of Demand for all Regions=var ( D )=∑ σ + 2 ∑ ρ σ σ
DD k
C (i) Delays contribute
(2) Reduction of Lead Times 2 more to lead time than
n ∑ xi hi, 0≤otherwise
x ij =1if market j is supplied by factory
K i y i for i=1 , … , n
production and transportation time
i DD (ii) Reduce information
(3) Reduction of DD Uncertainty (i) Aggregate
i=1 i>j
ij i j ∑ x ij ≥ D j for j=1 , … , m h =1
l t
distortion through sharing i=1 D j=quantity shipped ¿ plant i¿ market j
Assuming mean and std dev of periodic demand for all region is the same:
C
Network Design Deisions
(i) No. of manufacturing plants, production lines, distribution centers, cross
m ∑ x hi − ∑ x ie ≥ 0 for i=1 , … , n
D =k × D docking facilities (ii) Location of facilities (iii) Capacity of facilities (iv) Type
of products at facilities (v) Market served by facility ∑ x ij ≤ K i y i for i=1 , … ,n
1 factory to 1 market
Objective Function : h =1 e=1
t
σ CD =√ k σ 2D +k ( k −1 ) ρ σ 2D *Revisit Design Decisions after Market Changes, Mergers or Factor Cost j=1 n n m
c ij xxijie ≤ K i y i for i=1, … , n
MinCost =Min ∑ f i y i + ∑ ∑ D j ∑
Changes
If demand for each region is independent and mutually exclusive, Factors Influencing
Network Design
y i ϵ { 0,1 } for i=1 , … , n , x ij ≥ 0
e=1
ρ=0 (i) Strategic factors (ii) Competitive factors: +ve externalities and location to
split the market (iii) Political factors (iv) Infrastructure factors (v) Customer
All decision variables are non negative
Constraint
i=1 i=1 j=1
t
Accounting for Taxes, Tariffs and Customer Requirements

σ CD =√ k σ D response time and service level (vi) Total logistics cost (vii) Macroeconomic
factors: tariffs and tax incentives; exchange-rate and demand risk
Objective Function: n
∑ x ie ≤ W e y e for i=1 , … , n
∑ x ij =1
n
for
m
j=1 , … , m
Competitive Factors: Locating to split the market and capture largest share
k
m n n
e=1
Max Profit =Max ∑ r j ∑ xij −∑ i=1
d1=a+(1-b-a)/2
f i y i−∑ ∑ cij x ij −1 n m
Total Safety Inventory ∈ Decentralized Option=∑ F S ( CSL) × √ L ×σ i d2=(1+b-a)/2
j=1 i=1 i=1 m i=1 j=1
i=1
n
Framework for Network Design Decisions
(1) Maximize overall and long-term profitability of supply chain while
∑ D j x ij ≤ K i y i for i=1 , … ,n
providing appropriate responsiveness (2) Trade-offs during network design (3)
Constraint ∑ x ie −∑ x ej ≥0 for e=1 , … ,t
C i=1 j=1
Total Safety Inventory ∈Centralized Option=F−1S ( CSL ) × √ L× σ D
Design models used to: (i) decide on locations and capacities (ii) assign current
j=1
x
∑ ij j ≥ D for j=1 , … , m
demand to facilities and identify transportation lanes
*The process start by defining SC strategy aligned with competitive strategy.
t
Holding Cost −Savingson Aggregation per Unit Sold x , y ϵ { 0,1 } D ≥ 0
The robustness of the network should be checked to handle various risk and
uncertainties faced by DC and revised as demand and costs change.
i=1 ij i ,
j ∑ x ej ≥ D j for j=1 , … , m
−1 k m e=1
F ( CSL ) × √ L× H Phase I: Define Supply Chain Strategy / Design Locating Plants and Warehouses Simultaneously

¿ S

DC (∑ i=1
σ i−σ C
D ) (1) Clear definition of firm’s competitive strategy (2) Forecast likely evolution
of global competition (3) Identify constraints on available resources (Internal –
capital, growth strategy, network) (4) Determine broad supply chain strategy
Phase II: Define the Regional Facility Configuration
∑ x ij=K i y i for i=1 , … , n
j=1
l=number of suppliers y e , y i ϵ { 0,1 } , x hi , x ie , x ej ≥ 0
n=number of potential plant locations/capacity
Design Transnport Network

y i ϵ { 0,1 } for i=1 , … , n , x ij ≥ 0

Safety Inventory Holding Cost-Savings per Unit Sold on Aggregation (1) Forecast of demand by region or country (2) Identify fixed and variable Movement of product from 1 location to another. Transport is a significant cost
- Increases with the desired CSL
- Increases with the replenishment lead time L
costs, economies of scale or scope (3) Identify regional tariffs, requirements for
local production, tax incentives, export or import restrictions (4) Identify
All decision variables are non negative
t=number of potential warehouse locations
component. Shipper requires the movement of the product and Carrier moves or
transports the product.
competitors Transportation Decision affect SC profitability and influence inventory and
- Increases with the holding cost H
(5) Identify demand risk, exchange-rate risk, political risk (6) Production
Technologies (Cost, Scale/Scope, Impact, Support, Felexibilities)
Models for Identifying Potential Sites
Gravity Location Models
m=number of markets∨demand points facility decisions within a SC.
Air: Expensive but fastest, suitable for small high value, emergency shipments
Phase III: Select a Set of Desirable Potential Sites
(1) Hard and soft infrastructure requirements: (i) Production Methods (Skills x n , y n =coordinate location of either Sha=supply capacity source
market ∨supply at supplier
n h Variable depending on passenger / cargo
Cost component (1) Fix infrastructure and equipment (2) Labor and fuel (3)

needsm response time) (ii) Available Infrastructure Key issues (i) Location / no. of hubs (ii) Fleet assignment (iii) Maintenance
- Increase with the coefficient of variation
Phase IV: Location Choices and Market Allocation
F = cost to ship one unit for one mile between facility and market or supply n
n
K i= potential capacity of plant i schedules (iv) Crew scheduling (v) Prices and availability
σD (1) Factor Costs (Labor, Materials, Site Specific) (2) Logistics Cost (Transport
D n=quantity ¿ be shipped between facility ∧market ∨supply n Package Carriers: (1) Small package to about 150 lbs (2) Expensive (3) Rapid
of demand
Inventory, Coordination)
W e = potential warehouse capacityvalue-added
at location e consolidation is key factor
and reliable (4) Suitable for small and time-sensitive shipments (5) Additional
services (6) Shipment
D
Designing a Regional Network Configuration
d n=distance between facility at location ( x , y ) ∧the supply∨market n Truck: (1) Significant fraction of transport (2) Truck load: low fixed cost:
D j=annual demand ¿ market j
Inputs: (1) Demand (2) Desired response time (3) Fixed cost of opening a
facility (4) Variable cost of labor and material (5) Inventory holding cost (6) imbalance between flows (3) Less than Truckload (LTL): Small lots, hub and
- Decrease with increase correlation
spoke, may take longer than TL (4) Fatigue related accidents
2 2
coefficients
Square Root Law
ρ
Transport cost between regions (7) Sales price of product (8) Taxes and tariffs
(9) Potential facility capacity
Objective Function: Typically governed by cost
d n = ( x√ −x n ) + ( y− y n ) F i =¿ cost of locating plant at locationi
Rail: Suited for low-value and large shipments that are not time sensitive
(1) Move over large distances (2) High fixed costs in equipment and facilities
(3) Scheduled to maximize utilization (4) Transport time is long
Variable: (i) No. of Facilities / Market (ii) Capacity of Facilities / Market k
Possible disadvantages to aggregation
(iii) Fix and Variable Cost of Facility / Market F e =¿ cost of locating warehouse at location e
Water: (1) Geographic limitation (2) Large loads at low cost (3) Slowest (4)
- Increase in response time to customer order
- Increase in transportation cost to customer
Decision Variable: (i) Qty Shipped (ii) Binary decision on facilities Total Transport Cost =TC= ∑ d n D n nF Least expensive (5) Dominant in global trade
Pipeline: (1) High fixed cost (2) Primarily for petroleum and gas (3) Ideal for
Physical and Information Centralization (Aggregation)
Trade-Offs of Physical Centralization: Decrease holding cost and facility cost
Constraints: (i) Decision Variable should be 0 or positive (ii) Facilties’ Capacity
Constraint (iii) Demand should be met (Except when maximizing profit with n=1 c hi =cost ¿ ship 1 unit ¿ supply sourcelargeh factoy
and¿steady i structures suitable for predictable demand
flows (4) Pricing
Intermodal: (1) Use of more than 1 mode (2) Maybe the only option for global
different revenue, demand need not be met) (iv) Outgoing < Incoming Decision Variables:
but increase transportation cost
Capacitated Plant Location Model
( x , y )=coordinatelocation of facility c ie =cost ¿ ship1 unit ¿ factory i¿ warehouse etransfer between nodes
trade (3) Convenience for shippers – one entity (4) Key isseus - Exchange of
information to facilitate
Transportation Infrasturcture and Policies
Infrastructure such as ports, roads, and airports has a significant impact on
transportation. (1) Without monopoly, deregulation and market forces help
(1) Saving Matrix Method: (i) Identify distance matrix (ii) Identify savings
matrix (iii) Assign customers to vehicles or routes (iv) Sequence customers σ D =Standard deviationof demand per period
create an effective industry structure. (2) Pricing should reflect the marginal within route
impact on the cost to society. *Given its inherent monopolistic nature, most
transportation infrastructure requires public ownership or regulation. In the case
(i) Identify Distance Matrix:
2
L= Average lead time for replenishment
2
of public ownership, pricing based on average cost leads to overutilization and
congestion. It is important to use some form of congestion pricing under which
users are forced to internalize the increase they cause in network cost. (ii) Identify Savings Matrix:
√
Dist ( A , B )= ( x A −x B ) + ( y A − y B ) T =Review interval
Design Options for Transportation Network
(1) Transport be direct or via intermediate site (2) Intermediate site to stock or S ( x , y )=Dist ( DC , x ) + Dist ( DC , yCSL=Desired )−Dist ( x , y ) cycle service level
serve as cross-docking location (3) Deliver single or multiple destinations
Network Structure Pros Cons
(iii) Assign Customers: Add the customers based on max savings and capacity
(iv) Route Sequencing: (A) Farthest Insert (B) Nearest Insert (C) Nearest
Probability (demand during T + L ≤ OUL)=CSL
(2) Generalized Assignment Method: (i) Assign seed points for each route (ii) Mean demand duringT + L periods , D T+ L = ( T + L ) D
Direct Shipping No warehouse. Simple to High inventory (larger lot Neighbor (D) Sweep (Route Improvement: 2-OPT | 3OPT)
coordinate size)
Direct Shipping with Lower transport cost for Increase coordination and Evaluate insertion cost for each customer (iii) Assign customers to routes (iv)
Milk Runs small lots. Lower
inventories
complexity Sequence customers within routes
Applicability of Routing and Scheduling Methods
Std Deviation of Demand duringT + LTime=σ T +L =√ T + L × σ D
Via DC with Storage Lower inboud transport Increase inventory cost
Generalized Assignment Method Advantage is it is more sophisticated and
cost thru consolidation
Via DC with Cross-Dock Low inventory Low
and increase DC cost
Increase coordiation
superior to the solution obtained from the savings matrix method when the
delivery schedule has no constraints other than vehicle capacity. Disadvantage
ss=NORMSINV (CSL)×σ T + L
transport cost thru complexity
The generalized assignment method is recommended if the constraints are limit OUL=D T + L + ss
is that is difficult to generate good schedules when there are more constrains.
consolidation
Via DC with Milk Runs Lower outbound transport Further increase in to vehicle capacity or total travel time.

method is simple enough to be easily modified to include delivery time windows Average lot ¿Q=D T =D× T
cost coordination complexity Savings Matrix Method Advantage is its simplicity and robustness. The
Tailored Network Transport choices best Highest coordination
match needs complexity and other constraints and robust enough to give a reasonably good solution that
Compute Annual Cost (Transport + Inventory Holding Cost) can be implemented in practice. Its main weakness is the quality of the
(1) Determine no. of shipment solution. It is often possible to find better delivery schedules using more
(2) Determine trucking cost per shipment = Load + Transport sophisticated methods. The savings matrix method is recommended in case
(3) Determine cycle inventory and inventort holding cost there are many constraints that need to be satisfied by the delivery schedule.
(4) Annual cost = Annual trucking cost + Annual inventory holding cost Software packages for transportation planning and routing and scheduling of
* Direct shipments are the most effective when demand is large. deliveries are available from many supply chain software companies.
* Consolidation via intermediaries are the most effective when demand is small Vehicle Routing Problem
Same Delivery Networks (Mumbai Dabbawalas) Success Factors: (1) Low The problem usually met in transportation part of SC chain (last mile delivery).
Uncertainty of Demand (2) Temporal Aggregation of Demand (3) Use Definition: An Optimization problem with objective to find the best route for
Underutilized Tranport Resource **Responsive at Reasonable Cost each vehicle (min distance/time/cost)
Constraints: (i) Capacity constraints (Weight/Vlume/Insurance) (ii) Pickup and
Trade-Offs in Transportation Design delivery constraints (iii) Time window constraint
(1) Transport and Inventory Cost: (i) Choice of transport mode How to solve: Using solvers (Gurobi, Google OR-Tools, PuLP) – some kinds of
(ii) Inventory aggregation heuristics algorithms to solve. Quality of solution depends on how powerful the
(1) Reduce safety inventories (2) Transportation costs generally increase computer is
Use: (i) Inventory and Facility costs form large fraction of SC Total Cost (ii) For Fleet Routing Engine: Input (Customer’s locations (Demand nodes), Cargo
products with a large value to weight ratio (iii) Product with high dd uncertainty details (no. of cargo, weight, volume, insurance), Vehicle capacity (weight,
Conditions Favoring Aggregate Disaggregate volume), Time window for each customers). Process (Constraint preparation &
Transport Cost Low High VRP solving using Solver). Output (Optimized route & Cargo assignment for
Demand Uncertainty High Low each vehicle, route visualization on map)
Holding Cost High Low Machine Learning
Customer Order Size Large Small A computer program is said learn from experience E with respect to task T and
performance on T as measured by P improves with experience E
Inventory Aggregation
Typical steps to build Machine Learning
Value to Weight High - Reduce SC Cost Low – Increase SC Cost (1) Frame the problem and look at the big picture (2) Get the data (3) Explore
Demand Uncertainty High - Reduce SC Cost Low – Increase SC Cost the data to gain insights (4) Prepare the data to better expose the underlying data
Transport Cost Low - Reduce SC Cost High – Increase SC Cost patterns to Machine Learning algorithms (5) Explore many different models and
Customer Order Large - Reduce SC Cost Small – Increase SC Cost short-list the best one (6) Present your solution (7) Launch, monitor and
(2) Transport cost and customer responsiveness maintain your system
- High Responsiveness – High Transportation Cost Data Required: Performance and Features that can affect Performance
- Decreased Responsiveness – Lower Transportation Cost Machine Learning is Great For
Temporal Aggregation – Combining Orders across time (1) Problems for which existing solutions require a lot of hand-tuning or long
Temporal Aggregation of Demand reduces responsiveness but decrease lists of rules, ML can simplify code and perform better.
transport cost. Benefits of temporal aggregation declines with extended (2) Complex problems which does not have good solution from traditional
aggregation time window. approach and use best Machine Learning techniques to find solution
Mode Cycle & Safety Inv In-Transit Cost Transport Cost Transport (3) Fluctuating environments: ML system can adapt to new data
Time (4) Getting insights about complex problems and large amounts of data
Package 1 1 6 1 Forecast Demand: (1) Past data & features (factors affect demand) –
Air 2 2 5 2 min/max/std dev/mean (2) correlation in the past data
LTL 3 3 4 4 If no data: (1) Get rid of column with NA value (2) Impute using mean/mode
depending on context (3) Build simple prediction model to impute (4) Change 0
TL 4 4 3 3
Types of Machine Learning
Rail 5 5 2 5 (1) Supervised: Task driven (Regression / Classification) (2) Unsupervised:
Water 6 6 1 6 Data driven (Clustering) (3) Reinforcement: Algorithm learns to react to
1=Lowest | 6=Highest environment
Tailored Transportation Metrics Used to Measure Performance
Use of networks and modes based on customer and product characteristics (1) Mean Squared Error (2) Root Mean Squared Error (3) Mean Absolute Error
Factors affecting tailoring: (1) Customer density and distance (2) Customer size (4) Mean Absolute Percentage Error
(i) Transport cost based on distance (ii) Delivery cost based on no, of deliveries Smart Factory (Efficient and Flexible Manufacturing) (Digital Twin)
(3) Product demand and value Data Source: Machines / Materials / Method / Manpower / Environment /
Customer Density Short Distance Medium Distance Long Distance Supply Chain / Customer
High Density Own fleet with Cross-dock with Cross-dock with Analytics & Insights: Visualization & Trigger / Predictive Maintenance /
milk runs milk runs milk runs Predictive Quality / Optimization (Real Time > Warning > Auto Correct)
Medium Density 3PL with milk runs LTL carrier LTL or package Values: Quality / Flexibility / Productive / Cost Efficient / Sustainable
carrier Evolution (Reactive to Predictive): Auto Detect > Auto Diagnose > Auto
Low Density 3PL or LTL carrier LTL or package Package Carrier Prescribe > Auto Correct
carrier Using Deep Learning, Real Time Monitoring and Machine Learning to improve
accuracy, predictability, productivity, reduce downtime, fast time to resolution,
Product Type High Value Low Value
self-learn
High Demand Disaggregate cycle inventory Disaggregate all inventories
*Intellectual Property are usually not outsourced
Aggregate safety inventory Cheap mode of transport for
Cheaper transport for replenishment Goods In Transit = Demand X Transit Time
replenishment Average Inventory = Cycle Inventory + Safety Inventory + Good In Transit Inv
Faster mode for safety Periodic Review Policies
inventory Inventory levels are reviewed after a fixed period of time T and an order is
Low Demand Aggregate all inventories Aggregate safety inventory placed such that the level of current inventory plus the replenishment lot size
Fast customer orders Cheaper transport for equals a prespecified level called order-up-to level (OUL).
fulfillment replenishing cycle inventory
Routing and Scheduling in Transportation D= Average demand per period
Typical objectives include Min total distance, no. of vehicles and travel time.