lecture 04: forecasting and demand planningoran/classes/lscm_resources/handout/lscm... · lecture...

Demands101 Forecasting Planning Aggre. planning S&OP

Lecture 04:Forecasting and Demand Planning

Oran Kittithreerapronchai1

1Department of Industrial Engineering, Chulalongkorn UniversityBangkok 10330 THAILAND

last updated: May 4, 2017

LSCM v2.0 1/ 24


Outline

1 Understanding your demands

2 Forecasting SCM Demand

3 Operational Planning Concepts

4 Aggregate Planning

5 Sales and Operations Planning

General Reference: [?] [?] [?] [?]

LSCM v2.0 2/ 24


Game Plan: From Forecasting to Shipping...

Guessing customer’s demands ⇐⇒ ForecastingSmoothing resources ⇐⇒ Aggregate planning & Demand managementSourcing R/M needed to produce ⇐⇒ Purchasing & Material requirementEstimating producing time ⇐⇒ Production planingManaging stocks ⇐⇒ Warehousing & Inventory controlDefining appropriated service level ⇐⇒ Channel & Distribution managementEnsuring commitment and delivery products ⇐⇒ Shipping

LSCM v2.0 3/ 24


What are demand?

Are these terms difference?demand =? sale =? order =? forecast

Thinking Points: Recall your last dinning experienceHow to estimate resources restaurants should have?Which channel do restaurants receive ‘demand’ information? how reliable?What happen if restaurants estimate incorrectly?

LSCM v2.0 4/ 24


How should we manage demands?

Understanding nature of demandPrioritizing independent easy to monitor demandForecasting ‘reasonable’ demandCommunicating and updating demand forecastInfluencing quantity and time of demand

LSCM v2.0 5/ 24


Understanding nature of demands:

Aggregate VS Market: Why do we think about demand of our brand?Stochastic VS Deterministic: Why do we think about constant demand?Dependent VS Independent: Why do we think about independent demand?

How do we manage/model inventory?

Demanddep = f(Demandindi ,Demandind

j , ·)

Demandindi = demandcertain + demandrandom

⇓ ⇓• steady • safety stock• trend• seasonal

LSCM v2.0 6/ 24


Postponement: delay differentiation

Motivation: HP CustomizationMotivation: Inventory, nature of demand, multiple marketsIdea: delay the differentiation of product/processRelated concept: localization, simple and modular designTrade-offs: economy of scale, production cost VS inventory, transportationImplementation: cross-organization coordination (manufacturing,purchasing, R&D, market)

Product design: reducing inventory & leadtime by std parts & manuf.Process design: delaying process differentiation by technologyAgile supply chain: time based competitive

Example: DeskJet, Paint mixture, Benetton shirt, HP hard drive test

LSCM v2.0 7/ 24


Prioritizing demand with parieto principle

Idea: few items affects majority of activity

You don’t get what you want in lifeYou get what you focus on and expect.

—Tony Rush

Therefore: focus on few that are importantApplication: ABC analysis

Class A important (15-20 percentile → for 55-75% of activity)Class B some-what important (20-50 percentile → 15-15% of activity)Class C not important (50-100 percentile → 5-15% of activity)

LSCM v2.0 8/ 24


Example of ABC analysis

Item No. Annual dollar % Total Value22 95,000 40.6968 75,000 32.1327 25,000 10.7103 15,000 6.4382 13,000 5.5754 7,500 3.2136 1,500 0.6419 800 0.3423 425 0.1841 225 0.10- $223,450 100%

source: Chase and Jacob. 2011.LSCM v2.0 9/ 24


What we should know about forecasting

What: A process to understand systematic of demandsFacts: Inaccurateness, but all companies need it

Forecasting is, typically, incorrectForecasting is suitable for a group of productsForecasting is inaccurate as time horizon increases

Ideal: Valid for short time and need updatedAware Art & Science: roles of marketing & sale; common misconceptTypes: Qualitative, Time Series, Regression, Simulation

LSCM v2.0 10/ 24


What should we aware before forecast

Description: story, relationship with other dataTime Horizon: hour, day, week, year

Actual = Forecast + errorPattern of Data: seasonal, trend, cycleForecasting Model: assumption, data required, parameters, static VSdynamicAccuracy: measuring, how to improve

LSCM v2.0 11/ 24


Example: US Air Passengers 1949-1951

Month

AirP

asse

nger

s

Year 1949 Year 1950 Year 1951

6080

100

120

140

160

180

200

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

LSCM v2.0 12/ 24


Time Series Review: Workshop

Forecasting Accuracy: Actualt - Forecastt (i.e., MAPE, MSE)Naive Method: Forecast = actualSmoothing Tech: Forecast ≈ set of actual

Moving Average: Forecast = average of actualExponential Smoothing Forecast = combination of actual and last forecast

Trending: Forecast = expo smoothing + slope × periodSlope = gap of forecast

Seasonality: Forecastsea = rsea· Forecastde−sea

LSCM v2.0 13/ 24


Communicating and updating demand forecast

If you have to forecast, forecast often.

source: Edgar Fiedler

Sharing Points:Know data source: market goal =? operation cap, reliability VS urgencyKeep process simple/ comm. flat: verbal & constant, formal & informal,time frameRight communication channel: always confirm & name (scope), IT asenable (goldcity)

Believe the forecasting pattern, not absolute quantity or time

LSCM v2.0 14/ 24


Influencing demand

Cutting prices or putting things on sale is not sustainable businessstrategy.

source: Howard Schultz

Sharing Points:Other factors: service (physician schedule), experience (food), waiting time(BTS, bike), events (MOTOR SHOWS), quality (UBER)Marketing: focused social media (time spent)

LSCM v2.0 15/ 24


Planning Concept

Hierarchical: decision in upper-level → constraint in lower-levelFeedback: systematic correctionMan-Machine interface: ∃/ exceptions → including man’Single’ database & Integration: maintaining data integrity and accuracyTransparency: understanding logic and algorithm behind system

LSCM v2.0 16/ 24


Production Process

source: Chase & Jacabs. 2010. Operation and Supply Chain Management.

Project: product remains in a fixed locationWork center: similar equipment or functions are grouped togetherManufacturing Cell: dedicated area where similar products are producedAssemble line: processes are arranged according to the progressive stepsContinuous process: assembly line only the flow is continuous

LSCM v2.0 17/ 24


Closed-Looped Production Planning

MPSDemand Mgt.

PO

S. Forecasting

RCCP

Agg.Planning

L. Forecasting RRP

MRPBOM

InventoryCRP

Job Shop Purchasing

Long-Term Plan

Intermediate Plan

adopted from: Smith, S. 1989.

LSCM v2.0 18/ 24


Aggregate Planning

What: planning for family of productIdea: smoothing demands & key resources (workers, # machine)Occurred: beginning of fiscal year (unit: month of sale)Input: long-term forecasting + key resources, policy (overtime, subcontract,backlog, machine capacity)Output: production plan, capacity, inventory

LSCM v2.0 19/ 24


Red Tomato : Aggregate planningRed Tomato Tool, a US gardening tool manufacturer, usually faces with highly seasonaldemands and considers three distinguish strategies as follow:

Strategies Month Demand Month Demanda) adding seasonal worker Jan. 1600 Apr. 3800b) using subcontractor Feb. 3000 May 2200c) building inventory/backlog Mar. 3200 Jun 2200

The company sells each tool for $40 that requires 4 hour and has initial inventory andworkers in January of 1000 unit and 80 employees, respectively. A employee of RedTomato typically works 8 hours a day and earns $4 per hour during regular time.Because of labor law, no worker can work more than 20 days of regular time per monthand 10 hours of overtime per month.

Item Cost Item Cost Item CostMaterial cost $15/unit Hiring costs $300/worker RT cost $4/hourHolding cost $2/unit/month Layoff costs $500/worker OT cost $6/hourStockout cost $5/unit/month Labor required 4/unit Subcontract cost $30/unit

Determine the most suitable ’pure’ strategy and the optimal ’mixed’ strategy thatmaximizes profit if the company profile must have at least 500 units in June.

adopted from: Chopra, S. & Meindl 2016. pp. 227

LSCM v2.0 20/ 24


Red Tomato: Model and Excel

LSCM v2.0 21/ 24


Problems with aggregate planning

MethodsPrice taker: price can influence demand or negotiation (what if) → S&OPInventory of component: lack of components → MPS & MRPCapacity issues: simplified production constraints

ApplicationAggregate product: no information on exact product → communicationDelivery plan: frequency, product mixed, lot size → communication

LSCM v2.0 22/ 24


Sales and Operations Planning (S&OP)

What: manage supply or demand to smooth production given predictablevariabilityIdea: seasonal consumer demand, but fixed plant capacityManaging supply: capacity (flexible workforce & machine, subcontracting),investing (common component)Managing demand: shift or manipulate demand (using price) or promotion;may result to

Market growth:Forward buying:Stealing market share: depending on competitors

LSCM v2.0 23/ 24


Reference

[Bal07] Ronald H Ballou.The evolution and future of logistics and supply chain management.European Business Review, 19(4):332–348, 2007.

[CM07] Sunil Chopra and Peter Meindl.Supply chain management. Strategy, planning & operation.Springer, 2007.

[Goe11] Marc Goetschalckx.Supply chain engineering, volume 161.Springer, 2011.

[JC10] F Robert Jacobs and Richard B Chase.Operations and supply management: The core.McGraw-Hill Irwin New York, NY, 2010.

LSCM v2.0 24/ 24

lecture 04: forecasting and demand planningoran/classes/lscm_resources/handout/lscm... · lecture...

Documents