Anda di halaman 1dari 29

MBA 664

Analysis of Bullwhip effect in multi-product, multi-stage supply chain systems a simulation approach

Presented By: Pranav Agrawal (10505) Ankush Panwar (10118)

Contents
Introduction Bullwhip Effect Simulation model Analysis of Bullwhip Effect Case Study Discussion Conclusion

Introduction
Main objectives of a company are: o To synchronize the upstream and downstream flow of goods and services o To respond to uncertainties in customer demand without creating costly inventories Even if uncertainties are known, upstream and downstream do not always synchronize. Manufacturers are often surprised by the degree of variability of distributors orders Hence, a decision support tool is required to reduce and quantify this variability.

Bullwhip Effect

What this paper aims at?


Proposing a unique approach towards developing a flexible model to quantify and reduce the bullwhip effect A multi-product, multi stage supply chain is modeled using basic inventory policies, MRP and JIT approach. A simulation model is finally developed based on System dynamics modeling (SDM) and adaptive network based fuzzy inference system (ANFIS)

Important Keywords
SDM (System Dynamics Modeling) o A powerful simulation approach for studying and managing complex feedback systems ANFIS (Adaptive network based fuzzy inference system) o Implemented in SDM to increase its reliability for modeling soft variables which exist in supply chain environment MRP (Material Requirement planning) o A production planning and inventory control system used to manage manufacturing processes JIT (Just in Time) o An approach where materials, parts, and other goods are ordered only in quantities required to meet immediate production needs.

The simulation model


Supply chain agents + Departments that make ordering decisions = Supply Chain (SC) Units Each SC Unit contributes to Bullwhip effect Eg. Two SC units in the manufacturer in the figure given below

Supplier

Manufacturer

Retailer

Material (SC Unit)

Material (SC Unit)

Product (SC Unit)

Product (SC Unit)

The SC units in a single product, multi-stage supply chain

Proposed model
Supplier Manufacturer Retailer

Inventory Model

Linkage Model

Inventory Model

Linkage Model

Inventory Model

Linkage Model

Inventory Model

Contribution Model

Contribution Model

Contribution Model

Contribution Model

Layer 1: Supply Chain Network Layer Layer 2: Supply Chain contribution Layer Model structure of the single product, multi-stage supply chain

Simulation Model

SC Network layer

Contribution Layer

Inventory Model

Linkage Model

Basic ordering Policy

MRP

JIT

SC Network Layer
This layer is designed to create the specific network of a supply chain by linking SC units It is subdivided into two generic models: 1. Inventory Model 2. Linkage Model

Inventory Model
It is developed for modeling an inventory of a SC unit 3 main inputs in the model are No. of inventory, defect rates and decision when shortage of stock

Fig: The basic structure of the inventory model

Linkage Model
It is developed to model information and production flows between SC unit Two types of Product and Information Flows: 1. Distribution Linkage Model 2. Product Linkage Model

1. Distribution linkage Model


Model Information and product flows between inventories of SC units which distribute items without transformation 3 input variables which users have to assign: a) Order Lead Time: Period of time between customers order and order shipped to Customers inventory b) Delivery Lead Time: Period of time between order shipped from supplier and order arrives in the customers inventory c) Allocation Policy : It defines the ratio of product which an SC unit ships to its customers during supply shortage

2.Production Linkage Model


Model information and product flows between inventories of SC units which has product transformation or production 4 input variables: a) Order Lead Time b) Production Lead Time : Time period since materials are ready for production until order shipped to product inventory c) Allocation Policy for Production d) Bill Of Material (BOM) : No. of units material used for producing one unit product

Results From Linkage Models


Number of product shipped to a customer and the number of orders received from supplier against time scales

SC Contribution Layer
Model an ordering process to replenish a stock for each SC unit 3 sub-models in order to model 3 categories of ordering processes : Based on 1. Contribution model 1: Basic Inventory Policies 2. Contribution model 2: MRP 3. Contribution model 3: JIT

Contribution Model 1
Ordering process on the basis of basic inventory policies Firstly users assign ordering policy to identify number of orders released Six general policies are continuous inventory system, periodic order-up-to level inventory, min-max, scheduled ordering, base stock system and special ordering policies Secondly, future demand : Important variable to determine sufficient capacity using forecasting technique Simple moving average , exponential smoothing technique are common

Contribution Model 2
Ordering process based on MRP system Major input to this generic model for creating MRP Schedule are: 1. No. of time periods covered in MRP schedule 2. Equation to calculate gross requirement 3. Minimum inventory level to protect shortage 4. Lot sizing policy: To determine timing and size of an order quantity.

Contribution Model 3
Ordering process for a SC unit based on JIT approach Kanban Cards: Released after products shipped to customers in order to replenish it by placing same amount of order to the supplier Sometimes wait at a heijunka post to divide the order at into small increments for load smoothing Types of Kanban card accumulation before passing to the previous station (ordering policy) : 1. Lot-for-Lot policy: No accumulation and all cards passed to previous station for filling 2. Pattern policy: Accumulation at pattern post until reaching the specific production period 3. Lot making policy: Accumulation until quantity reaches a specific amount

Formula for calculating Bullwhip effect

Implement ANFIS in simulation model


SDM generates error when input parameters contain soft variables Soft variables (hard to measure) include mental data, human judgement, confidence levels etc. ANFIS implemented to increase the accuracy to model the soft variables ANFIS is a new generation of artificial intelligence tool which integrated some advantages between fuzzy logic and neural network techniques

Bullwhip effect Analysis


After creating the model, the users can reduce the bullwhip effect by following two steps: o Design the initial solution - Solution determined by improving the structure of the SC network, level of information sharing, operational efficiency of SC units and contributions of SC units o Match contributions of SC units along the supply chain - Adjusting contribution parameters of SC units manually or using the VENSIM optimizer to find optimal set of contributions

Case Study
Simulation Model
Supply chain of a beverage company (SSC Pty Ltd) was selected Study presents two-product, three-stage supply chain

Simulation model (Cont.)


Main products are two carbonated soft drinks SSC uses an MRP principle to control production planning and number of material orders released to its supplier Vendor managed inventory (VMI) approach is used to design the timing and the size of products shipped to their distributors SSC generates when the product stock is low The linguistic data low is a soft variable and is modeled using ANFIS in SDM 80% of the collected data was used to search optimized fuzzy interference systems Rest of the data was used to measure the accuracy of ANFIS

Case Study
Model Validation
To validate, the bullwhip effect of SC units was collected for a period of 90 days Results show that simulation results were consistent with actual data The error of bullwhip effect estimation varies between 0 and 9%

The bullwhip effect level of a beverage supply chain

Case Study
Bullwhip effect analysis
Using VMI strategy, bullwhip effect levels at manufacturer agent are lower than the distribution agent o VMI eliminates one layer of decision making o Eliminates delay and error in information flow at manufacturer

The reduction of bullwhip effect through the supply chain

Bullwhip effect analysis (Cont.)


Design the initial solution for reducing the bullwhip effect o After interviewing the agents in the supply chain, the structure of a SC network, the level of information sharing and the operational efficiency of each SC unit cant be changed Match contributions of SC units along the supply chain - VENSIM optimizer was used to search the optimal set of contributions by minimizing the total bullwhip effect 1. The manufacturer generates orders to fulfill distributors stocks 2. The manufacturer generates orders to fulfill its products and material stocks

Discussion

Anda mungkin juga menyukai