SMA 6304/MIT2.853/MIT2.854 Manufacturing Systems Lecture 19-20: Single-part-type, multiple stage systems Lecturer: Stanley B. Gershwin

Slides:



Advertisements
Similar presentations
M. Colledani, T. Tolio Dipartimento di Meccanica
Advertisements

Time to failure Probability, Survival,  the Hazard rate, and the Conditional Failure Probability.
Algorithm Analysis Input size Time I1 T1 I2 T2 …
Part 3 Probabilistic Decision Models
Statistical Inventory control models I
Disk Arrays COEN 180. Large Storage Systems Collection of disks to store large amount of data. Performance advantage: Each drive can satisfy only so many.
Capacity Planning and Queuing Models
Chapter 4 Inference About Process Quality
Statistical Analysis SC504/HS927 Spring Term 2008
Short-term Capacity Optimization
Introduction to Queuing Theory
Multiple Regression and Model Building
1 Chapter 7 Performance analysis of failure- prone production lines Learning objectives : Understanding the mathematical models of production lines Understanding.
Introduction into Simulation Basic Simulation Modeling.
1 Slide Process Analysis Fundamentals MGT Slide Process Definition  A process is a collection of operations connected by a flow of transactions.
Cost Behavior and Cost-Volume-Profit Analysis
Reliability Engineering (Rekayasa Keandalan)
 These 100 seniors make up one possible sample. All seniors in Howard County make up the population.  The sample mean ( ) is and the sample standard.
Experimental Design, Response Surface Analysis, and Optimization
McGraw-Hill Ryerson Copyright © 2011 McGraw-Hill Ryerson Limited. Adapted by Peter Au, George Brown College.
Central Limit Theorem.
Queuing Models Basic Concepts
INTEGRALS Areas and Distances INTEGRALS In this section, we will learn that: We get the same special type of limit in trying to find the area under.
8 TECHNIQUES OF INTEGRATION. In defining a definite integral, we dealt with a function f defined on a finite interval [a, b] and we assumed that f does.
Probability Distributions Random Variables: Finite and Continuous Distribution Functions Expected value April 3 – 10, 2003.
Chapter 5 Continuous Random Variables and Probability Distributions
Copyright © Cengage Learning. All rights reserved. 5 Integrals.
BCOR 1020 Business Statistics Lecture 11 – February 21, 2008.
BCOR 1020 Business Statistics
QUEUING MODELS Queuing theory is the analysis of waiting lines It can be used to: –Determine the # checkout stands to have open at a store –Determine the.
APPLICATIONS OF DIFFERENTIATION 4. In Sections 2.2 and 2.4, we investigated infinite limits and vertical asymptotes.  There, we let x approach a number.
WINTER 2012IE 368. FACILITY DESIGN AND OPERATIONS MANAGEMENT 1 IE 368: FACILITY DESIGN AND OPERATIONS MANAGEMENT Lecture Notes #3 Production System Design.
Introduction Cost-volume-profit (CVP) analysis focuses on the following factors: The prices of products or services The volume of products or services.
Chapter 13: Inference in Regression
Introduction to Management Science
Dr. Cesar Malave Texas A & M University
Copyright © Cengage Learning. All rights reserved. 10 Inferences Involving Two Populations.
14 Elements of Nonparametric Statistics
Copyright © Cengage Learning. All rights reserved. 3 Applications of Differentiation.
Physics 361 Principles of Modern Physics Lecture 8.
1 Chapter 5 Flow Lines Types Issues in Design and Operation Models of Asynchronous Lines –Infinite or Finite Buffers Models of Synchronous (Indexing) Lines.
ESTIMATES AND SAMPLE SIZES
Estimates and Sample Sizes Lecture – 7.4
LIMITS AND DERIVATIVES 2. In Sections 2.2 and 2.4, we investigated infinite limits and vertical asymptotes.  There, we let x approach a number.  The.
NETE4631:Capacity Planning (2)- Lecture 10 Suronapee Phoomvuthisarn, Ph.D. /
Queuing Theory Basic properties, Markovian models, Networks of queues, General service time distributions, Finite source models, Multiserver queues Chapter.
1 Systems Analysis Methods Dr. Jerrell T. Stracener, SAE Fellow SMU EMIS 5300/7300 NTU SY-521-N NTU SY-521-N SMU EMIS 5300/7300 Queuing Modeling and Analysis.
Integrals  In Chapter 2, we used the tangent and velocity problems to introduce the derivative—the central idea in differential calculus.  In much the.
Copyright © Cengage Learning. All rights reserved. 4 Integrals.
Chapter 16 Capacity Planning and Queuing Models
Copyright © Cengage Learning. All rights reserved. 2 Limits and Derivatives.
An Optimal Design of the M/M/C/K Queue for Call Centers
1 OUTPUT ANALYSIS FOR SIMULATIONS. 2 Introduction Analysis of One System Terminating vs. Steady-State Simulations Analysis of Terminating Simulations.
Copyright © 2014 by McGraw-Hill Higher Education. All rights reserved. Essentials of Business Statistics: Communicating with Numbers By Sanjiv Jaggia and.
Approximate Three-Stage Model: Active Learning – Module 3 Dr. Cesar Malave Texas A & M University.
© 2014 Carl Lund, all rights reserved A First Course on Kinetics and Reaction Engineering Class 25.
1 1 Slide © 2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole.
NETE4631: Network Information System Capacity Planning (2) Suronapee Phoomvuthisarn, Ph.D. /
1 Introduction to Statistics − Day 4 Glen Cowan Lecture 1 Probability Random variables, probability densities, etc. Lecture 2 Brief catalogue of probability.
1 1 Slide Chapter 12 Waiting Line Models n The Structure of a Waiting Line System n Queuing Systems n Queuing System Input Characteristics n Queuing System.
1 Chapter 7 Performance analysis of failure- prone production lines Learning objectives : Understanding the mathematical models of production lines Understanding.
SIMULATION EXAMPLES. Monte-Carlo (Static) Simulation Estimating profit on a sale promotion Estimating profit on a sale promotion Estimating profit on.
Sampling Distributions: Suppose I randomly select 100 seniors in Anne Arundel County and record each one’s GPA
Copyright © Cengage Learning. All rights reserved. 9.4 Models for Population Growth.
© 2010 Pearson Prentice Hall. All rights reserved Chapter Hypothesis Tests Regarding a Parameter 10.
Copyright © Cengage Learning. All rights reserved. 4 Integrals.
Reliability Engineering
Discrete-time Markov chain (DTMC) State space distribution
Topics to be Discussed Steady State and Transient Response.
Presentation transcript:

SMA 6304/MIT2.853/MIT2.854 Manufacturing Systems Lecture 19-20: Single-part-type, multiple stage systems Lecturer: Stanley B. Gershwin 2003 Stanley B. Gershwin

Flow Line also known as a Production or Transfer Line Machine Buffer Machine are unreliable Buffers are finite 2003 Stanley B. Gershwin

Flow Line Production output from a simulation of a transfer line Stanley B. Gershwin week Weekly Production

Single Reliable Machine 2003 Stanley B. Gershwin If the machine is perfectly reliable, and its average operation time is τ, then its maximum production rate is 1/ τ. Note: Sometimes cycle time is used instead of operation time, but BEWARE: cycle time has two meanings! The other meaning is the time a part spends in a system. If the system is a single, reliable machine, the two meanings are the same.

Single Reliable Machine ODFs 2003 Stanley B. Gershwin Operation-Dependent Failures A machine can only fail while it is working. Note: This is the usual assumption. IMPORTANT! MTTF must be measured in working time!

Single Reliable Machine Production rate 2003 Stanley B. Gershwin If the machine is unreliable, and then its maximum production rate is its average operation time is τ its mean time to fail is MTTF its mean time to repair is MTTR

Single Reliable Machine Production rate Proof 2003 Stanley B. Gershwin All rights reserved Average production rate, while machine is up, is 1/τ. Average production during an up period is MTTF/τ Average duration of an up period is MTTF. Average duration of up-down period : MTTF+ MTTR. Average production during up-down period : MTTF/τ Therefore, average production rate is (MTTF/τ) / (MTTF+MTTR)

Single Reliable Machine Geometric up- and Down-Times 2003 Stanley B. Gershwin Assumptions: Operation time is constant (τ). Failure and repair times are geometrically distributed. Let p be the probability that a machine fails during any given operation. Then p=τ/MTTF.

Single Reliable Machine Geometric up- and Down-Times 2003 Stanley B. Gershwin (Sometimes we forget to say “average”) Then the average production rate of M is Let γ be the probability that M gets repaired in during any operation time when it is down. Then

Single Reliable Machine Production rates 2003 Stanley B. Gershwin when it is up (short-term capacity), When it is down (short-term capacity), on the average (long-term capacity) So far, the machine really has three production rates:

Infinite-Buffer Line 2003 Stanley B. Gershwin Assumptions: A machine is not idle if it is not starved. The first machine never starved

Infinite-Buffer Line 2003 Stanley B. Gershwin The production rate of the line is the production rate of the slowest machine in the line- called bottleneck. Slowest means least average production rate, where average production rate is calculates from one of the previous formulas.

Infinite-Buffer Line 2003 Stanley B. Gershwin Production rate is therefore and Is the bottleneck.

Infinite-Buffer Line 2003 Stanley B. Gershwin The system is not in steady state. An infinite amount of inventory accumulates in the buffer upstream of the bottleneck. A finite amount of inventory appears downstream of the bottleneck.

Infinite-Buffer Line 2003 Stanley B. Gershwin All rights reserved

Infinite-Buffer Line 2003 Stanley B. Gershwin The second bottleneck is the slowest machine upstream of the bottleneck. An infinite amount of inventory accumulates just upstream of it. A finite amount of inventory appears between the second bottleneck and the machine upstream of the first bottleneck. Et cetera.

Infinite-Buffer Line 2003 Stanley B. Gershwin All rights reserved A 10-machine line with bottlenecks at Machines 5 and 10.

Infinite-Buffer Line 2003 Stanley B. Gershwin Question: ◎ What are the slopes (roughly!) of the two indicated graphs?

Stanley B. Gershwin Infinite-Buffer Line Question: ◎ If we want to increase production rate, which machine should we improve? ◎ What would happen to production rate if we improved any other machine?

Stanley B. Gershwin Zero-Buffer Line ◎ If any one machine fails, or takes a very long time to do an operation, all the other machines must wait. ◎ Therefore the production rate is usually less- possibly much less – than the slowest machine.

Stanley B. Gershwin Zero-Buffer Line ◎ Example: Constant, unequal operation times, perfectly reliable machines. ◎ The operation time of the line is equal to the operation time of the slowest machine, so the production rate of the line is equal to that of the slowest machine.

Stanley B. Gershwin Zero-Buffer Line Constant, equal operation times, unreliable machines ◎ Assumption: Failure and repair times are geometrically distributed. ◎ Define p i = τ/MTTF i =probability of failure during an operation. ◎ Define r i =τ/MTTR i probability to repair during an interval of length τwhen the machine is down.

Stanley B. Gershwin Zero-Buffer Line Constant, equal operation times, unreliable machines Buzacott`s Zero-Buffer Line Formula: Let K be the number of machines in the line. Then

Stanley B. Gershwin Constant, equal operation times, unreliable machines Zero-Buffer Line ◎ Same as the earlier formula (page6, page9) when K=1. The isolated production rate of a single machine M i is

Stanley B. Gershwin Zero-Buffer Line Proof of formula ◎ let τ(the operation time) be the time unit. ◎ Assumption: At most, one machine can be down. ◎ Consider a long time interval of length T τ during which Machine M i fails m i times (i= 1, …, k) ◎ Without failures, the line would produce T parts. All up Some machine down

Stanley B. Gershwin Proof of formula Zero-Buffer Line ◎ The average repair time of M i is τ/τ i each time it fail, so the total system down time is close to where D is the number of operation times in which a machine is down.

Stanley B. Gershwin Zero-Buffer Line Proof of formula The total up time is approximately where U is the number of operation times in which all machines are up.

Stanley B. Gershwin Zero-Buffer Line Proof of formula Note that, approximately, because M i can only fail while it is operational. Since the system produces one part per time unit while it is working, it produces U parts during the interval of length T τ.

Stanley B. Gershwin Zero-Buffer Line Proof of formula Thus, or,

Stanley B. Gershwin Zero-Buffer Line Proof of formula and Note that P is a function of the ratio p i /r i and not p i or r i separately. The same statement is true for the infinite-buffer line. However, the same statement is not true for a line with finite, non-zero buffers

Stanley B. Gershwin Zero-Buffer Line Proof of formula Questions: ◎ If we want to increase production rate, which machine should we improve? ◎ What would happen to production rate if we improved any other machine.

Stanley B. Gershwin Zero-Buffer Line P as a function of p i All machines are the same except M i. As p i increases, the production rate decreases.

Stanley B. Gershwin P as a function of p i Zero-Buffer Line All machines are the same. As the line gets longer, the production rate decreases.

Stanley B. Gershwin Finite-Buffer Line Difficulty: * No simple formula for calculation production rate or inventory levels. Solution: * Simulation * Analytical approximation

Stanley B. Gershwin Two Machine, Finite-Buffer Line Exact solution is available to Markov process model. Discrete time-discrete state Markov process:

Stanley B. Gershwin Two Machine, Finite-Buffer Line Here,where is the number of parts in the buffer; Is the repair state of means the machine is up or operational; means the machine is down or under repair.

Stanley B. Gershwin Two Machine, Finite-Buffer Line Several models available: ◎ Deterministic processing time, or Buzacott model: deterministic processing time, geometric failure and repair times; discrete state, discrete time.

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line

Stanley B. Gershwin Two Machine, Finite-Buffer Line ◎ Exponential processing time: exponential processing, failure, and repair time; discrete state, continuous time. ◎ Continuous material, or fluid: deterministic processing, exponential failure and repair time; mixed state, continuous time.

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Deterministic Processing Time

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Deterministic Processing Time Discussion: ◎ Why are the curves increasing? ◎ Why do they reach an asymptote? ◎ What is P when N=0 ? ◎ What is the limit of P as N→∞ ? ◎ Why are the curves with smaller r 1 lower?

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Deterministic Processing Time Discussion: ◎ Why are the curves increasing? ◎ Why different asymptote? ◎ What is n when N=0 ? ◎ What is the limit of n as N→∞ ? ◎ Why are the curves with smaller r 1 lower?

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Deterministic Processing Time ◎ What can you say about the optimal buffer size? ◎ How should it be related to r i, p i ?

Stanley B. Gershwin Two Machine, Finite-Buffer Line Should we prefer short, frequent, disruptions or long, infrequent, disruptions? And p 1 vary together and Answer: evidently, short, frequent failures. Why ?

Stanley B. Gershwin Two Machine, Finite-Buffer Line Questions: ◎ If we want to increase production rate, which machine should we improve? ◎ What would happen to production rate if we improved any other machine.

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Production rate vs. storage spac Improvements to non-bottleneck machine. Machine 1 more improved Machine 1 improved Identical machines

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Avg. inventory vs. storage spac ◎ Inventory increases as the (non-bottleneck) upstream machine is improved and as the buffer space is increased. ◎ If the downstream machine were improved, the inventory would be less and it would increase much less as the space increases

Stanley B. Gershwin Two Machine, Finite-Buffer Line Other models Exponential – discrete material, continuous time the probability that completes an operation in Down, in fails during an is repaired, while it is

Stanley B. Gershwin Two Machine, Finite-Buffer Line Other models Exponential – continuous material, continuous time the amount of material thatProcesses, While it is up, in the probability that fails, while it is up, in Is repaired, while it is down, in

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Other models Explain the shapes of the graphs.

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Other models Explain the shapes of the graphs.

Stanley B. Gershwin All rights reserved Two Machine, Finite-Buffer Line Other models

Stanley B. Gershwin Long Lines ◎ Difficulty: * No simple formula for calculating production rate or inventory levels. * State space is too large for exact numerical solution. * If all buffer sizes are N and the length of the line is k, the number of states is s =2 k (N+1) k-1. * if N = 10 and k=20, s = 6.41×10 25 * Decomposition seems to work successfully.

Stanley B. Gershwin Long Lines Decomposition

Stanley B. Gershwin Long Lines Decomposition ◎ Consider an observer in buffer B i. * Imagine the material flow that the observer sees entering and leaving the buffer. ◎ We construct a two-machine line (ie, we find r 1, p 1, r 2, p 2, and N) such that an observer in its buffer will see almost the same thing. ◎ The parameters are chosen as functions of the behaviors of the other two-machine lines.

Stanley B. Gershwin Long Lines Examples There-machine line-production rate.

Stanley B. Gershwin Long Lines Examples There-machine line-total average inventory

Stanley B. Gershwin Long Lines Examples Distribution of material in a line with identical machines and buffers. Explain the shape.

Stanley B. Gershwin Long Lines Examples Effect of a bottleneck. Identical machines and buffers, except for M 10.

Stanley B. Gershwin All rights reserved Long Lines Examples continuous material model. Eight-machine, seven-buffer line. For each machine, For each buffer (except Buffer 6), N=30.

Stanley B. Gershwin All rights reserved Long Lines Examples Which n i are decreasing and which are increas- ing? Why ?

Stanley B. Gershwin Long Lines Examples 2 buffers equally sized. 8 buffers equally sized; and Which has a higher production rate? 9-Machine line with two buffering options :

Stanley B. Gershwin All rights reserved Long Lines Examples Continuous model; all machines have R=.019, What are the asymptotes? Is 8 buffers always faster ? Total buffer Space

Stanley B. Gershwin All rights reserved Long Lines Examples Is 8 buffers always faster ? Perhaps not, but difference is not significant in systems with very small buffers.

Stanley B. Gershwin Long Lines Optimal buffer space distribution. ◎ Design the buffers for a 20-machine production line. ◎ The machine have been selected, and the only decision remaining is the amount of space to allocate for in- process inventory. ◎ The goal is to determine the smallest amount of in- process inventory space so that the line meets a production rate target.

Stanley B. Gershwin Long Lines Optimal buffer space distribution. The common operation time is one operation per minute. The target production rate is.88 parts per minute.

Stanley B. Gershwin Long Lines Optimal buffer space distribution. Case 1 MTTF=200 minutes and MTTR =10.5 minutes for all machines (P=.95 parts per minute)

Stanley B. Gershwin Long Lines Optimal buffer space distribution. Case 1 MTTF=200 minutes and MTTR =10.5 minutes for all machines (P=.95 parts per minute) Case 2 Like Case 1 except Machine 5. For Machine 5, MTTF=100 and MTTR =10.5 minutes (P=.905 parts per minute)

Stanley B. Gershwin Long Lines Optimal buffer space distribution. Case 1 MTTF=200 minutes and MTTR =10.5 minutes for all machines (P=.95 parts per minute) Case 2 Like Case 1 except Machine 5. For Machine 5, MTTF=100 and MTTR =10.5 minutes (P=.905 parts per minute) Case 3 Like Case 1 except Machine 5. For Machine 5, MTTF=200 and MTTR =21 minutes (P=.905 parts per minute)

Stanley B. Gershwin Long Lines Optimal buffer space distribution. Are buffers really needed? Yes. How were these numbers calculated?

Stanley B. Gershwin All rights reserved Long Lines Optimal buffer space distribution. solution Line Space

Stanley B. Gershwin Long Lines Optimal buffer space distribution. Observation from studying buffer space allocation problems: * Buffer space is needed most where buffer level variability is greatest

Stanley B. Gershwin Long Lines Profit as a function of buffer sizes There-machine, continuous material line.