THE MULTIPLE SET-UP ORDERING COSTS IN INVENTORY PROBLEM

Similar documents
A New Solution Method for the Finite-Horizon Discrete-Time EOQ Problem

c. Explain the basic Newsvendor model. Why is it useful for SC models? e. What additional research do you believe will be helpful in this area?

IP Reference guide for integer programming formulations.

Unit 6: Sequences and Series

CENTRALIZED EFFECT ON EXPECTED COSTS IN A MULTI-LOCATION NE'VSBOY PROBLEM

Self-normalized deviation inequalities with application to t-statistic

It is always the case that unions, intersections, complements, and set differences are preserved by the inverse image of a function.

MAT1026 Calculus II Basic Convergence Tests for Series

Sequences A sequence of numbers is a function whose domain is the positive integers. We can see that the sequence

Section 5.5. Infinite Series: The Ratio Test

A sequence of numbers is a function whose domain is the positive integers. We can see that the sequence

EXAMINATIONS OF THE ROYAL STATISTICAL SOCIETY

MATHEMATICAL SCIENCES PAPER-II

2.4.2 A Theorem About Absolutely Convergent Series

Online Appendices for: Optimal Recharging Policies for Electric Vehicles

6.3 Testing Series With Positive Terms

4.1 Sigma Notation and Riemann Sums

Ma 530 Introduction to Power Series

is also known as the general term of the sequence

Convergence of random variables. (telegram style notes) P.J.C. Spreij

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence

Series III. Chapter Alternating Series

The Choquet Integral with Respect to Fuzzy-Valued Set Functions

Stat 421-SP2012 Interval Estimation Section

Integrable Functions. { f n } is called a determining sequence for f. If f is integrable with respect to, then f d does exist as a finite real number

INEQUALITIES BJORN POONEN

A Hadamard-type lower bound for symmetric diagonally dominant positive matrices

Chapter 6 Infinite Series

INFINITE SEQUENCES AND SERIES

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS

Hoggatt and King [lo] defined a complete sequence of natural numbers

Random Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A.

OPTIMAL ALGORITHMS -- SUPPLEMENTAL NOTES

ROLL CUTTING PROBLEMS UNDER STOCHASTIC DEMAND

Ma 530 Infinite Series I

Math 61CM - Solutions to homework 3

Fundamental Theorem of Algebra. Yvonne Lai March 2010

Optimal inventory policy for items with power demand, backlogging and discrete scheduling period

The picture in figure 1.1 helps us to see that the area represents the distance traveled. Figure 1: Area represents distance travelled

Machine Learning Brett Bernstein

1 Duality revisited. AM 221: Advanced Optimization Spring 2016

The multiplicative structure of finite field and a construction of LRC

The log-behavior of n p(n) and n p(n)/n

Sequences. Notation. Convergence of a Sequence

Optimization Methods MIT 2.098/6.255/ Final exam

ON POINTWISE BINOMIAL APPROXIMATION

CHAPTER 10 INFINITE SEQUENCES AND SERIES

TEACHER CERTIFICATION STUDY GUIDE

Lecture 6 Simple alternatives and the Neyman-Pearson lemma

In number theory we will generally be working with integers, though occasionally fractions and irrationals will come into play.

Estimation for Complete Data

Lesson 10: Limits and Continuity

A note on log-concave random graphs

Chapter 9: Numerical Differentiation

An Introduction to Randomized Algorithms

Mi-Hwa Ko and Tae-Sung Kim

Independence number of graphs with a prescribed number of cliques

On Random Line Segments in the Unit Square

INFINITE SEQUENCES AND SERIES

Castiel, Supernatural, Season 6, Episode 18

PAijpam.eu ON DERIVATION OF RATIONAL SOLUTIONS OF BABBAGE S FUNCTIONAL EQUATION

( ) = p and P( i = b) = q.

Section 13.3 Area and the Definite Integral

The standard deviation of the mean

LECTURE 8: ASYMPTOTICS I

Infinite Horizon Production Scheduling in Time-Varying Systems Under Stochastic Demand

Output Analysis and Run-Length Control

STAT Homework 1 - Solutions

MA131 - Analysis 1. Workbook 9 Series III

Infinite Sequences and Series

f(x)g(x) dx is an inner product on D.

4.1 SIGMA NOTATION AND RIEMANN SUMS

Axioms of Measure Theory

1+x 1 + α+x. x = 2(α x2 ) 1+x

JANE PROFESSOR WW Prob Lib1 Summer 2000

A statistical method to determine sample size to estimate characteristic value of soil parameters

Stochastic Matrices in a Finite Field

Ada Boost, Risk Bounds, Concentration Inequalities. 1 AdaBoost and Estimates of Conditional Probabilities

Math 113 Exam 3 Practice

(b) What is the probability that a particle reaches the upper boundary n before the lower boundary m?

Sieve Estimators: Consistency and Rates of Convergence

7.1 Convergence of sequences of random variables

DIVISIBILITY PROPERTIES OF GENERALIZED FIBONACCI POLYNOMIALS

Lecture 19. sup y 1,..., yn B d n

Math 113 Exam 4 Practice

Discrete Mathematics and Probability Theory Spring 2016 Rao and Walrand Note 19

Sequences, Mathematical Induction, and Recursion. CSE 2353 Discrete Computational Structures Spring 2018

Preponderantly increasing/decreasing data in regression analysis

Statistical inference: example 1. Inferential Statistics

Topic 9: Sampling Distributions of Estimators

Definitions and Theorems. where x are the decision variables. c, b, and a are constant coefficients.

CEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering

A Note on the Symmetric Powers of the Standard Representation of S n

6 Integers Modulo n. integer k can be written as k = qn + r, with q,r, 0 r b. So any integer.

Topic 9: Sampling Distributions of Estimators

Seunghee Ye Ma 8: Week 5 Oct 28

Beurling Integers: Part 2

2.2. Central limit theorem.

CS161 Handout 05 Summer 2013 July 10, 2013 Mathematical Terms and Identities

CHAPTER 5 SOME MINIMAX AND SADDLE POINT THEOREMS

Transcription:

J. Operatios Research Soc. of Japa Vol. 19. No. 1. March. 1976 1976 The Operatios Research Society of Japa OPTIMAL BATCH (8,S) POLICIES FOR THE MULTIPLE SET-UP ORDERING COSTS IN INVENTORY PROBLEM JUNICHI NAKAGAMI Ch iba Uivl' r s ity (Received April 27. 1974; Rised Se,Ptemher 19. 1975) ABSTRACT Cosider the dyamic ivetory problem whe the orderig cost fuctioyl is liear with multiple set-up costs. I geeral, a optimal ivetory policy is sesitive to the form of the orderig cost, so that the purpose of this ote is to defie a ew policy, ie, a batch (s,s) policy ad to show the sufficiet coditios uder which this policy is optimal. 1. INTRODUCTION Cosider the sigle item, periodic :~eview, stochastic ad dyamic ivetory model whe the orderig cost fuctio is liear with multiple set-up costs rather tha oe with a sigle set-up cost. This type of cost is either covex or cocave, but has a :practical meaig whe the ordered quatity i each period is delivered by trasportatio's vehicle which has certai limited capacity. I geeral, a optimal ivetory policy is sesitive to the form of the orderig cost, so that util ow some types of ivetory policies have examied ad studied for several authors. Scarf [4J proved that a (s,s) policy is optimal for a liear cost with a sigle set-up,ad this case was ivestigated i detail by Iglehart [lj, Veiott [5J et al. Recetly Porteus [3J proved that a geeralized (s,s) policy is optimal for a cocavely icreasig cost. The purpose of this ote is to discuss the Lippma [2J's model i which the orderig cost has multiple set-ups, ad show the optimality of "batch (s,s) polices". 91

92 Juichi Naka,qami K+cM c(z) ~ ~ I I K o M 2M 3M Fig.l Graph of c(z) 2.FORMULATION OF THE MODEL I this sectio we discuss the model which explicitly allows for the ucertaity i demads ad make assumptios to keep the otatio simple. Less restrictive assumptios uder which the results of this paper still valid are give i Sectio 4. Let c(z) deote the orderig cost fuctio with multiple set-up as follows: (1) c(z) = K{~} + cz for z > 0, where c ~ 0, K, M > ad {z} is the miimum iteger ot smaller tha z. We illustrate this by Fig. 1. 'l-.1he we iterpret M as the capacity of a trasportatio vehicle, K as the cost of its use ad c as the uit cost of the treated item, the c(z) is more reasoable for if vehicles of the trasportatio are trucks the orderig cost is a fuctio oly of the umber of trucks required to satisfy the order ad ot of the fractio of truck space used. (if excess space Gaot be used.) It is specifically assumed that orders are delivered immediately, shortages are backlogged ad the objective is to miimize the total expected cost attributed to orderig, holdig ad pealty for shortages over periods. The quatities demaded i each period are idepedet, idetically distributed, oegative radom variables with COmmo p.d.f. ~('). Costs to be icurred periods i the future are dis couted by the factor a, where 0 < a ~ 1. Let holdig ad shortages costs charged o edig ivetory i each period be deoted by 1('), the the oeperiod expected holdig ad shortage cost for the level y of ivetory after Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

Optimal Batch Policies i Ivetory Problem 93 orderig is deoted by We assume g(y) exists for each y. As usual, let f (x) be the miimum expeeted cost over periods as a fuctio of the :"evel x of ivetory before orderig. We have, f (x) if [ c(y-x) + g(y) + a!~ f_l(y-~).(~)d~ J y>x ( 1,2,..., fo(x) - 0 ). Let (4) The G (x) f (x) = if [ G (y) + K{Y-x} J.. cx M y~x ( = 1, 2,... ; fo(x) " 0 ). Let Y (x) deote the optimal ivetory policy i the period, ie, the optimal level of ivetory after orderig i the first of periods whe the level of ivetory before orderig is x. Y (x)-x f (x) = G (y (x)) + K{ M } -cx 3. OPTIMALITY OF BATCH (s,s) POLICIES The we have for every x. I this sectio we shall give a defiitio of batch (s,s) policy ad some sufficiet coditios uder which this policy is optimal i the fiite horizo problem. Defiitio 1. A batch (s,s) policy is a. ivetory policy defied by parameters s, S with s < S ad M (>0), su(~h that Y(x) x for x > s Y(x) mi s-x ) S, x + M{M} for x < s. We illustrate this by Fig. 2. A bateh (s,s) policy has a followig S-s ecoomic iterpretatio. I case 2, where {M} ~ 2, M is smaller tha S-s, ie, the maager has small-sized trucks for trasportatio use as compared with a satisfig level regio (s,sj. the he orders a miimum amout of the item with full-loaded trucks so as to rai:3e the ivetory level upto the Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

94 Juichi Nakagami Case 1: {8-s} ~l 1 8 s --- x s-2m 8-2M s-m 8-M s Case 2: {S-s} > 2 M Y(x) x s-3m s-2m s-m s 8 Fig. 2 Batch (s,8) policy rigio (s,8j if the iitial level is less tha s; batch policy. I case 1, 8-s where {~} = 1, the maager has large-sized trucks, the he caot order the item with full-loaded trucks so as to raise the ivetory level ito the regio (s,8j. 80 that he raise the ivetory level ot to exceed 8 with trucks which are ot always full-loaded if the iitial level is less tha s; batch policy + (s,8) policy. to the well-kow (s,8) policy. If M is sufficietly large this policy is idetical Theorem 1. If G (x) is covex ad bouded below, the a batch (s,8) policy is optimal i period. Proof. From otatioal coveiece, we abbreviate the subscript. By our assumptio o G(x), there exist the smallest real umbers sad S with s < 8 (which may be :00.), such that Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

Optimal Batch Policies i Ivetory Problem 95 (6) G(S) ~ G(x) for all x, [G(x) - G(x+m)J < K for all x ~ s. S-s Case 1, {M} = 1 : (i) We have, for ay x ad y with s ~ x < y G(y) + K{y~x} > G(y) + K > G(x) [by (7)J. Thus it follows that (8) Y(x) = x o [s,oo) (ii) We have, for ay x ad y with S-H < x < s, x < y, Hece we get G(y) + K{y~x} ~ G(y) + K ~ G(S) + K [by (6)J. y ( x ) = S = mi [ S, x+m im s-x} J o [S-M,s). For ay x ad y with S-2M < x < S-M, x < Y ~ x+m, we have G(y) + K ~ G(x+M) + K (eluality holds iff y = x+m). Thus it is easily show by iductio that for ay x with S-(d+l)M ~ x < S-dM, d = 1, 2,..., (10) mi [ G(y) + K{y~x} J ~ G(x+dM) + dk. x+dm~y~x (iii) Therefore we have for ay x with S-(d+l)M ~ x < S-dM, d=l,2,..., mi [ G(y) + K{Y~x} J = mi [G(y) + K{y-~-dM} + dk J y~x y>x+dm [by (lo)j G(S) + (d+l)k if S-M < x+dm < s = [by (9)J G(x+dM) + dk if s ~ x+dm < S [by (8)J. Hece, if S-M < x+dm < s the {s~x} ( ) i s-x} d+l, so that Y x = S ~ x + M M ' ad if s < x+dm < S the {s~x} = d. so that Y(x) = x + M{s~x} < S. The, Y(x) = mi [ S. x + M{s~x} J o (_00, s). Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

96 Jrmichi Nakagami Case 2, {S-s} > 2 M = Similarly to Case 1, we ca prove that Y(x) = x o [s,oo), y(x) = x + M{s~x} < So (_oo,s), ard Y(x) = mi [S, x + M{s~x} ] o (_oo,s), ad hece a batch (s,s) policy is optimal i period, which completes the proof of the theorem. It is clear that ar. ivetory policy which is optimal to use i the first of perjods is also optimal to use whe there are periods left i a m periods problem. We will therefore oly be iterested i derivig coditios which isure that G (.) is covex for every ad hece that a batch (s,s) policy is optimal for the first of periods. Defiitio ;:0. A desjty ~(') is called M-idifferet, if it satisfies for ~ f,; < M. If we divide the demaded quatities by M, the M-idifferet desities give o iformatio ab01;t vrhich quatities left are likely to occur, that is, such desities are idifferet (igorat) of the remaiig quatities. M For example, let, for > 2 ' v > 0, the a M-idifferet desity ~(f,;) v Of M ~ M 1-2 ~ ~ < + 2 otherwise is give by, where v(o) is a geeralized probability desity defied o [~,oo) ad!dv() 1 =M"' Theorem 2. If g(o) is covex ad bouded below ad ~(o) is M-idifferet, the a batch (s,8) policy is optimal for ay fiite horizo problem. Proof. Here we will show by iductio that G (.) is covex for all. For = 1, Gl(x) = g(x) + cx is covex. Assume that Gk(o) is covex. The by Theorem 1 there exist two levels sk' Sk with sk < Sk such that (11) + dk - cx 1, 2,..., Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

Optimal Batch Policies i Ivetorory Problem 97 Hece we have (12) if x < mi (Sk,sk+M) - s*, ad fk(x) - fk(x-dm) is odecreasig i x. Now we exami G k + 1 ( ) defied by (4). We have from cotiuity ad piecewise covexity of f k ( ) give by (11) For x < s* we have (13) = Ei:O f~-f~(x-i;:-im) <j>(i;:+im) dl;: Ei:O f~-f~(x-i;:) <j>(i;:+im) 0_1;: [by (12)J = - f'(x-i;:) dl;: 1 fm- M 0 k [by the 1<. )'s M-idifferece] = 1- r*g' (I::) dl;: - c = costat (= -C) M sk x-s* Ad for x > s* let d* = {-M---}, we have [by (11) J. f x o - s * ( [f~(x-i;:) - f~ X-I;:-Cl*M)J<j>(I;:)dl;: + f~ f~(x-i;:-d*m)<j>(i;:)dl;: [by (12)J. The the secod term is -C, it is therefore sufficiet to show that the itegrat of the first term is o-egative ad o-decreasig i x. x -s* x -s* 1 xl < x 2 ' let di {-M---}, d:2 = {-t-} respectively, the di ~ d:2. For [by (12)J f' (x -1;:-d*M) k 1 2 < f k '(x 2-1;:) - f'(x -1;:-d*M) k 2 2 [by (12)J. Thus the proof of Theorem 2 is complete. Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

98 Juichi Nakagam 4. CONCLUDING REMARK I this ote we have show that a batch (s,s) policy is optimal i a stadard ivetory model whe the orderig cost is liear with multiple setups. The most crucial assumptio is that the demad pdf is to be M-idifferet. Our results will also valid for the ostatioary case; c (z) = c z ~ (.) is M-idifferet, the we have If G (.) is covex ad bouded below ad f (x) = if [ c (y-x) + g(y) + a J~ f l(y-~)~ (~)d~ J - y;,x where fo(x) = lower-bouded covex fuctio, for = 1,2,... Ufortuately covexity of f (x)'s will ot be obtaied for geeral demad pdf ad hece ay batch (s,s) policy may ot be optimal. However, i may cases the demad pdf is ot determied precisely, ad a M-idifferet pdf gives a good approximatio to the demad desity by exploitig a least-square method. For istace, whe the actual demad pdf is ~(.), the M-idifferet pdf ~(.) is give by i which v*() is the miimizig v() of the itegral J (~(~) Although it is a rough approximatio it is useful to put v*(') ~(~))2 d~ ~(. )/M Ackowledgemet The author expresses his thaks to Professor M. Sakaguchi of Osaka Uiversity for his helpfull discussios ad suggestios. Refereces [IJ Iglehart, D.L., "Optimality of (s,s) Policies i the Ifiite Horizo Ivetory Problem," Maagemet Sciece, 9 (1963), 259-267. [2J Lippma, S.A., "Optimal Ivetory Policy with Multiple Set-Up Costs," Maagemet Sciece, 16 (1969), 118-138. [3J Porteus, E.L., "O the Optimality of Geeralized (s,s) Policies," Maagemet Sciece, 17 (1971), 411-426. [4J Scarf, H., "The Optimality of (S,s) Policies i a Dyamic Ivetory Problem," i MathematicaZ Methcx1s i the SociaZ Scieces, K.J.Arrow, S.Karli, ad P.Suppes (eds.), Staford Uiversity Press, Staford,1960. [5J Veiott, A.F.,Jr., "O the Optimality of (s,s) Ivetory Policies: New Coditios ad a New Proof," J.SIAM AppZ. Math., 14 (1966),1067-1083. Copyright by ORSJ. Uauthorized reproductio of this article is prohibited.

2024 © sciencedocbox.com Privacy Policy | Terms of Service | Feedback