1 / 44

指導老師:林燦煌 博士 報告者:梁士明 200 5/4/25

A literature survey on planning and control of warehousing systems by JEROEN P. van den BERG P art II. 指導老師:林燦煌 博士 報告者:梁士明 200 5/4/25. Unit-load retrieval systems.

elita
Download Presentation

指導老師:林燦煌 博士 報告者:梁士明 200 5/4/25

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. A literature survey on planning and control of warehousing systemsby JEROEN P. van den BERGPart II 指導老師:林燦煌 博士 報告者:梁士明 2005/4/25 1

  2. Unit-load retrieval systems • Author:Goetschalckx, Ratliff[19] introduce duration of stay for individual load as alternative of COI(cube-per-order index 訂單體積指標 ,計算物品空間需求與暢銷性的關係) 2

  3. Unit-load retrieval systems Hausman et al.[3] introduce the cumulative demand function G(i)=i^s and show that a class-based policy with relatively few classes yields mean travel times that are close to those obtained by dedicated policy • i denotes a fraction of the products which contains the products with highest COI • s is a suitably chosen parameter, and s=0.139 if 20% products generates 80% of all demand 3

  4. Unit-load retrieval systems Graves et al.[2] observe furture travel time reductions when aloowing dual command cycles • Extended from Hausman et al.[3] • Analytic computations using a continuous rack and discrete computations using a rack with 30x10 locations • Determine the expected cycle time for combination of storage policies、sequencing strategies、queue length of S/R requests 4

  5. Unit-load retrieval systems Schwarz et al. verify the analytic results in [2],[3] with simulation • Closest Open Location rule is applied to select a location under randomize storage policy • Mean travel times with COL rule are comparable to analytic results which baes on arbitrary location selection 5

  6. Closest Open Location 靠近出口法則(Closest Open Location):將剛到達的商品指派到離出入口最近的空儲位上。 Refer:http://www.materialflow.org.tw/abstract/book4/chap3.html 6

  7. Chebyshev(柴比雪夫) travel • S/R machines can often move simultaneously along horizontal and vertical paths at speeds vx and vz. To reach a location (x,z) from (0,0) requires the Chebyshev measure travel time max(x/vx,z/vz). If rl is the rack length and rh the rack height Chebyshev travel require rl vx = rh vz • Rectangular building designs with I/O points at the eand of each aisle are often optimal for Chebyshev travel Refer : http://www.rh.edu/~ernesto/C_S2001/mams/notes/mams14.html 7

  8. Unit-load retrieval systems Guenov & Raeside[20] in experiments, an optimum tour with respect to Chebyshev travel may be up to 3% above the optimum for travel time with acceleration/deceleration 8

  9. Unit-load retrieval systems Hwang & Lee[21] provide a travel time measure that include acceleration/deceleration Chang et al.[22] consider various travel speeds 9

  10. Order-picking systems Organ pipe arrangement • Aisles closest to the center should carry the highest COI 10

  11. Control of warehousing operations • Batching of orders • Routing and sequencing • Dwell point positioning Focus on AS/RS 11

  12. Batching of orders • To reduce mean travel time per order • Orders in batch may not exceed the storage capacity of vehicle • Large batches give rise to response times • Orders at the far end of WH delayed • Trade-off between efficiency and urgency 12

  13. Batching of orders Two trade-offs • Static approach: select a block with most urgent orders and find a batching to minimize travel time • Dynamic approach: assign due date to orders and release orders immediately, then establish a schedule that satisfies these due date 13

  14. Batching of orders For static approach • select a seed order for batch • Expand the batch with orders that have proximity to seed order • Capacity can not be exceeded • Distinctive factor is the measure for the proximity of orders/batches 14

  15. Routing and sequencing • Unit-load retrieval operations • Order-picking operations • Carousel operations • Relocation of storage 15

  16. Unit-load retrieval operations Hausman et al.[3] only consider single command cycles 16

  17. Unit-load retrieval operations Graves et al.[2] study the effects of dual command cycles and observe travel time reductions of up to 30% 17

  18. Order-picking operations Ratliff & Rosenthal[56] present dynamic programming algorithm that solves TSP • In a parallel aisle warehouse with crossover aisles at both ends of ech aisle • Computation time is linear in the number of stops • Problem remains tractable if there are 3 crossovers per aisle 18

  19. Traveling salesman problem(TSP) • The salesman have to visit the cities in his territory exactly once and return to the start point • find the itinerary(行程) of minimum cost 19

  20. Order-picking operations Petersen[57] evaluates the performance of 5 routing heuristics in comparison with the algorithm of Ratliff & Rosenthal[56] • Best heuristics are on average 10% over optimal for various wh shapes, locations of I/O station and pick list sizes 20

  21. Order-picking operations Goetschalckx & Ratliff[58] give algorithm for order-picking in WH with non-negligible aisle width • Savings of up to 30% are possible by picking both sides of the aisle 21

  22. Order-picking operations Goetschalckx & Ratliff[59] propose a dynamic programming algorithm that the travel time of the order-picker is measured with the rectilinear metric • Determine the optimal stop position of vehicle when performing multiple picks per stop is allowed 22

  23. Order-picking operations Gudehus[1] describes band heuristic • Rack is devides into 2 horizontal bands • Vehicle visit the locations of lower band on increasing x-coordinate • Subsequentlt, visit upper band on decreasing x-coordinate 23

  24. Order-picking operations Golden & Stewart[60] • TSP for which travel times are measured by Euclidean metric has an optimal solution • Nodes on the boundary of the convex hull are visited in the same sequence 24

  25. Convex hull(凸包) • 求最小凸多邊形(convex polygon,沒有凹陷位)將平面上給定的所有點包含在裡面 Refer :http://www.geocities.com/kfzhouy/Hull.html 25

  26. Convex hull(凸包) • Akl & Toussaint[61] present a fast algorithm for finding the convex hull 26

  27. Order-picking operations Bozer et al.[64] present that use convex hull of the rack locations as an initial subtour • Locations in the interior of hull are inserted • For Chebyshev & rectilinear metric some locations can be inserted without increasing the travel time • also present an improved version of the band heuristic that blocks out a central portion of the rack 27

  28. Order-picking operations Hwang & Song[65] present a heuristic that considers the convex hull for Chebyshev travel and rectilinear hull for rectilinear travel to ensure safety of pickers • Below a predetermined height Chebyshev travel is performed • Above this height , rectilinear travel is performed 28

  29. Order-picking operations Daniels et al.[66] consider the situation where products are stored at multiple location and picked freely. It’s not acceptable because • Propagates aging of the inventory (not FIFO) • Increases storage space requirements (multiple incomplete pallets) 29

  30. Carousel operations Bartholdi and Platzman[67] present a linear time algorithm • Sequencing picks in single order • Assume time needed by robot to move between bins within the same carrier is negligible compared to the time rotating carousel to next carrier • Reduce the problem of finding shortest Hamiltonian path on a circle 30

  31. Hamiltonian path 由數學家 Euler 提出的:西洋棋的騎士能否走完一個空棋盤的六十四格,而且每格只走過一次。這條路徑,在圖論上稱為「Hamiltonian path」 ,而每個格子稱為「vertex」,每個格子能向外走出的步數稱為「該vertex的degree」。 • Refer:http://episte.math.ntu.edu.tw/java/jav_knight/ 31

  32. Carousel operations Wen and Chang[68] present 3 heuristics • Sequencing picks in single order • Time to move between bins may not be neglected • Based upon the algorithm in Bartholdi and Platzman[67] 32

  33. Carousel operations Ghosh and Wells[69], van den Berg[70] present optimal pick sequence • Multiple orders • Dynamic programming algorithm • Sequence of orders is fixed • Sequence of picks in orders is free 33

  34. Carousel operations Bartholdi and Platzman[67] present a heuristic for the problem with extra constraint • Order sequence is free • Picks within same order must be performed consecutively • Extra constraint: each order is picked along its shortest spanning interval 34

  35. Carousel operations Van den Berg[70] presents a polynomial time algorithm that solve the problem with extra constraint to optimality • At most 1.5 revolutions of the carousel above a lower bound for the problem without extra constraint • Reveal that the upper bound of one revolution presented by Bartholdi and Platzman[67] for their heuristic is incorrect 35

  36. Relocation of storage Jaikumar and Solomon[71] address the problem of relocating pallets with a high expectancy of retrieval to locations closer I/O station during off-peak hours • Assume there is sufficient time (travel time is omitted) • Present a algorithm to minimize the number of relocations 36

  37. Relocation of storage Muralidharan et al.[72] suggest randomized location assignment • Combines benefits of randomized storage (less storage space) and class-based storage (less travel time) • Respect to their turnover rate during idle periods 37

  38. Dwell point positioning Dwell point : the position the S/R machine resides when system is idle • Minimize the travel time from the dwell point to position of 1st transaction • If 1st operation is advanced, all operations within the sequence are completed earlier 38

  39. Dwell point positioning Graves et al.[2] select the point at the I/O station and Park[73] shows the optimality • If the probability of the 1st operation after idle period being a storage is at least 0.5 39

  40. Dwell point positioning Egbelu[74] presents LP-model that • Minimize the expected travel time • Minimize the maximum travel time to the 1st transaction 40

  41. Dwell point positioning Egbelu and Wu[75] use simulation to evaluate the performance of several strategies 41

  42. Dwell point positioning Hwang and Lim[76] treats this problem as a Facility Location Problem • Computational complexity is equivalent to sorting a set of numbers 42

  43. Dwell point positioning Peters et al.[77] presents an analytic model based on expressios found by Bozerand White[78] 43

  44. Question & Discussion

More Related