Presentation is loading. Please wait.

Presentation is loading. Please wait.

Operational Research & ManagementOperations Scheduling Introduction Operations Scheduling 1.Setting up the Scheduling Problem 2.Single Machine Problems.

Similar presentations


Presentation on theme: "Operational Research & ManagementOperations Scheduling Introduction Operations Scheduling 1.Setting up the Scheduling Problem 2.Single Machine Problems."— Presentation transcript:

1 Operational Research & ManagementOperations Scheduling Introduction Operations Scheduling 1.Setting up the Scheduling Problem 2.Single Machine Problems 3.Solving Scheduling Problems

2 Operational Research & ManagementOperations Scheduling2 Program

3 Operational Research & ManagementOperations Scheduling3 We are very grateful for the slides prepared by colleagues from other universities, in particular the slides of Siggi Olafsson has been used in our course extensively. Acknowledgement

4 Operational Research & ManagementOperations Scheduling Topic 1 Setting up the Scheduling Problem

5 Operational Research & ManagementOperations Scheduling5 Scheduling  Scheduling concerns optimal allocation or assignment of resources, over time, to a set of tasks or activities. – Machines M i, i=1,...,m(i th machine) – Jobs J j, j=1,...,n (j th job)  Schedule may be represented by Gantt charts.

6 Operational Research & ManagementOperations Scheduling6 Notation  Static data: – Processing time (p ij )on machine i – Release date (r j ) – Due date (d j ) – Weight (w j )  Dynamic data: – Completion time(C ij ) on machine i – Flow Time (F j = C j – r j )

7 Operational Research & ManagementOperations Scheduling7 Modeling  Three components for any machine scheduling model: – Machine configuration – Constraints and processing characteristics – Objective and performance measures  Notation:  |  |   Characteristics for  obviously present because of , are NOT mentioned.

8 Operational Research & ManagementOperations Scheduling8  : Machine Configuration  Standard machine configurations: – (1)Single-Machine models – (Pm)Parallel-Machine models – (Jm)Job Shop modelsjobs have different routes – (Fm)Flow Shop models:jobs have same order and same machines – (Om)Open Shop:routing also to be determined  Real world always more complicated: – (FJc)Flexible Job Shop:with parallel machines at each workstation – (FFc)Flexible Flow Shop:with parallel machines at each stage

9 Operational Research & ManagementOperations Scheduling9  : Constraints  (r j )Release dates  (prec)Precedence constraints  (s jk )Sequence dependent setup times  (prmp) Preemptions (resume or repeat)  (block)Storage / waiting constraints  (M j )Machine eligibility  (circ)Recirculation  Tooling / resource constraints  Personnel scheduling constraints

10 Operational Research & ManagementOperations Scheduling10  : Objectives and Performance Measures 1.Throughput (TP) and makespan (C max ) 2.Due date related objectives 3.Work-in-process (WIP), lead time (response time), finished inventory 4.(Setup Times)

11 Operational Research & ManagementOperations Scheduling11 1. Throughput and Makespan  Throughput – Defined by bottleneck machines  Makespan  Minimizing makespan tends to maximize throughput and balance load

12 Operational Research & ManagementOperations Scheduling12 2. Due Date Related Objectives  Lateness Minimize maximum lateness  Tardiness Minimize the weighted tardiness  Tardy job Minimize the number of tardy jobs

13 Operational Research & ManagementOperations Scheduling13 Due Date Penalties In practice Tardiness Late or Not Lateness

14 Operational Research & ManagementOperations Scheduling14 3. WIP and Lead Time  Work-in-Process (WIP) inventory cost  Minimizing WIP also minimizes average lead time (throughput time)  Minimizing lead time tends to minimize the average number of jobs in system  Equivalently, we can minimize sum of the completion times: or

15 Operational Research & ManagementOperations Scheduling Topic 2 Single-Machine Scheduling Problems

16 Operational Research & ManagementOperations Scheduling16 Classic Scheduling Theory  Look at a specific machine environment with a specific objective  Analyze to prove an optimal policy or to show that no simple optimal policy exists  Thousands of problems have been studied in detail with mathematical proofs! 3 Examples: single machine

17 Operational Research & ManagementOperations Scheduling17 1. Completion Time Models  Lets say we have – Single machine (1), where – the total weighted completion time should be minimized (  w j C j )  We denote this problem as

18 Operational Research & ManagementOperations Scheduling18 Optimal Solution  Theorem: Weighted Shortest Processing Time first - called the WSPT rule - is optimal for  Note: The SPT rule starts with the job that has the shortest processing time, moves on the job with the second shortest processing time, etc.  WSPT starts with job with largest

19 Operational Research & ManagementOperations Scheduling19 Proof (by contradiction) Suppose it is not true and schedule S is optimal – Then there are two adjacent jobs, say job j followed by job k such that Do a pairwise interchange to get schedule S’ jk kj

20 Operational Research & ManagementOperations Scheduling20 Proof (continued)  The weighted completion time of the two jobs under S is  The weighted completion time of the two jobs under S’ is  Now:  Contradicting that S is optimal.

21 Operational Research & ManagementOperations Scheduling21 More Completion Time Models SPT rule WSPT rule NP-hard preemptive SPT rule NP-hard polynomial algorithm

22 Operational Research & ManagementOperations Scheduling22 2. Lateness Models  Lets say we have – Single machine (1), where – the maximum cost for late jobs should be minimized (h max ) – subject to precedence constraints  We denote this problem as h j (C j ) denotes the cost for completing job j at time C j e.g. h j (C j ) = C j -d j (than h max = L max and EDD optimal)

23 Operational Research & ManagementOperations Scheduling23 Optimal Solution  Theorem: Lawler’s algorithm is optimal for  Lawler’s Backwards recursive algorithm (Minimizing Maximum Cost) : 1. Determine makespan 2. Determine job j * with smallest 3. Schedule job j * as last job in the sequence 4. Repeat same procedure with one job less (j * )  Proof by contradiction

24 Operational Research & ManagementOperations Scheduling24 More Lateness Models Lawler’s algorithm EDD rule B&B algorithm (App B2) same B&B procedure preemptive EDD rule Job j is interrupted when job k arrives with d k < d j

25 Operational Research & ManagementOperations Scheduling25 3. Tardiness Models  Lets say we have – Single machine (1), where – the number of late jobs should be minimized (  U j )  We denote this problem as

26 Operational Research & ManagementOperations Scheduling26 Optimal Solution  Theorem: Moore’s algorithm is optimal for  EDD rule with modification: 1.Three sets: J = empty; (complement of J) J C = {1..n}; (tardy jobs) J D 2.Determine job j * with smallest d j and add to J and delete from J C 3.If then remove job k from J with largest p j and add k to J D 4.Repeat step 2 and 3 until J C is empty  Proof by induction

27 Operational Research & ManagementOperations Scheduling27 More Tardiness Models Moore’s algorithm NP-hard special cases: d j = 0WSPT d j looseMS otherwise: apparent tardiness heuristic

28 Operational Research & ManagementOperations Scheduling Topic 3 Solving Scheduling Problems (Appendix C)

29 Operational Research & ManagementOperations Scheduling29 General Purpose Scheduling Procedures  Some scheduling problems are easy – Simple priority rules – Complexity: polynomial time  However, most scheduling problems are hard – Complexity: NP-hard, strongly NP-hard – Finding an optimal solution is infeasible in practice  heuristic methods

30 Operational Research & ManagementOperations Scheduling30 Different Methods  Basic Dispatching Rules  Composite Dispatching Rules  Branch and Bound  Beam Search  Simulated Annealing  Tabu Search  Genetic Algorithms Construction Methods Improvement Methods

31 Operational Research & ManagementOperations Scheduling31 Dispatching Rules  Other names: list scheduling, priority rules  Prioritize all waiting jobs – job attributes – machine attributes – current time  Whenever a machine becomes free: select the job with the highest priority  Static or dynamic

32 Operational Research & ManagementOperations Scheduling32 Release/Due Date Related  Earliest release date first (ERD) rule – variance in throughput times (flow times)  Earliest due date first (EDD) rule – maximum lateness  Minimum slack first (MS) rule – maximum lateness Current Time Processing Time Deadline

33 Operational Research & ManagementOperations Scheduling33 Processing Time Related  Longest Processing Time first (LPT) rule – balance load on parallel machines – makespan  Shortest Processing Time first (SPT) rule – sum of completion times – WIP  Weighted Shortest Processing Time first (WSPT) rule

34 Operational Research & ManagementOperations Scheduling34 Processing Time Related  Critical Path (CP) rule – precedence constraints – makespan  Largest Number of Successors (LNS) rule – precedence constraints – makespan

35 Operational Research & ManagementOperations Scheduling35 Other Dispatching Rules  Service in Random Order (SIRO) rule  Shortest Setup Time first (SST) rule – makespan and throughput  Least Flexible Job first (LFJ) rule – makespan and throughput  Shortest Queue at the Next Operation (SQNO) rule – machine idleness

36 Operational Research & ManagementOperations Scheduling36 Discussion  Very simple to implement  Optimal for special cases  Only focus on one objective  Limited use in practice  Combine several dispatching rules: Composite Dispatching Rules

37 Operational Research & ManagementOperations Scheduling Example of Composite Dispatching Rule Single Machine with Weighted Total Tardiness

38 Operational Research & ManagementOperations Scheduling38 Setup  Problem:  No efficient algorithm (NP-Hard)  Branch and bound can only solve very small problems (<30 jobs)  Are there any special cases we can solve?

39 Operational Research & ManagementOperations Scheduling39 Case 1: Tight Deadlines  Assume d j =0 Then  We know that WSPT is optimal for this problem!

40 Operational Research & ManagementOperations Scheduling40 Case 2: “Easy” Deadlines  Theorem: If the deadlines are sufficiently spread out then the MS rule is optimal (proof a bit harder)  Conclusion: The MS rule should be a good heuristic whenever deadlines are widely spread out

41 Operational Research & ManagementOperations Scheduling41 Composite Rule  Two good heuristics – Weighted Shorted Processing Time (WSPT )  Optimal with due dates zero – Minimum Slack (MS)  Optimal when due dates are “spread out” – Any real problem is somewhere in between  Combine the characteristics of these rules into one composite dispatching rule

42 Operational Research & ManagementOperations Scheduling42 Apparent Tardiness Cost (ATC)  New ranking index  When machine becomes free: – Compute index for all remaining jobs – Select job with highest value Scaling constant

43 Operational Research & ManagementOperations Scheduling43 Special Cases (Check)  If K is very large: – ATC reduces to WSPT  If K is very small and no overdue jobs: – ATC reduces to MS  If K is very small and overdue jobs: – ATC reduces to WSPT applied to overdue jobs

44 Operational Research & ManagementOperations Scheduling44 Choosing K  Value of K determined empirically  Related to the due date tightness factor and the due date range factor

45 Operational Research & ManagementOperations Scheduling45 Beam Search  Is B&B with restricted branching 1. Quick evaluation of all candidates 2. Choose the F best options (filter width) 3. Evaluate F options more thoroughly 4. Choose the B best options (beam width) for branching

46 Operational Research & ManagementOperations Scheduling46 Local Search N(S) = Neighborhood of a solution S = set containing all solutions that can be obtained by a simple modification of S Step 1: Choose a starting solution S 1 with value c(S 1 ); k = 1; Step 2: Evaluate all solutions S in N(S k ) Step 3: Choose as S k+1 the best solution only if c(S k+1 ) < c(S k ) and k = k+1; go to step 2 Otherwise stop (local optimum found)

47 Operational Research & ManagementOperations Scheduling47 Tabu Search  TS = LS with worse solutions allowed  Step 3:- Choose as S k+1 the best solution within N(S k ) unless the associated modification is on the Tabu List. - Add the modification S k -> S k+1 on the Tabu List. - Remove oldest entry of the Tabu List.

48 Operational Research & ManagementOperations Scheduling48 Graphically S k+2 SkSk S k+1 If c(S k+2 ) > c(S k+1 ) then S k+1 is a strong candidate for S k+3

49 Operational Research & ManagementOperations Scheduling49 Exercises  From chapter 3: 3.2, 3.4  From appendix C:C.1, C.4, C.6 (1 step), C.9  Single Machine Problems (see Blackboard)  Proof algorithm of Moore optimal for  Proof EDD optimal for


Download ppt "Operational Research & ManagementOperations Scheduling Introduction Operations Scheduling 1.Setting up the Scheduling Problem 2.Single Machine Problems."

Similar presentations


Ads by Google