Xiaofei Xu xiaofei@hit.edu.cn School of Computer Science and Technology

1. NICST’2013 New Information Communication Science and Technology for Sustainable Development: France-China international workshop Model and Problem Generation Tool of Project Scheduling Problem with Spatial Resources XiaofeiXu xiaofei@hit.edu.cn School of Computer Science and Technology Harbin Institute of Technology

2. Contents Introduction Concepts, features, classification of sRCPSP Problem generation tool for sRCPSP Demonstration of the tool A scheduling algorithm for sRCPSP

3. Motivation • Spatial resources usually belong to strategic/bottleneckresources of OKP and construction enterprises • High investment. • Throughput of factory highly depending on the utilization of spatial resources • Utilizing spatial resources efficiently and schedule activities in projects with limited capacity of spatial resources and other resources to shorten the make-span of projects and increase the output is of great significance! • Spatial resource constrained project scheduling problem (sRCPSP) • Ship-building process.

4. Background----Ship-Building process and spatial resources Component machining Palletized Components Collection and Distribution Purchase and inventory ship body assembling & system assembling Block assembling & system pre-assembling Assembling/testing at Dock

5. Outline What to solve? Specialty When to start activities? Where the activities are carried out? Spatial Occupied by activity group. Article structure What we do? Related research Lee, Koh, Park K, Koh S. et al . Researches on spatial resources and SRCPSP are still limited. Analyze! Modeling! Tool! Algorithm!

6. Concepts of spatial resource • Spatial resource is a kind of renewable resource with spatiality, which plays a role of holding or accommodating some objects in process. • It is required by a group of activities, rather than a single activity. Accordingly, the group of activities is called a spatial resource activity group. • Shipyard in ship-building process. • Block assembly (erection) platform in ship-building process.

7. Block assembly platform resource • The Block assembly platform resource is occupied by blocks in the ship-building process, which is one kind of bottleneck resources. Utilizing the site resource efficiently is of great significance to improvethe production efficiency.

8. Characters of Spatial Resource Divisibility Spatiality Directivity Use (σ1,σ2,σ3) to represent spatiality、divisibility and directivity,such as ( 1,d,o) represent 1-dimensional、can be divided and single-ended release space resource, it can be further simplified as 1do .

9. Classification of the task Spatial resource is occupied by activity group while not single activity. Movable activity Release activity Call activity There are three kinds of activity in the activity group. • Release the space • occupied by activity • group • Apply spatial resources • needed by activity group • Spatial resources are • movably occupied

10. Formulation of sRCPSP • The formulation of SRCPSP with objective • minimizing make-span is as following: (1)Minimizes the make-span of the project. (2)Denotes the renewable resource related constraints. (3)Forces activities to start after its immediately precedes. (4)Forces every activity to start not before zero. (5)Means that the requirement of every activity group should be satisfied. (6)Ensures that spatial resource blocks allocated to concurrent activity groups are not overlapped. (7)Denotes that activity group does not occupy spatial resource before starting or after ending. (8)Denotes the constraints of spatial resource availability.

11. A problem generation tool Based on ProGEN Incorporating spatial resources Has the full function of ProGEN A problem generation tool for sRCPSP. Code in java Enrich the tool set of project scheduling

12. Definition for Input • Base file (Input) describes the constraints for project scheduling problem. • Parameters about activity group: MinNofTaskG(MaxNofTaskG): The min(max)number of activity groups in every subproject; MinNofJobTask(MaxNofJobTask): The min(max) number of activities in every activity group; MinNofQJob(MaxNofQJob): The min(max)number of request activities in every activity group; MinNofMJob(MaxNofMJob): The min(max)number of movable activities in every activity group. • Parameters about spatial resources: NofSR：The number of spatial-resource types; SRAmin(SRAmax)：The min(max)kinds of the spatial resources of SRA type; SRAminDemand(SRAmaxDemand)：The min(max) needs of SRA type; SRARmin(SRARmax): The min(max)kinds of needs of activity group for SRA type; SRAF: The resource factor of SRA type; SRAS: The resource intensity of SRA type;

13. Definition for Output • Example file (Output) is used to describe a specific spatial resourcesproject scheduling problem instance. • AdditionalParameters (Input and output contain all of the PROGEN’s parameters respectively): • the information of activity group, • the space resources needed by every activity group, • the amount of available space resources • the direction of space resources.

14. The tool’s Generation method The generation process can be divided into four steps. Generate resource needs and available resources Generate specific parameters Generate project network Generate activity group • Generate the number of tasks, request tasks and movable tasks. • Generate activity group • The number of activities and modes. • The start time . • ……. • The work is very complex. • Generate resource's type. • Generate resource's level.

15. Demonstration of the tool • GUI of the tool.

16. Demonstration of the tool • An instance of example file.

17. Demonstration of the tool Project network with activity group has been generated.

18. A preliminary scheduling algorithm for sRCPSP When to place Where to place • The space resources restricted project scheduling algorithm. • Base on artificial bee colony ABC. • Solve the problem of regular and spatial resources scheduling. • The heuristic space scheduling algorithm. • Base on configuration space theory. • Solve the problem of optimal space layout.

19. Visual scheduling results The algorithm has a good performance in solving RCPSP with both traditional resource and spatial resource! Activity number: B-011 Duration: 9 days Start time: 2012-11-05 The occupation of each resource: [8,10,9,2,10] The utilization percent of each resource: [69%,71%,82%,56%,89%]

20. Summary sRCPSP Modeling Concepts Features Classification Formulation Summary Input and output Generation tool Generation method Base on ProGEN Demonstration Base on Configuration space theory Scheduling algorithm Consider regular and space resources restricted Click to add Title Base on Artificial bee colony ABC Click to add Title Demonstration

21. Further works • Thanks Prof. Alain Quilliotfor guidance in design algorithm for sRCPSP. • Further works: • Better model and algorithms • Domain specific sRCPSP • … • Collaborating with LIMOS on sRCPSP research

22. Q & A! Merci!Thank You!谢谢！ Xiaofei Xu xiaofei@hit.edu.cn School of Computer Science and Technology Harbin Institute of Technology