Integration of scheduling and multiple process plans

1
Integration of scheduling and multiple process
plans

• Factory Automation Lab.
• 3/18/1999
• Yang-Cha Chang

2
Contents

• Introduction
• IPPM (Integrated Process Planning Model)
• Process Net Model (Multiple process plan
  generation approach)
• Mathematical Model
• Conclusions

3
Introduction (1/3)

• Reason for the Integration of process planning
  and scheduling
• resource availability varies with the dynamic
  condition of the shop floor (machine breakdown)
• this change affects original schedule
• schedules or process plans should be changed to
  adjust to the current situation
• at the presence of alternative process plans,
  they can be considered to improve machine
  utilization and to reduce machining cost

4
Introduction (2/3)

• Multiple process plan generation
• Non-Linear Process Planning
• generates all possible process plans
• ranks process plans according to some criteria
• FLEXPLAN (Detand and Leuben, 1990)
• Closed Loop Process Planning
• generates only one process plan
• if process plan becomes infeasible, modify it or
  generate another.
• Distributed Process Planning
• conducts process planning and scheduling at the
  same time
• IPPM (Zhang, 1993), IPPS (Huang, 1992)

5
Introduction (3/3)

• Process plan selection
• objective
• load balancing, cost, profit,
• solution approach
• mathematical model, tabu search, other heuristics
• Integration of process planning and scheduling
• Proposal of integration schema
• Mathematical modeling
• Heuristics

6
An integrated model of process planning and
production scheduling

• Hong-Chao Zhang and Srinidhi Mallur
• Dept. of IE, Texas Tech Univ.
• I.J.CIM, 1994, Vol. 7, No. 6, 536-364

7
Model Objectives

• To truly integrate process planning and
  scheduling functions
• To generate process plans which reflect shop
  floor conditions
• To consider the objectives of both process
  planning and the scheduling function
  simultaneously
• To improve machine utilization and reduce
  machining cost and time

8
Scheme of Integration
Start Finish time
Manufacturing Resource Database
Route sheet
Job priority
Shop floor status
CAD Interface
Scheduling Module
Process Planning Module
Feature relationship
Available machines
Tolerance analysis
Final plans
Final plans
Decision Making Module Matrix Generation
Possible setups
Available Resource
Process Planning Criteria
Scheduling Criteria
9
Methodology of integration (1/2)

• 1. Decide the value of time_window.
• 2. Feature recognition
• 3. Develop feature relation graph

Final Part
THRD
CHMF1
CROV1
KEYWY
CHMF2
GROV2
HOLE
DIAM1
DIAM2
DIAM3
Raw Material
10
Methodology of integration (2/2)

• 4. Create alternative setups
• (1) Minimize number of setups
• (2) Minimize number of processing steps
• (3) Improve machining accuracy
• 5. Find the feasible process plans by fuzzy set
  modeling
• 6. Scheduling module
• expert system based on the system performance
  measurements
• dynamic priority rule selection
• compute start and finish time for selected jobs
• compute start and finish time for each
  job/machine pair

11
Conclusion

• Pioneering research that presents the scheme of
  integration
• Its components(modules) uses the AI techniques
• Scheduling module applies priority rules based on
  the dynamic shop floor status but which measure ?
• Feature recognition of complex part is still
  difficult.

12
Process Net Model Approach for Multiple Process
Plans

• Ji-Hyung Park and Min-Hyoung Kang
• CAD/CAM Research Center, KIST
• KSME International Journal, Vol. 12, No. 4,
  659-664, 1998

13
Process Net (1/2)

• AND-OR Graph
• and-split (a_s)
• and-join (a_j)
• or-split (o_s)
• or-join (o_j)
• Multiple process plans can be extracted from
  process net
• Input data for NLPP system
• process net can store multiple process plans in
  condensed form
• Net search should be carried out from the head
  node to tail node

14
Process Net (2/2)
Alternative process plans
15
System Configuration
Feature input module
Feature file
Process net generating module
Feature process net file
Process net file
Machine net generating module
Machine data file
Machine net file
Feature precedence matrix
Graphic output module
Process plan generating module
Process Plans
Process Plans
16
Example
Process plan (5)
Process net
17
Conclusion

• Process net representation can save significant
  storage for the process plan.
• This system can be used as a part of the
  integrated process planning system and can be a
  input to the scheduling system.
• Structure of process net needs to be extended to
  a various process types, for example plastic
  processing, heat treatment and non-conventional
  processes.

18
Mathematical model for job shop scheduling with
multiple process plan consideration per job

• Kun-Hyung Kim
• Hanyon Technology, Inc., Seoul, Korea
• Pius J. Egbelu
• Dept. of Systems Engineering, Iowa State Univ
• P.P.C., 1998, Vol. 9, No. 3, 250-259

19
Introduction

• Incorporating process planning and scheduling can
  produce schedules flexibility and adaptability.
• This model simultaneously selects a process plan
  for each job and generates job shop schedule.
• Objective is Min. makespan
• System output
• a set of selected process plans containing one
  plan per job
• schedule of the jobs on the machines based on the
  selected process plans

20
Math. Model (1/3)

• i job
• j process plan belonging to a job
• m machine
• h hth operation in a process plan of a job
• tijhm processing time of operation h in
  process plan j of job i on
• machine m
• Tijhm completion time of operation h in process
  plan j of job i
• Yijhpqsm 1, if operation h in process plan j
  of job i precedes operation s in
• process plan q of job p where operation h and s
  are on machine m
• 0, otherwise
• Xij 1, if process plan j of job i is selected
• 0, otherwise
• H A very large positive number

21
Math. Model (3/2)

• Min Z (1)
• s.t.
• For the last operation in process plan j of job
  i,
• Tijhm - H(1-Xij) ? Z (2)
• For every operation in process plan j of job i
  which has direct successor operation,
• Tijhm - Tij(h-1)g H(1-Xij) ? tijhm
  ?i,j,m,h (3)
• where for the first operation (h1) in process
  plan j of job i,
• Tijhm H(1-Xij) ? tijhm ?i,j,m,h (4)
• For every pair of operations that use machine m
  in process plan j of job i and the process plan q
  of job p,
• Tijhm - Tpqsm HYijhpwsm H(1-Xij) (1-Xpq) ?
  tijhm (5)
• Tpqsm - Tijhm H(1-Yijhpwsm) H(1-Xij)
  (1-Xpq) ? tijhm (6)

22
Math. Model (3/3)

• For every process plan in job i,
• ?jXij 1 (7)
• For the operations that use machine m in every
  process plan of job i and job p,
• -Xij ?qYijhpwsm ? 0 (8)
• -Xpq ?jYijhpwsm ? 0 (9)
• For every operation process plan j of job i,
• Tijhm ? 0 (10)

23
Preprocessing Method
Input all process plan combinations
Bounding Module
Select a process plan Sn with the lowest lower
bound on makespan and set T?
Apply mathematical Method (MPSX)
Select a process plan combination, Y, whose
Lower bound lt T and let YSn
Makespan Tn of Sn
TnltT
NO
YES
YES
Is there a process plan whose lower bound lt T?
TTn
NO
END
24
Bounding procedure (1/2)

• Step 0 Compute the total number of process plan
  combinations, R,
• R?iOi, where Oi Pi for ?i
• Step 1 For each Sn and n 1, 2, , R,
• Input data for process plan (Pij)
• Compute total processing time, Tij,
• Tij?ht(m,i,j,h)
• Compute cumulative processing time Mijm for ?
  Pij and m
• Compute
• (i) Gn(m) ? Mijn ? m
• (ii) Gn(?) MaxmGn(m) ? m

25
Bounding procedure (2/2)

• Step 2 For each process plan Pij?Sn considered
  separately, determine the possible start time
  STn(m,i,j,h) and completion time CTn(m,i,j,h).
• Step 3 Compute
• Earliest start time NSTn(m),
• NSTn(m)MinPijSTn(m,i,j,h)
• Latest completion time XCTn(m),
• XCTn(m)MaxPijCTn(m,i,j,h)
• Step 4 Lower bound LCn(m) of completion time on
  machine m,
• LCn(m) Gn(m) NSTn(m)
• Lower bound of makespan, XC(n),
• XC(n)MaxmLCn(m)

26
Experimental Results
27
Conclusions

• Process planning is the integrator of
  CAD/CAM/CIM.
• Process plannings flexibility enriches the
  production planning and controls quality.
• Through the recognition of the relationship
  between scheduling and process planning, it is
  advisable to maintain multiple process plans for
  a job.
• But computational effort for Simultaneous process
  planning and scheduling problem grows
  exponentially as the problem size increases.

28
Conclusions

• Heuristic method should be developed to solve
  problems that involve a large number of parts,
  process plans and machines
• Research Area
• multiple process plan generation and
  representation
• efficient and fast process plan selection
  algorithm
• efficient and fast algorithm for simultaneous
  process planning and scheduling

29
References

• Hong-Chao Zhang, IPPM-A Prototype to Integrate
  Process Planning and Job Shop Scheduling
  Functions,Annals of the CIRP Vo. 42/1/1993,
  513-518
• Khoshnevis, 1990, Integration of process planning
  and scheduling functions, Journal of Intelligent
  Manufacturing, 1, 165-176
• Paolo Brandimarte, Exploiting process plan
  flexibility in production scheduling A
  multi-objective approach, E.J.O.R. 114 (1999)
  59-71
• M. K. Tiwari and N. K. Vidyarthi, An integrated
  approach to solving the process plan selection
  problem in an automated manufacturing system,
  I.J.P.R. 1998, Vol., 36, No. 8, 2167-2184