On the performance of Workflow Processes with Distributed Actors: Does Place Matter

1
On the performance of Workflow Processes with
Distributed Actors Does Place Matter?

• Hajo Reijers1, Minseok Song1, and
• Byunduk Jeong2,1
• 1Eindhoven University of Technology, Dept. of
  Management Technology
• 2Korea Advanced Institute of Science and
  Technology, Dept. of Industrial Engineering

2
Agenda

• Workflow technology
• Research questions
• Case Study
• Approach
• Findings
• Conclusion

3
Workflow management
Commercial Workflow Systems
ViewStar
Lucent Mosaix
eiStream
WANG Workflow
Eastman
BlueCross
WANG
SIGMA
BlueShield
JCALS
CARNOT
Verve
Versata
MS2 Accelerate

• Technology to distribute work to the right
  person at the right time
• Commercial systems available for over 20 years
• WfM technology incorporated in many other
  technologies (ERP, Web Services, etc.).

VisualInfo
Netscape PM
iPlanet
jFlow
BEA PI
Continuum
DST AWD
DST AWD
ImagePlus FMS/FAF
Pavone
Onestone
Domino Workflow
NCR ProcessIT
Exotica I - III
FlowMark
MQSeries Workflow
OpenPM
FlowJet
Pegasus
AdminFlow
Chang
e
ngine
WorkManager
SNI WorkParty
Plexus FloWare
BancTec FloWare
Recognition Int.
COSA
BaaN
Ley COSA
Oracle Workflow
Digital Objectflow
DEC LinkWorks
Digital Proc.Flo.
AltaVista Proc.Flow
Beyond BeyondMail
Banyan BeyondMail
Fujitsu iFlow
Fujitsu Regatta
Teamware Flow
Staffware
FileNet WorkFlo
Visual WorkFlo
Panagon WorkFlo
FileNet Ensemble
ActionWorkflow
Action Coordinator
ActionWorks Metro
DaVinci
Xerox InConcert
TIB/InConcert
IABG ProMInanD
LEU
Olivetti X_Workflow
1985
1990
1995
2000
(Zur Muehlen, 2003)
4
Presumed merits

• Business processes execution is faster and more
  efficient

• It has become unimportant where actors reside
  physically

Longitudinal study (ongoing work)
This work
5
Research questions

• How is affected by
  workflow technology in terms of the geographical
  location of actors?
• How is affected by workflow
  technology in terms of the geographical location
  of actors?

processing time
transfer time
Task A
6
Research questions

• How is affected by
  workflow technology in terms of the geographical
  location of actors?
• How is affected by workflow
  technology in terms of the geographical location
  of actors?

processing time
transfer time
7
Hypotheses

• In a workflow setting
• The processing time of equivalent tasks is
  equally distributed, despite the geographical
  locations in which the tasks are performed.
• The transfer times of inter transfers and intra
  transfers are equally distributed.

8
Case study

• Urban Management Service of a local municipality
  in the northern part of the Netherlands
• Invoice handling process for purchases (e.g.,
  computers, furniture, pens, etc.)
• In principle, each of the 300 civil servants can
  be involved
• Tasks are performed across 10 different
  geographical locations
• Access to event log of 12,000 instances (invoices)

event, task name, actor, time stamp
9
(No Transcript)
10
Approach

• Translated proprietary event log into format of
  the ProM framework
• Selected five crucial tasks that are performed at
  various locations
• Selected two pairs of subsequent tasks with inter
  and intra transfers
• Filtered the event log (e.g. selected tasks,
  anonymized actors)
• Calculated processing times and transfer times
• Applied statistical analyses to determine whether
  differences exist

11
Findings (1)

• Processing time example
• Check by budget keeper on legitimacy invoice
  (CODFCTBF)
• Performed at 9 different locations
• Boxplot suggests differences

12
Findings (2)

• Transfer time example
• Check by local financial clerk followed by check
  budget keeper (ROUTEFEZ ? CODFCTBF)
• Both intra and inter transfers
• Boxplot suggests differences

13
Findings (3)

• Significant differences processing times over
  geographical locations
• for all tasks under consideration
• Kruskal-Wallis test at 95 confidence level
• Significant difference in transfer time between
  inter and intra transfers
• for all pairs considered
• Kruskal-Wallis test at 95 confidence interval

We reject our hypotheses place matters!
14
Evaluation

• Big surprise!
• No satisfactory explanation for differences in
  processing times
• local skills?
• informal norms?
• Likely explanation for differences in transfer
  times
• reports on open invoices in use
• colleagues that are close to you are activated
  more easily

15
Conclusion

• Workflow technology does not take away
  geographical barriers
• Various limitations
• one (!) case
• half a year of operation
• Implications
• Explicitly look for and implement procedures/
  tools/ housing opportunities etc. to stimulate
  interaction patterns

16
Questions?
17
www.processmining.org prom.sourceforge.net