Modeling Kanban Scheduling in
Systems of Systems
Alexey Tregubov, Jo Ann Lane
Outline
Problem statement
Research approach: Kanban process
Overview of KSS Network
Benefits of KSS
Simulation model
Simulation results








Health care example
Complex example
Future work

2
Problem statement
In SoS environments the following problems observed:
Lack of visibility:



What is going on at the project level?
What is the current status of SoS capabilities?
How to expedite certain urgent capabilities?
Everything is critical but nothing is done
Every constituent team establish their own priorities



3
Research approach: Kanban process

Kanban in manufacturing industry:


Kanban as part of lean concept:





Visualized work flow – kanban boards
Limit WIP
Feedback and collaboration to improve the flow
Kanban in System of System environments:



4
“Pull” instead of “push” in demand
Kanban is part of Just in time (JIT) delivery approach
Kanban in a software engineering (as a process):


It is a system to control the logistical supply/demand chain. Toyota,1953.
Kanban scheduling in work prioritization
Value based work prioritization based on SoS capabilities, balanced with
single system needs
Visualization
Overview of KSS Network
5
Benefits of using KSS
Eliminate waste



Minimize context switching
Limit work in progress
Make process more visible and transparent


Kanban boards
Increased value delivered earlier


Value-based work prioritization
Reduce governance overhead

6
Cost context switching in multitasking
120.00
100.00
Percent of Time on Project
Loss to Context Switching
80.00
60.00
40.00
20.00
0.00
1 project
7
2 project
3 project
4 project
5 project
Simulation model
Discrete event simulation:
 Inputs:



Event scenario: a sequence of events that describes how
network evolves over course of their execution
Team configuration: structure of teams, resource/specialties
allocation
Simulation configuration: stop condition
Outputs:



8
Sequence of network states
Analysis: various indicators of effectiveness
Simulation model: Kanban scheduling
All work items (WI) prioritized according to their
business value
Every WI has a class of service: Standard, Important, Date
Certain, Critical Expedite
Limiting work in progress: work in progress is never
interrupted unless





new work has a Critical class of service
work is suspended by prerequisites
Visualization: kanban boards for each team.

9
Health care example of KSS Network
10
Example: capabilities to requirements to products
11
Example: capabilities to requirements to products
Value = 60
12
Value = 20
Value = 10
Value = 10
Example: capabilities to requirements to products
Value = 60
20
20
13
Value = 20
40 10
50
10
Value = 10
Value = 10
Example: network structure & scenario
14
Example: outputs
15
Example: simple scenario
Value:
1200.00
1000.00
800.00
random
600.00
kss
400.00
200.00
0.00
0
16
5
10
15
20
25
30
35 Time
Example 2: complex scenario






10 teams (20 members each) + system engineering team.
20 new capabilities at start.
Each capability unfolds into 30 requirements on average
Each requirement unfolds into 9 tasks on average.
Each tasks takes 3-15 days.
There are 5 expedite tasks that cause blocked work
(blocked tasks)
17
Example 2: value comparison
Value
60000
50000
40000
30000
KSS
20000
Value-neutral (random selection)
10000
LIFO
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80
Time
18
Example 2: number of suspended tasks
Number of
Suspended Tasks
35
30
25
KSS
20
Value-neutral (random
selection)
15
10
LIFO
5
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80
Time
19
Example 2: work items in progress
WIP
KSS
250
Value-neutral (random selection)
200
LIFO
150
100
50
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80
Time
20
Example 2: total time spent (schedule)
Total schedule (person-days)
71
70
69
68
67
66
65
64
63
62
61
KSS
21
Value-neutral (random selection)
LIFO
Example 2: total effort
Total effort (days)
600
580
560
540
520
500
Effort required
if there are
interruptions
480
460
440
KSS
22
Value-neutral (random selection)
LIFO
Example 2: context switching
Effort on context switching
(person-days)
90.00
80.00
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00
KSS
23
Value-neutral (random selection)
LIFO
Conclusion: future work




Pilot the Kanban scheduling with several organizations
Fine-tune the simulator using empirical data and
organizations’ feedback
Scale up the cases we run through the simulator
Refine and calibrate cost models
24
Questions & answers
25
References







Waldner, Jean-Baptiste (September 1992). Principles of Computer-Integrated
Manufacturing. London: John Wiley. pp. 128–132. ISBN 0-471-93450-X.
"Kanban". Random House Dictionary. Dictionary.com. 2011. Retrieved April 12,
2011.
Ohno, Taiichi (June 1988). Toyota Production System - beyond large-scale
production. Productivity Press. p. 29. ISBN 0-915299-14-3.
Richard Turner, Jo Ann Lane, Goal-question-Kanban: Applying Lean Concepts to
Coordinate Multi-level Systems Engineering in Large Enterprises, Procedia
Computer Science,Volume 16, 2013, Pages 512-521, ISSN 1877-0509,
(http://www.sciencedirect.com/science/article/pii/S1877050913000550)
Torgeir Dingsøyr, Sridhar Nerur,VenuGopal Balijepally, Nils Brede Moe, A decade of
agile methodologies: Towards explaining agile software development, Journal of
Systems and Software,Volume 85, Issue 6, June 2012, Pages 1213-1221, ISSN 01641212, (http://www.sciencedirect.com/science/article/pii/S0164121212000532)
Woods D. Why Lean And Agile Go Together : [Digital document],
(http://www.forbes.com/2010/01/11/software-lean-manufacturing-technology-cionetwork-agile.html). Verified on 7/7/2013.
Gerald Weinberg, Quality Software Management: Vol1. Systems Thinking. 1991
26