Complex Systems Design & Management 2011 Rework: model & metrics Edmond TONNELLIER & Olivier TERRIEN Thales Systèmes Aéroportés (TSA)
2 /
TSA1015802-02 / 9 dec.2011
Where?
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
TSA: core businesses
3 /
Unmanned Air Vehicle systems Sensors, sensor systems, for combat platforms Maritime Patrol & Surveillance systems
TSA1015802-02 / 9 dec.2011
Fighter aircraft retrofit
N°1 in Europe cuttingcutting-edge technologies from research to manufacturing ~20% of revenues dedicated to R&T
Self-protection systems TSA: Thales Systèmes Aéroportés
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
TSA: locations
4 /
~3000 people in:
Elancourt (headquarters)
Brest
Pessac
In close collaboration with: with:
TSA1015802-02 / 9 dec.2011
Crawley & Leicester – Thales UK
Etrelles – Thales Microelectronics
Orsay & Ulm – UMS (JV Thales & EADS)
Leicester Crawley
Ulm
Elancourt Brest Étrelles Orsay Pessac
TSA: Thales Systèmes Aéroportés
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
5 /
TSA1015802-02 / 9 dec.2011
Why?
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Stakes & Issues
6 /
” “why has our schedule been delayed?
TSA1015802-02 / 9 dec.2011
WASTE
project Y?”
Poster CSDM2010
ur project X?” o f o d n e e th t is extra cost a th d se u a c t a “wh “how can we explain this cost shi ft on our
After several benchmarks, we have turned our first corner and started an initiative on rework CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Definition of rework
7 /
Rework: “work done to correct defects” Defect: “failure to conform to requirement” (even if this requirement has not been explicitly specified) Definitions from P.Crosby, ‘Quality is Free’.
“Incomplete or misinterpreted requirements at the start of a project resulting in rework in cascade through to subcontractors”
TSA1015802-02 / 9 dec.2011
“Low defined designs result in expensive reworks to meet the customers' true requirement”.
“Late changes in requirements cause high levels of rework throughout the life cycle of products”
Examples from external and internal interviews (benchmarks).
How can we keep inevitable rework under control?
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
8 /
TSA1015802-02 / 9 dec.2011
What?
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Behavioural model
9 /
Need
Development
Solution Work Really done
Work to be done
Work being done Work supposed done Defect Discovery
TSA1015802-02 / 9 dec.2011
Defect Correction
‘Re-work' is a looped phenomenon on solutions and/or processes
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Model of a looped phenomenon
10 /
1
WORK
OK
KO
1
Deliver the solution
5
REWORK
Integrate the solution
2
+
3
+
4
+
5
2
4
Look for the causes
3
Develop a solution
TSA1015802-02 / 9 dec.2011
Discover the problem
+
+ Monitoring = REWORK
Rework plays a central role in generating delays and overcosts KO means ‘not OK’ CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
11 /
TSA1015802-02 / 9 dec.2011
How much?
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Impacts on extra-delays and over-costs
12 /
TSA1015802-02 / 9 dec.2011
KO
Defect i ( Di )
1
Discover the problem
2
Look for the causes
3
Develop a solution
4
Integrate the solution
5
Deliver the solution
OK CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
impact
time
rework
Envelope
Surface
=
=
Correction Process ( Di )
Rework ( Di )
time
Mathematical model
13 /
tstop CP ( Di , t ) . dt
Rework ( Di ) = tstart
Di : Defect i
TSA1015802-02 / 9 dec.2011
CP : Correction Process
tstart : date of detection of the defect i tstop : date of closure of the correction process
‘Re-work' is an accumulation of over-activities to correct defects
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
14 /
TSA1015802-02 / 9 dec.2011
How? nd figures are a s h p ra g g in Follow ining materials a tr m o fr d te c extra
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Lever 1 : “volume per month”
15 /
al Usu h p G ra
New h p G ra 1 2 3
tstop CP Hours ( Di , t ) . dt
Rework Hours ( Di ) =
Prediction
TSA1015802-02 / 9 dec.2011
tstart
Volume of defects per month: month:
Workload per month: month:
Sorted by date (creation/closure), usual graphs display only volumes.
Using hours spent to correct defects, graphs display workloads.
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
1.1
To identify major events
2.2
To detect saturation
3.3
To predict future deliveries
Lever 2 : “duration per defect”
16 /
al Usu h p G ra
New h p G ra 1 2 3
tstop CP Days ( Di , t ) . dt
Rework Days ( Di ) =
5%
TSA1015802-02 / 9 dec.2011
tstart
Duration per month: month:
Distribution per defect: defect:
Sorted by date (creation/closure), usual graphs display average duration
Distributed by correction processes, graphs display the most disturbing defects
1.1
To detect increasing delays
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
1.2
To estimate the worst cases
2.3
To prioritize improvement actions
17 /
TSA1015802-02 / 9 dec.2011
When?
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
Lever 3 : “accumulation of workload”
18 /
i=n
Rework ( Di )
Rework ( 1, n ) = i=1
Typical S-curves on projects
3 2
TSA1015802-02 / 9 dec.2011
1
Divergent curves: curves:
Convergent curves: curves:
Increasing rework, 1 project not under control (‘firemen syndrom’)
Increasing rework, 2 But project under control
CSDM 2011 - Copyright@ E.Tonnellier & O.Terrien
NeverNever-ending curves: curves: Rework is like a ‘tail of comet’ 3 End of a project
Lever 4 : “levers per step”
19 /
1
OK
TCreation
T0
Discover the problem
+1
2
Look for the causes
+58
3
Develop a solution
+75
4
Integrate the solution
5
Deliver the solution
+94
TClosure
+105
T0 +1
1
2
+16
Improvement workshop
KO
…
+22 … +46 +51
tstop, Analyze CP ( Di , t ) . dt
Rework ( Di ) =
TSA1015802-02 / 9 dec.2011
tstart, Analyze
Map the current situation:
Evaluate improved situation:
Describe and quantify the current correction processes performed by local teams
Quantify the correction processes performed by local teams after improvement actions