Project Management tools for the Hadron Collider project Management tools... · 5/17/2010 1 Project Management tools for the Large Hadron Collider project Jurgen De Jonghe Pierre

5/17/2010

1

Project Management tools for the Large Hadron Collider project

Jurgen De Jonghe

Pierre A BonnalPierre A. Bonnal

CERN, European Laboratory for Particle Physics, Geneva, Switzerland

What is a large‐scale project?

Technically complex and demanding,i.e. that cannot be fully specified in the front‐end phase;

Long makespan: economical effects, variations of FX rates,

Involving many contributors, contractors, suppliers... i.e. many activities that are outsourced, result‐oriented.

LHC?

cryogenics superconductivity ultra high vacuum appscryogenics, superconductivity, ultra‐high vacuum apps, electronics, computing, civil engineering…

some 500 FTEs × 12 years (1995–2007);

material: more than 3 GCHF95 (3.75GCHF);

Over 6000 different suppliers and contractors

with ATLAS, CMS, ALICE, LHC‐b & Computing = LHC Programme2

5/17/2010

2

CERN’s EVMS | Project audit of 2001

The LHC Project was a “Must Win” project: under‐funded?

Technical Coordinator → “the project is behind schedule”

Project Administrator → “the project is under‐running”

Ad hoc link between cost control and scheduling system

The LHC Project Management Team was therefore not in position to demonstrate that EAC < TAB

Member States asked CERN Management and the LHC ProjectManagement to set up an appropriate Project Control System

3Jurgen De Jonghe, Pierre Bonnal CERN

EVM Basics


5/17/2010

3

EVM Basics


EVM Basics


5/17/2010

4

CERN’s EVMS | The key requirements

an EVM‐based project control system (ANSI standard 748)

a deliverable‐oriented physical progress monitoring system

interfaced to CERN’s corporate systems (accounting, contract management, human resource)

manages in‐kind contributions

web‐based (the WWW was born at CERN!!) but also uses Excel spreadsheets to interact with DBs

activity scheduling & time‐control reporting engines delayed

Jurgen De Jonghe, Pierre Bonnal CERN 7

CERN’s EVMS | Why a D‐O EVMS?

CERN’s arguments were the following:

In a distributed project environment, transparency of theh i l ti i i ifi tl i dphysical progress reporting is significantly improved

E.g. a “10 cold masses out of 20” physical progress statementis more informative than a “50% complete” statement!

Payment milestones of result‐oriented contracts refer toeffective deliveries

The finish dates of contract activities are always known!

More suited to implement a Line of Balance scheduling engine More suited to implement a Line‐of‐Balance scheduling engine

Finally, this is a trend in project management practices (ref. P.A. Howard. Deliverable‐oriented project management. ProjectWorld’98 Proceedings, 1998)


5/17/2010

5

CERN’s DO EVMS | Principles

“Standard Activity” (or “Standard Work Unit” in CERN’s PM jargon)



“Outsourced Activity”


5/17/2010

6


“Complex Activity”EVj = × Σj BACk

Σj ωl × DAl

Σj ωl


CERN’s EVMS | Main characteristics

It is deliverable‐oriented

It works with a lean project management team It works with a lean project management team

It is collaborative

it allows ad hoc alerting

transparency is promoted(discover overlapping or incomplete areas)

continuous re‐planning is encouraged(trend PV against baseline PV)


5/17/2010

7

LHC Project EVMS | Weak points

The LHC Project EVMS was introduced while AC = CHF 1 billion!!

Too many activities (work units) 12’000+ activitieseasily managed by the system, but difficult for humans

“Varying granularity” of activities:from a few kCHF to several MCHF

from a few weeks to several months

Project Engineers plan too optimistically (“Human nature”) Project Engineers plan too optimistically (“Human nature”)…to obtain budget (it’s easier to carry over)

Weak integration with schedule networks


LHC Project EVMS | Strong points

Despite the late introduction of this system to the project:

tailored tools tailored tools

Excel interface to interact with DBs very appreciated

strongly integrated with corporate databases

deliverable‐oriented approach revealed to be efficient

planned (baseline) vs. expected (trend) figures useful

“AC = EV rule” for in‐kind contribution worked well

incidentally: one repository of activities for all groups(eliminate overlap, detect holes…)

strong Project Leader support (essential)


5/17/2010

8

Outcome of EVM experience…

Re‐established trust of external partners.

CERN management was formally congratulated by g f y g yMember States.

Further LHC overruns were discussed in the context of proven project control via EVM.

The LHC Cost and Schedule Review Committee considered the application of EVM to be ‘world class’the application of EVM to be world class .

Project culture at CERN and cost awareness of project engineers have improved.


CERN’s EVMS | Further uses

Design and developed for the LHC Project, but implemented on several other projects

ATLAS Detector Project

LHC (machine and infrastructures) Project

CNGS (CERN Neutrino Beam to Gran Sasso) Project (completed on schedule, under budget in 2006)

EGEE Project(on going)

Linac4 Project(on going)


5/17/2010

9

Need for Linear & RepetitiveNeed for Linear & Repetitive Scheduling Methods


Scheduling

CPM (C i i l P h) / PDM (P d Di i ) CPM (Critical Path) / PDM (Precedence Diagramming)

Excellent companions to EVM

A critical task is not necessarily expensive…

Fragile for large sets of workunits


5/17/2010

10

Need for Linear Scheduling methods

Setting up a project activity network diagram for strict application of CPM or PDM is hard: 1800 cryo‐magnets need to be manufactured and installed

Single unit is complex: hundreds of operations in assembly

Schedule should also ensure efficient use of available crews/resources. Work load continuity between repetitive activitiesWork load continuity between repetitive activities


Introducing Repetitive & Linear Tasks

units

1800

meters

yk

k

time

1800

0

S F

k

time0

S F

xk

yk

k’

k

Sk Fk Sk Fk

k : Production rate (units/month)


5/17/2010

11

What do we want to schedule?


Master Schedule for Installation

22

5/17/2010

12

Repetitive Schedule for the Production of LHC Dipole Cryo‐magnets (Coils)

Many othercomponents

Manycomponents



ManufactureSC cable

Inner LayerAssemble

Inner Layer

unit no.

1250

1000Assemble Coils

Manycomponents Many other

components

componentsMany othercomponents

AssembleOuter Layer

AssembleCoil

AssemblePole

AssembleDipole

Cryo-magnet

ManufactureSC cable

Outer Layer

1000

750

500

250

1

1999 2000 2001 2002 2003 2004 2005 2006

Manufacture SC Cables

Assemble Inner & Outer Layers

Assemble Poles

Assemble DipoleCryo-magnets

Sub‐optimal scheduling

repetitive activities often have different production rates.

production rate imbalance → negatively impacting project performance by causing

Work stoppages,

Inefficient utilization of resources

Excessive costs


5/17/2010

13

Repetitive tasks scheduled with CPMunits

k

k

j

time

k

Sub-optimal !

j


RSM modeled with PDM constraints

jk jk jk units

k jj’j’’

time

k

jSS with Lagkj (span of k)

j’j’’FF with Lagkj (span of j)


5/17/2010

14

LSM modeling: “basic” PDM breaks down

jjk TSSS

j

k

meters

xj

yk

yj

j

jjjjk

yxTSSS

k

kj

j

jjjjk

xxyxTSSS

jkk

kj

j

jjjjk L

xxyxTSSS

timexk

Sj Sk Fj Fk

kj


Now do this 117 times …

<<>

<

k

jkj

j

=>

<>

>>

>=

==

<=

><

=<

<<

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

yjxj

ykxk

D D D H H HL L L

9 9

11 11 11

9

3 3 3

9 9 9 9 9 9

6 6 6<<=

<<<

=<

><

<=

==

>

k

k

jkj

jkj

jk

jk

jk

jk

DDD

EEE

EEA

EE

EE

E

B B B B

A

EE

EE

E

A

L L L

L L L L L L H H H

H H H

L L L L L L L L L

B B B B B

11

11

11

11 11 11 11 11 1111 11 11

11

11

11

1111 11 11

11

113

3

1

3

3

3

3 3

3

36

5 5 15

5 5 5 5 5 5 5 55

5 15 5

6

7 7 7 7 7 7 7 7 7

6

6 6

6

Details will be published in>=

<>

=>

>>>

>>=

>><

k

k

k

jk

jk

jk

jkj

jkj

jkj

C

G

G

JJ

C L

C L

J JJ J J JJ

G

G

F

F

C L

C L

F

F

C L

C L

F

F

G

G

G

G

G

G

G

G G G G

L L C L L C L L

K K K11

11

11

11

11

11

11 11 11 11

11

11

10 10 10

88 8

88 8

88 8

2 2 2 2 2 2 2 2 2

4 44

4 44

7 7 7 7 7 7

Details will be published in Journal of Construction Engineering and Management


5/17/2010

15

Analytical Linear Scheduling Method

While the formula’s and the number of cases to examine looks daunting, the idea is based on basic PDM:

Earliest Start dates: calculated by propagation in a forward pass

Earliest Finish dates are easily derived

Latest Finish dates: calculated by propagation in a backward pass

Latest Start dates are easily derived

Total and Free floats are derived from ES and LF (which identify critical activities)


Dealing with economic issues such as inflationDealing with economic issues such as inflation and currency exchange rate


5/17/2010

16

Economic Risks

Managing a Project Baseline is not a straightforward exercise: imprecise figures, unexpected events...

The comparison of actual figures with planned ones is not straightforward either: inflation, variations of FX rates...

Practices are, in this respect, not (so much) standardized!

Technical problems, programmatic problems… are the responsibility of the project management teams

But can a project management team be made responsible for But can a project management team be made responsible for economical or commercial unexpected events?


Project Management Reserve


5/17/2010

17

Reminder: What is a large‐scale project?

Technically complex and demanding,i.e. that cannot be fully specified in the front‐end phase;

Long makespan: economical effects, variations of FX rates,

Involving many contributors, contractors, suppliers... i.e. many activities that are outsourced, result‐oriented.

LHC?

cryogenics superconductivity ultra high vacuum appscryogenics, superconductivity, ultra‐high vacuum apps, electronics, computing, civil engineering…

some 500 FTEs × 12 years (1995–2007);

material: more than 3 GCHF95 (3.75GCHF);

Over 6000 different suppliers and contractors

with ATLAS, CMS, ALICE, LHC‐b & Computing = LHC Programme33

There are basically two approaches to cost estimate and budget large‐scale projects:

What is really included in PV?

the so‐called "Old world" approach.

the so‐called "US Government" approach.


5/17/2010

18

What is in PV as per the “Old World” approach?

1.The costs are estimated at WP/PP level, as if everything was bought or paid today.

2.After reviews, these estimates become budgets.

3.The WPs and PPs, and their budgets, are spread over th j t kthe project makespan.

4.The PV curves are then drawn.


What is in PV as per the “US Gov” approach?

1.The costs are estimated at WP/PP level.

2.The WPs and PPs, and their budgets, are spread over

the project makespan.

3.Estimated escalation rates are applied to all cost figures.

4 These time dependant cost figures become budgets4.These time‐dependant cost figures become budgets.

5.The PV curves are then drawn.


5/17/2010

19

What is in PV as per the “US Gov” approach?

FY R h &�O ti &I f tiE i t l

Project Categories

n/a2.32.82.72 6

1.0001.0231.0511.0801 108

n/a1.82.62.72 6

1.0001.0181.0451.0731 101

n/a0.80.90.51 0

1.0001.0081.0171.0221 032

n/a2.02.72.72 6

1.0001.0201.0471.0751 103

n/a2.12.52.92 8

1.0001.0211.0461.0761 106

20022003200420052006

FY Research &Development

�Operation &Maintenance

InformationTechnologies

EnvironmentalManagementConstruction


Escalation rate assumption for DoE projects (as of January 2002)downloadable from http://oecm.energy.gov/ [June 2005]

2.62.5

1.1081.136

2.62.4

1.1011.127

1.00.8

1.0321.041

2.62.4

1.1031.130

2.82.6

1.1061.135

20062007

“US Gov” vs. “Old World” approach


5/17/2010

20

Progress monitored in “US Gov” approach

if WPj is still planned.if WP is completed

0BACEV (t)

EV(t) in running dollars (r$).EVj(t)

jSEV(t) =

if WPj is completed.if WPj is on-going.

BACj

BACj × X%EVj(t) =


AC(t) in running dollars (r$).

expenses, as recorded in the ledger.t0

t

SAC(t) =

Progress monitored in “US Gov” approach


5/17/2010

21

Progress monitored in “Old World” approach

if WPj is still planned.f WP l d

0BACEV ( )

EV(t) in initial pounds (i£).EVj(t)

jSEV(t) =

if WPj is completed.if WPj is on-going.

BACj

BACj × X%EVj(t) =


AC(t) in initial pounds (i£).r£ into i£.

as recorded in the ledger.

(1 + i)nexpenses

t0

t

SAC(t) =



5/17/2010

22

The reality of project management practitioners is even more complicated! E.g.: outsourced activities.



“US Gov” vs. “Old World” approach

Can an EVM system (suited to large‐scale projects) be both (relatively) simple and accurate?


5/17/2010

23

Towards a hybrid approach...

1. The costs are estimated at WP/PP level, as if everythingwas bought or paid at t = t0.

2. After reviews, these estimates become budgets.3. The WPs and PPs, and their budgets, are spread over

the project makespan.


the project makespan.4. The PV curves are then drawn.


5. Every year or twice a year, the BACs (and the PVs)h i i d dj d d dthe remaining WPs and PPs are adjusted or updated to integrate enonomical affects.


6. The differential is either picked up from the Project Management Reserve (PMR) or from additional fundingsby, and agreed with, the project owners.

5/17/2010

24




This is the model we have chosen for the LHC Project.


5/17/2010

25

How to make this approach work: Initiating phase


How to make this approach work: Planning phase


5/17/2010

26

How to make this approach work: Executing phase


www.eva‐europe.eu

Project Management tools for the Hadron Collider project Management tools... · 5/17/2010 1 Project Management tools for the Large Hadron Collider project Jurgen De Jonghe Pierre

Documents