megaprojetos 2008 - apresentação - rob smith - megaprojects need new tools: why current pm tools...
DESCRIPTION
O MegaProjetos 2008 foi um evento pioneiro, que trouxe para quem participou uma visão completa sobre projetos de grande porte nacionais e internacionais. Estiveram presentes 450 participantes, representando 158 empresas inscritas e 17 empresas e instituições de peso como apoiadoras.TRANSCRIPT
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
1
Megaprojects Need New Tools:
Why Current PM Tools Don’t Deliver the Goods
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
2
Overview
• Why a fundamentally new project representation and methodology is needed for Megaprojects
• The Plexus™ representation and methodology
• Experience on real industry projects
• Demonstration
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
3
Fundamental research question
• “Why do large, mature, sophisticated organisations routinely fail to deliver complex new products successfully?”
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
4
Existing Tools
• “project management theory remains stuck in a 1960s time warp and that the underlying theory of project management is obsolete”
Professor Peter Morris,“Rethinking Project Management”
EPSRC Network 2004-2006
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
5
Importance of suitable
representations
• Multiply LXXXVII by LXXIII in your head without converting to decimal!
• The Roman Number system is not suited to arithmetic
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
6
Megaprojects Are Complex
Systems
• Involving large numbers of adaptive entities
• Interacting in multiple dimensions
• With stochastic effects
• And time evolution
– Extant project models are not suited to Complex
systems
– How do we derive an adequate models, then?
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
7
Product and Process Knowledge
Planning
Product Knowledge
Equations, Rules.
Process Knowledge Connections, Sequences
Stochastic data Workflow
Engineering
Domain
Management
Domain
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
8
Planning Paradox• Process Knowledge is the
responsibility of, and is generated within the Management domain
• The Engineering domain has an aversion to planning and treats it as low priority, non-core activity
• Planning data has inadequate input derived from and associated with Product Knowledge.
• Plans are too abstract, unrealistic and are more likely to fail
• Poor execution track record further erodes the Engineering domain’s interest and confidence in planning activities.
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
9
“Post-it Planning”
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
10
• Many, large, international, otherwise sophisticated organisations universally resort to “post-it” note planning
• Such planning sessions require labour intensive follow up activities
“Post-it Planning”
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
11
Iterations
• Complex products (such as in aerospace) are highly refined artefacts
• There construction involves many detailed iterations
• Example aircraft design…
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
12
Iteration in design
Determine
aerodynamic
shape
Span, Chord,
Section
Calculate
structural loads
Load casesDesign Load
bearing
structure
Wing box design,
Material volume, density
Calculate
structural mass
Mass of wing structure,
Systems, fuel
In reality projects have many complex iterative loops…..
Not allowed
In PM tools
Aerodynamicist
Flight Physicsengineer
Structural analyst
Mass properties engineer
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
13
Real Iterations!
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
14
Traditional Project management
networks
• Cannot model
iterations
• “Hard”, content-
free dependencies
• One-dimensional,
unsophisticated
visualization
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
15
The Goal
• Tools with a representation that– Can effectively capture process and
product knowledge in meaningful project models
• From which plans are derived
• And from which knowledge is reused
• And which allows for organic growth and replanning
• And which leads to real buy in from all project participants
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
16
Requirements
• Separate project logic from scheduling logic
• Allow for collaborative input
• Allow for multi-dimensional categorization
• Allow for organic growth and replanning
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
17
Plexus approach
• Detailed, multi-dimensional dependency network
• Collaboratively constructed
• For Detailed Simulation
• Multi-objective optimisation
• Analysis and Execution
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
18
Information Driven Planning
Apply Duration & Resource
Requirements
Publish
Schedule
Execute
Monitor and
Control
Re-schedule /
Re-optimise
Scope
Project
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
19
Experience• Project “war-room”
equipped with a data projector and a number of networked lap-top computers.
• Teams to switch between;– collective review of planning
using data-projector
– intensive sessions where sub-groups concurrently develop and refine the network using individual laptops.
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
20
Software Demo
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
21
Demo: Modelling In Plexus
• The Network View
• Focusing the context
• Adding:
– Hierarchy Elements
– Tasks
– Requirements
– Documenting Dependencies with Content
• Dropping and Satisfying Requirements
(collaborative planning)
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
22
Demo: Automatic Understanding
• Cyclic activity handling
• Automatic options and iterations
• Ignoring = Estimating = Risk
• Dependency Matrix View
• Multiple, user-defined plan hierarchies
• Using the project filter
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
23
Demo: Automatic Scheduling
• The DSM for quick schedules
• The resource model
• Discrete event simulation for non-conflicted resources
• Gantt chart view
• Coordinated, filtered views: Network, Matrix, Gantt
• Critical Path to any milestone
• Smart lookup
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
24
Demo: Optimizing Tradeoffs
and Analysis• The Optimizer and DES
• The Tradeoff Surface
– Views, Filters, Colors, Selections
• Multidimensional Export to MS Project
– With roundtrip
• Stochastic Analysis• And much more…
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
25
Demo: Execution and Replanning
• Roundtrip to MSP
• Updating of actual values and percentages complete
• Rescheduling
• Re-optimization that respects actual values
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
26
In Summary• Plexus provides a shift in the planning representation:
– Dependencies have content
– Collaboration provides appropriate content
– And buy in
– For reusable logic
– Iterations and options effect risk naturally
– Schedules are derived data
– That shows trade offs (risk/cost/time)
• Execution should include comparisons to baseline, switching on the trade off surface, reoptimizationwhen necessary, and remodelling
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
27
Benefits• Allows engineers to declare a detailed
representation of the relevant dependencies including iterations.
• Sophisticated viewing and navigation tools allow deep appreciation of context.
• Health checking shows up missing network data, inconsistencies or unconnected activities.
• Allows very fast generation of realistic, detailed networks
• Diagnostics automatically classify network nodes highlighting critical nodes, deliverables and inputs
• Powerful automatic layout algorithms to arrange the network in a logical, compact and visually appealing format
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
28
Questions?
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
29
Model Optional Activity!
• More (resource) cost, less risk:
• More cost and time, less risk:
Dr R Smith
Professor James Scanlan
Professor Phil Lawrence
30
Model Iterations Appropriately
• If a model has natural iterative refinement cycles, model them
• There are three ways: