Project Scheduling Principles and Strategies

advertisement
INT 02 - Advanced Schedule Analysis
Advanced Schedule Analysis
David T. Hulett, Ph.D.
Hulett & Associates, LLC
ICEAA / 2013
© 2014 Hulett & Associates, LLC
1
Summary







A schedule is a dynamic model – dates are outputs
Hammocks
Lags
Danglers
3-point estimates of durations
Out-of-sequence progress
GAO 10-point scheduling best practices
© 2014 Hulett & Associates, LLC
1
ICEAA 2014 Professional Development & Training Workshop
2
INT 02 - Advanced Schedule Analysis
Schedule – Dynamic Model of the Project,
not a Calendar on the Wall(1)

The schedule is a model of the project plan
– Activities from the WBS
– Relationship logic between predecessor and successor activities
– Resources applied to the activities
– Necessary external constraints


If the facts (e.g., activity durations) change, the dates
change because activities are linked and dates are outputs
Artificial constraints in the computer model can frustrate the
automatic calculation of the dates implied by changes in
durations
© 2014 Hulett & Associates, LLC
3
Schedule – Dynamic Model of the Project,
not a Calendar on the Wall (2)




A calendar uses constraints to set activities and events on particular
pre-determined dates as the input, not the output
– Calendar date constraints can be inserted into the schedule to fix
certain events (turnover) in time
– Even if the durations and logic make those dates impossible
There is a desire to finish the project on a date, but the schedule may
not support those dates.
Do not force dates onto the schedule – let the durations and logic
determine the dates
Re-plan if necessary
© 2014 Hulett & Associates, LLC
2
ICEAA 2014 Professional Development & Training Workshop
4
INT 02 - Advanced Schedule Analysis
Define Activities
Summary or Hammock Activities

Hammock summarizes activities at a lower level of detail

Linked to the detail activities
– Sometimes called Summary Activity (MS Project®)
– Logic attaches hammock to detailed activities
– Start-to-start with the first detail activity
– Finish-to-finish with the last detail activity
– Duration is passive, determined by the detail activities, since it
goes on as long as the detail activities are not finished

Hammocks are used for level of effort activities, to show
resources that are LOE
© 2014 Hulett & Associates, LLC
5
Sequencing Activities and
Adding Summary Activities in MS Project
In MS Project a LOE or hammock is a Summary Task
© 2014 Hulett & Associates, LLC
3
ICEAA 2014 Professional Development & Training Workshop
6
INT 02 - Advanced Schedule Analysis
“Level of Effort” (aka Hammock)
Task in Primavera P6
Construction Hammock starts with Construction on Unit 1 and ends with
Unit 2 at 250 days. Has©42014
successors,
two with each construction activity.
7
Hulett & Associates, LLC
Level of Effort work Needs to be
Hammocked to be Dynamic with Changes

Many schedulers schedule LOE tasks as Work-type or Task
Dependent
– Work or task activities have fixed durations, they do not expand or
–
–
–
contract with the work they support or manage
The activity may have the correct duration in the baseline, but…
If something happens to change duration of detailed tasks the
“static LOE” activity will keep its original duration
Get an incorrect duration and cost for the LOE resources
© 2014 Hulett & Associates, LLC
4
ICEAA 2014 Professional Development & Training Workshop
8
INT 02 - Advanced Schedule Analysis
Level of Effort work Needs to be
Hammocked to be Dynamic with Changes
Now, with Design Unit 2 shorter and Construct Unit 1 longer, the
hammock starts with Unit 2 and ends with Unit 1 at 230 days. Changes
start and finish activities and duration when durations change.
© 2014 Hulett & Associates, LLC
9
Using a Fixed Lag to Place Successor
Activity Improperly in Time

Lag places successor on July 1, receipt of fiscal year money
– This “works” only on day 1 before “things change”
– This logic may fail at the very first status date
July 1
Design Item
F-S Lag X days
Fabricate
Item
© 2014 Hulett & Associates, LLC
5
ICEAA 2014 Professional Development & Training Workshop
10
INT 02 - Advanced Schedule Analysis
Using Fixed Lags may Shift
the Successor Out Improperly

Suppose the design is delayed.
– With the lag fabrication will be pushed out
– Did we want this to occur?
July 1
Design Item
Fabricate
Item
© 2014 Hulett & Associates, LLC
11
Use “Start Not Earlier Than” Constraint

Using Start NET and F-S but no lag
– Successor still starts on its desired start date
Start NET
July 1
Design Item
Fabricate
Item
© 2014 Hulett & Associates, LLC
6
ICEAA 2014 Professional Development & Training Workshop
12
INT 02 - Advanced Schedule Analysis
The Problem with “Leads”

“Leads” are negative lags
– Something happens before something else is supposed to
– Software lets you do this

Leads are generally illogical
– Start “X days before something is forecasted to happen”
– “Turn left three miles before the freeway”
– Cannot tell when the activity or milestone will be recorded in
advance unless everything goes according to plan
© 2014 Hulett & Associates, LLC
13
Difficulty with
Negative Lags (Leads)

Negative lag may mean, “Start this activity 25 days before
that one finishes”
– Get out your crystal ball for this one
Predecessor
F-S -25 days
Successor
© 2014 Hulett & Associates, LLC
7
ICEAA 2014 Professional Development & Training Workshop
14
INT 02 - Advanced Schedule Analysis
Negative Lag (Lead) is a
Difficult Assumption in Practice

Start the successor 25 days before predecessor is predicted
to finish
– If predecessor takes 15 days longer, successor has started 40
days before finish
Predecessor
Delay
F-S -25 date becomes 40 day lead
Successor
15
© 2014 Hulett & Associates, LLC
Effects of adding Lags to Schedule Logic
70
F-F + 30 d
DESN501
80
S-S + 20 d
DRFT501
Where did the “+
20 d” and “+ 30
d” durations
come from?
What do they
signify?
Are they work?
© 2014 Hulett & Associates, LLC
8
ICEAA 2014 Professional Development & Training Workshop
16
INT 02 - Advanced Schedule Analysis
Use Activities instead of Lags
One problem is that the breakout of the activities may be artificial with
suspicious milestones
20
50
DESN501a
F-S
DESN501b
50
F-S
DRFT501a
30
DRFT501b
© 2014 Hulett & Associates, LLC
17
Lag Abuse

Lags are often abused to make successor start on specific
date
Predecessor
87-day lag
Successor


What does the 87-day lag represent?
Rx: find predecessors to determine the start date of the
successor
© 2014 Hulett & Associates, LLC
9
ICEAA 2014 Professional Development & Training Workshop
18
INT 02 - Advanced Schedule Analysis
Add a Lag in a Summary Schedule

S-S and F-F with a lag
– Successor starts after predecessor finishes + lag
– Lag often represents work that needs to be done before the
–
successor can start, or done after predecessor finishes
In a detailed schedule there would be activities and F-S logic
Electrical, Plumbing
& Ductwork
FF+ 8 d
SS+ 5 d
Drywall
Don’t worry so much about small lags
© 2014 Hulett & Associates, LLC
19
The Problem with Dangling Activity Logic



We need to avoid dangling activities – incomplete logic
Dangling activities are those for which the logic does not
automatically transmit changes (e.g., lengthening) of
duration to the proper successor
With dangling activities we cannot trust the results:
– Dates
– Critical path
– Float
© 2014 Hulett & Associates, LLC
10
ICEAA 2014 Professional Development & Training Workshop
20
INT 02 - Advanced Schedule Analysis
Problem with Dangling Activities
with S-S Logic
FS
Design Longer
Design
SS
Draft
Draft Longer
SS
FS
Build
Can Build
finish
before
Draft and
Draft
before
Design?
Figure A – Lengthening of S-S Danglers
© 2014 Hulett & Associates, LLC
21
S-S Dangler Logic: Primavera P6
Design 1A should finish before Build 1A does.
If Design 1A is longer (200 d), Build 1A is unaffected and FINISHES
BEFORE Design 1A
© 2014 Hulett & Associates, LLC
11
ICEAA 2014 Professional Development & Training Workshop
22
INT 02 - Advanced Schedule Analysis
Examples of Dangling Activities
with F-F Logic
FS
Design
Can Draft
Start
before
Design
and Build
before
Draft?
FF
Draft Longer
Draft
FF
FS
Build Longer
Build
Figure B – Lengthening F-F Danglers
© 2014 Hulett & Associates, LLC
23
F-F Logic in Software: Primavera P6
Build 1 A should start 50 days after Design 1.
If Build 1A is longer, it STARTS TOO EARLY
© 2014 Hulett & Associates, LLC
12
ICEAA 2014 Professional Development & Training Workshop
24
INT 02 - Advanced Schedule Analysis
A Solution: S-S and F-F
Design
SS
Base
FF
Draft
Design
Predecessor
Takes Longer
Design Longer
SS
FF
This is
OK
Draft
Design
Successor
Takes Longer
SS
FF
Draft
This is
OK
Draft Longer
Figure C – Closing Off Danglers, Activities Longer, Right Answers
© 2014 Hulett & Associates, LLC
25
S-S and F-F Logic: Primavera P6
Primavera allows two activities to be linked S-S AND F-F
(MS Project does not allow this)
© 2014 Hulett & Associates, LLC
13
ICEAA 2014 Professional Development & Training Workshop
26
INT 02 - Advanced Schedule Analysis
General Rule with Logic, Best Practice

ALL activities, except the first and last activity, MUST have
at least one "?-S" Predecessor relationship AND one "F-?"
Successor relationship,
Activity 101
Predecessor
F-S or S-S

Successor
F-S or F-F
These relationships must be “driving”
– A delay or lengthening in the predecessor has the most direct
impact on the successor
© 2014 Hulett & Associates, LLC
27
Three-Point Estimate of Duration


There is risk that the activities’ work will not be finished in
the duration allocated (threat), or that it might be finished
early (opportunity)
Take into account these uncertainties to make a better
(more realistic) estimate of duration
Beta or PERT estimate = (Opt. + 4xML + Pess.) / 6
Triangle estimate = (Opt. + ML + Pess.) / 3
© 2014 Hulett & Associates, LLC
14
ICEAA 2014 Professional Development & Training Workshop
28
INT 02 - Advanced Schedule Analysis
Compare the Triangular and Beta
Triang(250,300,480) and Beta(250,300,480)
Triang = 90%
Beta = 98%
Triang = 10%
Beta = 19%
10
Triang
Mean=343
9
Values in 10^ -3
8
7
6
5
Beta Mean=322
4
3
2
1
0
250
300
350
400
© 2014 Hulett & Associates, LLC
450
500
29
29
How do You Handle
Out of Sequence Progress?

An activity is not scheduled to start since its predecessor is
not completed, but it did start and progress is reported
– How do you handle this?

Two general alternatives
– Progress Override
– Retained Logic


Which is your software’s default?
Which do you want to assume?
© 2014 Hulett & Associates, LLC
15
ICEAA 2014 Professional Development & Training Workshop
30
INT 02 - Advanced Schedule Analysis
Project Override Approach:
The Task Manager Knows Best



The task manager for Build Unit 2 has started before Design
Unit 2 completes
He may know something we do not, that it is OK to start
early
Project override says Progress in the Field Overrides the
Schedule
© 2014 Hulett & Associates, LLC
31
Retained Logic Approach:
The Scheduler Knows Best


As much as possible of the original logic is retained
Make the remaining duration of Build Unit 2 wait until the
Design is completed before going any further
– The last 54 days of Build Unit 2 must wait until Design Unit 2 is
completed

Retained Logic says that the scheduler understands the
logic of the schedule better, maybe know design will change
© 2014 Hulett & Associates, LLC
16
ICEAA 2014 Professional Development & Training Workshop
32
INT 02 - Advanced Schedule Analysis
Avoid Scheduling Abuses
Do not turn the schedule into
a pretty “feel-good” calendar on the wall
that appears to support project date objectives
A schedule is a dynamic analytical and planning tool
The schedule will be used to
manage a real-life project
© 2014 Hulett & Associates, LLC
33
Sometimes The Scheduler Must
“Push Back” on Unreasonable Dates



Constraints often arise from target-setting too early or for
business or political reasons at the beginning
Instead, customers should specify WHAT should be done
Project manager should be the professional to plan:
– How it should be done
– How long it should take
– How much it should cost

Project managers must be willing to “push back” from
imposed dates that are unreasonable
© 2014 Hulett & Associates, LLC
17
ICEAA 2014 Professional Development & Training Workshop
34
INT 02 - Advanced Schedule Analysis
http://www.gao.gov/products/GAO-12-120G
35
© 2014 Hulett & Associates, LLC
Schedule Check Report in Pertmaster
© 2014 Hulett & Associates, LLC
(C) 2010-2013 Hulett &
Associates, LLC
36
18
ICEAA 2014 Professional Development & Training Workshop
36
INT 02 - Advanced Schedule Analysis
Details in Schedule Check Report
© 2014 Hulett & Associates, LLC
(C) 2010-2013 Hulett &
Associates, LLC
37
37
Using Acumen FUSE for Best Practices
© 2014 Hulett & Associates, LLC
(C) 2010-2013 Hulett &
Associates, LLC
38
19
ICEAA 2014 Professional Development & Training Workshop
38
INT 02 - Advanced Schedule Analysis
Apply GAO Best Practice Scheduling to
Assess Agency Schedules





The Government Accountability Office in the US has identified 10
criteria to review schedules
Review and assess the project schedule
Focus mostly on normal activities (not summary) and not complete (<
100% complete)
Can use tools such as Steelray, Fuse or Pertmaster’s schedule check
report to perform analysis
Often have to use native schedule and filter / sorting capabilities. Can
copy to Excel for sorts and report table creation
Refer to the GAO Cost Estimation and Assessment Guide, 2009, p. 218
© 2014 Hulett & Associates, LLC
39
BP1: List All the Work




Is all the work represented in the schedule?
Look to the WBS to see if it is all in the schedule
The WBS represents work – cannot be represented just by
milestones (events, not work)
Hard to tell if all of the work is really there even if WBS is
represented
© 2014 Hulett & Associates, LLC
20
ICEAA 2014 Professional Development & Training Workshop
40
INT 02 - Advanced Schedule Analysis
BP2: Sequence the Work without Lags,
Constraints or Danglers (1)




Are all activities and milestones sequenced using proper
network logic?
Open ends or dangling activities. They all need successors
from their finish dates and predecessors to their start dates
Inappropriate use of lags, especially to insert buffers or put
activities on dates. Use activities instead of lags
Leads (negative lags) are dubious, difficult to justify and to
use
© 2014 Hulett & Associates, LLC
41
BP 2: Sequence the Work without Lags,
Constraints or Danglers (2)



Inappropriate use of constraints to place activities on dates
that are different from those determined by predecessors
Finish-to-start logic is preferred but start-to-start and finishto-finish logic is OK if needed. Start-to-finish is dubious
No logic on summary or level-of-effort (LOE) tasks
© 2014 Hulett & Associates, LLC
21
ICEAA 2014 Professional Development & Training Workshop
42
INT 02 - Advanced Schedule Analysis
BP 3: Assign Resources to Activities





Are resources identified and assigned to activities?
–
Difficult for schedules to meet – may disqualify some schedules
Full descriptions of the resources, not necessarily by person but
by discipline, crew, time-dependent and time-independent costs
Resources should be leveled so the schedule is feasible – check
to see if resources are over-allocated
Resources should be costed so the schedule has the budget too
We are seeing management of hours done in other software –
the question is whether the schedule reflects these requirements
(source of so many SNET constraints?)
© 2014 Hulett & Associates, LLC
43
BP 4: Are Durations Realistic? (1)



Are the durations of key activities reasonable? Length and
realism given work and resources?
Generally a detailed schedule requires short activities that
are not longer than two review periods (e.g., 2 months)
Some activities in the far future are “planning packages” that
can be longer because not fully planned (e.g., “rolling wave
planning”)
© 2014 Hulett & Associates, LLC
22
ICEAA 2014 Professional Development & Training Workshop
44
INT 02 - Advanced Schedule Analysis
BP 4: Are Durations Realistic? (2)


Activity durations should be determined by work and
available resources, productivity and work environment
Often see LOE activities scheduled as normal – can cause
problems if it drives successors since LOE cannot drive
– Summary or hammock is better for LOE

Look for evidence of the basis of estimate – what data and
methods were used to determine durations, and are they
the same to determine costs
© 2014 Hulett & Associates, LLC
45
BP 5: Horizontal and Vertical Traceability


Is the schedule traceable vertically and horizontally?
Vertical traceability (integration)
– Find a summary schedule or high-level presentation of the
–

schedule and check the important dates with the detailed
schedules for the same date
May find problems with PowerPoint presentations to Congress
Horizontal traceability (integration)
– Harmed by open ends, dangling activities, reliance on lags and
constraints (see earlier criterion # 2)
© 2014 Hulett & Associates, LLC
23
ICEAA 2014 Professional Development & Training Workshop
46
INT 02 - Advanced Schedule Analysis
BP 6: Realistic Critical (Longest) Path





Is the critical path identified and realistic?
Check starting from the completion milestone and working
back to discover which activities are driving
Is the end date determined by its predecessors or is it
constrained?
Is there a break in the path, maybe fixed in time by SNET
constraints to put an activity in time
We are starting to focus on the Longest Path instead of
critical path
© 2014 Hulett & Associates, LLC
47
Review the Critical Path

Surprisingly, many PMs do not even review the critical path
of their schedules
– Based on zero total float


Some critical paths go through the least critical activities
– Level of Effort activities? LOE activities will never define
the project’s duration
Redefine “critical” to include any activity with, say, 20 days
of float
– Do you see your “intuitive” critical path yet?
© 2014 Hulett & Associates, LLC
24
ICEAA 2014 Professional Development & Training Workshop
48
INT 02 - Advanced Schedule Analysis
Critical Path to Which End Date?



Programs may focus on activities other than the very last
one in the schedule file
Find out from the program which is their final delivery
The total float (total slack) for an activity is calculated to the
last activity in the schedule file, and may not relate to the
final delivery of the program
© 2014 Hulett & Associates, LLC
49
Longest Path through the Schedule


The longest path is a newer concept, included in P6
Longest path starts at final delivery and works back toward the
start or actual dates by selecting the predecessors with zero free
float
– This represents the activities that are driving the final delivery
– Critical path is based on zero total float
– Longest path is based on zero free float

With backward pass constraints, the critical path may contain many
activities that have no impact on the final milestone
© 2014 Hulett & Associates, LLC
25
ICEAA 2014 Professional Development & Training Workshop
50
INT 02 - Advanced Schedule Analysis
BP 7: Is Total Float Realistic?


Do the “total float” or “total slack” values seem reasonable?
Sort the total float largest to smallest, check the largest float
to see if it reflects a project with a lot of flexibility
– Total float values reflect activity logic.
Incomplete logic can easily
lead to large float values



Fixing logic (open ends) will address total float along paths
Some high float is correct, others is due to incomplete logic
Many project managers do not look at their own total float
© 2014 Hulett & Associates, LLC
51
Review the Float





Surprisingly, many PMs do not even review the float in their
schedules
Some floats are 200+ days (40 weeks).
– Can this be right on your project?
Just a listing be total float (slack) in descending order will be very
informative
High total float is probably an indication that the schedule logic is not
complete.
Remember, an open end at the end of a path will cause high float all
along the path, so fix one and you may fix them all
© 2014 Hulett & Associates, LLC
26
ICEAA 2014 Professional Development & Training Workshop
52
INT 02 - Advanced Schedule Analysis
BP 8: Conduct a Schedule Risk Analysis






Has a schedule risk analysis (SRA) been conducted?
Has that risk analysis led to either risk mitigation?
Ask to see a schedule risk analysis if it has been done
Are there buffers for time contingency? Is it based on the SRA?
Sometimes there are vestiges of a risk analysis or partial risk
analysis or perhaps the beginnings of a risk analysis that was
not completed
This element informs the program that GAO expects them to do
it in the future
53
© 2014 Hulett & Associates, LLC
Risk Along a Contiguous Schedule Path

Path risk is the combination of the risks of its activities
Start
Design
Unit
Build
Unit
Test
Test
Unit
Unit
Finish
© 2014 Hulett & Associates, LLC
27
ICEAA 2014 Professional Development & Training Workshop
54
INT 02 - Advanced Schedule Analysis
Really Simple Schedule

This schedule finishes on September 3
–

7-day weeks, like a model changeover, refinery turnaround
If we can get into trouble with this simple schedule, we can get into
trouble with real project schedules
© 2014 Hulett & Associates, LLC
55
Add Duration Risk to the Schedule using
Triangular Distributions
This section features Primavera Risk Analysis, formerly Pertmaster,
owned by Oracle
© 2014 Hulett & Associates, LLC
28
ICEAA 2014 Professional Development & Training Workshop
56
INT 02 - Advanced Schedule Analysis
Risk Analysis using Monte Carlo

Monte Carlo Simulation of the schedule - 50-year old
method
– A simulation is made up of many (thousands) iterations
– Each iteration uses a new set of activity durations chosen at
random from the probability distributions
Iteration is just a CPM analysis using those durations
Iterate many times to reflect uncertainty in duration estimates
Collect the data, make probability distribution of results
–
–
–
 CPM is not even the most likely completion date and
may not be very likely, given risks
57
© 2014 Hulett & Associates, LLC
Monte Carlo Simulation Results for Really
Simple Schedule
CPM date is not even the most likely – That’s about 9/13
Entire Plan : Finish Date
100% 19 Oct 11
240
95% 29 Sep 11
220
90% 25 Sep 11
85% 23 Sep 11
200
80% 21 Sep 11
75% 19 Sep 11
180
70% 18 Sep 11
160
60% 15 Sep 11
Hits
55% 14 Sep 11
50% 13 Sep 11
120
45% 11 Sep 11
100
40% 10 Sep 11
35% 09 Sep 11
80
Cumulative Frequency
65% 16 Sep 11
140
80% Target is
9/21
30% 07 Sep 11
25% 06 Sep 11
60
20% 05 Sep 11
40
15% 03 Sep 11
10% 01 Sep 11
20
5% 29 Aug 11
0
0% 17 Aug 11
29 Aug 11
18 Sep 11
08 Oct 11
Distribution (start of interval)
CPM date is about 16% Likely to be met
© 2014 Hulett & Associates, LLC
29
ICEAA 2014 Professional Development & Training Workshop
58
INT 02 - Advanced Schedule Analysis
BP 9: Update (Status) the Schedule





Is the project correctly statused?
Are there actual dates in the future?
Are there activities or work that started or finished in the
past that do not have “actual dates?”
When activities are statused does this break the schedule
logic?
When an activity has started but is not finished, do the
actual and remaining durations agree with the Data Date?
© 2014 Hulett & Associates, LLC
59
BP 10: Baseline the Schedule



Has the schedule been baselined?
If a baseline is established, actual progress can be
compared to planned progress for variance calculations
Earned value concepts can be used to translate variances
into a new estimate of completion
© 2014 Hulett & Associates, LLC
30
ICEAA 2014 Professional Development & Training Workshop
60
INT 02 - Advanced Schedule Analysis
Summary







A schedule is a dynamic model – dates are outputs
Hammocks
Lags
Danglers
3-point estimates of durations
Out-of-sequence progress
GAO 10-point scheduling best practices
© 2014 Hulett & Associates, LLC
61
QUESTIONS?
David T. Hulett
Hulett & Associates, LLC
(310) 476-7699
www.projectrisk.com
David.Hulett@projectrisk.com
© 2014 Hulett & Associates, LLC
31
ICEAA 2014 Professional Development & Training Workshop
62
Download