Resource Allocation




Some definitions
Resource allocation, loading, leveling
Expediting and crashing projects
Goldratt’s “Critical Chain”
9-1
Some Definitions

Resource allocation permits efficient use of
physical assets



Within a project, or across multiple projects
Drives both the identification of resources, and
timing of their application
There are generally two conditions:


“Normal”
“Crashed”
9-2
Normal and Crashing


Normal: Most likely task duration, like
“m” in Chapter 8
Crash: Expedite an activity, by applying
additional resources



Specialized or additional equipment
More people (e.g., borrowed staff, temps)
More hours (e.g., overtime, weekends)
9-3
No Free Lunch: Crashing
Creates a Ripple Effect


Crashing buys time, but nothing comes free
Potential cost areas





Additional equipment/material
Extra labor
Negative effects on other projects
Reduced morale, from excessive hours/shifts
Lower quality, from the pressure of time,
inexperienced and tired staff

“If you want it bad, you’ll get it bad . . .”
9-4
Case: Architectural
Associates, Inc.


Projects uniformly run late, thus over
budget
Is that the problem, or just the
symptom?
9-5
Case: Architectural
Associates, Inc. (cont’d)

PROBLEM: Deterministic task schedules
cause workers to plan to meet schedule –
nothing more, nothing less


Parkinson’s Law: “Work expands to fill the time
available.”
RESULT: Issues arising early in each task can
be worked around, but late-occurring issues
can’t be absorbed in schedule

And most issues do arise late
9-6
Case: Architectural
Associates, Inc. (concluded)

The Solution:


Use probabilistic time estimates (optimistic,
pessimistic, most likely)
Have staff schedule work for effectiveness
and efficiency, not just to fill x-number of
days
9-7
When Trying to Crash a
Project . . .

Two basic principles

1. Generally, focus on the critical path



Usually not helpful to shorten non-critical
activities
Exception: When a scarce resource is needed
elsewhere, e.g., in another project
2. When shortening project duration,
choose least expensive way to do it
9-8
Compute Cost per Day of
Crashing a Project


Compute cost/time slope for each
expeditable activity
Slope = crash cost – normal cost
crash time – normal time
9-9
An Example (Table 9-1)
Activity
Predecessor
Days
(normal, crash)
Cost
(normal, crash)
a
-
3, 2
$40, 80
b
a
2, 1
20, 80
c
a
2, 2
20, 20
d*
a
4, 1
30, 120
e**
b
3, 1
10, 80
* Partial crashing allowed
** Partial crashing not allowed
9-10
Example (cont’d): Cost per
Day to Crash (Table 9-2)
Activity
$ Saved/
Day
a
40
b
60
c
-
d
30
e
70 (2 days)
9-11
A CPM Example, Figure 9-1
9-12
Another Approach to Expediting:
Fast-tracking/Concurrency



Different terms for similar concept
“Fast-tracking” (construction),
“Concurrent engineering”
(manufacturing)
Both refer to overlapping project phases

E.g., design/build, or build/test
9-13
CPM Cost-Duration, Figure 9-2
9-14
Fast-tracking/Concurrency
(cont’d)

Pros:




Can shorten project duration
Can reduce product development cycles
Can help meet clients’ demands
Cons:

Can increase cost through redesigns,
excessive changes, rework, out-ofsequence installation, and more
9-15
“Cost, Schedule, or Performance:
Pick Any Two . . .”



Assuming fixed performance
specifications, tradeoff areas must be in
time or cost
Time-limited or resource-limited
If all three dimensions are fixed, the
system is “overdetermined”


Normally, no tradeoffs are possible
But, something has to give . . .
9-16
Resource Loading

Resource loading: types and quantities
of resources, spread by schedule across
specific time periods


One project, or many
Identifies and reduces excess demands on
a firm’s resources
9-17
Resource Usage Calendar,
Figure 9-3
9-18
AOA Network, Figure 9-4
9-19
Modified PERT/CPM AOA,
Figure 9-5
9-20
Resource Leveling


Resource leveling minimizes period-by-period
variations in resource loading, by shifting
tasks within their slack allowances
Advantages



Less day-to-day resource manipulation needed
Better morale, fewer HR problems/costs
Leveling resources also levels costs, simplifies
budgeting and funding
9-21
Load Diagrams, Figure 9-6
9-22
Network Before and After
Resource Loading, Figure 9-7
9-23
Load Diagrams, Figure 9-8
9-24
Resource Loading Chart,
Figure 9-9
9-25
Constrained Resource
Scheduling

Two basic approaches

Heuristic



Rule-based, rules of thumb
Priority rules, tie-breakers
Optimization

Not finding an answer that works, but the best
answer
9-26
MSP Gantt with Resources,
Figure 9-10
9-27
MSP Load Diagram, Showing
Resource Conflict, Figure 9-11
9-28
MSP Load Diagram, Leveled,
Figure 9-12
9-29
Network for Resource Load
Simulation, Figure 9-13
9-30
Load Chart, Figure 9-14
9-31
Task a Decomposed, Figure
9-15
9-32
Hierarchy of Gantt Charts,
Figure 9-16
9-33
Sources and Uses of
Resources, Figure 9-17
9-34
Project Life Cycles, Figure
9-18
9-35
Flow Diagram for SPAR-1,
Figure 9-19
9-36
Goldratt’s Critical Chain

There are systemic problems that
plague project schedule performance


These problems are not randomly
distributed
If they were random, there would be as
many projects finishing early as late
9-37
Some Systemic Causes of Late
Projects

1. Thoughtless Optimism




Overpromising at project start
“Success-oriented” schedules
Lack of management reserves
2. Setting capacity equal to demand

Ignoring concepts of resource loading and
leveling
9-38
Some Systemic Causes of Late
Projects (cont’d)

3. “The Student Syndrome”



Delaying start of non-critical tasks
Parkinson’s Law: “Work expands to fill the
time available”
4. Multitasking to reduce idle time

Switching back and forth between projects
creates delays
9-39
Some Systemic Causes of Late
Projects (concluded)

5. Complexity of schedule drives delay


6. People need reason to strive


Uncertainty and complex paths join to make
trouble
There’s often no advantage seen to finishing early
7. Game playing


E.g., lower levels pad estimates, senior
management slashes them
Both can be equally arbitrary
9-40