Project Management Theory

1/21
Is There an Underlying Theory of Software Project Management?
Glen B. Alleman, MSSM
and
Annmarie Oien, Ph.D.
Abstract: Traditional project management methods are based on
scientific principles considered “normal science,” but lack a
theoretical basis for this approach. [35, 36, 69] These principles
result in linear step–wise refinement of the project’s outcomes by
applying a planning–as–management paradigm. Plans made in this
paradigm are adjusted by linear feedback methods. These plans
cannot cope with the multiple interacting and continuously changing
technology and market forces. They behave as a linear,
deterministic, Closed–Loop control system.
Using a Closed–Loop adaptive control paradigm, parallels are
drawn between this approach and agile project management
methods. From these, a comparison is made between project
management practices and the tenants of agile development
processes in terms of feedback control and emergent solutions with
a control system capable of adjusting to the changes brought about
by change of the dynamics of the process, disturbances, or some
other cause not established in the original plan.
This paper suggests that when managing in the presence of
uncertainties that create risk to project success, adaptive control
theory may be better suited as a model for project management in a
rapidly changing, dynamically evolving network of statistical
processes than traditional linear approaches.

2/21
1 Introduction
Because large-scale software projects increasingly affect the public good, the normal science
paradigm is proving insufficient to model their complexity and potential consequences. The
post normal science paradigm offer a better fit, using a robust management approach
predicated on a risk‒taking ethic. [13]
Project success is a frequent topic in the field of project management but is poorly
understood in terms of how to reach that success. [70] Since the early days of the
software industry, managing software development projects has been fraught with risk
to project success created by uncertainty. 1
While the technical content of products and
the methods used to build those products has changed over time, the fundamental issues
that determine the success or failure of a project have remained constant.
Traditional program control systems are not well suited to respond to changes
encountered on software development projects. In the developemnt software intensive
systems, research shows there are four primary root causes of project failure:
§ Unrealistic performance expectations,
§ Unrealistic cost and schedule estimates based on inadequate risk adjusted growth
models,
§ Inadequate assessment of risk and unmitigated exposure to these risks without
proper handling plans,
§ Unanticipated technical issues without alternative plans and solutions to maintain
effectiveness of the projects planned progress.
In the presence of these conditions, the success rate of software development has been
underwhelming when applying traditional methods in complex development software
development project environments. [30] The traditional, linear, step–wise approach to
software development has its roots in project management methods of the 1970’s. It
was clear then and has become cleared today, that this approach to managing projects
is inappropriate in many domains. [66, 67] What the project management literature is
missing in is an answer to the question – is there an underlying theory of project
management appropriate for complex, Software Intensive Systems development
projects? [53], [77]
A secondary question is – can a theory be constructed that is consistent with adaptive
feedforward control systems and agile development processes currently in use in
manufacturing, science, engineering, economics, biology, and ecology?
This paper describes an approach to applying theories in other domains that match the
behavioral aspects of software project management. The theory of Closed Loop
Adaptive Control Systems is one choice. Performance references, control loops, and
stochastic processes all have similar paradigms in dynamics systems and in project
management. In addition, the theory of complex adaptive systems and adaptive controls
for those systems have a similar paradigm in “agile” software development.
1 Poor management practices are one source of project failure. This paper does not address these management
processes, but instead addresses the failure modes from uncertainties that create risk encountered on the project.
Poor management is a risk, but research shows unaddressed reducible (Epistemic) and irreducible (Aleatory)
uncertainties are the primary source of project failure once management processes have been addressed.

3/21
The development of complex Software Intensive System of Systems (SISoS) 2
requires
substantial creativity and innovation as well as frameworks for engineering and
governing the development work. Predicting the outcome of the development work,
with a fixed or scarce set of resources and deadlines is difficult. When there are, external
market forces, incomplete or ill–formed requirements, and changing stakeholder needs,
there are three questions for consideration:
§ What technical development and project management methods are appropriate?
§ What theoretical aspects of project management can be applied in this
environment?
§ What gaps exist in current project management methods that can be closed to
increase the probability of success in the presence of uncertainty?
1.1 Project Management Theory
The current project management literature describes project management in terms of
initiating, planning, executing, monitoring and controlling, and closing of the project. This
literature often assumes project management takes place within the paradigm of management–
as–planning.
In this paradigm there are causal connections between management actions and project
outcomes. This view is that project management is an instrument to achieve a goal
rather than an individual organizations in their own right. 3
Feedback from this planning
process is based on after the fact variance detection. As a feedback control system, gaps
in the feedback include: delays used to correct the plans and execution before the
deviation grows too large, adaptive planning through adaptive feedback loops, and
feed–forward controls to direct the execution of the project based on inputs about future
needs of the stakeholders.
In the literature, project management methods are reduced to stable, technical, and
linear processes. 4
The impact on the project from external forces or from problems
within the project is given little attention. It is assumed in this traditional model that
“change” is an undesirable thing, when in fact change in the business systems world is
not only natural it is desirable. The conflict between “managing in the presence of
change” and “managing change” by attempting to control it is the source of many of
the gaps between traditional and agile project management.
An approach to defining a theory of project management can be found in [48]. It is
conjectured that a well–functioning bureaucracy, aided by scientific planning tools, can
efficiently deal with a project through these “normal–science” methods. This approach
2 Software Intensive System of Systems can be defined using IEEE 1471 to express the system and its evolution,
starting with the communication among the system stakeholders. A critical success factor is the evaluation and
comparison of architectures in a consistent manner to establish the framework for planning, managing, and
executing the activities needed for the system’s deployment. These persistent characteristics and supporting
principles guide the verification process of the system’s implementation compliance with the architectural
description.
3 The origins of industrial society can explain why much project management theory assumes that projects take
place within a single organization. This basic assumption is out of step with post-industrial society’s joint ventures,
and strategic collaborations.
4 Linear project management models are sometimes referred to as waterfall models. In these models it is assumed
that each phase of the project is completed in a fixed sequence, followed by the next logical phase. In the Agile
methods the linearity still exists, since the statistical processes and the resulting probabilistic outcomes are formally
addressed in the management control system.

4/21
assumes projects are carried out under conditions of complete rationality. 5
It also
assumes that projects are repetitive, with their requirements and stakeholder needs built
on existing knowledge from past performance. 6
The majority of software development projects are not conducted under conditions of
rationality. Software projects are not repetitive, stable, statistically stationary, or linear.
They are unique, driven by emerging requirements, technology, and market forces, and
contain many non–linear activities and stochastic processes. Technical development is
complex, the exact business and technical outcome is difficult to plan. The processes
used to manage the outcome may be chaotic. Projects are often subjected to forces
outside the control of the project manager, engineers, and stakeholders.
More importantly, the development and deployment of complex technical projects
creates a non–linear feedback loop between the development and the deployment
process. Once the project outcomes are deployed, the users have new and sometimes
disruptive requirements – once they know and understand how the delivered system
works.
A framework for examining this situation can be found in a similar approach in the
management of systems engineering activities. [63]
Figure 1 presents an overview of the elements and dimensions of project management.
The “control system” involved in project management is not shown, since this is a static
view of the elements and their interactions. The important aspect of Figure 1 is the
connection between the components of the problem domain and the solution domain.
As the problem grows, the linear non‒adaptive approaches to managing work in the
presence of uncertainty have lower probability of success.
5 All rational action embodies some precautionary principle. What kind of harm can be averted? What kinds of cost
are willing to be incurred by the stakeholders? In the rational context, risks can be pre–identified, production rates
are known, defects can be statistically analyzed, and requirements can be elicited up front.
6 This can be the case when Reference Class Forecasting is in place. But many times the needed information to construct
the reference class was never gathered from past projects. As well parametric and model based processes, like Agile
Function Point Analysis are not used. In the absence of this data, making informed decisions in the presence of
uncertainty will be difficult.

5/21
Figure 1 – Dimensions of Complex Project Management independent of the actual control system
used to manage the project. As complexity increases the non‒linear dynamics of the project and
its management demand methods other than traditional planning and control. Statistical process
control can address the issues using adaptive behavior feed forward control.
1.2 Managing the in Presence of Uncertainty
All project work is performed in the presence of uncertainty. Traditional project
management as well as Agile project management have no means to maintain stability
of project performance in the presence of this uncertainty. When a disruptive event
occurs, the project performance is disrupted, and the project must be re‒planned at some
level to correct the source of the disruption. The level can be at the lowest task level or
at the highest capabilities level. But no matter the level, in the presence of the
disruption, change must be made to the project’s plan.
Information is needed to manage in the presence of uncertainty and the disruptive
conditions the uncertainty creates. To manage is success, two issues must be addressed.
§ Information must be provided by the observed outcomes of the project. This is a
state estimation problem.
§ Second is the formulation of the stochastic control problem and the solution to the
problem in the form of a closed loop control system.
Three aspects of project performance management using any control system,
traditional, agile, or adaptive must provide guidance for:
§ Estimation – linear estimation, non-linear information, and uncertain information.
§ Identification – of the parameters that impact the performance of the project.
§ Control – of these parameters to maintain the desired project performance.
Program Office Scope of Interest
Balanced
Scorecard
Classification
Project Attributes
Scope (S)
Progress (P)
Behavior (B)
Program
Domain
Complexity
Programs
(Participative)
Systems
(Rational)
Projects
(Normative)
Tools
State
Transformation
Management
Value
Generation
Techniques
Flow
Management
S:Single
P:Time Boxed
B:Linear
S:Multidiscipline
P:Interdependent
B: Scheduled
S:Enterprise
P:Evolving
B:Non-Linear
Incremental
Stretch
BHAG*
Systems
Programs
Projects
Tools, Techniques, and Management
Techniques & Mgmnt
Tools
PERT
GANTT
WBS
Historical
and
Projective
Estimating
Tools
Balanced
Scorecard,
Monte Carlo
Simulations
Enterprise Project Management Platform
Earned Value Analysis Management
* Big Hairy Audacious Goals
1
3
2
Increasing Tools Complexity
Increasing
Problem
Domain
Complexity
Increasing
Project
Attribute
Complexity
1 Single purpose projects
2 Multiple Single purpose
projects
3
Enterprise Wide
Multiple or Single
purpose projects

6/21
The normative advice provided by traditional project management bodies of knowledge
– planning, execution, and control – forms a Closed–Loop linear system. This advice
is usually based on rules that specify which choices will maximize benefits to the
participants. Normative theory suggests a project is a series of sequentially related
activities.
Beyond this normative approach, project management is a set of multiple interacting
interdependent random activities behaving in a non–linear and adaptive manner. This
is an operational definition of a Complex adaptive system (CAS) that can be applied to
modeling project management activities. Adaptive control systems offer a simpler
model without the complex and intractable mathematics of CAS. The distinctions
between traditional and adaptive management can be summarized in Figure 2:
Traditional Methods Adaptive Methods
Planning drives results Results drive planning
Delivery focused on planned results Delivery focused on derived results
Defined process steps Self–adapting process steps
Figure 2 – Distinctions between Traditional and Adaptive Project Management methods. The
application of these methods is dependent on the complexity of the project and the dynamic
behaviors of those complexities.
1.3 A Focus on Information Technology Project Management
Software Intensive Systems (SIS) 7
projects traditionally use formal management
processes for the acquisition or development, deployment, and operation of the system
that emphasizes planning in depth. This approach organizes work into phases separated
by decision points. Supporters of this approach emphasize that changes made early in
the project can be less expensive than changes made late in the project.
SIS can be found in a variety of business and technical domains
§ Business information systems – the US Government is one of the large consumers
of ERP system, finance systems, logistics systems, personnel and payrolls
§ Network reliant systems – this is the traditional command and control systems
found in industry and government, where data is exchanged between physical
disparate systems, with large amounts of data used to assist humans in awareness
and decision making processes.
§ Infrastructure systems – enterprise systems of interconnected business, embedded,
or other systems. This infrastructure provides equipment and capability needed for
integrated complex systems to function properly.
§ Embedded systems – systems that interact with the physical, through sensors,
displays and human command and control for control applications in industry,
equipment, and products used to control other systems.
The embedded systems market is 100 times larger than the desktop software market. Hardly
any new product reaches the market without embedded systems any more. The number of
embedded systems in a product ranges from one to tens in consumer products and to hundreds
in large professional and industrial systems.
7 A software-intensive system is any system where software contributes essential influences on the design,
construction, deployment, and evolution of the system as a whole. [from ISO/IEC/IEEE 42010:2011].

7/21
Embedded systems are an important business domain for the application of the adaptive
project control paradigm based on Agile development processes, including capabilities
planning, programmatic and technical estimating, risk management and program performance
management. [Embedded Systems Roadmap 2002, published by the Technology Foundation
of the Netherlands (STW)]
In the past, when waterfall 8
is used as the approach to SIS, this framework contains
several erroneous assumptions that negatively impact SIS projects:
§ Planning – the assumption it is possible to produce a plan so that its implementation
is merely a matter of executing a defined set of tasks in a predefined order.
§ Plans for complex projects rarely turn out to be good enough for them to
remain intact throughout the project life cycle.
§ Continuous re–planning, re–adjusting of priorities, and re–analyzing the
consequences of these changes is common practice.
§ Unanticipated problems are the norm rather than the exception.
§ Change – It is not possible to protect against late changes.
§ All businesses face late changing competitive environments.
§ The window of business opportunity opens and closes at the whim of the
market, not the direction of the project manager.
§ Stability – Management usually wants a plan to which it can commit. By making
this commitment, they give up the ability to take advantage of fortuitous
developments in the business and technology environment [71].
§ In a financial setting this is the option value of the decision.
§ Deferring decisions to take advantage of new information and new
opportunities is rarely considered on IT projects [72].
8 The term waterfall has been used many times as a strawman by the agile community. In fact very few pure waterfall
projects exist today. This is not to say there are not abuses of the concept of waterfall – sequential development
based on the simple algorithm REPEAT [Design, Code, Test] UNTIL Money = 0. In practice,
development and deployment processes based on incremental and iterative methodologies are the norm. The
literature contains numerous references and guidelines to this iterative project management approach dating back
to the 1980’s [66].

8/21
1.4 Adaptive Control Systems and Agile Methods
In adaptive control systems dynamic characteristics are not constant because of changes to the
parameters and changes to the environment. The effects of small changes on the dynamic
characteristics may be attenuated in a feedback control system. If the changes are significant
a system must be in place with the ability to adapt. The adaption implies self-adjustment or
self-modifying in accordance with the unpredictable changes in the conditions. In adaptive
systems, the dynamic characteristics must be identified so the control parameters can be
adjusted to maintain optimum performance. Such systems accommodate the uncertainties
found in all project work, especially agile development where requirements are emerging. [56]
"To adapt,” means to change a behavior to conform to new circumstances. Intuitively,
an adaptive controller is a control system that can modify its behavior in response to
changes in the dynamics of the process and the character of the disturbances. [31]
Agile processes emphasis both the rapid and flexible adaptation to changes in the
process, the product, and the development environment [4]. This is a very general
definition and therefore not very useful without some specific context – which will be
developed below. Even agile processes are driven by linear, non-statistical algorithms
and are missing the statistical aspects of the underlying processes. To be adaptive the
control loop needs to
§ Provide control for non-linear processes
§ Adaptively tuned the control algorithm with no interruption to the controlled
process.
§ Be capable of fast response to changing conditions.
Before establishing this context, agile methods include three major attributes, they are:
§ Incremental and Evolutionary – allowing adaptation to both internal and external
events.
§ Modular and Lean – allowing components of the process to come and go depending
on specific needs if the participants and stakeholders.
§ Time Based – built on iterative and concurrent work cycles.
§ Self–Organizing – in the sense that normative guides have little to offer in terms
of structure and control. Agile methods rely primarily on heuristics and
participative processes rather than normative and rational methods and guidelines.
1.5 Project Management as a “Control System”
The general requirements for a control system start with it being absolutely stable. This is a
primary requirement. In additional to this absolute stability, the control system must have a
reasonable relative stability. That is the speed of response must be fast. The control system
must be capable of reducing errors to near zero or some small tolerable value. The
requirements of reasonable relative stability and steady-state accuracy are usually
incompatible. In designing the control system it is necessary to make the most effective
compromise between these two requirements. [56]
The vocabulary of the project management [17] is similar to that found in control
systems [43, 52]. With these terms, it will be clear that Project Management can be
modeled as a control system. With this modeling comes the ability to assess the
components of project management, the control system that provide feedback and
corrective action for maintaining the performance of the project.
Most importantly for this paper, the basis for introducing the notion of Adaptive
controls to manage projects in the presence of uncertainty and emergent behavior, often
found in Agile software development domains.

9/21
These terms include:
Project Management Process Control System
Monitoring – track, review, and regulate the
progress and performance of the project;
identify any areas in which changes to the
plan are required and initiate the
corresponding changes. [75]
Reference signal – an independent variable
(or set of variables) that defines the desired
output. The error signal is the arithmetic
difference between the reference signal and
the output signal. 9
Evaluating – an assessment of the project’s
progress to plan using some normative unit
of measure, usually money, or time.
Plant or process – is a continuous operation
or development marked by a series of
gradual changes that success one another in
a fixed way and lead toward a particular
result. An artificial or voluntary,
progressively continuing operation that
consists of a series of controlled actions or
movements systematically directed toward a
particular result or end.
Control – monitors and measures progress
against plan to identify variances and
provide corrective action, generating
feedback to the decision making process.
Controller – which uses feedback, in the
presence of disturbances, tends to reduce the
difference between the output of the system
and the reference input.
Figure 3 – Project Management and Control Systems Vocabulary. Making these connections is
the basis of applying adaptive controls in the project management domain.
2 Project Management as a Control System
Control systems play an important role in engineering, science, economics, and
biological systems. They also play an important role is creating models of other general
systems, either as models of these systems or as metaphors of the models of these
systems. [8].
Early control systems were based on linear feedback models. As the entities being
controlled became more complex, the classical control theory, which dealt with single
input and single output systems, became less useful. Multiple input and output systems
now dominate control systems theory and practice. Recently adaptive and optimal
control systems have been developed. Applications of modern control theory to non–
physical fields are also the norm. Biology, economics, sociology and other dynamic
systems are also common practice. Complex Adaptive Systems is a popular topic today.
Constructing a connection between control systems, especially adaptive control
systems and project management is the goal of this section.
2.1 Basic Problems in Control System Design
Before moving forward some comparisons between control systems and project
management systems will be helpful. Project management is a Close Loop Control
system.
9 In the case of a simple temperature controller, the reference signal is the desired temperature. The error signal is
the difference between the desired temperature and the current temperature. If this error is positive, then the
process is instructed to lower the temperature. If this error is negative, then the process is instructed to raise the
temperature. This is a very simple example but will serve to illustrate the point that project management has similar
terms and concepts as Closed–Loop controller.

10/21
A closed-loop control process assures that a system performs within control limits. In
closed-loop control, the system's output is fed back directly to change the system's
inputs. The way in which a thermostat works with a furnace to control room temperature
is an example of Closed–Loop control. Closed loop control starts with an explicit
objective (e.g., the desired room temperature), a measure of the status of the system
against that objective system (e.g., the difference between the actual and desired room
temperatures), and a mechanism for adjusting the system's inputs to correct the
difference and meet the objective (e.g., turning the furnace on or off).
Process Control Project Management
Process A natural and progressively continuing
operation or development marker by a
series of gradual changes that succeed
one another in a fixed way and lead
toward a particular result.
A step-by-step set of activities
needed to produce the
project’s outcome. Usually
performed in a linear manner
in traditional methods and
incrementally and iteratively
in agile methods.
Systems A combination of components that act
together and perform a certain
objective.
Project and product systems
are separated in traditional and
agile methods.
Disturbance A signal, which tends to adversely
affect the value of the output of a
system.
A performance outcome that
does not meet expectations.
Either cost, schedule, or
technical performance
shortfall
Feedback
control
An operation in the presence of
disturbances tends to reduce the
difference between the output of a
system and the reference input.
Project management feedback
of cost, schedule, and
technical performance in
mature approaches. Usually
only cost and schedule in less
mature.
Damping Damping is an influence within or upon
an oscillatory system that has the effect
of reducing, restricting or preventing its
oscillations.
The control points designed to
prevent chasing our tail when
project deliverables do not
meet expectations and hot
fixes are applied creating more
oscillations in the process flow
Feedback
control system
A system designed to maintain a
prescribed relationship between the
output and the reference input by
comparing these and using the
difference as a means of control.
Planned versus actual
measures on a time basis is
core to feedback. This
performance must be
Effectiveness, Performance,
Risk, and …ilities

11/21
Process Control Project Management
Closed loop
Control
Is one in which the output signal has
direct impact on the control action, as
shown in Figure 5. In a Closed–Loop
system the error signal, which is the
difference between the input and the
feedback, is fed to the controller to
reduce the error and bring the output of
the system to a desired value.
With measures of planned
versus actual, corrective
actions can be made. Closed-
loop assures a system
performs within control limits
by direct feedback of system's
output to change the system's
inputs.
Open loop
Control
Is one in which the output signal has no
direct impact on the control action, as
shown in Figure 5. In an open loop
system the output is neither measured
nor fed back for comparison with the
input. For each reference input there is
a fixed operating condition.
Planned performance and
actual performance data
assessed on a specific date to
determine the variance from
plan. With this information,
corrective actions can be taken
to return the project to its
planned performance.
Adaptive
control system
Method of control used to adapt to a
controlled system where parameters
vary or are initially uncertain.
For projects, information that
emerges from the execution of
the project is used to change to
control parameters.
Performance
index
Is a quantitative measure of the
performance, measuring the deviation
from the ideal performance? The
specification of the control signal over
the operating time is the Control Law.
Cost, schedule, and technical
performance measures used to
assess performance to plan by
compared planned to actual
performance.
Learning
control systems
Many open–loop control systems can
be converted to closed–loop control
system if a human operator is placed in
the loop. This operator compares inputs
with outputs and makes corrective
actions based on the resulting errors.
Project management should be
closed–loop, but statistical
processes usually not included
in the control loop like those
found in Learning Control
Systems
Figure 4 – Attributes of Control Systems and Project Management Systems. Not all attributes in
control systems can be found in traditional project management systems. Moving project
management to be more like close loop adaptive control systems.
Figure 5 illustrates Open Loop and Closed–Loop control systems. Only the Close Loop
Control system is applicable to managing projects. Project management is a Close Loop
control system. Some production processes can be Open Loop control as a monitoring
and reporting process.
The Open Loop control process has little value for project management but is the basis
of the Close Loop control process needed to make corrective actions in the presence of
variances in project performance.

12/21
Figure 5 – Open and Closed Loop Systems. Both can be found in the project management
domain. But only the Close Loop control system can provide indicators of performance variances
needed to take corrective actions to maintain the needed activities to arrive on or before the need
date, at or below the planned cost, and with the needed capabilities.
2.1.1 General Requirements for a Control System
Any useful control system must satisfy the following conditions:
§ The first requirement of any control system is stability.
§ In addition to absolute stability, the control system must have relative stability, that
is the speed of response must be fast and must show reasonable damping.
§ A control system must be capable of reducing errors to zero or to some small
tolerance level.
The requirement for relative stability and steady–state accuracy are actually
incompatible. The design of a control system becomes a tradeoff between these two
requirements.
2.1.2 Adaptive Control Systems
Adaptation implies the ability to self–adjust or self–modify with unpredictable changes
in conditions of environment or structure. In an adaptive control system, the dynamic
characteristics must be identified at all times so that the controller parameters can be
adjusted in order to maintain optimal performance.
Controller Process
Control
Signal
Input Output
Controller Process
Input Output
Measuring
Element
Open Loop System
Closed Loop System
Control
Signal

13/21
2.2 Basic Approach to Control Systems Design
One approach to the design of control systems, which will be useful here, is to use block
diagrams, which are pictorial representations of the functions performed by each
component of the system and the signals that flow between these components. 10
Figure
6 is a logical depiction of a Closed–Loop control system.
This system consists of two elements:
§ Block element – is the symbol of the operation performed on the input signal to
produce the output signal. The notation inside the block is usually the transfer
function of the block given as the Laplace function.
§ Error detector – produces an error signal, , which is the difference between
the reference input, and the feedback signal, . The choice of the error
signal is very important. Any imperfections in the error signal will be reflected in
the performance of entire system.
Figure 6 – A Logical Depiction of a Closed Loop Control Systems. This paradigm can be applied
to project management systems. The error signal is the difference between planned performance
and actual performance. The system under control is the baselined IMS and it’s BCWS
2.3 Adaptive Controls Design
Adaptive control is a specific type of control where the process is controlled in closed-loop,
and when: knowledge about the system characteristics are obtained on-line while the system
'is operating. Based upon refreshed information obtained during normal operation, specific
interventions in the control loop arc made in order to fulfill the control goal. Interventions can
be various, but they can be categorized as interventions obtained by changing: the signals,
parameters, and structure
In most feedback systems, small deviations in parameters values from their design
values will not cause any problem in the normal operations of the system, provided
these parameters are inside the loop. If the process parameters vary widely because of
environmental changes, then the control system will exhibit unsatisfactory behaviors.
In some cases large variations in process parameters will cause instability in non–
adaptive systems.
10 For the moment the specific notation used in Figure 6 will be ignored, since the interest is in applying control
systems theory to project management. The “functions” , , and , represent the reference,
error, and control signals respectively. These are functions of Laplace space rather than of time. For not familiar
with the Laplace transform it is defined as . By
transforming a time varying function to Laplace space it can be manipulated as an algebraic expression rather than
as a differential equation.
( )
E s
( )
R s ( )
C s
Input Output
( )
G s ( )
C s
( )
R s ( )
E s
–
+
Feedback
( )
R s ( )
E s ( )
C s
( ) ( ) ( ) ( )
st st
f t F s e dt f t f t e dt
¥ ¥
- -
= = =
é ù é ù
ë û ë û
ò ò
0 0
L

14/21
A simple definition of an adaptive control system is: a control system in which
continuous and automatic measurements of the dynamic characteristics of the process
are taken, comparisons are made with the desired dynamic characteristics, and
differences uses to adjust the system parameters – usually the controller characteristics
– or the generation of an actuating signal so as to maintain optimal system performance,
regardless of the environmental changes to the process.
Figure 7 – Adaptive Controller makes use of measurements and feedback adjustments from the
behaviors of the system in the presence of those measurements to adapt the control loop to the
emergent behavior of the system under control
To be called adaptive, some form of self–organizing features must exist. An adaptive
controller consists of the following three functions:
§ Identification of the dynamic characteristics of the process.
§ Decision making based on the identification of the process.
§ Modification or actuation based on the decisions made.
By performing these functions continuously, self–organization can take place to
compensate for unpredictable changes in the process.
2.3.1 System Behavior Identification
The dynamic characteristics of the process must be measured and identified
continuously. These measures should be accomplished through the effects produced by
the normal operation of the system. Identification may be made from normal operating
data or by the injection of test signals. Identification with normal data is possible only
when this data has adequate signal characteristics (bandwidth, amplitude, etc.) for
proper identification.
2.3.2 Decision Making In The Presence of Emergent Behaviors
Decisions are made on the basis of the process characteristics, which have an identified
or a computed performance index. Once the process has been identified, it is compared
with the optimal characteristics (or optimal performance), then a decision made as to
how the adjustable controller characteristics should be varied in order to maintain
optimal performance.
Controller
Input Output
Process
Decision
Id or PI
Measurement
–
+

15/21
2.3.3 Modification Based on Decisions Made
Modification refers to the changes of control signals according to the results of the
identification and decision processes. There are two approaches to modifying controls
signals:
§ Controller parameter modification – in which the controller parameters are
adjusted, to compensate for changes in the process dynamics.
§ Control signal synthesis – in which optimal control signals are synthesized based
on the transfer function, performance index, and desired transient response of the
process.
3 Project Management Theory as Control System Theory
With the control system theory established, let’s connect that theory with the needs of
software project management. This paper proposes the characteristics of software
development projects, especially agile software development, can be modeled using
adaptive control system theory.
3.1 Control Theory Summary
Control is a guiding a set of variables towards a common goal. Management Control
Theory may be seen as after–the–fact control or before–the–fact control. Control theory
suggests that where consequences are easily monitored, after–the–fact controls are
more effective. Where consequences are unique and hard to monitor, before–the–fact
control is appropriate.
4 Agile Project Management and Adaptive Control
What is needed now is some way to tie adaptive control theory to agile project
management. A simple approach is to compare the primary attributes of adaptive
control with agile PM methods.
Adaptive Control Agile Project Management
Identification of the desired loop
performance.
What is the project performance needed to
arrive on time, on budget, with the needed
capabilities?
Decision Making By assessing the planned outcomes against
with actual outcomes, decisions can be made
about the corrective actions needed to
maintain the planned performance.
Modification based on the decision made With this assessment information, changes to
the work processes, work intact, technical
processes, and resources can be made to
maintain the planned performance.

16/21
5 A Framework for Adaptive Project Management Processes
Traditional project management assumes linear feedback loops, stability in work process, and
no disruptive changes to requirements. Software development projects rarely possess these
attributes.
One question is are the methods described in traditional PM frameworks appropriate
for Adaptive or Agile Project Management? One place to look for traditional
frameworks is the Project Management Institute’s Project Management Body of
Knowledge®
.
First let’s look at the control block picture of the PMBOK’s functions. Figure 8
describes a simple view of PMBOK’s control elements.
Figure 8 – PMBOK Control Blocks. There is one feedback loop and two inputs to the process
under control. In PMBOK the process is based on a Plan and changes to the Plan are incorporated
in the Plan.
5.1 Gaps In Traditional Project Management
In Figure 8 there several things missing when viewed from a control systems process.
§ There is no capacity for work reference signal – the flow of control makes use of
performance reports to define the change control signal. These performance reports
have no reference signal by which create an “error” signal. The planned outcomes
are baselined, and the actual value compared to produce an error signal. But the
planned values are not assessed for the needed capabilities and the actual
capabilities of the actual values. It is not sufficient to desire a performance
measure. This desired performance measures must actually be achievable. This
missing assessment – what is our actual capacity for work, the achievable work –
is not part of the traditional project management. For project management to be
successful, the capacity for work must be part of the adaptive control loop.
§ There are multiple control signals – both plans and change control are used as a
control signal. Measuring variance from plan – set point minus measured value,
and changes to the set point are part of the multiple control signals. The coupling
between these two control signals masks the individual contributions to the control
loop. Separating these signals needed to isolate the corrective actions in the control
loop.
§ The dynamics and transfer function of each process is not specified. This includes
the sample rate and the response rate of each process. The dynamics of the systems
Control Execution Output
Plan Plans
Change
Control
Performance Reports
Controller Process

17/21
under control and the loop gain for controlling this system are not defined in the
traditional management process.
6 Bibliography
1. Alexander, Christopher, Notes on the Synthesis of Form, Harvard University
Press, 1964.
2. Alexander, Christopher, A Timeless Way of Building, Oxford University Press,
1979.
3. Agresti, New Paradigms for Software Development, IEEE Computer Society
Press, 1986.
4. Aoyama, Mikio, “Agile Software Process and Its Experience,” International
Conference on Software Engineering, 1998.
5. Ballard, Glenn, “The Last Planner System of Production Control,” thesis
submitted to the Faculty of Engineering, School of Engineering, University of
Birmingham, 2000.
6. Basili, Victor, “Iterative Enhancement: A Practical Technique for Software
Improvement,” IEEE Transactions on Software Engineering, 1(4), December
1975.
7. Bateman, T. S. and C. P. Zeithaml, Management: Function and Strategy, Irwin,
1990.
8. von Bertalanffy, Ludwig, General Systems Theory: Foundations, Development,
and Theory, George Braziller, 1976.
9. Boehm, Barry, “Getting Ready for Agile Methods, with Care,” IEEE Computer,
35(1), January 2002, pp. 64–69.
10. Charette, Robert N., “Large–Scale Project Management is Risk Management,”
IEEE Software, pp. 110–117, July 1996.
11. Christensen, Mark J. and Richard H. Thayer, The Project Manager's Guide to
Software Engineering's Best Practices, Computer Society Press, 2002.
12. Cleland, David I., Project Management: Strategic Design and Implementations,
McGraw Hill, 1998.
13. Charette, Robert N., “Large–Scale Project Management is Risk Management,”
IEEE Software, pp. 110–117, July 1996.
14. Cook, H. E., Product Management – Value, Quality, Cost, Price, Profit and
Organization, Chapman & Hall, 1997.
15. Davis, Alan M., “Fifteen Principles of Software Engineering,” IEEE Software,
11(6), pp. 94–96, November/December 1994.
16. Dempster, Beth, “Science versus Post–Normal Science,”
http://www.fes.uwaterloo.ca/u/mbldemps
17. Dorf, Richard C. and Bishop, Robert H., Modern Control Systems, Twelfth
Edition, Prentice Hall, 2011.
18. Duncan, William, A Guide to the Project Management Body of Knowledge,
Project Management Institute, 2000.

18/21
19. Earl, Michael, Jeffery Sampler, and James Short, “Strategies for Reengineering:
Different ways of Initiating and Implementing Business Process Change,” Centre
for Research in Information Management, London Business School, 1995.
20. Earl, Michael, “Information Systems Strategy: Why Planning Techniques are
Never the Answer,” Centre for Research in Information Management, London
Business School, 1995.
21. Erdogmus, H., “Valuation Of Complex Options In Software Development,” First
Workshop on Economics–Driven Software Engineering Research, EDSER–1,
May 17, 1999.
22. Flatto, Jerry, “The Role of Real Options in Valuing Information Technology
Projects,” Association of Information Systems Conference, 1996.
23. Funtowicz, S. and J. Ravetz, “Post–Normal Science: A New Science for New
Times,” Scientific European, pp. 95–97, March 1992.
24. Georgescu–Roegen, Nicholas, The Entropy Laws and Economic Progress,
Harvard University, 1971.
25. Feng, Gang, Adaptive Control Systems, Newnes, 1999.
26. Foster, Jason, James Kay, and Peter Roe, “Teaching Complexity and Systems
Theory to Engineers,” 4th
UICEE Annual Conference on Engineering Education,
7–10 February 2001.
27. Funtowicz, S. and Jerome R. Ravetz, “Post–Normal Science: A New Science for
New Times,” Scientific European, pp. 95–97, March 1992. Also in Futures,
25(7), pp. 739–751.
28. Funtowicz S, and Ravetz, “Post-Normal Science – An Insight Now Maturing,
Futures, 31:641–646, 1999.
29. Funtowicz S, and Jerome R. Ravetz, “Three Types of Risk Assessment and the
Emergence of Post–Normal Science”, in Krimsky S, and Golding D (editors),
Social Theories of Risk, Westport CT, Greenwood. Pp. 251–273, 1992.
30. Glass, Robert L., Software Runaways: Lessons Learned from Massive Software
Project Failures, Prentice Hall, 1998.
31. Adaptive Control Systems, 2nd
Edition, Karl Johan Astrom and Bjorn
Wittenmark, Prentice Hall, 1992.
32. Giglioni, Giovanni B. “A Conspectus of Management Control Theory: 1900–
1972, Academy of Management Journal, 17(2), June 1974.
33. Hallows, Jolyon E. Information Systems Project Management, AMACOM, 1997.
34. Harmsen, Frank, Ivo Lubbers, and Gerard Wijers, “Success–Driven Selection of
Fragments for Situational Methods: The S3 Model,” Design Methodology
Research Group, Department of Computer Science, University of Twente.
35. Hofstade, Geert, “The Poverty of Management Control Philosophy,” Academy of
Management Review, July 1979, pp. 450–461.
36. Howell, Greg and Lauri Koskela, “Reforming Project Management: The Role of
Lean Construction,” Eighth Annual Conference of the International Group for
Lean Construction, (IGLC–8), July 2000.

19/21
37. Jani, Arpan, “Decision Making in Software Projects: A Control Theory
Perspective,” Association of Information Systems, AMCIS 2003 Proceedings.
38. Jackson, M. C., “Towards Coherent Pluralism in Management Science,” Journal
of Operational Research, 50(1), pp. 12–22, 1999.
39. Jackson, E. T., “Teaching Project Management for the 21st
Century: Why it is
Important and What is New?” Carleton University, School Of Business
Administration, Ottawa, Ontario K1S 5B6, Canada.
40. Jones, Capers, “What it Means to be Best in Class,” Version 5, February 10,
1998.
41. Jones, Capers, Patterns of Software Systems Failure and Success, International
Thompson Computer Press, 1996.
42. Kogut, Bruce and Nalin Kulatilaka, “Strategy, Heuristics, and Real Options,” The
Oxford Handbook of Strategy (2001), Chapter 30, 2001.
43. Kogut, Bruce and Nalin Kulatilaka, “Strategy, Heuristics, and Real Options,” The
Oxford Handbook of Strategy (2001), Chapter 30, 2001.
44. Kogut, Bruce and Nalin Kulatilaka, “What is Critical Capability?” Reginald H.
Jones Center Working Paper, Wharton School, 1992.
45. Koskela, Lauri and Greg Howell, “Reforming Project Management: The Role of
Planning, Execution, and Controlling,” Ninth Annual Conference of the
International Group for Lean Construction, (IGLC–9), 2001.
46. Koskela, Lauri, “We need a theory of construction,” Berkeley–Stanford CE&M
Workshop: Defining a Research Agenda for AEC Process/Product Development
in 2000 and Beyond. Stanford, 26 – 28 Aug. 1999. Berkeley. University of
California; Stanford University, 1999.
47. Kuhn, T. S., Structure of Scientific Revolutions, Chicago University Press, 1962.
48. Lewis, Marianne, M. Ann Welsh, Gordon E. Dehler, and Stephen G. Green,
“Product Development Tensions: Exploring Contrasting Styles of Project
Management,” Academy of Management, 2002.
49. Luks, F., “Post–Normal Science and the Rhetoric of Inquiry: Deconstruction
Normal Science?”, Futures, 31(7), pp. 705–719, 1999.
50. May, Lorin J., “Major Causes of Software Failures,” Crosstalk, July 1998.
51. Maciarello, Joseph A. and Calvin J. Kirby, Management Control Systems: Using
Adaptive Systems at Attain Control, 2nd Edition, Prentice Hall 1994.
52. McCarthy, Dan, “Normal Science and Post–Normal Inquiry,” University of
Waterloo, Waterloo Ontario.
53. Morris, Peter W. G., “Science, Objective Knowledge, and the Theory of Project
Management,” ICE James Forrest Lecture, 18 March 2003.
54. Morris, Peter W. G., “Researching the Unanswered Questions of Project
Management,” Project Management Research at the turn of the Millennium,
Proceedings of PMI Research Conference, June 2000, pp. 87–101.
55. Nelles, Oliver, Nonlinear Systems Identification: From Classical Approaches to
Neural Networks and Fuzzy Models, Springer Verlag, 2000.

20/21
56. Ogata, Katsuhiko, Modern Control Engineering, 4th
Edition, Prentice Hall, 2002.
57. Osterweil, Leon J., “Software Processes are Software Too,” Proceedings of the
9th International Conference on Software Engineering (ICSE 1987), pp. 2–13,
March 1987, Monterey, CA.
58. Osterweil, Leon J., “Software Processes Are Software Too, Revisited,”
Proceedings of the 19th International Conference on Software Engineering
(ICSE 1997), pp. 540–548, May 1997, Boston, MA.
59. Pajares, Frank, “The Structure of Scientific Revolutions,” Outline and Study
Guide, Emory University, http://www.emory.edu/EDUCATION/mfp/Kuhn.html.
60. Potters, Marc, et. al. “Financial Markets as Adaptive Ecosystems,” May 31, 2001.
arXiv:cond–mat/9609172.
61. Ravetz, Jerome R., “What is Post–Normal Science,” Futures, 31(7), pp. 647–653,
1999.
62. Ravetz, Jerome R. and Silvio Funtowicz, “Post–Normal Science: An Insight now
Maturing,” Futures, 31(7), pp. 641–646, 1999.
63. Rechtin, System Architecture: 2nd
Edition, CRC Press, 2000.
64. Rockart, J. F. and C. V. Bullen, “A Primer on Critical Success Factors,” Center
for Information Systems Research, Working Paper No. 69, Sloan School of
Management, MIT, 1981.
65. Rockart, J. F., M. Earl, and J. Roos, “Eight Imperatives for the New IT
Organization,” Sloan Management Review, Fall, 1996, pp. 43–56.
66. Royce, Winston W., “Managing the Development of Large Scale Software
Systems,” Proceedings of IEEE WESCON, pp. 1–9, August 1970.
67. Royce, Walker, Software Project Management, Addison Wesley, 1998.
68. Selfridge, Oliver G. (editor), Adaptive Control of Ill–Defined Systems, Plenum
Publishing, 1984.
69. Shenhar, A. J. and D. Dvir, “Toward a Typological Theory of Project
Management,” Research Policy, 25(4), pp. 607–, 1996.
70. Shenhar, A. J., Levy, O., and Dvir, D., “Mapping the Dimensions of Project
Success,” Project Management Journal, Volume 28, Number 2, June 1997.
71. Sullivan, Kevin, P. Chalasani, S. Jha, and V. Sazawal, “Software Design as an
Investment Activity: A Real Options Perspective,” in Real Options and Business
Strategy: Applications to Decision–Making, edited by Lenos Trigeorgis, Rick
Books, 1999.
72. Szulanski, Gabriel, “Unpacking Stickiness: An Empirical Investigation of the
Barriers to Transfer Best Practices Inside the Firm,” INSEAD Study, Academy of
Management Best Paper Proceedings, pp. 437–441, November 1995.
73. Tao, Gang, Adaptive Control Design and Analysis, John Wiley & Sons, 2003.
74. Thorburn, W. M. “The Myth of Occam's Razor,” Mind 27:345–353, 1918.
75. Project Management Body of Knowledge, 5th
Edition, Project Management
Institute.

21/21
76. Eduardo, C., Sato, Y., Dergint, D. E. A., and Hatakeyama, K. “Project Based
Organizations as Complex Adaptive Systems,”
77. ISO/IEC/IEEE 42010, Systems and Software Engineering – Architectural
Description, 2001-12-01

Project Management Theory

More Related Content

What's hot (20)

Similar to Project Management Theory (20)

More from Glen Alleman (20)

Recently uploaded (20)

Project Management Theory