See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/343289801

Matrix-based Evaluation of Project Management Approaches

Chapter · September 2017

CITATIONS READS
0 665

2 authors:

Zsolt Tibor Kosztyán Csaba Hegedus

University of Pannonia, Veszprém University of Pannonia, Veszprém
168 PUBLICATIONS 525 CITATIONS 19 PUBLICATIONS 59 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Csaba Hegedus on 21 September 2020.

The user has requested enhancement of the downloaded file.

19TH INTERNATIONAL DEPENDENCY AND STRUCTURE MODELING
CONFERENCE, DSM 2017
ESPOO, FINLAND, SEPTEMBER 11 – 13, 2017

Matrix-based Evaluation of Project Management

Approaches
Zsolt Tibor Kosztyán1,2, Csaba Hegedűs2
1
MTA-PE Budapest Ranking Research Group, Hungary
2
Department of Quantitative Methods, University of Pannonia, Hungary

Abstract: One promising application of the matrix-based methods is handling and

modeling flexible (e.g. agile) project management approaches. The goal of this
research is to model both different kinds of flexible projects by matrix-based
techniques and project managers (as agents) who follow different kind of project
management approaches. The matrix-based project planning methods allow
characterizing customer claims as score values besides the time/cost and resource
demands. This study explores different kinds of flexible project structures with
various time/cost/resource/score constraints and determines, which project
management approach produces the most successful/shortest/cheapest projects.
Keywords: Project management, agile projects, agent-based modeling

1 Introduction
Project success becomes more and more critical to business performance, however, many
projects still suffer delays, overbudgeting and even failure. CHAOS Annual Reports (see,
e.g. (SGI, 2015)) shows that flexible project management techniques like agile project
management (APM) approaches produce more successful projects, while following
APM, project duration and the budgets can be kept more easily (Chow and Cao, 2008).
The inspiration of this study was to understand why and when agile project management
approaches produce expectedly more successful projects (Research Question 1,
furthermore: RQ1). Should we combine flexible (like agile) and rigid (like traditional)
project management approaches, or not (RQ2)?

1.1 Matrix-based project planning methods

Matrix-based methods, along with network planning techniques, are mainly used for
planning and scheduling production development projects according to (Eppinger et al.,
1994; Danilovic and Browning, 2007). While the first version of DSM assumes fix logic
structures, further studies (Yassine et al., 1999; Kosztyán and Kiss, 2010a; Tang et al.,
2010) showed that uncertain relations can be specified between two tasks. According to
(Kosztyán and Kiss, 2010b) probabilities or priorities or just so called score values (of
task completion) can also be assigned to the completion of the tasks. This matrix called as
Project Expert Matrix (PEM). In this way flexible dependencies and uncertain task
completions can also be modeled, which is a key point of modeling flexible, like agile,
projects.

DSM 2017 161

Part VII: Innovating Project Management

Since, specifying the logic structure is only the first step of project planning and
scheduling, according to (Danilovic and Browning, 2007), one way of extending DSM
method is to specify multiple domains (e.g. costs, resources etc.). Since the traditional
DSM focuses on a single domain, its analysis yields solutions deemed optimal from the
point of view of that domain only (e.g. dependency between tasks). However, the
Domain Mapping Matrix (DMM) provides means to represent interactions and relations
between domains (e.g. task-resources: Who will do these tasks?).
Despite flexible task dependencies and uncertain task completions may produces a great
number of possible project structures, (Kosztyán, 2015) proposes a fast polynomial
method that determines feasible project scenarios and project structures regarding time,
resource and cost constraints. Also orders these feasible project plans by their
probabilities/importance or according to other score values. In order to handle time, cost,
resource demands and constraints, Project Expert Matrix (PEM) is extended to the
Project Domain Matrix (PDM), similarly to extending DSM to DMM. DMMs are already
used for modeling project management problems (see e.g. (Danilovic and Browning,
2007; Browning, 2014), but the PDM can model also the uncertain task completions and
uncertain task dependencies.

1.2 Modeling project managers’ decisions

In this study, we model a decision maker (project manager), who follows a given project
management approach. Three approaches are distinguished in this study: Traditional
Project Management Approach (hereinafter: TPMA), Agile Project Management
Approach (APMA), and the combination of these approaches, which is a Hybrid Project
Management Approach (HPMA). Each management approach has a toolset of project
management techniques, which is implemented as a computer algorithm (hereinafter:
agent). Several projects, such as software development projects (SDP) and new product
developments (NPD) allow you to follow different kinds of project management
approaches, because tasks (e.g. features to be implemented) can be completed both in
serial and parallel way. Implementing tasks usually prioritized, and some of them will be
postponed to a following project or subproject (e.g., into a so-called sprint or iteration).
Following a traditional project management approach the question is how much it will
cost to implement all these features? Thus, the scope is given and has to be completed,
but time and cost are convertible if necessary. These problems can be specifies as trade-
off problems (see e.g. (Demeulemeester et al., 1996; Feng et al., 2000; Azaron et al.,
2005; Tareghian and Taheri, 2006)).
The traditional project management (TPM) approach is widely supported by traditional
project scheduling methods (see (Brucker et al., 1999) for an excellent summary of
traditional methods). All of these methods based on fix logic structure or a set of
predefined alternatives (Eisner, 1962; Pritsker, 1966). In contrast to traditional
techniques, the agile approach allows and sometimes requires to restructure the project.
In the APMA the question is how many of these features can be included within the given
budget and time interval? The goal is to realize the scope at the highest possible degree.
APMA and TPMA seem to be different approaches, however the real SDPs and NPDs
frequently combine the TPM and APM techniques. The combination of APM and TPM

162 DSM 2017

Z. T. Kosztyán, C. Hegedűs

methods (see e.g. (Fernandez and Fernandez, 2008; Rahimian and Ramsin, 2008; Tyagi
et al., 2014)) is often called as a hybrid project management (HPM) approach.
A HPMA combines traditional and agile approaches. This approach can also be modeled
by computer algorithms. Generating different kind of flexible project structures and using
different kind of agents (which simulates the project management approaches through the
decisions of project managers) may help us to answer: Why following agile project
management approaches produces more successful projects (RQ1)? Should we combine
approaches in order to increase the chance of project success (RQ2)?

1.3 Modeling project success

Although the reliability of Chaos Reports' results are criticized by several scholars (see,
e.g. (Eveleens and Verhoef, 2010; Lech, 2013), their objective classification of project
success (SGI, 2015) became widely used by both scholars and practicing project
managers; thus in this paper this classification is used.
This study follows the classification of the Chaos Reports (SGI, 2015) that categorizes
projects into 3 classes:
1. The project is successful; if project scope is satisfied within the
time/cost/resource constraints.
2. The project is challenged; if project scope is satisfied, time, cost or resource
demands are higher than the pertinent constraints, but these demands are within
an expanded (but permissible) tolerance.
3. The project is failed otherwise.
One of the most interesting results of the Chaos Report was that the IT projects managed
by agile project management approaches was 3 times more successful than the traditional
waterfall projects. This considerable result is confirmed by other surveys and scholars
(see, e.g. (Belout and Gauvreau, 2004; Lech, 2013; Dan, 2016)).
Most success criteria (like involving customers to the development process, strong
executive support, being capable to cope with emergent requirement, etc.) require
adaptive and flexible thinking for the project management. In the agile project
management approach the completion of the project is more flexible, and project
structure can adapt to the customer's claims.
In this research we focus only the success criteria of Chaos Report. Why could be more
successful a project management approach than another ones.

2 Applied methods
Since numerous (more than 10,000) project plans are generated and explored, only fast,
exact, polynomial algorithms are used as agents. For example studies on time-cost trade-
off problems (TCTP) have been using various kinds of cost functions such as linear
(Fulkerson, 1961), convex (Lamberson and Hocking, 1970), concave (Falk and Horowitz,
1972), discrete (De et al., 1995, 1997) or stochastic (Feng et al., 2000). In this study the

DSM 2017 163

Part VII: Innovating Project Management

continuous version of TCTP (furthermore, CTCTP) is used, because this method can be
solved within a polynomial computation time.
If the cost function is linear, the problem can be solved with a very fast algorithm (see,
e.g. (Ahuja et al., 1991; Orlin, 1993), therefore in the simulation phases, linear cost
function have been considered and the Orlin’s method is used in order to accelerate the
run of the proposed traditional and hybrid project management agents.

2.1 Generated project matrices and structures

We developed a project generator to produce different types of project plans. Only the
value of complexity (number of nodes, connectivity factor (cf) and ratio of supplementary
tasks and dependencies (hereafter flexibility factor, ff)) need to be specified to create the
logic domain of a single project. The result is a n by n+4 PDM matrix, which contains an
n by n logic domain (LD), n by 2 time domain (TD), and n by 2 cost domain (CD)
respectively.
The values of time and cost demands followed [0, 20] and [0, 30] uniform distributions,
respectively.
To define a project with a desired complexity an upper triangle matrix have been created.
The size of the matrix (the number of rows) will specify the number of nodes/tasks. The
connectivity factor sij ( a ) affects the number of the dependencies in the project's logic
domain. The probability of that the j-th task follows the i-th task is specified by the
sij (a) = min( j − i − 1 + a)−1,3 ,1) function, where j>i; i,j∈ℕ⁺ and a∈ℝ⁺. If a is 1 then
the critical path contains all the tasks, lower a values give fewer task dependencies in the
off-diagonal cells, therefore we get a more parallel project structure. In this simulation
a∈{0,1} and furthermore we called a as connectivity factor (cf). cf=0 produces a more
parallelized project structure. If cf∈ℤ⁺, the algorithm produces cf number of parallel
paths. In the next step the necessity of the task implementation and the previously defined
strict predecessor-successor relations between the tasks (selected randomly) are relaxed at
a given ratio of tasks and dependencies (ff alternates between 0.05 and 0.2 with a 0.05
step), and the score values modified from 1 to a uniform distributed value between 0.5
and 1.
Table 1 shows a PDM matrix, where cf=1, ff=0.2, n=30. Diagonal values of the LD
represent the score of task completions. 1 means, that the task has to be completed, if this
value is lower than 1 the (agile) project manager can decide whether this task will be
included in the project or will be excluded from it (and will be scheduled to the next
(sub)project). The off-diagonal value 1 represents a strict dependency between tasks, if
this value (score of dependency) is lower than 1, but greater than 0, an (agile) project
manager can decide how to complete these tasks i.e. it is performed in a parallel or a
sequential way.

164 DSM 2017

Z. T. Kosztyán, C. Hegedűs

Table 1. An example of a generated PDM (cf:=1, ff:=0.2. n:=30)

PDM 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 t min t max c min c max
1 0.71 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 10.59 12.20 15.59 17.08
2 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 10.27 10.87 24.38 28.66
3 0.00 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 15.95 17.45 16.71 20.68
4 0.00 0.00 0.00 0.92 0.67 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.56 1.94 8.51 9.33
5 0.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 15.55 15.87 22.12 27.45
6 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 0.00 0.90 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 4.99 5.89 21.05 23.39
7 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.14 3.89 15.45 16.13
8 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.62 3.70 14.52 16.17
9 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.88 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.16 18.76 26.78 27.04
10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 13.71 15.68 21.07 24.74
11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.64 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 10.21 10.93 22.25 22.28
12 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.52 0.82 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 13.94 15.97 7.60 8.57
13 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 9.87 11.29 11.60 12.20
14 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 17.19 18.04 10.51 11.53
15 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.21 18.73 16.18 19.32
16 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.61 1.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 15.75 16.67 22.45 26.97
17 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 5.94 7.29 20.53 25.47
18 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.97 1.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 11.10 12.94 16.49 18.52
19 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.57 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 7.05 7.79 9.17 9.69
20 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.98 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.45 17.92 26.13 26.20
21 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.87 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.40 19.36 20.44 22.71
22 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 14.11 15.02 21.45 22.56
23 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.56 1.00 0.00 0.00 0.00 0.00 0.00 0.00 10.22 12.60 14.81 16.22
24 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.96 0.00 0.00 0.00 0.00 0.00 17.64 19.56 16.13 16.17
25 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 0.00 0.00 0.00 13.97 16.17 12.33 14.04
26 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 14.13 15.11 9.79 10.36
27 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.71 0.00 1.00 15.62 16.92 22.02 26.91
28 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.96 0.00 15.22 17.01 6.83 8.24
29 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.52 0.92 13.82 14.25 18.11 19.81
30 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 8.93 9.96 17.68 18.48

In every simulation cf ∈{0,1}, ff ∈{0.05,0.10,0.15,0.20}, n∈{30,60,90} .

However, (Kosztyán, 2015)’s method allows cycles in a project plan, we assumed that
there is no cycle in the project (therefore the LD can be reordered as an upper triangular
matrix). For the sake of simplicity only finish to start dependencies are generated
between each pair of tasks.
Two kinds of time and two kinds of cost demands are assigned to each task. For the fast
calculation linear function is assumed between time and cost demands. The normal task
demands characterized as (tmax , cmin ) values, while so-called crash task demands
characterized as (tmin , cmax ) values.
Every implemented agent produces an n by n+2 project structure matrix (PSM), which is
a special PDM matrix. In the PSM every flexible task dependencies and every uncertain
n×n
task completion is decided to include to or exclude from the project ( LD ∈{0,1} ). TD
and CD are n by 1 vectors. Therefore, the total project cost (TPC) and the total project
time (TPT), can be calculated by traditional way.
The total project score (TPS) can be calculated as follows:
n
TPS = n
∏S i
, where Si=[PDM]i,i, if [PSM]i,i=1; Si=1-[PDM]i,i otherwise. (1)
i:=1

2.2 Specifying constraints

According to (Kosztyán, 2015) the minimal/maximal time/cost/score demands can be

calculated in one step. Minimal (maximal) project duration occurs, if every uncertain task
completion and flexible dependencies are excluded (included) and the minimal (maximal)

DSM 2017 165

Part VII: Innovating Project Management

durations are assigned to the tasks. Similarly, minimal (maximal) project cost occurs, if
every uncertain task completion is excluded (included) and the minimal (maximal) cost
demands are assigned to the tasks. Since every score value of the uncertain task
completion is greater than 0.5 in this case, according to Eq. (1) the minimal (maximal)
value of the total project score occurs if every task are excluded (included).
min max
Each range of the theoretical minimal ( cx ) and maximal ( c x ) values of
time/cost/score demands ( cx ) is treated as 100%, the theoretical minimal is the 0% (see
Eq.s (2-3)).

cx % = (cx − cxmin ) / (cxmax − cxmin ) (2)

cx (cx %) = cx %(cxmax − cxmin ) + cxmin (3)

In this study cx % ∈{0.0,.0.1,..,1.0}. “x”=”c” and “x”=”t” are upper cost and time
constraints, while “x”=”s” represents a lower score constraints, which models the
minimal requirements of the customers.

2.3 Agent-based simulations for characterizing project management approaches

Three project management agents (TPMa, APMa, HPMa) are implemented to test which
approach is least sensitive for the changes of the constraints and demands. When
implementing TPMa, we used a modified – matrix-based version of – (Orlin, 1993)'s
algorithm. In this case, the project manager must follow the project plan. There is no way
to restructure the project plan or exclude uncertain (less prioritized) tasks, however, in the
range tmin , tmax task durations can be reduced in order to keep deadlines. Every time
[ ]
reduction may cause the increase of costs, however, (Orlin, 1993)’s method will give us
the local minimum of project duration (considering that every flexible dependency and
every uncertain task completion is included to the project plan) if there are feasible
solutions.
When implementing APMa (Kosztyán, 2015)’s Exact Project Ranking algorithm is used.
In this case, there is no way to reduce time demands of tasks. We assume that every tasks
scheduled with normal task demands (tmax , cmin ), however, flexible dependencies can be
resolved, while the TPS is greater than the specified score constraint (cs) uncertain task
completions can be ignored from the project.
HPMa is based on (Kosztyán, 2014)’s algorithm. This agent can also reduce the project
duration and can restructure the project simultaneously.

166 DSM 2017

Z. T. Kosztyán, C. Hegedűs

3 Simulation results
In the simulation 11,420 PDM matrices and constraints had been generated. Every PDM
matrix has been solved by the three project management agents. TPT, TPC and TPS
values are calculated for every proposed project structures. Normalized project values are
specified for feasible projects as follows:
TPC TPT TPS
c% := max min
, t % := max min , s% := max min (4)
c − cc
c ct − ct cs − c s
If we consider only the successful projects, c%, t%, s% are lower or equal than 1. Lower
values mean lower project cost, duration and lower project scores.
While lower c%, t% values are positive property of project management agent, because it
means projects can be completed with less cost/time demands than the budget/deadline,
lower project scores may produce lower customer satisfaction.
After using an Analysis of Variance (ANOVA) method, all independent variable
(ff,cf,N,xPMa) were significant, and the adjusted Eta square were 0.734. Since the
limitation of pages, just the summary table is shown (see Table 2).
The simulation results show that agile project management approaches can almost triple
the ratio of successful projects of the traditional approaches. Table 2 shows that APMa’s
advantage is more significant (RQ1) if projects are more flexible (see ff factor), this
advantage is less considerable if the project is more rigid. Table 2 also shows, that hybrid
project management agent can produce even more successful project compared to the
agile agent.
While HPMa can produces the most of the successful and the shortest projects, Table 2
shows that APMa produces the cheapest projects, and if a project can be managed by
TPMa, this approaches guarantee only that all features will be implemented.

DSM 2017 167

Part VII: Innovating Project Management

Table 2. Summary of the simulation results (c%, s%, t% are calculated only for successful projects,
f/a= feasible / all projects) and the project triangle for the successful projects
Summary table c% s% t% f/a
xPMa n ff
0.05
cf =0 cf =1
0.36 0.63
cf =0 cf =1
1.00 1.00
cf =0 cf =1
0.13 0.53
cf =0 cf =1
0.47 0.24
Project Triangle for Successful Projects
0.10 0.50 0.69 1.00 1.00 0.26 0.68 0.30 0.12
30 APMa HPMa TPMa
0.15 0.55 0.85 1.00 1.00 0.36 0.81 0.20 0.06
0.20 0.69 0.82 1.00 1.00 0.47 0.87 0.13 0.07 t%
0.05 0.37 0.65 1.00 1.00 0.18 0.68 0.46 0.13
0.10 0.52 0.74 1.00 1.00 0.29 0.81 0.30 0.09
TPMa 60
0.15 0.60 0.85 1.00 1.00 0.43 0.84 0.19 0.08
0.20 0.67 0.80 1.00 1.00 0.58 0.89 0.12 0.03
0.05 0.35 0.74 1.00 1.00 0.20 0.76 0.50 0.13 47%
0.10 0.48 0.73 1.00 1.00 0.28 0.85 0.28 0.08
90
0.15 0.57 0.83 1.00 1.00 0.46 0.88 0.14 0.03
39%
0.20 0.65 0.89 1.00 1.00 0.56 0.89 0.09 0.04
0.05 0.21 0.19 0.87 0.87 0.51 0.54 0.20 0.36
0.10 0.32 0.29 0.84 0.86 0.42 0.54 0.35 0.56 34%
30
0.15 0.41 0.39 0.84 0.83 0.40 0.49 0.43 0.55
0.20 0.47 0.45 0.85 0.85 0.45 0.47 0.50 0.64
0.05 0.25 0.23 0.87 0.86 0.46 0.56 0.26 0.51
0.10 0.38 0.36 0.88 0.89 0.47 0.52 0.45 0.71 38%
APMa 60
0.15 0.43 0.42 0.85 0.87 0.42 0.48 0.48 0.72
0.20 0.47 0.48 0.85 0.88 0.42 0.46 0.62 0.72
0.05 0.25 0.22 0.89 0.84 0.50 0.55 0.31 0.61
50%
0.10 0.36 0.35 0.88 0.88 0.44 0.51 0.43 0.77
90
0.15 0.43 0.43 0.87 0.88 0.40 0.48 0.53 0.69 86% 53%
0.20 0.49 0.47 0.88 0.86 0.42 0.49 0.64 0.77
0.05 0.35 0.55 0.94 0.92 0.14 0.39 0.69 0.59
89%
0.10 0.45 0.55 0.90 0.88 0.21 0.45 0.72 0.65
30
0.15 0.49 0.57 0.87 0.84 0.24 0.42 0.72 0.62 s% c%
0.20 0.54 0.58 0.87 0.86 0.31 0.44 0.68 0.67
0.05 0.35 0.55 0.93 0.89 0.17 0.47 0.75 0.64
0.10 0.45 0.55 0.92 0.90 0.24 0.49 0.76 0.76 100% t% c% s% f /a
HPMa 60
0.15 0.49 0.57 0.88 0.87 0.27 0.47 0.74 0.75
0.20 0.50 0.57 0.86 0.88 0.34 0.44 0.76 0.75
APMa 47% 38% 86% 53%
0.05 0.32 0.53 0.94 0.85 0.18 0.46 0.82 0.76 HPMa 34% 50% 89% 73%
0.10 0.43 0.56 0.92 0.89 0.24 0.47 0.78 0.82
90
0.15 0.46 0.56 0.88 0.89 0.28 0.46 0.81 0.72 TPMa 39% 53% 100% 18%
0.20 0.52 0.56 0.90 0.86 0.33 0.47 0.78 0.79

4 Summary and conclusion

In this paper the agents model project management approaches. Simulated projects had
been analyzed to compare; which project management approach can be adequate. We
considered flexible project plans, which can represent an SDP or an NPD project
environment. Results show that there is not a single absolutely winning strategy. In every
decision of the project manager time, cost and quality requirements have to be
harmonized. If there is no chance to deviate from the original project plan and it is not
possible to restructure the project schedule, the only possible approach is the traditional
PMa (in this case HPMa=TPMa) and the only methodological tool is some kind of trade-
off method. If the project is flexible, agile approach produces the lowest mean values of
project cost, HPMa produces the lowest mean values of project duration but TPMa
produces the highest project scores (see Table 2 and RQ1). In all cases the hybrid
approach can produce the most of the feasible project plans, therefore agile and
traditional PMa should be combined (RQ2). If the project is not flexible, the hybrid and
the traditional project management agents are identical, while the flexibility factor is
increasing the behavior of hybrid and agile approach is getting very similar (see Table 2).

168 DSM 2017

Z. T. Kosztyán, C. Hegedűs

5 Limitation and future works

This paper is focusing on single projects, however in a multi-project environment (MPE),
more than one project are running at the same time, in such case the resource sharing is
also important within the running projects. In the next study we will focus on MPEs. The
multi-project manager can also be modeled by agents and can represent different kind of
approaches similarly to the single-project management agents used here. In this latter
case the collaboration and information/resource sharing of project managers can also be
modeled.
The other way is to extend the proposed framework is to simulate changes of project
parameters and requirements, while the projects are in progress. Our hypothesis is that
the results will be similar: HPMa should produce most of the feasible and most of the
successful projects, however, it should be analyzed, whether and when we should change
from one project management approach to another one. Can a delayed, overbudget
project be managed and recovered by another project management approach or not?

6 Acknowledgement
The research is supported by János Bolyai Fellowship, Hungarian Academy of Science.

References
Ahuja, R.K., Magnanti, T.L., Orlin, J.B., 1991. Some Recent Advances in Network Flows. SIAM Review
33, 175–219.
Azaron, A., Perkgoz, C., Sakawa, M., 2005. A genetic algorithm approach for the time-cost trade-off in
PERT networks. Applied Mathematics and Computation 168, 1317–1339.
Belout, A., Gauvreau, C., 2004. Factors influencing project success: the impact of human resource
management. International Journal of Project Management 22, 1–11.
Browning, T.R., 2014. Managing complex project process models with a process architecture framework.
International Journal of Project Management 32, 229–241.
Brucker, P., Drexl, A., Mohring, R., Neumann, K., Pesch, E., 1999. Resource-constrained project
scheduling: Notation, classification, models, and methods. European Journal of Operational
Research 112, 3–41.
Chow, T., Cao, D.-B., 2008. A survey study of critical success factors in agile software projects. Journal
of Systems and Software 81, 961–971.
Dan, S.N., 2016. Success Factors That Influence Agile Software Development Project Success. American
Scientific Research Journal for Engineering, Technology, and Sciences (ASRJETS) 17, 172–
222.
Danilovic, M., Browning, T.R., 2007. Managing complex product development projects with design
structure matrices and domain mapping matrices. International Journal of Project Management
25, 300–314.
De, P., Dunne, E.J., Ghosh, J.B., Wells, C.E., 1995. The discrete time-cost tradeoff problem revisited.
European Journal of Operational Research 81, 225–238.
De, P., Dunne, E.J., Ghosh, J.B., Wells, C.E., 1997. Complexity of the Discrete Time-Cost Tradeoff
Problem for Project Networks. Operations Research 45, 302–306.
Demeulemeester, E.L., Herroelen, W.S., Elmaghraby, S.E., 1996. Optimal procedures for the discrete
time/cost trade-off problem in project networks. European Journal of Operational Research 88,
50–68.

DSM 2017 169

 Part VII: Innovating Project Management

Eisner, H., 1962. A Generalized Network Approach to the Planning and Scheduling of a Research
Project. Operations Research 10, 115–125.
Eppinger, S.D., Whitney, D.E., Smith, R.P., Gebala, D.A., 1994. A model-based method for organizing
tasks in product development. Research in Engineering Design 6, 1–13.
Eveleens, J., Verhoef, C., 2010. The rise and fall of the chaos report figures. IEEE software 27, 30–36.
Falk, J.E., Horowitz, J.L., 1972. Critical Path Problems with Concave Cost-Time Curves. Management
Science 19, 446–455.
Feng, C., Liu, L., Burns, S., 2000. Stochastic Construction Time-Cost Trade-Off Analysis. Journal of
Computing in Civil Engineering 14, 117–126.
Fernandez, D.J., Fernandez, J.D., 2008. Agile project management–agilism versus traditional approaches.
Journal of Computer Information Systems 49, 10–17.
Fulkerson, D.R., 1961. A network flow computation for project cost curves. Management science 7, 167–
178.
Kosztyán, Z., 2014. Matrix-Based Time/Cost Trade-off Methods, in: HumanCapital Borders:
KnowledgeLearning QualityLife;Proceedings Management,
KnowledgeLearningInternationalConference2014. ToKnowPress, pp. 733–740.
Kosztyán, Z.T., 2015. Exact algorithm for matrix-based project planning problems. Expert Systems with
Applications 42, 4460–4473.
Kosztyán, Z.T., Kiss, J., 2010a. Stochastic Network Planning Method, in: Elleithy, K. (Ed.),
AdvancedTechniques Computing SciencesSoftwareEngineering. Springer Netherlands, pp.
263–268.
Kosztyán, Z.T., Kiss, J., 2010b. PEM–a New Matrix Method for Supporting the Logic Planning of
Software Development Projects, in: DSM2010Proceedings 12th International DSM
Conference, Cambridge, UK, 22.-23.07. 2010.
Lamberson, L.R., Hocking, R.R., 1970. Optimum Time Compression in Project Scheduling.
Management Science 16, B–597–B–606.
Lech, P., 2013. Time, Budget, And Functionality? IT Project Success Criteria Revised. Information
Systems Management 30, 263–275.
Orlin, J.B., 1993. A Faster Strongly Polynomial Minimum Cost Flow Algorithm. Operations Research
41, 338–350.
Pritsker, A.A.B., 1966. GERT: Graphical evaluation and review technique. Rand Corporation.
Rahimian, V., Ramsin, R., 2008. Designing an agile methodology for mobile software development: A
hybrid method engineering approach, in: ResearchChallenges Information Science, 2008.
RCIS 2008. Second International Conference. pp. 337–342.
SGI, 2015. CHAOS Manifesto.
Tang, D., Zhu, R., Tang, J., Xu, R., He, R., 2010. Product design knowledge management based on
design structure matrix. Advanced Engineering Informatics 24, 159–166.
Tareghian, H.R., Taheri, S.H., 2006. On the discrete time, cost and quality trade-off problem. Applied
Mathematics and Computation 181, 1305–1312.
Tyagi, M., Munisamy, S., Reddy, L.S.S., 2014. Traditional and hybrid software project tracking
technique formulation: state space approach with initial state uncertainty. CSI Transactions on
ICT 2, 141–151.
Yassine, A., Falkenburg, D., Chelst, K., 1999. Engineering design management: An information structure
approach. International Journal of Production Research 37, 2957–2975.

Contact: Csaba Hegedűs, Department of Quantitative Methods, University of Pannonia, 10

Egyetem Str., H-8200 Veszprém, Hungary, Phone: +36 88 624955, Fax: +36 88 624524,

170 DSM 2017

View publication stats