Executing Major Projects through C ontractors
Nicholas McKenna
Engineering Systems Division
Massachusetts Institute of Technology, Cambridge MA 02142
nickmck@ mit.edu
March 18, 2005
Abstract
Project based organizational structures are utilized in many industries. The firms engaged
in these endeavors, project sponsor and contractor alike, risk both capital and reputation
in the market-place with each new project. The relationship between project sponsor and
contractor influences the outcome of the project to a significant extent. Complex and
challenging projects are made more so by the adversarial relationships that frequently
exist between the sponsor and contractor(s). This paper presents a model for examining
the influence of the contractor/sponsor relationship on the execution of a project. The
focus is on the effects of the relationship, as determined by the financial performance of
the engaged firms and key project performance indicators (schedule, budget etc), on the
degree to which the firms engage with each other and the impact this has on project
performance. Analysis of the model indicates the importance of appreciating the project's
need for effective team integration in determining the financial arrangements.
1. Introduction
Project based organizational structures are utilized in many industries and exist on
many scales. At one end of the spectrum a “project team” may simply be a few
individuals within a firm assigned to solve a specific problem. At the other extreme a
project can involve thousand of individuals, employed by dozens of firms, spread across
the globe, acting together to deliver a particular outcome over the course of several years.
Examples of the second type of project are to be found in industries such as
aerospace/defense (for weapons system development, satellites, etc) and the energy sector
(for infrastructure developments such as offshore oil and gas platforms, refineries etc).
These projects are often described as Large Engineering Projects or LEPs. One element
that tends to characterize LEPs is their use of contracted firms to effect execution. While
many small projects are executed by teams that exist within a single firm, LEPs typically
involve a number of firms being brought together, through contract structures, by the
project sponsor to execute the project.
This research investigates the role that formal contract relationships between firms
play in determining major project outcomes. The following premises frame the
investigation:
1. Many product systems being developed through major project structures (utilizing
contractors) can be identified as integral systems (as compared to modular
architectures).
2. Integral systems require significant investment in integration activities
(specification development meetings, design reviews etc) in order to be
successfully developed.
3. Motivation for sustaining investment in integration activities is developed through
relationships between agents (firms, individuals) based on trust and mutual goals.
4. The firms engaged in a project organization will act to create value (as they
perceive it) for their shareholders.
These four premises when taken together can lead to unexpected outcomes. The need
to create shareholder value can generate adversarial relationships between the project
sponsor and contractor. This damages trust based relationships and undermines the
investment in integration activities, leading to sub-optimal project execution.
This research builds a formal model to investigate the mechanisms described above.
The research adopts the methodology of a case study and adds formal model building. A
case study of a recent major project undertaken by an integrated energy company was
conducted. A formal model was then developed that captures the dynamics of project
development and includes explicitly the relationship between project sponsor and
contractor. This paper presents the formal model.
2; Conceptual Background
The study of project based organizations, and the mechanisms that drive project
performance in general, has generated a rich literature that cuts across a number of
academic disciplines including organization theory, economics, product development and
system dynamics. This research draws upon aspects of that literature. This paper presents
an overview of some of these knowledge domains and establishes the linkage between
the existent theories in the appropriate knowledge area and the assumptions outlined in
the introduction.
Premise 1: Oil and Gas platforms are integral product systems.
This research limits its investigation to projects that feature integral product
systems architecture, such as offshore oil and gas facilities. The case study investigated a
large oil and gas infrastructure project, which I assert featured a product system with
integral architecture. The relevant literature provides a number of alternative definitions
for integral architectures that support this characterization. Integral systems are those that
are “designed with the highest possible performance in mind”! and where “modifications
to any one particular component or feature may require extensive redesign”.’ Another
definition of integral systems architecture has been offered that relates to the
decomposability of the system by function: in integral architectures functions are spread
across components resulting in more complex interfaces (Sosa, Eppinger and Rowles,
2000). Finally, an alternative perspective of the product system’s architecture is provided
by consideration of the system requirements from a mass and power transportation view.
Whitney (2004) suggests that certain physical systems, typically mechanical ones that
carry significant power, are constrained from utilizing modular architectures.
The oil and gas facility developed by the project under investigation featured a
large offshore platform that incorporated production, drilling and accommodation
modules. This type of facility, in addition to producing hydrocarbons at pressures that can
exceed 10,000psi and 200F, drills for reserves at depths of over 20,000ft below the
earth’ s surface. The facility as a whole must be designed and constructed to withstand
hurricane force wind and wave loadings while remaining on station in water depths of
over a mile. An inspection of the engineering complexity involved with developing such
a system reveals an architecture that satisfies the definitions of integrality enunciated
above.
Premise 2: Integral systems require integration activities during development.
The integral nature of the systems being development has important implications
for the development process. As Novak (2001) points out “the more interconnected are
the parts of a system, the more difficult it is to coordinate development”.
Communication between, and within teams, is essential for the successful development of
complex systems. Wheelwright and Clark (1992) have emphasized the importance of
‘ Ulrich K, T., Eppinger S, D., 2000, pg 184.
* Dhid.
3 Novak s., Eppinger S, D., 2001, pp 190.
communication with respect to improved project performance. As stated by Eppinger
(1997); “To assure that the entire system works together, the many sub-system
development teams must work together”.* Communication and information sharing is
central to the development of integral systems, “team members deal with imprecise
information and so must communicate to define problems or to reach consensus on the
solution of a problem”.° The literature thus certainly supports the notion that successful
projects require significant investment in critical integration activities that are
characterized by communication and information sharing. Examples of these critical
integration activities includes inter-team meetings such as design review meetings,
systems integration meetings, specification development meetings and a host of others. It
is therefore necessary to ask; what are the requirements for establishing this investment?
Premise 3: Sustained integration efforts require team trust and mutual goals.
A number of elements are needed to support the required communication. For
example, group cohesiveness has been described as a factor in determining project
outcomes. Keller (1986) noted that “cohesive project groups were able to achieve high
project quality and able to meet their goals on budgets and schedules.”® Generating
cohesive teams requires interpersonal and inter-organizational trust. As noted by
McAllister (1995), “researchers have argued that efficiency within complex systems of
coordinated action is only possible when inter-dependent actors work together
effectively. Trust between such actors is seen as a determining factor.”’ Investigations
into the phenomena of virtual and distributed teams have also noted the importance of
trust in generating the communication that is vital for project success. A recent study by
McDonough III et al, (2001) into the use of globally distributed product development
teams noted that “low levels of trust can have detrimental affects on the quality of
communication and interpersonal relationships.”® Trust becomes particularly important
as a function of complexity. McA llister references Thompson (1967) in observing that
“under conditions of uncertainty and complexity, requiring mutual adjustment, sustained
effective coordinated action is only possible where there is mutual confidence or trust.”®
Two principle forms of trust can be described: cognition based trust, grounded in
individual assessments in relation to peer reliability and dependability, and affect based
trust, grounded in notions of reciprocity founded by personal care and concern
(McAllister 1995). In either case “reliability and dependability expectations must usually
be met for trust based relationship to exist and develop, and evidence to the contrary
provides a rational basis for withholding trust.”!° A project using contractors for
execution provides ample opportunities for expectations not to be met. For example, a
contractor falling behind schedule, or increasing the cost of a project through variation
orders (sometimes known as change orders), can be interpreted as failing to meet the
reliability and dependability expectations of the project sponsor.
“ Eppinger S, D., 1997, pp 199.
° Sosa M, E., Eppinger S, D., Pich M, McKendrick D, G., Stout S, K., 2002, pp 46.
© KellerR, T., 1986, pp 723.
7 McAllister D., 1995, pp 24.
8 McDonough III E, F., Kahn K, B., Barczak G., 2001, pp 112.
° McAllister D., 1995, 25.
° Thid, pg 26.
Premise 4: Firms engaged on the project will act to create value for their shareholders.
Projects are mechanisms for delivering value to the project sponsor and contractor
alike. The project sponsor has initiated the project in order to generate value from the
product system being developed, whether it be through the sale of the product itself, or in
the case of an oil platform, from the sale of the hydrocarbons that the product system
delivers. For most projects the ultimate NPV delivered is affected by the development
costs of the product system. Project sponsors will typically be mindful of these costs and
seek to minimize them.
Contractors create value by charging sponsors for their particular skills and
services. The cost of the project is thus determined, in part, by the cost associated with
meeting the contractor's fees. It is often assumed that contractors will, ceteris paribus,
want to maximize their profits by charging as high a fee for their services as they can.
Sponsors will naturally want to contain these costs. In creating the project enterprise the
sponsor and contractor will establish a formal contract which stipulates the contractor's
scope of work, the sponsor’s expectations and the project's financial arrangements.
What System Dynamics tell us about projects.
The field of system dynamics has been particularly engaged with trying to
understand project behaviors. The nature of large scale projects, defined as they are by
highly nonlinear relationships between components, multiple feedback processes and
dynamic environments, makes system dynamics a particularly apt approach (Sterman
1992).
The persistence of poor project performance, despite the attention lavished on it, is
often cited (Ford and Sterman 1998, 2002, Lyneis, Cooper and Els 2001). A number of
areas have heen identified as causes for disappointing project performance:
Lack of adequate front end loading."
Unrealistic schedules.
Staffing. Either inadequate or poorly timed.
Over use of overtime.
Poor governance (Miller and Lessard 2000).
Poor processes. (i.e a lack of clearly defined requirements, reviews, metrics)
The system dynamics approach to understanding project pathologies has focused on
understanding the feedback structure of projects that lead to schedule delays and cost
overruns. The idea of the rework cycle is fundamental to this approach (e.g Cooper 1980,
Abdell-Hamid 1991, Repenning 2001, Ford and Sterman 1998, 2002). A number of
assumptions have characterized the systems dynamics models: first, the tasks carried out
by the organization are essentially homogenous, or are grouped into a few distinct
categories. Essentially though, each task is not generally differentiable in terms of
complexity, time to completion and skills required. This is clearly not true in real world
projects, but at the aggregate scale required for understanding the effects of delays,
feedbacks and policy decisions the distinction proves generally unimportant. Second, the
project organization is housed “under one roof”. This is not to say that management is not
"! Front end loading refers to the process of investing early in the project in activities that allow for areas of
uncertainty to be adequately investigated and defined.
distinct from staff engineers, or that there are not distinct phases of activities in a project
(Ford and Sterman 2002, Black and Repenning 2001, Repenning 2001). Indeed a key
behavior of the projects under investigation in the system dynamics literature has related
to the impact of allocation of resources to different phases of the project. Rather the
assumption of “under one roof” relates to the notion that the project model is contained
within the boundary of one firm or enterprise. Divergent financial incentives between
actors engaged in project execution have not been explicitly included previously. This
research addresses this gap in the literature.
3. Why variation orders cause problems.
It is important to recall an assumption stated above: the firms will act to create
value (as they perceive it) for their shareholders. For the contractors the creation of value
is achieved through a variety of contractual mechanisms. The first is the agreed rates or
lump sum value of the project. The project sponsor and contractors agree a price for
provision of services, the scope of services being set out in the contract documents. A
second mechanism for deriving value from the contract is the use of variation orders.
This mechanism is provided in contracts as, for all but the most trivial of projects, there is
uncertainty surrounding the scope, particularly for complex product system development
projects. This allows for changes to be made to the contract scope and additional costs
calculated.
Contractors are able to use the variation order mechanisms to generate additional
revenue from the project. In very large and complex projects there usually exists a certain
unavoidable amount of ambiguity to the contractual terms. It is almost received wisdom
amongst project sponsors that the contractors use variation orders as a primary source of
revenue. The variation order revenue mechanism can be described by the causal loop
below.
Bwana When a gap exists between the
Variation Orers contractor's desired financial
performance and the retum achieved on
Common Maia SPECific project this leads to pressure to
el secure revenue on that project. This in
Vario Orders 8) Seams fees tum leads to pressure to use contract
‘ “fees Loop mechanisms to raise revenue. The use of
‘ variations orders (V Os) consequently
Pressure to Use Contact increases. As the number of VOs
Revenue increase, revenue is generated from the
Figure 1. Variation Order - Revenue Loop project. This helps to close the gap
between expected and delivered
financial performance.
Of course, the use of variation orders does not just deliver revenue. Other
consequences exist.
VOs generate additional project
as ally Discovery of cost for the sponsor and when used are
+ ProblensiChanges likely to reduce the level of satisfaction
with the contractor s performance. This
Comemntaton ) is easy to understand if we recognize
a AR nee that the sponsor’ s managers are typically
Silia Biles —, assessed by their ability to deliver a
Conmarication Loop project on time and on budget. VOs
Strength of Working Satisfaction with usually hamper that ability. Satisfaction
(earstip nen) ae with the contractor’s performance is
correlated with the strength of the
working relationship between the
contractor(s) and the sponsor. As the
relationship is damaged by the V Os, the
incentive to invest in trust based
processes such as communication is
diminished.
Figure 2. Variation Order - Communication
Loop
A necessary consequence of reduced communication is reduced investment in
integration activities (specification meetings, design reviews etc). In highly integral
product architectures a reduction in these activities leads to an increase in errors as fewer
of the complex interactions between sub-systems are validated amongst the sponsor-
contractor design teams. Finding the sources of variations (rework errors) earlier allows
for the reduction in variation orders. As can be seen from the reinforcing loop described
above, a consequence of using variation orders is a damaged relationship between project
teams, reduced communication and integration activities and hence more of the errors
that create variation orders! Variation orders become a link between the need to secure
revenue and a damaged relationship between project sponsor and contractor.
Overall Desired
Revenue
Tnbogration 2 Hevenne front Performance
activities Early Discovery of Variation Orders:
i Potential v
+ Problems/Changes
Contractors Marginal
Communication P + Retum On Project
with contractor : Pale?
= R emapead 4e) Secure Revenue
Variation Orders
Variation Order - + Variation Order:
Communication Loop Revemie Loop
i +
oe Satisfaction with
slationship Inde Contractor Pressure to Use Contract
Cohort bee Performance Mechanisms to raise
Revenue
Figure 3. Variation Orders: A Linking Mechanism
Variation orders don’t just impact the time devoted to integration activities via
their effect on the sponsor/contractor relationship. V ariation orders also generate
additional work, or tasks, for the project team. Each variation order, at a minimum,
requires the development of documentation to support the claim, auditing, tracking,
attendance of meetings to resolve discrepancies, meetings to determine anticipated costs
and impacts on the project schedule and budget, in addition to actually carrying out the
project tasks that are identified in the VO. Thus variation orders also impact the
performance of the project by generating additional tasks and additional resource
pressures on the project.
More work means less resources (time, people) are available to invest in time
consuming activities such as the critical integration processes. The consequence of that
remains as described earlier. Here again we see that the use of variation orders in fact
leads to, again, more errors and more variation orders. However, the impact of variation
orders does not end here.
A further consequence is that pressure builds to service this additional work load
through the acquisition of additional resources. From the contractor’s perspective the
ability to staff the project has been determined, in part, by the terms (profit margins,
value of the bid etc) agreed for the original contract. Bringing more personnel onto the
project requires a budget to support that decision. This can lead to additional pressure on
the contractor’ s project team to deliver revenue to help pay for the additional resources
the variation orders generated. It is clear that once we put all of these feedback structures
together the decision to use variation orders has a number of consequences for the
execution of the project.
Integration Revenue from Performance
ates Ea Disowvery of Vartan Ores
Potential v
ProblemsiChanges
Defects/Reiori
Contractors Marginal
i 4 + Retum On Project
ae ~ Pressure To
aR Contract 4s) Secure Revenue +
Variation Orders
Vatation Order i Vasiaton Order Available Project
Commanicaton Loop Revenue Loop Bane
Effort Directed to -
i i ‘Budget Required
Project Execution _ Strength of Working _— 4 .
‘Activities Relationship Satisfaction with 7 ForHog
* Contractor Pressure to Use Contract,
(Relationship Index) _ on
+. Performance nets 1
Revenue
R Stating Pressure AB) Resource ising
Gs ;
Vasaton Orie
4 v mn Order pF
Effort Loop Resourves Loop Project Staff”
Effort Directed to
Managing Variation ‘ Desired Resources
Orders Additional Resources to for Project
‘Manage Variation 4
Orders
Figure 4. Variation Order Feedback Mechanisms
4. Project Model
4.1 Overview
A key structure in most project models is the rework cycle’. This structure was
first developed by Pugh-Roberts Associates in relation to the Ingalls Shipyard claim and
it has subsequently been revised and refined through many different applications (A bdell-
Hamid 1991, Repenning 2001, Ford and Sterman 1998, 2002). The rework cycle
constructed for this paper is illustrated below. It differs from the structure employed by
Ford and Sterman (1998), amongst others, in eliminating the stock of Unknown or
Undiscovered Rework. This removes a delay in processing and executing the rework
tasks, making the project model more efficient (thus making the model conservative in its
behavior) while maintaining the essential feature of distinguishing between work to do,
work completed correctly and rework.
The model adds the variation order cycle. Tasks with defects move either to the
stock of Task Rework or Variation Orders Submitted. The Variation Order Generation
rate establishes the percentage of tasks with defects that will be resolved through the
formal variation order process. Variation orders are then approved as rework tasks and
move to the stock of Task Rework. A certain percentage of these Variation Orders
Approved as Rework Tasks also generate new tasks which enter the stock of Project
Tasks to Do at the rate of V.O New Task Generation Rate.
” Lyneis J.M., Cooper K.G., Els S.A., 2001, pp 245.
Time Invested in
+ Integration Activities
a
Strength of ~ - Variation Orders
Sponsor/C ontractor Defects Apaiovell gs Rework
i i Variation Orders \
Relationship Sik
of
V.O. New Task - Variation Order +
Generation Rate Generation
Tasks Not Approved,
sent for Rework
Project Tasks
toDO Pa Tasks Completed Task Rework
Task Completion for Approval
Bae Task Rework Rate
Task Approval
Rate
Y
Tasks Completed.
Figure 5. Contract Model Rework Cycle
The variation order cycle captures the process whereby a certain percentage of
tasks that are identified as rework will generate claims for reimbursement as variation
orders. As discussed above, no contract can completely specify the tasks to be performed
and consequently some rework tasks can be subject to claims (variation orders for more
money and time associated with a task that now appears more complex than originally
thought, for example) by the contractor. In addition some of these variation orders
generate new tasks that had not previously been within the contractor’s scope (discovery
of the rework, and subsequent variation order claim, may also uncover gaps in the work
scope that need to be filled by including new tasks).
A further important feature of the model is the linking of integration time to the
New Work Defects Fraction. In previous models, defect or error rates are typically
determined by variables such as staff morale, fatigue, experience and schedule pressure.
The concept is that unmotivated, tired, inexperienced or harried staff makes mistakes in
executing the tasks leading to defects. These phenomena are well understood and
represented in numerous project models. The model presented here captures the idea that
a critical determinant of project success for complex systems is communication. When
teams in a complex project do not invest in integration activities (meetings, design
reviews, timely transfer of design specifications etc) elements of the project design
diverge and errors, or lack of fit constraints, are introduced. Thus the New Work Defects
Fraction is a function of the Fraction Time on Integration.
Ds =f(Ti), where Dy is the New Work Defects Fraction
T; is the Fraction Time on Integration
Variation orders also directly impact the relationship between project sponsor and
contractor and the financial performance of the project (each variation order represents a
claim by the contractor for more money). The model measures the financial performance
of the contractor and this determines in part the Percent of Rework Tasks submitted as
Variation Orders by Contractor, Desired Full Time Staff and the Initial Full Time Staff.
Completing the key structural elements of the model is the Relationship Index. This
composite variable captures the strength of the working relationship between the sponsor
and contractor and is a function of the Sustained Schedule Pressure, the Actual Staff to
Planned Staff Ratio and the Actual VO to Expected VO Ratio.
Indicated RI = 1/(VO Pressure* Sustained Schedule Pressure* Actual Staff to
Planned Staff Ratio)
Where VO Pressure =f (Variation Order Invoices/Expected Variation Orders)
The Relationship Index variable then determines in part the Fraction Time on
Integration and Percent of Rework Tasks submitted as Variation Orders by Contractor.
These relationships model the reinforcing loop “Variation Order - Communication
Loop”.
The model is simplified with respect to some of the feedback structures developed
in previous project models. The effects of overtime and staff fatigue are not included, and
nor are issues related to inexperienced staff (a “rookie” fraction). Schedule pressure
however is included. Leaving out some of these well understood mechanisms is thought
to, if anything, minimize the effects of the variation order- communication loop.
Including the effects of fatigue and staff inexperience would only exaggerate the effects
of poor communication on project performance.
4.2. Model Assumptions
The model is based on the assumption that the relationship between the project
sponsor and the contractor is established by the lump-sum contract agreed to at the outset
of the project. In other words, the model does not capture inter-firm relationships that
include cross-ownership mechanisms, profit sharing structures, joint-ventures, long-term
supplier relationships or similar. It is quite common for project sponsors to use a
competitive bidding process for lump-sum contracts as a means to select contractors. This
approach has been thought of as a mechanism through which the project sponsors can
control project risk and minimize the cost of the project, enhancing expected NPV.
The model includes a number of assumptions which reflect policy decisions made
by both the contractor and project sponsor. These assumptions are critical in determining
the model behavior and reflect the insights gained during interviews with senior project
managers conducted for the case study. A key assumption is that contractors will devote
time and resources to variation orders ahead of other project tasks. Variation orders
represent an opportunity to generate additional income for the contractor, over and above
the agreed contract. It follows that resources will be devoted to these activities as a
prionity:
Task Development Define Capacity = Work Capacity from Full Time Resources-
VO Generation Effort Drain
Where Work Capacity from Full Time Resources is the number of design
tasks/week that the contractor can attend to based upon resources (number of engineers)
available, their productivity, the level of effort each task requires and the current work
week. The capacity to do development tasks (Task D evelopment Define Capacity i.e the
design and engineering associated with completing the project) is determined by first
diverting resources to generating variation orders.
While generating variation orders are a priority for the contractors, the sponsor
does not approve them instantaneously. Each variation order is tracked, audited and
deliberated over by both the contractor and the project sponsor. This all takes time and
during this time resources are devoted to the process by both parties.
The model divides the remaining work capacity between integration activities and
task completion. Integration capacity (C;) is determined by the Fraction Time on
Integration. As discussed previously the amount of time devoted to integration activities
is a function of both the schedule pressure the contractor is under and the strength of the
working relationship.
Fraction Time on Integration = MIN(1, Ideal Fraction Time on
Integration* Integration Time Multiplier From Sched Pressure* Integration Time
Multiplier from RI)
It is assumed that when staff are under pressure to produce work they will cut
back on the time they spend attending meetings and design reviews (integration
activities) in a bid to work on “productive tasks”, such as producing the deliverables
specified in the contract. The strength of the relationship between the project sponsor and
contractor also determines, in part, the time devoted to integration activities. When the
relationship between project teams deteriorates (whether in response to schedule
pressure, or rising numbers of variation orders) the individuals in those teams are less
willing to spend time with each other. Thus a poor relationship leads to decreasing time
spent in integration activities. In addition, a poor relationship generates willingness to use
variation orders. If the relationship has become adversarial between the sponsor and
contractor then the contractor will feel justified in trying to “squeeze” the sponsor for
more money. The remaining capacity to do work, the Task Completion Capacity (Cro), is
divided between New Task Capacity, Rework Capacity and Task Checking Capacity.
Naturally, the contractor will hire/allocate staff onto the project in response to
schedule pressure. However, financial considerations also determine the contractor’ s
response to schedule pressure. A contractor feeling financial pressure will not be as
willing to shift staff onto the project or hire externally.
Finally, the project model is initialized with a number of benchmarked
parameters. These include Ideal Fraction Time on Integration, Benchmarked Percentage
of Rework Tasks that lead to Variation Orders and Typical New Task Correct Fraction.
These initial conditions reflect the assumption that sponsors and contractors will enter a
project with a track record of experience behind them. This experience leads them to
have expectations of what a project will require in terms of time devoted to integration,
how many variation orders they expect and what percentage of tasks will need to be
reworked.
5. Analysis and Results
The model was simulated for a range of exogenously determined agreed
engineering rates (dollars paid per engineer-hour worked). The model uses endogenous
rates of $100/eng*hr as the contractor's preferred rate and a break-even figure of
$70/eng*hr. The range of agreed rates spanned from below the contractor's cost
($60/eng*hr) to very healthy profits ($130/eng*hr). The model calculates a lump sum
cost for the project based on the agreed rate and an endogenously calculated estimate of
the staffing required for completion of all project tasks within schedule. This lump sum
cost represents the contract value agreed to by the project sponsor and contractor. Total
project costs for the sponsor include the lump sum costs and the variation order costs
generated during project execution. Variation orders require resources which are
frequently priced at a different rate from the agreed lump sum rate and the model reflects
this by setting the Variation Order Engineering Rate at $150/eng* hr. The sponsor’ s
project cost does not include the cost of lost revenue incurred by project delays. In reality
the “costs” of a delayed new vehicle launch, or delayed production from an oil and gas
facility, may far outweigh the costs associated with reimbursing the contractors.
Initially the model was run without the Relationship Index influencing contractor
behavior. i.e. no impact on Fraction Time on Integration or Percent of Rework Tasks
submitted as Variation Orders by Contractor. The first set of simulations, entitled
“Contractors Want Profits”, did include pricing effects on the contractor's policies for
establishing the initial staff level on the project, hiring staff and use of variation orders.
As can be seen below the overall project cost to the sponsor is greater than the agreed
lump sum across all agreed engineering rates. This reflects the fact that a certain
percentage of tasks will be completed with errors (established by the Typical New Task
Correct Fraction), and of these errors a certain percentage will generate variation orders.
This is the case no matter what the agreed rate for no project is perfectly specified or
perfectly executed. However, the slope of the project cost line is reduced as the “cheaper”
lump sum contracts generate more variation orders.
Project Sponsor Costs
18
16 =
= 14
Bp eS =a
= 10 =a
% 8 a
4
Bg
2
0
60 70 80 9 100 110 120 130
Engineering Rate
—e—Lump Sum Cost —»—Contractors Want P rofits
Figure 6.
Of course the pricing effects also included impacts on staffing levels which
contributed to the project finishing later for cheaper contracts (contractors use less staff
and will delay hiring new staff when they are squeezed by a cheap contract). The effects
of smaller than ideal staff, and delayed hiring, on project performance has been well
documented, and the results from this model are consistent with previous efforts. The
project was originally scheduled to be competed in 100 weeks, however the cheapest
contract ends in week 182, while the most expensive project finishes in week 123.
Expected Variation Order Revenue
45M
15M
0 2 30 % T00 125 150 175 200 25 750
Figure 7.
Including the effects of the Relationship Index (the simulation entitled
“Relationships Matter”) on the contractor’ s desire to use variation orders, and their
investment in integration activities, changed the sponsors project cost. Again we see
(Table 1 below) that projects under all contract price ranges are affected by the inclusion
of these feedbacks. However, the cheaper lump sum contracts exhibit stronger affects.
Agreed Project Project Completion Project Completion
Contract Rate| Sponsor Cost| Sponsor Cost) Date (week) | Sponsor Cost] Date (week)
($M) ($M) ($M)
Lump Sum Contractors | Contractors | Relationships | Relationships
($/eng*hr) Cost Want Profits | Want Profits Matter Matter
60 6.47 10.85 182 22.09 243
70 7.549 11.59 162 21.92 213
80) 8.627 12.25 148 21.5 191
90) 9.705 12.87 136 21.2 176
100 10.78 13.59 127 20.88 162
110 11.86 14.49 125 21.28 158
120 12.94 15.43 124 21.74 155
130 14.01 16.38 123 22.27 151
Table 1
Projects executed under higher priced contracts still experience cost growth which
can be explained by considering the following. All the projects in the simulation
experienced some delays and consequent schedule pressure, as discussed above. Schedule
pressure is an input into the Relationship Index and thus even those projects executed
under high price contracts experience some degradation of the relationship between
sponsor and contractor (indeed it could be argued that the fact that the sponsor is paying a
premium price would lead to increased antipathy towards a contractor who is behind
schedule). The “Relationships Matter” simulations incorporated the effect of this
degraded relationship on the use of variation orders and time spent on integration
activities. Irrespective of the contract price, a reduction in time spent on integration leads
to more defects and more variation orders, further degrading the relationship.
It is most interesting to note however the cost performance of the inexpensive
contracts relative to the more expensive ones. The model demonstrated price dependent
“tipping point” behavior. The cheaper contracts featured higher levels of cost growth
associated with variation orders (see Figure 8 below) and consequently had significantly
higher total project costs. Cheaper contracts ended up costing more than the expensive
alternatives. This is especially true if the cost of the delayed revenue from the product
being developed is included (See Figure 8).
Project Sponsor Costs Project Sponsor Costs
2 ——— SS i. ee
Cost ($ Million)
6 70 8 90 100 110 120 130
Engineering rate (s/eng*hr) rr ar a
[=a Relationships Matter
Engineering ate (Seng)
Figure 8.
The delay cost was calculated by assuming a project capital recovery
period not exceeding the project development time. i.e the revenue
from the product system being developed by the project would generate
the cost of the project in under 100 weeks. Each week delay therefore
incurred a significant penalty in lost eamings
Devoting sufficient time to integration activities is critical for the success of
projects developing complex product systems!’. In the model this notion was captured by
the Ideal Fraction Time on Integration variable. The initial value for this established the
required amount of time to be devoted to integration activities. In addition the Ideal
Fraction Time on Integration can be thought of as a rough proxy variable for the
complexity of the product system being developed: complex systems will require
proportionally more time to be devoted to systems integration than will simpler systems.
Carrying out a sensitivity analysis on the project model from the perspective of varying
levels of product system complexity revealed the effects system complexity has on the
project. As can be seen in Figure 9 below, projects developing more complex systems (i.e
with higher ideal integration time fractions) showed more variance in final project costs
as a function of agreed engineering rate. The projects developing simpler systems had
price outcomes more tightly clustered.
*3 Sosa M, E., Eppinger S, D., Pich M, McKendrick D, G., Stout S, K., 2002,
Project Sponsor Cost: A function of Complexity
and Price
25
e 24
&
z 23 —e— $60/eng*hr
# 22 —a— $70/eng*hr
8 21 ~» $80/eng*hr
¥ —s— $90/eng*hr
i 20 —*— $100/eng*hr
a 19 $110/eng*hr
18 —+— $120/eng*hr
0.1 0.2 0.3 0.4 0.5 —— $130/eng*hr
Integration Time Fraction
Figure 9.
6. Discussion
The results described above carry with them certain assumptions. It is useful to
restate these before considering more broadly the implications of the results. First, that
the model simulates a project developing a complex product system that is highly integral
in nature. Second, the development of such a product systems requires significant
investment in integration activities by the firms engaged in its delivery. Third, that the
motivation for the investment in integration is developed through relationships based on
trust and mutual goals. Finally, that the firms engaged in a project organization, sponsor
and contractor, act to create value for their shareholders by taking what they perceive as
the appropriate revenue enhancing, or cost reducing, actions. Linking these assumptions
together generated the causal structures described in section three. Modeling these
relationships in a system dynamics model and applying the motivation of financial self-
interest to each of the firms engaged in the project (sponsor and contractor) allowed a
number of findings to become evident.
6.1 Findings
The key findings are:
1, Projects developing complex integral product systems display price
sensitive “tipping-point” behavior.
2. Complex projects (those requiring significant integration efforts) are more
sensitive to price driven behaviors than simpler architectures.
The price sensitive tipping point, and the influence of product complexity, can be
seen clearly in the following figure. The plot shows a three dimensional map of the
project space with project costs on the vertical axis. This cost varies as a function of both
the agreed contract rate and the project complexity. Projects developing products with
“low” complexity (i.e. the leading edge of the plot below) show much less variability in
cost outcomes as a function of the engineering rate. By way of contrast the development
of “high” complexity systems (the far edge of the plot) shows marked variability and a
clear point of inflection.
Project Sponsor Cost: A function of Complexity and Price
Project Cost ($ million)
Project
Complexity
(isow,
05=High)
Figure 10.
The results from the sensitivity analysis showed that projects operating below the
contractors “preferred” returns demonstrate more variance of outcomes. This volatility
suggests that projects operating thusly are more likely to generate undesirable behavior in
the face of perturbations such as late changes and the like. In addition, and quite
intriguingly, the results indicated that projects which are developing highly integral
product architectures are more susceptible to the dynamics investigated than simpler
systems. This finding has a number of implications for the design of project
organizations. For example, the establishment of the project organization is frequently
carried out without detailed reference to the complexity of the underlying product
systems (in at least that while the contracts are written to ensure that the project teams are
established with the technical requirements considered, the financial aspects of the
complexity are treated separately). The results indicate that this can have clear and
detrimental consequences for project outcomes.
6.2 Multiple Contractors
Most sizeable projects are executed by more than one contractor and the project
enterprise usually features a number of engaged firms. The research findings suggest that
beyond a price sensitive tipping point project execution becomes increasingly difficult.
The “difficulty” has a number of dynamic cause-effect relationships including a reduction
in the time a contractor will devote to integration activities. Consideration of this finding,
in conjunction with the environment of networked contractors, suggests the possibility of
the project experiencing “contractor contagion”.
If one contractor reneges on investing time and resources on integration activities
with the project sponsor it follows that they will also, or are likely to, renege on investing
in integration activities with other contractors. Why would a contractor do this? In many
projects the contractors have formal contracts with the project sponsor for the delivery of
services, but not with other contractors. This has consequences for other contractors
working on highly integral systems and invokes the “variation order - communication
loop” described in previous sections. As integration meetings usually involve several of
the firms engaged on the contract, limiting effort in this area affects their work as well.
Through this mechanism we can see how the dynamics investigated by this research
could “spread” from contractor to contractor once an initial disruption (the initial
reneging) occurs. The idea of “contractor contagion” is analogous to the “fire fighting”
dynamic within the single firm, multi- project environment (Repenning 2002), suggesting
an opportunity for further research.
6.3 Alignment of Incentives
The findings of the research can also be framed in terms of alignment between the
contractor’ s and project sponsor's incentives. Alignment of the incentives between firms
is achieved when the risks, costs and rewards of doing business are distributed fairly
across the network (Narayanan and Raman 2004). In the model the alignment between
sponsor and contractor can be characterized as orthogonal. It can be seen that the firms
behave as they do because the financial incentives are not aligned. When project sponsors
drive down the initial lump-sum cost of a project, this is clearly at the expense of the
contractor’ s financial position. When contractors invoke variation orders to secure
revenue, this is not in the financial interests of the sponsor. This creates an adversarial
relationship which is an essential element of the competitive bid lump-sum contractual
relationship. The misalignment between the sponsor and contractor can generate
additional expense for the sponsor.
This suggests that improved project performance, through alignment of
incentives, requires an alternative enterprise architecture. The orthogonal architecture,
characterized by an adversarial element, may be improved by moving to a more fully
aligned architecture. Recognizing that the misalignment exists within a spectrum of
possible solutions provides an opportunity to address it. Under the structure modeled in
the paper a number of project pricing solutions deliver improved project performance in
comparison to the “zero-sum game” approach of minimizing up front costs (i.e every
dollar given to the contractor at contract award is a dollar off my NPV). However, it is
not explicitly evident to the project sponsor and contractor that alternatives exist.
Different enterprise architectures, an alliance or joint venture structure for example, may
make the tradeoffs explicit and allow for the misalignment to be minimized.
7. Implications for Research and Practice
As discussed above, the findings carry the promise of significant benefits for
project managers and the firms engaged in large engineering projects. The existence of
tipping point behavior related to pricing suggests a shift away from a “zero-sum game”
approach alluded to above. The implications of this are profound. Pushing for the lowest
price introduces significant project risk. However, the sponsors are wary of allowing the
contractors to capture an inappropriate share of the economic rent from a project. The
optimal pricing point for the project exists in a region near the contractors “preferred”
pricing structure (i.e the price at which they make their normal desired returns).
Negotiating the fair, and optimal price, for the contract requires understanding that all
parties need to be financially rewarded for their participation. This suggests a far more
open relationship than is currently the norm in the context of this research (oil and gas
projects). Studies of successful inter-firm relationships, usually in a supply chain context,
indicate that when firms develop close and consistent relationships they often involve an
“open book” philosophy, and an expectation of secure long term partnerships (Womack,
Jones and Roos 1991).
If project sponsors still choose to push for the lowest possible up front prices, and
relationship durations only as long as the next competitive bid, then this decision should
be made taking into account the following:
1 Projects operating in the price sensitive region are essentially unstable in the face of
changes. Therefore, a great deal of effort must be put into front end loading (FEL) to
ensure that the number of project changes is kept to an absolute minimum.
2 The lowest cost solutions are robust only for simple projects that are not highly
integral. For some projects in which the scope is very clear, and unlikely to change,
and which represent “standard” applications of technology then a low cost solution
may be appropriate.
While the discussion above sets out some steps to deliver effective projects, the
winning approach was best summed up by a senior project manager who said while
discussing how best to manage projects and contractors:
“projects that are approached as a win-win are very successful”
A number of issues were raised by the results that require further research. First,
the notion of “contractor contagion” requires further investigation. Virtually all projects
of any significance are executed by teams of contractors and it is worthwhile
understanding to what extent problems for one contractor transfer to other members of
the project team, and how that occurs. Second, altemative enterprise architectures and
structures that provide for improved alignment of incentives need research. It is proposed
in the next stage of this research endeavor that “alliance” project organizations will be
investigated. Alliance organizations, in an energy industry context, often feature an
explicit risk-reward pay-off structure to align contractor and sponsor interests. Finally,
the impact that integral product system architecture has on project performance and the
relationship to appropriate contract structures warrants serious study. As indicated in the
results, the level of effort required for integration has significant influence on the project
outcome and needs to be included in the determinations for the financial arrangements.
References
1;
10.
11.
12.
13.
14,
15.
16.
17.
Eppinger S, D., “A Planning Method for Integration of Large-Scale Engineering
Systems”, International Conference of Engineering Design ICED 97 Tampere,
August 19-21 1997.
Ford D, N., Sterman J, D., “Overcoming the 90% Syndrome: Iteration Management
in Concurrent development Projects”, Concurrent Engineering: Research and
Applications, Vol. 11, No. 3, 2003, pp 177 - 186
Ford D, N., Sterman J, D., “Dynamic modeling of product development processes”,
System Dynamics Review, Vol. 14, No. 1, 1998, pp 31 - 68
Gibbons R., “Incentives between Firms (and Within)”, forthcoming in Management
Science.
Keller R, T., “Predictors of the Performance of Project Groups in R & D
Organizations”, Academy of Management J ournal, Vol. 29, No. 4, 1986, pp 715-726
Kramer R, M., “Trust and Distrust in Organizations: Emerging Perspectives,
Enduring Questions”, Annual Review of Psychology, Vol. 50 pp 569 - 598
Lyneis J.M., Cooper K.G., Els S.A., “Strategic management of complex projects: a
case study using system dynamics”, System Dynamics Review, Vol.17, No.3, Fall
2001, pp 237-260
Martins L, L., Gilson L, L., Maynard M, T., “Virtual Teams: What Do We Know and
Where Do We Go From Here?”, Journal of Management, Vol. 30, No. 6, pp 805 -
835
Maznevski M, L., Chudoba K, M., “Bridging Space Over Time: Global Virtual team
Dynamics and Effectiveness”, Organization Science, Vol. 11, No. 5, September-
October 2000, pp 473 - 492
McAllister D., “Affect- and Cognition-Based Trust as Foundations for Interpersonal
Cooperation in Organizations”, Academy of Management J ournal, Vol. 38, No. 1,
1995, 24-59
McDonough III E, F., Kahn K, B., Barczak G., “An investigation of the use of global,
virtual, and collocated new product development teams”, The Journal of Product
Innovation Management, Vol. 18, Issue. 2, 2001, pp 110 - 120.
Miller R., Lessard D, R., The Strategic Management of Large Engineering Projects.
Cambridge, MA: Massachusetts Institute of Technology; 2000.
Narayanan V, G., Raman A., “Aligning Incentives in Supply Chain”, Harvard
Business Review, November 2004.
Novak S., Eppinger S, D., “Sourcing by Design: Product Complexity and the Supply
Chain”, Management Science, Vol. 47, No. 1, January 2001, pp 189 - 204.
Perrow, C., Complex Organizations: A Critical Essay. McGraw-Hill; 1986 oo Ed)
Reichelt K., Lyneis J., “The Dynamics of Project Performance: Benchmarking the
Drivers of Cost and Schedule Overrun” , European Management Journal, Vol. 17,
No. 2, 1999, pp 135 - 150
Repenning N, P., “Understanding fire fighting in new product development”, The
Journal of Product Innovation Management, Vol. 18, 2002, pp 285 - 300
18. Repenning N, P., “A dynamic model of resource allocation in multi-project research
and development systems”, System Dynamics Review, Vol. 16, No. 3, 2000, pp 173 -
212
19. Repenning N, P., Goncalves, P., Black, L, J., “Past the Tipping Point: The Persistence
of Firefighting in Product Development”, California Management Review, Vol. 43,
No. 4, 2001, pp 44 - 63
20. Sosa M, E., Eppinger S, D., Rowles C, M., “Designing Modular and Integrative
Systems”, Proceedings of DETC '00: ASME 2000 International Design Engineering
Technical Conferences and Computers and Information in Engineering Conference.
21. Sosa M, E., Eppinger S, D., Rowles C, M., “Identifying Modular and Integrative
Systems and Their Impact on Design Team Interactions”. Transactions of the ASME,
Vol. 125, June 2003, pp 240 - 252
22. Sosa M, E., Eppinger S, D., Pich M, McKendrick D, G., Stout S, K., “Factors That
Influence Technical Communication in Distributed Product Development: An
Empirical Study in the Telecommunications Industry”, IEEE Transactions on
Engineering Management, Vol. 49, No. 1, February 2002, pp 45 - 58
23. Sterman J, D., Business Dynamics; Systems Thinking and Modeling for a Complex
World, McGraw-Hill; 2000.
24. Ulrich K, T., Eppinger S, D., Product Design and Development. McGraw-Hill; 2000
(2° Ed).
25. Wheelwright S, C., Clark K, B., Revolutionizing Product Development. The Free
Press; 1992.
26. Whitney D, E., “Physical Limits to Modularity”, Engineering Systems Symposium
2004, http://esd.mit.edu/symposium/agenda_day3.htm
27. Williamson O, E., The Economic Institutions of Capitalism. The Free Press; 1985.
