A Framework for Measuring the Value-Added of Knowledge
Processes with Analysis of Process Interactions and Dynamics
José Cintrén, Wilawan Onkham, and Luis Rabelo
Department of Industrial Engi ‘ing and M:
University of Central Florida, Orlando, Florida 32816-: 2993, USA
Alfonso T. Sarmiento
Department of Industrial Engineering
University of La Sab Chia, Cundi ‘ca, Colombi
Abstract
In the current business landscape processes are heavily dependent on their use of intangibles and
knowledge to produce outputs. This shows the importance that intangible assets have in benefits
and value in cases such as project selection which cannot be appropriately managed without
considering the role of knowledge. This research develops a framework to measure the value that
processes add based on knowledge. It starts by considering current systems and analyzing
proposed changes to propose alternative systems in which system dynamics can then be used to
model the desired system for behavior measurements. The framework applies knowledge as a
way to generate value based on the concept of Kolmogorov complexity. Criteria for project
selection is then based on the amount of knowledge measured to generate change. The
framework is applied to a case study in a mobile weapon system using unmanned aerial vehicles
(UAV).
Keywords
Kolmogorov Complexity, Knowledge Value-Added, Return on Knowledge, Matrix of Change,
System Dynamics
1. Introduction
Corporations have traditionally measured success in terms of tangible assets. In highly technical
and information-based businesses, the value generated by company processes cannot be
measured using cost accounting, which accumulates costs and quantitative data for the purpose
of profit measurement. Investments can produce value through improvement or creation of
business processes (to increase efficiency) and through improvement of management decisions
by speedier and more accurate decision making (which makes them dynamic). Firms can
therefore attain value from knowledge-based processes but may not be able to account or
measure all or some part of that value. Capital budgeting models can measure the value of capital
investments but they rely on measures of cash flows. While tangible benefits can be assigned
cash values, intangible benefits providing business value cannot be measured under these
financial models. When intangibles are not measured the risks and uncertainty associated with
these assets is overlooked. An important question then becomes where and how can the value of
knowledge be measured and reported, within accounting models or as totally separate metrics
(Guthrie 2001). Another problem is the difficulty of quantifying intangibles when they are not
measured to begin with. Employee knowledge, training requirements, and learning curves are
some examples of the very important intangibles a manager can use for better decision making if
the information is available. Knowledge input is necessary in any business process and the value
of knowledge applied to business processes can be used as a measure of value added to the
business. Therefore since traditional methods for measuring return on investments or value
added are no longer applicable for current knowledge-based business models, the value earned
by executing such processes is better measured by accounting for value from knowledge rather
than mere monetary tangibles (Pavlou et al., 2005).
With an established need in the current technical- and technologically-based business landscape
to account for the impact of intangible assets and the importance of measuring the impact of
processes by knowledge management, different methodologies have emerged to quantify the
effects of intangibles and knowledge. Some methods identify knowledge as some remainder after
tangible capital has been accounted for while others use subjective means and assumptions. But
the reason processes are executed is to bring about changes to generate an output. In the case of
knowledge processes, how much change takes place on inputs by using knowledge can be
considered the most important aspect when executing these processes. And the amount of
change to an input by the use of knowledge can be measured by how much knowledge is used to
make the change. This last statement is based on both thermodynamic entropy and Kolmogorov
complexity, and can be summarized as follows: the required energy or complexity to generate or
describe a process output is a measure of change. The measurement of value added from
knowledge processes can be accomplished using complexity as the basis for value when
executing processes that convert inputs into outputs. Different from measuring value from
tangible cost and cash earnings, valuation of knowledge based on Kolmogorov complexity
principles can be used to calculate return from investments.
The application of knowledge valuation in this research will focus on measuring the returns on
investment and comparing the obtained metrics for decision making. The development of a
framework that takes into consideration more than just valuation metrics can greatly increase
process selection and decision making. This investigation will propose that considering process
interactions and complements as well as dynamic behavior of systems provides a significant
improvement to decision making that is based on value added from knowledge. The focus of the
research will be the development of a framework for process analysis and decision making.
Different from previous studies, the investigation will present a structured approach to analyzing
and measuring value based on process complexity along with process interactions and dynamic
behavior. Previous work has shown the appeal for measuring knowledge as a way to productivity
improvement, but knowledge-based processes also interact in dynamic system structures.
Reaching a consensus on why, what, and how the dynamic nature of systems affect the
measurements of value-added from knowledge processes becomes a new research area. This
research will propose to answer the question: Can knowledge provide better measures of value-
added from processes for decision making when aided with process selection and dynamic
modeling to provide higher returns on investment?
With knowledge management as a necessary aspect of organizational decision making, a method
to measure value-added from knowledge must then measure how much change a process has
when generating outputs as follows: the more knowledge used the more change that can take
place, and the more value that is generated. With a method for measuring the value added from
knowledge as described, what other aspects must be considered for a framework to provide a
structured and systematic method for alternative decision making? When alternatives are
available, how can one determine how changes or added alternatives would function together in
a system? Moreover, what behaviors may take place and how can we know if selected
alternatives will affect system stability?
2. Framework
2.1. Framework Overview and Description
The goal of the research is to combine methodologies in an ordered framework that analyzes
value-added based on decision on modifying the processes or selecting alternative processes
starts with the assumption that the processes are knowledge-based and their complexity
determines how much value they add as derived from Kolmogorov-complexity. With the
processes’ value defined from the knowledge they require to produce their expected outputs,
knowledge value metrics can be obtained for both current and proposed processes. The decision
on the resulting system of processes is the outcome of an analysis on how the processes function
together and complement each other as a system. The goals are to identify how critical process
are (more than merely adding more value), how they interact (do they reinforce or interfere in the
system?), how difficult they are to implement, and how do stakeholders feel about them. An
alternative system would now being analyzed not just by the main driver - value added from
intangibles - but by applying methods to determine if they should be considered at all. Since the
processes are expected to interact over periods of execution, analysis of system stability and
dynamic behavior after implementation will follow by modeling proposed systems. Since these
proposed system processes have never been executed, modeling would enable the identification
of behavior patterns as a way to test the process changes by studying before implementation.
This determines how the processes behave based on the interactions between them over time
periods in order to further analyze the actions taken for effectiveness, with the capability of
testing the process changes to affect behavior based on metrics of value-added
This initial part of the framework will analyze an “as-is” state of value added by the current
processes. Proposed processes will also be analyzed for value added before the framework
evaluates current versus proposed changes by recognizing complements between the processes’
technologies and practices. Since interactions can make it impossible to successfully implement
new, complex processes, this analysis has the goal of anticipating complex interrelationships that
surround system changes. During this phase of the framework decisions are made taking into
account interactions among all components of a proposed system. After complement analysis,
the review of results will yield a proposed “to-be” system that the framework will model to
understand the behavior over time of this new, complex system. This analysis will be based on
the complex systems are governed by both the influences that the processes they are made of
have on each other along with the time delays taking place during execution of processes. These
complex behaviors impact value added from knowledge processes over time and this phase of
the framework will analyze, by modeling, the stability and behavior of the proposed system.
Complex behavior modeling becomes a way to discover effects on value added and modification
of inputs and processes can be used to study the behavior of a complete system. Figure 1
demonstrates the framework in general terms based on the steps to accomplish process selection
based on value added by knowledge.
Values of Current Proposed
Knowledge-based petias
waedgepesed |, _»| Knowledge-based
4 Processes
Measured Measured
ZL
Analysis of Process based on | _,
“alprocess criticalities, interactions,
change transitions, and
stakeholder feedback
Modeling of resulting system for
feedback between processes
and process behavior over time
Higher Value-
Added System
Figure 1: Framework Flow Description
2.2. Analysis of Value Added from Knowledge
Information theory concerns the quantification of information. The most natural approach to
define a quantity of information is by viewing it in relation to individual objects instead of in
relation to a set of objects (from which the individual objects could be selected). The quantity of
information in an object can be defined by the number of bits needed to describe the object, and
a description of an object is useful only if the full object can be reconstructed from the
description. The processes of information exchange, of communication, transcription and
processing are covered under information theory. KVA applies the quantification of knowledge
following information theory by using a description of the knowledge needed to execute a
process and generate an output (be it by learning time, binary measures, or by quantity of
instructions). KVA requires that the outputs are useful as expected, being reconstructed from
knowledge that describes how to execute a process. KVA applies the principles of
Kolomogorov's Complexity (K-Complexity) Theory, an established and proven framework
widely used in the natural sciences to analyze structure creation in systems. It is a universal
measure of changes in the form of matter. With this in mind, creation of Kolmogorov complexity
(and the equivalent information) can be viewed as the universal activity of people, which
includes the creation of value in business proces:
Kolmogorov complexity is rooted in information theory (along with probability theory and the
concepts of randomness). While closely related to problems in information theories, K-
complexity aims to provide a way to measure ‘information’. KVA bases the concept of change
being proportional to and requiring knowledge on K-complexity theory. Under KVA the
changes in entropy are the product of processes that apply information. Kolmogorov uses the
“bit” as the unit of measurement, one bit typically defined in information theory as the
uncertainty of a binary random variable that is 0 or 1 with equal probability, or the information
that is gained when the value of such a variable becomes known. KVA essentially defines the
input (original or unchanged variable) as variable 0 and the output (a changed variable with
value added by information) as variable 1. The amount of information in an object may be
interpreted as the length of a description of the object.
Entropy is used by the KVA methodology when KVA is described by changes in structure that
can be measured as changes in entropy. Since businesses are complex open systems that
exchange information with their environments they are capable via processes to change inputs
(i.e., information, materials, etc.) into products. When these change processes take place, value is
added. This approach assimilates to the language of thermodynamics, where an input (a)
becomes and output (b) via a process (P). The change between an input (a) and an output (b) is
dependent on the knowledge needed to execute the process (P). The difference (i.e., change)
between the inputs from that of the outputs is the value provided by the people, systems, or
processes which acted upon the inputs. The major assumption of KVA is that change, and
therefore knowledge, are proportional to value. The output from a process is a function of its
input, such that: P(a) = b. A process P acts on input a to produce an output b. If a equals b, then
no value was added to the input by the process P. The following assumptions provide a
derivation of how valuation works under KVA based on KVA’s business application of
Complexity Theory:
e ifa=b, no value has been added, therefore
e value can be added only through changes to input, and
e "changes" can be described, therefore
¢ the minimum number of changes is equal to the length of the shortest description, so
e "value-added" = "number of changes" = "length of the shortest description"
With the relationship between entropy and change, KVA addresses value-added with the
assumption that if a business process P is such that no change takes place (output is equal to
input) then no value is added by the process. KVA also infers that the value-added by a process
can be proportionally associated with the change in entropy. A parallel between transformation
of information in information theory and transformation of substances in thermodynamic is made
to define entropy as it relates to information processing. In thermodynamics, the difference in
entropies is proportional to the amount of thermodynamic work needed for the change, say when
a substance is changed from a state a to a state b, such that:
AE = E(b) - E(a)
In the equation above the differences in the entropies of a and b is proportional to the amount of
change needed to make the change. The parallel with information theory comes in the form of
strings (vs. a thermodynamic “substance”). A string is generally defined as a data type which
stores data values in some sequence (usually bytes). The elements in a string stand for characters
according to some type of encoding. DNA, texts, and spoken languages are examples of
naturally occurring "strings". The complexity of a string is the length of the string’s shortest
description. In the case of information or knowledge processes versus a material or a substance
in thermodynamics, an information theory bit is proportionate to a unit of "complexity", and this
complexity is the underlying unit that is described as a unit of "knowledge".
If a description of x, d(x), uses the fewest number of characters it is of minimal length or minimal
description. The Kolmogorov complexity of x is the length of the number of characters in its
description d(x), and is defined as:
K(x) = |d(x)|
And when there is an amount of “thermodynamic” work or change needed to transform a string x
into a string y, the Kolmogorov complexity K(x) is defined as the length of the shortest
description of x. When that description of the complexity of x, K(x), is changed by a process, the
change or entropy is defined as a change in K-complexity:
AK = K(y) — K(x)
Where the change in information, or the entropy caused in the string, is the difference between
the complexity of input, K(x), and that of the output K(y). The calculation of value added comes
from the calculation of entropy or Kolmogorov complexity change that is caused when the
process transforms the input into an output. This again recalls the thermodynamics principles.
Entropy is a key measure of information in information theory and is usually expressed by the
average number of bits needed for storage or communication. Entropy quantifies the uncertainty
involved when encountering a random variable. This means that an event with a higher number
of likely outcomes will have more entropy than one with a smaller amount of likely outcomes.
To accomplish the calculation of value-added in business processes based on entropy or K-
complexity, a relationship between business change processes and the descriptions (e.g.
information strings) of those processes is used. To explain this, assume that a process can be
done two different ways: by an original process P or by a modified process M. The original
process P has an input a and output b; the modified process M has an input x and an output y.
(The M in this example is modified process, which can be thought of as U for a universal
computing machine or universal Turing machine)
Subsequently, if:
1) We map a to x in a one-to-one relationship such that a is a set of all inputs possible to
process P and x is the set of all possible inputs to process M, and
2) We also map / to y in a one-to-one relationship such that D is a set of all outputs
possible from process P and y is the set of all possible outputs from process M, then
3) M(x) = y if and only if P(a) = b
In the above, the M representing the modified process can be a computerized or information
technology process, using a computer program or software code which represents the process
and can be viewed as a description of the process outputs. The changes brought about by a
computer program in a computer are a reflection of the changes. This is because the K-
complexity in the program, as would a string, reflects the structure changes in the inputs from a
value-adding process.
To understand how Knowledge Value Added is applied for knowledge processes in the same
fashion as the ideas of entropy and K-complexity, we recall the concept of inputs thru processes
to generate outputs. KVA is based on the concept that in a process where change takes place
there is always knowledge used to change the input and generate the output. The change made to
an input is the value added to that input via the change process. The assumptions of KVA include
the reasoning that the value created by a process is relative to the change that the process affects
on the input and that change can be measured by the quantity of knowledge needed to generate
change. Knowledge can be defined by:
1) How much time it takes to acquire the knowledge (or learn how) to execute a process.
2) The amount of process instructions required to produce an output (Such as the number of
process description words, pages in a manual, and/or lines of code pertaining to each sub-
process.
3) Creating a set of binary yes/no questions so that possible outputs are represented as a
sequence of yes/no answers, to calculate the length of sequence of yes/no answers for
sub-processes. This is known as the Binary Query approach of KVA.
When KVA describes the amount of change in terms of the time needed to learn a process, these
units of learning time are proportional to an information bit, which is proportional to a unit of K-
complexity, which in turn is proportional to a unit of change. Change in KVA theory can be
described in any form as long common units are used. Hence, learning time, process description,
or binary query can all be used as descriptive languages for change measurement under common
units of output.
The KVA methodology is then based on the concepts of complexity theory and was developed
based on the concept of units of change or “units of complexity”. The information “bit” was the
best theoretical way to describe a unit of Kolmogorov complexity and therefore the knowledge-
based concept describes change in terms of knowledge required to make the change. KVA aims
to provide a measurement of the knowledge needed to produce outputs (by changing inputs). The
underlying assumptions are as follows: humans or technology change inputs to outputs thru
processes, and by describing process outputs via knowledge as the common units required to
produce outputs, revenue can be assigned along with cost.
With bits used to describe units of complexity, the “knowledge metaphor” of KVA defines
change based on the amount of knowledge needed to make changes under intangible activities.
KVA looks to standardize the outputs of knowledge processes in terms of the units of complexity
or change required to produce outputs. These outputs have value that is derived from the
knowledge needed to change the characteristics of the inputs. Value under KVA is an
assumption of the changes that knowledge brings about when generating an output. The value
added by a knowledge process is proportional to the amount of knowledge required to execute
the process. As value is measured based on the knowledge to create outputs, return on
investment, or under KVA return on process (ROP), is calculated applying value and cost. ROP
is what KVA calls its measure of value creation for processes with a predetermined output. ROP
is basically a return on investment in process. The derivation of ROP under KVA is as follows:
the internal performance V of a process is defined as
V=1/C
Where / is the amount of information or K-complexity to execute a process and C is the cost to
produce the specific amount of K-complexity needed for the process (while this explanation talks
about a single process, the performance of compounded processes can also be defined by using
weighted averages of component performances). To go along with performance V, a relation to
an external measure of performance or value is needed to account for the value added by / (as
information, knowledge or complexity). This relates to return on investment (ROI), where the
price of an output accounts for the money gained (or lost) and is the numerator in a ratio against
the cost to execute. For example, when a business obtains a monetary value from a process
output, that value correlates to the complexity of the process that generated the output. This can
also be seen when a customer pays for an output when a client pays a set price for a unit of
information no matter how it is produced. KVA derives return on knowledge (ROK) as the ratio
of the value that the complexity or knowledge of a process generates and the cost of the process.
Return on Knowledge (ROK) is the ratio of revenue allocated to a core area when compared to
its corresponding costs. With knowledge as a surrogate for common unit outputs, ROK
determines knowledge value to cost ratio for processes.
2.3. Change Analysis
Management of change, based on the importance of interconnections and considering that system
optimization requires cohesive processes, can be accomplished using the Matrix of Change
(MOC). Drawing from Quality Function Deployment (QFD), the proactive definition of
activities needed to meet requirements “permits quality and customer needs to be designed into
the product, not added on.” (Richardson 1997). QFD applies mechanisms that analyze
relationships and correlations by which requirements are translated to be successfully using a
matrix called the “House of Quality”, by applying values and priorities to the relationships
between needs and requirements. Effective management of change also requires recognition of
the critical role of interactions which “can make it impossible to successfully implement a new,
complex system in a decentralized fashion. Instead, managers must plan a strategy that takes into
account and coordinates the interactions among all the components of a business system.”
(Brynjolfsson 1997). QFD has proven successful in change management by early evaluation of
requirements and expectations. The proposed framework can relate this idea when it shows a
need or a “what” (value added from knowledge) with a “how” (process changes).
MOC provides effective change management as it recognizes complements between technology,
strategy, and practice by anticipating complex relationships that come about from change. MOC
considers the issues of stability under new changes, sequence of processes, pace of change,
implementation in new or available locations, and the sources of value added from the interests
of stakeholders. The analysis is accomplished in four steps: three matrices (current practices,
desired practices, and transitional state bringing current and desired together) and evaluations by
stakeholders to identify the importance of processes to activities by the stakeholders. The first of
the four MOC steps is to identify critical processes and practices that are broken down into
“constituent parts” or the processes expected to meet or improve practices or goals. The second
step classifies system interactions by matrices that identify processes as ones that increase
returns on processes they complement (reinforcing) or ones that decrease returns on processes it
competes against (competing). A grid based on QFD’s “House of Quality” starts in this step in
the way of triangular matrices: a horizontal for existing processes and a vertical for proposed
processes. These “interference matrices” use grid signs at the process junction locations: plus
signs (+) for reinforcing, minus (-) for competing, and no sign for weak or no interactions. The
plus and minus signs can be determined in different ways as many times it can be self-evident
but other formal theories can be used as well as empirical methods and surveying of personnel.
Step three identifies interactions of transitioning by implementing proposed processes by
combining the horizontal and vertical matrices from step two into a matrix to determine
interactions between existing and target practices using the plus and minus signs configuration
previously applied. The fourth step surveys stakeholders on how they perceive current and target
processes in terms of building a better system, output, or value-added. Those surveyed will use a
five-point Likert scale as follows to rate each process:
+2: Extremely important practice/process
+1: Important, but no essential practice/process
0: Indifference
-1: Some but not essential desire to change or reject a practice/process
-2: Strong desire to change or reject a practice/process
Figure 2 demonstrates completed MOC steps one and two (left) and three and four (on the right).
The measure of business value evaluated from stakeholder’s perspective for the purposes of this
framework will apply the quantifiable units of knowledge value-added as the basis for this step
of the MOC. It answers the MOC question: what are the greatest sources of value? The change
process has a higher chance of success by the identification of complementary structures and
analysis of interactions between processes provides the most important tool for decision making
without the significant commitments which change would otherwise incur. Such analysis also
provides a smooth transition into the next phase of the framework that will model the proposed
systems for behavior and stability during process executions over time.
.
— | Target Practices:
4 Altemative 1
§
“BAe
Blajfafta
BAG
B?| 2/2
5
2ypeda
Existing Practices
3 Existing Practice 1 2al-]+
+2.
+X |os Existing Practice 2 {Eee
26
3
Ff Existing Practice 3 4 +
Figure 2: MOC System and Transition Interactions
2.4. Dynamic Modeling Application
The systems resulting after MOC analysis in this research can be made up different combinations
including processes that have never been executed before and/or processes previously executed
within other systems. The dynamic complexity of knowledge-based processes makes it difficult
for managers to make decisions based on the behavior of these processes. Like most processes,
knowledge-based processes exhibit non-linearity that can make decision making difficult
because a simple change can produce complicated effects. These complexities imply a need to
understand the interactions that taking place in knowledge-based systems. A behavioral view of
system dynamics can focus on system (and process) characteristics that may “make or break” the
complete system. A behavioral model can provide reproduction of dynamic systems before any
change commitments. The implementation of changes then becomes a product from the insights
gained during the dynamic simulation modeling. The basis for modeling the behavior of process
that make up a complex system is the recognition that system structures are as important as the
individual processes, while there are properties of a complete system that cannot be explained or
even recognized by the behavior of individual processes. This research will analyze the behavior
of proposed processes functioning as a whole new system. With a proposed system composed of
processes generated from the first two major phases of the framework, this last phase will model
the system for behavior over time to understand complex issues and problems that arise from
dynamic behaviors.
The System Dynamics (SD) approach is unique in its study of the feedback and stock-and-flow
dynamics to display what could be severe non-linearity in systems that may appear simple. SD
uses visual representations of the information feedback and circular causality that conceptualize
the structure of complex systems, in turn communicating model-based insights (Sterman 2000).
Feedback loops are present when information from an action (e.g. a knowledge process) moves
through a system and can influence the system’s behavior. System dynamics modeling as the
last phase of the framework will analyze the dynamic behavior of alternative systems. The
modeling of alternatives would simulate processes as continuous steps in a system that begins
with an input and finishes with an output. SD modeling would be used to influence inputs,
value-added, and cost metrics. Figure 3 summarizes the methodologies used in terms of the tasks
that the framework looks to accomplish. The outputs of the SD model will help analyze the
behavior and success of alternative knowledge systems over time by providing KVA metrics.
Knowledge Value-Added __——
Values of
Values of Current |
Proposed
Current Knowles ised Knowledge-based
Processes Processes Processes
NE Measured | as
Matrix of Change
system Dynamics — fecdback between processes
Ng process behavior over me]
—_p
Higher Value-
Added System
Figure 3: Methodologies to accomplish the Framework’s objectives
3. Case Study
3.1. Case Study Introduction
A previous study on Department of Defense (DoD) acquisition programs looked to improve the
use of benefits in analysis of alternatives (AoA) by making a system dynamics model of a
military operation and integrating it with KVA in order to improve the accuracy of KVA
estimates in AoA processes. The main problem identified by the research was measuring the
benefits of material alternatives. AoA became difficult due to alternative diversity, metric
selection and performance measurement among other factors. Along with cost estimates pre-
dominating the AoA, the research arose from the difficulty of incorporating benefits from
materiel since many important benefits were intangible in nature. The goal of the research was to
include benefits in AoA in terms of common units, to enable better comparisons among
alternatives based on value instead of merely cost. When a materiel solution is needed, AoA is
used to meet criteria and reach decisions. When needs are derived in an area that can only be
met by new materiel, AoA helps comparison of options (for example manned or unmanned
aircraft vs a missile, chemical vs kinetic energy kill mechanisms, etc.). After lessons learned on a
Javelin anti-tank weapon system concept which had three missile technology alternatives to
award a development contract and where the chosen alternative was selected based on a
capability not a stated requirement that provided value. While there were lessons on
requirements, bureaucracy, and technology readiness, analyzing alternatives under a single
undefined and qualitative factor of performance (gunner survivability) ultimately drove the
chosen alternative. A parameter which promised the most of what was impossible to quantify
became the main factor when selecting alternatives and the process failed in reaching a final
solution faster and more directly due to insufficient articulation of benefits in the AoA process.
The Javelin program showed a need for common units of benefit estimates in AoA, leading to
inclusion of units of benefit along with cost.
3.2. Case Study Problem Description
Weapon acquisition programs typically conduct AoA to select material solutions based on
viability and costs to make decisions regarding further development and production. Concepts
are then analyzed as part of a material solutions analysis by which various cost estimates are
generated from cost comparisons. The emphasis on costs in the early stages of acquisition should
not become the main (and even less only) criteria for alternative selection. This practice caused a
feeling of disparity between costs and benefits from effective operations. The main problem area
stated by the Naval Postgraduate School (NPS) research looked to improve was the estimation of
benefits, but more importantly in common units. The main problem stated by the previous
research was the difficulty of defining common metrics to measure performance in order to
account for benefits from alternatives. This measurement of benefits using KVA was integrated
with dynamic modeling of a weapon system for unmanned aerial vehicles (UAV) to make
decisions on upgrading the system. The modeling uncovered synergies between the UAV
weapon system processes which (while being measured using common units) increased the
amount of alternatives to analyze. The research concluded that this measurement of benefits
along with modeling of the dynamics of the system’s alternatives was a major improvement from
decisions made using costs of alternatives (Housel and Cook 2005; Housel and Bell 2001;
Housel et al. 2001).
The case study being proposed in this paper will look to benchmark the findings from the NPS
research and more importantly improve on decision making by integration of common
measurement of benefits from intangibles (included research but done after dynamic modeling),
alternative decision making based interactions and complementarities between alternatives (not
included in original research), and dynamic modeling to analyzed changes in benefit values after
a new system has been defined (modeling used in previous research, but before any
complexity/benefit metrics were calculated). The NPS proposed as an item for further
investigation the ability to indicate the sub-processes that improve the alternatives. As an
example, while it was thought that increasing the “fuel capacity” alternative was the reason a
sub-process called “fire mission development” was improved, it was discovered from the
modeling that the actual cause for the improvement was an increase in “vehicle range” because
this alternative reduced the chance of losing a target if it was missed (versus not being able to re-
acquire a missed target and needing more time, fuel use, etc.). This will be researched under the
alternative decision making phase of the framework, which will provide a method to identify if
changes are to be implemented. The NPS study also made use of modeling to generate forecasts
of performance during acquisition, “comparing those forecasts with actual operations, and using
the results to improve the model fidelity with the system. The improved model can then be used
to analyze proposed changes or replacement of the system throughout its lifecycle” (Ford,
Housel and Dillard, 2010).
3.3. Matrix of Change
Under the NPS research a new version of the Predator UAV was being developed to enable it to
engage opposing UAVs. Only one improvement alternative was to be selected from three options
which stakeholders value differently: payload, dash speed, and range. The study’s analysis
focused on value compared to cost in terms of the capabilities of the systems. KVA was
integrated with SD to investigate how modeling weapon systems can improve the accuracy of
KVA ratios. Assuming a new version of the predator UAV is being developed to engage enemy
UAVs. This will increase the fraction of targets missed because UAVs are faster, more agile
(than land targets). There is access to only some limited resources to improve performance.
Stakeholders value payload, dash speed and range differently and want recommendations on
different improvements to select only one of the improvements. Examples of some of the
alternatives for improvement were:
e Increase size of power plant: can increase the vehicle’s payload, dash speed, or
combination of both; requires an increase in fuel capacity to not reduce range.
e Redesign transmission: will increase dash speed.
e Increase fuel tank size: will increase range but decrease dash speed unless power plant
increased.
e Reduce time required at base between trips to station: increases time the vehicle is on
station and available for missions.
The operation of the system with each potential alternative was simulated to calculate KVA
productivity ratios for subprocesses and for the whole system (the three subprocesses that are
impacted by the characteristics of the vehicle). Referencing the following table, the AoA
suggests that the increasing of fuel capacity 100% is the alternative that improves the system the
most. If there are inadequate resources to implement this alternative fully, then increasing fuel by
50% can be attempted (since it will still bring the highest improvement). The ones that do not
improve performance (last three with negative change from base case) can be eliminated from
consideration.
Table 1 summarizes the results.
This research proposes a MOC analysis before system dynamics modeling (as where the NPS
research performed SD in the beginning) where the alternatives are analyzed against available
practices or goals. For the NPS study this proposed framework will perform an MOC analysis on
the alternative practices against the system sub-processes. MOC will analyze the practices of
increased power plant size, increase fuel capacity, redesign transmission and reduce time at base
against fire mission development, weapons movement, and engaging targets. The stakeholders
being interested in payload, dash speed, and range will influence their view of target practices
differently.
Table 1: Predator Upgrade Alternatives Results
Sub-process KVA ratios Weapon System
Develop % Change
Fire Move Engage | KVA | from Base
Mission | Weapons | targets ratio Case
Predator Base Case 943 50 5,094 705 0.00%
y [Increase fuel capacity 100% 1,886 50 5,094 951 34.90%
g
& | Increase fuel capacity 50% 1,415 50. 5,094 831 17.90%
& Increase power plant 100% for payload 849 50 7,641 77 9.40%
3
2 Increase power plant 50% for payload 849 50 7,641 771 9.40%
€ | Redesign ission for 100% faster dash speed 943 100 10,188 741 5.10%
5
& Redesign ission for 50% faster dash speed 943 75 7,641 727 3.10%
2 Increase power plant 100% for dash speed 849 100 10,188 717 1.70%
& Increase power plant 50% for dash speed 849 15) 7,641 702 -0.40%
Reduce time at base 50% 943 oe §,094 699 -0.90%
+ +
Matrix of Change oe Matrix of Change +X+
Ss Case Study MOC attr: S
Rese Suse | OSS feet a
2)ldentiying System interacions | | _Atemative 1 Group Hoa iae sive a
= oy el |. ie
£ z = a SESE
i ee Glee bean se
oe ee ee g) 8) 2) 85) 3
oe ee ed ie #/3]5|22] 2
Slo] f]Ss| 2 e| 5) 2)23, £
B) 2) Ess] S| E] E/SE 5
Ble] S/fe] 5 £/ 2] 5s) °
#/ 2) 5 |e] ® =
Existing Practices ae Peeps 4 qo] +2] +1
& | Lowsuccess rate forFire aa
8 Mesion Exauutign g | LowsuccesstatstorFire | 5 +
a 2 At — — | Mission Execution
ES) Average Tumaround of =) #3]
ZA) 35 | “Weapon ovement f22| Snnewe [2] -[*|+
= Limited Target y a | Limited Target elt +
a Engagements fe Engagements
Figure 4: MOC for NPS Study
The MOC analysis shown in Figure 4 provides insights into complementarities of practices (and
assets) by the interaction signs in the matrix. Increased fuel capacity and shorter times at base are
practices with two reinforcing interactions and which reinforce each other, while larger plant and
redesign transmission only have one reinforcing relationship each. The questions of feasibility,
sequence of execution, location, pace/nature of change, and stakeholder evaluations offer
guidelines on how, when, and where to implement changes. The difficulty of transitioning to the
alternative processes is defined by +/- signs in the cross-sections for existing and target
processes. This MOC analysis would first eliminate the “reduce time at base” alternative based
on its interaction score and negative importance to the process while considering the remaining
three, with increased power plant demonstrating the easiest transition and “increase fuel
capacity” showing the strongest importance (+2).
3.4. Causal Loop and System Dynamics Model
After decisions on implementing new processes, system dynamics can model the resulting
proposed system. CLDs provide “maps showing the causal links among variables with arrows
from a cause to an effect” (Sterman 2000) capturing the dynamics of a modeled process and
applicable to the capture of hypothesis about dynamics’ causes and to demonstrate the feedbacks
of a specific process. Stock and flows structures are descriptions of variables with rates or
“flows” which can increase or decrease. These flows accumulate into the most important
information in a dynamic model as “stocks” which represent system states. An appropriate
systems dynamics model for measuring knowledge value requires variables for the state of the
system (stocks), for the increase and decrease of these stocks (flows), and variables that can be
linked to stocks and flows supporting the description of the model behavior. The resulting
system from the selected alternative processes in the MOC analysis is now put into a SD model
to the complete system processes of moving weapons and acquiring a target under the NPS
study. The modeling of alternatives would simulate processes as continuous steps in a system
that begins with an input and finishes with an output. Revenue and cost are used for knowledge
valuation based on KVA methods, and make up the stocks for each of the processes (these stocks
are considered return on knowledge stocks). System dynamics modeling would be used to
influence inputs, value-added, and cost metrics. This modeling allows graphing of stocks of
value-added to provide the KVA metrics to be analyzed for results in the framework. Figure 5
demonstrates the CLD of the alternative system from the processes selected in the MOC
analysis. This is the final phase of the framework and is to provide, by modeling before
implementation, insight on the behavior of the complex dynamics of a knowledge system over
time. Figure 6 represents a SD model that would be generated as the final major step in the
framework for this case study.
tedesign
‘Transmission
—Size of Powkr
Plant
Bo
+ Payload
Q
Cost
7
BI-Dash 7~\ J oad %
Speed ae. \
Méve v/ )
| ea
‘io
; 1) my ef
ey
] ae aN
A rie af Required
‘Time Required at Base
Between Trips to Fe Keowee
Time at
ea
‘The amount Complexty
of Processes <@—————
Figure 5: Causal Loop Diagram of UAV Weapons System
Probability of __
Available Asset at
Station
Leaving Rate from
Station to base
‘Avg. Time from
Dash Speed Task Complexity
Level
Fuel Capacity
J L
Work in Process * =
—& Production Start ‘Task Completion
Rate “h et
ea Size
Effect of Range
Range
Fire Mission
Performance
Number of Tasks
Effect of Size
——— Payload,
Power Plant
Size of Power
Redesign
‘Transmission
Avg Cost per T an
“raning
=H Experience Time
pe TES
—
Input Sum
Fire Miss
‘Time for Training —™ n
Relative Fire
Mission
~ Performance
_ = KVA Ratio J =
Figure 6: System Dynamics Model of UAV Weapon System
Fire Mission Performance KVA Ratio
600 400
450 300
4
300 Let 200
150 100
o 0 | ft
° 050 708090100 o 0 2 30 4 50 6 70 80 90 100
Time (how) Time (hour)
Fire Mission Performance : current, KVA Ratio : cutent
Figure 7: The Results of Fire Mission Performance and KVA Ratio
3.5. Model Optimization
The optimization option that comes with Vensim DSS provides an efficient tool for policy
analysis. An efficient Powell hill-climbing algorithm searches for the best set of policy parameter
values to maximize the objective function. The Powell hill-climbing algorithm was developed by
Powell (1964) and it is an optimization approach that searches the objective in a
multidimensional space by repeatedly using single dimensional optimization. The method finds
an optimum in one search direction before moving to a perpendicular direction in order to find an
improvement (Press et al., 1992). The main advantage of this algorithm lies in not requiring the
calculation of derivatives to find an unconstraint minimum of a function of several variables
(Powell 1964).
With the purpose of reducing the current level of Work in Process inventory (see Figure 8), we
apply policy optimization to the parameters that affect the Task Completion Rate (see Table 2).
This rate initially starts increasing until it reaches a peak and then stabilizes. The objective will
be to maximize the Fire Mission Performance while reducing the Work in Process inventory.
Work in Process
‘Task Completion Rate
|
ain
o 0 20 3 40 60
Work in Process : Inia Values.- ——
4050 6070S
‘Time (hour)
‘Task Completion Rate : Initial Values. ——
Figure 8: Work in Process inventory and Task Completion Rate before optimization
Table 2: Comparison of new parameter values with the original values
New values from the Original values of the
ptimization model
Fuel Capacity 2.06 1
Tank Size 1 1
Redesign Transmission 1 1
Size of Power Plant 1 1
Effect of Range 0.85 0.85
Effect of Size of Power Plant 1 1
Task Complexity Level 4 4
Range 2 2
Dash Speed 1 1
From Table 2 and Figure 9 we can conclude that although the Fire Mission Performance remains
almost the same, increasing the Fuel Capacity it is possible to the reduce and stabilize the Work
in Process inventory. The other parameters remain unchanged.
Work in Process Fire Mission Performance
40 600
i” 450
20 +00
0 \ so
o r
o 0 0 0 «0 50 6 1 80 0 100 -
‘Tare (ou) o 0 20 9 4 50 o 7 8 9% 100
‘Time (hour)
Work in Process: Optinn_vahes
Souk is Proeean aa ek Fire Optim_values
Fire Miss Tnita_vaes
Figure 9: Comparison of Work in Process inventory and Fire Mission Performance before and
after the optimization
4. Conclusions
Knowledge is clearly one of the most important strategic resources to remain competitive and
firms need to both create it and manage it. But effective decision-making in environments of
dynamic complexity requires expanded analysis and models that can describe these complex
behaviors. The proposed framework uses knowledge complexity as a more appropriate method
to measure the value of intangible knowledge processes. From there it performs analysis of
process alternatives based on their interactions, feasibility and the stability of a system with
modified and/or new processes. The framework then models the structures and feedbacks that
take place in processes while applying a knowledge valuation methodology as its mathematical
basis. The existing literature does not analyze interactions of knowledge processes before
changes are made and neither does it model the resulting systems in a dynamic fashion for the
purposes of controlling and comparing process variables. The proposed framework
methodologically selects candidate processes, studies their interactions, and dynamically models
the value added by knowledge for alternative decision making. The structured combination of
existing methodologies proposes a means to understanding how process investments affect
value-added while dynamically providing return on investment.
Acknowledgements
We would like to express our great appreciation to Dr. Thomas Housel of the Naval Postgraduate
School for his valuable and continuous support during the planning and development of this
research work. His availability and willingness to offer his time has been very much appreciated.
References
1. Ford, D., Housel, T., and Dillard, J. “System Dynamics Modeling for Improved
Knowledge Value Assessment: A Proof of Concept Study.” Master Thesis., Naval
Postgraduate School, August 2010.
2. Guthrie, J. “Management, Measurement and Reporting of Intellectual Capital.”
Knowledge Management Magazine, Friday, June 22, 2001.
3. Pavlou P., Housel, T., Rodgers, W., and Jansen, E. “Measuring the Return on Information
Technology: A Knowledge-Based Approach for Revenue Allocation at the Process and
Firm Level,” Journal of the Association for Information Systems, 6(7), 199-226/July
2005.
4. Cook, G. R., and Dyer, J. D. “Business Process Reengineering With Knowledge Value
Added Support of the Dept of the Navy CIO.” September 2003.
5. M. Curley. Managing information technology for business value: practical strategies for
IT and business managers, Hillsboro, Or. Intel Press, 2004.
6. Brynjolfsson, E., Renshaw, A. A., and Alstyne, M. “The Matrix of Change: A Tool for
Business Process Reengineering.” Management Review 38 (1997).
7. Richardson, T. Total Quality Management: Delmar Publishers, 1997.
8. Housel, T., and Cook, G. “An Approach to Valuing Intellectual Capital in Defense
Processes Using the Market Comparables Approach,” Naval Postgraduate School, IC
Conference 2005.
9. Sterman, J. D. Business dynamics: systems thinking and modeling for a complex world,
Boston: Irwin/McGraw-Hill, 2000.
10. Housel, T., and Bell, A. Measuring and Managing Knowledge, McGraw-Hill, New York,
2001.
11. Housel, T., El Sawy, O., Zhong, J. J., and Rodgers, W. “Measuring the Return on
Knowledge Embedded in Information Technology” Twenty-Second International
Conference on Information Systems, 2001.
12. Vensim DSS. Vensim Reference Supplement, 2003.
13. Press, W.H., Teukolsky, S.A., Vetterling, W. T. & Flannery, B.P. Numerical recipes in
C: the art of scientific computing. Cambridge, England: Cambridge University Press,
1992.
14. Powell, M.J.D. An efficient method for finding the minimum of a function of several
variables without calculating derivatives. The Computer Journal, 7(2), 1964: 155-162.
15. Housel, T and Kanevsky, V A. “Reengineering business processes: A complexity theory
approach to value added”; INFOR, Nov 1995.
. Press, W.H., Teukolsky, S.A., Vetterling, W. T. & Flannery, B.P. (1992). Numerical
recipes in C: the art of scientific computing. Cambridge, England: Cambridge University
Press.
. Powell, M.J.D. (1964). An efficient method for finding the minimum of a function of
several variables without calculating derivatives. The Computer Journal, 7(2), 155-162.
. Vensim DSS (2003). Vensim Reference Supplement.
a
x
oo
