Table of Contents
Go Back
A Dynamic Alignment A pproach to the Leverage Design in
System Dynamics Models: Use Strategem-2 As an Example
Showing Young Yi-ming Tu Ya-tsai Tseng
Department of Business Department of Department of Business
Management, National Sun Information Management, Management, National
Yat-sen University, National Sun Yat-sen Sun Yat-sen University,
Taiwan, R.O.C., University, Taiwan, R.O.C., Taiwan, R.O.C.,
886-7-5252000 ext. 4625 886-7-5252000 ext. 4717 886-3-5562959
young@ cm.nsysu.edu.tw ymntu@ mis.nsysu.edu.tw m8342607@ mis.nsysu.edu.tw
Abstract
This paper attempts to develop a dynamic alignment approach to leverage design of
system dynamics models. The approach developed in this research is a flow-based
design, different from traditional macro designs of aggregated system structures or loop
analysis. It is grounded on some major concepts in Synchronous management at plants
and uses the Drum-Buffer-Rope (DBR) system to align the various operations of rate
decisions. We take the famous long wave model (Strategem-2) as an example to
examine the performance of the proposed leverage design method. The redesigned
system shows that the original oscillation behavior is fully eliminated. By better
alignment and streamlining of flows, the discrepancy between the production rate and
desired production in Strategem-2 is dramatically reduced, even in higher degree than
some previous leverage design approaches proposed by other system dynamists (e.g.,
Sterman (1989), Ozveren and Sterman (1989)).
1. Introduction
System dynamics is a methodology to investigate the information-feedback
characteristics of complex managerial, economic, or social systems. The goal of system
dynamics is to obtain deep understanding of systemic behaviors and to help for the
design of improved system structure. The assumption that system behavior results from
interacting feedback loops, which force variables to grow or to decline, enables one to
understand how unwanted behavior arises and to analyze how a proposed policy design
may affect the dynamics of the system. System dynamics models are not sets of poorly
understood mathematical functions, but a formulation of acceptable formal decision
policies that describe how decisions result from the available information streams.
Nonlinear decision equations specifying the momentum of tangible and intangible flows
interact each other and constitute the underlying structure of dynamic systems. Hence,
the original decision policies and the proposed policies to improve system behavior are
no longer black boxes with poorly known operations and limits.
According to Macedo (1989), current methods used to conceive the best policy of a
system dynamics model can be classified into three families: the heuristic methods, the
modal methods, and the optimization methods. Heuristic methods largely rely upon a
sufficiently deep, intuitive understanding of the problem or by some simple principles
and rules of thumb (Forrester, 1961; Graham, 1976). Thus, heuristic methods do not
offer formal mechanisms to develop new policies for its lack of theoretical basis.
Relative to the heuristic methods, policy design methods of the modal and optimization
categories are too difficult for managers and designers to understand. Most of modal
methods build the desired policy with applications of modem control theory. These
methods require linearization of the system dynamics model, a procedure that is in
variance with the basic idea of system dynamics and limits these methods to be applied
in high dense complex problems (Mohapatra and Sharma, 1985). Optimization methods
use mathematical programming techniques to optimize a certain objective function and
to obtain the improved policy. Most researches in this category propose “black box”
policies. This is in variance with the basic principle of system dynamics (Burns and
Malone, 1974; Keloharju, 1982; Coyle, 1985). In summary, none of the existing
methods is fully satisfied and researchers continuously devote to develop better leverage
design methods.
This paper proposes a “dynamic alignment approach” which can be classified into
the heuristic methods, but has a more rigid theoretical basis. It is focus on the
synchronization of rate’s operation to generate the expected systemic behaviors. The
synchronization mechanism is derived from the Drum-Buffer-Rope (DBR) system in
Synchronous management. The DBR system is designed to manage work centers at
plants to streamline multiple flows. This research adopts this mechanism to manage the
relationships of the various rate policies within and between flows in system dynamics
models. We will introduce some major concepts in Synchronous management and DBR
system in more detail later. Principles and steps in the proposed dynamic alignment
approach will also be described further.
2. A Dynamic Alignment Approach
2.1. Theoretical basis and Guiding Principles of the Dynamic Alignment A pproach
Synchronous Management is a time-oriented description of the manufacturing
process, which is consistent with system dynamics as a dynamic analytic methodology.
“Synchronous management” is a newly developed manufacturing management
approach and is established on the theory of constraints (TOC), which is developed by a
physicist named Eliyahu M. Goldratt (Goldratt, 1987). Theory of constraints analyzes
systemic problems at plants by checking the consistent operations between constraint
resources and non-constraint resources. A constraint is defined as “any element that
prevents the system from achieving the goal of making more money”. With the
consideration of constraint capacity resources, Synchronous management identifies
several manufacturing principles to examine and manage the operation of plants.
Traditional way of thinking about a manufacturing plant reflects a resource focus or
resource orientation. Managers try to balance each work center’s output and input by the
balance of capacity (Srikanth and Podzunas, 1990; Umble and Srikanth, 1997).
Different from traditional management methods, synchronous management shifts from a
resource focus to a product-flow focus and stresses on the balance of flow between the
constraint resources and non-constraint resources. In the perspective of theory of
constraints, synchronous management manages a factory by coordination and
harmonization of the operations of critical work centers to streamline various flows.
And it is the Drum-Buffer-Rope (DBR) logistic system that is used in Synchronous
management to be responsible for the coordination work.
There are three components in the Drum-Buffer-Rope system. The first one is the
drum. Drum reconciles customer requirements with the system’s constraints. And a
drum is responsible for setting the pace for material release into the system. Usually the
drum at a plant is the schedule of master production plan. The second component is
buffer. Buffer refers to time buffers or stock buffers used at a few critical points in the
process to protect the system throughput. Because factories and plants designed by the
synchronous management approach are tightly coupling systems. Buffers are needed to
protect the tightly coupling systems from variations in customer requirement and work
centers. The last component is the DBR system is rope. Ropes are designed to provide
effective communication throughout the organization of those actions that are required
to support the master production schedule. Every aspect of the operation must be
synchronized to the requirements of the drum so that the planned product flow may be
executed. Ropes in the DBR system act as the communication link from the demand to
the drum, as well as from the drum to other schedule release points. A single rope
replaces the series of short ropes. It means that the pace of drum is directly transmitted
to those control points linked by ropes. With the drum, buffers, and ropes, critical
material release points operate at the same pace with the constraint resource, which also
muns synchronously to reflect market demand variations. It means that all the control
points or material release points centrally controlled by the drumbeat are responsive to
the demand variations at the same time. That’s why synchronized flows are possible in
synchronous management.
For system dynamics models, synchronous management to make each flow
synchronized has a much deeper implication. System dynamists have found that it is
phase-lag subsystems that transform the sinusoidal signal and allow it to continue
propagating (Graham, 1977; Mass and Senge, 1975). A phase-lag subsystem is defined
as a subsystem (possessing one or more levels) that produces a phase lag between its
input and its output, so that when the input reaches its steady-state value, the output
teaches its steady-state value only later. The phase-lag subsystem is identified to be
essential to producing oscillation. In a phase-lag subsystem, if the input to the
subsystem goes to its equilibrium value, the output goes to its equilibrium value only
afterwards. That suggests the disturbance propagates around a loop: an exogenous input
or initial condition disturbs a subsystem way from its equilibrium value, and even when
that subsystem returns to its equilibrium value, the loop has nevertheless transmitted the
disturbance to leave another subsystem out of equilibrium (Graham, 1977). A
synchronized-flow design can eliminate the phenomenon of propagation in sub-lag
system effectively, because all the streamlining flows operate at the same time and at
the same pace. When sudden exogenous changes occur, a synchronous system can
retum to its stable state quickly. The dynamic alignment approach developed in this
research is built on such a synchronous concept.
The goal of the dynamic alignment approach is to design a synchronous system
that can adapt to the environmental change as soon as possible. A synchronous system
in this research means a system that is not only synchronous to the extemal
requirements but also synchronous in the actions of the rates’ operations. From the
macro and systemic perspective, a synchronous system is a system that generates
exactly the quantity that extemal environment demands at exactly the time that the
demand is happened. In the analysis of the micro decision rules and policies, operations
of rates are designed to be responsive to the external disturbances and changes at the
same speed and at the same time. Based on the concepts of synchronous management
and the concept of propagation of variations in previous system dynamics research,
several principles guiding the dynamic alignment design are derived.
First, each level in the synchronous system has to remain its original state or its
proportion relationships to other levels in the system, unless some necessary changes
planned by designers. This principle is to maintain the original relationships among
levels and among rates to prevent any non-synchronous rate operations. Second, rates to
be synchronous with exodogenous input change must be the overt decisions and the
pseudo overt decisions. In this paper, pseudo overt decision is a finer classification
derived from the implicit decisions (Forrester, 1961). It is one kind of implicit decision
influenced by those resources that can be decided consciously by decision makers.
When resource limitations change at the same pace with the drumbeat, and these kind of
overt decision generated by the synchronous design are called as pseudo overt decisions.
In the dynamic alignment approach, such decisions can be treated as overt decisions. So,
this kind of decisions is called as pseudo overt decisions. This principle claims that
designers can manage and design only those decision rules of overt decisions and
pseudo overt decisions in synchronous systems. Third, to avoid any
non-synchronization due to the time delay between planned policy and actual actions,
pipeline stocks have to be supplemented. Fourth, a good synchronous design has to
maker sure all the rate that may be influenced by the exogenous change are taken into
account, including rates that are influenced directly and indirectly through the
relationships between each rate, connect directly by information wires or flows, or by
the wires and flows both.
2.2, Steps in the dynamic alignment approach
As mention above, the dynamic alignment approach is to design a system that can
prevent any oscillations due to the inconsistent actions of individual rates. The main
mechanism to align rate policies and to streamline flows used is the DBR logistic
system in synchronous management. With the DBR system, one can focus on the
synchronization design among rates within each flow and between different flows, thus
make sure the whole system react to the external disturbances as a whole. In the
dynamic alignment approach developed here, three major stages can be identified.
Firstly, one needs to decide the basic drumbeat and the pace of the system. Secondly,
those control points to be synchronized with the drumbeat have to be found out and the
quantity of each rate operation relative to the drumbeat has to be computed properly.
What is more important is this stage is to make sure the actual synchronization among
rates and flows are at present and avoid any fluctuations arise from the time lag between
the planned quantity and actual quantity of rate operations. Therefore, one needs to
design and implement a mechanism to solve the problem of time delay. Lastly, a buffer
size adjustment mechanism is designed to satisfy the buffer size and stable level
adjustment requirements to protect the system from internal variations of rates and to
offer the system a certain degree of flexibility in confrontation of external disturbances.
According to the guiding principles and the three major stages, there are eight
synchronous steps for the basic design of a synchronous system. We introduce the
eight-step basic synchronous design first. As to the advance mechanisms to make the
synchronous system more stable will also be described later.
(1) Choose the rate influenced directly by exogenous input as the drum. This step is
based on one assumption that only one exogenous variable is at present in most of
system dynamics models. Exogenous variables are variables that are generated
independently of the system. Seldom will more than a single exogenous variable be
justified, because using more than one exogenous variable impli3es a rather unlikely
condition that the exogenous variables are interlocked by control mechanisms
between themselves but all are free of interlocking ties with the variables of the
system under study (Forrester, 1961). Therefore, one can make sure that there is
only one drumbeat in the designed system. Because rate directly influenced by the
exogenous variables passes the change further to influence other levels and rates in
the system, that is chose as the drum and the change speed of the drum is the
drumbeat of the entire system.
(2) Choose the level directly influenced by the rate as the focus level.
(3) Make one of the rates influencing the focus level as the focal control point and
make it synchronize with pace of the drumbeat. This step is to make the focus
level influenced directly by the drum rate stay in its original state. According to
guiding principle 1, each level in a synchronous system should sustain its relative
state compared to other levels in the model. To make the focus level remain
unchanged, one should select one control point from all possible rates that make the
focus level decrease or increase directly. And according to guiding principle 2, the
control point selected should be an overt decision or a pseudo overt decision. The
quantity of this control point is the same as the drumbeat, because the drum and the
control point are of the same flow and have the same unit dimension. This step is a
synchronization of rates within a specific flow.
(4) Search for candidate rates influenced by the focus control point, make all of the
rates to be control points, if there is no more new focus level can be found. This
step is a synchronization design of rates belonging to different flows. These rates
connect to each other through information links, though rates seldom connect to
other rates directly in system dynamics models. Because the synchronization objects
in this step are of different flow, the unit consistency should be take into account.
That means the synchronized points in this stage move at the same time as the
drumbeat, but may not move at the same speed as the drumbeat. The speed of the
synchronized points depends mainly on the unit of the flow.
(5) Search for rates that influence the focus control point in original model, make
them be the control points. This step means also the synchronization design of
rates belonging to different flows. A little different from step 4, this step is to make
those rates influencing the focal control point to be synchronous with the drumbeat.
(6) Check if the focal control point is a pseudo overt decision in the original model.
If it is, make the source limit the pseudo overt decision to change in
synchronous with the drumbeat. This step is also a synchronization design for
different flows, but only applied in the pseudo overt decisions. Because pseudo
overt decisions are decisions that have some operational limitation. To disable these
limitations, one can make them also increase or decrease as the drumbeat.
(7) Choose one of control points set in previous steps to be the newest focal control
point.
(8) Reorient the focus level to the level influenced by the new focus control point.
And repeat step 3 to step 7 until all control points are fully taken into account
for the synchronization within and between flows.
The eight steps described above constitute the basic synchronous design of a
dynamic system. There lies a basic assumption in the basic synchronous design. That is
the information about the exogenous variable’s impact on the drum can be directly
acquired by those control points, thus synchronizations between extemal environment
and the system as a whole can be achieved. Information technology developments, such
as shard databases and high speed of communication networks can make this
assumption reasonable and practicable. In addition to the basic synchronous design for
rate operations, two supplement mechanisms are developed in the dynamic alignment
approach to make the synchronous design more stable. The first supplement mechanism
is for time delay problems in synchronous systems. In fact, Time delay has been a
bothering problem for it generates various kinds of oscillation behaviors. There are two
kinds of time delay problems in a dynamic system. One is information delay, and the
other is physical delay. The former is no more a problem in synchronous systems, for
information is shared in real time by each control points. The latter is the major problem
to be solved. Consistent with the basic assumption in the dynamic alignment approach
that rates can be connect directly to each other, one simple method and mechanism is
used to supplement quantity gap due to the time lag between decision points and actual
action outputs. The core mechanism to solve the time delay problem can be illustrated
by figure 1 listed below:
pipeline volumn
pipeline in
pipeline supplement rate
work in proced 3
‘work in process 1 work in process 2
inventory
productio\rate shighing rate
production lead time
customer order
Figure 1 supplement mechanism to solve time delay problem
In the example in figure 1, customer order rate is the only one exogenous variable
influence the balance state of the system. Shipping rate is the drum to guide the pace of
production rate. The information of shipping rate is transmitted directly to production
rate so that production rate is to be synchronous with the shipping rate. However, there
is a time lag between production rate and output rate. Without no management actions
to the time lag, behaviors of the synchronization design system would like inventory
change pattem in figure 2. From figure 2, one can observe that without synchronous
design, the system is oscillating due to a sudden increase of customer order rate. After a
basic design of the synchronous system, inventory quickly achieves its new equilibrium
point without any oscillation phenomenon, but inventory does not return to its original
state due to the pipeline lost. The logic to solve the time delay problem in synchronous
system is to capture the pipeline change in “every time interval (variable of “pipeline
in” in figure 1) and to supplement the pipeline change with a new double flow rate.
After the redesign of the synchronous system, the new behavior of inventory change is
as curve 3 in figure 2. From figure 2, one can observe that inventory goes back to its
initial state quickly without any lost.
1: inventory 2:inventory 3: inventory
125.00
90.00
Curve 2: A basic synchronous system
55.00 , .
1.00 15.75 30.50 45.25 60.00
Figure 2 Comparison of behaviors with and without time delay
Lastly, this paper discusses another supplement mechanism in the dynamic
alignment approach. Because the synchronous system is a tightly coupling system,
variations in any operation of the synchronous rate will have impacts directly and
indirectly to other rates. To design a strong synchronous system, this research also
develops related mechanisms to adjust buffer size and the equilibrium level of variables.
The buffer size adjustment mechanism can be shown as figure 3.
work ipfrocess 1 work in process 2 work in process 3 invent
en
production rate
shipping "te
inventory buffer chayge rate
fé| size adjustment
custyfner order
production lead time
customer order change rate
Figure 3 mechanism of buffer size adjustment
Figure 3 is also a simple example of production chain. In that example, managers
may want to adjust the inventory level as the customer order rate change. For example,
per unit of customer order rate change may trigger 0.5 units adjustment of safety
inventory quantity at the same direction. Logic of the adjustment mechanism in figure 3
is to capture the first differential value, which is added by the inventory buffer change
rate to change the inventory level. Simulation result of the adjustment can be shown as
figure 4.
93.00
89.50
86.00
1.00 15.75 30.50 45.25 60.00
Figure 4 simulation behaviors with and without adjustment mechanism
In this section, we described the complete dynamic alignment approach. The
dynamic alignment approach contains three parts: basic synchronous design procedures,
time delay adjustment mechanism, and buffer size adjustment mechanism. With these
procedures and mechanisms, the synchronous designed system can be responsive to the
environmental change and has the ability to maintain its stability. In next section, the
dynamic alignment approach will be applied to the famous long wave model
(Strategem-2) and the redesigned system will be simulated and compared to other
leverage design methods to test the performance of the dynamic alignment approach.
3. A Case Study
Strategem-2 is a simulation game of the long wave model. It represents the capital
self-ordering feedbacks created by the fact that capital is an input to its own production.
The decision point in Strategem-2 is to balance the supply and demand for capital. That
is to minimize the average absolute deviation between supply (production capacity PC)
and demand (desired production DP) over the T periods of the game. Several system
dynamists have used the Strategem-2 as the experimental case in their research. With
Strategem-2 model, Sterman (1989) experimented human being’s misperception
behaviors of dynamic problems. In that research, Sterman suggested an optimal policy
to minimize the average score and got an average score that equals to 19. Ozveren and
Sterman (1989) proposed an optimal leverage design approach with control theory
techniques. With their approach, the average score lows down to 15. To compare with
these researches, this article uses the same values NCAT=2, ALC=20, COR=2. Also,
the score function is the same with previous researches in order to make the present
results easily comparable to the simulation results designed by Sterman (1989) and
Ozveren and Sterman (1989). But notice that DT is set to be 0.1 in this paper for the
consideration of the length of minimum delay in the redesigned model (Forrester, 1968).
Because there is no buffer size required in the Strategem-2 model, this article only
illustrates synchronous procedures and adjustment mechanism of time delay problem.
At initial, all variable in Strategem-2 are in equilibrium. The exogenous variable
arouses disturbance and oscillations is NGS (New orders of the Goods Sector). It is a
tate containing the exogenous changes, thus is the drum of the synchronous design in
the Strategem-2 model. Then, the next step is to decide control points to be synchronous
with the drumbeat of NGS. In Strategem-2, there is only one decision point NKS (New
order of the Capital Sector) for players to control. How to make NKS move
synchronously with the change of NGS is the main problem to solve. In our
synchronous design with the dynamic alignment approach, the policy of NKS is
formulated as below:
NKS=CD + order change New orders of the Capital Sector (units/year)
Order change=original order state- NGS (unit/year)
Original order state=NGS (unit/year)
In Strategem-2, just a new policy of NKS is not sufficient to make the system
sustain its stable state, for there are some positive loops hidden in the structure. Those
positive loops can amplify minor oscillations generated from the discrepancy between
production capacity (PC) and desired capacity (DP). Though NKS and NGS are set to
be synchronous, there still exists a delay to increase capital stock. To avoid any
unwanted behaviors that may be generated from that delay, we apply the adjustment
mechanism of time delay to reduce the discrepancy between PC and DP. One
supplement rate is added to the model. The supplement rate depends on two factors.
One is the discrepancy between NGS and GCA and the other is the discrepancy
between NKS and CA. The former is to reduce the impact of delay impact between
capital orders and capital acquisition, and the latter is to reduce delay of goods orders
and the capital acquisition rate of consumer goods sector. Related equations are listed as
below:
Order supplement = pipeline of NGS and GCA out +
pipeline of NKS and CA out
pipeline of NGS and GCA out = pipeline of NGS and GCA
pipeline of NKS and CA out = pipeline of NKS and CA
pipeline of NGS and GCA =pipeline of NGS and GCA + (pipeline
of NGS and GCA in- pipeline of NGS
and GCA out)
pipeline of NKS and GCA =pipeline of NKS and GCA + (pipeline
of NKS and GCA in- pipeline of NKS
and GCA out)
pipeline of NGS and GCA in =NGS-GCA
pipeline of NKS and GCA in =NKS-GCA
pineline of NGS and CA
Jing nf NGS and GCA in vipeline of NGS and GCA Out
Figure 5 Structure of the synchronous designed Strategem-2 model
The redesigned system structure is shown in figure 5. With the synchronous
designed Strategem-2, we further test its stability in confrontation with environmental
changes. The step input is used to compared with previous researches. Simulation
results of key variables are shown in figure 6 and figure 7. To compare with the other
researches, value of NKS is multiplied 2 times in figure 6 and figure 7, because the DT
in this experiment is just 0.1 year. From the two figures, one can observe the
oscillation phenomenon is completely eliminated in the redesigned system. All the
variables go to equilibrium quickly. For example, the FDS returns to 1 after the 4-year
deviation. Table 1 is the summary of key variables and it also compares the simulation
result to previous researches.
Table 1 Comparison of the simulation result
[Approach FDS |Average Score
Sterman (1989) 62% 19
'Ozveren and Sterman (1989) 80% 15
[The dynamic alignment approach [81% 13
1 800.007"
2.
uy 400.00
1.00 18.50 36.00 53.50 71.00
Figure 6 Simulation results (1)
1: FDS
1.00 7
200.00
0.50
100.00
0.00
0.00
1.00 18.50 36.00 53.50 71.00
Figure 7 Simulation results (2)
In addition to the step input , we also test other kind of demand variations, such as
the sin-wave test as Ozveren and Sterman (1989) do. The result shows that the average
score in the redesigned system can be dramatically improved also. In fact, other than the
Strategem-2 model, the dynamic alignment approach is experimented in other system
dynamics model, for example, the industrial dynamics model by Forrester (1961) and
market growth model (Forrester, 1968). All the experiments show that system behaviors
can be improved dramatically.
4.Conclusion.
Most of current high leverage related researchers use moder control theories and
complex mathematic algorithms to find optimal solutions. They focus on system
dynamics as a generic theory and methodology to solve dynamic complexity problems.
This paper attempts to connect system dynamics with other social science domains
other than mathematics and control theory to develop a leverage design approach.
Different from these researches, this paper focus on the tangible flows only. The
dynamic alignment approach developed in this research is proved to be effective in the
dynamic problems with tangible flows mainly. The dynamic alignment approach is built
on the one assumption that each rate is responsive to the exogenous input change at the
same time. This implies that the objective and limitation of this approach is focus on the
tangible flows, and the extensive usage of information technology and inter-net or
intra-net applications is the premise.
Reference
Burns, j. and D. Malone, “Optimization techniques applied to the Forrester model of the
world,” IEEE Transactions on Systems, Man, and Cybernetics, SMC-4(2), pp.
164-171, 1974.
Coyle, R.G., “The use of optimization methods for policy design in a system dynamics
model,” System Dynamics Review, 1(1), pp. 81-91, 1985.
Forrester, J.W., Industrial Dynamics, 1961, Cambridge, MA: Productivity Press.
Forrester, J.W., Principles of Systems, 1968, Cambridge, MA: Wright-Allen Press.
Goldratt, E.M., The Theory of Constraints Journal. 1987, Great Barrington: North River
Press.
Graham, A.K. Principles on the Relationship Between Structures and Behavior of
Dynamic Systems, Ph.D. Dissertation, M.LT., 1977.
Keloharju, R., Relativity Dynamics, Series A-40, 1982, Helsinki: School of Economics.
Macedo, J., “A reference approach for policy optimization in system dynamics models,”
System Dynamics Review, 5(2), pp. 148-175, 1989.
Mass, NJ. and P.M. Senge, “Understanding Oscillations in Simple Systems,” in System
Dynamics Group Working Paper D-2045-2, MIT, 1975.
Mohapatra, P.K.J. and S.K. Sharma, “Synthetic design of policy decisions in system
dynamics models: a modal control theoretical approach,” System Dynamics
Review, 1(1), pp. 84-117, 1985.
Ozveren, C.M. and J.D. Sterman, “Control theory heuristics for improving the
behavior of economic models,” System Dynamics Review, 5(2), pp. 130-147,
1989.
Srikanth, M.L. and A.E. Podzunas, Synchronous Manufacturing: Principles for
World-Class Excellence, 1990, CT, New Haven: The Spectrum Publishing
Company.
Sterman, J.D., “Misperceptions of Feedback in Dynamic Decision Making,”
Organizational Behavior and Human Decision Process, 43, pp. 301-335, 1989.
Umble, M.M. and M.L. Srikanth, Synchronous Management: Profit-Based
Manufacturing for the 21 Century-Implementation Issues and Case Studies, 1997,
Guilford, Connecticut: The Spectrum Publishing Company.
Back to the Top
