A Microworld Customized for an Oil Refinery of a
Petroleum Corporation
Showing H. Young Lihlian Hwang
Associate Professor Doctoral Student
Department of Business Management Department of Business Management
National Sun Yat-sen University National Sun Yat-sen University
O. Box 59-35, Kaohsiung, Taiwan P.O. Box 59-235, Kaohsiung, Taiwan
Tel(Fax): 886-7-5252367 Telephone Number: 886-3-3272365
E-mail address: hwanglih @tpts1.seed.net.tw
Abstract
An oil refinery tries to decrease the headcount, reduce the cost, and elevate the
ratio of high value added products to face competition. A system dynamics model is
developed in this research. In the model, decreasing the headcount and reducing the
cost lead to a decrease in unit product cost compared to the base case. However, the
two policies have little effect on cumulative gross margin. As for the policy of higher
ratio of high value added products, it leads to a significant decrease in cumulative
gross margin. These results suggest that the high value-added product ratio is a
sensitive parameter in determining cumulative gross margin. The research described
in this paper is part of a project. The project includes three stages: conceptualization,
formalization, and building a microworld. This paper focuses on the second and third
stages. For a detailed description of the first stage see Hwang and Hu (1999).
Introduction
Because Taiwan government is opening up the domestic petroleum market from this
year and the average refining costs of the oil refinery is higher than the average refining
costs of the competitors, it is important for the managers of the oil refinery to understand
the consequences and interactions of their policies. However, there is considerable
evidence which shows that managers are not good at intuiting the dynamic behavior that
will be produced by the interaction of their policies (Sterman, Repenning, and Kofman
1997, Paich and Sterman 1993, Sterman 1989).
A project that regards the application of system dynamics to develop a microworld to
support the examination, evaluation and reformulation of the policies in the oil refinery
was carried out (Hu and Hwang 1999). The project includes three stages:
conceptualization, formalization and building a microworld. Due to page limitation, this
paper focuses on the stages of formalization and building a microworld. As for the stage
of conceptualization see Hwang and Hu (1999).
Formalization
There are many ways to capture knowledge for models (Andersen and Richardson
1997; Richmond 1997; Vennix, Akkermans, and Rouwette 1996), but group model
building is still more art than science (Andersen, Richardson and Vennix 1997).
Andersen and Richardson (1997) state that “we tend not to do extensive equation writing
“live” in front of a group because there is rarely enough time for this when the whole
team is assembled and usually only a subset of the whole client team is interested in
formulation details. However, we do use two simple techniques to elicit valuable
formulation information from groups.” Ford and Sterman (1998) describe an elicitation
method that uses formal modeling and indicate the method improved model accuracy and
credibility. The experts in Ford and Sterman’s (1998) research were familiar with the
system dynamics approach to modeling product development projects and several had
received training in systems thinking. However, high-level managers and representatives
from the divisions in the refinery who paticipated in this research were not familiar with
the system dynamics approach, and the system of the refinery has many interacting loops,
therefore, participants’ knowledge was elicited in the stages of conceptualization and
building a microworld. Figure | presents a conceptual model developed in the conceptual
stage of the project (Hwang and Hu 1999).
disaster desired refining
loss goal P FC i sag
g | amount of crude oil refining amount
gal
ra gap <«__ - gap <— of oil cost goal
| disaster y
training —— loss stich rate
check pipeline gap
ie ietanc: of equipment 7
erosion resistance malfunction
oil tanks rate of cost
skill level | equipment J ‘
sales gross ;
quality _A volume margin _ reduce cost of
product Sy A purchasing
price —~» engineering
revenue maintenance
headcount
preferential retire f
no recruitment ~
headcount / aoe ps
headcount C average pay
goal ~~, gap <_
Figure 1. A conceptual model developed in the conceptual stage
In the stage of formalization, a system dynamics model was bulit on the basis of the
conceptual model, responses of questionnaires, interview data, and archival data. Figure 2
presents a sample view of the system dynamics model that contains three sectors and 160
equations. The three sectors represent the Petroleum Corporation, the oil refinery, and the
market. Complete model equations are available from the authors.
S for RLVAP unfilled customer orders of LVAP
E of OWR level on refine
LVAP -
refine for LVAP. shipments for LVAP
ys
desired refine by M
|
PO from H and
oil refine {for HVAP shipments for HVAP
: oil wait for refinement HVAP
oil received
PO from Hf and EC)
; S for RHVAP
E of OWR level on refine unfilled customer orders of HVAP
months supaly OWR direct H
desired refine by M
Headcount
desired refine by M
potential output from EC goal of refined oil by Y
AVG E of maintenance on R goal of RO in 87 months per Y
barrel per KL MCEHCCM
quality refine equipment capacity
Figure 2. Sample view of the system dynamics model
The model’s dynamic behaviour has been examined by the authors. For example,
dimensional-consistency test, extreme-conditions test, and behavior-sensitivity test were
used to validate the model (Forrester and Senge 1980).
Policy tests
Since the government is opening up the domestic petroleum market from January
1999 and the average refining costs of the oil refinery is higher than the average refining
costs of the competitors, the oil refinery tries to decrease the headcount, reduce the cost,
and elevate the ratio of high value added products in order to improve gross margin.
Because of the oil refinery has decided not to recruit anyone, these policies can be
expressed as follows:
Increase in the number of quits.
Increase in the ratio of reducing cost.
Increase in the ratio of high value added products.
Table 1 and Figure 3 show the effect of 10% increase in the number of quits on
cumulative gross margin relative to the base case. The cumulative gross margins rise until
2001 but then fall due to increased competition. The higher quits policy has little effect
on cumulative gross margin compared to the base case. This is because gross margin
equals revenue minus product cost and revenue equals volume multiplied by price and
price is determined on the basis of the unit product cost. Unit product cost was marked up
by a percentage to yield a base price level, which could be adjusted to reflect market
conditions. Table 1 and Figure 4 show the effect of 10% increase in the number of quits
on unit product cost relative to the base case. The higher quits policy leads to a decrease
in unit product cost compared to the base case. However, lower unit product cost falls
base price level, leading to a erosion of revenue. Lower revenue offsets the effect of
lower product cost. By 2003 cumulative gross margin is 2.1 percent lower than the base
case.
Table 1. Policy analysis
2003
Variable [Base Case |Higher Quits }%A [Higher ratio |%A_ |Higher ratio of |%A
lof reducing HHVAP
cost
cumulative | -937,202,635] -956,853,399} 2.1] -920,747,528] -1.76|-6,187,503,212) 560)
IGM
product 4,684 4,681} -0.06 4,665} -0.41 4,730} 0.98
cost per
unit
customer 564,924 564,924 0 564,924] 0 564,924 0
order rate
lof HVAP
price of 5,038 5,034] -0.08) 5,032) -0.12) 5,029) -0.18}
IHVAP
Notes: Higher Quits: 10% increase in the number of quits
Higher ratio of reducing cost: 10% increase in the ratio of reducing cost
Higher ratio of HVAP: 10% increase in the ratio of high value added products
Table | and Figure 5 show the effect of 10% increase in the ratio of reducing cost on
cumulative gross margin relative to the base case. The cumulative gross margins rise until
2001 but then fall due to increased competition. The higher ratio of reducing cost policy
has little effect on cumulative gross margin compared to the base case. This is because
gross margin equals revenue minus product cost and revenue equals volume multiplied
by price and price is determined on the basis of the unit product cost. Unit product cost
was marked up by a percentage to yield a base price level, which could be adjusted to
reflect market conditions. Table 1 and Figure 6 show the effect of 10% increase in the
ratio of reducing cost on unit product cost relative to the base case. The higher ratio of
reducing cost policy does indeed lead to a decrease in unit product cost compared to the
base case. However, lower unit product cost falls base price level, leading to a erosion of
revenue. Lower revenue offsets the effect of lower product cost. By 2003 cumulative
gross margin is only 1.76 percent higher than the base case.
Figure 3. Effect of 10% increase in the number of quits on cumulative gross margin
(run 2) relative to the base case (run 1)
Figure 4. Effect of 10% increase in the number of quits on unit product cost (run 2)
relative to the base case (run 1)
Figure 5. Effect of 10% increase in the ratio of reducing cost on cumulative gross
margin (run 2) relative to the base case (run 1)
Figure 6. Effect of 10% increase in the ratio of reducing cost on unit product cost (run
2) relative to the base case (run 1)
Table | and Figure 7 show the effect of 10% increase in the ratio of high value
added products on cumulative gross margin relative to the base case. The cumulative
gross margins rise until 2001 but then fall due to increased competition. The higher
ratio of high value added products policy leads to a significant decrease in cumulative
gross margin. By 2003 cumulative gross margin is 560 percent lower than the base
case. This is because customer order rate of high value added products is the same as
in the base case as can be seen in Table | and Figure 8 and price of high value added
products fell compared to the base case as can be seen in Table 1 and Figure 9.
Figure 7. Effect of 10% increase in the ratio of high value added products on
cumulative gross margin (run 2) relative to the base case (run 1)
Figure 8. Effect of 10% increase in the ratio of high value added products on customer
order rate of high value added products (run 2) relative to base case (run 1)
The oil refinery tries to decrease the headcount, reduce the cost, and elevate the ratio
of high value added products to improve gross margin. In the model, however, decreasing
the headcount and reducing the cost have little effect on cumulative gross margin. This is
because lower unit product cost falls base price level, leading to a erosion of revenue.
Lower revenue offsets the effect of lower product cost (Table | and Figures 3-6). As for
the policy of higher ratio of high value added products, it leads to a significant decrease
in cumulative gross margin. This is because customer order rate of high value added
products is the same as in the base case and price of high value added products fell
compared to the base case (Table 1 and Figures 7-9). The above simulations suggest that
the high value-added product ratio is a sensitive parameter in determining cumulative
gross margin in the case.
Figure 9. Effect of 10% increase in the ratio of high value added products on price of
high value added products (run 2) relative to the base case (run 1)
Building a microworld
In the stage of building a microworld, the system dynamics model was further
developed to a microworld. An user’s guide was designed, which contains how to use the
microworld to create their own scenarios, experiment with alternative policy options, and
get the feedbak from computer simulation results. In the computer-based learning
environment for the oil refinery, players can take the role of the director of the refinery,
create their own scenarios about oil products price and market demand, experiment with
alternative policy options for managing cost and improving gross margin.
To lessen the “video game” syndrome, the steps of play shown in Table 2 were
designed to encourage the players to state their expected outcomes prior to executing
their policies, then explain the gap between their expected outcomes and the computer
simulation results ( Isaacs and Senge 1992). The visual interface allows the participants
to explore the interactions between different scenarios and alternative policies. A
questionnaire was also designed to get feedback from the players about the items of
scenarios, policies, and results on interface.
About 16 hours were spent on introducing the microworld to the representatives
from the divisions in the refinery who paticipated in this research. The user’s guide, the
steps of play, the questionnaire were given to participants in this meeting.
Because participants didn’t fill out the questionnaire, the authors tried to interview the
participants. However, only one of the participants tested the model’s dynamic behaviour
and returned his feedback in this meeting. The main reason for non-response was that
some personnel were busy at the time the task was administered; others were not familiar
with the actual data of the refinery. One of the busy personnel recommended us another
three high-level managers who were familiar with the actual data of the refinery.
Table 2. Steps of play
1. Please write down your policies to maximize cumulative gross margin:
2. Please write down the reasons for selecting above policies:
3. Please write down expected outcomes prior to executing your policies:
4. Please enter decisions in computer.
5. Please note the gap between expected outcomes and the simulation results:
6. Please explain the gap between expected outcomes and the simulation results:
After introducing the microworld to the three recommended high-level managers, one
of them tested the model’s dynamic behaviour and returned his feedback with much
enthusiasm. The microworld was modified on the basis of the feedback from the users.
The final meeting was set to introduce the new version microworld to the high-level
managers and obtain their opinions about the value and effects of the microworld. One of
the high-level managers felt that the microworld could be used for seeking leverage
points hidden in the system through understanding the interactive relationships from
policyies under objectives of the oil refinery. This manager also suggested that the
microworld sould be used in their manager training program and the project should
continue. Another manager felt that the microworld will benefit the oil refinery, if the
microworld could probe deeply into the kinds of high value-added products, poduction,
demand, and competition.
Summary and future prospects
Due to the government is opening up the domestic petroleum market from this year
and the average refining costs of the oil refinery is higher than the average refining costs
of the competitors, the oil refinery is decreasing the headcount, reducing the cost, and
tries to elevate the ratio of high value added products to face competition. In the model,
decreasing the headcount and reducing the cost do indeed lead to a decrease in unit
product cost compared to the base case. However, the two policies have little effect on
cumulative gross margin. This is because lower unit product cost falls base price level,
leading to a erosion of revenue. Lower revenue offsets the effect of lower product cost.
As for the policy of higher ratio of high value added products, it leads to a significant
decrease in cumulative gross margin. This is because customer order rate of high value
added products is same as the base case and price of high value added products fell
compared to the base case.
The results from this paper point that the high value-added product ratio is a sensitive
parameter in determining cumulative gross margin in the case. The poduction, demand,
and competition of high value-added products should be probed deeply.
Acknowledgements
The research presented in this paper is based on a project that is funded by a grant
from the National Science Council (NSC 87-CPC-H-1 10-003). The authors are thankful
to the participants of the oil refinery for the time they devoted to the research and the
feedback they provided to improve it.
References
Andersen, D. F. and G. P. Richardson. 1997. Scripts for group model building. System
Dynamics Review13: (2) 107-129.
Andersen, D. F., G. P. Richardson and J. A. M. Vennix. 1997. Group model building:
Adding more science to the craft. System Dynamics Review 13: (2) 187-201.
Ford, D. N. and J. D. Sterman. 1998. Expert knowledge elicitation to improve formal and
mental models. System Dynamics Review 14: (4) 309-340.
Forrester, Jay W., and Peter M. Senge. 1980. Tests for building confidence in system
dynamics models. In A. A. Legasto Jr. , J. W. Forrester, & J. M. Lyneis (ed.), System
Dynamics, Vol. 14 Series: TIMS Studies in the Management Sciences. pp. 209-228.
New York: North-Holland.
Hu, Gary and Lihlian Hwang. (1999). The interaction of decisions under multiple
objectives and its time delay effects: A case study of Kaohsiung Oil Refinery of China
Petroleum Corporation. The National Science Council. NSC 87-CPC-H-1 10-003.
Taipei, Taiwan.
Hwang, Lihlian and Gary Hu (1999). Steps in conceptualizing a system for system
dynamics model-building: a case study of an oil refinery of a petroleum corporation.
Proceedings of the 1999 International System Dynamics Conference. New Zealand.
Isaacs, W. and Senge, P. M. (1992). Overcoming limits to learning in computer-based
learning environments. European Journal of Operational Research 59(1), 183-196.
Paich, Mark and J.D. Sterman.(1993). Boom, bust, and failures to learn in experimental
markets. Management Science 39: (12) 1439-1458.
Richmond, B. 1997. The strategic forum: Aligning objectives, strategy and process.
System Dynamics Review 13: (2) 131-148.
Sterman, J.D., (1989). Modeling managerial behavior: misperceptions of feedback in a
dynamic decision making experiment. Management Science 35: (3) 321-339.
Sterman, J.D., Nelson P. Repenning, and Fred Kofman. (1997). Unanticipated side
effects of successful quality programs: exploring a paradox of organizational
improvement. Management Science 43: (4) 503-521.
Vennix, J. A. M., H. A. Akkermans, and E. A. J. A. Rouwette. 1996. Group model-
building to facilitate organizational change: An exploratory study. System Dynamics
Review 12: (1) 39-58.
