-1017-
MRP and JIT: Teaching the Dynamics
of Information Flows and Material
Flows with System Dynamics Modeling
Merrill E. Warkentin .
University of Nebraska-Lincoin
Abstract
System Dynamics modeling is used as an instructional aid for the
teaching of production and inventory management techniques. The
roles of Material Requirements Planning (MRP) and just-in-time
(JIT) systems in production and inventory management are pre-
sented and discussed. By modeling these manufacturing systems,
the student can acquire an appreciation of the dynamic relation-
ships between the elements of each system. Some elements of the
Dynamo models of these systems are presented. The future oppor-
tunities and research needs are discussed.
Introduction
Every instructor of Production and Operations Management (P/OM)
has experienced the frustration of trying to bring about a com-
plete understanding of the several production and inventory
Management (P & IM) models. The Economic Order Quantity (E0Q)
method, the Material Requirements Planning (MRP) methodology, and
the newer "just-in-time" (JIT) approach are three tools used for
organizing and controlling the manufacturing assembly line.
These systems are usually described in a lecture, are compared
and contrasted, and are left for the student to conceptualize.
The limited homework and classroom problems serve as the only
"hands-on" experience with such manufacturing management tech-
niques. While well-written sources exist, it is often difficult
to achieve thorough understanding of the dynamic characteristics
of each system. Without any actual experience in observing the
behavior of a shop floor under MRP or JIT conditions, the student
can never really gain a full appreciation for the advantages,
routine difficulties, and special-case problems of each system.
The technique of Dynamo modeling, developed during the 1950's by
Jay W. Forrester at MIT's Sloan School of Management (Meadows
1980, p. 30), can be used to facilitate the instructor of such P
& IM techniques. Because the primary goal is the teaching of the
systems' characteristics under varying conditioners, Dynamo
modeling is especially suited to this task. It provides the
ability to generate generic standard models which can be "fine-
tuned" and subjected to many individual internal and external
events and changes.
-1018-
Production and Operations Management
The production process can be viewed as a simple open system with
inputs (capital, labor, managerial skills, etc.), the conversion
process itself, environmental and other influences (economic
fluctuations, political-legal-regulatory impacts, random factors,
etc.) , outputs (goods and services), and the various flows of
information such as feedback from output monitoring (inventory
levels, sales volume, plant efficiency, quality considerations,
etc.). The information feedback elements of this system are com-
plex and varied, and can be both positive and negative. Figure l
shows the basic structure of this simple model. One major goal
(desired behavior) of this system is efficiency (as measured by
the relation between output quantities and input quantities).
Another primary consideration is the level of quality of the out-
put (as measured by consumer response, return rates, product test
results, etc.).
Figure 1
The Basic Production Process Model
Environmental and Other Influences Ye
Late Deliveries oe
Economic Conditions (business cycles, long waves \.
vo interest rates, random, etc.)
Pa Political/Legal/Regulatory environmental impacts \
va Labor Turnover and Strikes \
Shop floor factors (equipment breakdowns, etc.)
Lf Ecological Impacts and Resource Availability \
eit Random factors, etc.
f INPUTS OUTPUTS
{ Capital (bldg, ma Finished Goods
, | equip, land,etc) \ Intermed Goods |
\ abor raw materials goods and / i
{ Managerial Skill}” purchased parts services / J /
\ U j
(— feedback /
onversio / hipments|
Process /
/
contr
ra
\ nl
SUPPLIERS
on
actual vs. ial vs. desired
-1019-
However, the real work of the manufacturing manager occurs within
the conversion process itself. The methods employed for organiz-
ing and controlling the conversion of inputs into outputs have
lately been given a great deal of attention by the P/OM profes-
sion. While many large companies with complex production systems
have used Master Production Scheduling (MPS) for many years, the
technique has only recently filtered down to many smaller compa-
nies with smaller budgets. At the same time, there is growing
interest in new production management concepts. The American
Production and Inventory Control Society (APICS) has a Zero
Inventory Committee which advocates the conversion to stockless
production. This rising interest in just-in-time (JIT) manufac-
turing has coincided with the growing concern over price competi-
tion and quality competition from the Japanese manufacturers.
Managers of American factories have employed various methods for
organizing the production activity. Perhaps the oldest and sim-
plest technique used to organize and control the mass-production
assembly line is the economic order quantity (E0Q) method, where
stocks are replenished when they reach the reorder point (ROP).
This is a lot-for-lot “pull system" for stocking each inventory
location in the system. When inventory levels at one point
(purchased parts inventory, work in process, finished goods
inventory, etc.) fall to some prescribed quantity, more parts are
ordered (pulled). The order quantity is generally large so that
scale economies (from purchasing, transportation, order-
processing, and setup costs) can be realized. Yet the lot size
or run size must not be so large that excessive carrying costs
are incurred. Thus we speak of the "economic order quantity" or
EOQ. Figure 2 graphically represents this derivation. Manufac-
turers that employ this algorithm "do so because of a difficulty
in associating parts requirements with the schedule of end
products." (Schonberger 1983, p. 64)
Figure 2
Economic Order Quantity
Total cost —_
Order-processing
and setup cost
Material requirements planning (MRP), is yet another popular
method for managing the production process. MRP, a critical com-
-1020-
ponent of an overall manufacturing planning and control system,
is a method for planning and controlling inventories so that
sequential work centers are provided with the materials they
require to produce the output planned by the Master Production
Schedule (MPS). Finished goods inventory (FGI) levels are deter-
mined by the output of the demand management function. The
intermediate inventory levels (work in process, WIP) are computed
by the MRP system from current inventory levels and planned oper-
ations affecting inventories. Timing is very critical in the MRP
system; future needs are set by the MPS, and the problem is cal-
culating when to operate each work center in the system (and when
to order). Each step in the process is "backscheduled"; the
timing and quantity of the final stage can be easily calculated
and used to calculate the second-to-last stage which is used to
drive the production schedule for the stage before it. There is
little flexibility in the order quantities because of lot size
constraints (EOQ, setup costs, etc.). Equipped with the MPS, the
EOQs, the setup times and costs, the bill of goods (BOG) informa-
tion, and the current inventory levels, the MRP system can use a
high-speed computer to solve the set of equations so that the
production schedule for each work center (and thus the WIP
levels) can be carefully calculated. The master schedule of
finished goods is translated into a schedule for hundreds of
parts requirements. This computer output is used to determine
the material flow for the entire plant. It should be noted that
MRP-organized plants also require "shop floor control" to fine-
tune the computer-generated schedule. It is the job of the
expediters and shop managers to override the schedule determined
by the computer if this becomes necessary. Figure 3 shows the
basic design of this manufacturing planning and control system.
A third method for organizing the manufacturing process is the
"just-in-time" (JIT) system. This concept is also referred to as
stockless production, zero inventories, the Kanban system, and
the pull system. Actually these terms are related, but each
describes a unique concept. "The JIT idea is simple: Produce
and deliver finished goods just in time to be sold, subassemblies
just in time to be assembled into finished goods, fabricated
parts just in time to go into subassemblies, and purchased
materials just in time to be transformed into fabricated parts."
(Schonberger 1982, p. 16) The just-in-time/total quality control
(JIT/TQC) system is much more than an inventory replenishment
system; it becomes a major influence on every aspect of the pro-
duction process from purchasing through distribution. The reduc-
tion of lot sizes triggers a chain reaction of benefits in the
plant. (Schonberger 1982, p. 18) In a JIT system, the ideal lot
size is one. One of the benefits of minimum lot sizes is lower
scrap with improved quality. "If a worker makes only one of a
given number of parts and passes it to the next worker immedi-
ately, the first worker will hear about it soon if the part does
not fit at one of the next work stations. Thus, defects are dis-
covered quickly and their causes may be nipped in the bud; pro~
duction of large lots high in defects is avoided." (Schonberger
1982, p. 25) The JIT system with its Kanban inventory control
system is gaining in popularity in American plants because of the
-1021-
Many advantages it holds over the MRP system in most cases, For
an example of a Kanban system, see Schonberger 1982, pp. 221-224.
Figure 3
Manufacturing Planning and Control System
Resource Production Demand
——ens —
planning planning - management
Rough-cut Master
capacity |<} production Front
plannina scheduling
y
Material te
al
enton
Routing Bills of fequiraiont ventory
file material alanain statu:
9 data
4
Detailed Timed-phased
capacity requirement
planning records
F | cre
/ Material
and
/ capacity plans
\ ier Purchasing
\ release |
\ Back
end
Vendor
Shop- floor tollow-up
control systems,
The real distinction between these systems is their performance
"under duress", their operation in the dynamic setting of the
-1022-
factory floor. If demand is certain, if labor levels are given
(no turnover, no strikes, etc.), if there are no machine break-
downs, and so on, any of these systems can be successfully en-
ployed as a management tool. But when production problems arise,
and they will, each system exhibits a unique behavior with
respect to the manufacturing system variables (see Figure 4).
The dynamics of each system are the focus of this project.
P/OM Education Using Dynamo Modeling
The task of teaching Production and Operations Management con-
cepts to young students can be both challenging and rewarding.
One sometimes frustrating element of this charge is the descrip-
tion of the dynamics of inventory and information flows. on the
shop floor under various conditions and system controls. One
innovative approach to enlightening the students is the use of a
production game where each individual (or team of individuals)
represents one work center in an assembly process. Each team has
responsibility of managing its inventories through forecasting
and ordering inputs. By introducing various parameter changes
(such as demand pulses) into the game conditions, the partici-
pants can observe how the components react to the information
flows with ordering activity and inventory adjustments.
A better approach to teaching these concepts is to allow the
students to model the inventory management systems with the
Dynamo simulation compiler. This System Dynamics approach is
especially suited to the task. It allows the user to test a
great variety of conditions and influences easily and quickly.
In order to use Dynamo modeling to teach P/OM, the instructor
must first introduce the students to the basics of System Dynam-
ics theory and Dynamo simulation language techniques. The goal
is to teach the students as much of this body of knowledge as is
necessary to facilitate the education of the production systems.
Primary emphasis will be on MRP and JIT. Students must under-
stand simple model behavior -- growth and decay, cyclical activ-~
ity, and so on. They must also understand the information flows,
positive and negative feedback, and general impact relationships
between rates, auxiliaries, levels, and overall system para-
meters. By teaching the student these concepts, the teaching of
any system will be substantially enhanced. Because the goal is
not Dynamo modeling expertise, only basic techniques need to be
presented; the students must possess basic proficiency at
modeling the systems.
The overall production system (in Figure 1) is first presented to
the students. This basic model contains two levels and three
rates with additional auxiliary variables. The students are
introduced to these variables using the terms "stocks" and
"flows". The students are also presented with the information
feedback element of this first model. The lecture and discussion
should revolve around the feedback and control loops. By concen-
trating on these elements, the students will gain an appreciation
-1023-
for the dynamic nature of the production system, its components,
and its environment.
It is a reasonably straightforward task to diagram and describe
the systems, their components, and the basic relationships to a
large class of Introductory P/OM. But the real goal of any good
instructor is to enlighten the students in the way these systems
"really work"; that is, how they behave under realistic condi-
tions. Only through a simulation exercise can the student gain a
real appreciation for the way in which these systems operate
under various environmental and internal influences. By care-
fully constructing a model of a JIT system, for example, the
student must not only carefully think through the many relation-
ships in the system, but he must also ensure that each relation-
ship will be valid under extreme conditions (when production
falls to zero, for example). By performing the validation tests,
the student can ensure that the modeled system is logical and
realistic. (If stock levels become negative, for example, the
structure is probably inaccurate.
Figure 4
Manufacturing System Variables
Inventory turns Percentage of purchase
Cost of inventory investment discounts taken
Month's supply on hand Percentage of stockouts
Dollars of back orders Machine utilization
Days to fill an order Indirect labor
Customer service level Average shop order time
Number of open purchase orders Percentage of order split
Number of open shop orders because of shortages
Percentage of orders expedited
These system components and their behavior under various system
parameter combinations can be taught to the P/OM students using
the Dynamo modeling exercise. After a brief introductory lec-
ture, which might be accompanied by an instructional film, the
instructor can direct the class through a group modeling session.
Through this group decision process, the students can debate and
formulate the system structure. At the next class meeting, the
students can follow up and make any necessary changes to the
equations in the model. Finally, they can run the model under a
variety of model parameters. This process will ensure that they
are able to witness the behavior of each component of the model
under various conditions. In this way, the system's dynamic
aspects can be elucidated.
Figure 5
Water Flow / Material Flow Analogy
1. With high levels of work in process inventory (WIP), the
material (water flow) accumulates in inventory locations (deep
pools). These bottlenecks, which slow the flow through the
system, cannot all be seen from the surface.
2. By reducing the level of WIP, the obstructions (rocks) are
exposed. By solving these problems, the WIP level can be safely
reduced again until new obstructions (more rocks) are uncovered.
This process can continue incrementally until most of the
hindrances have been remedied.
TN
3. As the sources of material impedence are removed (the channel
is cleared of rocks), the material flows through the system at a
uniform rate of flow; the WIP is not allowed to accumulate.
Therefore, any quality problem will be discovered by the next
work center or inspection station. When problems are discovered,
only a small lot will need to be reworked or discarded, rather
than a large amount of inventory which has been accumulating like
water in a pool.
a
\ SNe
a a
-1025-
Material Stocks and Flows
It is said that the function of inventories is the decoupling or
separation of the sequential stages in the production and distri-
bution of a product so that successive stages can operate inde-
pendently. If raw materials and purchased parts are delivered
late, the maintenance of safety stock will preclude the need to
shut down operations. Similarly, if buffers are maintained at
work centers throughout the plant, the stopping of one machine
due to malfunction or labor shortage will not dictate the need to
shut down other work centers. Yet the maintenance of buffer
stocks can also be viewed as a system for hiding inefficiencies,
bottlenecks, and other problem situations which should be exposed
and solved. Figure 5 depicts the water flow analogy (Hall 1983,
p. 13, and others) which demonstrates the potential gains from
reducing the work in process inventory level. This analogy
demonstrates a primary reason for implementing a JIT or stockless
production system. By uncovering these problems (rocks), the
entire production system can be made to operate more efficiently.
In addition, quality can be enhanced as potential sources of
defects are exposed and removed. It has been said that the
Japanese manufacturing companies produce small quantities of
output "just in time" while their American counterparts produce
wastefully large quantities "just in case."
The dynamics of these stocks and flows can be targeted in the
model formulation stage. Rather than concentrating on exact
measurement of the components of the Dynamo models, the qualita-
tive impact of the information flows must be emphasized to the
student modelers. And rather than emphasizing the mechanics of
the production process, most students would learn a great deal
more by discussing and modeling the general nature of the
information flows and system controls.
MRP and JIT Dynamo Models
A hypothetical model which might be generated by the P/OM student
for the JIT system is presented in Figure 6. While there can be
no generic production system model because the BOGs and informa-
tion delays are unique to each real-world system, the students
will each generate unique models from which they can learn about
these production systems. The MRP model is driven by a complex
set of auxiliary variables which represent the Master Production
Schedule and the Material Requirements Planning function. The
JIT model, however, is simpler in design. Each work center is
driven by the next with a single item of information -- the level
of WIP between -- triggering the activity. In other words, work
center #3 (a rate variable) does not become active (processing
WIP) until the inventory location ahead of it falls to some small
quantity. This can be visualized as essentially a chain of rates
and levels with some delays built into the system.
-1026-
Figure 6
Just-in-Time Model
Work Center #2
WIP5
—
Kica | wea
WIPE WIP7 WIPS
\ ? i
3 Ih,
: WTR Weg
irre arse raen
SR
Purchasing E
‘ P - Forecasted - KY
Actual wees Se ae Channels of
Demand ~ 7o- ---- + Distribution
/ oF
(F) Optimism Constant _~ RN
/ \ \ - y yh
3 , yo. N
Suppliers Industry ) Ped \
Factors \ Industry i
\ Factors )
}
and Raw Market
Material ; Share Gross \
Sources National pre \
Perceived Product Price BS
Quality Pau} ~~
Level
Quality Buffer
Module
-1027-
Conclusions
The many complexities of these systems of production could not be
explored in this brief paper, but the reader can appreciate the
relative difficulty of teaching the dynamic nature of their
material and information flows. By modeling the several systems
used in manufacturing, the student gains a greater appreciation
of the relative complexity/simplicity of each system and of the
systems' components and relationships. More importantly, as the
student gains an appreciation for these relationships and the way
in which each component (level or rate) affects, or is affected
by, other system components, the overall dynamic behavior of the
system becomes evident. After repeated simulation runs are
followed by careful analysis, the students are able to acquire a
comprehensive understanding of the manufacturing management
systems. This approach to teaching P/OM concepts offers the
prospect of greater comprehension of production systems and of
their dynamics. This better understanding can be achieved in a
classroom setting -- a distinctly appealing prospect in a field
of study which is heavily practitioner-oriented and often
difficult for many students to grasp.
A pilot study for the development of a general packaged approach
to this exercise is being developed. By experimenting with sev-
eral groups of students, areas of learning difficulty will be
exposed and specific course procedures will be targeted for fur-
ther development attention. This learning package should be
presented upon completion of a set of required readings. This
course of learning may be structured as a series of microcomputer
modules that each individual student can "check out" from a
software library and study at his or her own pace. However, the
flexibility of this approach would be gained at the expense of
the group interaction learning process.
The overall objective of the development of a useful tool for
facilitating the instruction of P/OM concepts, particularly the
MRP and JIT systems, requires careful planning, analysis, and
design. Additional research toward this goal should include an
analysis of the process which students normally experience when
they are introduced to these inventory management systems. Do
they readily envision the dynamic nature of the MRP and JIT sys-
tems? Do homework problems which show the student how to gener-
ate a simple schedule from demand data lead to systems thinking
on the part of the average student? Does the introduction of
system dynamics modeling overload the student with mental con-
structs and detract from his understanding of the production
systems or does it unequivocally add to his understanding of the
production process? These and many other questions need to be
addressed as this project continues.
References
y San Jose: CA: IBM
hniesteeins Industry Support Center, 1960.
-1028-
DeMott, John S. "Manufacturing is in Flower," Time, March 26,
1984, pp. 50-52. (with "In Quest of Quality," p. 52)
Forrester, Jay W. Industrial Dynamics. Cambridge, Mass.: MIT
Press, 1961
Forrester, Jay W. "System Dynamics - Future Opportunities," TIMS
14, pp. 7-12, North-
Holland Publishing Company, 1980.
Garvin, David A, "Product Quality: An Important Strategic
Weapon," Business Horizons, March-April 1984, pp. 40-43.
Hall, Robert W. Zero Inventories. Homewood, Ill.: Dow Jones-
Irwin, 1983.
Huang, Philip Y., Rees, Loren P., and Bernard W. Taylor III, "A
Simulation Analysis of the Japanese Just-in-Time Technique
(with Kanbans) for a Multiline, Multistage Production
System," Decision Sciences 14, 1983, pp. 326-344.
Meadows, Donella H. "The Unavoidable A Priori," in Elements of
the System Dynamics Method, edited by Jorgen Randers,
chapter 2, pp. 23-56. Cambridge, MA: MIT Press, 1980.
Morecroft, John D. W.
i , Ph.D. thesis, Sloan School of Management,
MIT, 1979.
Schonberger, Richard J. “Applications of Single-Card and Dual-
Card Kanban," Interfaces 13:4, August 1983, pp. 56-67.
Schonberger, Richard J. and Abdolhossein Ansari. ""Just-In-Time"
Purchasing Can Improve Quality," i
Materials Management, Spring 1984, pp. 2-7.
Schonberger, Richard J. Japanese Manufacturing Techniques: Nine
’ in Simplicity, New York: The Free Press,
1982.
Vollman, Thomas E., William L. Berry, and D. Clay Whybark.
, Homewood,
Illinois: Dow Jones-Irwin, 1984.
