In Proceedings of the Symposium on Industrial Engineering and Management (pp. 249-256). Toronto: Canadian Society for Mechanical Engineering.
Modeling Techniques to Support Abnormal Situation Management in the
Petrochemical Processing Industry
Greg A. Jamieson and Kim J. Vicente
Cognitive Engineering Laboratory
Department of Mechanical and Industrial Engineering
University of Toronto
Individual levels of the Abstraction Hierarchy
ABSTRACT (AH) are related to adjacent levels by a means/ends
We describe the use of the Abstraction relationship. When transitioning between levels, an
Hierarchy framework to model a petrochemical operator can exploit these relationships to ask three
system. The framework provides a description of crucial questions (see Figure 1). By entering any
the physical and functional relationships within the level of abstraction operators are implicitly asking
plant which reveal opportunities for operator action themselves the question, “What?” More
in both normal and abnormal situations. We specifically, “What is the form, function, or purpose
demonstrate that the Abstraction Hierarchy can be that I am interested in?” When traversing up the
meaningfully applied to petrochemical systems as it levels of abstraction, the operator can ask the
has been in other domains and we discuss some question “Why?”. That is, “Why does the structural
implications for future research. description of the plant include equipment X or
function Y at this level?” At the adjacent level
1. INTRODUCTION above, the operator should find the answer to that
The inability to effectively manage abnormal question in a more abstract function or purpose.
situations exacts an annual $20 billion toll from the Similarly, when stepping down to lower levels of
U.S. petrochemical industry . This occurs despite abstraction, the operator can ask the question,
continual technological developments in advanced “How?” For example, “How can function Y or
control systems and training of operational purpose Z be realised?” The adjacent level below
personnel. Many of these developments have been should denote the functions or equipment that are
based on an implicit assumption: that the range of pertinent to answering such questions. The
representative events considered by designers are WHY:WHAT:HOW questions form a window that
sufficient to cover all possible contingencies. can be vertically translated through the levels of
However, experience with large industrial systems abstraction. That which constitutes a WHY in one
inevitably leads to the conclusion that unanticipated window can serve as the WHAT if the frame is
events will occur regardless of the extent of moved up one level of abstraction.
engineering analysis and planning. Further, it is that
very unanticipated variability which represents the WHY?
greatest threat to plant safety and productivity .
How does one model a plant so that the Purpose WHY? Z1 Purpose WHAT? Z1
representation is both useful and meaningful to
operators forced to contend with unanticipated Function WHAT? Y1 Y2 Function HOW? Y1 Y2
Equipment HOW? X1 X2 Equipment X1 X2
The Abstraction Hierarchy Representation Figure 1: The shifting WHY:WHAT:HOW window
The Abstraction Hierarchy  is a multi-level in the Abstraction Hierarchy.
representation of the structure of a plant. Although
the number and nature of levels are not fixed, in the The AH can be complemented by a second
process control domain we have found it useful to dimension to describe the physical aggregation of
include models of production and safety goals, first components at various levels of resolution.
principles, general functions, plant equipment, and However, the nature of the relationship described
equipment location and appearance. Each level of along the aggregation dimension is conceptually
the Abstraction Hierarchy constitutes a complete distinct from the means/ends relationship described
system model and is distinguished by a specific along the abstraction dimension. The various levels
language employed at that level. of aggregation are ordered by a part/whole
relationship. For example, a casing, impeller and a The McFarlane et al. model is a highly
motor might be aggregated to form a pump. The simplified model of an FCCU. However, it is
individual components can be treated separately or designed to capture major system dynamics in order
they can be treated at a lower level of physical to explore various control configurations . The
resolution (i.e., higher aggregation) as a pump. important criterion in this case is the degree to
These orthogonal dimensions of abstraction and which the model is representative of the
aggregation define an area over which a number of complexities of existing FCCUs. The McFarlane et
different plant models are described (see Figure 2). al. model is multivariable, nonlinear, and features
Although each of these models reveals a unique set strong interactions between sub-systems. Further, it
of information about the modeled system, we imposes both mechanical and operational constraints
emphasize that each is a complete model within its that would be found in actual applications .
particular cell in the full abstraction/aggregation Given these characteristics, the McFarlane et al.
area. Navigation through this area facilitates an model seems to strike a good balance between
understanding of both the physical and functional representativeness and simplicity.
relationships between plant elements. A major advantage of the McFarlane et al. 
Finally, it is often helpful to specify what an model is that a full description of the model
abstraction hierarchy is not. The AH is not a equations is provided. The modeling techniques
specification of events, situations, or plant states. described here can only yield models as precise as
These concepts are temporally restricted whereas the engineering process models on which they are
the structural representation provided by the AH is based. Our goal is to transform the algebraic and
relatively invariant over time. That is to say, the AH differential equations that engineers use to describe
is event-independent . The model content is not the plant into a model that is psychologically
specified via a finite set of abnormal events which relevant to operators challenged with controlling
are anticipated by designers. Further, the AH does these processes .
not specify operator tasks or goals. While such a
specification can be useful, tasks and goals can vary 2. MODELING THE STRUCTURE OF THE
while structure remains constant. FCCU
In this section we apply the AH modeling
The McFarlane et al. FCCU Model framework to the McFarlane et al. model of the
The Abstraction Hierarchy has been employed FCCU. In the following section, we highlight some
previously to model a simulated thermal-hydraulic lessons learned from the application of the
process , a simulation based on an existing framework to this novel domain.
pasteurisation process , aircraft engineering
systems , a power plant feedwater system , and The McFarlane et al. Abstraction Hierarchy
conventional  and nuclear  power production. Space
In order to test the feasibility of employing the AH Figure 2 provides an overview of the regions in
in the petrochemical domain, we have employed a the abstraction/aggregation space which have been
partial plant simulation described by McFarlane, defined for the McFarlane et al. FCCU. Eight cells
Reineman, Bartee, and Georgakis . Henceforth in the space have been found to be useful in
we will refer to the simulation as the McFarlane et describing the plant. The arrow between the
al. model. This simulation focuses on the Physical Function and Generalized Function levels
reactor/regenerator section of a Fluid Catalytic at the Component level of aggregation indicates that
Cracking Unit (FCCU). Within a refinery, an FCCU we have constructed a diagram detailing the
breaks down high boiling point input feeds and transition between these cells.
separates valuable products from waste products.
Within the FCCU, the reactor breaks up the
hydrocarbon chains by combusting input feed with
the help of a catalyst. The regenerator serves to
clean the catalyst used in the reactor. The FCCU is
the economic heart of a refinery. To a large extent,
its successful operation determines whether or not
the refinery will be profitable . The critical
nature of its employment makes the FCCU an ideal
candidate for exploring the techniques described in
Figure 3 provides a representation of the levels
of aggregation. It emphasizes the manner in which
System (S) Sub-system Unit (U) Component
(SS) (C) lower level nodes are aggregated to form higher
level nodes. We give the levels of aggregation a
Functional cursory treatment in this paper because part/whole
$$ hierarchies are common in system descriptions.
What is important at this stage is that the reader
recognize that the four levels of aggregation cut
Function AF-SS AF-U AF-C (mass) across the five levels of abstraction (see Figure 2).
(AF) AF-C (energy)
Thus, at each level of abstraction there are four
Generalized possible complete descriptions of the plant.
Function GF-SS GF-U GF-C
PART - WHOLE (LINKS)
Physical CATALYST PRODUCT SEPARATION
HEAT AND FEED
SUB-SYSTEM AIR INPUT REGENERAT ION GENERATION AND OUTPUT
Form SUBSYSTEM SUBSYSTEM SUBSYSTEM SUBSYSTEM INPUT SUBSYSTEM
Figure 2: The abstraction/aggregation topography UNIT LIFT AIR COMBUSTION REGENERAT OR REACTOR
SUPPLY AIR SUPPLY UNIT UNIT
MAIN WET GAS
FEED HEAT FEED
for the McFarlane et al.  model FCCU.
UNIT UNIT UNIT UNIT UNIT UNIT
In reviewing Figure 2, the reader might
question why only 8 of 20 cells in the AH space E
have been described. We emphasize that all cells of
the AH are valid and potentially useful Figure 3: Levels of Aggregation with emphasis on
representations of the plant. However, in relationships between levels.
constructing an AH we must evaluate which cells
are likely to be most useful to operators. Each cell The Level of Physical Form
has been evaluated in terms of what information is The reader will likely notice that no cells at the
added/lost when transitioning from one cell to Physical Form level are defined in Figure 2. This is
another. When a representation is constructed for a because there is no physical instantiation of the
given cell, we evaluate what information is McFarlane et al. FCCU. The plant is exclusively
contained therein which is not contained in other simulated and thus has no Physical Form to model.
cells. If there is little or none, we sacrifice use of In an operational plant this level would be filled
that cell. Through this evaluation process we can with representations of the location and appearance
limit the presence of marginally useful information of the physical components.
that might drain the limited cognitive resources of
the operator. The Level of Physical Function
Experimental observations of problem solving The Physical Function (PFn) representation
behavior (from which the AH concept was initially resembles a traditional piping and instrumentation
formulated) have shown that such behavior typically diagram except that instrumentation is not specified
falls along the diagonal from the Physical by the AH (see Figure 4). Also absent from this
Function/Component cell to the Functional representation are the various controllers employed
Purpose/System cell . Not surprisingly, the in the FCCU model. In constructing an AH, we pay
evaluation process described in the preceding particular attention to restricting our descriptions to
paragraphs frequently yields a set of representations the means/ends structure of the plant elements.
that fall along the same diagonal. When operators While control systems are crucial to the successful
think about the purposes and functions of a plant, operation of a modern petrochemical plant, they do
they tend to adopt a coarse unit of analysis (e.g., not lend themselves to characterisation by
system, subsystem); when operators think about means/ends descriptions. In our work, we employ a
physical properties of a plant they tend to adopt a different framework (not described in this paper) to
fine unit of analysis (e.g., component). model the behaviour of control systems.
The lines connecting the nodes at the PFn level
Levels of Aggregation represent physical relationships between the
structural components. For example, the two U- abstraction, each complete yet unique in the type of
bend lines connecting the reactor and regenerator information it provides.
imply that there is a direct physical connection Figure 5 details the transition between the cells
between these units. While this statement may of the PFn and GF levels at the Component level of
strike the reader as being obvious, higher levels of aggregation. It is the manifestation of the arrow
abstraction do not follow this rule. The reader between the cells visible in Figure 2. Such a
should be careful to understand the meaning of representation is not typical of AHs and was initially
connections between nodes at each level. constructed as a memory aid for the authors. In
review, however, we realised that it was a valuable
The Level of Generalized Function explanatory tool and we have continued to find it
The level of Generalized Function (GF) reveals useful. This representation provides a very detailed
information about heat transfers and flows of explanation of how each node at the PFn level is
commodities (e.g., catalyst, feed, products). We related to its associated end(s) at the GF level.
have found it useful to discuss chemical reactions at Conversely, each node at the GF level is connected
this level also. The primary reason for including to its mean(s) at the PFn level. Thus, a given node
reactions at the GF level is that the language can have a single or multiple means and ends,
typically employed here lends itself well to emphasizing the homomorphic nature of the AH
discussing reacting commodities. Further, chemical representation.
reactions are equally subject to the mass and energy There are a couple of noteworthy features in
first principles that are described at the Abstract Figure 5. One is the empty node at the PFn level
Function level (see below). In other words, we can which is connected to the Gas Oil Flow node at the
talk about chemical reactions at the GF level and GF level. This flow is different from other flows in
emphasize which commodities are reacting, in what this simulation in that it has no regulating valve.
proportions, with certain products and heat transfers. Our conversations with process engineers indicate
At the Abstract Function level we can talk about that it is not uncommon for flowrates in FCCUs to
relevant mass and energy relations. Thus, the same be determined by upstream processes over which the
chemical reaction can be treated at two levels of FCCU operators have no control. This lack of
Fraction- Wet Gas
Lift air Steam
Combustion Fuel Gas
Figure 4: The Physical Function level at the Component level of aggregation.
Wash Oil Diesel Fresh Feed Gas Oil Slur ry Heat Fuel g as Combusti on 1, Heat Exhaust, Air Sunction Suction Comb. Ai r Dis-
Flow, F1 Flow, F2 Flow, F3 Flow, Fg o Recycle Transfer 3 Flow, F5 F5Hfu /lm Transfer 1, Qloss Flow, F V6 Pressur e, Throug h- char g e
Flow, F4 UAfT lm/lm P1 put, F6 Press.,
V1, C 1 V2, C 2 V3, C 3 V4, C 4 V5, C 5 Furnace, T3 V6, C 6
Lift Air Dis- Lift Vent Heat Slur ry/WG Fracti onator Flar e Gas Wet Gas Suction Wet Gas Dis-
Atmospher ic Spill Air
Press, Patm Throug h- char g e Flow, F V8 Flow, F10 Transfer Separ ation Press., P5 Flow, FV12 Flow, F V11 Presure, Compressor char g e
Flow, FV7 put, F8 Press., P7 Flow, F11 Press.,
Blower, s a Wet Gas
V7, C 7 V8, C 8 V9, C 9 Fracti onator V12, C 12 V11, C 11 Compressor
Wet Gas Spent Reg enerated Catalyst/WG Heat Cr acking Coke Pressur e at Catalyst Reactor Catalyst/ Heat
Anti- Surg e Catalyst Catalyst Separ ation, F wg Transfer, Reaction, Production, Riser Bottom, Inventor y, W r Press., Hydr ocar bon Transfer 3
Flow, F V13 Flow, Flow, Qff , Qslurr y, Qcracki ng F coke Pr b P4 Separ ation
Fsc, Qsc Fr gc, Qr gc Qsr, Qfr
V13, C 13 U- bend U- bend Diseng ag i ng
Cycl ones Riser, Tr Reactor Str ipper
( sl ide valve) ( sl ide valve)
Reg enerated Catalyst Reg enerator Combusti on Reg enerator Fluidi- Stack g as Spent Lift Air Fl ow, Catalyst / Stack
Catalyst Pressur e
Catalyst Outflow, F sp Pressur e, P6 2, Q C Bottom zation Flow, F sg , Catalyst air ,li ft, Gas Separ ation
Reser voir , at bottom
Reser voir , Wsp W r eg Pressur e, P r gb Qfg , Qe Flow, cat,l ift cat,l ift, of l ift pipe,
Qair, QH Pblp
Stand Diseng ag i ng Fluidized Lift Cycl ones,
Pipe, l sp Section Bed, T r eg V14, C 14 Pipe T cyc
Figure 5: Detail of transition from the Physical Function to Generalized Function level.
opportunity for action is clearly reflected in the Changes in the state of the equipment will lead to
AH. In a fault management situation involving multiple changes in flows and heat transfers.
this flow, a well designed interface should make it The GF-Component cell representation is
clear to the operator that he has no capability to shown in Figure 6. The nodes at this level represent
affect this flow. A second point of interest is that general functions of the plant, e.g. flows, heat
many nodes at the PFn level have multiple ends. transfers. The connections between these nodes
represent causal relationships. Note that
Catalyst/ Fracti onator WG Anti- sur ge
Flar e Flow, F V12
Hydr ocar bon Press., P5 Fl ow, F V13
Wet Gas, Wet Gas
Catalyst/WG Slur ry/WG Compressor
Reactor Separ ation, F wg Separ ation Flow, FV11
Flow, F11 Dischar g e
Catalyst Press., Pvru
Heat Tr ansfer 3 Inventor y,
Coke Fl ow, F coke Product
Press. at Outflow
Lift Pipe, P blp Slur ry Flow, F4 &
Cr acking Reaction, Heat Tr ansfer 3
Reactor Qcracki ng Wash Oil Flow, F1
Lift Air Press., P4
Blower Spent Cat.
Lift Air Dischar g e Flow,
Throug h- Press., P3 F sc, Qsc Heat Tr ansfer 2
Press. at ( latent , Qsr, Qfr, Fresh Feed Diesel Fl ow, F2
put, F8 Flow, F3
bottom of and sensi bl e,
Lift Air Fl ow, r eactor ri ser , Qff , Qslurr y)
F9, air ,li ft, Pr b Heat Tr ansfer 1, Gas Oil Flow, Fg o
Combusti on UAfT lm/lm
Lift Vent cat,l ift, Qair, 2, Q C
QH Reg en. cat.
Flow, F sp
Fr gc, Qr gc Combusti on 1, Exhaust,
F5Hfu /lm Qloss
Spill Air Catalyst
Flow, F10 Fluidizati on
Stor ag e 2,
Wsp, Wr eg Reg enerator
Press. at Fuel Gas
Atmospheri c Air Suction Bottom of Flow, F5
Pressur e, Patm Flow, FV6 Reg en., Pr gb
Cycl one Exhaust
Flow Flow, Fsg ,
Comb. Ai r Combusti on Ai r Qfg, Qe
Blower Sucn. Combusti on
Throug hput, F6 Air Flow, F7
Combusti on Comb. Ai r
Vent Flow, F V7 Blower Disch.
Figure 6: The Generalized Function level at the Component Level of Aggregation.
causal relationships need not be coupled with Generalized Function - Unit Level
physical relationships such as those described by the Lift Air
connections between nodes at the PFn level. The Storage Storage
GF-Component cell is quite complicated. The Combus tion
extensive connections between the nodes reflects the Feed
high degree of interaction between plant functions. Generalized Function - Subsystem Level
The influence of pressure propagations is a major Air
driver of this inter-dependency. Note that the
complexity increases around those nodes related to Heat and
the reactor and regenerator units. Whereas the Figure 7: The Generalized Function Level at the
functionality around the Feed Input and Heat Unit and Subsystem levels of Aggregation
Transfer Units is essentially sequential (which is
typical of AHs we have dealt with previously), the The representations employed in the AF level
functionality in the reactor and regenerator is are adapted from Multi-level Flow Modeling
circular. (MFM) . We should clearly note, however, that
The GF level at the Unit and Subsystem levels we do not adopt all of the MFM rules of syntax.
of aggregation are shown in Figure 7. Note that we Thus, we do not claim that these representations are
employ the same language to describe the functions examples of MFM.
at these levels. Thus, units and subsystems are MFM prescribes six types of functions; source,
discussed in terms of their functions as flows, sink, store, balance, transport, and barrier (see
reactions, and heat transfers. Note how much Figure 8). A source occurs when mass or energy
simpler these representations are compared to the crosses a system boundary into the system.
GF-Component level. Moving up a level of Similarly, a sink occurs when mass or energy
aggregation allows the operator to think about the crosses a system boundary away from the system. A
same system in fewer terms, exploiting hierarchy to store represents a point in the system at which mass
reduce memory demands. or energy can accumulate. A balance describes a
conservation of mass or energy without
The Level of Abstract Function
The Abstract Function (AF) level reveals Source Bal ance Store
information about mass and energy relationships in
the plant. Connections between nodes at this level
Si nk Transport Barri er
again reflect causality. Note that at this level of
abstraction the various commodities are no longer Figure 8: The functions of MFM employed in the
distinguishable, they are all represented as masses. Abstract Function Level of the AH.
Further, different types of heat transfer are treated as
Feed Pr eheat Uni t
Fuel Gas V5 Furnace
Fuel Feed Uni t
Wash Oil V1
Diesel Oi l V2 V3
To VRU V4 Cr acking
V13 Wet Gas Compr essor
Flar e Gas V12 Fracti onator Diseng ag er
Combusti on Ai r Blower
Catalyst Combusti on
Reg eneration Air Blower
Reg enerator Bed V9 Steam Turbine
Exhaust Dr ive
Lift Air Bl ower
Lift Pipe Blower Steam
Figure 9: The Abstract Function level for mass relationships at the Component level of aggregation (figure split
due to size and resolution limitations).
installation as a whole. This may appear to be a
the use of a store, usually in the form of an simplistic statement, but the rest of the AH
exchange. A transport function indicates that mass demonstrates that, in order to achieve this purpose,
or energy has been moved from one physical an extensive range of functions must be properly
location to another. A barrier is used to indicate a arrayed. We have experimented with including a
prevention of mass or energy transport. couple of other purposes for the FCCU. The most
The functions at the AF level most likely to be prevalent among these were safety-related.
confused are barrier and balance. A heat exchanger However, out conversations with process engineers
is an example of a common piece of process have convinced us that these concerns are either
equipment that serves to exemplify and distinguish coincident with production interests or not worthy
between these two functions. In a typical shell and of mention.
tube heat exchanger, energy is transferred from the
hot side flow to the cold side flow. As such, the 3. DISCUSSION
heat exchanger acts as a balance to energy because Prior to this exercise, the Abstraction Hierarchy
the energy is conserved without being stored. In had not been employed to model petrochemical
contrast, the two flows never come into contact with processes. The results of our efforts shown here
each other (a major advantage in nuclear systems). indicate that it is indeed feasible to extend the AH to
Thus, the heat exchanger acts as a barrier to mass this domain. While the transition has necessitated
because it prevents physical contact between the some modifications and extensions of the AH
commodities. concepts, it has not posed any insurmountable
The AF level at the Component level of obstacles or demanded any changes in philosophy.
aggregation for mass representations is shown in In the following paragraphs, we will discuss some of
Figure 9. The parallel representation for energy the peculiarities associated with employing the AH
relationships has not been included due to space in the petrochemical domain.
Figure 10 shows the separate mass and energy Dealing with advances in technology.
portions of the AF level at the Unit level of In reviewing the AH described here, two
aggregation. Corresponding representations for the process engineers noted that advances in FCCU
AF level at the Sub-system level of aggregation are technology are manifested at the Physical Function
not provided. level of Abstraction. In other words, the higher
level functions which comprise an FCCU are seldom
Mass modified by new technology. This observation has
strong implications for employing an AH throughout
the life cycle of a plant. Modifications of plant
equipment are to be expected, although their actual
form cannot be anticipated far in advance. If the
immediate effects of those changes are manifested
in a single level of abstraction, modifications could
be restricted to that level. Information systems and
Energy displays could be designed flexibly in areas where
changes are likely. Such an approach would
alleviate the need to overhaul the AH when slight
(but influential) process modifications are
This observation can also be extended to
creating AHs for other FCCUs. If differences
Figure 10: The Abstract Function Level at the Unit between multiple plants also lie primarily at the PFn
Level of Aggregation (mass and energy treated level of abstraction then it is likely that higher levels
separately). of abstraction will be relatively consistent between
plants. This suggests that once an AH has been
The Level of Functional Purpose created for a full scale FCCU, it can be adapted to
The overview of the AH space (Figure 2) has a other FCCUs with modifications to low levels of
pair of dollar signs in the Functional Purpose- abstraction only. If this extension holds then it has
System cell. The overall purpose of an FCCU is to strong implications for the flexible application of a
contribute to the financial viability of the particular AH modeling effort.
Abstraction Hierarchy is event-independent, it
New developments in AH methodology. accommodates unanticipated variability by
Extending the AH to the petrochemical domain describing the invariant relations  of the
has challenged our understanding and appreciation plant that constrain both operator and control
for the modeling technique itself. Two particular
principles have evolved from this application. First,
we have concluded that not every node in the AH
needs to have an associated quantitative variable.
This research was sponsored by the Honeywell
Previously we had assumed that each node could be
Technology Center (Peter Bullemer, Grant
quantified in some manner. Our experience with the
FCCU has convinced us that this is not necessary.
Qualitative labels can be employed as place holders
as long as the distinction is clearly drawn. For
 P. T. Bullemer, and I. Nimmo, 1994,
example, the cyclones in the reactor and regenerator
“Understanding and Supporting Abnormal Situation
clearly serve the function of separating commodities
Management in Industrial Process Control
(see Figure 6). However, there is no model variable
Environments: A New Approach to Training”, In
that characterizes this function. Despite our
Proceedings of the 1994 IEEE International
inability to attach a quantitative parameter to this
Conference on Systems, Man, and Cybernetics (pp.
function it is still important that it be spelled out in
391-396). Piscataway, NJ: IEEE.
the plant model. Operators can still use the
 K .J. Vicente, and J. Rasmussen, 1992,
qualitative concept (i.e., the object) in their
“Ecological Interface Design: Theoretical
reasoning processes, even though there is no
Foundations”, IEEE Transactions on Systems, Man,
quantitative value for it.
and Cybernetics, 22, pp. 589-606.
The second principle is that not all higher level
 J. Rasmussen, 1985, “The Role of Hierarchical
functions must be connected to lower level nodes.
Knowledge Representation in Decision Making and
In other words, not all ends have means which can
System Management”, IEEE Transactions on
be efficiently described at the adjacent lower level.
Systems, Man, and Cybernetics, SMC-15, pp. 234-
The introduction of pressures to the GF level
provides a good case in point (see earlier
 K. J. Vicente, and F. Tanabe, 1993, “Event-
discussion). In this case, there is frequently no node
Independent Assessment of Operator Information
at the PFn which acts and a means for the pressure.
Requirements: Providing Support for Unanticipated
In previous applications of the AH this problem was
Events”, In Proceedings of the Topical Meeting on
not encountered. We elected to allow nodes at
Nuclear Plant Instrumentation, Control and Man-
functional levels to remain unconnected because it
machine Interface Technologies (pp. 389-393). La
would be misleading to suggest to operators that a
Grange park, Illinois: American Nuclear Society.
node at a lower level could be employed to affect a
 A. M. Bisantz and K. J. Vicente, 1994, “Making
higher level function. In other words, we opted for
the Abstraction Hierarchy Concrete”, International
no information over misleading information.
Journal of Human-Computer Studies, 40, pp. 83-
 D. V. C. Reising and P. M. Sanderson, 1996,
We stated earlier than the greatest threat to
“Work Domain Analysis of a Pasteurization Plant:
plant productivity and safety is the reality of
Using Abstraction Hierarchies To Analyze Sensor
unanticipated variability. Because all events cannot
Needs”, In Proceedings of the Human Factors and
be foreseen by plant designers, event-based
Ergonomics Society 40th Annual Meeting (pp. 293-
responses to abnormal situations must eventually
297). Santa Monica: Human Factors Society.
fail to encompass the range of situations operators
 N. Dinadis and K. J. Vicente, in press,
will face. This statistical fact need not lead to the
“Designing Functional Information for Aircraft
conclusion that there is nothing that we, as
System Status Displays”, International Journal of
engineers, can do to support operators faced with
abnormal plant situations. The Abstraction
 N. Dinadis and K. J. Vicente, 1996, “Ecological
Hierarchy is a framework which allows a
Interface Design for a Power Plant Feedwater
description of opportunities for acting on the plant
Subsystem”, IEEE Transactions on Systems, Man,
which is independent of transient states. The model
and Cybernetics, 43, pp. 266-277.
formed by the AH is applicable in all situations,
 C. M. Burns and K. J. Vicente, 1995, “Physical
anticipated and unanticipated. Because the and Functional Displays in Process Supervision and
Control (CEL 95-11)”. Toronto: Cognitive
 J. Itoh, A. Sakuma, and K. Monta, 1995,
“Ecological Interface for Supervisory Control of
BWR Nuclear Power Plants”, Control Engineering
Practice, 3, pp. 231-239.
 R. C. McFarlane, R. C. Reineman, J. F. Bartee,
and C. Georgakis, 1993, “Dynamic Simulator for a
Model IV Fluid Catalytic Cracking Unit”,
Computers in Chemical Engineering, 17, pp. 275-
 N. P. Lieberman, 1991, “Troubleshooting
Process Operations”, Tulsa: Penwell.
 M. Lind, 1994, “Modeling Goals and Functions
of Complex Plants”, Applied Artificial Intelligence,
8, pp. 259-283.
 S. N. Kavuri and V. Venkatasubramanian,
1992, “Combining Pattern Classification and
Assumption-based Techniques for Process Fault
Diagnosis”, Computers in Chemical Engineering,
16, pp. 299-312.