Professional Documents
Culture Documents
Material Handling Equipment Selection: New Classifications of Equipments and Attributes
Material Handling Equipment Selection: New Classifications of Equipments and Attributes
www.elsevier.com/locate/eswa
Abstract
Conveyor equipment selection is a complex, and sometimes, tedious task since there are literally hundreds of equipment types
and manufacturers to choose from. The expert system approach to conveyor selection provides advantages of unbiased decision
making, greater availability, faster response, and reduced cost as compared to human experts. This paper discusses the development
of a prototype expert system for industrial conveyor selection. The system, which was developed on Level V Object, provides the user
with a list of conveyor solutions for their material handling needs along with a list of suppliers for the suggested conveyor devices.
Conveyor types are selected on the basis of a suitability score, which is a measure of the fulfillment of the material handling requirements
by the characteristics of the conveyor. The computation of the score is performed through the Weighted Evaluation Method, and the
Expected Value Criterion for decision making under risk. The prototype system was successfully validated through two industrial
case studies.
q 2003 Elsevier Ltd. All rights reserved.
Keywords: Weighted evaluation method; Decision making; Expected value criterion; Conveyor selection
Therefore, engineers end up usually selecting pieces of The system is designed to recommend a truck type
equipment that they are most familiar with, and thus, these alternative that meets the user specified requirements,
may not represent the most cost-effective solution. while optimizing the given evaluation criteria (e.g. mini-
mum cost, minimum downtime, etc.). Moreover, the
program attempts truck selection through few inquiries
2. Survey of related research from the user. This is accomplished through pattern-directed
inference (Malmborg et al., 1987).
Heuristic-based systems are becoming increasingly
popular in material handling equipment selection. Their 2.3. MATHES—material handling equipment selection
importance stems from the fact that they consider both expert system
qualitative and quantitative factors while selecting and
evaluating pieces of equipment. It was developed by Fisher and Farber, 1988. It
Intelligent computer systems have been developed as addresses the problem of selecting a material-handling
expert systems (ES) and decision support systems (DSS). device from a total of 16 possible choices. It is a rule-
Expert systems have been widely applied to material based system that uses 172 rules to arrive at the final
handling equipment selection problems. They function by decision. MATHES uses four main parameters: path,
inquiring the user for desired attribute values, which are volume of flow, sizes of unit, and distance between
used to select the most cost effective equipment type from a departments. Moreover, minor parameters such as palleti-
pre-established set of alternatives. The knowledge base zing, accumulation, and fragility are also considered for
usually consists of rules drawn from textbooks, equipment some scenarios. The system generates multiple solutions
literature, or from consultation with experts. An expert with different certainty factors. Its evaluation capability is
system can also be programmed to select among available limited to a small number of attributes. It also suffers from
commercial material handling models. Some of the most the drawback of not considering common types of
successful applications of expert systems to the selection of conveyors, and trucks for interdepartmental moves (Fisher
material handling devices are discussed below. & Farber, 1988).
2.4. Mathes II
2.1. SEMH-Selection of equipment for material handling
MATHES II was developed in 1990 by Honng. It follows
SEMH was developed in 1984 by Hatem Nasr. It starts the
the same procedure as MATHES, but it possesses a larger
equipment selection process by identifying the character-
working scope and greater consultation functions. Its main
istics of the material to be handled, i.e. type, weight, size, etc. goal is to select suitable material handling equipment
Based on these characteristics, the system searches its alternatives for material handling activities. MATHES II is
knowledge base to recommend on the degree of mechaniza- capable of performing economic analysis for particular
tion, and the type of material handling equipment to be groups of material handling equipment. It can also provide
employed. SEMH is implemented as a rule- based system. specific information on equipment suppliers (Honng, 1990).
The knowledge base consists of 39 rules divided between two
knowledge bases: the level of mechanization and the 2.5. EXCITE—expert consultant for in-plant
equipment-selection knowledge bases. The system draws transportation equipment
its knowledge mainly from textbooks, handbooks, and
similar sources. SEMH is also capable of learning new It was developed by Swaminathan S.R et al. in 1990 as a
facts inferred from previous consultations (Nasr, 1984). part of his MS thesis at the University of Alabama. It
addresses 35 equipment types, and 28 material, move, and
2.2. A prototype expert system for industrial method attributes. The knowledge base is written in OPS83,
truck type selection and the system uses 340 rules for decision making. EXCITE
operates by asking the user for relevant material, move, and
This expert system, developed by Malmborg, Agee, method attributes. The inference engine uses forward
Simons, and Chowdhury, considers 17 equipment attributes chaining to arrive to a decision. The program lists the
and 47 devices for industrial truck selection. Developed in most viable equipment types, and preference factors based
PROLOG, it uses a combination of forward and backward on flexibility, maintainability, and cost attractiveness
chaining for inferencing. In the forward reasoning stage, the criteria. It focuses on selection of devices for individual
system attempts to deduce a truck type recommendation moves only, and the interface requirements are explicitly
based on known relationships, and specified requirements. considered to ensure that the recommended device is
In the backward reasoning stage, the system tries to appropriate for the given requirement. A certainty factor
establish which truck type is most suitable for the task at module called Mathand allows the user to specify measures
hand by inquiring the user on equipment attributes. of belief for attributes, and incorporate certainty factors in
D.J. Fonseca et al. / Expert Systems with Applications 26 (2004) 615–623 617
the rules. It also offers limited explanation facilities, which material handling operations take place. An object oriented
permits the user to follow the reasoning process. EXCITE paradigm is used for knowledge representation in
does not consider pieces of equipments that are used for MAHSES, which supports modularity, multiple reasoning,
storage activities, such as automated storage and retrieval inheritance, message sending communication, and hierarch-
systems. Commercial models of equipment are frequently ical structures. MAHSES carries out its inferencing on the
not considered by its knowledge base (Swaminathan et al., basis of user supplied attribute values and the heuristic rules
1999). stored in the knowledge base. MAHSES is currently limited
to the selection of micro load AS/RS design. However, it is
2.6. EMHES—expert system for material handling feasible to expand MAHSES capabilities to other material
equipment selection handling and storage alternative designs (Kim & Eom,
1997).
EMHES was developed in 1992 by Attia et al. EMHES is
a prototype expert system which incorporates knowledge
based systems technology and database management 3. Expert system development
techniques to assist in the selection of cost effective material
handling equipment alternatives. EMHES is divided into 3.1. Development of the knowledge base
two main components: the knowledge base and the
reasoning system. The knowledge base consists of a A considerable amount of information and data was
database which includes a working memory and a large collected on the different types of conveyors mentioned
set of decision rules on material handling equipment listed in Tables 1 and 2 through textbooks, handbooks,
selection. The database includes a material handling journal articles, interviews with material handling experts,
equipment catalog, and a vendor’s equipment availability and promotional literature supplied by the vendors. Vendor
report. The knowledge representation is in the form of IF- supplied literature was used as the primary source because it
THEN statements. The reasoning system is based on five led to the construction of a system most in tune with the
independent modules, i.e. the user interface module, the industrial world. The accuracy of that information was
selection module, the optimization module, the modification validated with the use of journal articles and material
module, and the simulation module (Attia, 1992).
Table 1
2.7. ICMESE—intelligent consultant system for material Unit handling conveyors
handling equipment selection Chain conveyors
Flat top chain Chain-on-edge
ICMESE was developed by Park, 1996. This system Slat chain Drag chain
overcomes the limitations of its predecessors, i.e. consider- Roller conveyors
ation of a limited number of equipment types and attributes, Chain driven Line shaft
lack of consideration for storage and warehousing equip- Belt driven
ment, and exclusion of commercial models. It addresses 50 Belt conveyors
equipment types, and reaches a final conclusion on the basis Roller bed belt Slider bed belt
of 29 attributes which are grouped under (1) move Trash belt
attributes, (2) material characteristics, (3) operation require- Tow conveyors
ments, and (4) area constraints. First, ICMESE collects Overhead Underfloor
information on the attributes by inquiring from the user. It Flush
then matches its rules to the specific attribute values to Gravity conveyors
select an equipment type. ICMESE retrieves all the Skatewheel Chute
commercial models for the selected equipment type, and Gravity roller
lists them on the screen. Finally, the program evaluates the Monorail systems
performance of the selected equipment model by simulating Overhead power and free Overhead monorail
its operations using SLAMI, and Fortran 77 (Park, 1996). Inverted power and free Automated electrified monorail system
Vertical conveyors
2.8. MAHSES—material handling selection expert system Reciprocating Opposed shelf
Continuous Arm
MAHSES was developed by Kim & Eom, 1997. It is Sortation conveyors
a hierarchical object oriented expert system composed of Right angle pusher Line shaft diverter
two main modules. Module 1 addresses the selection of Panograph sorter Roller diverter
material handling alternatives as well as storage systems for Motor driven sorter Sliding shoe sorter
Pneumatic puller Tilt tray sorter
electronic assembly. Module 2 addresses the selection of the
Powered wheel diverter Belt carriage sorter
design, assembly flow, and departmental layout, where
618 D.J. Fonseca et al. / Expert Systems with Applications 26 (2004) 615–623
Table 2 Table 4
Bulk handling conveyors Bulk conveyor attributes
Belt conveyors 1. Loading: it refers to the feeding of the bulk materials on to the conveyor
Rubber belt grade 1 Mobile stripper Loading can be either controlled or uncontrolled
Rubber belt grade 2 Stationary stripper 2. Discharge: it refers to the material transfer that can take place at the head
Rubber belt grade 3 end, head or foot end, at one place in between (single intermediate), at many
places with equal intervals in between (multiple intermediate), and at
Pneumatic conveyors
various places with unequal intervals between the head and foot end
Pressure dense phase system Portable vacuum pressure system (variable intermediate)
Fan type system Dilute phase system 3. Angle of inclination: it is the slope at which the bulk material needs to be
Hooper loader system Vacuum dense phase system transported
Screw conveyors 4. Complexity: it refers to the conveyor profile that can be straight-line or
Troughed and shafted Tubular and shafted compound. Straight-line conveyors have a constant angle of incline, and no
Troughed and shaftless Tubular and shaftless discontinuity in between, while a compound conveyor is made of two or
more separate conveyors joined at their ends
Bucket conveyors 5. Horizontal carry: it refers to the maximum horizontal distance that a
Gravity discharge Pivoted bucket conveyor can move the material
Chain conveyors 6. Vertical lift: it refers to the maximum vertical distance that can be
Deep pan apron Flight traversed by the conveyor device
Hinged pan apron Sliding chain 7. Material size: it refers to the size of the transported bulk materials
Shallow pan apron Slat 8. Angle of repose: it is the angle made by the uppermost layer of the bulk
Enmasse material with the horizontal
Bucket elevators 9. Temperature: it refers to the hotness of the transported material
10. Corrosiveness: it refers to the eroding ability of the transported material
Belt type centrifugal discharge Super capacity bucket style F
Chain type centrifugal discharge Super capacity bucket style G 11. Friability: it refers to the brittleness of the materials
Chain type continuous discharge Super capacity bucket style H/HL
Belt type continuous discharge Positive discharge 3.2. Computation of material handling device scores
Internal discharge
Vibrating conveyors
The score of the various conveyor types is a measure
of the fulfillment of the material handling requirements by
handling books. The knowledge base development was the characteristics of the conveyor. The computation of the
accomplished through the following three steps: score utilizes the concepts of the Weighted Evaluation
Method and the Expected Value Criterion for decision
† Identification of a list of conveyor types, one or many of making under risk. The conveyor type’s score is computed
which the system can recommend to the user for his/her in the following fashion:
material handling needs
† Identification of conveyor attributes (Tables 3 and 4) to If Attribute 1 has Value 1 OR Value 2
be matched with the material handling requirements Then Score ¼ Score þ 100 p (wt. of Attribute 1/Swt. of
given by the user all attributes) p Prob. 1
† Development of the heuristic rules for conveyor selection. Else Score ¼ Score þ 100 p (wt. of Attribute 1/Swt.
of all attributes) p Prob. 2
Table 3 ..
Unit conveyor attributes .
1. Plane of movement: it refers to the movements of the conveyors to the
horizontal, inclined or vertical plane If Attribute N has Value N
2. Power source: it refers to the gravity or electric power that makes the Then Score ¼ Score þ 100 p (wt. of Attribute N/Swt. of
convey or move the materials all attributes) p Prob. X
3. Speed: it refers to the required speed of the movements of the materials
Else Score ¼ Score þ 100 p (wt. of Attribute N/Swt. of
4. Temperature: it refers to the hotness of the transported materials
5. Loading capacity: it refers to the weight carrying capacity for a conveyor all attributes) p Prob. Y
type Score of the Conveyor 1 ¼ Score
6. Horizontal carry: it refers to the horizontal distance that needs to be End
covered by the conveyor
7. Product shape: it refers to the shape of the transported material that can
The programming logic compares each user supplied
be of regular or irregular type
8. Accumulation: it is the ability of the conveyor type to pile up transported attribute value against the conveyor characteristics. On
materials along the path finding a match, the score of the conveyor type gets
9. Flexibility: it is defined as the ability of the conveyor to adapt to changes. incremented by a quantity equal to the weight given to the
High flexibility conveyors can be moved form one place to the other, while corresponding attribute over the sum of the weights for all
low flexibility conveyors can be moved only after incurring very high
the other attributes, multiplied by the probability that the
expenditure
attribute in question is in conformance with the conveyor
D.J. Fonseca et al. / Expert Systems with Applications 26 (2004) 615–623 619
characteristics. This probability value is computed as determine the suitability of the conveyor group. The
follows: knowledge base is searched using a search order defined by
the developer. Conveyor type suitability or unsuitability is
Probability ¼ conf þ ð1 2 confÞ p ½ðx 2 1Þ=ða 2 1Þ ð1Þ
determined through the rules embedded in the When-Needed
where conf refers to the confidence that the user has on the method associated with that particular conveyor group.
supplied information. Since other attribute values could After displaying a list of suitable conveyor groups, the
also give a positive match, the component ð1 2 confÞ p expert system inquires the user on the selected conveyor
½ðx 2 1Þ=ða 2 1Þ is added to attain the final probability groups. The responses to these queries, along with the
value, where x refers to the number of favorable events, and attribute values, are used to identify suitable conveyor
a is the total number of different values of the attribute. solutions with corresponding conveyor scores.
When the user supplied attribute does not concur with the The identification of suitable conveyor subtypes is
conveyor characteristics, the Else statement of the rule also performed using the same scheme of Pursue and When-
increments the score, but by a smaller quantity. To illustrate, Needed methods. The conveyor score is calculated
let us suppose that the user has stated with a 90% confidence through a combination of the Weighted Evaluation Tech-
that the attribute value is Value 1, while Value 2 is the one nique and the Expected Value Criterion for decision making
that matches the conveyor characteristic. Therefore, there is under risk. The conveyor score is a reflection of the
still a 10% chance that the user was incorrect, and therefore, compatibility of the conveyor to the material handling
the score should be incremented accordingly. The prob- requirements. Only conveyor solutions with a score of 80 or
ability of such an event is computed as: above are displayed to the user.
Probability ¼ ð1 2 confÞ p ½x=ða 2 1Þ
where conf is the confidence the user has on the provided 4. Use of the system
information, x is the number of favorable values, and a is
the total number of different values of the attribute. 4.1. System’s input
A conveyor can attain a maximum score of 100 if it
satisfies all the material handling requirements of the user As it has been mentioned before, one of the main
who has supplied all attribute values with complete objectives established in the system design phase of this
certainty (confidence value ¼ 100). study was to burden the user with the minimum number of
queries on material handling features as possible. This was
3.3. Development of the inference procedures accomplished using a forward chaining inferencing mech-
anism, which moves in a step by step manner to identify the
The prototype expert system can identify suitable appropriate conveyor category, conveyor groups, and then,
conveyor solutions from a knowledge base of 76 conveyor the most suitable conveyor type.
types that have been demarcated into three categories of The first input screen presented to the user is the Conveyor
unit, bulk, and sortation conveyors. Category Selection screen which asks the user to select one of
For unit and bulk handling conveyors, the system three conveyor categories (Fig. 1). The expert system then
identifies suitable conveyor groups as an intermediate presents the user with a list of attributes needed to determine
conclusion, and then lists numerous conveyor subtypes in suitable conveyor solutions (Fig. 2). The user is asked to
the final conclusion screen. However, suitable sortation input the weights or importance values for these attributes.
conveyors are presented in a single stage itself. The expert system compares material handling require-
The inference procedures identify the suitable conveyor ments supplied by the user against the conveyor character-
groups through a scheme involving the use of Pursue istics for the purpose of selecting the best conveyor system.
commands and When-Needed methods. The user is asked The user is asked to input the material handling requirements
to express his material handling requirements in terms of in two stages for unit and bulk handling conveyors. However,
attribute values. The user also provides weights for the identification of sortation conveyors is accomplished in a
various attributes as well as confidence values for the single stage only. In Stage 1 (Fig. 3), the handling
information supplied. After the user has supplied the required requirements are expressed as attributes. The user is asked
information, the system prompts the user to click on the to select for each attribute one of the many ranges of attribute
‘Continue’ pushbutton to begin the device assessment. This values that best describes the work at hand. This information
pushbutton is attached to a When-Changed method which is then used to identify suitable conveyor groups. In Stage 2
then fires a series of Pursue commands. Each Pursue (Fig. 4), specific questions pertaining to individual conveyor
command is followed by an attribute representing a particular groups are asked to find out the most suitable conveyor
conveyor group. The value of this attribute indicates the subtypes. In both Stages 1 and 2, the user is also asked to
desirability or undesirability of the conveyor type for the task provide confidence values for the information submitted.
at hand. Hence, when the Pursue commands are fired, The system takes the confidence values into consideration
the inference engine searches the knowledge base to while computing the conveyor’s suitability score.
620 D.J. Fonseca et al. / Expert Systems with Applications 26 (2004) 615–623
The input screens are user friendly, wherein the definitions presented with a list of candidate conveyor groups as well as
of technical terms are provided to the user. All suitable a series of queries on handling requirements. The system
conveyor solutions and attributes appear as hyperlinks in blue uses the query responses along with the attribute values to
color, and their definitions can be accessed by clicking on calculate a score for all the suitable conveyor subtypes, and
them. Moreover, all screens are equipped with help menus. then, sorts out those conveyor subtypes that attained a score
of 80 or above. The list of conveyor subtypes that represent
the best match to the handling requirements of the user is
4.2. System’s processing
finally displayed to the user.
The expert system requires the weighted importance
of attributes, confidence values, and material handling 4.3. System’s output
requirements from the user. It then matches the handling
requirements against the characteristics of conveyor groups The main goal of the system’s output interface is to
to identify suitable conveyor groups for the task. The user is provide all the necessary information in a clear and easy to
understand manner. The output to the user includes the Conclusion screen is not presented for sortation conveyors
preliminary and the final conclusion. The Intermediate as they are identified in a single stage itself. After specific
Conclusion screen lists the conveyor groups suitable for the data particular to the conveyor groups are entered by the
task at hand, while the final Conclusion screen lists the user, the final Conclusion Screen reporting the most suitable
various conveyor types along with their suitability indexes. conveyor subtypes and their corresponding suitability
Both screens display the names of the conveyor groups and scores is displayed (Fig. 6).
conveyor types as hyperlinks. Upon clicking on them, the
user can access relevant information on the devices including
a list of manufacturers and suppliers for each conveyor type. 5. Validation of the system
The first set of outputs to the user is the Intermediate
Conclusion screen which lists the suitability/unsuitability The validation of the prototype expert system was
of each conveyor group (Fig. 5). The Intermediate accomplished by the use of two ‘real life’ case studies.
The case studies together tested the system’s unit and bulk painting, assembling, and packaging are the main operations
handling conveyor expertise. The case studies were involved in making these lawn and garden products.
evaluated by the expert system, and the results along with The component parts need to be spray painted before being
the reasoning process were compared against those obtained assembled into lawn products in the Assembly department.
from selected human experts. Encouraging results were Spray painting involves surface preparation through the
obtained from both case studies. washing of the component parts with a special washer
solution. The components are then dried with the use of a
5.1. Case study 1: overhead monorail conveyor heater and a blower that blows hot air over them. The washed
at Lambert Inc., Ansonia, Ohio and dried parts can then be spray painted. Once spray painted,
the parts need to go back to the heater to dry the paint. The
Lambert Corporation is a manufacturer of lawn and garden company required a conveyor to transport the parts through
products, such as lawnsweepers, dethatchers, plug aerators, the washer (an enclosure that sprays the washer solution on
seed cultivators, and ramps. Welding, stamping, spray parts), two heaters, and the paint booth. Also, the required