- Email: [email protected]
On a microcomputer integrated system for structural engineering practices
ON A MICROCOMPUTER INTEGRATED SYSTEM FOR STRUCTURAL ENGINEERING PRACTICES WORS.+~C K.~>o~~-Nu~uLcH.~I+ Department of Civil Engineering. University of Tokyo. Tokyo I I?. Japan
( RtTeiwd -I Fehrunp 1985) .ibstract-This paper describes a comprehensive use of microcomputer for professional practices of structural engineers. Transition from the conventional method to a computer-based procedure calls for a complete rethinking of an accurate. unambiguous definition of all the elementary activities, their logical orders as well as inter-relationships. An integrated database software system IS then proposed for analysis and design of structures. on microcomputer. With a common data bank. all previously processed data can serve directly as a basis for subsequent activities. such as computer-aided drafting. computer-aided construction planning, etc. Several program strategies to attain a maximum efficiency within existing microcomputer constraints are discussed.
design into one process. but to streamline their interaction through the use of a standardized data Analysis and design of structures constitute the two structure. Their traditional distinction as well as major activities in structural engineering practices. their logical orders remains unchanged. A data bank Generally, structural analysis and design are two serves to pool all data units at one place, accessible mutually coupled processes. While design variables and amendable at various phases of analysis and affect the analysis, results of the analysis dictate design operations. the choice of these variables in the design. TradiThe idea of an integrated FEM/CAD system is tionally, analysis and design are handled as though not new. Over a decade ago FEM/CAD was the being independent. Thus theoretically, a trial and exclusive province of large aerospace companies. error process must be conducted to eventually meet The long occupation of a time-sharing system durthe requirements of both at the same time. By hand ing a CAD session may be too costly for smaller calculations, the attempt to adhere to this consistengineering firms. Now this situation has changed ency may require such a big effort. It may not be drastically with microcomputers. Despite the low worthwhile if member designs are merely based on microprocessing speed, the insignificant operating the result of a simplified analysis. The provision of cost as well as the attractive input/output capabila factor of safety is assumed to help offset errors ities often make FEMiCAD on a microcomputer caused by stress misrepresentations as a result of very effective. In practice. the low processing analysis over simplifications. This often leads to a speed does not necessarily pose a problem as design in which the margin of safety fluctuates human responses during a CAD take a much longer greatly from place to place. Such a design is far from time, whereas lengthy FEM operations can always optimalit). be unattended overnight. After the advent of computers, a rigorous analToday, professionalism in software writing is ysis can be automated with great generality using more critical than it used to be on the mainframe. the finite element method (FEM). This is possible Optimizations of program and data units in relation because of the objectivity of structural analysis, for to a microcomputer environment are crucial to alwhich an exact algorithm can always be established leviate excessive demands for microcomputer time without ambiguity. and human attendance. The present paper will outFor design, computer automation is not feasible line useful programming strategies for the develdue to (a) the lack of a unique algorithm for design opment of an integrated FE&l/CAD program. synthesis. and (b) the lack of “engineering sense” in computer. Lacking intelligence, however. the computer has great durability. precision, and a perSTRL‘CTWAL \lODEL fect memory. Computer aided design (CAD) aims Rigorousness of structural modeling depends at combining the roles of human designer and comlargely on the computing tool available to the strucputer, using the strengths of both to complement tural engineer. With manual calculations, he can and supplement each other. either simplify the structure into a model for which The objective of an integrated software system a solution is available, or retain its rigorous form is not an attempt to couple structural analysis and but strive to obtain an approximate solution. In both cases, correctness of the result is often ques+ Associate Professor, on leave from Asian Institute tionable. With computer. it is possible to formulate of Technology, P.O. Box 2754. Bangkok, Thailand. INTRODUCTIOY4
33
a more
representative
reflect
behaviors
structural
model.
of the real structure
which
more
will
are defined
closely
and comprehensively. In the past 25 years, been
so phenomenal
considered
that
today.
as an industrial
analysis.
FEM
problems
2. The design
stresses among
in the
by ,Ci( ). where
engineers
same group.
and
on office
mi3. The quality
Consider
a structural
structural nodes
elements
(joints).
standard
element
Several design
the entire section,
interaction
I. Structural tural
analysis
constitute activities
signer’s
definition
the
will be prefor all these
data units.
it is pos-
process.
is an attempt
subjected
to a set of system This
process
parts
parameters
true.
code,
process
applied
as a function
loads f,
“material
“primary
properties”
A(
joint
) represents
of model design
(I,,,. Primary
design
design
module
g,
(zp. and variables
(
This
DESIGN
sizes,
numbers. They
these
REVISION 1
ENGINEERING
OK.
DRAWINGS
I.
Conventional
analysis
from may
as need
together
as
“subjective”
etc. They
a
varb
to be re-
It is worthwhile. into a replaceable
library”.
is not acceptable process
could
a revision
of CI,
be represented
of detailings
bq
“tn”.
determines
variables”
and placements
depend
=
design mainly
the re-
CI,,. such as
of reinforcing
on ~1,. The general by D(
can be denoted
bars. form
) where
D(ri,“‘, S”‘. I,,. I, 1 Designer). engineering
are prepared
drawings
based
various A suitable
tivity
is also suggested.
to secure
DETAILING
)
activities
units.
(5)
and related
on the final
Note
in practice
I
as there
size and reinforcement
procedure
the
doc-
results
of
are demands
data
mode serves
factor.
that
design
problem
of a member
with
other
for continuation along
of member
design
is also desirable
ma)
members.
continuous
cases, the use of similar
for a set of members
associated
mode for each ac-
The interactive
a “Designer”
& DOCUMENTS
and deign
and
computerization
not be an uncoupled
In many Fig.
mathwhich
g of the system. up. N, . as well as the geometry Figure 3 depicts the inter-relationships among
ANALYSIS
FINAL
The
as well
codes
of
“objet-
can be incorporated
time to time.
“secondary
uments
NO
types
The
calculations,
program.
them
5. The process maining
6. Finally
I STRESS
They
from
to bring
u,“’ DESIGN
other
an anal-
MODELING
MEMBER
(3)
I,, can be various
regulations,
such as “a
1 PRELIMINARY
as tuo
I,, and I,.
strength
environment
of this process
1
equation
I,. Desiytrer).
) in a way that. for a design group
etc.
STRUCTURAL
This pro-
displace-
geometry
variables”
with
efficiency.
I, can be specifications
city
vised or updated
U(
ysis
for
4. If the design
stresses;
design
requirements.
as well
of information
therefore,
can be symbol-
uJ and s’ are, respectively,
be evaluated
by an inequality
of information
formulas
is necessary.
in which
practical
of the computer
type
by
ments and element
will
information,
are universally
with
of the struc-
variables.
type
ematical
design
to determine
and displacements)
(2)
adequacy.
judgements
supplementary tive”
model
represented
analysis
describing
and design ically
a for
I.
the entire
analysis
(stresses
and K(
and design
associated
to streamline
responses
Q(
design
model
a new clear-cut
and their
then
in Fig.
be denoted
) represents quality characteristics of the ) represents corresponding minimum requirements. This K( ) could be subjected to de-
where
and
of structural
a general
between
With
activities
process
shall
up. CI,,. 11’) 2 ti(I,,.
Q(S)“,
or be
typical
process
etc.
properties
together
The
are presented
In this sented.
thus
group.
other
can have
of material
and
design
and design
at
column,
171
.i*’ of all elements
as
to any of the
such as beam.
values
variables,
completing
interconnected
of the same type
similar
member
sible
types.
elements
assigned
of
may belong
group
stresses
of a design
cess can be represented
as an assemblage
(members)
An element
the
= .Cf(s”).
to its strength
plus several
ISTER.4CTIOS
model
S”’ of a member
This
S”’
respect AND DESICS
u hich intluence
in
FEM
crocomputers.
0ALYSIS
quantities
can be determined
structural
to implement
practical
has
is alread!
for
to structural
age is the ability
large-scale
of FE&l
standard
A new challenge
the modern solve
the evolution
as those
analysis.
members. parameters
from
the view-
Microcomputer rMember
(deslgn)
7-
g : geometric
data
s:
f:loads
-
OF MEMBER
sD
i :
USER
Fig. 2. Interactions
GROUP
~
design
~.shsme”t -
NE<_E>SARY
INTERACTION
among activities
stress
I i
i
in analysis and design
processes.
point of a construction’s efficiency. The introduction of member design groups, therefore, allows the designer to specify a similar design for a set of members.
PROGRAM
practices
35
Derelopmenr philosophy
___d,lp,
/ANALYSIS
-
/
-prelm.
qz.$sD,
engineering
ever, with the current trend of hard disk rechnology, this problem will disappear soon.
elastic oroperties primary design variables secondary design variables design stresses given
new
system for structural
grouo IO.
PROPERTIES h:
integrated
STRATEGY
Limitarions of microcomputer
It is imminent that the microcomputer will become an integral part of the structural engineering profession. There is a need, therefore, to understand what a microcomputer can or cannot do. Generally, characteristic constraints of microcomputers are the following: 1. Limitation of incore memory. Microprocessor has a maximum limit on the size of addressible RAM space. After a portion is reserved for various system utilities, the remainder is then available for the user’s program and data. 2. Low speed ofj7oating-point arithmetic. Aside from the low microprocessor speed itself, a series of machine code instructions must be performed internally for each arithmatic operation in order to maintain a degree of precision. However, in some machines, a coprocessor for floating-point arithmetic can be added specifically to speed up these operations. 3. Low speed of disk data transfers. When data units must be shuttled between core and disk frequently, the total time for disk data transfers can exceed even the microprocessing time itself. How-
The development philosophy for engineering programs on a microcomputer can be totally different from that used for a mainframe. due to their great contrasts in capacity. operation, cost, and speed. For a mainframe, the expensive CPU requires a program to be highly efficient in terms of computer time. The cost of solving large FElM problems on a mainframe is governed mainly by the computer cost, and partly by the cost of engineers for data preparation and output digestion. as depicted in Fig. 3. For a microcomputer, the first and main concern has been whether practical, engineering problems can fit into microcomputer memory. If not. what are the necessary steps for achieving the same goal, even at the expense of excessive computer time. Since the cost of operating a microcomputer is negligible, the overall cost is likely to be dominated by the engineer’s costs. With this consideration, the adopted strategy is as follows: 1. To further save engineer’s time on data preparations and output digestions, the microcomputer should assist the engineer with its powerful interactive and graphic capabilities. During the input process, an error handling algorithm is necessary to prevent accidental fatal errors, which may lead to the total loss of all previously input data. 2. Time consuming, objective processes should be automated without a requirement for attendance. Therefore, a lengthyjob can be run overnight. 3. For subjective processes, the necessary machine/user interactions can proceed smoothly if the user is constantly informed of his possible actions or options, as well as useful information. Advanced
a) costs
for
note = : automated {G
: interactive
mainframe
1
_iTIJE
computmg
stage.
unattended.
stage.
attendance
1
required
Fig. 3. Cost components of a typical engineeringjob: frame vs microcomputer.
main-
Table I. FE%1 CAD criteria. CRITERIA FOR GOOD FEWCAD PACKAGE I.
constraints.
and stratcgxs
I+ICROCOWUTER CONSTRAINTS
GENEALITY (COMPLEXITY 0F CODES)
LI!lITATION OF tNCORE xE?lORY
: (a) PROGRA'i MODULARIZATION
RESOLUTION
(b) MODULE SEGMENTATION AND OVERLAY 2.
HIGH CAPACIn (DATA CAN BE VOLUMINOUS)
DISK DATA TRANSFER IS VERY TIME CONSUMING
~SOLII'KION : (a) OPTIMAZATION OF DATA BLOCKS (b) DATA STORAGE IN BINARY FORM RIGOROUSNESS
3.
FLOATING-POINT NUMBER CRUNCHING IS RELATIVELY SLOW
(HEAVY FLOATING-POINT NUMBER CALCULATION)
RESOLUTION : (a) MANUAL REDUCTION INTO EXPLICIT FORM (b) USE COMPILATION (c) INSTALL ARITHMATIC CO-PROCESSOR 4.
PRACTICALITY (REQUIRE co~ons~~s~s)
RESOLUTION : (a) ALLOW HUMAN INTERACTIONS WHEN NECESSARY (b) INSTALL 'KNOWLEDGE' & 'EXPERIENCES STATISTICS' (TOWARD AN 'EXPERT' SYSTEM)
L
interactive
facilities
“mouse”)
pen. ciency
LACK HUMAN INTELLIGENCE AND EXPERIENCES
(such
can
as graphic
be used
of the interactive
tablet,
to enhance
process.
light
order
efti-
systematically.
the
to generate,
retrieve,
or refresh
data
units
and thus increase
the productivity. MISS-A
XIICROCOMPUTER STRUCTURAL
I summarizes
Table practicality
on microcomputers, them.
More
paperill
and various
details
necessary
programs,
are
their
for
obstacles
steps to overcome
available
in
a previous
of the author.
The
first
larization. fined
some criteria
of FEMICAD
resolution
as a small
a well-defined tually
calls
In general, logical
function.
independent
for
program unit
program
modu-
can be de-
of data processor
Individual
and only
“administrative
a program module
modules
communicate unit”
with
are muwith
and
the
the
“data
bank.” Three
overlaying
schemes
gram units are depicted able for
linking
continuity taining
a core
units,
one at a time, of
Scheme An
must
finite
calls
be allow
a number
of research
pro-
units
with
no
while
main-
Scheme
(c) is useful
in a series
of satellite
for data c.xtributions. subroutines
The liemploys t
multitask
efficient data units
the past few years,
works[2-91 were conducted at the Asian Institute of Technology toward achieving a complete microcomputer package for structural engineering practices. The ultimate goal is an ambitious database FEMKAD program MISS which will handle a general multistory R.C. building project in the following aspects: (a) automated analysis by FEM; (b) computer aided design subjected to the AC1 (or a user supplied) code; (c) computer aided production of engineering drawings and related documents; (d) computer aided preparation of the bill of quantities
(b) is common
of a large module
element
integrated
numerical will
program
SYSTEM FOR
ENGINEERS
(a) is suit-
(c).
tremely sible,
Scheme
data continuity.
when brary
data.
segments
some
connecting
a set of independent
of in-core
for chaining
for
in Fig. 4. Scheme
During
INTEGRATED
program
data management. must
elaboration.
Whenever
ex-
t)APPEND
b)CHAIN
pos-
be cast in a form suitable for A standard
used consistently various
requires
1, LINK
modules
for
data
all data
the ease of
THREE
structure units.
This
access
in
OVERLAYING
SCHEMES
Fig. 4. Three microcomputer overlaying gram segmentation.
schemes for pro-
\Iicrosomputer
integrated
system for structural
engineering
practice5
37
this valuable form of data bank may replace or supplement the engineering blue prints as the permanent record of a building project. as required by the owner or a city building authority.
The paper described the development of a microcomputer integrated software system for application to professional activities in the structural engineering field. The objectives of this paper are twofold: ( I) to stimulate enthusiasm among structural engineers about the great potential of microcomputers for supplementing their professional activities; and (2) to show that the emergence of inexpensive microcomputers will not drastically change the principal roles and responsibilities of structural engineers (at least not until the coming age of artificial intelligence). REFERESCES
Fig. 5. Typical
modules of microcomputer tem for building projects.
integrated sys-
and breakdown costs; and (e) computer aided and monitored construction scheduling and resources planning. A key factor toward achieving this goal is the introduction of a centralized data bank, which serves as a common denominator to all processing modules. as shown in Fig. 5. Basically, data units in the data bank are continuously processed until they meet both the requirements of analysis and design. In the subsequent activities of (c)-(e) then, the data bank simply serves as an information center. With a standard data structure. an unlimited number of future modules can be added and integrated without disturbing the existing ones. Future expansions may include a module to update relevant data units in order to keep account with revisions made during the construction phrase. This will enable as-built engineering drawings to be prepared automatically and accurately. This author anticipates that. as a future practice.
and S. Udomlurgchai. DevelI. W. Kanok-Nukulchai opment Strategy of a Microcomputer Finite Element Analysis Program-MICROFEAP, in Engineering Sofrwarefur Microcompurers. (Edited by B. A. Schreffer et al.). Pineridge Press, Swansea (1984). S. Udomlurgchai. Development of General Purpose Finite Element Program on ;Microcomputer (APPLE II). M.S. Engng thesis. Asian Institute of Technology. Bangkok (1983). S. A. Abueva, Ultimate Strength Design of Reinforced Concrete Members on Microcomputer, Special Study No. SSPR-ST-83-I. Asian Institute of Technology. Bangkok (1983). 4. W. Wilaingarm, hlicrocomputer Application to Lead Lag Precedence Network and Bar-Chart, Special Study No. SSPR-ST-83-3. Asian Institute of Technology, Bangkok (1983). 5. R. Piyasena. Development of a Structural Analysis Interpretive Treatise for Microcomputer, MS. Engng thesis, Asian institute of Technology, Bangkok 11983). Development of a Standard Ele6. S. Attasaeranewong. ment Library for MICROFEAP, M.S. Engng thesis. Asian Institute of Technology, Bangkok (1984). Software Development for Three-DiI. S. Rattanalert, mensional Tall Building Analysis, M.S. Engng thesis, Asian Institute of Technology. Bangkok (1984). 8. Y. R. Liu, Towards the Development of a Fully-integrated Computer-Aided Analysis. Design. Drawing and Documentation System on iMicrocomputer. M.S. Engng thesis. Asian Institute of Technology. Bangkok (1984). 9. Y. N. Vidya. A Computer aid in the Take Off of Work Quantities of a Building Work, M.S. Engng thesis. Asian Institute of Technology. Bangkok (1983).
( RtTeiwd -I Fehrunp 1985) .ibstract-This paper describes a comprehensive use of microcomputer for professional practices of structural engineers. Transition from the conventional method to a computer-based procedure calls for a complete rethinking of an accurate. unambiguous definition of all the elementary activities, their logical orders as well as inter-relationships. An integrated database software system IS then proposed for analysis and design of structures. on microcomputer. With a common data bank. all previously processed data can serve directly as a basis for subsequent activities. such as computer-aided drafting. computer-aided construction planning, etc. Several program strategies to attain a maximum efficiency within existing microcomputer constraints are discussed.
design into one process. but to streamline their interaction through the use of a standardized data Analysis and design of structures constitute the two structure. Their traditional distinction as well as major activities in structural engineering practices. their logical orders remains unchanged. A data bank Generally, structural analysis and design are two serves to pool all data units at one place, accessible mutually coupled processes. While design variables and amendable at various phases of analysis and affect the analysis, results of the analysis dictate design operations. the choice of these variables in the design. TradiThe idea of an integrated FEM/CAD system is tionally, analysis and design are handled as though not new. Over a decade ago FEM/CAD was the being independent. Thus theoretically, a trial and exclusive province of large aerospace companies. error process must be conducted to eventually meet The long occupation of a time-sharing system durthe requirements of both at the same time. By hand ing a CAD session may be too costly for smaller calculations, the attempt to adhere to this consistengineering firms. Now this situation has changed ency may require such a big effort. It may not be drastically with microcomputers. Despite the low worthwhile if member designs are merely based on microprocessing speed, the insignificant operating the result of a simplified analysis. The provision of cost as well as the attractive input/output capabila factor of safety is assumed to help offset errors ities often make FEMiCAD on a microcomputer caused by stress misrepresentations as a result of very effective. In practice. the low processing analysis over simplifications. This often leads to a speed does not necessarily pose a problem as design in which the margin of safety fluctuates human responses during a CAD take a much longer greatly from place to place. Such a design is far from time, whereas lengthy FEM operations can always optimalit). be unattended overnight. After the advent of computers, a rigorous analToday, professionalism in software writing is ysis can be automated with great generality using more critical than it used to be on the mainframe. the finite element method (FEM). This is possible Optimizations of program and data units in relation because of the objectivity of structural analysis, for to a microcomputer environment are crucial to alwhich an exact algorithm can always be established leviate excessive demands for microcomputer time without ambiguity. and human attendance. The present paper will outFor design, computer automation is not feasible line useful programming strategies for the develdue to (a) the lack of a unique algorithm for design opment of an integrated FE&l/CAD program. synthesis. and (b) the lack of “engineering sense” in computer. Lacking intelligence, however. the computer has great durability. precision, and a perSTRL‘CTWAL \lODEL fect memory. Computer aided design (CAD) aims Rigorousness of structural modeling depends at combining the roles of human designer and comlargely on the computing tool available to the strucputer, using the strengths of both to complement tural engineer. With manual calculations, he can and supplement each other. either simplify the structure into a model for which The objective of an integrated software system a solution is available, or retain its rigorous form is not an attempt to couple structural analysis and but strive to obtain an approximate solution. In both cases, correctness of the result is often ques+ Associate Professor, on leave from Asian Institute tionable. With computer. it is possible to formulate of Technology, P.O. Box 2754. Bangkok, Thailand. INTRODUCTIOY4
33
a more
representative
reflect
behaviors
structural
model.
of the real structure
which
more
will
are defined
closely
and comprehensively. In the past 25 years, been
so phenomenal
considered
that
today.
as an industrial
analysis.
FEM
problems
2. The design
stresses among
in the
by ,Ci( ). where
engineers
same group.
and
on office
mi3. The quality
Consider
a structural
structural nodes
elements
(joints).
standard
element
Several design
the entire section,
interaction
I. Structural tural
analysis
constitute activities
signer’s
definition
the
will be prefor all these
data units.
it is pos-
process.
is an attempt
subjected
to a set of system This
process
parts
parameters
true.
code,
process
applied
as a function
loads f,
“material
“primary
properties”
A(
joint
) represents
of model design
(I,,,. Primary
design
design
module
g,
(zp. and variables
(
This
DESIGN
sizes,
numbers. They
these
REVISION 1
ENGINEERING
OK.
DRAWINGS
I.
Conventional
analysis
from may
as need
together
as
“subjective”
etc. They
a
varb
to be re-
It is worthwhile. into a replaceable
library”.
is not acceptable process
could
a revision
of CI,
be represented
of detailings
bq
“tn”.
determines
variables”
and placements
depend
=
design mainly
the re-
CI,,. such as
of reinforcing
on ~1,. The general by D(
can be denoted
bars. form
) where
D(ri,“‘, S”‘. I,,. I, 1 Designer). engineering
are prepared
drawings
based
various A suitable
tivity
is also suggested.
to secure
DETAILING
)
activities
units.
(5)
and related
on the final
Note
in practice
I
as there
size and reinforcement
procedure
the
doc-
results
of
are demands
data
mode serves
factor.
that
design
problem
of a member
with
other
for continuation along
of member
design
is also desirable
ma)
members.
continuous
cases, the use of similar
for a set of members
associated
mode for each ac-
The interactive
a “Designer”
& DOCUMENTS
and deign
and
computerization
not be an uncoupled
In many Fig.
mathwhich
g of the system. up. N, . as well as the geometry Figure 3 depicts the inter-relationships among
ANALYSIS
FINAL
The
as well
codes
of
“objet-
can be incorporated
time to time.
“secondary
uments
NO
types
The
calculations,
program.
them
5. The process maining
6. Finally
I STRESS
They
from
to bring
u,“’ DESIGN
other
an anal-
MODELING
MEMBER
(3)
I,, can be various
regulations,
such as “a
1 PRELIMINARY
as tuo
I,, and I,.
strength
environment
of this process
1
equation
I,. Desiytrer).
) in a way that. for a design group
etc.
STRUCTURAL
This pro-
displace-
geometry
variables”
with
efficiency.
I, can be specifications
city
vised or updated
U(
ysis
for
4. If the design
stresses;
design
requirements.
as well
of information
therefore,
can be symbol-
uJ and s’ are, respectively,
be evaluated
by an inequality
of information
formulas
is necessary.
in which
practical
of the computer
type
by
ments and element
will
information,
are universally
with
of the struc-
variables.
type
ematical
design
to determine
and displacements)
(2)
adequacy.
judgements
supplementary tive”
model
represented
analysis
describing
and design ically
a for
I.
the entire
analysis
(stresses
and K(
and design
associated
to streamline
responses
Q(
design
model
a new clear-cut
and their
then
in Fig.
be denoted
) represents quality characteristics of the ) represents corresponding minimum requirements. This K( ) could be subjected to de-
where
and
of structural
a general
between
With
activities
process
shall
up. CI,,. 11’) 2 ti(I,,.
Q(S)“,
or be
typical
process
etc.
properties
together
The
are presented
In this sented.
thus
group.
other
can have
of material
and
design
and design
at
column,
171
.i*’ of all elements
as
to any of the
such as beam.
values
variables,
completing
interconnected
of the same type
similar
member
sible
types.
elements
assigned
of
may belong
group
stresses
of a design
cess can be represented
as an assemblage
(members)
An element
the
= .Cf(s”).
to its strength
plus several
ISTER.4CTIOS
model
S”’ of a member
This
S”’
respect AND DESICS
u hich intluence
in
FEM
crocomputers.
0ALYSIS
quantities
can be determined
structural
to implement
practical
has
is alread!
for
to structural
age is the ability
large-scale
of FE&l
standard
A new challenge
the modern solve
the evolution
as those
analysis.
members. parameters
from
the view-
Microcomputer rMember
(deslgn)
7-
g : geometric
data
s:
f:loads
-
OF MEMBER
sD
i :
USER
Fig. 2. Interactions
GROUP
~
design
~.shsme”t -
NE<_E>SARY
INTERACTION
among activities
stress
I i
i
in analysis and design
processes.
point of a construction’s efficiency. The introduction of member design groups, therefore, allows the designer to specify a similar design for a set of members.
PROGRAM
practices
35
Derelopmenr philosophy
___d,lp,
/ANALYSIS
-
/
-prelm.
qz.$sD,
engineering
ever, with the current trend of hard disk rechnology, this problem will disappear soon.
elastic oroperties primary design variables secondary design variables design stresses given
new
system for structural
grouo IO.
PROPERTIES h:
integrated
STRATEGY
Limitarions of microcomputer
It is imminent that the microcomputer will become an integral part of the structural engineering profession. There is a need, therefore, to understand what a microcomputer can or cannot do. Generally, characteristic constraints of microcomputers are the following: 1. Limitation of incore memory. Microprocessor has a maximum limit on the size of addressible RAM space. After a portion is reserved for various system utilities, the remainder is then available for the user’s program and data. 2. Low speed ofj7oating-point arithmetic. Aside from the low microprocessor speed itself, a series of machine code instructions must be performed internally for each arithmatic operation in order to maintain a degree of precision. However, in some machines, a coprocessor for floating-point arithmetic can be added specifically to speed up these operations. 3. Low speed of disk data transfers. When data units must be shuttled between core and disk frequently, the total time for disk data transfers can exceed even the microprocessing time itself. How-
The development philosophy for engineering programs on a microcomputer can be totally different from that used for a mainframe. due to their great contrasts in capacity. operation, cost, and speed. For a mainframe, the expensive CPU requires a program to be highly efficient in terms of computer time. The cost of solving large FElM problems on a mainframe is governed mainly by the computer cost, and partly by the cost of engineers for data preparation and output digestion. as depicted in Fig. 3. For a microcomputer, the first and main concern has been whether practical, engineering problems can fit into microcomputer memory. If not. what are the necessary steps for achieving the same goal, even at the expense of excessive computer time. Since the cost of operating a microcomputer is negligible, the overall cost is likely to be dominated by the engineer’s costs. With this consideration, the adopted strategy is as follows: 1. To further save engineer’s time on data preparations and output digestions, the microcomputer should assist the engineer with its powerful interactive and graphic capabilities. During the input process, an error handling algorithm is necessary to prevent accidental fatal errors, which may lead to the total loss of all previously input data. 2. Time consuming, objective processes should be automated without a requirement for attendance. Therefore, a lengthyjob can be run overnight. 3. For subjective processes, the necessary machine/user interactions can proceed smoothly if the user is constantly informed of his possible actions or options, as well as useful information. Advanced
a) costs
for
note = : automated {G
: interactive
mainframe
1
_iTIJE
computmg
stage.
unattended.
stage.
attendance
1
required
Fig. 3. Cost components of a typical engineeringjob: frame vs microcomputer.
main-
Table I. FE%1 CAD criteria. CRITERIA FOR GOOD FEWCAD PACKAGE I.
constraints.
and stratcgxs
I+ICROCOWUTER CONSTRAINTS
GENEALITY (COMPLEXITY 0F CODES)
LI!lITATION OF tNCORE xE?lORY
: (a) PROGRA'i MODULARIZATION
RESOLUTION
(b) MODULE SEGMENTATION AND OVERLAY 2.
HIGH CAPACIn (DATA CAN BE VOLUMINOUS)
DISK DATA TRANSFER IS VERY TIME CONSUMING
~SOLII'KION : (a) OPTIMAZATION OF DATA BLOCKS (b) DATA STORAGE IN BINARY FORM RIGOROUSNESS
3.
FLOATING-POINT NUMBER CRUNCHING IS RELATIVELY SLOW
(HEAVY FLOATING-POINT NUMBER CALCULATION)
RESOLUTION : (a) MANUAL REDUCTION INTO EXPLICIT FORM (b) USE COMPILATION (c) INSTALL ARITHMATIC CO-PROCESSOR 4.
PRACTICALITY (REQUIRE co~ons~~s~s)
RESOLUTION : (a) ALLOW HUMAN INTERACTIONS WHEN NECESSARY (b) INSTALL 'KNOWLEDGE' & 'EXPERIENCES STATISTICS' (TOWARD AN 'EXPERT' SYSTEM)
L
interactive
facilities
“mouse”)
pen. ciency
LACK HUMAN INTELLIGENCE AND EXPERIENCES
(such
can
as graphic
be used
of the interactive
tablet,
to enhance
process.
light
order
efti-
systematically.
the
to generate,
retrieve,
or refresh
data
units
and thus increase
the productivity. MISS-A
XIICROCOMPUTER STRUCTURAL
I summarizes
Table practicality
on microcomputers, them.
More
paperill
and various
details
necessary
programs,
are
their
for
obstacles
steps to overcome
available
in
a previous
of the author.
The
first
larization. fined
some criteria
of FEMICAD
resolution
as a small
a well-defined tually
calls
In general, logical
function.
independent
for
program unit
program
modu-
can be de-
of data processor
Individual
and only
“administrative
a program module
modules
communicate unit”
with
are muwith
and
the
the
“data
bank.” Three
overlaying
schemes
gram units are depicted able for
linking
continuity taining
a core
units,
one at a time, of
Scheme An
must
finite
calls
be allow
a number
of research
pro-
units
with
no
while
main-
Scheme
(c) is useful
in a series
of satellite
for data c.xtributions. subroutines
The liemploys t
multitask
efficient data units
the past few years,
works[2-91 were conducted at the Asian Institute of Technology toward achieving a complete microcomputer package for structural engineering practices. The ultimate goal is an ambitious database FEMKAD program MISS which will handle a general multistory R.C. building project in the following aspects: (a) automated analysis by FEM; (b) computer aided design subjected to the AC1 (or a user supplied) code; (c) computer aided production of engineering drawings and related documents; (d) computer aided preparation of the bill of quantities
(b) is common
of a large module
element
integrated
numerical will
program
SYSTEM FOR
ENGINEERS
(a) is suit-
(c).
tremely sible,
Scheme
data continuity.
when brary
data.
segments
some
connecting
a set of independent
of in-core
for chaining
for
in Fig. 4. Scheme
During
INTEGRATED
program
data management. must
elaboration.
Whenever
ex-
t)APPEND
b)CHAIN
pos-
be cast in a form suitable for A standard
used consistently various
requires
1, LINK
modules
for
data
all data
the ease of
THREE
structure units.
This
access
in
OVERLAYING
SCHEMES
Fig. 4. Three microcomputer overlaying gram segmentation.
schemes for pro-
\Iicrosomputer
integrated
system for structural
engineering
practice5
37
this valuable form of data bank may replace or supplement the engineering blue prints as the permanent record of a building project. as required by the owner or a city building authority.
The paper described the development of a microcomputer integrated software system for application to professional activities in the structural engineering field. The objectives of this paper are twofold: ( I) to stimulate enthusiasm among structural engineers about the great potential of microcomputers for supplementing their professional activities; and (2) to show that the emergence of inexpensive microcomputers will not drastically change the principal roles and responsibilities of structural engineers (at least not until the coming age of artificial intelligence). REFERESCES
Fig. 5. Typical
modules of microcomputer tem for building projects.
integrated sys-
and breakdown costs; and (e) computer aided and monitored construction scheduling and resources planning. A key factor toward achieving this goal is the introduction of a centralized data bank, which serves as a common denominator to all processing modules. as shown in Fig. 5. Basically, data units in the data bank are continuously processed until they meet both the requirements of analysis and design. In the subsequent activities of (c)-(e) then, the data bank simply serves as an information center. With a standard data structure. an unlimited number of future modules can be added and integrated without disturbing the existing ones. Future expansions may include a module to update relevant data units in order to keep account with revisions made during the construction phrase. This will enable as-built engineering drawings to be prepared automatically and accurately. This author anticipates that. as a future practice.
and S. Udomlurgchai. DevelI. W. Kanok-Nukulchai opment Strategy of a Microcomputer Finite Element Analysis Program-MICROFEAP, in Engineering Sofrwarefur Microcompurers. (Edited by B. A. Schreffer et al.). Pineridge Press, Swansea (1984). S. Udomlurgchai. Development of General Purpose Finite Element Program on ;Microcomputer (APPLE II). M.S. Engng thesis. Asian Institute of Technology. Bangkok (1983). S. A. Abueva, Ultimate Strength Design of Reinforced Concrete Members on Microcomputer, Special Study No. SSPR-ST-83-I. Asian Institute of Technology. Bangkok (1983). 4. W. Wilaingarm, hlicrocomputer Application to Lead Lag Precedence Network and Bar-Chart, Special Study No. SSPR-ST-83-3. Asian Institute of Technology, Bangkok (1983). 5. R. Piyasena. Development of a Structural Analysis Interpretive Treatise for Microcomputer, MS. Engng thesis, Asian institute of Technology, Bangkok 11983). Development of a Standard Ele6. S. Attasaeranewong. ment Library for MICROFEAP, M.S. Engng thesis. Asian Institute of Technology, Bangkok (1984). Software Development for Three-DiI. S. Rattanalert, mensional Tall Building Analysis, M.S. Engng thesis, Asian Institute of Technology. Bangkok (1984). 8. Y. R. Liu, Towards the Development of a Fully-integrated Computer-Aided Analysis. Design. Drawing and Documentation System on iMicrocomputer. M.S. Engng thesis. Asian Institute of Technology. Bangkok (1984). 9. Y. N. Vidya. A Computer aid in the Take Off of Work Quantities of a Building Work, M.S. Engng thesis. Asian Institute of Technology. Bangkok (1983).