Professional Documents
Culture Documents
INTRODUCTION
69
1. The engineer has control over the Code of Practice to be used, the
unit system to be used, the material types and properties, company pol-
icy and architectural design constraints, and the rules used by the pro-
gram. The engineer's control is protected by the use of a password which
allows him to access the knowledge base of the program.
2. The designer, who may in practice be a draftsman or a technician,
accesses only the functions that assist him in the design and/or evalu-
ation of a concrete column. He may therefore execute designs that draw
upon the knowledge of his superior, the engineer, and of the experts
who originally participated in the development of the program. He is
guided through the design process by a series of decision trees which
are controlled by an inferencing subroutine.
ENGINEER'S ENGINEER'S
INTERFACE INTERFACE
SYMBOLIC RULE
DESIGN
PROCESSING
AMD INFERENCE PROPOSAL - •
KNOWLEDGE
ANALYSIS
BASE
DATA
BASE
decision trees cover issues such as requirements for seismic design, con-
crete cover to reinforcing, punching shear for columns supporting con-
crete slabs, and others. The rules processed set the spacing limitations
for reinforcement, minima and maxima for reinforcement content, and
constructibility limits on the gross column dimensions. For example, a
rule that limits the minimum spacing of longitudinal reinforcing bars in
a column to allow for adequate concrete placement is expressed as
spacing : 2 : 1.5 : in. : concrete placement
which has the standard form
subject : operator : value : unit : reason
An example of the type of expert "rule of thumb" used in the design
module is provided by the procedure for selection of appropriate bar
sizes for columns. From the experience of designers, it was established
that the size of the main reinforcing bars was proportional to the min-
imum outside dimension of the column section (Fig. 2). As the bar sizes
are discrete, a set of possible sizes is sensible for a given column, The
program also uses accumulated data to aid its selection of b/h ratios for
rectangular columns. Through an option available, each column de-
signed using the system is evaluated by the engineer with respect to the
b/h ratios and the bar sizes. The program will then adjust itself to in-
corporate this evaluation (experience) in proposing future designs.
Neither the decision trees nor the symbolic rules are part of the pro-
gram; they are stored in knowledge base files and are manipulated by
71
0 5 10 15 20 25 30 35
72
73
2. Set the neutral axis depth c equal to the neutral axis depth com-
puted from balanced failure condition for the section.
3. Compute the value of Pu and update c using a modified secant nu-
merical method until P„ = P (see Fig. 5 for description).
4. Compute eux and euy and compare with ex and ey to decide whether
the section is adequate or not.
POSSIBLE SENSIBLE
POINTS OF ACTION
OF ULTIMATE LOADING
0 X
P(GIVEN)
NUMBERS INDICATE
COMPUTATION
SEQUENCE
BAL ^1^2
74
CONCRETE
TU
In
STRESS STRESS f
Y
STEEL
E=0.002 +
2000008,
*m fy
f
y
/,E=29x10*
0 0.002 STRAIN ° STRAIN
axis depth and is then iterated until the vertical load is within 1% of the
given load. (This requires up to four iterations using a modified secant
method, as shown in Fig. 5, for the solution of the nonlinear relationship
between neutral axis depth and vertical load.) The calculations use the
stress-strain diagrams for the materials (Fig. 7) given in the design codes,
and no assumption of an equivalent stress block is required. In addition,
each reinforcing bar is considered individually in its true position.
APPLICATION EXAMPLE
PROGRAM RELIABILITY
Testing of the program design module to date has shown that its pro-
posals are either very close to those of human expert designers or more
efficient. Sample results are given in Table 1. The EIDOCC designs are
in some cases more sensible than other designs, e.g., ACI SP-17A gives
a 15x15 in. column with 8 #10 bars and #3 ties for a design problem
where the vertical load is P = 208 kips and the moments are Mx = 2,064
kip-in. and My = 828 kip-in. (no further design information is supplied).
Clearly the column is required to be significantly stronger about the x-
axis than about the y-axis and if no other restrictions (architectural, avail-
ability of forms, constructibility) apply, a rectangular column would be
preferable to a square column. The EIDOCC initial design proposal is
26x12 in. with 12 #8 bars and # 3 ties. Refining and analyzing shows
that a 20x12 in. column with 10 #8 bars and #3 ties is adequate. (This
section has 6% more concrete but 25% less steel than the 15x15 in. sec-
tion.) Other examples are shown in Table 1.
The accuracy of the analysis module was tested by comparison with
experimental results achieved by Bresler (1960) and Ramamurthy, with
analysis results of Bresler (1960) and Salmon and Wang (1973), with de-
sign chart results of ACI SP-17A (1973) and Park and Paulay (1975), and
with program results of Taylor (1985) and Ehsani and Rosenbaum (1985).
The experimental results are actual failure loads and eccentricities for
model columns; EIDOCC analysis of the same column sections gave fail-
ure loads and eccentricities that were on average 2% lower than the ex-
76
CONCLUSION
(Yes/No/Explain) -> Y
Does the architect require a circular shape?
(Yes/No/Explain) -> N
Does the architect have specific dimension limits for this column?
(Yes/No/Explain) -» N
4) Fix Column Area.
In the Y Direction:
Is the column braced against sidesway?
(Yes/No/Explain) -» Y
In the X Direction:
Is the column braced against sidesway?
(Yes/No/Explain) -> Y
Does the column support a concrete slab at any point in its height?
(Yes/No/Explain) - ^ Y
Is the slab supported by beams?
(Yes/No/Explain) -> Y
5) Proportion Steel and Adjust Section.
Is the column to be cast in place?
(Yes/No/Explain) -> Y
Is the concrete to be cast against and permanently exposed to earth?
(Yes/No/Explain) -» N
Is the concrete to be exposed to earth and weather?
(Yes/No/Explain) -> E
This implies exposure to both temperature and moisture changes.
Is the concrete to be exposed to earth and weather?
(Yes/No/Explain) -> Y
6) Choose Bar Placement Pattern.
The following bars are available.
Those preferred are marked by asterisks ' #*/
Bar number 4 5 6 7 8 9 11 14 18
Preference
Bar Diameter 0.38 0.50 0.63 0.75 0.88 1.00 1.13 1.37 1.75 2.26
79
APPENDIX II.—REFERENCES
80
APPENDIX III.—NOTATION
81