Sei sulla pagina 1di 12

Knowledge-Based Systems 15 (2002) 177±188

www.elsevier.com/locate/knosys

Knowledge based-system for alternative design, cost estimating


and scheduling
Abdulrezak Mohamed a,*, Tahir Celik b,1
a
Assistant Prof., Department of Civil Engineering, Eastern Mediterranean University, Gazimagusa, Mersin 10, Turkey
b
Prof., Department of Civil Engineering, Eastern Mediterranean University, Gazimagusa, Mersin 10, Turkey
Received 1 August 2000; accepted 2 February 2001

Abstract
The traditional design, cost estimating and scheduling procedures focus on assessing costs and schedule after design decisions are made
and conveyed to construction drawings and design that is, they do not have a mechanism that allow the designer to perform a rapid `what if'
analysis on design alternatives and cost analysis of different types of building materials and alternative schedule analysis at early stage of
project without accompanying of detailed design and drawing. Moreover, the available scheduling tools are not integrated to design and cost
estimating system and hence require manual input of activities lists, activities sequencing, resources, costs and durations to generate project
schedule and they are unable to consider the effect of productivity factors and the resulting uncertainty inherent in each activity. This paper
presents an integrated knowledge based system for construction cost estimating and scheduling. The knowledge based system supports an
automated alternative design analysis with on line schematic drawing, material selection, crew selection and productivity analysis for
generating activities sequencing, normal durations, productivity adjusted durations, cost estimate, list of materials selected and their
respective quantities and list of assigned crew. An analysis of alternative design, cost estimating and scheduling for a residential building
using the integrated system is presented. q 2002 Elsevier Science B.V. All rights reserved.
Keywords: Cost estimating; Building design; Planning and scheduling; Productivity; Expert system

1. Introduction resulting costs, and have no alternative material selection


integrated to the cost system.
Building design and material type used are the two main Once the client's wants are translated into detailed design
parameters that have signi®cant impact on the cost of a and drawings, it is very costly and time consuming to revisit
building. However, the traditional design Ð cost estimating these decisions because the project must be redesigned and
procedures generally focus on assessing costs after design redrawn to accommodate changes. Because redesign is so
decisions are made and conveyed to construction drawing as costly and time consuming, the engineer/architect selects
in quantity take off. On the other hand, comparative cost one solution with which he/she is most familiar without
estimating (unit cost per square meter) uses historical data, investigating different design solutions and different build-
which is collected by averaging buildings of different ing materials that satisfy the client's needs and ®nancial
design, height and materials. Even today's Computer constraints.
Aided Design (CAD) incorporate mechanism to take off More over, computer based schedule analysis for project
quantities from completed design and drawings; however, management have been used since late 1950's with the
basically are same as the traditional quantity take off esti- introduction of Critical Path Method (CPM). Obviously,
mating because a single speci®ed design still controls the advances in computer technology have greatly facilitated
the use of CPM and many Personal Computer (PC) oriented
scheduling Softwares such as Primavera project planner,
Microsoft project, Auto plan, Artemis, project scheduler
* Corresponding author. Tel.: 190-392-630-2359; fax: 190-392-365- etc. become available at reasonable prices. Despite the
0710.
advances in computer technology, these computerized
E-mail addresses: abdulrezak.mohamed@emu.edu.tr (A. Mohamed),
tahir@emu.edu.tr (T. Celik), celik@iit.edu (T. Celik). algorithmic scheduling systems require complete input of
1
Tel.: 190-392-630-1443; fax: 190-392-366-4479. activity description, precedence relationships, duration and
0950-7051/02/$ - see front matter q 2002 Elsevier Science B.V. All rights reserved.
PII: S 0950-705 1(01)00155-1
178 A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188

allocation of resources for each activity. In fact, they are 2. Knowledge based systems application and object
more of schedule report generators and calculators rather oriented models in construction
than scheduling tools that aid and provide alternative
solutions to schedulers and project managers. This state- Knowledge based expert systems have been a subject of
ment is not intended to down play the major improvements considerable research in civil engineering in recent years,
that computer automation have made in avoiding mathema- especially in the area of construction engineering and
tical errors found in hand calculation of activities: early management [1±7]. This section brie¯y describes several
start, early ®nish, late start, late ®nish, ¯oat and identifying previous work that used knowledge based systems and
the critical path. However, basically, most of the task at ®rst object oriented models for schedule generation and/or cost
painstakingly is performed manually by the scheduler. The estimating.
scheduler should formulate the net work, sequences Construction Planex [1] uses a bottom±up approach to
activities, and calculates activities durations and cost in generate construction plan. It generates element activities
order to prepare input data for these scheduling tools. for the building components, aggregates the element activ-
Because forming schedule manually is very costly and ities into project activities, assigns crew type and size,
time consuming the scheduler cannot perform several sequences the project activities, estimates costs, and dura-
schedule alternative to evaluate time integrated cost that tions. However, Construction Planex requires detailed
indicate the timing of capital needs for the design decisions descriptions of design elements (dimensions and co-ordi-
and construction methods selection. nates) and it deals only with excavation and erection build-
This paper presents an integrated knowledge based ing skeleton and does not interact with any computer-aided
system for alternative design decisions, material selection, design systems. GHOST [2] takes as an input a set of
cost estimating and alternative schedule analysis by activities to be performed and produces as output scheduling
simulating the different factors that affect productivity. by setting up precedence relationships among activities.
The knowledge based system Alternative Design Cost SIPE [3] generates and sequences repetitive activities but
Estimating and Scheduling (ADCES) has two parts. The does not consider construction costs and durations. ESCHE-
®rst main part is for alternative design, cost estimating DULER [4] is a rule based hybrid prototype system for
and schedule analysis in the predesign stage with auto- construction planning and scheduling. It uses rules to
mated on line schematic drawing displays. Here the standardize activity description, determines activity
designer/user inputs general information of project, such sequences, and adjusts user input activity durations due to
as building type, ¯oor area, roof type etc. and selects variations in productivity and does not consider cost.
material types from series of material selection displays, CONSCHED [5] generates activities list, activities sequen-
assigns number of crew from series of crew selection cing and durations. However, CONSCHED requires user
displays and assigns productivity factors that re¯ect possi- input quantity and productivity rates to generate activity
ble project site conditions from a series of productivity durations and does not consider cost. HISCHED [6] gener-
factor selection displays. Then the system by integrating ates activities list, determines dependencies between activ-
the user input and selection with experts knowledge, ities and schedules activities. ICEMS [7] is an integrated
building regulations, design criteria, and current cost system for alternative design decision, material selection
data, generates quantities, cost, activity lists, normal and cost estimating mainly for predesign analysis.
activity durations, adjusted activity durations, bill of CONSCOM [8,9] is an object oriented information model
materials and list of assigned crew. The system allows for construction scheduling, cost optimization and change
the designer engineer/architect to perform a rapid cost order management. The model can be used by the owner as
analysis of design alternatives, schedule alternatives and an intelligent decision support system in schedule reviews,
different use of materials prior to expending resources for progress monitoring and cost-time trade off analysis for
detailed engineering analysis and drawings. Once the change order particularly for highway construction.
designer reaches to project design and cost that meets This paper like the aforementioned works addresses the
the client's needs he/she could proceed with greater integration of alternative design, cost estimating and
ease and con®dence for detailed engineering analysis scheduling. It addresses the problem at two stages: (a) At
and drawings. The second part of the system is material predesign stage with general information of project provides
selection, cost estimating and alternative schedule analy- early cost and schedule estimate. (b) After detailed design
sis when detailed design and drawing are available. The for alternative material selection, cost estimating and alter-
input for the second part is received in two ways: either native schedule analysis.
the designer inputs manually into edit table the dimen-
sions of building from completed drawing and design or if
a CAD and Structural design soft ware is used the result- 3. System development
ing output can be imported to ADCES from Microsoft
Excele after few adjustment that ®ts the format of This section outlines the development and implementa-
LEVEL5 object design class. tion of the system and describes its basic components.
A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188 179

Fig. 1. Structure of knowledge base.

3.1. Structure of knowledge base through three sources: (1) literature review, (2) experts
interaction, (3) foreman delay survey.
The structure of knowledge base has the following
components as shown in Fig. 1. 1. Literature review was conducted in three directions (a)
(1) The knowledge based system is implemented on literature about construction cost estimating, bidding and
object oriented environment by using LEVEL 5 scheduling: [13±22]. (b) Literature about factors that
OBJECT w for Microsoft w Windowse release 3.6 by affect productivity on construction site and methods of
Information Builders Inc.[10,11]. LEVEL 5 Object has quantifying the productivity levels: [23±31]. (c) Expert
an integrated array of powerful tools: Graphical User system application and object oriented models in
interface development, forms and display builders and construction management [1±9], [32±38].
has the capability to chain more than one knowledge 2. Expert interaction were done with three parts: (a) Regular
base together. It accesses to all common data base formats visits were paid to top three construction companies in
and SQL, interfaces to external programs, has communi- the region and structured interviews were conducted with
cation paradigms, text ®les, timers, and custom interfacing the companies' design engineers, cost estimators and
options and can be used as DDE server. (2) Microsoft schedulers. For the purpose of this study, the buildings
Excele version 5.0 is used for storing data of several were classi®ed as residential buildings, of®ces and
project trials, and drawing graphs for easy comparison schools and further classi®cations are made as low-rise,
purpose. It stores weather data, average temperature, mid-rise and high-rise. The knowledge concerning the
humidity, and average rain. It is also used to receive building types were grouped under six categories:
design data from PROBINA w in order to make few dele- general, architectural, structural, mechanical, electrical
tions and adjustment that suits the format of LEVEL 5 and ®nishes. Special forms were prepared to tabulate
OBJECT w design knowledge library classes. (3) PROBI- the knowledge extracted from experts and separate
NA w is a Computer Aided Design and Drafting and struc- records were kept for each company. The design experts
tural design soft ware version 8.5 for Microsoft Windows (Engineers and Architects) were requested to give their
by Prota Bilgisayar [12]. (4) Microsoft Project for engineering judgment and dimensions of building
windowse or Primaverae based on user selection are elements for different category. To supplement their
used for generation of schedule report. experiential design knowledge a thorough and extensive
study were done for different built historical projects.
The elicited knowledge were then put into their respec-
4. Knowledge acquisition tive design knowledge library classes. For the ®nishing
part of the building, the architects and design engineers'
An important part of building knowledge based system is practical knowledge and judgments have been consid-
acquiring the knowledge needed to achieve a desired level ered in similar manner. (b) Regular meetings were
of performance. The system's knowledge was acquired conducted with professional engineers and architects
180 A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188

Fig. 2. On-line schematic drawing display for selected design parameters and input.

responsible for building new faculty buildings, repair and ties and structure of a group of objects and attributes
maintenance of the university campus. (c) Lecturers in describe the object's important characteristics. The values
the Department of Civil Engineering have been the main of these attributes for a particular instance are obtained from
domain experts for building design criteria according to the following sources: (1) querying the user; (2) methods,
TS500 (Turkish Standard 500). Practical equations were rules, demons, facets or the PURSUE command, and (3)
formulated for the calculation of required reinforcements knowledge library classes.
for different structural elements for predesign analysis. In ADCES knowledge is represented in three main types
(d) Materials costs that are not available in the unit prices of classes: (a) knowledge library classes, (b) selection
list, which are published by the Ministry of Public Work classes, and (c) project classes.
and Housing were collected directly from material
suppliers. (a) Knowledge library classes contain design, construc-
3. Foreman delay survey was carried out in four different tion crew, cost and productivity level of information.
sites over a period of nine months in order to identify They have prede®ned instances and serve as a data
factors that affect productivity and quantify loss factors base. The library classes will not be modi®ed or adjusted
(percentage) for the different factors under different during estimating and scheduling processes with the
conditions for the locale. Tucker's method is adopted exception of cost library. However, all library classes
in analyzing and evaluating the foreman's delay survey can be changed from instance editors to re¯ect the
[39]. changes in construction location and time. The library
classes are categorized into ®ve groups.
(i) Design library classes: consist of a collection of
classes which contain design information for different
building category. In addition to the different knowl-
5. Knowledge representation edge of structural elements of building category, the
major important part of ADCES is its ability on
After the knowledge acquisition stage reached a suf®cient displaying on-line architectural schematic drawings
point that allow an ef®cient system to be developed, the of building selections made particularly for predesign
knowledge representation had been started. The knowledge analysis as shown in Fig. 2. Each schematic drawing
in the system is represented by a group of objects. Each has a corresponding class that contains its dimensions
group of objects is represented by a class with its attributes. and is displayed in an edit table from which the
In LEVEL5 OBJECT w, a class de®nes the general proper- designer can make changes for his particular run by
A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188 181

Fig. 3. Edit table for displayed schematic drawing.

over writing the values in the edit table as shown in Fig. quanti®ed with mathematical equations. It has been
3. This approach is a very quick and effective way for mostly dealt in probabilistic approach. For the purpose
the architects/design engineers to communicate with of this study a similar approach to Benjamin and
the client's desires particularly at brie®ng stage. Greenwald [41], and Car [42], has been adopted.
Once the client is satis®ed with on-line schematic Threshold values are established and based on the
drawing or commented on modi®cations, the design threshold values, the productivity levels are deter-
engineer proceeds to the second important step; mined.
performing alternative design analysis, material selec- (b) Selection classes: contain design and scheduling para-
tion and schedule analysis. meters, type of materials, type and number of crew and
(ii) Cost library class: contains the unit prices and it is productivity factors. Each type of design and scheduling
attached to the edit table and is displayed in cost parameters, type of materials etc. are represented by
display as shown in Fig. 4 from which the user can compound attributes and the compound attributes are
change the unit prices for his/her particular run by attached to series of design selection, material selection,
editing on the edit table. crew selection, and productivity factors selection displays
(iii) Crew library class: contains information about the as shown in Figs. 6±8. The compound attributes also have
crew type, crew description and daily crew productiv- attached methods, demons and production rules, which
ity as shown in Fig. 5. are part of the different of process modules.
(iv) Productivity level library class: contains produc- (c) Project classes: contain QTO and cost class, activity
tivity levels for the different productivity factors under class, schedule class, aggregated cost class and list of
different conditions. selected material class, as shown in Figs. 9±11 respec-
(v) Weather library class: contains the average tively. The instances and attribute values of these classes
temperature, humidity and average rain and their corre- are created and determined by process modules. The
sponding productivity levels. The productivity level process modules consists of production rules, methods,
related to cold weather and hot/warm weather is demons and system commands. There are six types of
obtained from Koehn and Brown [40] mathematical process modules that perform different types of processes
equations. The values of these mathematical equations and these are (i) Quantity take off process modules, (ii)
for cold and hot weather for the year, is imported from cost generation process module, (iii) schedule process
Microsoft Excel to weather library class. The effect of module, (iv) productivity analysis process module, (v)
rain unlike temperature and humidity, which has been material aggregation process modules and vi) cost aggre-
quanti®ed with mathematical equations, has not been gation process modules.
182 A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188

Fig. 4. Unit prices list display.

6. Operation of the system 1. User selects the design and scheduling parameters from a
series of design selection displays. If the cost estimate and
ADCES performs the following sequence of operations in the selected scheduling is predesign, ADCES presents the
generating cost, quantities and schedule of project: user predesign input display for editing the general project

Fig. 5. Crew description display.


A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188 183

Fig. 6. Typical material selection display.

information such as block number, ¯oor number, ¯oor 2. Based on user design and scheduling selection and input
area, ¯at area, roof type, etc. For each input of ¯oor ADCES creates instances for the design elements in the
area, ¯at areas and selection of design parameters QTO class. By integrating the experts judgment, engi-
ADCES displays a typical schematic drawing. neering criteria and user selection and input ADCES

Fig. 7. Typical crew selection displays.


184 A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188

Fig. 8. Typical productivity factor selection display.

generates quantities for each element. Similar process values of instances of after detailed design knowledge
modules used to generate quantities for predesign analy- library are imported from CAD and structural design
sis are also used for after detailed design; however, the soft ware or edited manually whereas for predesign
design knowledge libraries are different, the attribute analysis they are prede®ned.

Fig. 9. Generated schedule display.


A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188 185

Fig. 10. Aggregated cost display table.

3. ADCES displays unit prices display, as shown in Fig. 4 found it to be less/more than the market prices during
before material selection display in order to give the the estimating period. This is very important particu-
user a thorough information about material cost. The larly in countries where in¯ation rate is high.
user can change the unit prices of any item if he/she 4. ADCES displays material selection displays for the user

Fig. 11. Generated bill of materials display table.


186 A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188

Fig. 12. Interactive and iterative processes between user and system.

to select the type of material he/she would prefer to use 7. ADCES displays crew description display before crew
as shown in Fig. 6. selection displays in order to give the user thorough
5. By using the current unit prices of materials selected, information about crew type, equipment and crew
ADCES generates costs, and breaks the cost of each daily output. Then user assigns crew number from
design element into material, labor and equipment cost. crew selection displays as shown in Fig. 7.
6. ADCES creates instances in activity class and 8. Using the generated quantities and assigned crew sizes
sequences activities and determines quantities of activ- ADCES generates normal activities durations and
ities by aggregating from QTO class. ADCES assigns crews to each activity.
sequences activities based on physical, and/or resources 9. ADCES display productivity analysis displays. If the
related relationship. user selects perform productivity analysis, ADCES
A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188 187

displays a series of productivity factor selection display it would take several days for an experienced engineer to
for the user to specify the productivity factors that take off quantities, ®nd cost and produce schedule for each
re¯ect the job and site conditions of the project as of the project. The timely decisions provided by ADCES
shown in Fig. 8. Then ADCES generates productivity makes it a valuable asset to engineers and schedulers parti-
impacted durations as shown in Fig. 9. cularly at the predesign stage.
10. ADCES aggregates cost either according to Building
Speci®cation Institute (BSI) masterformat or Construc-
tion Speci®cation Institute (CSI) masterformat based on 8. Conclusions
the user selection from design and scheduling displays
Traditional design, cost estimating and scheduling takes
as shown in Fig. 10.
many man hours of detailed effort to measure the quantities
11. ADCES lists the selected materials and aggregates the
from design drawings and then compute the costs separately
quantities of each selected material for the whole
and lastly the designer/scheduler translates the design infor-
project as shown in Fig. 11.
mation to activity lists, and formulates the network to
12. The user can repeat the process several times by
prepare input for the traditional scheduling tool. This is
changing the building design, building materials and
costly and time consuming to be used to search for alter-
assigning different crew sizes and methods until he
native design solutions, cost analysis and alternative
meets the optimum point that satis®es his/her cost and
schedule analysis particularly at predesign stage. In this
schedule constraint. Interactive and Iterative process
study, an automated construction cost estimating and
between the user and system is shown in Fig. 12.
scheduling is presented. The automated system addresses
the integration of cost estimating and scheduling at two
stages; at predesign stage and after detailed design stage.
ADCES has several notable features such as the automated
schematic displays, the alternative design analysis at prede-
7. System evaluation sign stage, the alternative material and crew selection at
both stages, and alternative schedule analysis simulating
The performance of ADCES predesign analysis and after
the different productivity factors at both stages. More
detailed design analysis was evaluated with respect to low-
notable is its timeliness and accuracy which makes a true
rise and mid-rise residential buildings. The performance is
valuable value engineering tool prior to design and a
rated against the measures of accuracy and timeliness. To
valuable analysis tool after design.
perform the evaluation, ®rst historical data of several built
low rise and mid rise residential buildings have been
obtained from construction companies. The new cost of Acknowledgements
each building is found by using current unit prices. Then,
ADCES is used for generating the quantity, material cost, The writers would like to thank the Eng. Fuat Kutlu
labor cost, equipment cost and list of materials selected and general director of Korman Const. Co. and all the staff of
their respective quantities for the same projects. The results Korman Const. Co. and their subcontractors, Eng. Yurdaer
obtained from ADCES predesign analysis and selection, Garip of Makro Teknik Co. and Eng. M. Tunar of Project
were compared with actual values. The total costs were Technique for lending their expertise and providing us
found to vary between the range of 2.1±4.21% depending valuable data in developing ADCES.
on the complexity of the design. These differences are insig- The writers have received valuable input from several
ni®cant comparing to the timely decisions and alternative colleagues in the Depts. of Civil Eng., architecture and
design and cost analysis mechanism provided by ADCES at Mechanical Eng.: Prof. Dr A. Topcu, Dr Erden Karaboga,
the moment when they are mostly needed. Arch. Fevzi Ozerbey, and Dr Mustafa Dagbasi. We grate-
In the second case, a manual schedule was prepared inde- fully acknowledge their valuable contributions.
pendently, and then compared with the schedule produced
by ADCES in terms of: (i) whether the schedule includes all
the activities necessary for completion. (ii) Are the sequence References
of activities produced reasonable and agree to the ones
[1] C. Hendrickson, D. Martinelli, D. Rehak, Hierachical rule based
produced manually and (iii) Are activities durations gener- activity duration estimation, J. Constr. Engng Mgmt 113 (2) (1987)
ated reasonable. In all respects, the evaluation and compar- 288±301.
isons proved that the schedule generated by ADCES were [2] D.S. Navinchandra, R.D. Logcher, GHOST: project network genera-
satisfactory. The time required to generate the quantities, tor, J. Comp. Civ. Engng 2 (3) (1988) 239±254.
costs, bill of materials, activities durations, activities [3] N. Kartam, R. Levitt, Intelligent planning of construction projects,
J. Comp. Civ. Engng 4 (2) (1990) 155±176.
sequence and crew assignment was found to be 8±16 min [4] O. Moselhi, M.J. Nicholas, Hybrid expert system for construction
depending on whether the productivity analysis will be planning and scheduling, J. Constr. Engng Mgmt 116 (2) (1990)
performed or not and complexity of the project. Whereas 221±239.
188 A. Mohamed, T. Celik / Knowledge-Based Systems 15 (2002) 177±188

[5] O. Shaked, A. Warszawski, Expert system for scheduling of modular [23] C. Oglesby, H.W. Parker, G.A. Howell, Productivity Improvement in
construction projects, J. Constr. Engng Mgmt 118 (3) (1993) 488± Construction, McGraw Hill, New York, 1989.
506. [24] J.R. Baldwin, et al., Causes of delay in construction industry, J. Constr.
[6] O. Shaked, A. Warszawski, Knowledge based system for construction Div., ASCE 97 (2) (1972) 177±187.
planning of high rise building, J. Constr. Engng Mgmt 121 (2) (1995) [25] C.T. Grimm, N.K. Wagner, Weather effect on mason productivity,
172±182. J. Constr. Div., ASCE 100 (3) (1974) 319±335.
[7] A.N. Mohamed, T. Celik, An integrated knowledge based system for [26] R.D. Logcher, W.W. Collins, Management impact on labor produc-
alternative design, material selection and cost estimating, Expert Syst. tivity, J. Constr. Div., ASCE 104 (4) (1978) 447±461.
Appl. 14 (3) (1998) 329±339. [27] J.D. Borcherding, D.F. Garner, Work force motivation and produc-
[8] A. Karim, H. Adeli, Object-oriented information model for construc- tivity on large jobs, J. Constr. Div., ASCE 107 (3) (1981) 443±453.
tion project managment, J. Constr. Engng Mgmt, ASCE 125 (5) [28] H.N. Ahuja, Nandakumar, Simulation model to forecast project
(1999) 361±367. completion time, J. Constr. Engng Mgmt 111 (4) (1985) 325±342.
[9] A. Karim, H. Adeli, CONSCOM: An OO construction scheduling and [29] H.R. Thomas, Learning curve models of construction productivity,
change management system, J. Constr. Engng Mgmt, ASCE 125 (5) J. Constr. Engng Mgmt 112 (2) (1986) 245±258.
(1999) 368±376. [30] H.R. Thomas, Impact of material management on productivity,
[10] Level Object for Microsoft Windows reference guide manual, release J. Constr. Engng Mgmt 115 (3) (1990) 370±385.
3.6, Information Builder Inc. 1995. [31] H.R. Thomas, Effects of schedule overtime on labor productivity,
[11] Level Object for Microsoft Windows, Getting started guide manual, J. Constr. Engng Mgmt 118 (1) (1992) 60±76.
release 3.6, Information Builder Inc. 1995. [32] H. Adeli, Expert System in Construction and Structural Engineering,
[12] Probina for Microsoft Windows, version 8.5. Reference guide Chapman and Hall, New York, 1988.
manual, Prota Bilgisayar, Inc., Ankara, 1995. [33] H. Adeli, Knowledge Engineering, vol. I, McGraw Hill, New York,
[13] J.J. O'Brien, Preconstruction Estimating Budget through bid, 1990.
McGrawHill, New York, 1994. [34] H. Adeli, Knowledge Engineering, vol. II, McGraw Hill, New York,
[14] R.L. Peurifoy, W.B. Ledbetter, Estimating Construction Costs, 1990.
McGrawHill, New York, 1985. [35] M. Arockiasamy, Expert Systems Applications for Structural, Trans-
[15] R.C. Smith, Estimating and Tendering for Building Work, Longman portation and Environmental Engineering, CRC press, Inc. NW, Boca
Scienti®c and Technical Books, London, 1986. Raton, Florida, 1993.
[36] T. Celik, et al., Development of an expert system in concrete produc-
[16] P.J. Cook, Estimating for Constractors, R.S. Means company Inc,
tion procedures, Concr. Int. 11 (8) (1989) 37±41.
1982.
[37] J.P. Ignizo, Introduction to Expert Systems: the Development and
[17] R. Pilcher, Principles of Construction Management, McGraw Hill,
Implementation of Rule based Expert System, McGraw Hill, New
London, 1992.
York, 1991.
[18] R.H. Clough, G.A. Sears, Construction Contracting, John Wiley, New
[38] S. Mohan, Expert system application in construction and management
York, 1994.
and engineering, J. Constr. Engng Mgmt 116 (1) (1990) 87±99.
[19] J.J. O'Brien, CPM in Construction Management, McGraw Hill, New
[39] R.L. Tucker, D.F. Rogge, W.R. Hayes, F.P. Hendrickson, Implemen-
York, 1993.
tation of foreman delay surveys, J. Constr. Div., ASCE 108 (4) (1982)
[20] M. Skitmore, Contract bidding in Construction Strategic Management
577±591.
Molelling, Longman Scienti®c and Technical, London, 1989.
[40] E. Koehn, G. Brown, Climatic effect on construction, J. Constr. Engng
[21] H. Adeli, A. Karim, Scheduling/cost optimization and neural
Mgmt 111 (2) (1985) 129±137.
dynamics model for construction, J. Constr. Engng Mgmt, ASCE
[41] N.B. Benjamin, T.W. Greenwald, Simulation effects of weather on
123 (4) (1997) 450±458.
construction, J. Constr. Div., ASCE 99 (1) (1973) 175±190.
[22] H. Adeli, M. Wu, Regularization neural network for construction cost
[42] R.I. Carr, Simulation of construction project duration, J. Constr. Div.,
estimation, J. Constr. Engng Mgmt 124 (1) (1998) 18±24. ASCE 105 (2) (1979) 117±127.

Potrebbero piacerti anche