Rafael Sacks, Ph.D.Research Scientist

College of ArchitectureGeorgia Institute of Technology

Atlanta, Georgiaand

LecturerFaculty of Civil and Environmental

EngineeringTechnion – Israel Institute of Technology

Haifa, Israel

Charles E. Eastman, Ph.D.Professor and PhD Program HeadCollege of ArchitectureandProfessorCollege of ComputingGeorgia Institute of TechnologyAtlanta, Georgia

Design professionals worldwide have applied thetechnology of computer-aided design and drafting(CAD) on a broad scale, primarily to increase theefficiency of manual design and drafting methodsand to promote standards, rather than to improvethe process itself. Even with improvements in thetechnology, however, errors in design and draftingremain common. Taking the 2-D CAD technologyfurther, the application of three-dimensionalintegrated parametric modeling of precastbuildings at the assembly and piece levels mayenable producers to greatly reduce design errors,resulting in significant improvements in projectquality, cost, and schedule. An examination of anumber of case studies of precast/prestressedconcrete projects has revealed that the commoncauses of construction problems are design,detailing, and drafting errors, a lack ofcoordination between different disciplines, andinadequate management of changes. An analysisof the cases presented in this paper indicates thatthe application of 3-D top-down modeling andautomated production of shop drawings holds thepotential to eliminate most of the sources of error.

Although computer-aided drafting has become preva-lent in all branches of the construction industry, asignificant portion of construction dollars is still

spent on correcting errors made in the design stage.1 Build-ing parts that do not align correctly, spatial conflicts be-tween components of different systems, and work that mustbe demolished because drawings were not updated to

Process Improvements in PrecastConcrete Construction UsingTop-Down Parametric 3-DComputer Modeling

2 PCI JOURNAL

Ghang LeePh.D. Candidate

College of ArchitectureGeorgia Institute of Technology

Atlanta, Georgia

reflect design changes are among thecommon errors.

A recent comprehensive study ofseven large construction projects (em-ploying structural steel, masonry, andcast-in-place concrete constructionmethods) showed that design errorsaccounted for an average of 26 percentof all construction defects.2 A similarfield survey of cast-in-place reinforcedconcrete construction revealed numer-ous and diverse reinforcing bar con-structibility problems. These deficien-cies arose largely from inadequatedetailing, lack of construction experi-ence among designers, poor coordina-tion between the design of the variousdisciplines (structural, electrical, andother trades), and insufficient involve-ment of contractors in detailing.3

In a recent survey of U.S. precast/prestressed concrete producers, 41percent of respondents reported en-countering problems in productiondue to ambiguities in design “often”or “very often.”4 This situation stillremains despite the industry’s fulladoption of computers for design anddrafting work. In another recent sur-vey of producers, all of the respon-dents reported using CAD drawings,with 96.3 percent produced in house,and the remainder outsourced to con-sultants).5

The Precast Concrete Software Con-sortium (PCSC) is currently specify-ing and procuring three-dimensional(3-D) automated and integrated designand management software for itsmembers.*

One primary goal of the PCSC is toenable precast producers to reducelead time on projects from months tojust one week,6 and to make produc-tion-related activities (such as pro-curement, control, and shipping) moreefficient. Another goal is to generatethe engineering information needed tosupport automation, such as bendingof reinforcement, steel cutting, andmold design.

In addition to these benefits, the au-

May-June 2003 3

thors hypothesize that 3-D computermodeling of buildings, if performedwith well-structured top-down para-metric dependencies between assem-blies, pieces and components, has thepotential to reduce or eliminate manysources of error. Case studies of fail-ures and successes provide an effectiveand convenient resource for initial ex-amination of this hypothesis; its proofwill require extensive implementationand adoption of such systems.

In this research, eight case studiesof precast/prestressed concrete pro-jects, each of which required signifi-cant remedial work, were collected,documented, and examined. The stud-ies allow qualitative tracing of the rootcauses of the errors that led to the re-work. In some cases, sufficient de-tailed cost information was providedto allow a quantitative assessment ofthe impact of the rework on projectbudgets and schedules.

In the first part of this paper, we de-scribe the integrated assembly andpiece modeling approach to computer-ized design. Next, we classify the de-sign and drafting errors reported in thecase studies. Each classification is il-lustrated with examples tracing thecause and effect of the error. Lastly,we trace the ways in which each typeof error would be prevented in such adesign environment.

INTEGRATED PARAMETRICASSEMBLY AND PIECE 3-D

MODELINGThe following two principles for

precast concrete modeling softwareare central to reducing the incidenceof errors and consequent rework:

1. Modeling versus drafting. In-stead of generating multiple and dis-crete drawings to represent a buildingand its parts, the operator builds an in-tegrated model of the building assem-bly and its components. Both assem-bly and piece drawings are generatedfrom the model. Drawings are reportsof the information, rather than con-taining the information itself.

2. Maintenance of integrity fromthe assembly to the parts, ratherthan from the parts to the assembly.Instead of composing a buildingmodel as a collection of instances oftypical pieces with fixed geometry, thegeometry of each piece is derivedfrom the spatial topological relation-ships between it, its neighbors, and thebuilding grid. In this way changes athigher levels of an assembly can bepropagated to lower level parts auto-matically.

Fig. 1 illustrates the principle: thebeam is automatically sized to fit be-tween the columns, and the corbelsupports are automatically sized to fit

Fig. 1. Parametric model of a beam between two columns.

* The PCSC is an LLC formed in 2001 by a group ofPrecast/Prestressed Producer companies in order to collectively specify and foster procurement of 3D parametric modeling software, and to develop a Precast Product Model for integration of precast engineering systems with other information systems.

4 PCI JOURNAL

the beam. Any change made to any ofthe independent dimensions (LAB, w1,w2, or c) will result in propagation ofthe change to the beam and to the sup-ports. Also, if the beam is removed,the supports are automatically re-moved (i.e., recognition of connec-tions between pieces as a separate log-ical entity is crucial to enabling thisbehavior).

The first principle was incorporatedin pioneering 3-D modeling softwarefor precast concrete design (EDGE),8

which has enabled its developers tosignificantly reduce the frequency of

errors in their projects. The secondprinciple has yet to be applied in pre-cast concrete software. It has, how-ever, been applied with success in anumber of other industries, includingstructural steel detailing.

Unlike traditional CAD files, thebehavior of pieces in a top-down para-metric building model closely mirrorsthe conceptual thinking of an engineerin executing the design of a precastconcrete building. The engineer isconcerned first with the structure as awhole assembly, then with the piecesthat make up that assembly and the

(a)

(b)

(c)

connections between them, and lastlywith the details of each individualpiece. When changes are made at anylevel, the elements at a lower levelshould adapt to the changes made.

The approach is effectively illus-trated in the sequence of capturedscreen shots shown in Fig. 2. Standardor user-defined parametric cross sec-tions are extruded to form the basicvolume of each piece (Fig. 2a shows aspandrel beam). All of the pieces areplaced in the assembly (in Fig. 2b,columns are red, spandrels are yellow,and double tees are cyan). The userdoes not define the length of the span-drel; rather, the system automaticallysets the length parametrically as thedistance between the columns.

Next, connections are modeled (seeFig. 2c). These are selected from aparametric library of connections andautomatically adapted to fit the appro-priate pieces. The resulting piecemodel can be seen in Fig. 3a, and,with embedded hardware, in Fig. 3b.If any change is made to the positionor cross section of any of the pieces inthe assembly, the software automati-cally propagates the effect of thechange to all the other pieces and con-nections, ensuring that the integrity ofthe model as a whole is maintained.

Piece prestress and reinforcementdesign will be performed directly inthe model using plug-in professionalsoftware. At any time, productiondrawings and bills of material can beautomatically generated by the system.The drawings are derived directlyfrom, and are, therefore, fully consis-tent with, the 3-D model. Any subse-quent change must be made to the 3-Dmodel to ensure that all future piecedrawings and assembly drawings willbe mutually consistent.

This corner spandrel, shown in Fig.3, was taken from an office building(not included in the case studies).Preparation of the full piece-ticketdrawing file using conventional 2-DCAD, including all dimensions and abill of materials, required more thanone week. Modeling the piece and itsimmediate neighbors, including allconnections and reinforcement in the3-D prototype software, and generat-ing the piece tickets and bill of materi-als (without annotations and exploded

Fig. 2. Top-down 3-D modeling of a precast concrete spandrel (prepared using Teklaprototype Xengineer software7).

May-June 2003 5

details) required approximately twohours.

It could be argued that top-downbuilding modeling can be done usingconventional CAD systems, even in 2-D, with sophisticated and disciplineduse of drawing layers and model/paperview separation. However, in mostsystems, individual pieces are insertedin assemblies as instances of piece pro-duction series (piece-marks) (e.g.,“blocks” in AutoCAD® or “cells” inMicrostation®). Their parameters areset at the time of insertion, and so as-sembly geometry is driven “bottom-up” from the CAD blocks.

This means that any localizedchange to one piece in a series re-quires the user to separate that piecefrom the series, create a new piece-mark, adjust the changed locations ofother dependent elements in the as-sembly, and produce a new piece-mark drawing. In contrast, none of thiseffort is necessary in systems in whichtop-down parametric dependencies aremaintained between assemblies,pieces, and components.

The PCSC has specified and ten-dered for a comprehensive 3-D andknowledge-rich software design tool,which will be integrated with other de-sign, analysis, scheduling, accounting,and production managementsoftware.6 Identified in this specifica-tion are the following priorities:

• The 3-D modeling software mustsupport a top-down design processwith the three distinct phases of as-sembly layout, assembly detailing, andpiece detailing.

• All assemblies, pieces, and con-nections must be parametrically re-lated to a building grid and to eachother, and changes must be propagatedautomatically so that integrity is main-tained.

• The 3-D computer model must bethe only repository and source for allproduct design information. Drawings,bills of material, and other documentsare to be reports of the project infor-mation, rather than separate reposito-ries of that information.

• Most of the routine layout, analy-sis, and detailing tasks are to be auto-mated.

These priorities distinguish the pro-posed system solicited by the PCSC

(a)

(b)

Fig. 3. Spandrel piece model.7

from traditional CAD drafting, ascommonly practiced, in terms of thetwo principles established above.

COMMON DESIGN ANDDRAFTING ERRORS

Seven precast/prestressed concreteprojects in this study were examined:four parking decks with structuralpieces, one office building with archi-tectural and structural pieces, one in-door arena with precast concrete rak-ers and walls, and one jail complexwith precast concrete boxed cell mod-ules (see Table 1).

The largest project had 3211 precastcomponents, covered an area of688,000 sq ft (63,915 m2), and had acontract price of $13.4 million. Thesmallest project had 259 components,covered 75,000 sq ft (6968 m2), andhad a contract price of $1.2 million. Intotal, 22 distinct and significant errorswere uncovered, each of which af-

fected the project duration and cost. Inthe most severe case, the estimatedcost of a single error amounted to 9.9percent of the contract price.

These case studies were collectedfrom companies participating in thePCSC, and were selected to reflect thevariety of design, drafting, and coordi-nation problems common in theirdaily work. Even though these compa-nies were helpful in assessing the na-ture and the effect of the errors en-countered, they provide only a rough,empirical indication of howwidespread such errors are, or howbroad their cumulative financial im-pact is on the precasting companies,their clients, and the other companiesin the construction supply chain.

The identified errors were classifiedinto the following five categories:

1. Errors in design;2. Errors of inconsistency between

assembly drawings and piece produc-tion drawings (i.e., shop tickets),

6 PCI JOURNAL

caused by errors in drafting assemblydrawings;

3. Errors of inconsistency betweenassembly drawings and piece produc-tion drawings caused by piece detail-ing errors;

Table 1. Precast concrete project case studies.

Project key Description Precast piece Floor area Other data Number of Contract value(1) (2) type (3) (sq ft) (4) (5) pieces (6) (7)A County jail 3D modular cells 688,000 Façade area 140,000 sq ft 3211 $13,400,000 B Multi-use building Structural & architectural 220,000 700 $5,000,000 C Parking structure Structural 75,000 259 $1,161,000 D Basketball arena Structural - stadium 15,000 seats 528 $4,000,000 E Parking structure Structural 95,000 332 $1,850,000 F Parking structure Structural 460,500 2400 cars 1442G Parking structure Structural 162,000 815 $1,943,000

Note: 1 sq ft = 0.0929 m2.

Fig. 4. Two double tees bearing on Spandrel SP-12.

commonly designs double tee stems tobe supported in pockets in the interiorface of the spandrels. The engineer ex-plained that, in this case, problemsarose for the following reason:

“The tee was drawn and detailed onthe shop drawings as if both stemswere to bear in 6 in. [152 mm] deeppockets in the spandrel. Due to theproduction schedule, the framing anddouble tees were drawn and checkedseveral months before the spandrelsneeded to be. However, when the ele-vations were drawn and the spandrelswere checked against the elevations,one stem of the tee was shown heldback and bearing on a haunch [seeFig. 4]. Normally, when this error isfound before erection, the stem can becut back and properly reinforced.However, in this case, the stem beingdiscussed was dapped 10 in. [254mm]. This prohibited the stem frombeing cut back, because it would bevery difficult to properly reinforce thestem after removing the dapped sec-tion. We looked at placing this tee inanother location in the building andre-pouring a corrected tee in that pro-duction slot (of the alternative tee), butto no avail. The tee had to be thrownaway and re-poured correctly. A totalcost of a 12DT28 with all the materi-als, labor, and disposal is about $750per cu yd [$981/m3]. Therefore, thistee cost us about $8000.”

Assembly Drawing Errors

Assembly drawings are the mediumthat enables design engineers to de-velop, record, and communicate theirconcept of the building as a whole.They are usually developed at the startof a project. The main purpose ofpiece drawings, on the other hand, is

Fig. 5. Precast stairwell.

4. Errors caused by lack of coordi-nation between different building sys-tems; and

5. Errors caused by inadequate man-agement of design and detailingchanges.

Other types of errors that can occuron precast projects were not includedin this investigation. For example, er-rors in bills of material (e.g., wrongquantities, wrong items specified, oritems missing) are commonly cor-rected by field personnel, and not re-ported as errors in design or drafting.The five identified error types are de-scribed below, each with an examplefrom a case study.

Design Errors

The design errors in the case studyinclude errors of judgment or detailingin the engineering design decisionsthat determine a building’s assemblydetails. Errors in structural calcula-tions, or in setting prestress or rein-forcement, are not included for thecurrent purpose. Design errors wererare in the cases reported, althoughtheir effects can be far-reaching.

Project B provides an example. Inthis five-story multi-use building, thefloors are composed of double tees,which are supported at the edge of thebuilding on spandrels. This precaster

May-June 2003 7

to define the individual pieces of abuilding for production. If disparitiesbetween the assembly drawing set andany piece drawing are introduced, it islikely that the resulting piece producedwill not function properly in the over-all structure.

This was the single most commontype of error found in this study. Thistype of error is common to all con-struction industry trade sectors inwhich parts are prefabricated off-siteaccording to custom project specifica-tions; these include structural steel,HVAC, curtain-walls, ironwork, rein-forcing bar fabrication, and others.

To illustrate the nature and the po-tential impact of inconsistencies be-tween assembly and piece productiondrawings, consider Project G, a typicalprecast concrete parking structure. Theinterior ramp spandrels in this buildingwere detailed with the batter length(the distance by which the top and bot-tom edges of a spandrel must be in-creased to account for its slope) sub-tracted instead of added. As a result,75 spandrels were cast too short, cost-ing the precaster $193,000 (9.9 per-cent of the contract value).

Another example is provided by thedesign of a geometrically complexstairwell that was built using separateprecast pieces for landings and stair

(a)

(b)

Fig. 6. Spandrels in Project F: (a) Elevation; (b) Piece detail.

Fig. 7. Spandrelconnections to walland column: asdesigned.

sections (see Fig. 5). Frustration withconventional methods led this pre-caster to pursue an ad-hoc top-downdesign of the pieces using 3-D solid-modeling software, although with no

automation. The engineer noted:“Each set of stairs connected a series

of curved landings in a triangular pat-tern. Location of doors, railings, andelectrical units in relation to the stair

8 PCI JOURNAL

locations were critical. Calculating lo-cations and dimensions manually de-voured much valuable time and createdfrustration when calculated figureswould not agree with the informationgiven. [Applying 3-D modeling]cleared up many misunderstandingsand brought everyone into agreement.It also assured us that the landings andthe stairs all fit together properly.”

Piece Detailing Errors

In current 2-D design practice, thedetailed cast-in accessories for eachcomponent piece are not shown in assembly drawings. Such details areshown in piece drawings, in which thepieces are designed and drawn for production. Improper coordination between this detail and the overall assembly can introduce inconsisten-cies between multiple assembly draw-ings and piece drawings.

A typical example of this type oferror occurred in Project F. On theeastern façade of the elevator core,spandrel beams connect the cornercolumn to the wall panels at all eightlevels of the structure (see Fig. 6a).The connections to the corner columnare designed to consist of threadedbars passed through holes through thewidth of the columns and screwed intosockets embedded in the spandrel,which sits in a recess in the column onits outer face (see Fig. 7). At their op-posite ends, the spandrels are designedto connect to the wall panels withwelded plates.

When the first spandrel was hoistedinto place, it became clear that the twoconnection hardware types had eachbeen embedded at the wrong ends ofthe piece (see Fig. 8). The boltsthrough the columns could not be an-chored, and the welds to the wallplates could not be made because therewere no connection plates embeddedin the spandrel. The detailing error canbe seen in the piece drawing in Fig. 6b.Work on the core was halted for con-sultations, as it appeared that the span-drels would have to be abandoned andthat erection would have to wait for re-placement pieces. The elevator shaftswere on the critical path of the generalcontractors’ project schedule. It wasdecided that the slab connections to thespandrels would suffice to hold thespandrels in place until new field con-nections could be designed, fabricated,and implemented. The event had anegative effect on all three key mea-sures of project success, namely, time,cost, and quality.

Similar detailing errors were re-ported in Project C:

• “One column was missing all thecorbels necessary to support an entirestack of double tee stems. (This is anunusual condition, and even though itwas clear in plan view in the layoutdrawings, the fabrication drawing de-tailer referenced the building eleva-tions where the support requirementwasn’t shown.) We had to bolt onmajor remedials which shutdown erec-tion for several days.”

• “Several ramp columns were de-

Fig. 8. Spandrel connections to wall and column: as built.

tailed too short. This wasn’t discovereduntil after most of the deck had beenerected, and necessitated $50,000 [i.e.,4.3 percent of contract value] worth ofshoring, jacking, and shimming.”

Building System Coordination Errors

This classification includes all con-flicts between precast pieces and partsof other building systems. These er-rors result from insufficient coordina-tion between different system designs.They are common and insidious, and,as with the other error types reported,they are often not discovered until thetime of erection.

Project A, a large prison (see Fig. 9),suffered over $500,000 (3.7 percent ofcontract value) in overruns as a directresult of lack of coordination betweenthe precast concrete cell modules andvarious cast-in-place, mechanical,plumbing, and architectural systems.The project manager reported:

“One of the biggest problems of allwas the coordination of openings forducts, vents, draws, sprinklers, andother utilities. Usually when an open-ing was added or changed, it affectedmany other adjacent modules.”

A unique problem in this case wascoordination of steel anchors embed-ded in the exterior walls of each prisoncell module for the sliding cell doors.The doors of all the cells along a roware connected together with a mecha-nism that allows automatic opening ofall the doors in an emergency. As de-sign of the mechanism developed, theanchor positions and sizes had to bepropagated to each cell modulethroughout the building. This se-quence is complex to monitor becausethe door mechanisms are designed atthe assembly level, but the modularcells are drawn on separate piece-markdrawings.

Errors Resulting fromDesign Changes

Changes in architectural designs orother building systems require that pre-cast designs be updated to match. Thedifficulty is exacerbated by the rela-tively long duration of design detailingin most precast projects; late changesmust be coordinated through assembly

May-June 2003 9

drawings and a complete set of piecedrawings. Nevertheless, owners and ar-chitects expect the precaster to respondquickly to changes submitted beforephysical production of each piece. In-adequate management of thosechanges often results in significant re-work on the site, and, in certain cases,the need to replace incorrect pieces.

In the case of Project D, the indoorarena, precast concrete raker beamsand walls were supplied to rest on acast-in-place substructure. Many elec-trical and railing embeds were re-quired. An architect’s design change,which occurred relatively late in theproject, aggravated the task of coordi-nation. Precast erection, scheduled for16 weeks, was extended by more thanone month for correction of railingposts, lighting fixtures, and other em-beds and holes, all at the precaster’sexpense. Fig. 10 shows a situation inwhich the location of a 6 x 6 in. (152 x152 mm) hole was changed in a cast-in-place wall, but not updated in theproduction drawing of the adjacentprecast piece. The impact is not onlyin the cost of rework and scheduledelay, but also in the quality of the fin-ished product and damage to the pre-caster’s reputation.

Project A, the prison project pro-vides an additional example: “Holesfor the shear pins (in the bottom of themodule walls) were field drilled in thecast-in-place slab on grade. Due tochanges in the modules used to makeup the space, the location of these keptchanging daily; some module loca-tions had three sets of holes by thetime the module arrived.”

THE POTENTIAL FORELIMINATION OF ERRORSThe following features of a knowl-

edge-rich integrated assembly and part3-D modeling system can contribute toeliminating or reducing errors:

1. The logical relationships betweenconnections and pieces are embeddedwithin the system. This feature com-bined with the parametric behavior ofthe assembly and of the pieces meansthat the spatial integrity of the 3-Dmodel is maintained as changes aremade without the need for the user topropagate the changes.

2. The 3-D model is the singlesource for all the product information;2-D drawings are generated as reportsfrom the 3-D model information. Thissingle-source concept means that in-consistencies cannot occur. Errors ofcoordination between assembly draw-ings and piece drawings are essen-tially eliminated.

3. Automated detailing, such asplacing connection hardware and mak-ing all the necessary geometricaladaptations to the connected pieces,

removes further opportunities forhuman error. Even in unique designsituations, where automated detailingcannot be applied and the detailingmust be done manually, the detailingis done in the context of all otherpieces so that the probability of mak-ing an error is significantly reduced.

4. Any building system that impactson the precast pieces can be importedor directly modeled in three dimen-sions (possibly requiring significantadditional 3-D modeling layout work).

Fig. 9. Prison construction using precast modular cell units.

Fig. 10. Misaligned 6 x 6 in. (150 x 150 mm) holes between a precast piece and acast-in-place wall.

10 PCI JOURNAL

If these are updated over time to re-flect all changes, then any piece draw-ings produced will correctly show anyholes that are required, and any auto-mated detailing procedures can ac-count for the building systems’ com-ponents. An associated benefit is thatthe lead time required to producepiece drawings is reduced frommonths to days, so that changes can beaccommodated much later in the pro-cess than is currently possible.

5. The 3-D building model providesa platform for automated designchecking routines. For example,pieces without adequate connections,spatial conflicts, and other errors canbe automatically identified and re-ported to the user.

Case Study Errors Re-examined

The errors encountered in the casestudies were each re-examined in lightof the 3-D modeling system as specifiedby the PCSC. For example, the mir-rored spandrel connections in Project Fcould not have occurred if the shopdrawings had been produced automati-cally from a 3-D model. This is becausethe stage of transfer from assembly topiece drawings, currently performed by

a human operator, is eliminated entirelyin the proposed software paradigm(Items 1, 2, and 3). The level of confi-dence is considered “complete.”

Errors related to coordination betweenbuilding systems, such as in Project A,are dealt with as described in Item 4.However, these can only be consideredeliminated with “medium” confidence –that is, human error can still be intro-duced if the various building systems arenot updated in the 3-D model.

Table 2 summarizes this analysis forall the cases. It lists the features of sucha system that would apply in each caseand provides an assessment of the like-lihood that each would be eliminated.Of the 22 errors listed, all are consid-ered eliminated with at least mediumconfidence, 16 (73 percent) with atleast high confidence, and 11 (50 per-cent) with complete confidence.

CONCLUSIONSModeling a building in a computer,

rather than drafting multiple represen-tations of it and of its parts in draw-ings (whether CAD or manual), holdsthe potential to reduce the occurrenceof errors and the need for rework inconstruction projects. Employing 3-D

CAD is necessary, but not sufficient:the building model must be developedin an integrated parametric fashion,must be comprehensive, must cover asmuch of the project scope as possible,and must drive the production of alldrawings and reports, if the benefitsare to be fully realized.

The seven case studies of precastconcrete construction projects showthat errors related to design and draft-ing occur despite the use of 2-D CADtechnology. It seems reasonable to as-sume that many more errors occurred,which were intercepted and corrected;the case studies do not allow estima-tion of their frequency or severity. Allof the errors reported in these cases re-sulted in the precast concrete produc-ers losing time and money.

The PCSC has specified, and is cur-rently procuring, 3-D modeling andknowledge-rich software for precastassembly design and detailed engi-neering. The authors expect that intro-duction and use of such software willeliminate multiple types of errors thatare common today. Additional bene-fits will accrue from the drastic reduc-tion in time required to produce bothassembly and piece drawings.

The PCSC has also begun develop-

Table 2. Probability of elimination of case study errors.

Error System Confidence Error Project Classification Feature Level

Holes for shear pins drilled three times over A Change management 1, 4 CompleteCould not maintain integrity through changes C Change management 2 Complete

Holes did not align on different pieces D Change management 1, 4 CompleteArchitectural changes not dealt with correctly D Change management 1,2 High

Sliding door mechanism alignment A Systems coordination 2, 4 HighHoles for ducts, draws, vents, sprinklers A Systems coordination 4 Medium

Railing and electrical embeds not detailed correctly D Systems coordination 4 MediumLighting and railing embeds not coordinated D Systems coordination 4 MediumInverted T detailed too long for CIP support E Systems coordination 1, 4 High

3-D design for placement of electrical, doors, railings F Systems coordination 1, 4 High/MediumDesigned pocket instead of haunch for double tee at end of spandrel B Design error 1, 3 High

Angles on slabs and panels incorrect A Drafting error 1, 2 CompleteRamp columns too short (shored and jacked) C Drafting error 1, 2 CompleteHorizontal block-outs for sloped spandrels E Drafting error 1, 3 Complete

Incorrect length for batter on sloped spandrels G Drafting error 1, 2, 3 CompleteVarying wall thicknesses due to varying modules and triangular building A Drafting errors 1, 2 Complete

Column detailed without double tee corbels C Piece detailing 5 CompleteNo end finish detailed for strands in spandrels E Piece detailing 5 MediumSpandrels with wrong connections (mirrored) F Piece detailing 1, 2, 3 Complete

Incorrect thickness for ledger beam bearing pads F Piece detailing 1, 3 HighSpandrel and double tee detailed with different connections B Piece detailing 1, 2, 3 Complete

Block-outs missing from inverted tee beam supported on CIP concrete EPiece detailing;

1, 3, 4, 5 MediumSystems coordination

May-June 2003 11

ment of a precast data model, whichwill enable integration of all of the in-formation technologies throughout theprecast construction business process.(For background information on thesubject of Building Product Modeling,see Reference 9). This will extend thebenefits gained in adopting 3-D com-puter modeling by allowing immediatecommunication of engineering changes

to scheduling, procurement and othernon-engineering activities, further re-ducing errors, and improving the over-all management of changes as they in-evitably occur.

ACKNOWLEDGMENTThis work was funded in part by the

Precast Concrete Software Consortium,

a consortium of major precast/pre-stressed concrete producers fromCanada, the United States, and Mexico(http://dcom.arch.gatech.edu/pci). Thecontribution of detailed case studies byconsortium member companies isgratefully acknowledged.

The authors wish to thank the PCIJOURNAL reviewers for their helpfulinput on this paper.

REFERENCES

1. Love, P. E. D., Mandal, P., Smith, J., and Li, H., “Modelingthe Dynamics of Design Error Induced Rework in Construc-tion Projects,” Construction Management and EconomicsRoutledge, London, V. 18, No. 5, 2000, pp. 567-574.

2. Josephson, P. E., and Hammarlund, Y., “The Causes and Costsof Defects in Construction: A Study of Seven Building Pro-jects,” Automation in Construction, Elsevier, V. 8, 1999, pp.681-687.

3. Navon, R., Shapira, A., and Shechori, Y., “Automated RebarConstructibility Diagnosis,” ASCE Journal of ConstructionEngineering and Management, V. 126, No. 5, September-Oc-tober 2000, pp. 389-397.

4. Arditi, D., Ergin, U., and Gunhan, S., “Factors Affecting theUse of Precast Concrete Systems,” ASCE Journal of Architec-

tural Engineering, V. 6, No. 3, 2000, pp. 79-86.5. “PCI and CPCI Software Survey,” Precast/Prestressed Con-

crete Institute, Chicago, IL, 2001.6. Eastman, C. M., Sacks, R., and Lee, G., “Software Specifica-

tion for a Precast Concrete Design and Engineering SoftwarePlatform,” PCSC Research Report, Georgia Institute of Tech-nology, Atlanta, GA, 2001.

7. Tekla 2002. Tekla Xengineer 1.1, www.tekla.com, Espoo, Fin-land.

8. EDGE 2002, PTAC Consulting Engineers, Pensacola, FL,2002, http://www.ptac.com/software/index.html.

9. Eastman, C. M., Building Product Models, CRC Press, BocaRaton, FL, 1999.