distribution restriction statement - dtic · for the design of civil works projects. 3. references...

DEPARTMENT OF THE ARMYU.S. Army Corps of Engineers

CECW-ED Washington, DC 20314-1000 ETL 1110-2-536

Technical LetterNo. 1110-2-536 31 December 1994

Engineering and DesignNONLINEAR INCREMENTAL STRUCTURAL ANALYSIS

OF ZINTEL CANYON DAM

Distribution Restriction Statement

Approved for public release; distribution is unlimited.

Report Documentation Page

Report Date 31 Dec 1994

Report Type N/A

Dates Covered (from... to) -

Title and Subtitle Engineering and Design: Nonlinear Incremental StructuralAnalysis of Zintel Canyon Dam

Contract Number

Grant Number

Program Element Number

Author(s) Project Number

Task Number

Work Unit Number

Performing Organization Name(s) and Address(es) Department of the Army U.S. Army Corps of EngineersWashington, DC 20314-1000

Performing Organization Report Number

Sponsoring/Monitoring Agency Name(s) and Address(es)

Sponsor/Monitor’s Acronym(s)

Sponsor/Monitor’s Report Number(s)

Distribution/Availability Statement Approved for public release, distribution unlimited

Supplementary Notes

Abstract

Subject Terms

Report Classification unclassified

Classification of this page unclassified

Classification of Abstract unclassified

Limitation of Abstract UU

Number of Pages 67

DEPARTMENT OF THE ARMY ETL 1110-2-536U.S. Army Corps of Engineers

CECW-ED Washington, DC 20314-1000

Technical LetterNo. 1110-2-536 31 December 1994

Engineering and DesignNONLINEAR INCREMENTAL STRUCTURAL ANALYSIS

OF ZINTEL CANYON DAM

1. Purpose

This engineer technical letter (ETL) provides anexample of the use of nonlinear incremental structuralanalysis (NISA) for a massive concrete structure(MCS).

2. Applicability

This ETL applies to all HQUSACE elements, majorsubordinate commands, districts, laboratories, andfield operating activities (FOA) having responsibilitiesfor the design of civil works projects.

3. References

a. EM 1110-2-2006, Roller CompactedConcrete.

b. EM 1110-2-2200, Gravity Dam Design.

c. ETL 1110-2-365, Nonlinear IncrementalStructural Analysis of Massive Concrete Structures.

4. Discussion

a. Background.NISA is a developing technol-ogy which has been used on several recent civil

works projects and valuable experience is gained witheach new usage of the method. NISA is used primar-ily as a supplemental tool for the design of MCS inorder to: improve the cracking performance of aparticular type of structure which has exhibited unac-ceptable cracking in the past; to more accuratelypredict the structural behavior of an unprecedentedstructure, or to develop more cost effective structuresby revising the geometric configuration, materials orconstruction parameters.

b. Application to Zintel Canyon Dam.Appen-dix A presents a study conducted of the Zintel Can-yon Dam in the Walla Walla District using the NISAmethod of analysis. The dam was constructed usingRCC and the NISA study was conducted by CENPWafter completion of construction. The study providesinsight into the use of NISA as an analytical tool fordesigners.

5. Action

The enclosed study demonstrates the proceduresnecessary to conduct a NISA analysis of a RCCstructure. NISA should be performed according tothe requirements of ETL 1110-2-365.

FOR THE DIRECTOR OF CIVIL WORKS:

PAUL D. BARBER, P.E.Chief, Engineering DivisionDirectorate of Civil Works

ETL 1110-2-53631 Dec 94

APPENDIX A: NONLINEAR INCREMENTAL STRUCTURALANALYSIS OF ZINTEL CANYON DAM

A-1. Introduction

a. Purpose.

(1) The U.S. Army Corps of Engineers, CivilWorks, has been developing guidelines and proce-dures for determining thermally generated stressesresulting in construction of massive concrete struc-tures. The original product of the research and devel-opment was documented in ETL 1110-2-324, SpecialDesign Provisions for Massive Concrete Structuresand has since been superseded by ETL 1110-2-365,Nonlinear Incremental Structural Analysis of MassiveConcrete Structures.

(2) The Walla Walla District was commissionedto perform a thermal stress analysis of Zintel CanyonDam. The goals in performing this study are to:

(a) Perform a nonlinear, incremental, structuralanalysis (NISA) to evaluate the effects of tempera-ture, Roller Compacted Concrete (RCC) materialproperties, and the subsequent volume changes on thecracking potential of Zintel Canyon Dam. The pur-pose of such an analysis is to evaluate costs andperformance so that appropriate design features andrequirements may be established. Obviously, sincethe dam is complete, no design modifications will bedone. However, some observations of the effective-ness of such an evaluation can be made.

(b) Demonstrate the implementation of the NISAprocess for an RCC structure and compare analyticalperformance with observed performance of the struc-ture. In the process, evaluate the NISA method toprovide recommendations on what measures may beimplemented to make the process more serviceable.By performing a NISA of a current project at theDistrict level, valuable insight would be gained onwhether this analytical method is a suitable tool fordesigners and implementable at the district level.

(c) Evaluate the suitability of the Corps of Engi-neers’ guidance in performing nonlinear, incrementalstructural analyses for Zintel Canyon Dam. At thetime the NISA was performed, the guidance wascontained in ETL 1110-2-324. A new document,ETL 1110-2-265, has been published, but most of theprocedures for performing a NISA remain the same.

b. Scope. This work is limited to the NISAevaluation of Zintel Canyon Dam. The analysesperformed for the project are not intended to be com-prehensive evaluations as described in the ETL, butabbreviated evaluations, more appropriate for a struc-ture of this type and function. There are two reasonsfor this abbreviation. The structure is a simplegravity design, containing no contraction joints andimpounding no permanent reservoir. Since transversecracking of the structure poses no threat to the safetyof the structure or to the public and does not impactthe function of the project, a relatively simple analy-sis is sufficient. Secondly, the level of funding forthis study was not sufficient to perform extensiveanalyses. Funding only permitted simple modelingand limited evaluations. However, the extent ofanalysis was sufficient to evaluate the thermal stressperformance, to provide recommendations on districtimplementation of the NISA, and to evaluate theguidelines specified in the ETL.

c. Report. This report is organized to specifi-cally address the study goals. Section 2 providesbackground on the development of finite elementanalyses for mass concrete structures which led to thedevelopment of the current procedures for performingthermal stress evaluations and Corps guidance on thesubject. The NISA evaluation for Zintel CanyonDam, which includes the details of performing theevaluation and the project specific results, conclu-sions, and recommendations, is contained in Section 3and the referenced appendices. Recommendations areprovided, in Section 4, on implementing a NISAevaluation. Finally, recommendations for modifica-tion of the guidance, to make it more useful in appli-cation of the NISA process, are contained inSection 5.

A-2. Background

a. Development of the NISA process.

(1) Mass concrete structures are different frommany concrete structures in that material propertieshave a significant effect on the state of stress in thestructure. These material properties are not onlydependent on the type and quantity of material, buton the age of the concrete, temperature of the

A-1

ETL 1110-2-53631 Dec 94

concrete, and the state of stress of the material.Further, certain material properties exhibit nonlinearperformance and somewhat unusual behavior at earlyages. Consequently, definitive analyses of massconcrete structures require a very complex analysisprocedure requiring the definition of many variables.

(2) A further complicating factor is that mostmass concrete structures, such as dams, are con-structed over a long period of time. Consequently,thermal stresses develop during the construction phaseand may be significantly affected by subsequentconstruction activities.

(3) The evolution of an analysis package toadequately model these variables has been a long andtedious process. One of the earliest attempts todevelop an analysis system was in 1966 during thedesign of Dworshak Dam. A finite element analysispackage was developed, under contract, by research-ers at the University of California, Berkeley. Thissystem was likely one of the first such analysis toolsever developed. Over the years, the finite elementanalytical process has progressed to the point wheremany commercial vendors have provided quitesophisticated analytical tools for the evaluation ofstress and strain in concrete structures. Unfortu-nately, most of these systems are general purposecomputer codes that do not address the specific issuesof the time-dependent behavior of mass concrete.

(4) More recently, the Corps of Engineers hasinitiated the development of supplemental codes andtechniques to enhance and refine the analytical pro-cess. The Corps selected for general use, ABAQUS,a general purpose finite element code. ANATECHResearch Corporation, under contract to the Water-ways Experiment Station (WES) developed softwaresubroutines to be used with the ABAQUS generalpurpose finite element code. These subroutines weredesigned to allow the user to input accurate, timedependent and cracking material properties of con-crete into the ABAQUS model.

(5) ETL 1110-2-324 dated 30 March 1990, “Spe-cial Design Provisions for Massive Concrete Struc-tures,” was published, providing policy guidance todesigners for execution of a NISA for the design ofmass concrete structures.

b. Applicability of ETL-1110-2-324. The guid-ance provided by this ETL has been followed in theanalysis and design of several structures. Most

notable was the analysis of Lock and Dam 26 nearSt. Louis, Missouri, and the current design ofOlmsted Locks and Dam on the Ohio River. Severalconcerns have been raised regarding the implementa-tion of the ETL. The analytical process, as outlinedin the ETL, is extremely comprehensive and expen-sive. Certainly, not all Massive Concrete Structuresrequire a NISA to be performed. The results of thisstudy can be used to determine the applicability ofthe current ETL guidance for future projects that mayor may not require a NISA be performed.

A-3. Zintel Canyon Dam Thermal CrackingEvaluation

a. General. The purpose of performing a ther-mal cracking evaluation for Zintel Canyon Dam wasto determine the consequent cracking of the structureresulting from thermally generated volume changes.Since the evaluation was performed some time afterthe construction of the dam, actual conditions, such asambient temperatures, RCC placing temperatures, andplacing schedules were used in the model. However,additional laboratory work to characterize RCC mate-rials was not done because of limited funds. Thisincluded creep properties, thermal properties, andtensile strain capacity. Instead, these material proper-ties were estimated. In addition, the limited access tocomputer resources necessary to perform the analysesfurther limited the depth of the investigation.

b. Project description.

(1) Zintel Canyon Dam is a straight axis con-crete gravity structure. The length of the structure is520 ft across at the crest and 126 ft above the foun-dation at the deepest point. The structure is con-structed of 70,600 cu yd of RCC. The outflowspillway is 160 wide with a crest 16 ft below the topof the dam. The spillway flows are contained bycast-in-place concrete training walls anchored to theRCC mass and RCC gravity training walls borderingthe stilling basin. The dam, stilling basin, and stillingbasin training walls are founded on basalt rock (seeappendix C for a more complete project description).

(2) The project provides flood protection to thecity of Kennewick, Washington. It is located onZintel Canyon, a 19-square mile water course whichthreatens the city with winter snow melt and summerthunderstorm events. The water course is otherwise a

A-2

ETL 1110-2-53631 Dec 94

dry streambed. The structure will impound the100-year flood for no more than 20 days. A self-regulating outlet provides reservoir drawdown at acontrolled rate. The structure is designed to requireno manned operations in the event of a flood.

(3) After final excavation of the foundation, therock surfaces were cleaned and covered with a wet-mix shotcrete and foundation concrete. RCC wasplaced in 12-in. thick horizontal layers on and againstthese foundation surfaces. Interfaces of the RCC andthe foundation concrete were bonded with a beddingmortar. Similarly, the RCC lift joints were fullybonded with the same bedding mortar. RCC place-ment began on 6 July 1992 and was completed on15 October 1992. The 126 RCC lifts were placed inapproximately 75 placing days during a 100-dayperiod. In general, the process was to deliver theRCC to the dam on a conveyor. Front-end loadersreceived the RCC and transported it to the desiredlocation. The RCC was spread with a small dozerand compacted with a vibratory roller.

c. ABAQUS model and data.

(1) After developing the mesh for a two-dimen-sional transverse model through the spillway section,it became apparent that a three-dimensional modelwould become extremely large. Although the CRAYcould handle the computational analysis during thisstudy, the time required to execute such a modelwould have been extensive, due to the system work-load. In addition, the geometry of the dam does notlend itself to easy input generation for ABAQUS andas a result, extensive efforts to generate the modelwould be required. A three-dimensional model wouldincrease the cost of the study significantly beyondinitial estimates. Because of these factors, the studywas limited to two two-dimensional models. Thermalanalyses of this nature have most often been doneusing two-dimensional models. A two-dimensionaltransverse model usually gives good analytical resultsthat can be used to predict cracks that propagateinward from the surface and cracks originating fromthe foundation. However, recent projects, most nota-bly Upper Stillwater Dam, experienced significantcracking. These cracks propagate vertically from thefoundation and are oriented perpendicular to thelongitudinal axis. The Bureau of Reclamation per-formed a thermal analysis to predict the crackpotential. They utilized a transverse, two-dimensionalmodel and a longitudinal model of a horizontal plane.

Because of the magnitude of the cracks at thefoundation/RCC interface, it was necessary to deter-mine the suitability of a longitudinal model of avertical plane for use in predicting crack potential.As a result, two 2-dimensional models were analyzed.One model computed stresses resulting from a trans-verse model and the other computed stresses from alongitudinal model in a vertical plane. While thesetwo-dimensional models provide less accurate solu-tions than a 3-dimensional model, the accuracy isappropriate considering the size of the project andavailable funding for the study. The limitation of thetwo-dimensional, longitudinal model, is that it isassumed that symmetry exists on either side of thelongitudinal plane, which is not the case. Thisassumption may cause a shifting of the thermal gradi-ent from its actual location. In the case of the trans-verse model, the full section is modeled which moreaccurately models the thermal gradient. Appendix A,Figure A-1, shows the location of the assumed two-dimensional plane through the dam. Thermal con-tours of both models, shown in Appendix A,demonstrate a good correlation of peak temperaturesbetween the two models.

(2) Because RCC exhibits similar material char-acteristics as conventionally placed mass concrete, theanalytical process for determining thermal gradientsand stresses in RCC is practically the same. Theanalytical procedures have been well documented byprevious analyses and authors and should follow thegeneral guidance established in ETL 1110-2-324.However, RCC construction generally occurs over arelatively short time frame with numerous lift joints(usually 1 to 2 ft in height). Conventionally placedmass concrete usually involves placements with liftheights of 4 to 7 ft with 5- to 7-day restrictionsplaced on form removal. The exposure and depth ofeach lift in conventionally placed concrete will gener-ally define the limits of the number of steps neces-sary to perform the incremental analysis. In contrast,continuous placement of RCC on some projects hasachieved four 1-ft lift heights in 24 hr. In a 7-dayperiod, RCC can achieve changes in elevation of 25to 30 ft, depending on the project specifics and resul-tant production rates. Therefore, before a NISA canbe performed for RCC, a comprehensive study ofproduction rates must be completed in order to selecttime steps, the number of steps, and element meshsize. The results of the production rates, in combina-tion with the element size, and parametric studies

A-3

ETL 1110-2-53631 Dec 94

described in the following paragraph, will define theelement mesh.

(3) The mesh sizes for the two models wereestablished by the equation provided in the ABAQUSuser manual and restated in ETL 1110-2-234. Resultsfrom this equation were found to be very restrictiveon the element size. Instead, a simple parametricstudy was performed to determine if a larger elementsize could be used without creating numerical insta-bilities in the thermal model. The results of the para-metric study indicated that the maximum length ofthe element in the direction of heat flow could be48 in. For both models, 48 in. was the maximumsize of element used in any direction for a 6-hr timeinterval. A 6-hr time interval was chosen based onproduction rate of RCC and because it satisfied themaximum time interval required to compute earlyheat gain in the concrete. Results of the parametricstudy are shown in Appendix B. Mesh size was thencorrelated with the RCC placement schedule of thedam and was designed to capture heat gains at theearly ages of construction. Appendix A, Figure A-1represents a time history for construction and theanalysis for the transverse and longitudinal modelrespectively, as well as the initial conditions, bound-ary conditions, and input values used. Boundaryconditions include the insulating effects of upstreamprecast facing panels, free surface convection, soil(rock) conditions, downstream stilling basin slab,average daily temperatures for preconstruction, duringconstruction, and postconstruction. Initial conditionsinclude RCC placement temperatures, and initialfoundation temperatures. The user subroutineDFLUX was used, in conjunction with ABAQUS, togenerate time-dependent heat fluxes for the thermalanalysis. Parameters used in DFLUX included adia-batic heat gain (time and temperatures) for the RCCmix as well as initial placement times for each lift.Adiabatic heat gain curve is plotted in Figure A-1.Results of the thermal analysis for both the transverseand longitudinal model are represented in contourplots in Appendix A, Figures A-2 to A-11 and timehistory plots of maximum nodal temperatures inAppendix A, Figures A-12 and A-13. The maximumtemperature reported in the transverse model isrepresented by node 2330, Figure A-12c, and in thelongitudinal model by node 3213, Figure A-13c.Maximum temperatures and temperature differentialcorrelate well with predicted temperatures calculatedby using approximate computational methods forZintel Canyon Dam. As stated previously, effects ofmaximum heat gain between the two models were

nearly the same. However, there is a notable differ-ence in the rate of coding. The two-dimensionalelements do not have the capability to conduct heat inthe out-of-plane direction. With convection beingmodeled along the top surface only, of the longitudi-nal model, the elements sustain a higher thermalgradient over a much longer period of time.

(4) Discussion of stresses for both models shouldbe limited to principle tensile stresses. However,software developed for plotting stress histories is onlycapable of plotting stress in orthogonal directions andfor shear stresses. Because Zintel Canyon Dam nor-mally has no pool and experiences a very shortduration reservoir impoundment, cracking posed noconcerns related to seepage. Of concern are theorientation of cracks that will compromise the stabil-ity of the structure. A feature of the ABAQUS-basedNISA that sets it apart from others is that it allowsmaterial properties and relationships to be user-defined. UMAT is the subroutine that provides atime-dependent cracking material model. The subrou-tine allows input of specific material properties andcalibration of the model-predicted properties againstactual observed material performance. While somematerial testing had been done for Zintel CanyonDam, extensive evaluation of time-dependent proper-ties and creep performance had not been determined.Consequently, calibration of the UMAT materialmodel could not be performed. The material modelgenerated for Olmsted Locks and Dam was usedexcept that the material constants were replaced withactual or estimated values for Zintel Canyon DamRCC materials. This means that the UMAT predictedperformance for Zintel Canyon Dam was based onOlmsted material relationships. No data were avail-able to shift the relationship curves. Without actualdata to calibrate the material model, changes wouldbe arbitrary and not necessarily an improvement overusing the Olmsted data. To perform a reliable NISA,these material properties need better definition. Morecomplete definitions of the development of the modu-lus of elasticity and creep are critical to accurateresults. For smaller scale projects, standard relation-ships for a range of materials should be developed sothat the analyst can select the performance relation-ship that most likely models the materials being eval-uated. Only the larger projects will have the fundingto perform a complete battery of laboratory evalua-tions. For the longitudinal model, the maximumprinciple stress occurred at the foundation/RCC inter-face at element 1796 and are presented by stresscontour plots in Appendix A, Figures A-14 to A-19

A-4

ETL 1110-2-53631 Dec 94

and A-20 to A-25. ABAQUS calculated a maximumprinciple stress of 715 psi which occurs during thecooling period when maximum temperature differen-tials occur. Earlier testing of the RCC mix indicateda 28-day tensile strength capacity of 200 psi. The715-psi principle stress calculated is far in excess ofthe tested direct tensile capacity. This may be attrib-uted to the fact that the analyses were performedusing the original version of UMAT. That versioncontained inconsistencies which resulted in the crack-ing threshold to be computed in an unconservativemanner. This inconsistency has been corrected in thecurrent version of UMAT which was not availablewhen this study began. However, newer versions ofsoftware available at WES have the capability topredict and plot direction and magnitude of cracks.Observation of the stress contours for the longitudinalmodel indicate higher stresses occur at the foundationinterface and in the upper reaches of the abutmentand at the spillway. This can be attributed to thetemperature differential occurring between theexterior and interior elements and the rigidity of thefoundation. At the time of year for postconstructioncooling, the average ambient temperature is decreas-ing causing a larger temperature differential fromsurface elements to interior elements. Time historystress plots of various points of high stressesobserved in the contour plots are presented in Appen-dix A, Figures A-26 to A-32 and A-33 to A-40. Thestress time histories presented in Figures A10.4 andA10.6 indicate that some cracking may haveoccurred. Jumps in stress, as indicated in these plots,typically do not occur unless a crack has formed.Discussion of these results are presented below.Likewise, the transverse model predicted stresses thatare higher than the limited cracking stress. The high-est stress occurred at element 217 at the foundation/RCC interface. In addition, a region of high stressoccurs along the downstream exposed face of thedam.

d. Discussion.

(1) A simple thermal analysis was done for theproject during the design phase of the project. Thisanalysis indicated that during the normal summerweather conditions at the site, the structure may crackat three locations. Two-crack locations wereestimated to be located where the foundation changesfrom a horizontal surface at elevation 635 to thesloped abutments. The third crack was speculated tobe in the center of the spillway. None of thesecracks pose a threat to the stability of the structure

since the orientation will be in the traditionalupstream-downstream direction. In addition, since thedam is almost continuously dry, and channel flowscarry a phenomenal silt load, this cracking of thestructure is not of great concern. No additionalexpense was warranted in lessening the crackingpotential by reducing the placing temperatures or byinstalling transverse joints.

(2) Postconstruction inspections revealed onecrack in the structure located high on the left abut-ment as a result of a slope change in the foundation.No other cracks are apparent. Less cracking has beenobserved because the restraint provided by the foun-dation is probably lower than full restraint assumedby the simple analysis.

(3) Examination of the temperature history plotsindicates that the longitudinal model cools at a muchslower rate. The benefits of performing a longitudi-nal model are significant when time and costs are ofconcern for smaller projects, or projects of this typewhere certain cracking will not adversely effect theperformance of the structure. However, in this case,the thermal gradients of the longitudinal model shouldbe calibrated with the more accurate transverse modelto produce nearly the same rate of cooling. This maybecome a significant factor in the analysis because:

• While cooling is at a slower rate than mightbe expected, this will cause volume changesand hence the maximum stresses to occur ata later age in the model. Since the higherstresses occur later in time, the aging modu-lus will be higher. Since the criteria forcracking is partially based on stress, thepotential for cracking will be unconservativefor this case.

• The thermal stresses are being applied at alater age; therefore, the effect of creep willplay a less significant role in stress relax-ation. This may be conservative; however,as long as a large amount of effort has beenexpended to accomplish a material investiga-tion, it would be prudent to spend the sameamount of effort to calibrate the longitudinalmodel. This will ensure that the analysis willincorporate the more accurate creep data inthe earlier time steps when cooling would beexpected to cause higher stresses. Hence, theeffects of stress relaxation due to creep willbe incorporated into the analysis.

A-5

ETL 1110-2-53631 Dec 94

(4) Calibrating the longitudinal model may bedone by adjusting the thermal conductivity of differ-ent element sets, allowing for higher conductivity inthe earlier time steps and reducing the conductivity inthe later time steps where the convective surfaceplays a more significant role in cooling.

(5) Examination of the stress history plots indi-cates stresses are still increasing as a result ofdecreasing ambient air temperatures. For both mod-els the cooling period should, at a minimum, beapplied until thermal stresses begin to decrease andattain a steady state. For this analysis the coolingperiod was 9 months for the transverse model and3 months for the longitudinal model. Admittedly, thisanalysis fell short of predicting maximum stresses.However, review of the principle stress contours fromthe longitudinal model reveals high stress areas wherecracking is likely to occur. One area is at the centerof the spillway, the second, at the intersection of thespillway and non-overflow section, and the third, inthe upper reaches of the abutments. With the versionof UMAT used for this study, it is difficult to predictwhere cracking will initiate first. However, wherelocations of high stresses occur, the model may bemodified to depict the location of transverse joints ifcracking is undesirable in those regions. Other condi-tions that affect the stress in the dam are the assumedfoundation restraints. Both models include fullyrestrained boundary conditions at the RCC/foundationinterface. By visual observations of the rock andpostconstruction coring of the foundation, theassumed restraint conditions used in the model couldbe modified to provide a more flexible restraint con-dition or an adjusted foundation modulus. Beforeproceeding with any further analysis, the most recentversion of UMAT should be used to include theredistribution of stress that occurs after cracking. Forthis study, no further calibration of the longitudinalthermal model was completed. This is mainly due tolimited scope of the study and the fact that crackingis not of great concern for Zintel Canyon Dam.

e. Conclusions.

(1) Performance of the analysis leads to severalconclusions and recommendations for subsequentsteps to proceed with further evaluation of cracking.The first step would be to make the required adjust-ments in the foundation modulus and/or the restraintconditions. The model for Zintel Canyon Damincluded the foundation in the stress analysis.Adjusting the foundation modulus to tested values

would be more appropriate in this case. If the foun-dation conditions are modeled by the use of springs,the degree of restraint may be adjusted by softeningor stiffening the spring constants. In the case ofZintel Canyon Dam, the foundation rock was highlyfractured. Therefore, the degree of restraint providedby the concrete to rock interface may be less thanfully restrained. However, unless sufficient datasupport softening the spring constants, the results ofthe stress analysis would be unconservative in thiscase. Secondly, the model can be modified by insert-ing a transverse joint at the mid-point of the spillwayor at the corners of the spillway/non-overflow inter-section. Additional measures could include reducingthe RCC placement temperature and mandating place-ment schedules to avoid hot seasons.

(2) For Zintel Canyon Dam, considering thefrequency of reservoir impoundment and the functionof the structure, the observed cracking is well withinacceptable levels for the project. There would be nobenefit for implementing these measures.

(3) The results of this NISA analysis were, ingeneral, consistent with the results of the approximatethermal analysis performed during design of the pro-ject. Both indicate that cracking may occur in threeareas. Based on that comparison and the observedperformance of Zintel Canyon Dam and other RCCdams, the NISA for Zintel Canyon Dam provided noadditional information to attain the desired objectivesstated in the ETL. This statement is based on thefact that:

• Some cracking would be acceptable as longas structural stability is not compromised.

• Joints for control of cracking is not necessaryfor serviceability conditions since there is nopermanent reservoir.

• Because of the combinations of the above,real cost savings have been achieved bymaximizing production rates for placement ofRCC.

• Unusual loadings, extreme loadings, orsevere operational conditions do not exist.

(4) This is not to say that a NISA should not beperformed for RCC structures. Each structure issubject to unique conditions and loadings resulting inunique structural features. These factors are

A-6

ETL 1110-2-53631 Dec 94

evaluated and developed by a team of responsibleengineers who must determine the level of analysesand consultation necessary to obtain any of thedesired objectives stated in the ETL. The ETL can-not provide guidance for all possibilities.

A-4. NISA Application Recommendations

a. Applications.

(1) This work would not have progressed with-out the aid of several WES personnel. These individ-uals, currently running ABAQUS problems on theCRAY, provided a significant amount of consultation.This type of aid will likely be necessary should dis-trict users begin to utilize this software andequipment.

(2) This evaluation was performed over alengthy period of time. The time period was muchlonger due to a start-stop approach employed for thestudy. Several problems were identified during thisperiod that seem to be shortcomings of the currentNISA process. Admittedly, some of the problemsencountered were due to the protracted approachtaken. A start-stop operation is rarely an efficientoperation.

(3) Accessing the CRAY computer provides avariety of problems. Remote access for field use ofABAQUS at this time is not possible. becauseABAQUS is currently site licensed for use at WES.Districts preparing to embark on such a study willfind it necessary to utilize WES personnel at WES orperhaps negotiate a contract with the ABAQUSowner (Hibbitt, Karlsson, and Sorenson Inc.) to allowthe District access to ABAQUS via the WES Cray.Our conclusion is, that while the CRAY may providesignificant computing capability, the logistical prob-lems to off-site users may provide more problemsthan solutions for off-site users.

(4) There has been a tremendous amount ofeffort expended in developing NISA via ABAQUSand the user subroutines. The logical next step is todevelop desktop software packages that would bebeneficial to Districts with smaller projects requiringfinite element analysis or larger projects wherepreliminary two-dimensional modeling is necessary

prior to embarking on a full scale three-dimensionalNISA. The Cray-based software usage would then bereserved for those rare monumental projects. It isrecommended that a micro-based version of thissoftware be utilized for most applications.

(5) A critical deficiency in the ABAQUS analyti-cal package is the lack of graphic preprocessing andpostprocessing routines. Adequate preprocessingwould eliminate much of the input generation errors.Currently, there is no interactive preprocessing capa-bility. Further development of preprocessing andpostprocessing routines should match formatsprovided by some of the more common processingroutines currently being used by designers.

(6) For these analyses, input to the user subrou-tines required changing the FORTRAN code to pro-vide heat flux information, creep and shrinkagecharacteristics corresponding to appropriate elementsets. Although many engineers can decipherFORTRAN code, many involved in the work may nothave produced any programs for years. This can betime consuming and is a waste of effort in the design.A more appropriate means for entering data would bevia batch files to the user subroutines as is currentlypermitted. UMAT is an extremely comprehensivesubroutine that is the crux of the time-dependentstress analysis. Subsequent to this study, the capabil-ity for entering data to the user subroutine through abatch file was implemented.

(7) For Districts to utilize ABAQUS in order toexecute a NISA as part of the design process, and toachieve the stated objectives in the ETL, ABAQUSmust be readily available to District designers. With-out these labor saving additions, ABAQUS use bydesigners outside of WES may never develop.

(8) Wide usage by Districts will no doubt requiresophisticated support services in the form of trainingand user support. This service is best provided byWES. Program orientation is recommended in theform of a periodic PROSPECT course on the use ofthe NISA/ABAQUS system. Secondly, a staff mem-ber(s) must be available to service the ABAQUSsystem and provide user support. User support mayrange from troubleshooting user problems to workingas part of the design team.

A-7

ETL 1110-2-53631 Dec 94

A-5. ETL Recommendations

a. Future guidance.

(1) Future guidance concerning mass concreteshould address whether performing a NISA is manda-tory. Current guidance is unclear i.e., a reader caneither assume that a NISA is mandatory (ETL 1110-2-324 paragraph 7a) or that the need for such aninvestigation is subject to consideration (ETL 1110-2-324 paragraph 7b). It should be noted that a newETL has been published, that incorporates updatedinformation based on NISA’s that have been per-formed by WES that may address these issues. Theconsiderations for when to implement a NISA versusother less comprehensive analyses need to be devel-oped and included in any comprehensive policydocument.

(2) Certain basic questions must be addressedprior to embarking on a NISA. These include:

• Why do a thermal study?

• Is a thermal study appropriate for this struc-ture? If so, what level of analysis isnecessary?

• What are the basic principles?

• How do I do a thermal study?

(3) It is recommended that future documentationaddress these issues. The document should addressthe basic issues of the goals and desired objectiveswhen performing thermal studies and supplementedwith more specific information, for all levels ofanalyses.

(4) Guidance in the ETL references acceptablebandwidths to be applied to the mechanical propertiesof the concrete for estimated data. There are severalways to generate reasonable estimates of this data atthe time of the analysis. Further guidance, that wouldbe beneficial, should reference other sources thatcontain the methodologies to estimate the data.

(5) We concur with the recommendation of theETL that the full intended benefit of a NISA requiresthe combined efforts of the structural designer, mate-rials engineer, cost engineer, and the geotechnicalengineer.

A-8

ETL 1110-2-53631 Dec 94

Figure A-1. Finite element mesh and model data

A-9

ETL 1110-2-53631 Dec 94

A-11

ETL 1110-2-53631 Dec 94

A-12

ETL 1110-2-53631 Dec 94

A-13

ETL 1110-2-53631 Dec 94

A-14

ETL 1110-2-53631 Dec 94

A-15

ETL 1110-2-53631 Dec 94

A-16

ETL 1110-2-53631 Dec 94

A-17

ETL 1110-2-53631 Dec 94

A-18

ETL 1110-2-53631 Dec 94

A-19

ETL 1110-2-53631 Dec 94

A-20

ETL 1110-2-53631 Dec 94

A-21

ETL 1110-2-53631 Dec 94

A-22

ETL 1110-2-53631 Dec 94

A-23

ETL 1110-2-53631 Dec 94

A-24

ETL 1110-2-53631 Dec 94

A-25

ETL 1110-2-53631 Dec 94

A-26

ETL 1110-2-53631 Dec 94

A-27

ETL 1110-2-53631 Dec 94

A-28

ETL 1110-2-53631 Dec 94

A-29

ETL 1110-2-53631 Dec 94

A-30

ETL 1110-2-53631 Dec 94

A-31

ETL 1110-2-53631 Dec 94

A-32

ETL 1110-2-53631 Dec 94

A-33

ETL 1110-2-53631 Dec 94

A-34

ETL 1110-2-53631 Dec 94

Figure A-26. Stress Histories for Element 71

A-35

ETL 1110-2-53631 Dec 94


A-36

ETL 1110-2-53631 Dec 94


A-37

ETL 1110-2-53631 Dec 94


A-38

ETL 1110-2-53631 Dec 94


A-39

ETL 1110-2-53631 Dec 94


A-40

ETL 1110-2-53631 Dec 94


A-41

ETL 1110-2-53631 Dec 94


A-42

ETL 1110-2-53631 Dec 94

Figure A-34. Stress Histories of Element 309

A-43

ETL 1110-2-53631 Dec 94


A-44

ETL 1110-2-53631 Dec 94


A-45

ETL 1110-2-53631 Dec 94

A-37. Stress Histories for Element 1364

A-46

ETL 1110-2-53631 Dec 94


A-47

ETL 1110-2-53631 Dec 94


A-48

ETL 1110-2-53631 Dec 94


A-49

ETL 1110-2-53631 Dec 94

APPENDIX B: ELEMENT SIZE

B-1. Parametric Study

a. General. The integration procedures used inthe program for transient heat transfer analysisrequires a relationship between the minimum timestep and the element size. The equation to establishthis relationship is given as:

(B-1)∆ t > (ρc/6k)∆ l 2 or < l (6k∆t/ρc)

where:

∆t = time step

ρ = density

c = specific heat

k = thermal conductivity

∆l = element dimension

b. Adiabatic heat gain. Adiabatic heat gain inconcrete begins within the first 12 hr after placementand can continue rapidly until a maximum is attained.Therefore, when performing incremental time depen-dent stress analysis for concrete, it is important tokeep the time steps sufficiently small during the earlystages of the analysis. Input of the appropriate prop-erties for Zintel Canyon Dam into Equation B-1yields a maximum length of element, using a 6-hrtime interval, of 27 in. Analysis for a 12-hr timeinterval yields a 38-in. element. A 12-hr time inter-val is not a good choice for calculating early heatgain in the concrete, while placing a 27-in. restriction

for a 6-hr time interval doubled the size of the model.A 48-in. step height nearly matched production ratesfor daily concrete placements, however, did not fit thecriteria established in ABAQUS. Therefore, thisstudy focused on a 48-in. element size to determineits reliability in reporting temperature data.

c. One-dimensional heat flow.This study is asimple one-dimensional heat flow problem, usingmaterial properties for Zintel Canyon Dam. Twomodels were generated, one with a 24-in. elementsize in either direction and the other with a 48-in.element size in either direction. Depicted in Fig-ure B-1 are the two finite element meshes, and boun-dary conditions used for the study. One exterior rowof boundary nodes were held at a constant 50 degwhile the ambient surface conditions along the oppo-site face was a fixed 90 deg. The thermal modelscalculated nodal temperatures in 0.25 day incrementsfor a period of 10 days. Corresponding nodal tem-peratures were compared from both models to deter-mine accuracy, and if stable heat gain was beinggenerated. Figure B-2 contains plots of nodal tem-peratures for both the 24- and 48-in. meshes forvarious times. The only inconsistency was at timet= 0.5 days, for the 48-in. mesh, where a slight incon-sistency in the heat gain exists. This can be seen inFigure B-2.1. Figures B-2a through B-2d, indicatenearly identical heat gain, when comparing nodaltemperatures at the same time steps of the twomodels. Because this amount of inconsistency was atsmall, and only occurred at one time step, it wasconsidered negligible and would not effect the out-come of the study. Therefore, we decided to use atime step of 6 hr (0.25 days) and a maximum elementsize in any direction of 48 in.

B-1

ETL 1110-2-53631 Dec 94

Figure B-1. Finite element mesh and input data

B-2

ETL 1110-2-53631 Dec 94

B-3

ETL 1110-2-53631 Dec 94

APPENDIX C: ZINTEL CANYON PROJECT

C-1. Project Description

a. General. The Zintel Canyon Project(Figures C-1 through C-4) was authorized for con-struction by resolution of the House and Senate Com-mittees on Public Works, December 1970, underauthority of section 201, Flood Control Act of 1965(Public Law 298, 89 Congress). The project wasconstructed substantially as authorized. Detentionstorage was reduced from 2,560 to 1,260 acre-ft sincethis was considered the optimum economic size of thedam. This alternative will not prevent damages insome areas of Kennewick or avoid the use of streetsas a channel during flooding in excess of 50 years(100-year thunderstorm). Zintel Canyon Projectincludes a dam and a floodway channel with requiredstructures that carry the combined flows from thedam and areas below the dam through a developedsection of Kennewick to a discharge point at theColumbia River. Zintel Canyon Dam is a 90-ft reten-tion straight axis concrete gravity structure totalingapproximately 70,000 cu yd of roller compactedconcrete (RCC). The purpose of the dam is to pro-vide flood protection to the city of Kennewick, Wash-ington, by impounding flood flows behind the dam upto the 100-year return frequency, and discharging thatvolume over a 20-day period. The Floodway Chan-nel improvements consists of a buried conduitdesigned to pass up to a 50-year composite floodlevel. The 78-in. buried conduit is designed to carry400 cfs from its intake at West 7th Avenue and Van-couver to the outlet in the Tri-City Country Club GolfCourse. From there the natural channel is designedto pass 620 cfs through the Burlington Northernrailroad fill (Figure A-1). Downstream of the railroadfill the channel is designed to provide standard proj-ect protection. The project is co-funded by theU.S. Army Corps of Engineers (75 percent) and thecity of Kennewick (25 percent). The project islocated in a semi-arid region of eastern Washingtonand borders on the south end of Kennewick, Wash-ington. The basin, a well defined water course calledZintel Canyon, is normally dry and drains approxi-mately 28 square miles of the north side of the HorseHeaven Hills of which approximately 19 square milesin area is upstream of the dam. The drainageupstream of the dam collects winterstorm and thun-derstorm runoff, thereby providing a 100-year floodstorage volume of 1,260 acre-ft.

b. Geology and foundation. Zintel Canyon islocated on the southwest flank of the Pasco Basin, astructural feature formed by downward folding andfaulting of the Columbia River Basalt formation.Erosion and deposition has modified the structuralfeatures by partially filling the basin with sedimentsand covering the rock slope with a mantel of fine-grained materials. Bedrock is close to the surfacewithin the drainage area of Zintel Canyon and wherethe dam was located. The foundation rock was com-posed of hard, dense basalt with closely spaced frac-tures. The moderately unweathered pieces werebounded by weathered fracture surfaces. Fracturefillings, particularly near the surface, were filled withsilt and clay. Because the rock would easily dislodgewhen the joint filling dried, as well as from subse-quent construction activities, the exposed foundationrock was covered with a minimum 8-in. layer ofpumped concrete prior to RCC placement.

c. Dam, spillway and outlet.The dam is astraight axis concrete gravity structure with a crestlength of 520 ft and a 160-ft, centrally located,ungated overflow spillway. The height of abovefoundation is 126 ft and 86 ft above the existingchannel with a 20-ft crest width in the nonoverflowsection. The slope of the downstream face was.85 horizontal to 1 vertical to facilitate free formingof the downstream face. The upper 24 ft of thedownstream face of the dam (adjacent to the spill-way) was constructed using vertical concrete facingpanels as was the upstream face. An 80-ft longhydraulic jump-type stilling basin was located at thetoe of the structure. This stilling basin consists of a12-ft-thick RCC base slab integrally constructed withan RCC endsill and RCC gravity training walls. Thespillway was designed to discharge a flow of38,950 cfs. The full width of the spillway crest wassurfaced with a two-foot thickness of wet-mix shot-crete for a distance of 30 ft, until it transitions to thenatural RCC surface. A fixed orifice in the intaketower regulates discharge to a maximum of 60 cfs.This discharge rate was sized to drain the reservoir in20 days and produce minimal flows in the down-stream channel. An intake tower, attached to theupstream face of the dam, provides inlet control forincreasing heights of sediment deposition. The tower,a typical U.S. Soil Conservation Service design, con-sists of a double weir overflow at the top and portal

C-1

ETL 1110-2-53631 Dec 94

intakes at 5-ft intervals along two sides of the tower.These portals, sealed as sediment from periodicimpoundment, accumulates against the structure. Thestructure is designed to be an unmanned project.Discharged water flows through a 48-in. outlet pipecast in the RCC dam and training wall into an impactbasin. Subsequent low velocity flows are channelizedand eventually discharged into the natural channel.

d. Floodway channel.

(1) A natural water course below the dam,incised into the canyon, channels water flow until itreaches the city limits where the natural channelwidens out at west 7th Avenue and Vancouver. Atthat point the channel improvements consist of aconcrete intake structure with trash racks and anearthen dike to funnel flows of 400 cfs into a buriedconduit consisting of a 78-in. reinforced concretepipe. The conduit proceeds east on West 7th Avenuethen north on Rainier Street to the golf course whereit flows out from a concrete stilling well and followsa natural drainage path through the golf course.From Canal Drive, which borders the golf course onthe north side, the water flows through a 6- × 8-ftconcrete box culvert with a capacity of 620 cfs underCanal Drive to the Burlington Northern Railroad fillwhere a 78-in. diameter steel culvert was jackedthrough the fill to be able to pass flows up to 620 cfswith 3-ft of freeboard on the railroad fill.

(2) Downstream of the railroad fill, a 200-ftfloodway dike was constructed to elevation 383.5between Highway 395 fill to the high ground near theUnion Pacific Railroad. An opening was left in thedike to allow train traffic to continue, with a stockpileof material near by to fill in the opening when flowsexceed. The lower Zintel Canyon water course, alsoknow as Tweedt Canyon Drain, is a combination ofnatural flow channels, low bridge crossings, andculverts crossings under embankments. Dependingupon flow magnitude, water will either flow com-pletely through the area in a series of existingchannels and culverts or escape the watercourse andproceed to the east of Highway 395 overpass to theColumbia River.

e. Construction operations (Photos C-1 throughC-7).

(1) Crushed basalt rock (140,000 tons) for theRCC was produced from a quarry located only a fewhundred feet upstream of the dam right abutment.

The quarry area was developed using a dozer (CatD9C) and ripper. A crushing operation was set up,and consisted of a primary jaw crusher, an impactcrusher, and two roller crushers. The RCC mixrequired 29 to 32 percent of each rock product,2.5-in. rock, 3/4-in. rock, fine aggregate, and approxi-mately 6 percent added silt. Approximately 50 per-cent of the total required aggregate was producedprior to the start of RCC placement.

(2) Design parameters require the RCC to attaina minimum compressive strength of 1,400 psi at oneyear of age. Static stability requires cohesion valuesof 35 psi at the base of the structure, and lesser val-ues in the upper regions of the dam. Subsequentdynamic analyses determined that lift joints also hadto attain cohesion values of 50 psi in the upperregions of the structure. It was determined that thespecified construction system had to provide jointquality that resulted in shear cohesion values exceed-ing 50 psi. The resulting mix attains a 1-year com-pressive strength of 2,200 psi, and displays laboratorycohesion values of 95 psi and 150 psi for unbeddedand mortar-bedded lift joint configurations, respec-tively, at exposures of 24 hr at 70˚F. The paste-to-mortar ratio is approximately 0.50, the mortar volumeis 23 percent, and the workability level is approxi-mately 15 sec, measured using the modified vebeapparatus.

(3) Great economy is achieved when RCC pro-duction and placement can proceed uninterrupted at aconsistent production rate. Repeatedly changing mixdesigns (e.g., for upstream and downstream richerRCC zones) creates placing problems, and limitsequipment selection. Consequently, only a singleRCC mixture was produced for Zintel Canyon Dam,so that continual plant changes were not required.This is especially beneficial for continuous mixingoperations, since there is usually no convenientmethod of instant and frequent mix changes. Severalother mixes were used on the project. A highercement content mix, with an air entraining admixture,was used for the top two ft of the stilling basin slab,as well as for the top four lifts of the dam. A lowcement content mix, with an air entraining admixture,was used for the top four lifts of the training walls.

(4) Precast panels for vertical face constructionwere fabricated in a commercial precast facility100 miles from the project and then trucked to thesite. The panels, 4 ft by 16 ft in width and 4 in.thick, were keyed along the horizontal joints. The

C-2

ETL 1110-2-53631 Dec 94

panels were anchored into the RCC with 8-ft coilrods (3/4-in. diameter) and end plates. Panels wereused for the vertical faces of the stilling basin trainingwalls and for the above-grade vertical surfaces of theupstream face of the dam.

(5) Panels were placed in a checkerboard config-uration so that intermediate panels were supported bypreviously placed and anchored panels. This elimi-nated the need for external bracing. The checker-board method of panel installation is a veryeconomical panel system, however, tight alignmenttolerances are difficult to achieve. The specifiedalignment tolerances were purposely broad so thatsuch a panel installation system could be utilized.

(6) The sloping surfaces were to be a free-formed RCC slope. In order to dress these slopes,the free slopes had to be trimmed with a backhoebucket periodically. This produced the relativelyuniform appearance of the slope, and removed theuncompacted RCC on the slope.

(7) RCC was conveyed from the plant to theplacement and discharged directly into front-endloader (Cat 980) buckets. The material was driven tothe specific placement location and dumped onto theuncompacted RCC surface. The dozer (Cat D4)spread the material in 14-in. thick layers. Compac-tion was done with a 10-ton double drum vibratorroller (Ingersol Rand DA-50), and supplemented witha smaller roller (Ingersol Rand DH-22). Edge com-paction was done with a rammer (Wacker). SinceZintel Canyon Dam required only moderate shearperformance at the lift joints, bedding mortar wasapplied to the lift joints to assure shear and tensilestrengths, and vehicle transportation on the surfacewas allowed to reduce project costs. This arrange-ment did not jeopardize the lift joint quality and stillprovided significant equipment cost savings.

(8) RCC was placed in lifts 12 in. thick andmortar was applied to each lift surface. To minimizethe impacts of mortar application, the contractorformulated a system to pump mortar to the placementand shoot the mortar on the surface. The mortar mixwas modified with “a high range retarder” to produce

phenomenal extended set times and reasonablestrength performance. This process proved to be veryeffective in reducing manpower dedicated to mortarplacement and provided uniform coverage of mortar.The retarder is a product originally developed todelay the setting of concrete, in transit mixers, forextended periods of time.

(9) Placement began in the key trench, with aplacement of 16 lifts, totaling 1,800 cu yd. The RCCwas conveyed to the placement and dropped to therock or RCC surface by elephant trunk followed bydozer spreading and compaction. The placement areathen expanded to the stilling basin slab, with 12 liftsaveraging 1,400 cu yd. RCC was conveyed to load-ers, and subsequently transported to the placementlocation. Loaders traveled as much as possible onfresh RCC surfaces rather than the older surfaces thatwere being prepared for the next lift. Upon comple-tion of the stilling basin slab, the placement areanarrowed to 84 ft and continued to narrow as thedam’s height increased. The RCC lifts for the stillingbasin training walls were placed concurrently witheach lift of the dam placements.

(10) Production rates averaged 50 cu yd/hr dur-ing the early key placements and the upper lifts (inthe upper section of the dam). Typical productionrates of 200 to 225 cu yd/hr were maintained duringplacement of the stilling basin and main dam lifts.The typical placing sequence was: 1) vacuum accu-mulated debris, ponded water, and segregated aggre-gate; 2) air clean the surface; 3) wet the surface;4) apply bedding mortar; and 5) place the RCC.

(11) A drilling program commenced approxi-mately 6 months after completion of the RCC place-ments. The purpose of the drilling was to remove6-in. diameter cores from the structure and the foun-dation to evaluate the actual engineering properties ofthe RCC and the foundation rock. This testing pro-vided excellent information for future design effortsusing RCC. The testing showed that shear cohesionof the RCC lift joints more than doubled with the useof bedding mortar on the lift surfaces from 85 psi forunbedded lifts to 205 psi for bedded lift joints. Theparent RCC containing no lift joint, tested at 290 psi.

C-3

ETL 1110-2-53631 Dec 94

Figure C-1. Basin map

C-4

ETL 1110-2-53631 Dec 94

C-5

ETL 1110-2-53631 Dec 94

C-6

ETL 1110-2-53631 Dec 94

C-7

ETL 1110-2-53631 Dec 94

C-8

ETL 1110-2-53631 Dec 94

C-9

ETL 1110-2-53631 Dec 94

C-10

ETL 1110-2-53631 Dec 94

C-11

ETL 1110-2-53631 Dec 94

C-12

ETL 1110-2-53631 Dec 94

Photo C-6. Constricted placement conditions for RCC placement

C-13

ETL 1110-2-53631 Dec 94

C-14

distribution restriction statement - dtic · for the design of civil works projects. 3. references...

Documents