En`ineerin` Failure Analysis\ Vol[ 4\ No[ 2\ pp[ 136Ð148\ 0887Þ 0887 Elsevier Science Ltd[ All rights reserved\ Pergamon Printed in Great Britain

0249Ð5296:87 ,08[99 ¦ 9[99

PII] S0249Ð5296"86#99916Ð6

FAILURE OF A FLEXIBLE PIPE WITH A CONCRETE

LINER

MARK TALESNICK� and RAFAEL BAKER

Department of Civil Engineering\ Technion\ Israel Institute of Technology\ Haifa 21999\ Israel

"Received 04 September 0886#

Abstract*This study documents the functional failure of a concrete lined steel sewage pipe[ Symptoms of thepipe failure are presented[ Failure of the pipe system can be attributed to incompatibility between themechanical behavior of the pipe and the methodology employed in its design[ The underlying cause of thefailure may be traced to a lack of su.cient back_ll sti}ness[ In situ testing was used to evaluate the sti}nessof the side back_ll[ The existing pipeÐtrench system condition was analysed numerically and a criteriondeveloped for the consideration of the structural integrity of the pipeline[ Þ 0887 Elsevier Science Ltd[ Allrights reserved[

Keywords] Corrosion protection\ _tness for purpose\ pipeline failures[

0[ INTRODUCTION

The present paper documents a failure of a large diameter concrete lined steel sewage pipe\ buriedin a clay soil pro_le[ The project consisted of a 2[4 km long gravity pipe in central Israel whichfailed before being placed in service[ The present contribution documents the failure of this pipeÐtrench system[ Field and laboratory testing provided signi_cant insight into the probable cause"s#of failure[ The case study accentuates some basic design principles\ as well as the use of simple _eldtests as an e}ective diagnostic tool to evaluate site conditions[

1[ DESIGN\ CONSTRUCTION AND SITE CONDITIONS

The sewage pipeline was designed and constructed in central Israel during 0881Ð0883[ The designcalled for a steel pipe with an inner diameter of 019 cm and a wall thickness of 9[53 cm[ The innersurface of the pipe was lined with an aluminum based cement of between 0[7 and 1[1 cm thickness[The primary purpose of the inner liner was to provide protection of the steel pipe from the a}ectsof the corrosive sewage ~owing inside[ The outer surface of the pipe was covered by a 1[4 cm thickconcrete layer[

The design of the pipeÐtrench system was based on a ~exible pipe criterion[ This implies thatthe pipe maintains structural and functional integrity by mobilizing lateral resistance from thesurrounding soil[ The pipe was designed to withstand static soil loads alone[

A design section of the pipeÐtrench system is shown schematically in Fig[ 0[ The pipe invert wasfounded at a depth of between 3[4 and 4[4 m below the ground surface\ depending on the naturaltopography[ The natural soil consists of a highly plastic clay "CH\ liquid limit] vl � 51)\ plasticityindex] Ip �25)#[ A perched water table "depths of as little as 1Ð2 m# exists in part of the projectarea[ The design speci_ed the excavation of a 1[4 m wide trench "twice the pipe diameter#\ placementof a 19 cm thick layer of poorly graded gravel "GP# with a particle size between 05 and 19 mm[ Thepipe was placed directly on the gravel layer[ Following placement of the pipe section the designspeci_ed that "a# dune sand "SP# with calcareous concretions "D59 �9[06 mm and D09 �9[01 mm#be placed around the pipe to a height of 29 cm above the pipe crown elevation^ "b# above that

� Author to whom correspondence should be addressed[

136

137 M[ TALESNICK and R[ BAKER

Fig[ 0[ Typical design section of the trenchÐpipe system[

height\ natural clay material should be returned to the excavation to the original ground elevation^"c# the lower 89 cm of the sand back_ll be compacted in layers to a design dry density of 84) ofthe maximum density according to ASTM standard D0446 "gd max �06[0 kN:m2#^ and "d# allmaterials placed above the compacted sand layers be dumped in without compaction[ Constructionof the pipeline was completed in mid 0883[ The pipeline was abandoned "before any sewage ~owedalong its length# in mid 0884 because of severe cracking of the inner concrete liner[

2[ OBSERVATION OF PIPE FAILURE

Upon observation of the internal liner cracks\ a survey of the pipe condition was initiated[ Thesurvey included measurement of vertical and horizontal pipe de~ections\ visual description of theinner pipe surface and elevation of the pipe invert[ The survey was performed along most of the 2[4km length[ The survey was carried out by the Technion Foundation for Research and Devel!opment*Building Materials Testing Laboratory[

Results of the survey indicated that vertical pipe de~ections greater than 2) "of the pipe diameter#were common over signi_cant sections of the pipeline length[ In places the de~ections reached morethan 7)[ Severe cracking of the inner pipe liner was noted over substantial sections of the pipeline[Open cracks and peeling of the liner was observed at many locations[ Longitudinal cracks withapertures greater than 9[24 mm were found in pipe sections which had undergone vertical de~ectionsof 1[9) and less[ Cracking of the internal pipe liner resulted in a substantial reduction in theprotective capability of the concrete liner against corrosion of the steel pipe[ Typical results per!taining to one 019 m pipe segment are shown in Fig[ 1[ The survey indicated signi_cant deviationsof the measured pipe invert level from the design elevation[ Over signi_cant portions of the pipelinelength\ the measured invert elevation was found to be as high as 14 cm below the design level[However\ it must be noted that over several other segments along the pipeline length the surveyedinvert level was found to be above the design elevation[

138Concrete lined ~exible pipe failure

Fig[ 1[ Typical data obtained from damage survey[

3[ GENERAL DESIGN PERSPECTIVE AND PURPOSE OF INVESTIGATION

It is common to de_ne two major categories of soil!pipe systems]

Flexible pipes[ In this case the pipe is prevented from collapsing through the mobilization of soilreaction[ In order to mobilize the soil reaction the pipe must deform[ A successful design in thiscase depends on the ability of the pipe to retain its functional and structural integrity under thedeformation required to mobilize soil resistance[ This case represents a typical soil structure inter!action problem[

Ri`id pipes[ The common design assumption for this category of pipe is that their load carryingcapacity is independent of the reaction of the surrounding soil\ and pipe deformation is neglected[

It is not obvious to which of the above categories the present pipe belongs[ On one hand\ beingbasically a thin!walled steel pipe its unrestrained load carrying capacity is rather low\ making it anatural member of the ~exible pipe category[ On the other hand\ the brittle inner concrete liner maybe damaged "cracked# at deformations below those required to mobilize su.cient soil reaction[

It appears\ therefore\ that the pipe under consideration represents a borderline case which doesnot obviously belong to either one of the common design categories[ Proper pipe design requiresanalysis of the soil pipe system\ rather than use of standard design methodologies[

The objective of the present investigation was to determine the cause"s# of damage and the areasresponsible[ For this purpose it was necessary to determine mechanical properties of the pipe section\and soil conditions in the _eld[ A secondary objective of the investigation was to study the suitabilityof the pipe as a structural shell for a more ~exible insert which would act as a barrier between the~owing corrosive sewage and the steel pipe[ For this purpose it was necessary to evaluate thestructural integrity of the pipe in its present\ damaged\ condition[

149 M[ TALESNICK and R[ BAKER

4[ EXPERIMENTAL PROGRAM

The experimental program consisted of two components[ The _rst was laboratory testing of pipesections in order to determine their sti}ness "sti}ness factor�EI#\ vertical de~ection or strain\which induces cracking in the inner pipe liner and collapse loads[ The second was a _eld investigationwhich included opening of test pits at several sections along the pipeline[ Excavation of the test pitsallowed for visual description of the soil!trench cross section\ and performance of dynamic conepenetration "DCP# tests within the sand back_ll alongside the pipe[ The _eld investigation waslimited to a 229 m pipeline segment[

4[0[ Results of tests on pipe sections

Ring compression "bending# tests were carried out on three sections of pipe[ Each section wasplaced in a hydraulic press and loaded across its vertical diameter by a line load along the fullsegment length[ Throughout loading of each test section\ vertical and horizontal de~ections weremonitored[ Visual physical damage to the inner pipe lining "cracking# was also recorded[ Figure2"a# presents the experimental load deformation curve of one of the pipe sections together withobservations with respect to crack development throughout the test[ Figure 2"b# shows that theresults for the three sections are fairly similar[

Based on the data presented in Fig[ 2 it is possible to obtain the following information]

"0# The collapse load of the pipe section is between 49 and 44 kN:m[ Collapse occurred at verticalde~ections of 52Ð76 mm which correspond to diametrical strains of 4Ð6)[ It is noted that thesevalues characterize the unsupported behavior of pipe sections[

"1# The maximum moment acting in the pipe section at the collapse load may be determined byeqn "0#\ after Timoshenko and Gere ð0Ł[ For the pipes tested the maximum moments at collapsevaried between 4[5Ð5[9 kN m:m\

Mmax �P =R

1= $

p

1−0%\ "0#

where P is the collapse load per unit length as noted above\ and R is the pipe radius["2# The sti}ness factor of the pipe "EI# can be determined based on the linear section of the force

de~ection curve using eqn "1# ð0Ł[

EI�PR2

3 =Dy0p−7p1\ "1#

where Dy is the vertical pipe de~ection under load per unit length P[The calculated sti}ness of the three pipe sections was found to be approximately 02[4 kN m[ Itis noted that the EI is an inherent property of the pipe section which is independent oflateral support conditions[ This experimentally determined pipe sti}ness is representative of thecomposite pipe cross section\ which includes both concrete layers and the steel core[

"3# Severe cracking of the inner liner wall "de_ned as a crack opening of 9[2 mm ð1Ł# occurred at avertical diametric strain of approximately 0[1)[ The working assumption used throughout theinvestigation has been that cracking occurs at the same strain value irrespective of the supportconditions[ Obviously the load required to impose this strain level is dependent upon lateralsupport conditions[

4[1[ Results of _eld investi`ation

Dynamic cone penetration testing was performed at several stations along the investigated portionof the pipeline[ Technical details of the testing procedure and interpretation of results may be foundin ð2Ł[ The testing was performed following excavation of the _ll material down to the pipe crown[Two or three DCP soundings were performed within each excavation to a depth of approximately0[5Ð0[7 m[ The end point of the sounding was located at a depth of approximately 9[4 m below thepipe invert[ The plots shown in Fig[ 3 are typical results found at six stations[ It is noted that\ in

140Concrete lined ~exible pipe failure

Fig[ 2[ Pipe loadÐdeformation tests] "a# including damage observations\ "b# comparison of results for threesections tested[

general\ ~atter portions of depthÐblow count curves represent material more resistant to penetration[The slope of the depthÐblow count curve is called the DCP number "mm:blow# which characterizesthe sti}ness of the material at a particular depth[ In general a lower DCP number would indicate

141 M[ TALESNICK and R[ BAKER

Fig[ 3[ DCP sounding data[

142Concrete lined ~exible pipe failure

sti}er material[ In homogeneous soils low DCP numbers infer dense materials[ Figure 4 shows thedistribution with depth of the DCP numbers as inferred from the results shown in Fig[ 3[

At three locations along the pipeline segment considered\ test excavations were opened to depthsof 9[4Ð9[5 m below the pipe invert[ The excavations were made at locations where DCP soundingshad been performed[ Groundwater was encountered in each of the excavations[ In order to enablevisual examination\ water in the excavations was pumped out[ The examination revealed thefollowing qualitative features in each of the test pits "see Fig[ 5#[

"0# Sand back_ll of thickness between 09Ð24 cm was found below the pipe invert[ It is noted thatthe design called for the pipe to be placed directly on the gravel layer[ The best availableinformation indicates that the pipe was laid out according to the design speci_cations[

"1# Below the sand back_ll a layer of natural clay subgrade approximately 4Ð14 cm in thicknesswas found[ The thickness of this intermediate layer increases from the invert of the pipe towardsthe trench wall "see Fig[ 5"a##[

"2# Below the intermediate clay layer the gravel base was found\ and below it\ the natural claysubgrade[

The sand back_ll in the zone of the pipe haunches was found to be very loose\ signi_cantly lessdense than the sand _ll in the upper part of the trench[ The gravel layer was seen to be completelyimpregnated by a mixture of the natural clay subgrade and the sand back_ll[

Figure 6 shows very good correlation between the actual soil pro_le revealed by the visualexamination "Fig[ 5# and the results of the corresponding DCP sounding shown in Fig 3[ Thelocation of the discontinuities in the distribution of DCP numbers shown in Fig[ 4 are generallyconsistent with the layer boundaries in the lower portion of the trench pro_le[ Breakpoint A shownin Fig[ 6 implies that the sand below mid pipe elevation "haunch zone# is considerably looser thanthe sand above this level[ Breakpoint A is a common feature of all the plots shown in Fig[ 3 andFig[ 4[

Despite variations in the absolute value of the DCP numbers\ each of the sounding pro_les shownin Fig[ 4 have the following common features]

"0# There is a marked increase in DCP number at depths between 64Ð034 cm below the pipe crownwhich corresponds to the bottom part "haunches# of the pipe section[

"1# There is a marked decrease in DCP number at elevations corresponding to the visually observedgravel layer below the pipe invert\ followed by an increase in DCP numbers as the soundingentered the natural clay subgrade[

5[ INTERPRETATION AND ANALYSIS OF FAILURE

The vast majority of _eld measured pipe de~ections "as shown for example in Fig[ 1# exceed the0[1) limit found to induce severe liner cracking of pipe sections in the laboratory[ As a result theextensive damage observed in the internal pipe liner in the _eld this is not surprising[

Steel pipes are usually considered to be ~exible and they are designed in accordance with {{~exibledesign methodologies||[ However\ in the present case the deformations associated with such a designfar exceed the limiting capability of the inner pipe liner to withstand cracking[ As a result\ althoughthe pipe section may remain structurally sound\ it loses its functionality due to cracking of the liner[

Although it is impossible to specify a sharp criterion de_ning a ~exible pipe\ the value of 1)vertical de~ection is often noted in the literature as the boundary between ~exible and rigid pipesð3\ 4Ł^ i[e[ a ~exible pipe should be capable to withstand 1) de~ection without damage[ Accordingto this criterion the present pipe does not belong to the ~exible pipe category and should not havebeen design based on this methodology[

It is worthwhile to note that design standards of ~exible pipes allow vertical pipe de~ections tobe as high as 4[9Ð6[4) ð5\ 6Ł[

The large vertical deformation of the pipe and cracking of the pipe liner appear to be related toinsu.cient back_ll sti}ness as observed in the _eld investigation[ The existing sti}ness of the sandback_ll may be inferred on the basis of the DCP tests performed alongside the pipe[ Using empirical

143 M[ TALESNICK and R[ BAKER

Fig[ 4[ DCP number versus depth pro_les[

144Concrete lined ~exible pipe failure

Fig[ 5[ Schematic of visual observations in test excavation] "a# cross!section\ "b# longitudinal section[

Fig[ 6[ DCP sounding pro_le\ excavation pro_le composite[

relations between DCP numbers\ laboratory CBR values "California Bearing Ratio# and elasticmoduli it is possible to establish the following relation ð2\ 7Ł]

E�015\399

DCP9[60zlog DCP\ "2#

where E is the elastic modulus "in kPa# and the DCP number is in mm:blow[Applying eqn "2# to the DCP numbers established below breakpoint A "Fig[ 4# the elastic moduli

shown in Table 0 were inferred[The data in Table 0 show good inverse correlation between moduli inferred on the basis of DCP

results and measured pipe de~ection in the _eld\ that is\ lower moduli result in larger pipe de~ections[Such a relation should be expected on the basis of the Spangler equation ð8Ł "eqn "3## which formsthe basis of standard design procedures for ~exible pipe ð5\ 09Ł[

Dy�K =W =Dl =R

2

EI¦9[950 =E? =R2\ "3#

145 M[ TALESNICK and R[ BAKER

Table 0[ Field measured pipe de~ection\ DCP\ moduli values

Depth DCP no[ Elastic modulus Pipe de~ectionStation no[ "cm# "mm:blow# E "kPa# d "cm#

8 39Ð039 054 464 5[809¦29 89Ð029 51 1499 2[700¦16 64Ð049 24 4499 0[300¦14 74Ð034 024 674 2[801 84Ð039 35 2799 1[401¦49 × 54 49 2399 2[9

where Dy�pipe de~ection "m#\ W�soil cover loads\ taken as average prism load "kN:m#\ K�bed!ding constant "non!dimensional#\ Dl �de~ection lag factor "non!dimensional#\ R�pipe radius "m#\EI�pipe sti}ness factor "kN m#\ E ?� soil reaction modulus "kPa#[

Equation "3# was utilized with W�79 kN:m "corresponding to a depth of 3Ð4 m of soil cover#and a bedding constant K of 9[00[ The choice of the bedding constant was based on the visualexamination and it corresponds to poor bedding conditions below the pipe invert[ The de~ectionlag factor Dl\ which accounts for pipe creep and dynamic loading\ was taken as unity[ The pipesti}ness factor\ EI\ was taken to be 02[4 kN m based on the results from laboratory tests[

Figure 7 shows a comparison between the predicted de~ections based on Spangler|s equation andthe data shown in Table 0[ The open symbols shown in the plot will be referred to at a later stage[The _gure shows good correspondence between the predicted and measured results\ thus supportingthe assumption that the large _eld de~ections were due to insu.cient sti}ness of the soil back_ll[More signi_cant however is the fact that use of a very simple and cost e}ective _eld tool "DCP#coupled with empirical correlations "DCPÐE relation and the Spangler formula# make it possible topredict reasonably well\ the expected de~ections of the pipe[ In the particular case under con!sideration such an approach provides an excellent diagnostic tool to assess the pipe condition"cracking# along the length of the pipeline[

6[ STRUCTURAL STABILITY OF PIPELINE

The secondary objective of the present work was to investigate the possibility of using the existingpipeline as a structural shell to an extremely ~exible insert which would provide protection from

Fig[ 7[ Pipe de~ection vs side back_ll sti}ness[

146Concrete lined ~exible pipe failure

the corrosive e}ect of the sewage[ The insert would in e}ect functionally replace the damagedinternal concrete liner[

Design of engineering structures is frequently based on the notion of safety factors with respectto strength[ In the case of a pipe section it is reasonable to de_ne a safety factor as]

FSMom �Mult

Mmax

\ "4#

where FSMom is a safety factor with respect to moments\ Mult is the yield moment of the pipe crosssection\ as determined on the basis of the laboratory testing and\ Mmax is the maximum momentexisting in a pipe section as loaded in the _eld[ Essentially the basic engineering question to beanswered is whether the pipe in its existing deformed state\ has a su.ciently high safety factorallowing it to be utilized as structural element protecting the ~exible insert[

In general the load distribution acting on a pipe section in the _eld is unknown\ therefore\ thereis no straightforward approach to estimate the moments "Mmax#[ To overcome this di.culty it wasdecided to analyze the pipeÐtrench system numerically using the commercially available softwarecalled FLAC "Fast Lagrangian Analysis of Continua ð00Ł#[ FLAC is a two!dimensional explicit_nite di}erence program for the computation of engineering mechanics[ The program simulates thebehavior of structures built of soil\ rock and other materials which may undergo plastic ~ow whentheir yield limits are reached[ It allows for the presence of structural members which may be modeledas beams or cables[ The pipe was represented as a series of beam elements having a total thicknessequal to the composite pipe section thickness[ An equivalent section modulus "EI# as determined inthe laboratory tests was used "i[e[ no attempt was made to model the internal composite structureof the pipe section#[ The di}erent soil layers were modeled as elasto!plastic MohrÐCoulomb mediaeach assigned representative soil deformation and strength parameters[ The discretization schemeand chosen material properties are shown in Fig[ 8[ A plane strain problem with a single axis ofsymmetry "AA?# was considered[ One half of the soilÐtrench system was represented by 873 two!dimensional solid elements\ and the other by 02 one!dimensional beam elements[

Since soil behavior is stress history dependent\ we found it important to follow\ in a numericalsense\ the _eld construction sequence[ Toward this end the following three numerical steps weretaken]

"0# Establishment of the initial\ at rest\ state of stress in an homogeneous half space of the claypro_le[

"1# Establishment of stresses and strains in each element resulting from {{excavation|| of the trenchpro_le "ABCD*Fig[ 8#[

"2# Establishment of the stresses and strains resulting from placement of the pipe and back_lling ofthe trench[ It is noted that the initial conditions for this step are the stresses and strainsestablished in the previous stage[

In order to test the suitability of the numerical system as a predictive tool a parametric studyrelating vertical pipe de~ection to the sti}ness of the sand back_ll was performed[ The results ofthese calculations are shown as the open symbols in Fig[ 7[ The numerical results compare very wellwith the _eld values[ Both the numerical computations and the _eld values fall below the curverepresenting the Spangler model[ Such an outcome is reasonable considering the fact that theSpangler formula is a design tool rather than a predicitive one[ It should be noted that the datashown in Fig[ 7 involves three di}erent {{types|| of elastic moduli\ namely] modulus of soil reactionlabeled as E? in the Spangler equation^ conventional modulus of elasticity E as used in FLAC^ anda sti}ness modulus based on the DCP results[

Despite these di}erences in de_nition of the sand back_ll sti}ness\ the correspondence of the datais quite remarkable[ It is not clear whether this is a general phenomenon^ or true only in thisparticular case[

For each assumed value of soil modulus the numerical scheme yields not only the verticalshortening of the pipe diameter "used in Fig[ 7#\ but also the distribution of the beam momentsaround the pipe circumference[ It is possible therefore to plot the maximum moment developed inthe pipe section as a function of the vertical shortening of the pipe diameter as shown in Fig[ 09[The dashed line in the _gure represents the maximum moment de~ection relation for an unrestrained

147 M[ TALESNICK and R[ BAKER

Fig[ 8[ FLAC input] "a# grid\ "b# system pro_le and parameters[

pipe section "laboratory conditions#[ The plot illustrates that for any given pipe de~ection themaximum moment in an unrestrained pipe is greater than the maximum moment in a buried one[Including within Fig[ 09 the de_nition of the safety factor "eqn "4##\ it is possible to construct asafety factor!de~ection relation\ shown as the curve through the triangular points in the _gure[ Thiscurve can be used to assess which pipe segments are su.ciently safe to be used as a structural shell[For example\ assuming a required safety factor of 1\ all pipe segments which have undergonede~ections greater than approximately 4 cm would be considered unsuitable[ The advantage of thisapproach is its simplicity[ Pipe de~ection is a simple parameter to measure\ whereas moments in thepipe section are not[

7[ CONCLUSIONS

The following conclusions can be drawn from the present investigation into the failure of thispipeline]

"0# {{Flexible|| pipes with rigid liners must be designed with care[ Flexible pipe design methodologiesmay be applicable\ however\ the deformation limitations of the liner must be carefully

148Concrete lined ~exible pipe failure

Fig[ 09[ Maximum moment\ safety factor\ de~ection plot[

considered[ Careful control over the sti}ness of the trench back_ll material is of the utmostimportance[ In the particular case considered\ de~ection of the pipe was in places 5Ð6 times thedeformation initiating severe damage[

"1# DCP sounding has proven to be a simple _eld tool capable of estimating soil sti}ness moduliof the trench back_ll materials[ Field measured de~ections\ predicted de~ections based onstandard design procedures and de~ections predicted by use of a sophisticated numerical tech!nique\ correspond remarkably well[

"2# A criterion for the structural stability of the pipeline in terms of moments has been presented[This criterion makes it possible to utilize the measured de~ections in a decision process aimedat establishing which parts of the pipeline are suitable for use as a protective structural shell[The advantage of the criterion is due to the fact that it directly relates safety factor to themeasurable quantity of de~ection[

REFERENCES

0[ Timoshenko\ S[ and Gere\ J[ M[\ Theory of Elastic Stability\ 1nd edn[ McGraw!Hill\ New York\ 0850[1[ AASHTO Designation T179] Standard practice for concrete pipe\ sections or tile[2[ Livneh\ M[ and Ishai\ I[\ Pavement and material evaluation by a dynamic cone penetrometer[ Proceedin`s of the 5th

International Conference on Structural Desi`n of Asphalt Pavements[ Ann Arbor\ MI\ 0876[3[ Moser\ A[ P[\ Buried Pipe Desi`n[ McGraw!Hill\ 0889[4[ Howard\ A[ K[\ Pipe bedding and back_ll[ Geotechnical Branch\ Division of Research\ Bureau of Reclamation\ United

States Department of the Interior\ Engineering Research Center\ Denver\ Colorado\ 0870[5[ AWWA "American Water Works Association#\ Steel pipe*a guide for design and installation\ AWWA Manual M00\

1nd edn[6[ ASTM Standard F568] Standard practice for Poly"Vinyl Chloride# "PVC# large!diameter plastic gravity sewer pipe and

_ttings[ ASTM Standards\ Vol[ 97[93 Plastic pipe and building products[7[ Yoder\ E[ J[\ Principles of Pavement Desi`n[ Chapman and Hall\ 0848[8[ Spangler\ M[ G[\ The structural design of ~exible pipe culverts[ Bulletin 042\ Engineering Experiment Station\ Iowa

State University\ 0830[09[ ASTM Standard D2728] Standard practice for the underground installation of {{Fiberglass|| "glass!_ber reinforced

thermosetting resin# pipe[ ASTM\ Vol[ 97[93 Plastic pipe and building products[00[ Itasca\ FLAC "Fast Lagrangian Analysis of Continua#\ Itasca Consulting Engineers\ Minneapolis\ Minnessota\ 0881[