REPRINT
STINA
Project for the Application of the AASHO
Road Results in the Nordic Countries
FINAL REPORT
The Secretariat of the Nordic Council
Projekt Organization
Coordinator:
Magnus Holmgren
National representatives:
Denmark Mogens Rasmussen
Erik Bärenholdt Finland Esko Kankare
Jon Skulason Rasmus Nordal Iceland
Norway
Geir Refsdal Sweden Olle Andersson
Note
This booklet is a reprint of the English summary of the project technical report, included in NU A 1977:3 4, published by the Nordic Council, Gamla Riksdagshuset, Stockholm, Sweden.
The Secretariat of the Nordic Council of Mi-nisters, Oslo, Norway
Public Roads Administration, Copenhagen
The National Danish Road Laboratory,
Ros-kilde _
Technical Research Centre of Finland. Road and Traf c Laboratory, Otaniemi
Public Roads Administration, Reykjavik Norwegian Institute of Technology, Trond-heim
Public Roads Administration, Norwegian Road Research Laboratory, Oslo
Royal Institute of Technology, Stockholm in the STINA project representing Swedish Road
9. Summary
Background, purpose and planning
The design specifications for road pavements which are in use in different countries are mostly based on accumulated experience during all the time heavy vehicles have been
in use. For quanti cation of the relation
be-tween the traf c load ( ow and axle weight) and the design parameters of the pavement the American road association AASHO took in 1946 the initiative of the most compre-hensive road experiment ever made. To illu-strate the extent of the experiment it can be mentioned that the total cost was $27 million, it lasted from 1958 till 1960, and resulted in 300 million data which were ana-lyzed. The results comprise relations
be-tween serviceability, number of axle
passa-ges, axle load and pavement design
para-meters.
Quite naturally the results have been
re-ceived with great interest, not only in the
US but also in Europe. Thus the applicabili-ty of the results at other locations than the site of the experiment has been debated. Especially differences in the composition of the traf c and differences in climate and subgrade properties have been payed atten-tion to. Also in the Nordic countries the AASHO results have naturally been studied and to some extent applied, although syste-matic applicability test on a large scale have not been performed. Against this back-ground, and due to the common interest in
176
the question, a Nordic cooperative project, named by the acronym STINA, was started. The purpose of the project was to give an improved base for pavement design, especial-ly with respect to in uence of subgrade, traf c load and climate, and to illustrate the in uence of traf c load on investment and
operation costs.
No appraisal of the AASHO road test and its results was included in the STINA pro-ject.
After programming the work started in the spring of 1974 and was concluded in the fall of 1976. Sponsored by the Nordic Coun-cil of Ministers, who also employed the pro-ject coordinator, the work was done in the member countries by national project groups, each led by a national project leader. The work was divided into ve main activiti-es, and each main activity was assigned to one national project group for detailed plan-ning, coordination and reporting. The main acitivities were
inventory abroad
inventory in the Nordic countries eld trials and sampling
laboratory investigations analyses
The rst two acitivites were accounted for in a progress report, published as two parts of Nordisk utredningsserie , denoted NU 1975:Il and NU 1975:12.
[i-wentory abroad
The purpose of this inventory was to find out by literature surveys and by visits to se-lected countries, how application and deve-lopment of the AASHO road test results had progressed during the last few years. Visits were paid to the US and Canada and to Europe, where interviews were made at road research institutions and road authorities.
The subgrade is in many countries intens-ly studied with respect to stress-strain be-haviour and fatigue properties. The influence of the subgrade upon road pavement failure was however found to be incomplete.
During the study visits special attention was given to the way in which the AASHO load equivalency factors were used in treat-ing traf c load in the different countries. It then appeared that the equivalency factors according to the four power law had been largely accepted, lacking other alternatives. An important finding was the vehicle equi-valency factors, based on traffic load ana-lyses and differential traffic counting. Such factors have also been used in Finland for several years. The influence of Climate was in writing the AASHO interim guides ac-counted for by assigning regional factors to different areas, representing the average climatic influence of the area. Considerable skepticism against these factors was met, since they account for only part of the climatic influence upon road design.
The serviceability concept was another important product of the AASHO road test. The serviceability index is a quantitative
measure of the state of the road as
experi-enced by the motorist and is used as a
meas-ure of the state of the road as experienced by the motorist and is used as a measure of the state of the road also in analysis of road failure.
The failure model concept is used to quantify the progress of the failure of the road in relation to the variables causing fail-ure. Several failure models developed abroad were recorded for use in analysis of road fail-ure in Nordic countries.
Inventory ofNordic conditions
The purpose of this inventory was to
com-pile data from the Nordic countries required for examination of the applicability of AASHO road test results and other results derived therefrom.
The inventory comprised road structures,
rheology, climate, traffic load, failure data
and test road data. The data collection was made by sending questionnaires to road au-thorities and research institutions. In addi-tion road appraisal for determinaaddi-tion of serv-iceability was performed in the member countries and on a number of test sections in the member countries and on a number of test sections in the vicinity of Stockholm International Airport (Arlanda). These sec-tions were also used for roughness measure-ments in order to establish correlation with data from subjective ranking.
These inventories gave an account of the occurence of different subgrade materials in the road networks and the permitted
traf c loads in different countries. The
climate inventory resulted in maps of the Nordic countries showing zones of mean
temperature, mean precipitation and mean
freezing index.
Rz'deabz'lity ranking and roughness meas-urements of selected test sections resulted in a serviceability index formula (equiva-lent to the AASHO PSI formula) and a re-lated Nordic terminal serviceability index.
The road damage inventory gave an
indi-cation of the occurrence of damages, their kind and causes.
In conjunction with development of a Nordic serviceability index formula, atten-tion was directed towards the existence of different roughness meters in the Nordic countries and the possible use of other meters than the CHLOE meter for service-ability assessment. Correlation studies were therefore performed using the CHLOE me-ter, the bump integrator and the towed straight-edge. All instruments were used at the same time for measurements on the test sections near Stockholm International
Air-port.
These measurements resulted in empirical
formulae for conversion of data from the other two instruments into CHLOE data
and in serviceability formulae bases upon
these other two instruments.
Field tests and sampling
The purpose of the field tests was to start an inventory of road engineering properties, predominantly bearing capacity, of Nordic subgrades and their seasonal variation. The measurements also comprised moisture
con-tent, temperature, water table and frost
depth and were performed ten times in 1975, beginning at frost break. Bearing
ca-pacity was measured by conventional static
plate loading on the subgrade and on the surface of the road adjecent to the eld
station.
During recent years it has become more and more customary to simulate the traffic load in bearing capacity measurements by
using a dynamic type of measurement.
Dur-ing plannDur-ing of the STINA project, however, equipment for such measurements was not
available at resonable cost, and therefore all the measurements were made by using
con-ventional static plate bearing equipment. The measurements were made at field stations prepared for this particular purpose. Between measurements the subgrade at these stations was provided with a thin pavement structure, which was thermally and hydrolo-gically equivalent to the specified pavement structure required by such a subgrade. Two subgrade types were selected in each country
(in Iceland only one), selection being made
on the basis of most trafficked subgrade under the public road network.
A multipurpose sample was taken from each field station for geotechnical analysis and classification. Sampling was also made for deformation and fatigue testing. From stations with friction material this sampling was done once and for all and from stations with cohesive material the sampling was done in connection with each field testing.
178
The winter 1974 75 was unusually mild in all Nordic countries except Iceland, and therefore there was practically no frost break period. A few more measurements were therefore made in the spring and early summer of 1976. This winter was however also milder than average.
In plate bearing testing a procedure was
used which was specially developed for the
purpose, This implied loading from zero to
each load level three times, the level being
increased step-wise until a predetermined
deflectiOn was reached.
The results of these field tests was report-ed as static eleastic moduli, computreport-ed from the total deflection and from the rebound de ection (Et and Ee), since both moduli are used in road design in the Nordic coun-tries. Regression analysis of the two moduli gives linear relationships with very high cor-relation coefficients, the analysis being done separately from each field stations. The in-tercept of each linear relation is however
considerable, which illustrates that the
mo-duli are not mutually interchangable, The intercept reflects the permanent de ection, which is exceptionally great with the Ice-landic lava. This reflects the high porosity of the lava material.
The moduli showed at several places a
considerable variation with load level, which
illustrates the non-linear properties of these materials and also to some extent the bedd-ing effect of the plate load test.
The seasonal variation of bearing capacity at each field station was reviewed by draw-ing a curve showdraw-ing modulus against time, the modulus value being selected at the loading level corresponding to the weight of the pavement required by that subgrade type according to the local design specification. In spite of the mild winters a fairly normal course of each seasonal curve was obtatin-ed, having low bearing capacity in the spring and high values during the later half of the year. At some stations the bearing capacity ofthe adjacent road showed a minimum dur-ing late summer. This is a natural result of the softening of thick bitumen bound
ment layers during the summer months, whereas the softening during the spring period was not very pronounced after the mild winters. _
As an inventory of the bearing capacity of Nordic subgrades the work done has to be
considered as a-start. However, a common
method for measurement has been
develop-ed, and it is recommended that the
measure-ments are pursued for a considerably pro-longed period.
Laboratory in vcstigations
The laboratory investigations of subgrade materials aimed at deformation and fatigue properties comprised undisturbed samples from cohesion materials and laboratory com-pacted samples of friction material, compac-tion then being made at moisture contents and to bulk densities prevailing during eld tests. The measurements on cohesion materi-als were made by applying haversine shaped
loads in a triaxial cell, whereas friction
ma-terials were tested also by applying haversine shaped loads in a triaxial cell, whereas
fric-tion materials were tested also by applying
haversine shaped loads but in an open-face cylinder with a dead load on the free soil surface.
Undisturbed tube samples were also im ported from three places outside
Scandi-navia, namely 100p number 1 of the AASHO
road test site, the Alconbury Hill test road in England and the Hiltpoltstein test road in West Germany. Alconbury Hill is a test sec-tion on the London-Edinburgh motorway used for follow-up studies of the AASHO road test, and Hiltpoltstein is the site of an
extensive road test in Germany.
A sampling technique adapted for these and the Nordic field stations had to be devel-oped. Altogether 128 tube samples were col-lected. Some of them were cancelled and
some were stored for future research, while
69 samples were analysed at the road
labo-ratory at Roskilde, Denmark, and at the
In-stitute of Technology at Trodheim, Norway. The friction material investigations were
made at the Swedish Road and Traf c Re-search lnstitute at Linköping, Sweden.
The undisturbed samples were also sub-jected to CBR testing and static testing in a triaxial cell. These tests were used predoni-mantly for determining the strength pro-perties of the undisturbed samples under static conditions. Dynamic testing consist-ed in applying 20000 load pulses at each
stress amplitude in a stepwise increasing
sequence of amplitudes.
During testing the rebound de ection and the permanent de ection of the samples were recorded continously. The former was used for computation of the rebound mo-dulus Mr, which was recorded as a function of the stress amplitude.
No systematic variation of Mr with mois-ture content could be established except at
frost break, and therefore the rebound
mo-dulus was reported as a function of stress amplitude, showing the mean curve and the envelopes generated by the standard devia-tion. The frost break curves were reported separately, where necessary.
The modulus curves can mathematically
be represented by a power law, where the ex-ponent lies in the interval Olo 0.57 for all materials tested. The modulus value corre-sponding to the stress 0.1 athmospheres was denoted material modulus and was de-signated by the synbol M0. It was consider-ed as a parameter expressing the dynamic properties of the material. This quantity showed a noteworthy correlation with the CBR value, the factor of prOportionality beting 100, if the modulus is expressed in kg/cm2. This is an interesting con rmation of earlier work in this field. The present re-sult is however confined to undisturbed cohesive soil samples.
The permanent de ection has been fitted into a power function, comprising the num-ber of load applications. the stress amplitude and the rebound modulus. From this equa-tion it is possible to derive a stress equival-ency factor for comparison of different
stress amplitudes, which in contrast to the
expo-nent values in the interval 15-~20.
This equation of the permanent
deflec-tion can be interpreted in the following way:
- there exists a critical stress level in cohesive subgrade materials
- at stresses below this level, failure seldom
occurs in cohesive subgrades
- at stresses above this level failure occurs after only a very limited number of load applications
-- - it is not realistic to use load equivalency factors for cohesive subgrade failure __ the subgrade criterion should be based
upon the maximum permitted stress (Which of course varies with season)
Equivalency factors should on the other hand be used in treatment of the contribu-tions of the pavement itself to the failure of the road, but it is then not dependent upon the failure process in the subgrade.
At the analysis of corresponding measure-ments on friction subgrade materials a simi-lar arithmetic expression was used. The
cor-relations were however rather weak, and an
equivalency factor of the same kind could not be derived. In order to give more con-clusive results these measurements require a considerable extension beyond what was availabe within the limits of the STINA pro-ject.
The studies of subgrade materials was combined with the field tests. searching a re-lation between the modulus values found in laboratory tests and those found in the field tests. The stress dependent field moduli could be fitted into a power relation similar to that of the laboratory rebound moduli. The factors of proportionality found in these analyses differed rom one another (field and laboratory) by a factor of 2 or more. Consindering dissimilarities in the tests such as temperature and depth under the subgrade surface this agreement can be considered satisfactory.
A nalyscs
Improved basis for analytical pavement
de-180
sign was one of the main purposes of the STINA project, and this implied predomi-nantly load equivalency factors and the con-tribution from the subgrade to road failure criteria. Analyses of these factors were part-ly made hand in hand, and the results were checked against oneanother.
Traffic load
By consideration of the wearing wourse, the pavement or the subgrade separately or in combination it is possible to derive equiva-lency functions with a wide range of expo?
nents. The most extreme exponent values,
pertaining to choesive soils, can as explain-ed above be ascribexplain-ed to a fracture process and can be replaced by a critical stress criteriOn instead of a fatigue criterion. Exponents derived in other ways remain in the region 2.5 - 5, the low end of the range being derived from permanent de ections of the road surface. Considerations of the whole road structure lead to exponent values in the range 3 5. Such values are obtained eg from a linear elastic model using the exponent value 4 and modification of this model by assuming non-linear pro-perties of the pavement materials.
A conservative conclusion from these observations is that the exponent value of4 as derived from the AASHO road test has not been seriously contradicted under Nordic conditions. A range of3 5 or possib-ly 3.5 4.5 can be stated a applicable. On the other hand computations of the equiva-lent number of 10 ton axles based on real axle load distributions show a moderate in uence of the magnitude of the expo-nent. ln analyses of the traffic load in con-junction with pavement design therefore, the assumption of the magnitude of the expo-nent is unimportant as long as its value is limited to the interval 3 5.
Subgrade criteria
Determination of the critical values of the subgrade has earlier been made by
tion of the subgrade strains during the various phases of the AASHO road test and therefore determining a relation between the number of load applications to failure and the subgrade strain amplitude (sometimes
referred to as the Shell criterion). Failure
was then defined as the occurrence of ter-minal serviceability index.
In the STINA project a similar approach was tried but not applied to the AASHO road test but to the road pavements built according to standard speci cations. This procedure was applied to the specifications in Finland, Norway and Sweden. Iceland has the same specifications as Norway and the Danish specifications are based upon an analytical approach, which would make the proposed test inadequate.
A design period of 10 20 years was
assumed, which in combination with known
traffic flows gave a basis for the number of passages to failure. The pavement layer parameters were taken from experience in the various countries.
This procedure implies quite a few simpli-fications and hence the high and low ex-tremes were computed in each combination of inputs. These extremes defined limiting lines of the computed relation between subgrade strain and number of passes of equivalent standard axles. The subgrade criterion diagram therefore was represented by a band rather than a curve. In two-sided logarithmic representation the limiting lines are straight and have a slope of 1/4, confirm-ing the existence of an exponent value near 4.
Between the extreme curves derived from the Swedish standard specifications the corresponding curve derived from the AASHO road test can be drawn. The Nor weigian and the Finnish specifications gave a similar result, although the agree-ment with the Shell criterion was not in all cases quite so good.
The Shell criterion is based upon sub-grade strain, whereas there are other criteria systems based Upon subgrade stress. The basic difference between these criteria is
not considerable but can be of importance when non-linear elasticity is assumed.
Climate
The influence of the climate upon pavement design has been given some attention in the STINA project. The AASHO test site repre-sents only one climate, but the change of
climatic conditions during the test was
ac-counted for by using a set of seasonal weight ing factors, based upon deflection measure-ments of the untrafficked lOOp number 1. Similar weighting factors have been calculat-ed from the field station results in the
STINA project (adjacent roads). The season
curves then obtained were at some stations not similar to those obtained at the AASHO road test. This is partly due to the absence of typical frost break periods during the course of the STINA project. It should also be noted that the sections underlying the AASHO seasonal weighting function curve did not contain thick bitumen stabilized layers. Since no serviceability measurements were made at the test stations the usefulness of the method cannot be checked but such
measurements are recommended in the fu
ture. Tests of the regional factors recom-mended in the AASHO interim guides were not checked within the STINA project.
Failure models
The AASHO pavement design system is based on the semi-empirical failure model derived in conjunction with analyses of the AASHO road test data. The model gives by a system of formulae an arithmetic relation between serviceability index, number of passes. axle load and the bearing capacity of the pavement. Later on the AASHO model was extended to taking care of the properties of the subgrade and the non-traffic load influenced pavement failure. Other models but of different arithmetical structure have been developed in other areas in North America. These models, including a British failure model (based on Alconbury Hill test road data) of quite different
compo-sition, were tested within the STINA project by application to data from Nordic test roads.
The inventory of Nordic test road data gave a rather limited set of data in view of this purpose, predominantly because traffic load data were missing and data for assess
ment of serviceability were missing at many
test roads. Swedish test road traffic data could largely be aquired afterwards by traf-fic counting and be extrapolation back-wards in time. The test roads included in the analysis were those at Edsvalla, NykrOppa and Barkarby-Staket. The later is not actual-ly a test road but has been subjected to ex-tensive measurements by the CHLOE meter for a number of years. In addition the Vorm-sund test road in Norway could be included in the analysis.
The Vormsund test road data could be fitted to three of the models tested, namely the AASHO, Ontario and Alconbury Hill Models. To the data from the Nykroppa test road only a modi cation of the British Al-conbury Hill model could be fitted, whereas to the other Swedish test road data the AASHO Texas and Ontario models could be fitted.
The Texas and Ontario models (one being a modification of the other) both have a component which is independent of traffic load and describes mainly climate induced failure. This is also true for a later modifi-cation of the AASHO model. Fitting of Swedish test road data required such a load independent component. A factor applied to the time variable in these models can be used as a climate parameter which describes the failure directly induced by the climate. The goodness of fit to Nordic test roads was similar to that in the home areas in spite of the rather limited amount of data.
These models are expected to find their use in development of rehabilitation strategi-es and in calculation of the cost rstrategi-esponsibili- responsibili-ty of different responsibili-types of vehicles.
182
Rehabilitation strategies
In order to include in a rational way all the
factors entering into Optimization of rehabi
litation strategies the methods of systems engineering have been taken into considera-tion during later years. These strategies re-quire quantification of the state of failure of the road, and for this purpose the AASHO serviceability concept has been very helpful. An American system denoted SAMP has been treated within the STINA project. It has been adapted for Nordic conditions and extended for wider application. The re-sulting system, denoted by the acronym SYLVIA is comprehensive and contains a separable part, named SUSANN, which treats regular maintenance. These systems could not be developed further than to principles, but for implementation the SAMP 5 system was adapted to Nordic conditions and is offered as a FORTRAN program (denoted NSAMP) ready for test-ing by Nordic agencies.
Vehicle equivalenqv factors
In analyses of the traffic load spectrum it has been found that the number of vehicle types running on the roads is quite limited. Each vehicle type usually has a miximum nominal axle load. It has therefore been tried at some places to assign to each vehicle type a vehicle equivalent factor, being the number of equivalent standard axles per vehicle. In this way the expensive traffic load analyses could be replaced by differen-tial traffic counting for assessment of the number of vehicles of each type.
This method has been in use in Finland since 1964. It has also been tried in Minne-sota, where however the correlation between the number of standard axles found by the two methods was rather weak. A similar analysis has been made within the STINA project, using a very comprehensive traffic analysis made by the Swedish Road Ad-ministration in the middle of the sixties. The analysis was limited to a small
tion of roads of different types and loca-tions. lt comprised, however. all vehicle types except passanger cars whose contribu-tion to the total traf c load was found to be negligible. The vehicle equivalence factor (VEF) is identical with the truck factor which has been in use for a long time, but the truck factor averages all heavy vehicles. VEF was found to vary not only with the vehicle types but also with region and type of road. The exponent of the equivalency factor formula is however quite unimportent within the range 3.5 5. ln road design the exponent value is therefore unimportant. if VEF is used. A fair agreement between Fin-nish and Swedish VEF :s was found.
Cost respons!!)l'lify
The structure of rehabilitation strategies shows that there is quite a strong relation-ship between certain road operation costs and traffic load. Within the STINA project therefore the possibilities have been studied to use these relationships for allocation of the traffic load dependent part of the Opera-tions costs to vehicles of different axle loads. Thereby a certain rehabilitation strategy was assumed and the decrease in annual cost caused by exclusion of one vehicle type (or axle load) was computed. Repetition ofthis computation excluding one vehicle type (or axle load) at a time and comparing the cost reductions was assumed to re ect the cost responsibility of the component excluded.
lt can then be assumed, that the original strategy is maintained except for a change (i.e. cost) of the treatment due to the ex-clusion of one traffic component. The calcu-lations, all inputs considered, become rather involved, but a few examples, based upon drastic simplifications. were performed. The two alternatives of excluding one vehicle type at a time and excluding one axle load at a time were tried. The former alternative inplies the use of vehicle equivalency factors. As expected the influence of the magnitude of the exponent of the equivalency factor formula is much less if the vechicle approach
is used. On the other hand, in the present work great differences were found between vehicle types that are very similar.
Pursuit of this approach could certainly give important contributions in this problem
area, but this would require a considerable
reseach effort.
Future aspects
In the final chapter of the report there is a list of possible follow-up projects. which could be carried out by all or a few of the member countries. One of these projects has already gone into operation and will be finishedby the end ofJune 1977.
The STINA project has Opened up new varieties of c00peration in Nordic road re-search and has contributed to the progress in this field on the same time as it has sti-mulated further cooperation of a similar kind. '
List of results
Subdivided according to the declared pur-poses of the STINA project the results may . be condensed in the following manner.
Subgrade
0 The start of an inventory of road engi neering prOperties, covering nine typical Nordic subgrades during one climate season.
' Experimental technique for seasonal as-sessment of subgrade bearing capacity. . Experimental technique for determination
of road engineering properties, especially those important for bearing capacity, by sampling and laboratory testing of sub-grade materials.
0 Improved knowledge of the deformation and fatigue properties of subgrades and subgrade materials, including non-linear elasticity.
0 Check of subgrade criteria used for ana-lytical pavement design.
stresses in cohesive subgrade soils.
O lilucidation of the relation between sub-grade properties determinad by field and laboratory testing.
Traffic load, road failure
O Establishment of the influence of sub-grade properties on the failure of roads built on cohesive soils.
0 Analysis of vehicle equivalency factors based on axle arrangements.
0 Elucidation of the applicability of the four power rule in the Nordic countries. O Elucidation of the applicability of road failure models in Nordic countries. 0 Serviceability formula applicable in the
Nordic countries.
O Inventory of road failure mechanisms in the Nordic countries including pertinent road engineering data.
Climate
O The seasonal variations of road bearing capacity. documented examples.
G Illustration of the use of the climate parameter in failure models for quanti -cation of the climate induced road failure. 0 Inventory of climate data in the Nordic
countries.
Cost Optimization
0 Test of a new approach to calculation of the cost responsibility of different vehicle types in road failure. based on
rehabilita-tion strategies.
0 Suggestions for design and rehabilitation
strategies in the Nordic countries.
0 Modification of an American design and rehabilitation system. prepared for testing and implementation in the Nordic coun-tries (NSAMP).
