Using Road Weather Information Systems (RWIS) to optimize the scheduling of load restrictions on Northern Ontario s low-volume roads.

Transcription

1 Usng Road Weather Informaton Systems (RWIS) to optmze the schedulng of load restrctons on Northern Ontaro s low-volume roads. Sarah Baz, M.A.Sc. Canddate Department of Cvl Engneerng, Unversty of Waterloo 200 Unversty Avenue West, Waterloo, ON, Canada N2L 3G sbaz@engmal.uwaterloo.ca Susan L.Tghe, Ph.D., P.Eng. Canada Research Char and Assocate Professor of Cvl Engneerng Unversty of Waterloo, sltghe@cvmal.uwaterloo.ca Phone: (59) Ext. 3352, Fax: (59) Carl T.Haas, Ph.D. Canada Research Char and Professor of Cvl Engneerng Unversty of Waterloo, chaas@cvmal.uwaterloo.ca Phone: (59) Ext , Fax: (59) Bran Mlls Adaptaton and Impacts Research Group Meteorologcal Servce of Canada, Envronment Canada Faculty of Envronmental Studes, Unversty of Waterloo Phone: (59) Ext , Fax: (59) Bran.Mlls@ec.gc.ca Max Perchanok Mantenance Research Coordnator Hghway Standards Branch Ontaro Mnstry of Transportaton 30 St. Paul Street, St. Catharnes, Ontaro L2R 7R4 Phone: (46) , Fax: (46) Max.Perchanok@ontaro.ca Paper prepared for presentaton at the Preservng Low Volume Roads Investments Sesson of the 2007 Annual Conference of the Transportaton Assocaton of Canada Saskatoon, Saskatchewan The authors would lke to acknowledge the contrbutons of the Mnstry of Transportaton of Ontaro and the Hghway Innovaton Infrastructure Fundng Program (HIIFP) for fundng ths research. In addton, the authors would lke to acknowledge the MTO Regonal staff n partcular Ken Mossop, Mckey Major, Bran Scott, Doug Flegel and Vc Lawrence, for ther assstance. - -

2 ABSTRACT Seasonal shfts n mosture and temperature wthn the pavement can affect the bearng capacty of the roadway, especally n Northern Ontaro where low volume roads are challenged by frequent freeze-thaw cycles. Seasonal load restrctons (SLR) are mplemented every year on Ontaro s secondary and tertary hghways to protect the road nfrastructure. However, each tme loaded trucks are allowed on a weakened pavement, premature deteroraton of the roadway occurs and mantenance expenses ncrease. Conversely, when the payload s unnecessarly restrcted, the transportaton and resource ndustres are penalzed. These economc losses are due to the reactve approach of SLR tmng, hstorcally based on patrols and expert judgment nstead of systematc montorng of the pavement condton. Road Weather Informaton Systems (RWIS) form a network of automated clmatc statons desgned largely for supportng wnter mantenance operatons n Ontaro. A study conducted by the Ontaro Mnstry of Transportaton and based on two expermental stes located n Northern Ontaro found a reasonable correlaton between frost thckness n the roadway and RWIS varables. As a follow-up to ths analyss, frost predctors have been developed ths year and are presented n ths paper along wth the results of crackng smulatons usng the Mechanstc-Emprcal Pavement Desgn Gude and typcal Northern Ontaro condtons. Based on varables provded by the onlne RWIS database and Envronment Canada forecasts, models would ndcate changes n relatve pavement strength. Real-tme and cost-effectve gudelnes could then be derved and translated nto a decson-support tool that asssts MTO engneerng professonals n the mplementaton of load restrctons

3 INTRODUCTION AND BACKGROUND Load restrctons polces n Ontaro are manly targetng the low volume pavements whch are generally bult to lower standards than those of the Kng s hghways. Ths analyss focused on the secondary (500-seres and 600-seres) hghways of Northern Ontaro, most of whch are remote access ndustral and resources roads and are thus subject to heavy traffc loadng. These roadways were bult to handle year-round unrestrcted traffc, apart from specfc stes lmted by certan crcumstances. However, t s the combnaton of sprng thaw cycles wth heavy truck loadng whch s suspected to be responsble for the major part of pavement damage. Therefore, each year, a seasonal reduced load perod, namely a Sprng Load Restrcton (SLR) perod, s put nto effect on the varous low volume routes regrouped as Schedule 2 Hghways, usually throughout March, Aprl and May []. Also, overszed load permts - Wnter Weght Premums (WWP) -, whch are usually allowed as long as the pavement structure s frozen and thus assumed to be able to cope wth hgher loads, are restrcted durng an SLR perod. Even though load restrcton perods are commonly desgnated as half load perods, secton 22 of the Hghway Traffc Act [2] specfes the load restrcton lmt to be 5,000 kg per sngle axle. Vehcles exceedng ths lmt have to take alternatve routes or be subject to the penaltes descrbed n the Act. These weght restrcton polces are currently establshed usng hstorcal date thresholds and expert vsual observaton nstead of systematc and real-tme montorng of the pavement condton. Therefore, such polces can ether penalze the transportaton and resource ndustres each tme that the payload s unnecessarly restrcted, or lead to rreversble pavement damage when overloaded trucks are allowed on a weakened roadway. WWP certanly provde ncreased capacty, but n the absence of proper montorng of the roadway condton, delayed overloads can further compound pavement damage. To address ths problem, ths study proposes ste-specfc models that can relate a few clmatc varables to the frost depth and thckness n the pavement on the one hand, and (as part of some future work) to strength thresholds obtaned from hstorcal data and from pavement deflecton testng on the other hand. TARGET PAVEMENTS AND TYPICAL DISTRESSES Northern Ontaro roads manly dffer n the surface structure whle most of ther structural layer (from the base down to the subgrade) generally remans constant between pavement types [3]. Apart from a few roads whch are stll gravel-surfaced snce ther constructon n the 950s, most of Ontaro s secondary hghways have now become hardsurfaced (paved) wth asphalt-concrete [4]. Moreover, btumnous surface treatments may - 3 -

4 be found on some portons of those low-volume roads where the rejuvenaton of old asphalt-concrete surfaced pavements or the use of sealng coats was needed. Therefore, ths study addresses pavements that fall under one of the followng categores: Gravelsurfaced, asphalt-surfaced or surface-treated roads. The standard condton ratng schemes used n Ontaro by the Mnstry of Transportaton for the evaluaton of the road surface condton nvolve the use of specfc ndcators, such as functonal performance ( rdng qualty ) and structural damage ( extent and severty of the roadway s surface dstress ) [5]. The typcal surface dstresses assocated wth each type of road nvolved n ths project were dentfed to be: Surface roughness, ruttng, thermal fracture (transverse crackng), and fatgue crackng, composed of top down crackng (longtudnal crackng) and bottom-up crackng (reflectve crackng). MECHANISTIC-EMPIRICAL ANALYSIS Part of the work conssted n examnng how the performance of low volume pavements was affected by Northern Ontaro s specfc clmatc and traffc constrants, and to evaluate the benefts of mplementng load restrctons on those roads. The new Mechanstc-Emprcal Pavement Desgn Gude (MEPDG) software was selected as t s capable of evaluatng the performance of new and rehabltated pavements gven ther structural profle, and of calculatng the damage assocated wth varous clmatc and loadng condtons. The MEPDG was used to examne how load restrctons nfluenced the key pavement dstress ndcators of two types of low volume roads (asphalt and gravel pavements, both surface treated) over a twenty-year desgn lfe. The physcal condton of the pavement structure was evaluated n terms of fatgue crackng and ruttng and the functonal performance n terms of rde qualty and comfort, usng the Internatonal Roughness Index (IRI) [6]. Snce ts frst development n 997 by the AASHTO Jont Task Force on Pavement (JTFP) under Natonal Cooperatve Hghway Research Program Projects -37 and -37A [6], the MEPDG software has consderably mproved along wth progress n model accuracy and computer scence, as well as ncreased amount of avalable pavement data. The latest verson of the software (MEPDG v.0.9) became avalable onlne n July 2006 [7] and was used to conduct the analyss. Ths verson reflects changes recommended by the NCHRP -40A ndependent revew team, the NCHRP -40 panel, the general desgn communty, varous other re-searchers, and the Project -40D team tself [8]. In partcular, the Enhanced Integrated Clmatc Model (EICM) was greatly mproved. Thanks to ths feature, the software s able to calculate hourly temperatures and mosture wthn each pavement layer and wthn the subgrade over the entre desgn perod. A fnal verson ncorporatng addtonal enhancements (MEPDG v..0) should be released n the followng months. However, for the purpose of assessment, the avalable software was deemed to be approprate and sutable

5 The MEPDG mpose that the user nputs the pavement structure by layers and wth the nature and thckness of each layer. In the case of unbound materals, the ste-specfc seve analyss dstrbuton and the compacton state can be entered. In the case of bound materals (asphalt-concrete), the bnder and aggregates propertes can be specfed. In ths study, the MEPDG smulatons were carred out on a number three level of analyss as t s suted to low volume pavements and allows natonally calbrated values to be adopted when no nformaton s avalable [6]. The man purpose of the MEPDG smulatons presented heren was to assess qualtatvely the beneft of reducng loads durng certan perods of the year n comparson to other perods of year, wth the no restrctons case as the baselne. MEPDG nputs of nterest Two typcal structures were entered nto the software: One gravel road and one asphaltsurfaced road (as llustrated n Fgure on the next page and detaled n Table ). Both roads receved a thn asphalt overlay n year number ten of ther desgn lfe (startng arbtrarly from August 980), as thn overlays are commonly accepted as one of the preservaton technques for low volume roads pavements [9]. In accordance wth the Ontaro Provncal Standards Specfcatons (OPSS), a Superpave PG bnder was selected for these overlays. 2.5 cm Asphalt surface treatment 2.5 cm Asphalt surface treatment 2.5 cm Old asphalt overlay 2.5 cm Crushed gravel 5.2 cm Granular base 5.2 cm Crushed gravel base 50.8 cm Granular subbase 50.8 cm Granular subbase Clay subgrade Clay subgrade Fgure Target pavements (left: asphalt road ; rght: gravel road ) The typcal traffc volumes were retreved from the Mnstry of Ontaro hstorcal archves, n whch the Annual Average Daly Truck Traffc (AADTT) s defned as the - 5 -

6 twenty-four hour two-way traffc volume, averaged for the perod of January st to December 3 st [0]. The traffc volumes were then converted nto equvalent sngle axle loads wth default load dstrbutons for each of the 4 to 3 traffc classes. Smulatng seasonal varatons of traffc loadng throughout the twenty-year desgn lfe of the pavement was acheved by settng the monthly axle load dstrbutons factors equal to zero for all the loads exceedng 5,000 kg (,000 lbs to 4,000 lbs). Snce the total volumes were kept constant, the axle load dstrbuton factors had to be adjusted for loads 3,000 lbs to 0,000 lbs so that the sum of all the factors equalled 00 per cent. It should be noted that the MEPDG software does not allow the user to change the dstrbuton factors for perods of less than one month. Table 3 llustrates the varous scenaros that were selected and smulated wth the MEPDG software and performed for each of the four typcal sectons presented n Table. A total of sxty-four MEPDG runs were performed. The outputs of nterest and ther assocated desgn thresholds (as assumed n the MEPDG Gude for Flexble Pavements []) are summarzed n Table 2. These thresholds are commonly used as trggers for pavement repar, rehabltaton, and reconstructon decsons. To account for the varous uncertantes n predctng future pavement deteroraton, a relablty level of 75% was selected n accordance wth the Desgn Gude recommendatons for low volume pavements [2]. Results of the MEPDG analyss For each of the performance crteron used n the analyss (IRI, longtudnal crackng, reflectve crackng and total ruttng) and for each of the four typcal stes (Northeastern asphalt road, Northeastern gravel road, Northwestern asphalt road and Northwestern gravel road), tme to falure was computed and s presented n Table 4. Values greater than twenty years ndcate that the parameter lmts were not reached durng the twentyyear desgn lfe. In Fgures 2 and 3, the baselne case ( no load restrctons scenaro) s compared to the other load restrcton scenaros defned n Table 3, usng the termnal values (values at year twenty) of two of the performance crteron (IRI and ruttng). The purpose of Fgures 4, 5 and 6 s to evaluate and compare the ablty of each load restrcton scenaro to ncrease the pavement servce lfe. In short, none of the scenaros quanttatvely ncreased pavement servce lfe more than sx years. The results ndcated that n general, longer perods of load restrctons could protect the pavement nfrastructure, but one should keep n mnd that longer SLR perods mght not have an overall economc beneft. Nevertheless, the tme to falure varatons observed were qualtatvely relevant and a few conclusons could therefore be drawn from the results of ths MEPDG analyss

7 Frstly, t was found that the performance crtera provdng us wth the most nformaton about the mpact of load restrctons duraton and schedulng on the pavement servce lfe for both asphalt and gravel roads n Northern Ontaro were surface roughness, ruttng and longtudnal crackng. The IRI reached ts assocated crtcal threshold wth up to two years varaton dependng on the loadng scenaro. Longtudnal crackng changed consstently on Northeastern Ontaro asphalt roads wth up to sx years of servce lfe ganed wth scenaros #0, #3 and #5 as per Table 3. Mnmal dfferences were noted on gravel roads. Ths s not surprsng though, as gravel road performance s dffcult to model. Reflectve crackng dd not change regardless of the amount of load restrctons, or schedule. The duraton of load restrctons was made to vary between one and fve months n the analyss, whch revealed the followng: - The greatest gan n pavement servce lfe (based on IRI, ruttng or longtudnal crackng) s obtaned wth ether four or fve month load restrctons. Therefore, the fve-month SLR duraton appears to have no beneft. - For a four-month load restrcton duraton (scenaros # 9 and #3 as per Table 3), the March to June schedule s slghtly more effcent than the February to May schedule. - The three-month duraton (scenaros #8, #2 and #5 as per Table 3) s roughly one percent less effcent than the four-month duraton, and the March to May and Aprl to June schedules are equvalent. - The two-month duraton (scenaros #7, #, #4 and #6 as per Table 3) can be up to three percent less effectve than the three-month duraton n ncreasng the pavement servce lfe, and the March to Aprl and Aprl to May schedule are the most sgnfcant contrbutors. - The one-month SLR provdes lmted beneft. However, the analyss showed that load restrctons should mperatvely be n place durng Aprl, as ths month appears to be a major contrbutor to pavement preservaton. THE FROST DEPTH MODEL The second part of ths paper descrbes the results of the second phase of a two-year ongong study n Northern Ontaro. It nvolved the nstallaton and nstrumentaton of two expermental stes n Northeastern Ontaro (New Lskeard, Hghway 569) and n Northwestern Ontaro (Thunder Bay, Hghway 527). The purpose was to develop and calbrate a model that relates frost depth n the pavement of a low volume road located n Northern Ontaro to the ar temperature and pavement temperature on ths ste [3]. The calculaton of the Freezng and Thawng Index developed n the prelmnary study mentoned earler s summarzed n Equatons and 2. These ndces are beng used n the frost depth predcton models descrbed later. Ar and pavement temperature were - 7 -

8 retreved from the onlne database of the Road Weather Informaton System (RWIS), whch s a network of clmatc statons that enable mantenance personnel to make mantenance decsons based on real tme road data [3]. Envronment Canada forecasts wll be also used to predct pavement condton up to fve days n advance, based on the fve-day ar and pavement temperature forecasts. The Freezng Index (FI) s defned as follows [3]: FI FI FI + = T = FI T < 0 FI + = 0 () Where Number of days after the frst day of below 0 C ar temperatures T Noon ar temperature on day number (n C) FI Freezng Index on day number. The Thawng Index (TI) s defned as follows [3]: TI TI TI + = T = TI T + < 0 TI ref ( T + = 0 T ref ) (2) Where Number of days after the frst day of below 0 C ar temperatures T Noon ar temperature on day number (n C) TI Thawng Index on day number. T ref Calbraton parameter The reference temperature (T ref ) was obtaned by relatng daly ar temperatures wth the correspondng pavement surface temperatures [3]. It was found to be equal to mnus C on the New Lskeard expermental ste, and to mnus C on the Thunder Bay expermental ste. Based on the frost thckness model ntroduced durng the prelmnary study n 2005/2006, predctors for the lower and the upper frnge of frost were developed ths year and are presented heren. So far, they have only been calbrated for the Northeastern Ontaro regon usng the frost data collected on the New Lskeard expermental ste from November 2005 to May In ths secton, the man calculaton steps are summarzed. No results wll be presented concernng the Northwestern regon, as no consstent amount of data could be collected ths year. Instrumentaton and hardware ssues on the Hghway 527 expermental ste are beng currently nvestgated

9 In order to use the frost depth model, two major dates were dentfed. It should be noted that they are ste-specfc and that they are lkely to change from one year to the other. The frst perod starts on the frst day on whch the average ar temperature falls below zero degrees Celsus. It wll be ndexed as day 0 and wll trgger all the calculatons (Freezng Index (FI), Thawng Index (TI), frost depths). In the year 2005/ 2006, on the Northeastern expermental ste, ths day corresponded to the 0 th of November The transton from perod A to perod B (defned below) allows the thawng phenomenon to gradually overcome the freezng process. After day 0, the calculaton of frost depth changes from Equaton 3 (perod A) to Equaton 4 (perod B) and from Equaton 5 (perod A) to Equaton 6 (perod B). On Fgure 7 below, perods A and B are dentfed. - Perod (A): The mosture located n the pavement s freezng; the TI rses constantly. But as a result of a hgh number of freeze-thaw cycles n that regon, desgnated as a low-freeze area, warmer perods cause the surface ce to thaw and the TI to dsplay postve values. The combnaton of those two phenomenon llustrates Equaton 3 and Equaton 5 below. It can be noted that from Feb.8 th, 2006 to March 8 th, 2006, the pavement was freezng very rapdly and no warm days occurred after that; the TI stayed equal to zero. - Perod (B), startng from the day ndexed as 0 : Begnnng of the thawng season. The frozen layers of the pavement are thawng from the top-down but also from the bottom-up, what llustrates Equaton 4 and Equaton 6 below. FI, TI (deg.days) Freezng Index (FI) Thawng Index (TI) =0 Perod A = 0 Perod B 00 0 /0/05 /20/05 /30/05 2/0/05 2/20/05 2/30/05 /9/06 /9/06 /29/06 2/8/06 2/8/06 2/28/06 3/0/06 3/20/06 3/30/06 4/9/06 4/9/06 Days (from Nov.0, 2005) Fgure 7 Freezng and Thawng ndces on the Northeastern expermental ste (Hghway 569), year 2005/

10 The calculaton steps for the lower and upper depths of frost are summarzed below: Lower frnge of frost Durng perod (A) Durng perod (B) LFFD = 5.537* FI (3) LFFD LFFD * o 2 = TI (4) Where LFFD Lower Frost Frnge Depth for day LFFD 0 Lower Frost Frnge Depth for day 0 FI Freezng Index for day TI Thawng Index for day Number of days after day 0 0 Frst day after whch the TI rses consstently above zero Upper Frnge of Frost Durng perod (A) UFFD = 5.537*( FI TI ) 2 (5) Durng perod (B) UFFD UFFD TI 0 2 (6) Where UFFD Upper Frost Frnge Depth for day UFFD 0 Upper Frost Frnge Depth for day 0 FI Freezng Index for day TI Thawng Index for day Number of days after day 0 0 Frst day after whch the TI rses consstently above zero The depths of the upper or the lower frost frnge are negatve numbers and ther unt s the centmetre. If one depth value happens to be postve, t wll be replaced by zero. Moreover, as soon as the lower depth of frost and the upper depth of frost meet (whch means that there are no frozen layers wthn the pavement anymore) they wll be equalled and they wll reman constant. At the end of ths paper, Fgure 8 shows that the upper and lower depths of frost calculated wth ths model are close to the values measured on the New Lskeard (Hghway 569) expermental ste. In general, the amount of frost predcted s less than the amount observed on-ste, whch s sutable snce t wll translate nto more conservatve loadng recommendatons. The second season of data collecton s beng currently prepared. Expermental data wll be used to determne the coeffcents of calbraton denoted a and b n Equaton 7 below and further valdate the frost predctors for both Northeastern and Northwestern - 0 -

11 Ontaro stes. Therefore, future focus s to ensure that sensng, acquston and loggng of the subsurface pavement temperatures wll be acheved properly and wth as lttle dsrupton as possble. FFD =.537 *( a FI b TI ) (7) 5 Where FFD s the Upper or the Lower Frost Frnge Depth for day SUMMARY AND CONCLUSIONS The MEPDG analyss was drected at quantfyng how pavement dstresses were mpacted by varous loadng scenaros. The fndngs are encouragng and ndcate that the approprate tmngs can result n up to sx years addtonal lfe for the pavement. The proper selecton of the PG bnder s also very mportant. Although the MEPDG does need to undergo extensve calbraton for Canadan condtons, the results presented do seem reasonable and provde valuable nsght nto the mpact of traffc loadng on a pavement structure weakened by envronmental constrants. To fulfll the ultmate objectve of ths project, whch s to assst the Mnstry of Transportaton of Ontaro n makng effectve real tme Sprng Load Restrctons and Wnter Weght Premums decsons usng RWIS and Envronment Canada data, pavement performance wll have to be related as consstently as possble to the key clmatc varables provded by neghborng weather statons. Ultmately, both RWIS and Envronment Canada data wll be used to estmate frost depths on Northern Ontaro s low volume roads. FUTURE STEPS Based on the economc and structural performance crtera of the transportaton agences, frost depths thresholds wll be developed and used to trgger the mplementaton of load restrctons or surpluses. For ths purpose, hstorcal deflecton data wll be supplemented by actual testng usng a portable verson of the Fallng Weght Deflectometer (FWD) devce (Dynatest Prma 00 Portable FWD). Deflecton data wll be collected throughout the upcomng thawng season, from the begnnng of March 2007 to end of Aprl 2007, and t wll be used to correlate the pavement bearng capacty to the depth of frost n the sol. Crtcal frost depth thresholds wth the assocated pavement strength wll be dentfed: The least acceptable strength threshold wll trgger load restrctons; the least strength recovery threshold wll allow the removal of weght restrctons and even the placement of load surpluses. Moreover, a lfe cycle cost analyss wll be performed n order to fnd a consstent balance between the reducton of mantenance expenses and the ncrease of vehcle payload. - -

12 In the long term, ths methodology could provde Northern Ontaro s hghway offces wth an accurate means of predctng when to enforce or remove load restrctons on ther roads. For each partcular locaton, ar and pavement temperatures wll be determned by extrapolatng from the nearest RWIS and Envronment Canada statons and then used n the frost depth model to calculate and/or predct the on-ste depths of the upper and lower frnges of frost. Safety margns wll be determned based on an analyss of varance of the predcted data wth actual condtons. The research team wll contnue to collect data to ensure the models can be calbrated and further valdated n the future

13 REFERENCES [] Last accessed 03/2/2007. [2] Reduced Load Perods. Hghway Traffc Act, R.S.O. 990, Chapter H.8, Secton 22. [3] Tghe et. al. Development of Tools for Improved Sprng Load Restrcton Polces n Ontaro. Unversty of Waterloo, Cvl Engneerng, 2006 Annual Conference of the Transportaton Assocaton of Canada, Charlottetown, Prnce Edward Island, May [4] Last accessed 02/0/2007. [5] G.J.Chong, G.A.Wrong, SP-025- Manual for Condton Ratng of Gravel Surface Roads, Research Development Brand, Ontaro Mnstry of Transportaton, August 989. [6] Appled Research Assocates, Inc., ERES Consultants Dvson. Gude for Mechanstc Emprcal Desgn of New and Rehabltated Pavement Structures. NCRHP Project -37A. Washngton, D.C., [7] Last accessed 03/28/2007. [8] Research Result Dgest 308. Changes to the Mechanstc-Emprcal Pavement Desgn Gude Software through Verson 0.900, July NCHRP, September [9] Pavement Preservaton wth Thn Overlays. Better Roads, June [0] Last accessed 02/3/2007. [] Desgn of New and Reconstructed Flexble Pavements. Gude for Mechanstc- Emprcal Desgn of New and Rehabltated Pavement Structures, Chapter 3. NCHRP -37A, March [2] Low Volume Roads Desgn. Gude for Mechanstc-Emprcal Desgn of New and Rehabltated Pavement Structures, Part 4, Chapter 4. NCHRP -37A, March

14 [3] Desgn Relablty. Gude for Mechanstc-Emprcal Desgn of New and Rehabltated Pavement Structures, Appendx BB. NCHRP -37A, August [4] Last accessed 03/04/2007. [5] Last accessed 03/04/2007. [6] Last accessed 04/2/

15 TABLES Table Baselne scenaros. Typcal Regon Northwestern Northeastern Lattude Longtude Elevaton 37 m 40 m Pavement type Asphalt road Gravel road Asphalt road Gravel road Upper layer nature and thckness Superpave PG cm Crushed Gravel 2.5 cm Superpave PG cm Crushed Gravel 2.5 cm Base nature and thckness Granular A cm Crushed Gravel 5.2 cm Granular A cm Crushed Gravel 5.2 cm Subbase nature and thckness Granular A cm Granular A cm Granular A cm Granular A cm Subgrade Clay Clay Clay Clay nature and thckness Sem-nfnte Sem-nfnte Sem-nfnte Sem-nfnte Secton ID Hghway 527, Tertary Road 8 Hghway 569, S JCT Hghway 2-way traffc [0] 200 AADTT 500 AADTT % Trucks 00 Operatonal 56.3 km/h Speed Table 2 Analyss parameters used n MEPDG applcaton (20-year desgn lfe). Performance Crtera Lmt **Relablty *Termnal IRI (n m/km) % Asphalt longtudnal crackng (n m/km) % Asphalt reflectve crackng (n %) 45 75% Asphalt thermal crackng (n m/km) % Total Ruttng (permanent deformaton) (n mm) % * The ntal IRI was set at.578 m/km. ** The relablty level of a gven performance crtera s defned as the probablty that ths crtera s less than the crtcal level over the desgn lfe [3]

16 Table 3 Sprng Load Restrcton (SLR) scenaros. No SLR SLR n February SLR n March SLR n Aprl SLR n May SLR n June Scenaro # Scenaro #2 Scenaro #3 Scenaro #4 Scenaro #5 Scenaro #6 Scenaro #7 Scenaro #8 Scenaro #9 Scenaro #0 Scenaro # Scenaro #2 Scenaro #3 Scenaro #4 Scenaro #5 Scenaro #6 Table 4 Years to reach performance lmt for the four baselne stes. Baselne stes Years to reach the IRI performance lmt Years to reach the Longtudnal crackng performance lmt Years to reach the Reflectve crackng performance lmt Years to reach the Transverse crackng performance lmt Years to reach the Total ruttng performance lmt *NER AC road *NWR AC road *NER Gravel road *NWR Gravel road > 20 N/A N/A.7.7 ** Not ** Not Applcable Applcable >20 >20 >20 > > * (as per Table ) NER: Northeastern ste; NWR: Northwestern ste; AC: asphalt road. ** Reflectve crackng s defned as crackng occurrng n the Hot-Mx asphalt layer located below the asphalt overlay. Therefore, t s not defned on gravel roads, even f they have receved an overlay

17 FIGURES Surface roughness: IRI (m/km) Northeastern gravel road Northwestern gravel road Northeastern asphalt road Northwestern asphalt road Desgn threshold year desgn lfe (n months) Fgure 2 IRI changes over a 20-year desgn lfe ( no restrctons scenaro). Total ruttng (mm) Fgure 3 Northeastern gravel road Northwestern gravel road Northeastern asphalt road Northwestern asphalt road Desgn threshold year desgn lfe (n months) Ruttng changes over a 20-year desgn lfe ( no restrctons scenaro)

18 Load restrcton scenaro (as per Table 3) Gravel road Asphalt road Fgure Years to reach IRI threshold over a 20-year desgn lfe Lfetme (defned as the number of years to IRI falure) for the sxteen load restrctons scenaros (Northeastern regon)

19 6.0% 5.5% 5.0% Asphalt road Gravel road % decrease n termnal IRI 4.5% 4.0% 3.5% 3.0% 2.5% 2.0%.5%.0% 0.5% 0.0% Load restrcton scenaro (as per Table 3) Fgure 5 IRI changes over the sxteen scenaros (Northeastern regon). 40.0% 37.5% 35.0% Asphalt road Gravel road 32.5% % decrease n termnal ruttng 30.0% 27.5% 25.0% 22.5% 20.0% 7.5% 5.0% 2.5% 0.0% 7.5% 5.0% 2.5% Fgure 6 0.0% Load restrcton scenaro (as per Table 3) Total ruttng changes over the sxteen scenaros (Northeastern regon)

20 Depth from the pavement surface (cm) (From model) Upper depth of frost (From model) Lower depth of frost (Ste data) Upper depth of frost (Ste data) Lower depth of frost Days (from Nov.0, 2005) Fgure 8 Comparson between the predcted and the on-ste depths of frost (Northeastern Regon, year 2005/2006)