ROAD EMBANKMENT DESIGN ALTERNATIVES OVER PERMAFROST ' By David C. Esch,' , ABSTRACT The side slopes of roadway embankments in warm permafrost areas often cause particularly severe problems from long term thaw-related settle- ments. This situation results primarily from the snow plowed from the roadway, which thickens and increases the insulating cover on the side slopes. Soils underlying the slopes therefore often do not totally refreeze each winter, as normally occurs beneath the cleared portion - of the roadway. These annually deepening thaw zones beneath the slopes result in progressive settlements of the outer edges of the roadway and often cause longitudinal cracking of the roadway surface. Experimental installations of air convection ducts, in conjunction with insulation layers and embankment toe berms, were made during 1973 and 1974 on a newly constructed roadway embankment approximately 25 miles west of Fairbanks. Performance has been monitored since that time. Results through the thawing season of 1977 are presented, and demonstrate that satisfactory annual refreezing can be assured by this method, preventing long term thaw-settlements of snow covered embank- men t slopes. INTRODUCTION: Previous studies by the Alaska Department of Highways of roadways constructed over permafrost have shown that progressively deeper thaw- ing beneath the roadway side-slopes often still occurs, even in areas where the embankment thickness beneath the roadway centerline is ade- quate to prevent thawing into the underlying permafrost. Because the roadway side sl'ope areas are well drained and generally only sparsely vegetated, their mean surface temperature during a given thawing sea- son will generally be higher than that of a well vegetated surface overlying permafrost. This will result in deeper thawing beneath the side slopes, an increase in the thickness of the annual thaw layer, and consolidation and settlements if the underlying permafrost has a moisture content higher than it can retain in a thawed state. Since a considerable portion of a roadway embankment may be supported struc- turally by the soils beneath the side slopes, any settlement in these areas will be reflected by embankment slumping and lateral spreading and cracking (Fig. 1). Because the snow cover on the side slopes is increased by the snow plowed from the roadway surface, the side slopes will also not cool and freeze back as efficiently as either the adja- cent undisturbed ground or the travelled roadway surface areas. There- fore, annually deeper thawing of underlying permafrost often occurs beneath the roadway side slopes. 'Engineer of Tests, Alaska Department of Transportation, Fairbanks 1 *! J i Literature reviews indicate that very little study has been given to side slope effects of roadways over permafrost. The problems which result from side slope thaw are intensified by increased fill heights and probably by the steeper side slopes comonly used on high fills for economic reasons. To study the benefits of several alternate embankment slope designs in controlling thaw of permafrost beneath roadway side slopes, a research project was instituted in 1973 under funding of the Highway Planning and Research program of the Federal Highway Administration. Six different combinations of insulation layers, toe berms, and air ducting systems were installed on a newly constructed roadway and have been monitored since the completion of construition in 1974. Thaw and Consolidation Zones Fig. 1 .--Typical Roadway Distress from Thaw beneath Embankment Side Slopes. PROJECT LOCATION AND DESCRIPTION The site selected for the experimental embankment construction is located at Bonanza Creek, approximately 25 miles west of Fairbanks, on the "Parks Highway"; the primary trucking route between Fairbanks and Anchorage. In this area, a new roadway segment was to be routed across a generally undisturbed muskeg, underlain by ice-rich silt permafrost soils, with segregated ice. The variations in frozen water contents of the silt are shown by Figure 2. Overlaying the permafrost was a thin (1' to 2') peat layer, covered with a surface layer of sphagnum moss, with some areas of sedge covered tussocks also present. Vegetation consisted of scattered black spruce, tamarack, willow, and alder. In the interior of Alaska, such muskeg environments are ideal for the protection of permafrost at shallow depths. This site was selected because the new routing required a relatively high and uniform embankment height, ranging from 22 to 25 feet, over very poor soils. The critical nature of these soils was evidenced by the old roadway embankment crossing this muskeg approximately 500 feet to the north, which had exhibited distress similar to that depicted in Figure 1. On-site soil temperatures were monitored for one year prior to construction, and indicated that the permafrost in this area ranged from 28O to 30°F in mean annual temperature. 2 ' .\ (Frozen) Moisture Content Yo .. 0. .*. 0 . e .. .. e . m . -0 *-.. 0. Fig. 2--Frozen Moisture Contents of Silt FounZation Soils DESIGN OF EMBANKMENTS: Thermal analyses of the proposed embanknent, using the "Modified Berggren" calculations approach (l), indicated that thawing beneath the cleared roadway portion of the embanhent would reach to a depth of 12 to 15 feet, but that thaw beneath the lower portions of the side slopes would penetrate into the underlying permafrost. Three different basic types of modifications were considered for field evaluations to retard this side slope thawing. These included the use of rigid foam insulation layers, the construction of toe berms, and the installation of air convection ducts. Insulation layers have been utilized beneath roadways and airfields, to prevent thawing of underlying pernafrost, (2), (3), in addition to xuch more common usage beneath roadways in warmer regions for frost heave prevention. Both experience and thermal calculations have demonstrated that insulation layers are most beneficial when placed at or near the ground surface, and installed with a layer of thaw-stable soil between the insulation and the soil to be protected from thawing. An additional consideration made in ezbankiaent designs for this project xas that the insulation layers be nearly horizontal, both to simplify construction operations, and to preveat the insulation from creating a slippage plane, as might occur if it were installed parallel to the side slopes. Insulation layer widths were selected from one-dimensional thaw depth analyses and estimations of thaw depths within the embank- Zen t . Toe berms were designed with a thickness of six feet, the calculated thickness required to prevent the first year's thaw from penetrating beneath the berm. The berm width of twenty feet was selected to pro- vide a major contrast to the no-berm condition, with a 100 foot transition length to permit studies of the effect of berm width. Berms ;:ere designed for use of excess silt caste materials from an adjacent section. The air convection ducts were included in this study to determine their benefits in removing heat and thereby refreezing the soils be- 3 neath the insulative snow cover. Because of the many variables which affect the efficiency of convective flow air ducts, the ducting system was not fully analyzed in the design stage, but was selected, installed, and experimentally evaluated. A detailed discussion of air ducts as related to building foundations is presented in Reference 1. The ducts function in convection only in winter, as the adjacent soil warms the air in the duct and causesit to rise up the exhaust stack, at the same time drawing in cool air at the inlet. Two different layouts for the ducts were utilized on the north and south sides of the embankment, as shown by Figure 3. The ducts were constructed from 8 inch galvanized corrugated metal pipe, installed with the inlet ends above the depth of maximum snow cover, and with 10 foot high exhaust stacks. The 50 to 100 foot long buried portionsslopedslightly upward toward the exhaust stacks. Inlet end dampers were provided to permit positive summer air flow prevention, and rain and snow hoods were added. Fig. 3--Overall View of Embankment Details CONSTRUCTION OPERATIONS: Prior to construction, the embankment area was hand cleared, and all small trees and brush were placed in an even layer beneath the embank- ment. To provide a working pad for the experimental installations, and to assure a well frozen embankment foundation, the initial three feet of the embankment was constructed in September of 1973, and periodically cleared of snow during the subsequent winter. This resulted in lowering the soil temperatures to 25' to 26'F at a depth of 15 feet. In early April of 1974, prior to the start of seasonal thawing, the berms were constructed, and insulation layers and convection ducts were installed. The remainder of the embankment fill was conpleted by early June, 1974, followed by pavement pla e.ent in July of 1975. Insulation Was expanded polystyrene foam, HI-35CRy by DOW Chemical co. 4 INSTRUNENTATION DETAILS \ The thermal performance of the various design features is evaluated from temperatures measured with a system of 352 thermocouples and 20 thermistors installed both in horizontal strings and in 18 vertical borings made through the embankment. TherEocouples are of the copper- constantan type, and are connected through double pole rotary switches located on roadside posts. Readings of both thermocouples and thermi- stors are made monthly with a Hewlett-Packard >lode1 3465B Digital Volt- Ohm meter, using an ice-bath reference junction for the thermocouples. The combination of these two sensors was selected to take advantage of the accuracy of better than +. O5OF from calibrated thermistors, while avoiding the high cost of a Full thermistor installation. Accuracy of the thermocouples used has generally been better than +0.3'F, as judged by comparisons with adjacent thermistors and agreement between temper- ature indicated and probe measured thaw depths. Sit& air temperatures and wind velocities are obtained by battery operated recorders housed in a Weather Bureau type weather station.