Using Passive Refrigeration To Stabilize Foundations In Cold Climates

Passively refrigerated piles and gravel pads were selected as the foundation system for the new hospital in Kotzebue, Alaska

Ashrae Journal; September 1993

Passive refrigeration using two-phase thermosyphons is being used to provide a stable foundation for the new $40 million hospital currently under construction at Kotzebue, Alaska. This project is reportedly the largest (in area) designed to date on a site underlain by highly variable and unbonded saline permafrost. The 80,000 ft2(7430 m2) hospital is a comprehensive healthcare facility that will service the residents of northwest Alaska. Two different types of foundation systems using thermosyphons for subgrade cooling are being used for the project.

Clearing drifts from installed piles.
Clearing drifts from installed piles.

By definition, permafrost is soil or rock whose temperature remains below 32°F (0°C) over at least two consecutive winters and the intervening summer.1 Permafrost soils prevail in the arctic and sub-arctic regions of North America as well as Asia and Europe.

Predominate permafrost soils are generally not thaw-stable and consolidate upon thawing. Typically, settlement of structures due to thawing permafrost is not uniform and differential settlements are high.

The most common technique used to prevent permafrost degradation (thawing) beneath a structure is thermal isolation.

Structures are constructed on piles or piers with an air space between the ground surface and the structure. The air space is typically 3 to 6 ft (1 to 2 m) high to allow natural ventilation between the base of the structure and the ground surface.

At-grade structures founded on thaw-unstable permafrost must rely on some form of subgrade cooling system to intercept heat that could thaw the underlying subsoil, or the structures must be designed to accommodate large differential settlements. Both active (mechanical) and passive refrigeration systems have been used for subgrade cooling beneath at-grade structures with varying degrees of success.

Subgrade cooling is also used to lower the temperature of permafrost soils to increase soil strength. The strength of permafrost soils is proportional to temperature. That is, the colder the soil temperature, the greater the strength of the permafrost.2

Two-phase thermosyphons are the preferred technology for the majority of projects that require subgrade cooling, having proved their reliability since 1960.3,4 Over 120,000 units were installed on the trans-Alaska pipeline system for subgrade cooling.1

Clearing drifts from installed piles.

Two-phase thermosyphons are relatively simple devices that transfer heat against gravity. The typical unit is constructed of pipe, closed at both ends, and charged with refrigerant.

The condenser (above ground) portion of the unit can be bare or finned, depending on heat transfer requirements. The evaporator portion of the unit can have almost any configuration as long as slope remains between the evaporator and condenser.

Refrigeration of the subgrade occurs when the condenser temperature is lower than the soil temperature at a depth where the liquid portion of the refrigerant is pooled.

Condensation occurs, initiating evaporation of pooled refrigerant and, hence, subgrade cooling. Condensate returns to the evaporator portion of the pile by gravity and re-evaporates, provided the temperature differentials still exist.

Thermosyphon units operate during the period when the air is colder than the ground, typically during the period from October through April.

Thermosyphons have been manufactured using propane, butane, CFCs, HCFCs, anhydrous ammonia and carbon dioxide for the refrigerant. The choice of refrigerant depends primarily on the allowable internal pressure capabilities of the thermosyphon vessel, the quality and stability of the refrigerant available, and the preferences of either the customer or the manufacturer.

Site conditions

The community of Kotzebue is located in northwest Alaska between Kotzebue Sound and a substantial brackish water lagoon. To the east and south of the lagoon, low hills dominate the landscape of the peninsula. West of the lagoon, a gravel spit extends upon which most of the community's development is clustered.

The entire Baldwin Peninsula lies within the region of continuous permafrost. The depth to the base of permafrost is approximately 240 ft (73 m) in the Kotzebue area.6;7

The gravel and sandy gravel deposits that form the Kotzebue Sound side of the spit are generally 30 to 40 ft (9to 12 m) in depth and are underlain by fine-grained deposits of sand, silt, clay and organics. The soils in the hills to the east and south are predominately ice-rich fine-grained deposits.

The transition zone between the coarse grained spit deposits and the fine grained ice-rich deposits are interfingered with varying quantities of all the soil types found in the area and are generally capped by a tapering strata of coarse grained material.

Salt (a byproduct of the marine and lagoonal deposition processes) is present in relatively high concentrations in the fine grained subsoils. The salt not only causes a freeze point depression of the permafrost, but more importantly, changes its rheological properties due to an increase in unfrozen moisture content over a broad temperature range.

Early structures in Kotzebue were small and used conventional footings on the gravelly soils near the beach where few foundation problems were experienced. As the community grew, the size of structures increased and new construction was forced eastward where the surface granular deposits were thinner and frequently interbedded with fine-grained lagoon sediments, organics and some ice.

Even on the better portions of the spit, as the size of the structures increased, settlement became a problem as the thaw bulbs beneath the structures penetrated into the ice-rich fine-grained materials.

Field and laboratory studies were conducted in 1986 and 1989 to document actual conditions at the eight-acre hospital project site.8,9 A total of 23 borings were drilled to depths ranging from 35 to 101.5 ft (11 to 31 m),

The subsurface conditions generally consist of a surface layer of tundra overlaying gravel that grades into gravelly sand, sand and silt. Clays and occasional thin gravel zones were encountered below approximately 70 ft (21 m). With the exception of the surface active layer, the soils were perennially frozen (permafrost).

The upper 15 ft (5 m) of gravel and sand generally contained ice-rich zones with ice crystals, random ice and stratified ice. Below 75 ft (23 m), segregated ice was observed ranging in thickness from 0.25 in. to over 6 in. (6 to 150 mm). No massive ice formations were encountered.

Temperature data, field observations and laboratory tests indicated that saline, marginally bonded and unbonded zones of soils exist throughout the area investigated. Most of the unbonded soils were observed between 30 and 80 ft (9 and 24 m) in depth: This also corresponds to those soils that typically contained high salt concentrations.

The soil conditions and characteristics important to the design of a foundation at this site are: soil temperature, thaw settlement characteristics, presence/absence of soils with a depressed freezing point(salinity), compressibility, and bearing capacity of thawed soils or unbonded permafrost, adfreeze/friction and shear strength of both frozen and thawed material, and the creep properties of frozen soils.

Two different foundation systems were used for different portions of the hospital complex. The main building structure is elevated approximately 5 ft (1.5 m) above the grade on piles. Entries, water tanks and a garage are being built at grade.

Piling system

The piling for the elevated hospital structure were critical to the overall structure's performance. Analyses indicated that, if properly designed and installed, either a deep driven pile system or a passively refrigerated shallow, slurried pile system could provide adequate support for the main structure.

Because no end-bearing strata were found at this site, the controlling element in both designs would be the long-term creep rate of the frozen soil adjacent to the pile. By contrast, such piles have a much higher capacity for resisting peak loads of short duration.

Creep is defined as the slow, progressive movement of a loaded frozen soil without thawing or consolidation (no total volume change). Movement is in the shearing mode; the soil "flows" rather than "settles".

Creep is caused by pressure-melting of Ice in the soil at points of soil grain contact, migration of unfrozen water to areas of lower stress, breakdown and plastic deformation of the pore ice, and rearrangement of soil particles. Frozen soils exhibit creep under long-term loads as low as 5% to 10% of their rupture strength.10

In a simple pile that has developed an adfreeze bond with the surrounding soil, creep manifests itself in the bond zone at the pile/soil interface. Both bond strength and creep rate are primarily a function of soil type, ice content (ice is a viscoelastic material), temperature and salinity.

Salinity causes freeze point depression (concentrations of 5 to 20 ppt yield depressions in the 0.5° to 2°F (0.3° to 1.1°C) range and, at a given temperature, salinity can dramatically reduce adfreeze strength and increase creep rates; even beyond accepted values for pure ice.

It is generally accepted that the presence of salt can adversely affect adfreeze designs by several orders of magnitude. In the final analysis, long-term capacity of the foundation system is a function of sustainable adfreeze bond strength (creep rate) and piles urface areas.

The design of an adfreeze pile system in frozen soils is usually developed using either empirical sustained adfreeze stress data published for various soil/pile types at varying temperatures, or by applying an accepted creep relationship to calculate total settlement rates over the life of a structure. For saline soils, the creep settlement approach is more appropriate.

Both pile systems considered were analyzed and priced. The passively refrigerated, relatively shallow pile system was determined to be the most economical for the project.11 This pile system develops bearing in the upper 20 to 30 ft (6 to 9 m) of the subgrade, where the soil is more uniform and lower in salinity. To achieve the design soil temperatures required to meet the established creep criteria, each structural piling was manufactured as a two-phase thermosyphon.

For this project, a total of 360 thermo helix-piles were manufactured. Each unit was built as an ASME U-stamped vessel using 20 in. (508 mm) diameter steel pipe for the shell and standard semi-elliptical heads for the end-closures.

A 24 in. (610 mm) OD by 20 in. (508 mm) ID helicoid flyte was attached to the lower 20 ft (6 m) of the pile shell. A 1.5 in. (38 mm) nominal piping loop (backup loop)was installed in each unit to facilitate active refrigeration if required. No external fins were installed on the condenser portions of the units.

A non-pressurized portion of pipe was added to the top of each pile to allow for installation tolerances and provide a cut-off length of 1.6 ft (0.5 m).

The piles were charged with R-22 and the backup loops were charged with carbon dioxide. The piles were installed during November and December of 1992 in 30 in. (762 mm) diameter drilled holes. The annulus between the pile and in-situ material was backfilled with a sand/water slurry that was placed as though it was high strength concrete.

Upon freezeback, the helicoid flyte on the piles ensured that a shear bond would exist between the piles and the slurry backfill. Because the backfill was engineered to be stronger than the in-situ material, the potential shear (creep) failure is at the edge of the drilled hole.

The piles were designed to maintain a ground temperature profile equal to or colder than 27°, 26° and 25°F (- 2.8°C,  -3.3°C and -3.9°C) at depths of 5, 12 and 20 ft (1.5, 4 and 6 m), respectively, below the 4 in. (102 mm) of polystyrene board insulation that was placed 1ft (0.3 m) below the ground surface on any day between September 1 and September 15 of any year. This would be the maximum end-of-summer ground temperature profile.

Additionally, ground temperatures were to be equal to or colder than 22°, 23° and 26°F (-5.6°, -5.0° and –3.3°C) at the above referenced depths between March 1 and March 15, 1993, so that loading (construction of the building shell) could proceed. The installed pilings have a nominal rating of 65 kips (290 KN); see Figure 1.

Design average heat removal rates were approximately 500 Btu/h (146 W) per pile for the winter period. This corresponds to pile conductances of 15 to 90 Btu/h - °F (4° to 26 W/°C) for air velocities between 0 and 15 mph (0 to 7 m/s). The key to achieving the design conductances is in providing enough refrigerant to maintain a wetted turbulent flow condition on all internal surfaces.

The exposed portion of each pile was painted with a white fusion bond epoxy to optimize emissivity and reflectivity characteristics. No fins were used on the condenser ends of the piling.

Calculations indicated that too much cooling could create frost heave forces capable of displacing large sections of earth upward. With significant amounts of unfrozen pore water beneath the piles in the high salinity zone, this phenomenon could occur.

In an effort to monitor the long-term thermal regime of the soils under the new building, thermistor strings were installed adjacent to several piles. By recording subsurface soil temperatures on a regular basis, variations from expected and required soil/pile temperatures can be noted. Soil warming near a particular pile could be caused by such problems as a loss of refrigerant charge or chronic, localized snow drifting that has reduced stick-up conductances.

Because slight increases in temperature would most likely be associated with accelerated long-term creep versus short term failure, corrective actions could be taken before the structure is adversely affected.

By February 4, 1993, the thermistor strings verified that the specified March temperature profile had indeed been met. In fact, on that date, no monitoring thermistor temperature was warmer than 16°F (-9°C).

With moderating weather, erection of the structural steel for the main hospital structure began in May 1993. The building uses conventional arctic construction, with a steel frame and insulated wall panels. The single program floor of the hospital will be approximately 14 ft (4 m) above grade because of the separation between the base of structure and grade and the 9 ft (3 m) high maintenance/utility structure beneath the finished floor.

At-grade foundation systems

The two options explored to gain access to the structure included raised grade entries and at-grade entries. The raised grade option involved either large pile supported deck and ramp structures, or earth structures to provide ambulance and pedestrian access.

The structural deck and ramp structures were clearly preferred, allowing ready vehicular and pedestrian access without causing massive snow drifting problems. Unfortunately, this option was cost prohibitive. The earth structures presented a major snow drifting potential that was unacceptable.

At-grade entries with conventional spread/continuous footing systems were analyzed. On permafrost, these systems are generally subject to excessive settlement without site modification. The modification techniques most often used include: over-excavation and replacement with non-frost-susceptible gravel fill; pre-thawing and consolidation; or construction of refrigerated gravel pads.

The over-excavation design would require 30 to 40 ft (9 to 12 m) of in-situ material to be replaced. Lack of a good gravel source in Kotzebue and timing of the earthwork so as not to thermally affect the installed piling made this alternate unattractive.

Likewise, the time required for pre-thawing and consolidation along with thermal effects on the adjacent piling made the second alternate unacceptable.

Refrigerated gravel pads are commonly used in the Arctic to support moderate size at-grade structures using thermosyphons with near-horizontal evaporators laid at a nominal slope beneath the insulated floor. This alternate was economically and technically feasible for the at-grade entries and the associated garage structure and water tanks.

At-grade foundations using thermosyphons have three main components that must be optimized during the design: insulation; non-frost-susceptible (NFS) fill; and thermosyphons. These components are usually installed with the insulation separating the floor from the NFS fill and the evaporator sections of the thermosyphons placed in the NFS material at 3° to 6° slopes on 7 to 16 ft (2 to 5 m) centers.

These passive at-grade systems provide seasonal (winter) cooling to maintain permafrost at the bottom of the gravel pad and below. In winter, the thermosyphons withdraw heat from below the building and freeze the NFS pad. At the end of the cooling season (spring), the thermosyphons become dormant and the heat loss from the building begins to warm the pad from the top down.

Over the summer, any thawing that occurs must be contained within the NFS pad. With the onset of winter, the thermosyphons begin cooling the subgrade and the cycle continues (see Figure 2).

Thermosyphons have been used to provide subgrade cooling for at-grade structures as large as 70,000 ft2 (6500 m2) and have been installed with conventional evaporators as long as 195 ft (59 m).

Proprietary one-directional loop type evaporators as long as 1,283 ft (391 m) are presently operating in Kotzebue on a system with a total combined evaporator length of 14,436 ft (4400 m). 12 Condensers are sized to fit evaporator configurations with fin areas ranging from 30 to 3,400 ft2 (3 to 316  m2).

The entries and associated at-grade structures for the new Kotzebue Hospital were designed with a conservative attitude befitting the mission of the facility. A thickness of 10 in. (254 mm) of extruded polystyrene board insulation was used beneath the level of the conventional footings to limit heat transfer and thawing of the 6.5 ft (2 m) minimum thickness of NFS material below the insulation.

To compute thaw depths into the NFS material, the Modified Berggren Method was used.13 Job performance specifications required that thaw not penetrate into the lower 2 ft (0.6 m) of the NFS material. Design maximum summer thaw was 3.8 ft (1.2 m) below the base of the insulation.

Performance data for thermosyphons fitted with 3in. (75 mm) nominal evaporators and 70 ft2 (6.5 m2) finned condensers were used to compute an equilibrium spacing for the systern.3 For redundancy, a safety factor of two was used on the number of units required. Spacing of evaporators ranged from 8 to 10 ft (2.4 to 3 m) depending on the configuration of the structures being protected.

Average freezing season design heat transfer rates for the thermosyphons beneath the at-grade foundations range from 609 to 835 Btu/ h (178 to 245 W) depending on the length and spacing of evaporators. Thermosyphon conductance is primarily a function of the airspeed across the finned condensers and ranges from 36 to 145 Btu/h•°F (19 to 77W/°C) for air velocities of 0 to 15 mph (0 to 7m/s).

The average winter air velocity in Kotzebue is 13 mph (5.8 m/s). However, because of the proximity of the buildings and the potential for wind blockage, a more conservative 5 mph (2.2mls) was used for design.

The installation of NFS fill, thermosyphons and subgrade insulation will proceed in the Summer of 1993. Construction of the at-grade structures associated with the project is scheduled for the spring of 1994 after complete freezeback of the NFS fill to ensure minimal movement of the foundations over the life of the structure.


Passively refrigerated piles and gravels pads were determined to be the most appropriate foundation system for the new hospital at Kotzebue, Alaska.

The two-phase thermosyphons used are simple yet elegant devices with no moving parts, no power requirements and no replenishment of working fluid. They remove approximately 1.2 x 109 Btu (3.5 x 105 kWh) annually from the subgrade to stabilize foundations on saline permafrost.


By Edward Yarmak, Jr., P.E. and Jay A. Farmwald, P.E.


1.Johnson,G. 1981. Permafrost-Engineering, Design and Construction. Ottawa, Canada: National Research Councilof Canada.

2. Vialov, S. 1965. Rheological Properties and Bearing Capacity of Frozen Soils. CRRELTranscript74. Hanover, New Hampshire: U.S. Army Cold Regions Research and Engineering Laboratory.

3.Heuer, C , etal. 1985. "Passive techniques for ground temperature control:' Thermal Design Considerations in Frozen Ground Engineering. New York, New York: American Society of Civil Engineers. pp. 72-154.

4. Richardson, P. 1979. "Tough Alaska conditions prove new pile design's versatility" Alaska Construction and Oil. Seattle, Washington: Vernon Publications. February, pp. 20-28.

5. Heuer, C. 1979. The Application of Heat Pipes on the Trans-Alaska Pipeline. CRREL Special Report 79-26. Hanover, New Hampshire: U.S. Army Cold Regions Research and Engineering Laboratory. July.

6. Brown, 1.,Pewe, T. 1973. "Distribution of permafrost in North America and its relationship to the environment: A review 1963-1973:' Proceedings of the 2nd International Conference on Permafrost-North American Contribution. Washington, DC: National Academy of Sciences. pp.71-100.

7. Ferrians, O. 1965. Permafrost Map ofAlaska. Washington, DC. US Geological Survey.

8. Shannon & Wilson Inc. 1990. Foundation Engineering Studies: Proposed New Kotzebue Hospital. Fairbanks, Alaska.

9. Shannon & Wilson Inc. 1990. Geotechnical Field and Laboratory Studies; Proposed New Kotzebue Hospital. Fairbanks, Alaska.

10. Smith, D. 1986. Cold Climate Utilities Delivery Manual. Montreal, Quebec: Canadian Society for Civil Engineers.

11. Farmwald, 1. 1992. "Passively refrigerated adfreeze piles; A foundation system for.the new PHS hospital in Kotzebue, Alaska:' Presented at USPHS professional meeting in Cincinnati, Ohio. May.

12. Long, E. Yarmak, E., 1988. "A subgrade cooling and energy recovery system:' Proceedings of the Fifth International Conference on Permafrost. Trondheim, Norway: Tapir Publishers.

13. Lunardini, V. 1981. Heat Transfer in Cold Climates. New York, New York: Van Nostrand.

About the authors

Edward Yarmak, Jr. is the chief engineer of Arctic Foundations Inc., Anchorage, Alaska. He received a BS in civil engineering from the University of Alaska at Fairbanks and a MS in engineering from the University of California at Berkeley. He is a member of the American Society of Civil Engineers, the National Association of Corrosion Engineers, the Society of Manufacturing Engineers, and the Construction Specification Institute. He is currently serving as the Alaska Section chairman for NACE.

Jay A. Farmwald, is the chief of planning, design and construction for the Alaska Area Native Health Service section of the U.S. Public Health Service, Anchorage, Alaska. He received a BS in environmental engineering from the U.S. Air Force Academy, a MS in civil engineering from Stanford University and a MS in arctic engineering from the University of Alaska. He is a member of the American Society of Civil Engineers and currently serves as the secretary of the ASCE national awards committee for the Technical Council on Cold Regions Engineering.