Ground freezing

What it is

Natural frost occurs many places around the world, in the top layer of the ground during wintertime. Far north and on the South Pole, there is permafrost. When the frost is created by humans, as a method for civil and earth works or in the mining industry, we use the term artificial ground freezing (AGF) or just ground freezing.

Ground freezing is a method utilizing property changes when the temperature is lowered below the freezing point. The ground becomes waterproof and the strength increases significantly. The method is normally used as temporary support and waterproofing, or just as a waterproof barrier against water flow through the ground.

Completely waterproof

When the temperature in the pore water in the ground drops below its freezing point, the water changes phase from liquid to solid. The free pore water freezes first, while the layer closest to the mineral grains has chemically bound water freezing last and that can remain unfrozen even at very low temperatures. Still, for all practical purposes, the ground is 100 % waterproof, also with respect to pore pressure and groundwater control.

Even above the groundwater level, frozen ground becomes waterproof as ice is built onto the frozen material, either from moist in the air or from entering water. It is the balance between added and removed heat that determines whether the frozen body expands or contracts.

Since ground freezing provides completely waterproof constructions, the method is excellent to preserve the groundwater level and pore pressure on the outside of for instance a construction pit or tunnel, and thus avoiding settlements caused by pore pressure reduction in the surroundings.

High strength

Frozen ground represents a complete multiphase system which constitutes solid mineral grains, viscoplastic ice in the pores, liquid as chemically bound water around the mineral grains and gas with different saturation.

Compression strength versus temperature for different soils and ice, after Sayles (1966)
Compression strength versus temperature for different soils and ice (after Sayles 1966).
The strength increase when the ground freezes is caused by the phase change from water to ice. After the free pore water has frozen, more and more of the bound water will freeze as the temperature decreases. Most materials get higher strength with lower temperature, including frozen soil. Since temperature partly determines the strength, it is a key design parameter.

Removing heat

Energy flows from higher to lower energy levels. In order to freeze the ground artificially, a temperature gradient must be created, so the heat can flow towards colder regions and get removed from the lowest level. This is normally done by putting freezing pipes in the ground and circulate a colder medium, which both creates the temperature gradient in the ground and continously removes the heat entering the pipes.

Different materials have different thermal properties and the ability to conduct heat for different soil and rock types vary strongly. This affects the required initial freezing time. Temperature monitoring is part of the quality control to ensure the temperature criteria have been fulfilled according to the design.

Installation of freezing pipes

Because ground freezing often is done in unstable ground, it is common to install casing during drilling. The freezing pipes are fitted to the freezing system used and can be open single pipes, parallel or coaxial double pipes.

Open/direct freezing system

The simplest system is open freezing pipes down in the ground. A cold fluid with low boiling point is poured in the pipe where the heat from the ground makes the liquid boil. The evaporated gas transports the heat out of the system, and it is common to use liquid nitrogen, which is extracted from the air at air separation plants. Air contains 78 % nitrogen. As the liquid boils and evaporates, more liquid is added. The method is often called nitrogen freezing. In an open freezing system, the medium is both a refrigerant and coolant.

Nitrogen boils at -196 °C, and one liter liquid expands to 672 liter gas at atmospheric pressure and 0 °C. Nitrogen does not have any odor or color. The density compared to air at same temperature is 0.967, but because the gas is colder than the surrounding temperature when let out in the air, it will be denser. It has to be planned where and how the cold gas is let out, so it will not concentrate too much at low points and displace air and offset the oxygen concentration, which is a potential suffocation risk. Oxygen meters are always used as a measure to mitigate this risk during nitrogen freezing.

Figure: Principle for nitrogen freezing.
Figure: Principle for direct freezing with nitrogen.

Closed/indirect freezing system

In a closed system, the liquid circulated in the pipes will normally only transport heat from the pipes to a heat exchanger in a freezing plant. The liquid circulated must not freeze at working temperatures. A common liquid is brine, and the method is therefore termed brine freezing.

A brine often used is a mixture of water and CaCl2. This salt is a main constituent in seawater and is regularly used on our roads for thawing ice during the winter and to bind dust during the summer. The brine is cooled down to a temperature in the range -20 °C and -40 °C. Freezing by means of a freezing plant heat exchanging with a coolant circuit is called indirect freezing.

Inside the freezing plant the refrigerant is circulated in a separate circuit where the cooling process takes place: Evaporation, compression, condensation and choking. The compressors and pumps are powered by electricity.

The heat from the freezing plant must be removed and happens by direct water cooling, cooling tower or dry cooler. In the end the heat is let out to the water or air in the environment.

Figure: Principle for brine freezing.
Figure: Principle for indirect freezing with brine.


The freezing plants can use a variety of refrigerants, depending on temperature range and purpose. In the past, refrigerants based on chlorine fluoride compounds were commonly used, but due to their negative climate effects (several thousand GWP), but they are now banned in more and more countries.

The best refrigerant for the climate is the natural media ammonia, NH3, which does not affect the climate or the ozone layer (Ozon depletion product, ODP=0, Global warming potential, GWP=0). In high concentrations in air it is toxic to humans. If let out in water, too high concentrations is toxic to fish and other sea creatures. However, the media is naturally present in nature and small concentrations do no harm. The refrigerant has high efficiency and is commonly used for both ground freezing and other commercial purposes.

Another natural refrigerant which has gained popularity recently is carbon dioxide, CO2, which also has zero ozon depletion, and a global warming potential of one (ODP=0, GWP=1).

When and why

Excavation below groundwater table

If water leaking into the excavation is a problem or water ingress will make the ground unstable and unsafe, then ground freezing should be considered for waterproofing or stabilization. Likewise, if the surroundings are sensitive to settlements and cannot sustain groundwater draw down. Ground freezing provides a safe and stable construction pit or soil tunnel with full groundwater control. The groundwater influence area is limited to the construction area itself.

Temporary support

Ground freezing is a temporary aid, used to get something done or built. The fact that the ground thaws and goes back to its original unfrozen state after use is a big advantage in all cases where it is desired to remove the construction after use and is not part of the permanent construction. This can for instance be temporary support extending into a neighbor property, blocking of natural groundwater flow, temporary underpinning of parts of a building on direct foundation or closing of waterfilled tunnel for maintenance and refurbishment.

Permanent as part of a heat pump

More and more heat pumps have been installed throughout the last decades, collecting heat from the ground. The needs of ground freezing and heat production can be combined, reducing overall costs. In arctic areas ground freezing is used to ensure a permanent stable foundation, utilizing the heat drawn from the ground to heat the buildings. Same principle is used for instance when artificially frozen ice rinks are combined with heating of swimming pools or soccer fields.

Comparison of methods

External assessments

According to the textbook "Ground engineering equipment and methods", published by Frank Harris in 1983, sheet piles, injection, groundwater lowering, and compressed air are the preferred choices among Europe's contractors. But sheet piles were reported to be too expensive for depths greater than ca. 20 m and specially if the area exceeded 200 - 300 m2. Compressed air was only practically feasible with water pressure less than 35 m head. Groundwater lowering and injection were only possible in a limited range of grain sizes. Such limitations are not applicable to the artificial ground freezing method.

Comparison of methods depending on the soil's permeability was done by J.S. Harris in 1995. The figure below shows artificial ground freezing works irrespective of grain size, soil type and permeability.

Figure: Applicability of geotechnical processes according to soil type and permeability (after J.S.Harris 1995).
Figure: Applicability of geotechnical processes according to soil type and permeability (after J.S. Harris 1995).

Technical Designbasis for InterCity, BaneNor, 2019-08-15, says in chapter 8.4.1 about soil tunnels:

"Tunnels in rock with very low Q-values are normally excavated with rock bolts, pipe umbrella or spiling. Often a similar concept is tried when overburden is low or lacking. A big difference for tunnels in soil is that the soil usually has to be stabilized in advance of the tunnel excavation and that the stabilized mass cannot be part of the permanent construction."

Normally, and always below groundwater table, a waterproof construction is needed in the permanent condition. In soil injection is generally not suitable as waterproofing method and unsuitable as stabilization method.

Part of Bane NOR's soil tunnel definition is tunnels where only the upper part lacks overburden.

Figure: Different techniques for soil improvement.
Figure: Different techniques for soil improvement (after Bane NOR).
Figure: Methods for tunneling through different soils.
Figure: Methods for tunneling through different soils (translated after Bane NOR).
Figure: Methods for tunnel excavation after ground improvement.
Figure: Methods for tunnel excavation above and below groundwater level after ground improvement (translated after Bane NOR).

Feasibility of different ground improvement methods for different ground conditions. Ground freezing can be used for all conditions.

Figure: Applicability of different methods in different soils.
Figure: Applicability of different methods in different soils (translated after Byggegropveiledningen, Norwegian Geotechnical Society).

Geofrost's assessments

Geofrost's assessment of various methods' feasibility for tunneling and construction pits is shown below.

Figure: Method comparison for tunnel support.
Figure: Method comparison for tunnel support, Geofrost's assessment.
Figure: Method comparison for construction pits.
Figure: Method comparison for construction pits, Geofrost's assessment.

Grouting does not work as a waterproof seal in soils

Grouting in rock fissures and fractures

Pregrouting in rock is a well-known method to control water and stabilize fractured rock. This is the dominating mitigation in tunneling in Norway where major parts of the tunnels are in hard rock.

The fissures may be regarded as a two-dimensional plan where the rock on each side gives confining resistance to grout penetrating it. The further from the injection point the larger area to cover and the pressure drops accordingly resulting in range limitations. The different grouting mixtures have different particle sizes and therefore different possibilities to penetrate the smaller/finer/thinner fractures. The grout goes where the resistance is least, larger fractures before smaller, smooth and plane fractures before rough and winding ones, all the time seeking the less hydraulic resistance. Grouting in general works very well, if dealing with fractures in rock.

Grouting in soil

On the European continent, grout in frictional soil is used for many purposes, but not for waterproofing and stabilization of excavations below groundwater table. It is distinguished between grout compressing the ground (compaction grouting, hydraulic fracturing) and grout without deformation in initial grain structure (permeation grouting, fissure grouting, bulk filling).

The pores in soil are the space between each single mineral grain. These are connected to each other through narrow passages, where there is no direct contact grain to grain. The pore volume in soil may be regarded as a three-dimensional space. When the grout is to be spread in the three-dimensional space instead of in a two-dimensional fissure, the pressure drops much faster as distance from the source increases.

The grout, which follows the path of least resistance, will have large problems dividing and distributing evenly to cover all empty spaces. When the ground does not absorb the grout, the desired effect is not achieved. In sand and finer soils, it is impossible to force the grout into all the small pores with pressure.

Photo: Grout lumps in gravel.
Photo: Grout lumps (orange) in gravel.

If the pressure is increased, splitting may occur, and the grout follows the crack instead of being distributed into the pores.

Photo: Grout fingers silt.
Photo: Grout splitting fingers (orange) in silt.

Not even grout with ultrasmall particles or viscous fluid/gel works for waterproofing of soil. The permeability stops it. With a permeability (hydraulic conductivity) of 10 -3 cm/s (typical for medium sand), and a hydraulic gradient of 1, even water will not be able to penetrate further in than 3.6 cm/hour. The result is mainly compression.

It is generally difficult to grout in soil and impossible to ensure even distribution in the soil with a waterproof result.

Even though grouting can reduce leakages to be handled by pumps, it is not the right method to maintain the pore pressure outside the excavation where it is important to avoid settlements. The grout may compress the ground by splitting and increase stability above the groundwater level, but below it cannot exclude washout and collapse.

Environmental benefits

CO2 and climate

Ground freezing has generally a low CO2 footprint. Using Geofrost retaining walls in Norway for one year, the CO2 footprint will be reduced by 80 % compared to a AZ-48 sheet pile wall.

Energy and use of resources

Making of construction materials requires energy, both for frozen ground and steel or concrete. A comparison between sheet piles and Geofrost retaining walls shows equivalent energy consumption for freezing half a year and using a light GUGN sheet pile. Similarly, freezing almost 2 years equals the energy needed to produce the steel for an AZ48 sheet pile wall. For a wall of Ø406 x 10 mm piles equivalent freezing time to consume the same amount of energy used to make the steel is almost 3 years.

Material transport is reduced to transportation of freezing pipes since the ground itself is the construction material. The freezing pipes constitute a small share of the earth support volume.

Noise and vibrations

Noise and vibrations are limited to drilling of holes for freezing pipes. Freezing plants and equipment are designed to meet applicable requirements and standards.

Pore pressure and influenced area

There is no groundwater draw down, pore pressure reduction and associated risk for settlements because the frozen body and support is completely waterproof. The influence zone does not reach further out than the frozen soil, limiting potential settlements to the volume that has been frozen.

The coolant circulates in a closed circuit. No chemicals are left in the ground. Groundwater outside the frozen region is unaffected.

Other aspects

Warm climate

Plenty of freezing plants are working in tropical areas. Ground freezing can be executed during the summer and in warm climate as well as in cold climate. The temperature must be lowered from a higher start temperature for initial freezing in warm climate than in cold climate, but it is the phase change from water to ice that requires most of the energy. There will also be more heat to remove during maintenance freezing. Down in the underground, the annual mean temperature is more important than the peak daytime temperature.

Photo: Warm climate in Hong Kong.
Photo: Geofrost freezes in Kowloon, Hong Kong.
Hong Kang has an annual mean temperature of 28 °C. They were skeptical to ground freezing in warm climate. First time ground freezing was done in Hong Kong was in 1995, when Geofrost participated in the first phase of The Strategic Sewage Disposal Scheme in Kowloon. The summer climate is warm and humid with a daily average temperature often above 32 °C.

High water pressure

High water pressure does not delay the freezing process. Both below high mountains and deep below fjords the water pressure is high. These pressures are not a problem for ground freezing. Largest water pressure Geofrost has frozen so far is 1400 kPa, at the bottom of the frozen construction used for the Oslofjord subsea tunnel.

Figure: Principle of the temporary support of the subsea tunnel under the Oslofjord.
Figure: Principle of the ground freezing support of the subsea tunnel under the Oslofjord.

Running water

Moving water represents extra heat added, which must be removed in connection with ground freezing. If water flows through the volume to be frozen, one must ensure sufficient freezing capacity to be able to remove the additional heat.

Insulation of exposed frost structure

To keep the frozen body intact if exposed to warm environments, it has to be covered by a layer where the temperature gradient goes from sub-zero to the surface temperature. The layer can be thawed soil, shotcrete or insulation.

Figure: Temperature profile through frost structure and toward the environment.
Figure: Example of two temperature profiles from center frost structure with a temperature of -40 °C, towards the surroundings with a temperature of +20 °C. Top figure: Stable situation where thawed soil acts as insulation. Bottom figure: A layer of insulation is covering the excavated frost.

Concreting against frozen structure

It is fully possible to use the frozen body directly as formwork. Already during the 1950s this was studied thoroughly both in laboratories and in situ during construction of many shafts in Great Britain.

Two things to consider:

  • Heat balance at time of casting:

    The volume of the concrete must be large enough to generate sufficient heat - long enough - to start the curing process. Then it is not of importance if a minor part of the concrete freezes, as the curing will continue when the concrete thaws and will eventually develop full strength.

    Casting directly toward the frozen body will usually make a layer thaw. This layer may eventually freeze again and the thawing will normally have insignificant effect on the frozen body stability.

  • Concrete temperature at casting:

    Too high start temperature together with the curing top leads to an unnecessary large temperature gradient during refreezing. Too low temperature will delay the curing process and may lead to reduced concrete quality. 19-20 °C is typically an ideal concrete temperature.

Frost heave

Volume expansion during phase change from water to ice is per definition not frost heave and is often negligible since the excess water is pressed out of the pores during water-ice transformation. A parallel to this is water pipes bursting due to freezing. It is the water pressure building up, if the water cannot escape, that bursts the pipe, not the ice itself.

Frost heave is caused by the growth of ice lenses from water flowing to the freezing front. The frost heave's direction of force is perpendicular to the freezing front, while the ice lenses are parallel to the freezing front. To obtain frost heave, or ice lense growth in the ground, all the following three conditions must be satisfied:

  1. 1. The temperature has to be below the freezing point.
  2. 2. There must be free access to water.
  3. 3. The ground has to be frost heave susceptible.

To which extent the third condition is satisfied, frost heave susceptible ground, depends on:

  • Amount of fines, that is particles smaller than 0.02 mm.
  • Velocity ratio between freezing front progression and water suction to the freezing front.
  • The pressure on the soil in the freezing front's direction.

Photo: Frost heave.
Photo: The surface has lifted due to water suction to the freezing front. Here the ice lense has grown multiple centimeter high ice crystals. It is spring and the meltwater has not drained away due to impermeable frozen barrier below, resulting in ice lense growth at sub-zero temperatures at nighttime.
The winter gives slow freezing from the surface where the overburden load is minimal. Ice lenses grow in frost heave susceptible ground and lifts the surface.

Seasonal frost and artificial ground freezing have different effect on the ground.

Prof. Williams (1967) has published a comparison for how deep it is possible to get ice lense growth, depending on groundwater level and soil type expressed by the following formula:

pi – pw =
2 σiw / rc
pi = ice pressure = total weight of soil above
pw = water pressure = pore pressure
σiw = boundary level energy air-water (surface tension)
rc = radius corresponding to the size of the largest continunous opening through the pore system

Frost heave susceptibility classes have been made for Norwegian soils and climate.

Table: Norwegian frost susceptibility classification, (translated after Bane NORs «Teknisk regelverk» 2019-09-09).
Table: Norwegian frost susceptibility classification. Figure: Examples of grain size distribution curves in different frost susceptibility classes.
Figure: Examples of grain size distribution curves in different frost susceptibility classes, (translated after Bane NORs «Teknisk regelverk» 2019-09-09).

The temperature gradient during artificial ground freezing is much larger than during seasonal frost, and hence the freezing velocity is much faster and the permeability in the normally frost susceptible fine grained soil classes is too small to feed the freezing front with water for ice lens growth. The frost rarely reaches deeper than 2 m from the surface in Norway, while artificial ground freezing is used much deeper, where the confining stresser are much larger. Hence, seasonal frost and artificial ground freezing have different effects on the ground and the Norwegian frost heave susceptibility classes are not applicable for artificial ground freezing.

Seasonal frost - winter

Figure: Seasonal frost - winter.
Figure: Principle of frost penetration with seasonal frost - winter.
  • Slow freezing
  • Small surface load
  • Ice lenses grow in frost heave susceptible ground and lifts the surface.
  • Frost heave susceptibility classes are made for Norwegian climate and temperature gradients.

Ground freezing

Figure: Ground freezing.
Figure: Principle for frost penetration with artificial ground freezing.
  • Quick freezing
  • In fine grained soil (clayey), the permeability prevents water suction for ice lense growth.
  • In coarse grained soil (frictional), the ice develops from the pore centers, displacing the excess pore water.
  • Only at the frozen body's outer region, ice lenses may grow during long lasting freezing.

Without ice lense growth (frost heave), there is no frost heave settlements, neither by seasonal frost nor artificial ground freezing.

By artificial ground freezing the initial freezing is so fast that the permeability in the normally frost heave susceptible fine grained soil types effectively prevents water supply to the freezing front, so ice lense growth is minimal. The soil pressure increases with depth and the possibility for ice lense growth decreases similarly. Penner (1978) points out that the water suction from the surroundings to the freezing front decreases with increasing earth pressure and that the limits depend more on soil type than freezing temperature.

Thaw consolidation

Seasonal frost and artificial ground freezing have different impact on the ground. Settlements due to thawing requires both that frost heave has occurred during the freezing period and that the excess water from the thawed ice lenses is prevented from being absorbed in the ground in step with the thawing. The resulting increase in pore pressure reduces the effective stress and settlements occur if the loading exceeds the bearing capacity. The settlements will further increase if the loading is dynamic, as is the case with traffic loading on roads in the frost thawing period. For seasonal frost, the excess water in the top layer is common because there is a frozen waterproof layer below preventing drainage of the excess water.

Difference in freezing velocity and boundary conditions during thawing makes the pore system behavior different for seasonal frost and artificial ground freezing, and hence different impact on the ground.

Table: Pore system behavior in freezing and thawing, and following consequences for seasonal frost and for ground freezing.
Table: Pore system behavior in freezing and thawing

Generally, very little ice lenses occur from artificial ground freezing and the thawing takes place at a slow rate from the outer surfaces without any drainage barriers. Thaw consolidation is rarer and generally less for ground freezing than for seasonal frost. Still, it should always be considered which order of magnitude and which consequences potential thaw consolidation may cause.

Quick clay and soft clays where natural water content is higher than the liquid limit are exceptions that require special attention. Such soils require pretreatment with moderate amounts of lime/cement to avoid thaw consolidation.

Quick clay

The water content of quick clay is higher than the liquid limit and causes settlements if it is released and the clay reconsolidates. Therefore, the quick clay's sensitivity with respect to deformations has to be included in the geotechnical assessment for ground freezing.

Photo: Quick clay undisturbed and after remoulding.
Photo: Quick clay undisturbed and after remoulding.

Tests show that mixing in a moderate amount of lime/cement in quick clay before freezing will prevent settlements from thawing.

Case studies

Freezing 120 mbsl in The Oslofjord subsea tunnel

Ground freezing was required to get through a rock depression filled with permeable glaciofluvial material, with crown 120 m below sea level. Geofrost had the full responsibility for all the activities from design to procurement, construction, commissioning and hand over of the frozen structure.

The longest tunnel on the Oslofjord Crossing is the 7.2 km long undersea crossing itself. The 78 m2 highway tunnel comprises three lanes - one lane going downhill and two lanes on the uphill slope – and has an inclination of 7 % to minimize tunnel length. It was the 19th subsea road tunnel to be built in Norway by The Norwegian Public Roads Administration (NPRA). SRG was the main contractor.


The Oslofjord is situated in a major regional rift belt, the Oslo grabend, in which the total vertical displacement is about 2 000 m. The bedrock consists of deep eruptive rocks and dykes of Permian age, dominated by coarse-grained granite, traditionally termed Drammens-granite. The rift belt crosses the Oslofjord tunnel.

The rift belt here consists of several faults and weakness zones, some of the major ones following the fjord over long distances. The seismic survey showed three wide channels in the threshold of the fjord, eroded by glaciations. In each of these channels, major weakness zones were confirmed. The "Hurum weakness zone" having the lowest seismic velocity of 2 600 m/s was further investigated, among other methods by two penetrating core holes. One of the holes was made by directional core drilling along the assumed tunnel alignment. The other core hole was made from the other side. Core material consisted of familiar crushed rock and clay. They both missed the depression filled with frictional soil.

The 15 m wide zone of loose glacifluvial deposits at the bottom of a deep channel above the "Hurum weakness zone", was found by probe drilling from the tunnel head. It is believed that the channel was cut by a glacial melting river, leaving behind a permeable soil deposit containing sand and gravel in addition to the larger rounded blocks of rock; a glacifluvial deposit. Without any fines the zone was highly permeable with constant hydrostatic water pressure of 12 bars at the crown.

Photo: Frozen glacifluvial material after blasting.
Photo: Frozen glacifluvial material after blasting.

Tunnel excavation

There were 3 headings: one from the east side of the fjord and one in each direction from a 730 m long adit at the sea front on the west side of the fjord. The full face of the tunnel was excavated by conventional drill and blast methods, with an average advance of 30 to 40 m a week at each tunnel face.

It was from the eastbound heading from the adit the glacial zone was first encountered by a probe hole. As a routine, three 30 m long probes were drilled in the crown, having a 15 m overlap. It was soon after the overlap with the previous probe, the new probe struck water at enormous pressure, and it was clear that the probe had intersected with the full hydrostatic pressure of the 120 m head of the fjord above. Further investigations of the area showed that almost half of the cross section would run into the loose glacifluvial soil. The lower half contained fractured rock of the crushing zone.

It was decided to excavate a lower bypass tunnel, spiraling away from the main tunnel alignment, passing some 20 m below the glacial channel, and rising up to the main tunnel alignment on the other side of the zone. From there the tunnel proceeded eastward under the fjord as well as backwards to the problem zone. The bypass tunnel later became the drainage sump, replacing the one originally designed.

Grouting was the preplanned solution for progressing through this zone. Continued drilling problems indicated that the soil zone would be difficult to stabilize by grouting. The attempts to inject grout and seal the zone had no effect. In fact, more than 700 tons of cement-based grout was pumped into the zone without the slightest effect in cutting off the water ingress or reducing the water pressure. Large volumes of grout/concrete would be needed and it would be difficult to control the result, if continued grouting were to be the final solution. It was therefore, after 5 months of grouting trial, decided that freezing was necessary to stabilize the zone before excavation. The decision was based primarily on the assumed higher and more controllable safety of the freezing method. The bypass tunnel had taken the problem zone out of critical time schedule.

Geofrost, as a subcontractor, was responsible for both design and execution of the artificial ground freezing works, including drilling. 46 m was left between the two tunnel faces. It included the approximately 15 m wide zone of permeable glacifluvial and morainic deposits at the bottom of the deep channel, and some good quality rock on each side.

Design of the frozen structure

Geofrost designed the frozen structure based on the Berggren creep model for frozen soil and design loads from GeoVita. Third party design verification was done by SINTEF.

Ordinary core drilling failed to give undisturbed samples suitable for laboratory testing, except for two small pieces. As a result, the frictional material flushing out of the drill holes were gathered, compacted and saturated with sea water before frozen and tested. All tests were performed at The Norwegian Institute of Technology (NTNU).

Due to the relatively low strength caused by the salinity of the porewater at normal ground freezing temperatures, it was decided that the design temperature should be as low as -28 °C. This temperature was both obtainable in the laboratory and could practically be achieved in the field. The cycle of drilling, blasting, excavate and produce the permanent lining of a section was planned to take one week.

Before excavation could start, three different requirements had to be fulfilled for the frozen structure: 1) The temperature of the main frozen structure should be -28 °C or lower. The length of each drill and blast round was then given as a function of the temperature and thickness of the frost structure. As an example, a thickness of 3 meters at -28 °C resulted in an allowed unsupported length of 2.7 m. 2) The soil part of the face should all be frozen to ensure safe working area and support of the stabilizing frozen structure. 3) The complete circumference around the tunnel should be impermeable to avoid water seepage through the invert and thereby thermal erosion of the frozen structure.

Field experiences

A drilling chamber was established at one side of the zone. In the main frozen structure through the soil area there were two rows of freezing pipes, elsewhere only one. Down the hole hammers were used for drilling, and done through safety valves. Due to the high water pressure the normal air driven hammers were exchanged by water driven hammers with good results. It is believed to be the first time in the world water driven hammers were used in soil.

Drilling was extremely difficult. The soil consisted of very good quality boulders in a matrix of loose sand and gravel. Earlier grouting had no positive effect on permeability and hole stability. Drill rods and grouting equipment left in place from these works, as well as rock bolts, gave the drillers a hard time. Deviation measurements were carried out for all holes. Despite all the complications, only 12 of 115 holes were abandoned.

Coaxial freezing pipes were installed inside the casing to circulate brine which was cooled by an ammonia based freezing plant, placed in the tunnel.

To ensure the design requirements were fulfilled, both before starting and during excavation, Geofrost carried out a thorough temperature measurement program.

Photo: Tunnel face with freezing pipes installed.
Photo: Tunnel face with freezing pipes installed.

Excavation scheme

The full face of 130 m2 was excavated by means of short rounds of drill and blast and then the full concrete lining, before advancing to the next section. The excavation started from the opposite end of the installations. Because the freezing pipe pattern was coned, the thickness of the frozen structure decreased as tunneling proceeded. For this reason, excavation sections were planned to decrease from 3.0 to 1.5 m in length.

Excavation and lining were undertaken by the main contractor, SRG. The low temperature of the contour at approximately -30 °C was considered when designing drill pattern and loading of the holes. Detonating fuse was used in the two outermost rows. Igniter was electronic. Otherwise anolit was used for the rest of the face of 130 m2. There were no problems with drilling in the frozen material and work proceeded as planned, with good results.

At the most, 40 % of the face consisted of soil. The bottom layer was a morainic material, well rounded, containing all fractions up to boulders of 3 to 4 m3. Above there was a layered glaciofluvial material, and stones from most parts of southern Norway was recognized in the zone.

The vault and face were shotcreted, with layers up to 20 cm, to avoid stones falling from the surface when heat from the working machinery rises and thaw the surface. Additives in the shotcrete prevented any problems with shotcreting the very cold surface.

Concrete lining was 1.2 m in the bottom and 1.0 m in the vault. It was designed to take full water pressure. Strength requirements of the concrete in the lining before next blast, was 40 MPa. This was reached after approximately 20 hours as prescribed.

Photo: Frozen contour with freezing pipes appearing as cone is decreasing.
Photo: Frozen contour with freezing pipes appearing as cone is decreasing.

Closing remarks

It was possible to plan, design and control all operations necessary for the project fulfilment. Drilling for the freezing pipes was the most challenging task. There were neither stability problems nor any water leakages. Shotcreting and concreting against the exceptionally cold ground surface worked very well without any problems. A very difficult situation was solved safely in a controlled manner.

Temporary ice plug in Skoddeberg headrace tunnel

The ice plug concept developed and tested by Geofrost, makes it possible to temporarily close a water filled tunnel.


The ice plug concept developed by Geofrost is based on two facts:
1) Materials are gaining strength when frozen
2) The sealing may easily be removed after use.

The concept is ideal for temporary works in connection with maintenance and refurbishment in water filled hydro tunnels when there is no working upstream gate or other easy way to cut off the water. The ice plug method has a low impact on the environment and is an efficient alternative to cofferdam building or reservoir lowering.

During concept development, both theoretical research and laboratory testing were performed before the full-scale test in one of Statkrafts tunnels.

Full-scale test for the research work was carried out in a bypass tunnel in the Røssåga waterways, situated between Mosjøen and Mo i Rana in northern Norway. In this tunnel, with a cross section of 50 m2, a 6 m long artificially frozen ice plug was holding back the water reservoir. Water pressure equaled a water column of 20 m. When the tunnel was emptied, the plug was loaded horizontally by more than 1 000 tonnes. This load test lasted for one month. Then the ice plug was removed efficiently by thermal erosion. (Berggren & Sandvold 1995)

Case: Skoddeberg headrace tunnel

Hålogaland Kraft AS planned maintenance works for Skoddeberg Hydro Power Plant. To be able to do these works, the waterway had to be emptied, but the gateway in the headrace tunnel from the 1950s was leaking severely.

Their first design approach was to lower the water level in the reservoir and build a coffer dam. The works should be done during winter season when the water level normally is at its minimum. However, there were safety concerns about this solution since flooding had occurred last winters. If that would happen again, the water level in the reservoir could rise by a meter per day and a reasonable coffer dam would be overrun quite soon.

While searching for other solutions, the ice plug concept appeared. Geofrost proposed two different alternatives, one based on brine freezing and one on nitrogen freezing. The more rapid nitrogen freezing was chosen in order to minimize the stop in energy production, although more expensive.

The date for closing down the hydro power plant was chosen to minimize the loss of potential power during the stop and all activities were coordinated and planned accordingly.

Photo: The Skoddeberg lake with the intake whirl in the middle of the picture.
Photo: The Skoddeberg lake with the intake whirl in the middle of the picture.

Liquid nitrogen (LIN), with a temperature of -197 °C at atmospheric pressure, was used for the initial freezing. The nitrogen, delivered by tank lorries, boils in the freezing pipes and are then let back to the atmosphere. For economic reasons, nitrogen was used initially and then brine at -40 °C was used for the ice plug maintenance.

When the ice plug had achieved the required temperature and thickness, loading could start. The leaks were large enough to lower the water level downstream in the gateway shaft and the surge chamber. When water surface reached the tunnel level and thus a much larger volume of water was to be removed, a downstream gate was partly opened to speed up the emptying, while keeping a gentle loading rate.

Besides that the leaking gate was to be replaced, some rock support work in the surge chamber had to be done, and an extra turbine was to be installed in the hydro power station. Due to the new turbine, a branching of the penstock also had to be built. All these works, done by the client, were coordinated to be performed as quickly as possible as soon as access was provided by emptying the tunnel for water downstream the ice plug.

Photo: Tunnel entrance: down the shaft
Photo: Tunnel entrance: down the shaft.

To assure the ice plug was developed as designed, temperature measurements were performed and compared with design calculations and monitoring was continued throughout the maintenance period.

Photo: The ice plug seen from inside the emptied tunnel.
Photo: The ice plug seen from inside the emptied tunnel.

The cold winter months north of the polar circle revealed demanding working conditions. However, works ran smoothly. As shown in the time schedule, the power plant interruption was only 6 weeks, whereas only 2 of these weeks were needed for establishing and removal of the ice plug.

The cofferdam solution would have been approximately 3 times as costly as the ice plug. In addition, weather conditions turned out so badly that the cofferdam solution would have been impossible to carry through that winter. Heavy rain in late autumn resulted in a full reservoir when it was supposed to be at the lower regulation level. The execution ran smoothly and the project was completed 3 weeks earlier than planned.

Closing remarks

Geofrost has developed an ice plug concept for maintenance works in hydro power plants. The research project proved that the concept was feasible, and this case history has proven commercial benefits. Apart from being safe, the ice plug concept is predictable in time and cost.


Existing Norwegian guidelines

There is no own standard for ground freezing, neither in Norway, Europe or the rest of the world. Ground freezing is however referred in The Norwegian Public Roads Administration's guidance V221 and in Bane NOR's Technical design basis for Intercity.

All geotechnical design follows the Eurocode with its national annexes. Eurocode 7 applies for geotechnical design for buildings and constructions. This includes for instance foundations of buildings, support constructions, fillings and slope stability.

Figure: Grunnforsterkning, fyllinger og skråninger. Håndbok V221. Statens vegvesen / The Norwegian Public Roads Administration).
                  - Teknisk Designbasis for InterCity. Rev05A. Bane NOR. -
             Eurocode 7 - Geotechnical design - Part 2: Ground investigation and testing. European standard.
Grunnforsterkning, fyllinger og skråninger. Håndbok V221. Statens vegvesen/The Norwegian Public Roads Administration.
Teknisk Designbasis for InterCity. Bane NOR.
Eurocode 7 - Geotechnical design - Part 2: Ground investigation and testing. European standard.

Construction pit guide (Byggegropveiledningen 2019)

Byggegropveiledningen utgitt av NGF 2019
Construction pit guide published 2019 by Norwegian Geotechnical Society.
After a large research project (BegrensSkade) revealed the causes of many of the damages to adjacent buildings after earth and foundation works, the Norwegian Geotechnical Society started the works on making a construction pit guide. This has been an enormous effort of voluntary work by the industry which has taken time to publish. Ground freezing is treated in chapter 6.20 and 7.11.


Since there is no international standard for design and execution of ground freezing works, a CEN-group (CEN TC288 WG20) has been created with the assignment to get such a standard in place. Geofrost is represented in this group.

Ground freezing history

Mining industry in the 19th century

Already in 1852 French engineers discovered the useful properties the ground gets when it freezes. During the very cold and harsh winter that year, they were about to make a mining shaft through a layer of soil usually saturated and unstable, but which now were possible to pass relatively easy.

First known planned artificial ground freezing project to utilize the strength and waterproofness was in 1862 in connection with a mining shaft in Wales. Brine was circulated through freezing pipes that were installed through the instable soil layer. The principles were the same as those that were patented in 1883 in Germany and which still is used today: Freezing pipes are installed through the layer to make stable and/or impermeable, then it is frozen and finally safely excavated.

Shafts dominating also in the 20th century

In 1950 the depth record for ground freezing from surface was 900 m for a shaft in Canada.

Experience was gradually built, but originally ground freezing was done by trial and error. Design methodology and computational models were lacking.

Frozen ground is a complex material where the properties in addition to stress, depend on both temperature and time. Based on much experience from permafrost, Vyalov et. al. published in 1962 an empirically based model. Later many more arose.

In the mining industry ground freezing proved being both reliable, safe and cost efficient, and is therefore often chosen as method early in the design phase.

In civil works however, the method has been regarded as being costly. The known qualitative benefits are often overlooked and the method is often considered a last resort when the problems arise. A contributing reason to this may be that the responsibility for temporary works and choice of method has been by the contractor. In tenders it is then unavoidable that the contractor chooses the method that appears to be the cheapest.


Geofrost was founded in 1986, as the first specialized ground freezing company. With its background within geotechnics and engineering geology, they realized what fantastic tool ground freezing would be for civil works. By using mobile containerized and module-based systems the costs were driven down so it could be the preferred method every time it was technically and environmentally the best option.

International Symposium on Ground Freezing (ISGF)

To exchange experiences and further develop the ground freezing method in a more efficient way, prof. H.L. Jessberger arranged the first international symposium on ground freezing in Bochum, Germany in 1978. The conference gathered approximately 100 participants from all over the world within all disciplines related to ground freezing.

ISGF was set to be arranged every third year, but eventually stopped. The symposiums held were:

  • ISGF´78: Bochum, Germany
  • ISGF´80: Trondheim, Norway
  • ISGF´82: Hanover, USA
  • ISGF´85: Sapporo, Japan
  • ISGF´88: Nottingham, UK
  • ISGF´91: Beijing, China
  • ISGF´94: Nancy, France
  • ISGF´97: Luleå, Sweden
  • ISGF´00: Louvain-la-neuve, Belgium

21st century

Ground freezing gets more used as the projects get more complex, areas more constrained and the need to utilize areas with challenging ground conditions increases. Groundwater control has also become more important with deeper basements nearby old buildings with direct foundation.