I've been designing helical pile foundations for close to fifteen years and in that time the one topic that has consistently generated more confusion, more back-and-forth between engineers, contractors, and building officials, and more quiet uncertainty than almost anything else in the practice is frost adfreeze: what it is, how it works, what values to use, why the values in the references don't always match each other, and whether the pile you just designed is actually going to stay in the ground through a Canadian winter the way you intended it to.
The reason that confusion exists is not that the topic is poorly understood in the research community, because it actually isn't. There is a reasonably solid body of published work on the mechanism, on the values, and on where those values come from, spanning Canadian, American, Russian, and Chinese research going back decades. The reason the confusion persists in practice is that no single Canadian resource has ever assembled all of that into a coherent picture that a practitioner can read from start to finish and walk away from with a clear understanding of what is happening physically, why the design values are what they are, and how to apply them to the specific soil and groundwater conditions on a real project.
This article is my attempt to be that resource.
Frost Heave and Frost Adfreeze Are Not the Same Thing
The first thing to understand about frost adfreeze is that it is not the same thing as frost heave, and that distinction matters more than almost anything else in this topic because the two phenomena get conflated constantly in practice and the design values that exist in the literature come from measuring one of them, not the other, and applying the wrong value to the wrong problem is how you end up either over-designing a pile to the point where the economics don't work or under-designing it to the point where it moves.
Frost heave is what happens to the soil itself when it freezes. When water in frost-susceptible soil turns to ice, it expands by approximately nine percent of its volume and that expansion has to go somewhere, and the direction it goes is upward because that is the path of least resistance, and the magnitude of that upward movement depends almost entirely on how much water was available in the soil at the time of freezing, how quickly that water could migrate toward the freezing front as the frost advanced downward through the soil profile, and how frost-susceptible the soil is in terms of gradation, plasticity, and structure. A silty soil sitting above a shallow water table in a wet fall can heave dramatically. A well-drained granular soil with no water table anywhere near the frost zone barely moves at all.
Frost adfreeze is something different. It is not about what the soil does to itself. It is about what happens at the interface between the frozen soil and the steel shaft of the pile passing through it. As the soil freezes around the shaft, ice bonds directly to the steel surface, and as the frozen soil above tries to heave upward, that bond attempts to drag the pile upward with it. Whether the pile actually moves depends entirely on the relationship between the strength of that ice-steel bond and the resistance the pile is able to develop through its anchorage in the unfrozen soil below the frost zone, plus whatever dead load the structure above is pressing downward on the pile head.
Those are two completely different problems governed by two completely different physical mechanisms, and the design values that exist for one of them do not apply to the other.
Three Forces That Get Treated as One
Once you understand that frost heave and frost adfreeze are different phenomena, the next thing to understand is that even within the adfreeze problem itself there are three distinct forces operating at very different magnitudes, and the published literature contains values for all three of them, and if you do not know which one you are looking at when you pull a number from a reference, you can be off by a factor of ten or more in either direction.
The first is the thermodynamic expansion pressure of freezing water. Ice formation in confined conditions can generate pressures into the hundreds of megapascals under the right circumstances, which is orders of magnitude larger than any pile anchorage could ever resist. The reason this does not simply rip every pile out of the ground every winter is that soil is not a sealed container. Water migrates away from the growing ice front rather than being compressed, so the expansion energy goes into heaving the ground upward and feeding ice lenses rather than building up pressure against the pile shaft. This force is real and enormous but it is not what acts directly on the pile and it is not what you are designing against.
The second is peak adfreeze bond strength, which is the peak shear stress at the interface between the frozen soil and the pile surface at the moment before any relative movement has occurred. It is the grip strength of a fresh ice-steel bond, the maximum force per unit area that the frozen soil can transmit to the pile before the bond ruptures and the soil begins to slide past the shaft. For steel against frozen silt under the right conditions, measured peak values in the literature range from about 276 kPa in US Army cold-regions testing up to over 1,200 kPa in Russian field measurements. These are real numbers measured under real conditions, but the conditions under which they were measured are not the conditions that exist on a helical pile shaft in a seasonal frost zone in Ontario or Manitoba or Alberta, and that is the critical point that gets missed constantly in practice.
The third is residual sliding friction, which researchers call tangential heave stress (THS), and this is the shear stress that acts on the pile shaft after the initial ice-steel bond has broken and the heaving soil is sliding past the shaft surface rather than carrying the pile with it. This is the force that governs seasonal frost design on helical piles in Canada. It is what Penner measured in Ottawa in 1974, what Penner and Goodrich measured in Thompson Manitoba in 1983, and what every field study conducted in actual seasonal frost conditions keeps producing: somewhere in the 70 to 150 kPa range depending on soil type and moisture conditions, regardless of which country the study was done in or which research group conducted it.
The key point is this: the 100 kPa and 150 kPa values in the Canadian Foundation Engineering Manual are not conservative lower bounds of the 276 kPa US Army value or the 1,200 kPa Russian value. They are measurements of a completely different physical quantity measured under completely different conditions, and understanding why those three numbers are so different from each other is the key to understanding the entire frost adfreeze design problem.
What Actually Happens to a Pile Over a Canadian Winter
The reason the residual sliding friction is so much lower than the peak bond strength, and the reason the seasonal Canadian field values keep clustering between about 70 and 150 kPa, comes down to how frost actually advances through soil during a Canadian winter, because it does not happen the way most people intuitively imagine it.
Frost does not flash-freeze two metres of soil overnight. A Canadian winter delivers frost progressively, with the freezing front advancing incrementally through the soil profile over weeks and months as the cumulative effect of cold temperatures works its way downward. The upper few centimetres of soil might freeze in the first cold snap of November. Another few centimetres might freeze in December. The frost front continues to advance through January and February until it reaches its maximum depth sometime in late winter, and then it begins to retreat as spring temperatures work their way back down from the surface.
For the pile shaft, this means the shaft is experiencing a depth-varying stress profile rather than a uniform condition from surface to frost depth. The soil that froze earliest, nearest the surface, has had the most time to bond, break, re-bond, and slip again, and research suggests that as little as one to three millimetres of relative displacement between the pile and the surrounding soil is enough to break the initial ice-steel bond and transition the interface from peak bond strength into residual sliding friction. The zone near the advancing frost front, where the soil has just frozen and the bond is fresh, may still be at or near peak bond strength, and that fresh-bond zone migrates downward through the winter as the frost front advances.
What Penner and subsequent researchers measured in the field is the integrated average of that depth-varying, time-varying stress profile across the full frozen length of the shaft over a complete winter season, not the single instantaneous peak on a laboratory coupon. Qu (2021), re-analyzing Penner's raw monthly data, showed that adfreeze stress peaks near 150 kPa when the frost front is only about one third of the way down, then drops to a stabilized range of 50 to 75 kPa as frost reaches full depth, and proposes a characteristic design profile of 100 kPa at 1 metre frost depth declining to 90 kPa at 2 metres and 80 kPa at 3 metres to capture that trend. That average over a full winter keeps coming out between roughly 70 and 150 kPa for typical Canadian clays, silts, and saturated granulars, and that is the residual sliding friction value you are designing against.
Why the Russian and American Values Are So Much Higher
When you put Penner's 100 kPa, the US Army's 276 kPa, and a Russian 1,200 kPa side by side, the temptation is to think one of them must be wrong. None of them are wrong. They are measuring different mechanisms at different temperatures and under different freezing histories, and comparing them directly as though they describe the same design quantity is the source of most of the confusion in this topic.
The Russian value around 1,200 kPa was measured at a soil temperature of approximately minus twelve degrees Celsius at the pile interface in permafrost conditions. The US Army 276 kPa figure is explicitly described in TM 5-852-4 as the adfreeze stress before the initial break in bond between frozen silty soil and an eight-inch steel pipe, which means it is a peak bond strength value measured under controlled laboratory conditions on a freshly frozen sample, not a residual value measured after months of progressive seasonal freezing and displacement cycles.
The connecting variable is soil temperature at the pile interface, not air temperature. An air reading of minus forty degrees Celsius in Timmins does not mean the soil at one and a half metres depth is anywhere near minus forty. Soil is an insulator and the frozen profile carries a thermal gradient, with temperatures near the surface in mid-winter perhaps in the minus five to minus ten degree range, at mid-depth perhaps minus two to minus five degrees, and near the frost front approaching zero degrees. Adfreeze bond strength is strongly temperature-dependent, becoming much weaker as the soil temperature approaches zero from below, which is why the permafrost values measured at minus twelve degrees are so much higher than the Canadian seasonal frost values where the soil temperature at the shaft interface is typically in the minus one to minus six degree range.
The Russian permafrost values are accurate measurements of real conditions in permanently frozen ground. The US Army laboratory values are accurate peak bond strength measurements under controlled conditions. Qu (2021), writing specifically about Canadian helical pile practice, states explicitly that the TM 5-852-4 supporting tests were conducted in permafrost areas with temperatures below minus 40 degrees Celsius using piles placed with silt-water slurry, and that these data "should not be directly used in pile design" for helical or driven piles in non-permafrost environments. The Canadian seasonal field values are accurate measurements of residual sliding friction averaged over a full winter in soil that is only modestly below zero and has experienced progressive freezing and displacement. They are all internally consistent. They are just not describing the same design quantity, and using any of them in a context they were not derived from will produce a number that is either wildly unconservative or wildly over-conservative.
The Field Studies Behind the Canadian Design Values
The two most commonly used adfreeze design values in Canadian practice are 100 kPa for steel in fine-grained soils and 150 kPa for steel in saturated gravel. Both come from the same research methodology applied to two different soils in two different Canadian locations, and with that physical picture now established, their origin becomes much clearer and more defensible than they appear as bare numbers in a reference table.
Penner's 1974 study, published in the Canadian Geotechnical Journal, was a field program, not a laboratory experiment. He installed steel, concrete, and wood columns in Leda clay at test sites in Ottawa and monitored them through multiple Canadian winters, measuring the actual uplift forces that developed on the columns as the ground froze around them each season. He reported the results as average adfreeze stress over the frozen column length, which is total measured uplift force divided by the frozen surface area of the column. For steel columns he found values ranging from approximately 100 to 255 kPa with an overall average around 120 to 130 kPa, and the 100 kPa figure that entered Canadian practice is a rounded near-mean of that dataset sitting below Penner's measured average rather than at it.
Two things about that study matter for how the value gets applied in practice. The first is that Leda clay is a highly sensitive post-glacial marine clay specific to the Ottawa and St. Lawrence River valley, and it is not representative of the silty tills, glaciolacustrine silts, and mixed overburden soils that characterize most Canadian helical pile sites. The 100 kPa value reflects the behaviour of that specific soil in that specific climate over those specific winters, and its application to other soil types is a working assumption supported by the general alignment of the compiled international dataset rather than by direct measurement in those other soil conditions. The second is that Penner documented a diameter effect in his data, with smaller diameter columns producing higher unit adfreeze stress per unit area than larger ones, and standard helical pile shafts in the 76 to 178 mm OD range fall at the small end of that relationship, though the extrapolation from Penner's driven pipe piles to hollow round tubular helical shafts has not been formally validated in published field research.
The 150 kPa value for saturated gravel comes from Penner and Goodrich's 1983 study at Thompson Manitoba, using the same methodology on steel pipe piles installed in saturated gravel. The higher value in gravel compared to clay is not because gravel is more frost-susceptible in the conventional heave sense, because gravel is typically classified as non-frost-susceptible by gradation and produces minimal heave in normal conditions. The mechanism is the moisture regime: gravel's high hydraulic conductivity provides a continuous water supply to the ice forming at the steel interface, which maintains a stronger and more sustained ice-steel bond even after the initial peak bond has broken and the interface has transitioned to residual sliding friction. The practical consequence is that saturated gravel with a water table in the frost zone is not benign from an adfreeze perspective, even though gravel is typically considered a better engineering soil than clay in almost every other respect, and treating a gravel layer as safe because it is granular is a mistake that the 150 kPa value in the CFEM is specifically intended to address.
An argument could be made for a third design value sitting below both of these. Penner's original study included concrete and wood columns alongside steel, and the CFEM carries forward a value of 65 kPa for concrete or wood in fine-grained soils derived from those same Ottawa field measurements. For helical piles that are grouted in the frost zone, where the grout becomes the contact surface with the frozen soil rather than the steel shaft, there is a reasonable case that the 65 kPa concrete value is the more appropriate design parameter than the 100 kPa steel value. The practical challenge is that no research has been done specifically on grouted helical pile shafts in seasonal frost conditions, and the assumption that grouted helical piles behave like Penner's concrete columns in Leda clay is an extrapolation that has not been validated in the published literature.
The Lightly Loaded Structure Problem
The framework described above, where the progressive seasonal freeze mechanism keeps the net sustained adfreeze force on the shaft in the residual sliding friction range of 70 to 150 kPa, applies to piles that are sufficiently anchored below the frost zone and carry sufficient dead load that the initial ice-steel bond can break before the pile starts to move. That transition from peak bond strength to residual sliding friction requires only a small amount of relative displacement, with research suggesting one to three millimetres is sufficient, but it does require some movement, and for a pile with meaningful dead load and deep anchorage that movement happens almost immediately at the onset of freezing each winter without the pile going anywhere significant.
For a pile with very little dead load and minimal geotechnical resistance developed below the frost zone, the situation is fundamentally different. If the total upward force generated by even the first few centimetres of freezing soil bonding to the shaft exceeds the sum of the dead load and the sub-frost anchorage resistance at that early moment in the season, the pile begins to move upward before the bond has had any opportunity to slip into the residual regime. A pile that is moving upward along with the frozen soil is still at or near peak bond strength at the interface, not residual sliding friction, and as the frost front continues to advance through the winter and adds more bonded shaft area to the uplift force, the pile continues to be carried upward.
This is the condition that governs lightly loaded structures: solar racking foundations, light decks, small additions, carports, and similar applications where each individual pile may carry only a few kilonewtons of dead load. For these structures the question is not simply what residual sliding friction value to use in a design calculation, but whether the pile is adequately anchored and loaded enough to ensure that the bond actually breaks before the pile starts to move, and if it is not, then the peak bond strength for the specific soil and temperature conditions governs rather than the residual value. This distinction is not well articulated in most Canadian design guidance and it represents a genuine gap between what the CFEM provides and what a complete design for lightly loaded structures in frost-susceptible soil requires.
The Galvanized Steel Question
Most helical piles installed in Canada today arrive on site hot-dip galvanized, and a reasonable intuition might be that the smoother zinc surface would develop lower adfreeze bond than bare or oxidized steel, making the CFEM values conservative for galvanized installations. The available laboratory data do not reliably support that intuition.
Testing conducted by Sailors Engineering Associates across several surface conditions found peak adfreeze bond strengths on galvanized steel of approximately 678 to 778 kPa, compared to about 725 kPa on bare factory-protected steel and about 1,073 kPa on lightly rusted bare steel. The galvanized surface did not produce meaningfully lower peak bond strength than bare factory-protected steel, and the explanation appears to be that the zinc crystalline microstructure produced by hot-dip galvanizing creates surface roughness at the microscale that promotes ice bonding even though the macroscale surface appearance is smoother than oxidized bare steel. Residual sliding friction values across both bare and galvanized surfaces converged into a relatively narrow band in the 75 to 190 kPa range, suggesting that surface condition has a much larger influence on peak bond strength than on residual friction.
The practical implication for design is that it is not appropriate to apply a reduction factor to the CFEM 100 or 150 kPa values on the basis that the piles are galvanized. Current evidence suggests that bare and galvanized steel should be treated similarly for residual sliding friction design. Engineered anti-friction coatings specifically designed to disrupt the ice-steel bond can reduce peak adfreeze bond strength substantially in laboratory conditions, but the durability of that reduction through years of freeze-thaw cycling in field conditions has not been documented in the published literature, and relying on a coating system for adfreeze mitigation without that long-term performance data is a judgment call that deserves explicit documentation in the design file.
What the Research Does Not Yet Know
The gaps in the published knowledge base are as important to understand as the established findings, because those gaps define where engineering judgment is required to fill the space between what the research provides and what a complete design demands.
There has never been a published Canadian field study measuring adfreeze forces specifically on helical pile shafts. Every primary Canadian field study used large-diameter driven pipe piles. The diameter effect documented in Penner's data suggests that standard helical pile shaft sizes may develop higher unit adfreeze stress than the CFEM values reflect, but the scaling from driven pipe piles to hollow tubular helical shafts has not been formally validated.
There is no published adfreeze data for most of the soil types that Canadian helical pile practitioners actually encounter. The research is anchored at Leda clay on one end and saturated gravel on the other, with thin scattered laboratory data covering some silts and tills and essentially no pile-specific field data for the mixed tills, glaciolacustrine silts, sandy clays, and organic soils that characterize most Canadian project sites.
There is no standardized laboratory test protocol for measuring residual adfreeze stress on pile shaft surfaces, which means that values from different research groups were produced under different conditions using different methodologies and cannot be compared to each other with full confidence. The coefficient of variation across the compiled dataset is approximately 0.36, which reflects both genuine variability in the phenomenon and genuine variability in how it was measured.
There is no published reliability calibration of adfreeze design values for Canadian seasonal frost conditions that would allow a statement along the lines of: the 100 kPa value corresponds to a specific exceedance probability over a defined design life. At best the current evidence supports saying that 100 kPa sits in the upper-middle of the compiled field residual values, without being able to quantify precisely what exceedance probability that represents.
The Design Framework and Where the Code Sits
The Canadian Foundation Engineering Manual, 5th Edition (2023), addresses frost adfreeze in Section 14.6.1 with three design values: 100 kPa for steel in fine-grained soils, 65 kPa for concrete or wood in fine-grained soils, and 150 kPa for steel in saturated gravel, all traced back to Penner's Ottawa and Thompson field work. Table 6.2 includes a footnote stating that where maximum frost penetration depth is used as the design input, a geotechnical resistance factor of 1.0 shall be used to calculate tensile resistance to frost uplift.
That footnote reflects a limit states calibration logic that is worth understanding explicitly. The resistance factor and the load characterization in a limit states check are calibrated together as a matched pair, and the intent of this provision is that the conservatism lives on the load characterization side, in the requirement to use the maximum frost penetration depth rather than an average seasonal depth, rather than in a reduced resistance factor on the resistance side. Using an average frost depth with a resistance factor of 1.0 does not satisfy the intent of the provision. Using maximum frost depth with a resistance factor below 1.0 double-counts conservatism the calibration does not require. The two go together: maximum frost depth and resistance factor of 1.0 as a matched pair for this specific limit state.
The manual is explicit that it is not a building code and not a substitute for engineering judgment, describing itself in its own preface as neither a textbook nor a prescribed code of practice. The frost adfreeze provisions give practitioners anchor values grounded in the best available Canadian field data and a calibration rule for the resistance factor, and they defer to engineering judgment for everything between and beyond those two anchor points. That deference is appropriate given the state of the research. The gap between what the manual provides and what a complete frost adfreeze design requires for the full range of Canadian soil conditions, groundwater regimes, and structure types is substantial, and filling that gap with documented engineering judgment informed by the international research literature is both necessary and consistent with how the manual intends to be used.
References
- Canadian Geotechnical Society. Canadian Foundation Engineering Manual, 5th Edition. NRC Research Press, Ottawa, 2023.
- Penner, E. 1974. Uplift forces on foundations in frost heaving soils. Canadian Geotechnical Journal, 11(3): 323-338.
- Penner, E., and Goodrich, L.E. 1983. Adfreeze stresses on steel pipe piles, Thompson, Manitoba. Proceedings, 4th International Conference on Permafrost. National Academy Press, Washington, D.C., pp. 979-983.
- Aldaeef, A.A., and Rayhani, M.T. 2024. Characterization of adfreeze shear behavior at the interface of frozen clay till and steel piles. ASCE Geo-Congress 2024, Vancouver. DOI: 10.1061/9780784485330.068.
- Emami Ahari, H., Ajmera, B., Pant, R., Huang, C., and Liu, Y. 2025. Estimating tangential heave stress on solar piles. Canadian Geotechnical Journal, 62: 1-8. DOI: 10.1139/cgj-2024-0118.
- Johnson, J.B., and Buska, J.S. 1988. Measurement of frost heave forces on H-piles and pipe piles. US Army Cold Regions Research and Engineering Laboratory, CRREL Report 88-21, Hanover, NH.
- Levasseur, P.P., Maher, M.L.J., and Dittrich, J.P. 2015. A case study of frost action on lightly loaded piles at Ontario solar farms. CGS GeoQuébec 2015, Paper 735. Quebec City, QC.
- Andersland, O.B., and Ladanyi, B. 2004. Frozen Ground Engineering. 2nd ed. ASCE Press / Wiley.
- US Army Corps of Engineers. 1983. Soils and Geology: Frost Action in Soils. TM 5-852-4, Washington, D.C.
- Foundation Technologies Inc. / Sailors Engineering Associates. SlickCoat Adfreeze Bond Stress Reduction Test Report.
- Qu, G. 2021. Pile design against frost heave for lightly weighted structures in northern regions. Wood Group Canada, Oakville, Ontario. Unpublished conference paper. Available at: researchgate.net/publication/355029896.
- Hoeve, T.E., and Trimble, J.R. 2018. Rationalizing the design of adfreeze piles with limit states design. CGS GeoEdmonton 2018, Paper 136. Edmonton, AB.
PileConnect keeps a curated reference of the codes, standards, and evaluation documents behind all of this at pileconnect.com/codes-and-standards. Use it to get oriented, not as a substitute for the adopted code in your jurisdiction or the judgment of a licensed engineer.
Cory Goulet is a P.Eng. with 15 years of helical pile design experience and the founder of PileConnect, a free directory of helical pile installers across North America.