The case for building thermodynamics.
The thermodynamics will continue until morale improves.
A building’s present condition is the accumulated result of physical processes acting through time. Understanding how something failed therefore requires more than identifying its condition today. It requires thinking about gradients, transport, rates, coupling, history, and irreversible change. I believe building science would benefit from thinking more broadly in these terms, and I think the honest name for it is building thermodynamics.
A great deal of excellent work along these lines has already been done. I am not claiming that building science has missed thermodynamics. It hasn’t. I am suggesting that this way of thinking has proved useful enough that we ought to carry it farther, into more of the materials, processes, failures, and durability problems that make up the buildings we study.
Classical thermodynamics is the study of energy, its interactions with matter, and the constraints that determine which changes can happen and in which direction. Non-equilibrium thermodynamics deals with systems away from equilibrium, where differences in temperature, chemical potential, and pressure drive flows and irreversible change. Together with transport and kinetics, it answers the next question: “how fast?”
Our understanding of thermodynamics is a modestly recent development with a muddled beginning involving multiple people working in loose and sometimes contentious contact between the late 1700s and mid 1800s. So, what we call thermodynamics has been around for only about 200 years, as opposed to about 2,200 years for Archimedes’ principle of buoyancy or about 400 years for Galileo’s principle of relativity (the laws of mechanics are identical in all uniformly moving frames, not Einstein’s relativity).
We learned thermodynamics the same way we have learned everything else about Nature: we watched something happen and tried to figure out why. The Nobel Prize-winning physicist Richard Feynman, in The Feynman Lectures on Physics, compared our effort to understand nature to observing a great game of chess without having read the rules. “We do not know what the rules of the game are,” he wrote; “all we are allowed to do is to watch the playing.” By observing the moves, we gradually infer the rules that account for what we see.
Same with thermodynamics: there is no published rulebook for Nature. We infer the rules from what Nature does. Or, as Feynman put it, “The rules of the game are what we mean by fundamental physics.”
The Laws of Thermodynamics are about as fundamental as the rules get. Albert Einstein was particularly impressed by their extraordinary reach and described classical thermodynamics as “the only physical theory of universal content” that would never be overthrown when applied within the limits of its basic concepts.
That grand place in the canon of physics does not mean thermodynamics is all of physics, or that every building problem is a thermodynamic one. Classical thermodynamics describes equilibrium states and the processes connecting them, using a handful of measurable properties like pressure, temperature, volume, and composition, plus the constraints and the direction a process can run. What it does not give us is the rate. It cannot tell us how fast steel corrodes, how quickly water diffuses through a membrane, how long a polymer creeps before failing, or how fast a crack advances across a pavement stone. Those questions need transport theory, kinetics, mechanics, material properties, and, where coupled fluxes matter, non-equilibrium thermodynamics.
Building science is generally presented as a grouping of various concepts: heat transfer, air exchange, the properties of liquid and gaseous water, et cetera. Of course, Nature never separates any of these things. They are all happening at the same time and affecting one another. Luckily, thermodynamics already connects much of what building science studies. And it reaches into failure and degradation topics the field typically treats as somebody else’s department: material durability, adhesion, stone fracture, metal corrosion, material aging, and more.
Thermodynamics isn’t for everything – wind loading, structural statics, acoustics, and daylighting mostly need other tools – but it is useful for a surprising amount of what we do. Building scientists and engineers have long used it for energy and mass balances, psychrometrics, heat transfer, moisture transport, drying potentials, and HVAC load calculations. And some have already carried that thinking considerably farther. John Straube, Joe Lstiburek, Hugo Hens, and others have spent decades connecting transient heat, air, and moisture behavior to freeze-thaw damage, corrosion, biological growth, material durability, service life, and other building failures. Building science already does this exceptionally well in certain areas. The question is why we stop there.
The term “building thermodynamics” itself has appeared before, particularly in energy and systems contexts. This is not really a proposal about vocabulary. The proposal is to carry that way of thinking farther. When we explain a failure that took years to happen, we should be suspicious of an explanation with no time in it. An explanation can account for what we see and still point at the wrong mechanism if it never asks how fast, how long, or how many times.
Building scientists famously call buildings “systems,” and rightly emphasize that changing one component affects others. Add insulation and you can reduce drying potential. Increase air sealing and you change pressure fields. But there is a little more to it than that.
A building is an open, nonequilibrium physical system. Energy and matter move within it and across its boundaries. Materials absorb and release moisture, heat moves, air moves, chemical reactions occur, and materials change. What happens to one becomes part of the conditions affecting the others.
The word nonequilibrium is important. Einstein praised classical thermodynamics, with an important qualification about the limits of its basic concepts. Equilibrium is one of those limits. Buildings rarely have much use for equilibrium.
Going beyond equilibrium does not mean going beyond the Laws of Thermodynamics. It means applying them to a system that never really settles. “Equilibrium” is Oz; our buildings are Kansas. We need non-equilibrium thermodynamics. We need building thermodynamics.
In truth, we already have some of this. WUFI is probably the best-known example. It solves coupled equations for heat and moisture transport through building assemblies, accounting for changing temperature and moisture conditions, material properties, storage, transport, and the interaction between heat and moisture. The calculations involve fluxes, driving differences, transport resistances, and changes with time. That is non-equilibrium thermodynamic thinking, already sitting on thousands of consultants’ desktops. Yet few realize that the same way of thinking can be extended across more areas under the building science umbrella.
The problem is that WUFI is often where the thinking stops rather than where it starts. We apply it to walls and roofs and then set it down. We do not apply it to the stone, the sealant, the fastener, or the adhesive. And we do not apply it to the question of how any of them got to be the way they are. The framework is in the building already. It is only being used on part of the building.
Consider a cracked pavement stone. Alan Arnold Griffith, the father of fracture mechanics, developed Griffith’s criterion in 1920: microscopic flaws concentrate stress in brittle materials, and a crack can propagate when extending it releases energy at least as fast as creating the new crack surfaces requires it. That is thermodynamic thinking joined to the specialized physics of fracture. It is accounting, except the books are kept in energy rather than money. The accounting tells us if extending the crack is energetically favorable.
Now add water, temperature, and time. That’s a dynamic system with gradients and fluxes. How fast will the crack grow? Now we’re in non-equilibrium territory.
Temperature gradients run through the stone as sunshine and weather change. Rainfall and drying create differences in moisture content and water potential that can drive liquid water through the pore structure. Capillary pressure is part of that potential in an unsaturated stone, not a separate mechanism. Those differences help move water through the stone and toward cracks, while changes in temperature and moisture can also change the stresses around them. Water changes things at the crack tip. It can affect the energetic resistance to crack growth and it can participate in chemical reactions at highly stressed bonds near the crack tip. Those reactions have activation barriers, so their rates depend on things like temperature, moisture, stress, and mineralogy. Now time matters.
Cracks grow even when the mechanical driving force sits below the critical level required for rapid fracture. They advance slowly, sometimes over years, at a rate set by stress at the crack tip, crack-tip chemistry, temperature, moisture, and the material itself. There is no contradiction with Griffith. His energy criterion tells us when crack extension becomes energetically favorable under the assumptions of the model. It says nothing about how quickly environmentally assisted processes will extend a crack below that threshold.
What we casually call “a cracked pavement stone” is therefore not merely a strength problem. Non-equilibrium thermodynamics asks how it got that way: what moved, what changed, what drove those changes, how fast they happened, and how they accumulated over time. The crack is what we see today. The physics that produced it may have been at work for years.
Freeze-thaw damage in porous materials is another example. Commonly, frost damage is explained as water expanding 9% on freezing: the water in a pore that is 100% filled expands when it freezes, that expanded volume is greater than the pore volume, the surrounding material can’t stretch to accommodate that new size and BOOM, the material breaks.
The 9% is real, and in a nearly saturated material it matters. But the familiar picture of water simply freezing in a sealed pore, expanding, and breaking the pore wall is incomplete. Pores are not sealed containers. Water can move as freezing occurs, ice can grow by drawing water from elsewhere in the pore network, and damaging pressure can develop in more than one way. What we think we know just ain’t so. The physics in porous materials is considerably more interesting.
Keep in mind, an outdoor paver does not freeze uniformly all at once. As conditions cool, it exchanges heat with the surrounding air and loses thermal radiation to its surroundings, including the sky. On a clear night, radiative cooling to the sky is especially strong. The surface responds faster than the interior, so whenever conditions are changing there is a temperature gradient through the thickness. And outdoors, conditions are almost always changing. Freezing starts near the cold face and works its way in. For most of the freezing process, part of the material is frozen and part is not.
Before any of this can happen, ice has to get started. Cooling water below thirty-two degrees makes ice thermodynamically favorable, but that does not mean ice appears immediately. A stable ice nucleus first has to form. In a real stone or concrete, surfaces, impurities, defects, or ice entering from somewhere nearby can provide places for that nucleation to occur. Once a stable ice crystal exists, pore size becomes important. Ice can grow into the larger pores at temperatures where water in smaller pores and narrow passages remains liquid. For ice to occupy a smaller pore, the ice-water interface has to curve more sharply, and that curved interface carries an energetic cost. The smaller the pore, the greater the undercooling required for ice to be thermodynamically favored there. This is the Gibbs-Thomson effect.
Freezing releases heat, but that alone does not decide whether freezing will happen. Entropy gets a vote too. At constant temperature and pressure, the accounting is done with Gibbs free energy. Freezing lowers the enthalpy of the water, which favors ice, but liquid water has greater entropy than ice, which favors the liquid state. Temperature determines how heavily that entropy difference counts in the balance. Think of that as the vig paid to the Universe. That entropy term is the Second Law showing up in the arithmetic. For pure bulk water at ordinary atmospheric pressure, thirty-two degrees Fahrenheit is where the two sides of the account balance: ice and liquid water have the same chemical potential, so there is no thermodynamic push toward either one. Drop the temperature and the balance changes. Ice becomes the favored state. The farther below the equilibrium freezing temperature the water gets, the greater the thermodynamic push toward ice.
That push has a name. Saying ice is favored is the same as saying it sits at a lower chemical potential than liquid water. Chemical potential is a measure of how badly Nature wants a molecule to be somewhere else, or in some other form. What matters is the difference. Give Nature a difference in chemical potential and it has somewhere to go. Chemical potential is only one kind of thermodynamic driving quantity. Which one matters depends on the problem: gravitational potential for water flowing downhill, temperature differences for heat leaving a hot pan, pressure differences for air rushing out of a punctured tire. But the idea is the same. No additional force has to come along and do the pushing. A difference left standing is the push. That is the Second Law again, the same entropy accounting that set the freezing point a moment ago.
The water in the tiny passages can still be liquid even though ice has formed in the larger pores. At that temperature, the greater curvature required for ice to occupy those small pores can keep the liquid state thermodynamically favored there. But that liquid water can move through the connected pore network toward ice already growing in a larger pore. Adding molecules to an existing crystal does not require ice to invade the smaller pore itself. The difference in chemical potential provides the thermodynamic push, and liquid water moving through the pore network feeds the growing ice. At an atomic scale, ice crystals don’t start at one size and expand to a greater size like a loaf of bread in an oven. They grow because new water molecules are added to them, like a Claymation creature.
The crystal grows in the pore while liquid water remains in the finer passages around it. It cannot simply grow through those narrow passages, because the curvature required there makes ice unfavorable at that temperature. So, it is boxed in. But the chemical-potential difference is still there, and liquid water can still reach the crystal. A very thin unfrozen liquid film can remain between the ice and the mineral surface, allowing molecules to reach the crystal even under confinement. The crystal therefore continues to grow while exerting pressure on the surrounding pore walls. That is crystallization pressure.
The fancy name for this tendency to draw liquid water toward a freezing region is cryosuction. The word makes it sound like the crystal itself is doing the pulling. It isn’t. Differences in chemical potential provide the push, while the pore network provides the path and its resistance to flow. The ice is where the water is headed, not what is dragging it there. This is closely related to frost heave and ice lensing in soils. Water keeps arriving, so the crystal keeps growing, so the crystal keeps pressing. Eventually the material gives.
All of this needs a number. The difference between the temperature the water is actually at and the temperature it would freeze at in a glass on your counter is called the undercooling, and that difference sets how much pressure can develop. The Clausius-Clapeyron equation puts it at roughly 100 PSI per degree Fahrenheit. Five or ten degrees is starting to be real pressure. In the finest pores of cement paste, water remains liquid twenty or thirty degrees below normal freezing, because the curvature of the ice-water interface depresses its equilibrium freezing temperature. That unfrozen water is then drawn through the pore network toward growing ice. The farther below its equilibrium freezing temperature the water gets, the greater the driving force toward ice, and the greater the pressure difference that can develop.
Seen this way, two things that never fit the simple 9% story fall into place. First, water is not trapped inside individual pores. It can move through the pore network as freezing occurs, so a water-filled pore is not necessarily a sealed container in which the same water simply freezes and expands in place. Second, having some empty pore space does not by itself protect the material. What matters is how much water is present, where the available space is, whether that space is connected and accessible, and whether water can move there fast enough as freezing proceeds. That is why degree of saturation, pore structure, and transport rate matter so much to frost resistance.
Now look at what is actually coupled here. Temperature helps determine where ice can exist, while pore geometry helps determine which pores can contain ice at a given temperature and which continue to hold liquid water. Pressure, dissolved materials, and other conditions can shift that balance too. Together, those conditions establish differences in chemical potential that can drive liquid water through the pore network toward growing ice. So, a temperature difference does not simply push the water toward the cold. It changes the thermodynamic conditions, and those changed conditions can produce a movement of mass. Meanwhile, the phase change releases heat into the same material whose temperature helped govern the freezing in the first place. Heat is moving, water is moving, and phase change ties the two together. Classical thermodynamics tells us which states are favored and what the equilibrium relationships must be. To understand how the system actually moves from one state toward another, and how fast, we need transport and non-equilibrium thermodynamics.
And this same thinking helps explain something about air entrainment in concrete that equilibrium considerations alone cannot: why the spacing of the air voids matters. Tiny air bubbles are deliberately distributed through the cement paste, and somehow they make concrete much more resistant to freeze-thaw damage. Even in Canada. The usual explanation is that the bubbles provide somewhere for water to go as freezing occurs. That is part of the story, particularly as the concrete approaches saturation. But there is more going on.
An air void is a big open space, which makes it a favorable place for an ice crystal to grow once ice reaches it. Because the ice-water interface in the void has relatively little curvature, ice grows there at temperatures where water in the much finer pores of the surrounding paste remains liquid. That difference in conditions draws liquid water from the surrounding paste toward the ice in the void. Same cryosuction we just walked through. As water leaves, the pore water in the paste comes under suction and the paste contracts. Measure it. Concrete without air voids expands as it freezes. Properly air-entrained concrete contracts. That contraction offsets the expansive strain from ice forming elsewhere in the pore system. Concrete is weak in tension and strong in compression, so trading some of the first for some of the second is a good deal.
But an air void only helps the paste within some distance of it. So, the fix has a radius. That is what the spacing factor is trying to capture: how far any point in the cement paste is from a nearby air void. Smaller spacing means a shorter trip for water that has to move as freezing proceeds. And now time is back in the problem. Water movement through the paste takes time, while freezing is progressing too. If the voids are too far apart, water cannot redistribute fast enough and damaging pressure builds. That is a rate argument, start to finish. Equilibrium thermodynamics alone cannot tell us what air-void spacing will provide adequate freeze-thaw protection.
Notice what that leaves of the 9% story. If frost damage were just water expanding in a pore too full to hold it, the fix would be a question of volume: how much water is in there and how much room it has. Air content is what goes on the batch ticket, because it is easy to measure in fresh concrete. But hitting the air content does not protect anything by itself. Concrete can carry its specified air and still come apart if the voids are too large or too far apart. What governs is the spacing, and spacing is a distance. Distance matters because water has to move, and moving water through cement paste takes time. Once distance, resistance to flow, and time matter, the simple 9% story is no longer enough. The number most of us design to is telling us that this is a transport problem too.
There is another explanation worth mentioning, because it came from the OGs of concrete frost research themselves. T. C. Powers and Richard Helmuth at the Portland Cement Association proposed a hydraulic-pressure theory in the 1940s and then, in 1953, a more complicated picture built around dissolved salts. Growing ice rejects most of the dissolved material, the liquid left behind becomes more concentrated, and that lowers the chemical potential of the water and depresses its freezing temperature. Solute rejection is real, and it does change the conditions under which ice and liquid water coexist. But it does not replace the pore-size effect. Curvature does the same job through geometry, and dissolved salts are not necessary for frost damage. Stone saturated with distilled water still spalls. (Strictly speaking, calling the salt version osmosis is questionable, since true osmosis needs a semipermeable barrier and an ordinary pore network does not provide one. The terminology matters less than the physics. Anyone who wants to argue the label instead of the mechanism can go pound sand.)
Stone, concrete, mortar, and soil can all suffer damage from crystal growth in their pore systems. Salt crystallization damage has much in common with frost damage, although the details are different and pore structure matters enormously. Air entrainment was found by accident and adopted because it worked. The bubbles can provide pressure relief, but they also shorten the distance water has to move to reach an air void and provide larger spaces into which ice can grow. Nobody was wrong about what to do. Understanding why it works requires us to put temperature, pore structure, phase change, water movement, pressure, and time into the same problem. That is the non-equilibrium part.
Two examples: a cracked pavement stone and freeze-thaw in porous materials. In both, the familiar explanation catches part of the physics but leaves out processes that can decide the outcome. Add transport, rates, and time and you get a different mechanism, not just a longer version of the same one.
So, what does any of this change on a Tuesday morning?
It changes the questions we start with. None of this is exotic. It is just not where most of us begin.
Buildings are sort of like people, every once in a while they need a checkup. For people that’s a doctor visit. For buildings it’s a condition assessment. Usually, the question is what shape the patient is in. Fine. But a doctor does not just look at you. A doctor asks what you have been doing for the last thirty years. A snapshot tells you very little about how you got there. How did it get that way? What was it exposed to? What was pushing on it? For how long? How many times did it happen? That history is usually the thing that explains what you are looking at.
Testing changes too. The usual question is whether a material meets the specified property. But what happens to that property after the material has spent twenty years in an actual building? A number from a fresh sample in a lab is a number about a fresh sample in a lab. It may not be the number that matters after twenty years of getting hot and cold, wet and dry, expanding and contracting. Better to also test materials after appropriate conditioning, and to build mock-ups that see realistic combinations of temperature, moisture, pressure, sunlight, and movement. Buildings don’t experience those things one at a time.
Investigation probably changes the most. A crack, a stain, a debonded membrane, a corroded fastener, a wet wall: these are things we can see. But each is also the result of a process, or many processes, that may have been going on for years. The job is not just to give the condition a name. It is to reconstruct enough of the building’s physical history to explain how it got there.
And durability turns out not to be an intrinsic material property at all. We talk about durable materials the way we talk about tall people, as if durability were something the material has. It isn’t. Durability is what comes out of a material, an environment, a design, and a history interacting over time. Same material, different building, different answer.
None of this turns transport theory, kinetics, mechanics, chemistry, or fracture mechanics into branches of thermodynamics. They aren’t, and thermodynamics does not replace any of them. The building is doing one thing. We need several tools to describe it, and that is our limitation, not Nature’s.
Thermodynamics gives those tools some common ground. Energy and matter still have to be accounted for. Differences in temperature, pressure, chemical potential, and other driving quantities provide the push, while kinetics, material properties, and resistances determine how fast the process goes. And some of it is irreversible. Whatever happened is still in there.
A wet wall, a cracking stone, an adhesive slowly letting go, a corroding fastener, a piece of wood taking on moisture, a roof cooking in the afternoon and radiating to a clear night sky. We need different specialized tools to understand each of them. But they are all happening in the same building. They exchange energy and matter. They share materials and interfaces. What one of them does becomes the condition the next one has to live with.
Sometimes that runs in a circle. A caused B, and then B changed the conditions controlling A, and the thing started feeding on itself. A crack lets in water, the water widens the crack, the wider crack lets in more water. Corrosion products take up more room than the metal they came from, so the corrosion cracks the concrete, and the crack delivers more of what the corrosion needs. Once a failure starts paying for its own continuation, the rate stops being constant and the history stops being a straight line. These are the ones that surprise people, because nothing about the present condition tells you the process is accelerating.
We have been calling this heat, air, and moisture. But what have we really been studying? Energy and matter moving because something is driving them, energy being stored and released, matter reacting and changing phase, and materials slowly changing as all of it happens. None of it happens independently, and some of it only becomes important because it happens over and over again.
We have been calling it building science. Maybe we should call it building thermodynamics.
A Few Other Places to Look
Below is a list of building science topics that might benefit from this sort of building thermodynamics approach. Basically, whenever a driving difference exists, something must move or react, the rate matters, and the process changes the material or system so that the next cycle starts from a different state. Great work has already been done on many of these subjects. I’m advocating for more of it, and for more of that sort of thinking across the building science community.
- Adhesive and membrane debonding
- Corrosion of concealed metals
- Sealant aging and joint failure
- Wood decay and biological deterioration
- Polymer and foam aging
- Efflorescence, subflorescence, and salt crystallization
- Coating blistering and osmotic blistering
- Roofing and pavement thermal fatigue
- Concrete and masonry deterioration (other than freezing)
- Moisture hysteresis and material memory
- Differential movement and buckling of siding, decking and flooring
- Wood checking, splitting and cupping
- Fastener withdrawal and connection loosening
- Concrete drying shrinkage and restraint cracking
- Early-age concrete cracking
- Gypsum board deterioration after repeated wetting
- Roof membrane shrinkage
- Plasticizer migration
- Insulating-glass seal failure
- Insulating-glass gas loss
- Low-E coating and edge-of-glass corrosion
- Rising damp and salt accumulation
- Metal flashing fatigue
- Stone bowing and marble façade distortion
- Coating chalking and color change
- Fire-retardant-treated wood degradation in hot roof assemblies
- Roof sheathing deterioration beneath vapor-open or vapor-closed assemblies
- Mold growth
- Thermal insulation aging
