Exergy
A kilojoule of heat from a furnace and a kilojoule of heat from a lukewarm radiator satisfy the first law identically. They are not worth the same amount, and every engineer knows it without being able to say why in the vocabulary the first law provides. Exergy is the quantity that finally lets you say it: not how much energy a thing has, but how much of it could ever become work.
Two topics have already been built out of the second law: entropy, which measures how badly a process falls short of reversible, and isentropic efficiency, which scores one device against its own reversible twin. Exergy is what you get when the same idea is turned into a property of astate rather than a property of a process: a number attached to a kilogram of steam or a tank of compressed air that says, once and for all, how much work it could ever be made to yield. It is built from entropy directly, and it is worth reading as the second law finally handing over a currency.
Why energy alone is not enough
Take 1000 kJ of heat from a reservoir at 500 K, and 1000 kJ from a reservoir at 1200 K. The first law records them as identical: the same number of joules, ready to be spent on whatever the energy balance calls for. They are not remotely equal in what they can buy. Pair each one with a Carnot engine rejecting to the same surroundings at \(T_0 = 300\) K, and the maximum work each can yield is set by the efficiency already derived in the second law:
The 500 K heat is worth 400 kJ of work at best. The 1200 K heat, identical in every way the first law can see, is worth 750 kJ. The difference is not a detail of engine design; it is fixed the moment the temperature of the source is fixed, before any hardware has been drawn. Energy has a size. It does not have a quality, and the first law was never built to report one.
Exergy, also called availability, is the second law supplying that missing quality as a number. It is defined relative to a reference environment, the dead state, at pressure \(P_0\) and temperature \(T_0\), ordinarily taken as the actual local atmosphere, and it answers one question precisely: starting from a given state and ending in equilibrium with that environment, with every process along the way as reversible as physics allows, how much work can be extracted? Once the substance has reached \(P_0\) and \(T_0\) it is said to be dead, not because anything has been destroyed but because a system at the same pressure and temperature as everything around it has no further pressure gradient and no further temperature gradient through which to do anything at all. It can still hold energy, plenty of it, in general, and none of that energy is exergy, because none of it can be converted without something external to convert it against.
This is also the honest answer to why the dead state has to be specified before exergy means anything. Energy is a property of a state on its own. Exergy is a property of a state and an environment together, and changing the environment (a colder climate, a different atmospheric pressure) changes the exergy of every substance in it without changing a single one of their internal properties. That is not a flaw in the definition. It is the definition doing its job: exergy is explicitly a statement about what the surroundings will let you get away with, and the surroundings are part of the question.
One distinction is worth making early, because it saves confusion later. Exergy is not the same as useful energy in the loose, everyday sense, and it is not simply energy scaled down by some fixed fraction. It depends on the state, on the dead state, and on nothing else: not on how the conversion is actually carried out, which may throw most of it away, and not on what the energy is eventually used for. It is a ceiling on usefulness, fixed by thermodynamics alone, and every real conversion falls somewhere under it.
Two reservoirs, the same 1000 kJ, against the same dead state.
Exergy of a closed system
Building the formula is a matter of running the definition literally: take the system from its given state to the dead state by the most work-generous route available, and add up what comes out. For a stationary closed system, ignoring kinetic and potential energy for the moment, the result, derived by combining the first and second laws for a reversible process ending at \(P_0, T_0\), is
Each term earns its place. The first, \(u - u_0\), is the internal energy still available above the dead state’s floor, the obvious part, and the only part energy accounting alone would have offered. The second term is a correction for the atmosphere: as the system contracts or expands on its way to \(v_0\), the surrounding air at \(P_0\) does work on it or has work done against it, and that exchange has to be credited or charged before theuseful work is what is left. It is exactly the flow-work idea from control volumes and steady flow, reapplied to the boundary between the system and an atmosphere that never stops pushing.
The third term is the one that makes this a second-law quantity rather than a first-law one, and it is worth sitting with. Any reversible process that ends at the dead state must reject entropy to the surroundings if the system’s own entropy is to fall to \(s_0\), and rejecting entropy reversibly to a reservoir at \(T_0\) costs heat, in the amount \(T_0(s - s_0)\), by the very definition of entropy from two topics back. That heat has to leave the system rather than becoming work, so it is subtracted. A state with a great deal of internal energy but also a great deal of entropy (steam close to its saturation line, say, rather than superheated well clear of it) gives up less of that energy as work than the bare \(u - u_0\) would suggest, and this term is where that penalty is paid.
Where the system is moving or elevated, the kinetic and potential terms are added on exactly as they were in every energy balance since heat, work and energy, because both are already fully convertible to work in principle and neither needs a second-law correction:
Two states rarely need an exergy value each on their own; what a problem usually wants is the exergy change between them, which drops the dead-state terms that do not move:
which is computable from a property table exactly as any other property change is, once \(P_0\) and \(T_0\) have been fixed. A flowing stream carries the flow-work term already folded into enthalpy, so its exergy, usually written \(\psi\) to mark it as the flow version, drops the separate \(P_0 v\) term and uses \(h\) instead:
The two formulas are the same idea wearing the same clothes as the \(u\)-versus-\(h\) split from control volumes: the non-flow property for matter that stays put, the flow property for matter that has to be pushed across a boundary, and the difference between them is exactly the same \(Pv\) that separated \(u\) from \(h\) in the first place.
Fix a reference once: the surroundings, at rest, at P0 and T0.
Irreversibility and exergy destruction
Entropy is generated by irreversibility and never destroyed. Exergy runs the other way: it is never generated, and it is destroyed by exactly the same irreversibility, in an amount fixed by the entropy that irreversibility produced. The relation is called the Gouy–Stodola theorem, and it is short enough to be worth memorising outright:
Recall the number from the entropy topic: 1000 kJ falling from a reservoir at 800 K to one at 300 K, with nothing else inside the boundary, generated \(S_{\text{gen}} = 2.08\) kJ/K. Price that generation at a dead-state temperature of \(T_0 = 298\) K and the destroyed exergy comes to
Six hundred and twenty kilojoules of work potential vanished, for an exchange in which no work was ever asked for and nothing crossed the outer boundary at all. That is the sharpest way to see what an uncontrolled temperature drop actually costs: not the heat, which balances to the last joule, but the exergy, which does not, and which cannot be recovered by any device built afterwards. A smaller temperature difference would generate less entropy and destroy correspondingly less exergy, which is the thermodynamic reason a well-designed heat exchanger uses a large area to keep its driving temperature difference small. The trade was already named in reversibility, and this is the price of ignoring it, in currency rather than in kelvin per kilojoule.
The units are worth pausing on, because they are what makes exergy useful rather than merely correct. Entropy generation is measured in kJ/K, which is not a quantity anyone budgets in. Multiplying by \(T_0\) converts it to kJ, the same units as fuel cost, capital cost and every other line in an engineering decision. That conversion is the entire reason exergy analysis exists as a discipline: it turns the abstract fact that \(S_{\text{gen}} \ge 0\) into a specific number of kilojoules that a plant is throwing away in a specific valve or a specific heat exchanger, which is a number that gets acted on where a kJ/K figure would not.
Because exergy destroyed and entropy generated are locked together by a single positive constant, every statement already made about \(S_{\text{gen}}\) carries straight across. Destruction is zero only for a fully reversible process, positive for every real one, and never negative. A calculation returning \(X_{\text{destroyed}} < 0\) reports an impossible process exactly as a negative \(S_{\text{gen}}\) did.
The same 1000 kJ falling from 800 K to 300 K, no engine in between.
Second-law efficiency
The ordinary thermal efficiency of a heat engine, \(\eta_{\text{th}} = W_{\text{net}}/Q_H\), answers a first-law question: what fraction of the heat supplied came out as work. It cannot by itself say whether the engine is close to doing as well as it possibly could, because the ceiling it should be measured against, the Carnot efficiency between the same two reservoirs, is nowhere in the formula. An engine running at\(\eta_{\text{th}} = 0.35\) sounds mediocre until the reservoirs are named. Between reservoirs at 800 K and 300 K the reversible ceiling is \(\eta_{\text{rev}} = 1 - 300/800 = 0.625\), so that engine is recovering better than half of what was ever on offer; the same 0.35 between reservoirs only 50 K apart, where \(\eta_{\text{rev}}\) might be 0.06, would be claiming an engine that violates the second law outright.
Second-law efficiency, \(\eta_{II}\), builds the ceiling into the ratio directly. For a work-producing device it is the actual work divided by the maximum, reversible, work available from the same states:
For the heat engine above this reduces to a ratio of two ordinary efficiencies, since both share the same \(Q_H\):
which is a genuinely different statement from \(\eta_{\text{th}} = 0.35\): it says the engine is capturing fifty-six per cent of what thermodynamics would permit, not thirty-five per cent of the fuel’s energy. The same idea applies to the isentropic-efficiency machines of the previous topic, with one distinction worth keeping straight: isentropic efficiency compares the actual work to the work of an isentropic process reaching the same exit pressure, while second-law efficiency compares it to the reversible work available between the actual end states, dead-state included. The two numbers are related but not identical, and a device can have a respectable isentropic efficiency while still destroying a large amount of exergy, if the process it is idealised against was itself far from the dead state.
Work-consuming devices invert the ratio, for the same reason compressor and refrigerator formulas always invert relative to their work-producing counterparts: the smaller quantity has to sit on top for the result to stay below one.
For a refrigerator or heat pump the same logic is written in terms of the coefficient of performance, \(\eta_{II} = \mathrm{COP}/\mathrm{COP}_{\text{rev}}\), and for a general process with no single work term to point to, the broadest form of all simply compares exergy recovered to exergy supplied:
which is the version that survives even when nothing resembling a turbine or an engine is anywhere in the picture. A well-insulated hot water tank losing heat slowly to a cold room has a second-law efficiency too, and it is this last formula that computes it.
The reversible engine between the same two reservoirs sets the ceiling.
An exergy balance on a control volume
Everything above generalises into one balance the same way the energy balance generalised in control volumes and steady flow, with exergy destroyed standing in the position irreversibility always ends up in: on the side of the equation that is never allowed to go negative. Heat does not carry its full magnitude across the boundary as exergy; only the Carnot-weighted share does, because that is genuinely all a reversible engine sitting at the boundary could extract from it:
Work carries all of itself, since work is already the most useful form energy can take and needs no discount. Mass carries \(\dot m\,\psi\) in and out at every inlet and outlet, using the flow exergy from the earlier section. Collecting every term and adding the one that is new to this balance (destruction, always subtracted, never negative) gives the general rate form:
At steady state the right-hand side is zero, exactly as it was for mass and for energy, and the destruction term can be moved to the other side and identified with the entropy generated inside the same control volume:
Reading this balance is the practical skill the whole topic has been building towards. Solve the energy balance on a throttle, say, and the answer is that nothing has changed: enthalpy in equals enthalpy out, and the device looks thermodynamically inert. Solve the exergy balance on the same throttle and a positive \(\dot X_{\text{destroyed}}\) falls straight out, because a pressure drop with no work recovered from it is irreversible by construction, whatever the energy balance says. The two balances are not in conflict; they are answering different questions, and a device search that only ever runs the first one will walk straight past a throttle, a mixing chamber, or an under-designed heat exchanger without ever being told that anything was lost there at all.