Our first piece on this law treated Local Law 97 as what it looks like from the compliance office: limits, periods, fines, pathways. Local Law 97 in the Hospital Plant is that view, and it holds. But a hospital’s emissions number is not actually decided in a compliance office. It is decided in the plant — by the air handlers, the chiller plant, the boilers and the reheat coils — because in a hospital the mechanical systems are not one load among many. They are most of the load, running all of it, all the time.
This is the companion piece: Local Law 97 read through the mechanical plant, for the people who run it. The law’s arithmetic is fixed. What a facilities team controls is the machinery the arithmetic is applied to.
The limit, translated into plant terms
A general hospital — “Hospital (General Medical & Surgical)” in the Department of Buildings’ property-type table — carries the highest emissions limit in the law for the 2024–2029 period: 0.02381 metric tons of CO2 equivalent per square foot per year, a ceiling it shares only with laboratories and data centers. An office building’s limit over the same years is 0.00846. The law itself concedes, in its own numbers, that a hospital is close to three times as emissions-intensive as the building type the law is usually discussed through.
The concession ends in 2030. Under the Department of Buildings’ rule (1 RCNY 103-14), the hospital limit for 2030–2034 falls to 0.00734 tons per square foot — a cut of roughly 69 percent, among the deepest single steps in the law. Run the arithmetic on a 500,000-square-foot hospital and the shape of the problem appears: an allowance of about 11,900 tons a year becomes about 3,670 tons a year. A hospital whose plant performed exactly at today’s limit and changed nothing would face roughly 8,200 tons of excess emissions in 2030, fined at $268 per ton, per year — about $2.2 million a year, every year, until the plant changes.
Numbers like that are not closed by behavior programs or lighting retrofits. They are closed by machinery: by how much air the building moves, how it makes hot water and chilled water, and how much of its own waste heat it uses twice. That is the subject of everything below.
Ventilation is the floor the number stands on
Start with the load that cannot be negotiated. ANSI/ASHRAE/ASHE Standard 170 — the ventilation standard for health care facilities, and the document behind most of what New York’s hospital engineers design to — sets minimum air change rates by room type in its Table 7.1. An operating room requires 20 total air changes per hour, at least 4 of them outdoor air, held at positive pressure. Airborne infection isolation rooms and protective environment rooms each require 12 total air changes per hour with 2 of outdoor air; an isolation room’s air is exhausted directly outdoors, all of it, by design. A standard patient room requires 4 total air changes with 2 of outdoor air. Pressure relationships between rooms are part of the same table, and they are clinical requirements, not comfort settings. (Our explainer walks the standard itself: What Is ASHRAE Standard 170?, and the ASHRAE 170 edition tracker follows which edition a given project is held to.)
Every one of those air changes has to be heated, cooled, humidified or dehumidified, filtered and moved — and fan energy rises much faster than airflow, roughly with its cube, so a 20-air-change operating room consumes air-side energy out of all proportion to its floor area. In an office tower, ventilation is a dial. In a hospital, ventilation is the floor the emissions number stands on, and the floor is set by infection control.
The standard itself hands back exactly two dials, and plant teams should know both cold. Standard 170 permits an operating room’s airflow to be set back when the room is unoccupied, provided the pressure relationship is maintained. And an airborne infection isolation room that is not in isolation use may drop to 6 total air changes per hour, again with its negative pressure held. Those are real savings — an unoccupied OR at setback is one of the few places a hospital can simply move less air — but they are scheduling savings at the margin. Everything beyond them has to come from recovering energy out of air the building is still required to move.
Reheat: the waste stream hiding in plain sight
The most characteristically hospital waste stream in the plant is reheat. Humidity control in clinical space is typically achieved by cooling air below its dew point to wring moisture out — and then reheating it back up to a temperature a patient room or an operating room can accept. The result is a building that heats and cools the same air, in the same hour, all summer. In most building types, simultaneous heating and cooling is a malfunction. In a hospital it is the design condition. What is not fixed by design is how large that stream runs — and where the reheat energy comes from.
This is where the heat recovery chiller has become the central machine of hospital decarbonization. A conventional chiller makes chilled water and throws the extracted heat — plus the heat of its own compressor work — away to a cooling tower. A heat recovery chiller makes chilled water and hot water at the same time, routing that condenser heat into the building’s heating loops instead. The temperatures line up with the hospital’s needs: condenser water in the 90–140°F range is directly useful for terminal reheat, preheat and domestic hot water preheating, and engineering references for heat-recovery operation put the combined coefficient of performance — the cooling delivered and the heating delivered, added together from one machine — as high as about 7. Every unit of reheat supplied from condenser heat is a unit the boiler or the steam meter does not supply, and it is a unit recovered from a cooling load the building was going to run anyway.
The plant-room translation of Local Law 97 is mostly this sentence: the cheapest ton in a New York hospital is usually a ton of heat the building already made, threw away at the tower, and then bought back from the boiler.
The coefficient table is a design brief
Local Law 97 does not count energy. It counts energy multiplied by a coefficient for each fuel, and the Department of Buildings’ schedule of those coefficients — set in rule 1 RCNY 103-14 — is the most important design document in New York hospital engineering right now. Read it as an engineer and it reads as instructions.
Natural gas counts at 0.00005311 tons per kBtu in every period, forever. The law never forgives a gas molecule. District steam starts at 0.00004493 and declines slowly — 0.00004320 in 2030–2034, 0.00003200 in 2035–2039, 0.00002500 in the 2040s. Grid electricity starts as the dirtiest input a hospital buys — 0.000288962 tons per kWh in 2024–2029, which works out to about 0.0000847 per kBtu, well above gas or steam per unit of energy — and then it collapses: 0.000145 in 2030–2034, 0.0000867 in 2035–2039, and zero in the 2040–2049 schedule. The crossover lands at 2030: from 2030 onward, a kBtu of grid electricity counts for less than a kBtu of steam, and less than a kBtu of gas.
Put one unit of work through all three machines — 1,000 MMBtu of heat delivered to a building loop — and the ordering becomes concrete. An on-site gas boiler at 80 percent seasonal efficiency burns about 1,250 MMBtu of fuel to deliver it and is charged about 66 tons of CO2e, in 2026 and in 2046 alike. Purchased district steam is charged about 45 tons in the current period, 43 in the 2030s’ first half, 32 late in that decade, 25 in the 2040s. A heat pump delivering the same heat at a coefficient of performance of 3 draws about 97,700 kWh and is charged about 28 tons today, 14 tons in 2030–2034, 8.5 tons in 2035–2039 — and nothing at all in the 2040s.
Two practical conclusions fall straight out of that table. The first is the steam trap, and it is worth stating plainly because the instinct runs the other way: converting a hospital from district steam to on-site gas boilers raises its counted emissions per unit of delivered heat — roughly 45 tons against 66 on the example above — because the steam coefficient is charged on steam delivered while the gas coefficient is charged on fuel burned, boiler losses included. Steam has real problems, and our earlier piece treats the strategic question in full; but as pure compliance arithmetic, on-site gas is a step backward. The second conclusion is the order of operations the coefficients impose: recover the building’s own waste heat first, because recovered heat is charged to no fuel at all; electrify the heat that remains, because electricity is the one input whose coefficient is scheduled to fall to zero; and treat any fossil plant installed now as a thirty-year position against a published schedule.
Recovering energy from air you cannot recirculate
The air side has its own version of the same discipline, with an infection-control edge. The isolation room whose entire air volume is exhausted directly outdoors is throwing away air that was filtered, heated or cooled, and humidity-controlled — by design, and correctly. The engineering question is whether the energy leaves with it. Heat can be pulled from an exhaust stream without the airstreams ever touching: run-around coil loops and heat pipes move heat between exhaust and supply through a sealed fluid circuit, which is precisely why they are the standard answer on clinical exhaust. Rotating enthalpy wheels, which recover more energy per pass, carry a leakage path between the two airstreams, and that carryover is why they are commonly kept off isolation and laboratory exhaust. The design rule the law quietly rewards is unglamorous: every exhaust stream in the building should be audited for a recovery device its air classification allows, because energy recovered from exhaust is ventilation the building gets to keep without paying for twice.
The same logic extends past the clinical core. Kitchens, laundries, sterile processing and the general exhaust of a large hospital are all conditioned air leaving the building on a schedule the plant team can read. None of it is as constrained as isolation exhaust, and much of it runs around the clock. A hospital’s exhaust map is, from the compliance side of the desk, a list of pre-paid energy looking for a coil.
The cooling plant, and the one input that improves on its own
Cooling is the plant’s electrical heart — chillers, cooling towers, condenser and chilled water pumps — and it is the one major load whose counted emissions fall even if the plant team does nothing, because the electricity coefficient halves at 2030 and falls again behind it. That is an argument for patience about exactly nothing. At the 2024–2029 coefficient, grid electricity is the most expensive carbon a hospital buys per unit of energy, so every kilowatt-hour the cooling plant saves in the current period is saved at the worst rate the schedule will ever offer. High-efficiency chiller selection, variable-speed drives across the plant’s pumps and tower fans, and condenser-water temperature reset are mature, unglamorous measures whose Local Law 97 value is at its maximum right now and declines — on paper — as the grid cleans up behind them. The fine schedule pays $268 a ton for reductions in this period too.
There is a sequencing subtlety worth naming. Measures that save electricity look worse in a naive 2035 projection than they are, because the projection applies a cleaner coefficient to them; measures that eliminate fossil fuel look better every year, because the gas coefficient never moves. A capital plan that ranks projects on a single year’s arithmetic will systematically under-buy electrification and over-buy electric efficiency late. Ranking the same projects across the published coefficient schedule — the arithmetic above, run per measure — is the correction, and it costs nothing but honesty about which year’s coefficient a measure will actually live under.
Retro-commissioning is the cheapest ton in the building
None of the machinery above performs at its nameplate in a building that has drifted, and hospitals drift like every other complex plant: control valves that no longer close, steam traps failed open, sensors reporting temperatures the rooms are not at, air handlers left in override after a long-forgotten emergency, heating and cooling valves open against each other in the same unit. Local Law 87’s audit and retro-commissioning cycle — the ten-year obligation hospitals already carry alongside Local Law 84 benchmarking — exists to find exactly this. What Local Law 97 changes is the value of finding it: every ton a retro-commissioning pass removes is now also $268 a year of fine exposure retired, in the period where the hospital is closest to its limit.
The plant version of the compliance asset list is therefore short and familiar: retro-commissioning treated as maintenance rather than paperwork; submetering that separates the plant’s end uses — chilled water, heating water, steam, the big air handlers — so a coefficient can be managed system by system instead of argued about at the utility meter; and a controls culture in which an override is an event with an owner and an end date. A hospital cannot manage an emissions limit it can only see in aggregate, twelve months late, in a benchmarking report. It can manage the five systems that produce the number.
One equipment generation to get there
The last mechanical fact is about time. Chillers, air handlers and boiler plant are twenty-to-thirty-year decisions. The 2030 limit — the 69 percent cut — arrives inside the service life of equipment being specified in New York hospitals today, and the 2035 and 2040 coefficients will be met, or missed, substantially by the plant that this decade’s capital plans install. That is what makes Local Law 97 a mechanical problem in the end: not that the law is about machinery, but that its deadlines are written in the machinery’s own units of time.
The teams that will meet the 2030 limit are not waiting for a technology. They are recovering heat they already make, electrifying the heat that remains in the order the coefficient table dictates, taking the two setbacks the ventilation standard legitimately offers, and keeping the plant commissioned tightly enough that the nameplates mean something. The law’s arithmetic is published. The machinery is known. The work is sequencing — and, as our first piece argued, the time to sequence it is while it can still ride along with the renovations the building already has planned.
Run a hospital plant in New York City? Tell us what this missed — corrections and practitioner quotes are credited by name, by organization, or fully anonymously, your choice. We would rather cite the people who run these buildings than describe their work from outside.
Frequently asked questions
What is a hospital’s Local Law 97 limit for 2030–2034?
For a general hospital — the Department of Buildings property type “Hospital (General Medical & Surgical)” — the 2030–2034 limit is 0.00734 metric tons of CO2 equivalent per square foot per year, set in rule 1 RCNY 103-14. The same hospital’s 2024–2029 limit is 0.02381, so the 2030 step is a cut of roughly 69 percent, among the deepest in the law.
Why can’t a hospital comply by turning down its ventilation?
Because the ventilation rates are infection-control requirements, not comfort settings. ANSI/ASHRAE/ASHE Standard 170 sets minimum air change rates by room type — 20 total air changes per hour in an operating room, 12 in isolation and protective-environment rooms, 4 in a patient room — along with the pressure relationships between rooms. The standard offers two legitimate reliefs: an unoccupied operating room may be set back while its pressure is held, and an isolation room not in isolation use may drop to 6 air changes. Beyond those, compliance has to come from recovering energy, not from moving less air.
Does converting from district steam to on-site gas boilers improve a hospital’s Local Law 97 position?
On the law’s arithmetic, no — it makes it worse. Purchased steam is charged at 0.00004493 tons per kBtu of steam delivered in 2024–2029. Natural gas is charged at 0.00005311 tons per kBtu of fuel burned, so a boiler at 80 percent efficiency is charged about 0.0000664 per kBtu actually delivered — roughly half again the steam figure. Steam supply carries its own strategic questions, but a conversion to on-site gas raises counted emissions per unit of heat unless it is paired with heat recovery or electrification that changes the equation.
Which mechanical measures usually matter most for a hospital’s number?
Practice points to a consistent order. First, heat recovery aimed at the reheat load — recovering condenser heat from the chiller plant to serve reheat, preheat and hot water instead of rejecting it at the tower. Second, electrification of the remaining heating load, which is the only input whose emissions coefficient is scheduled to fall to zero. Third, exhaust-air heat recovery wherever the air classification allows it, and retro-commissioning to keep all of it performing. The right order for a specific building comes out of its own submetered end uses, which is why the metering matters.
When do the fines actually start?
The first compliance period, 2024–2029, is already running, and emissions above a building’s limit are fined at $268 per ton per year in that period — though the limits for these years were drawn to catch the worst performers, and the city’s first-period rules included good-faith-effort provisions for buildings actively working a plan, with adjustments available for hospitals and other special cases. The limits that bind most hospitals are the 2030–2034 limits. Annual reporting is required either way, and there are separate penalties for failing to file.