HVAC and dehumidification for grow roomsHVAC and dehumidification for grow rooms

Purpose and scope

A grow room is a machine that turns electricity into light, light into plant, and water into vapour. The lights are the half everyone budgets for. The climate plant, cooling, heating, dehumidification, is the half that hauls the heat and the water back out, every hour of every day, and it typically eats 30–60% of an indoor facility’s energy bill^[1].

When it’s sized right you barely think about it. When it’s guessed, a mini-split off a BTU chart and a dehumidifier that looked big in the shop. You find out in week 6 of flower, at 2 a.m., when the room is cold, saturated, and growing Botrytis instead of flower.

This paper is written for someone speccing their first serious room or sanity-checking a mechanical quote. No refrigeration background assumed. By the end you should be able to do your own load arithmetic, read a dehumidifier spec sheet with suspicion, and explain exactly why the humidity pins to 85% the moment the lights go out.

Definitions

Evidence and limitations

Sensible and latent loads

A comfort air conditioner in an office does one job: remove dry heat. A grow room asks for two: remove heat and remove water. Engineers split these as the sensible load (temperature) and the latent load (moisture), and grow rooms are unusual because the latent side rivals or exceeds the sensible side for much of the cycle, a ratio almost no comfort-cooling equipment is built for^[2][4].

Where the sensible load comes from: the lights, overwhelmingly. In an insulated, sealed room, essentially every watt of electrical input becomes heat that the cooling plant must remove, and lighting is the single largest share^[2][3]. Dehumidifiers, fan motors, pumps and people add the rest.

Where the latent load comes from: the plants. A cannabis leaf is covered in tiny pores called stomata. When they open in the light, the plant draws water up from the roots and releases it as vapour into the air — the same way your skin releases sweat to cool you down. The plant isn’t wasting water; evaporation cools the leaf and pulls more water (and dissolved nutrients) upward from the root zone. Transpiration is this continuous, light-driven release of water vapour through the stomata. Transpiration is the latent load: of the water delivered to the root zone, roughly 95%+ passes up through the plant and out the stomata as vapour, Desert Aire’s engineering note puts it near 99% of water taken up by the roots^[2], and facility engineers design on 80–95% of total irrigation returning to the air^[7]. Your crop is not decoration sitting inside the climate system. Your crop is the humidifier, running at hundreds of litres a day, powered by your lights and throttled by VPD^[5].

Calculating room heat loads

Everything from here on uses one worked room so the numbers stay honest: 40 m² of flowering canopy (~430 sq ft), 10 × 700 W LED fixtures, sealed and CO₂-enriched, insulated internal walls, two 800 W dehumidifiers, ~500 W of circulation fans and pumps, two crew working in it. Swap in your own numbers, the method is the point.

Water balance and dehumidification load

Here is the single most useful sizing fact in this paper: your dehumidification requirement is written by your irrigation schedule. Not by room volume, not by plant count charts, by litres per day. Water that goes in and doesn’t leave down the drain leaves through the air^[8][7].

1. 1
   Count the water in

   40 m² of canopy × 6 L/m²/day at mid-flower = 240 L/day. Your irrigation controller already knows this number exactly.
2. 2
   Subtract collected runoff

   25% runoff captured to drain = 60 L that never touches the air. 240 − 60 = 180 L/day stays in the room. (Runoff left standing in trays evaporates, then it’s latent load again. Drain it.)
3. 3
   Subtract what the plant keeps

   Plant tissue holds only a few percent of uptake, call it 5 L/day. The rest transpires^[2]. ≈175 L/day becomes vapour.
4. 4
   Convert for spec sheets

   175 L ÷ 0.473 = ≈370 US pints/day. Quest’s shortcut (gallons fed minus drained, times 8) lands on the same answer^[8].

Cooling-load calculation

Every forum will tell you “3,000–4,000 BTU per 1,000 W of light”. Here’s the secret: that isn’t horticultural wisdom, it’s a unit conversion. 1 W of electricity makes 3.412 BTU/h of heat, always, by physics^[6]. The folklore is roughly right about the lights and silent about everything else. Which is how rooms end up 20–30% short.

1. 1
   List every watt in the room

   Lights 7,000 W. Dehumidifiers 1,600 W. Fans and pumps 500 W. Two crew ≈240 W^[6]. Envelope gain: ~0 here, real if your room has a hot roof or sun-struck wall.
2. 2
   Sum it

   7,000 + 1,600 + 500 + 240 ≈ 9.3 kW of sensible load.
3. 3
   Convert to trade units

   9.3 kW × 3,412 = ≈31,700 BTU/h = ≈2.6 tons of cooling.
4. 4
   Add margin, not hope

   20–25% covers hot ambients, dirty coils and derate with age^[6]: spec ≈11.6 kW (≈40,000 BTU/h, ≈3.3 tons).
5. 5
   Split it across units

   Two smaller units beat one big one: staged capacity for light loads, and a failure loses half your cooling, not all of it (Section 12).

Dehumidifier sizing

Dehumidifier sizing is the water balance from Section 05 plus two corrections everyone skips: when the moisture arrives, and what the machine actually removes at your conditions rather than at the rating point on the box.

1. 1
   Start from the vapour load

   Example room: ≈175 L/day ≈ 370 pints/day total.
2. 2
   Split day from night

   Transpiration doesn’t stop in the dark, stomata close partially, media keeps evaporating. Assume 70/30: day 122 L over 12 h (≈10 L/h), night 53 L over 12 h (≈4.4 L/h). Check the split against your own condensate volumes and correct it. It varies with cultivar and night climate.
3. 3
   Day duty

   With lights on, the AC coils condense a share of the moisture while cooling; dehumidifiers top up. This is the easy shift.
4. 4
   Night duty, the sizing case

   With lights off there is no sensible load, the AC stops, and its incidental moisture removal stops with it^[2]. The dehumidifiers alone must carry 4.4 L/h. That is 105 L/day of removal rate.
5. 5
   Derate the nameplate

   Ratings are quoted warm, commonly 26.7 °C (80 °F) / 60% RH, and refrigerant units remove less as the room cools^[10]. If the manufacturer’s curve shows ~⅔ of nameplate at your 19 °C (66 °F) night, you need ≈160 L/day of nameplate running to hold the night: e.g. two 80 L/day units flat out.
6. 6
   Then apply N+1

   Two units exactly covering the night means one failure ends the crop. Fit three 80s (or two 120s) so any single unit can die on a Saturday night without drama (Section 12).

Per fixture, the example works out to 17.5 L (≈37 pints) per light per day, a handy pub-quote number, but notice it’s downstream of the irrigation schedule, not a property of the light. Quest’s field guidance of 0.5–2 pints per square foot of canopy per day brackets the same range^[9], our room computes to ~0.86 pints/sq ft. When a rule of thumb and your arithmetic agree, you can trust the arithmetic; when they disagree, trust the arithmetic anyway.

HVAC equipment classes

Four families of cooling, plus the dehumidifiers that bolt onto all of them. Engineering firms describe the same ladder: packaged DX with standalone dehus at entry level, DX with hot-gas reheat in the middle, chilled water at scale^[3].

Standalone vs ducted dehumidifiers. Standalone (hang-above-canopy or floor) units are cheap, movable and simple, but they dump their heat and noise in the room and their condensate wants managing. Ducted/inline units sit outside the canopy, plumb their drains properly, and share the air-handling path, at higher install cost. Either way: plumb the drain. A bucket is a humidifier with extra steps.

Lights-off humidity spike

Watch any grow room’s trend graph and you’ll see the same signature: the moment the lights cut, RH leaps 15–25 points in under an hour. Three things happen at once, and every one of them pushes the same direction:

1. The sensible load vanishes. 7 kW of light heat disappears in one second. The AC, which was condensing moisture as a side effect of cooling, ramps to zero and takes its moisture removal with it^[2][9].
2. The air cools, so RH rises with no new water at all. Air’s capacity to hold vapour roughly halves per 10 °C (18 °F) drop. Cool the example room’s 26 °C (79 °F) / 55% air to 19 °C (66 °F) and it sits at ≈84% RH, same grams of water, smaller container. (Quest’s version of the same arithmetic: 24 °C (75 °F) at 57% becomes ≈80% at 18 °C (64 °F)^[9].)
3. The crop keeps transpiring. Slower in the dark, but far from zero, and wet media keeps evaporating all night. In the example room that’s still ≈4.4 L of new vapour every hour.

Why this window matters so much: bud rot (Botrytis cinerea) thrives in cool, near-saturated, still air, and dense late-flower colas hold exactly that microclimate internally^[11]. The room hits its highest RH at its lowest temperature, dew forms on whatever surface sits below the dew point (at 26 °C (79 °F) / 55%, that’s any surface under ~16 °C (61 °F), duct skins, exterior walls, cold glass), and the crop is at its most vulnerable stage. The night latent capacity you sized in Section 07 is not a comfort feature. It is mould control.

HVAC supply, return and circulation

The climate plant conditions air; the room still has to distribute it. Two systems, two jobs: the air handling loop delivers conditioned air and drags moist warm air back to the coils, sealed grow rooms typically turn the room’s air over 20–40 times per hour, recirculating ~100% of it to keep CO₂ and keep outside contaminants out^[3], while circulation fans stir the canopy so no leaf sits in its own humid boundary layer (that story is the airflow design paper).

• Supply high, return low. Dry supply air is less dense paths matter less than geometry: pushing supply across the ceiling and pulling return at floor level forces air through the canopy, where the load actually is.
• Don’t short-circuit the air path. A supply diffuser blowing straight into a nearby return conditions the duct, not the room. Sensors near that path read beautifully while the far corner rots.
• Place dehumidifiers deliberately. Discharge aimed along a wall or aisle, not blasting one bench of plants with hot dry air; intake sitting in the moist zone, not in its own dry plume, a unit re-breathing its own discharge reads a dry room and idles while the canopy stays wet.
• Watch compressor short-cycling. A grossly oversized single-stage AC satisfies the thermostat in minutes and shuts down before the coil ever gets cold and wet enough to condense much. You get temperature control and no dehumidification, plus compressor wear. Staged or inverter capacity, plus minimum-run timers, is the fix.

Condensate management and reuse

Everything the coils and dehumidifiers condense has to go somewhere, in the example room, ~175 L/day of it. That flow is a maintenance liability, a free instrument, and a potential water resource, in that order.

• A maintenance liability: every unit needs a trapped, sloped drain or a reliable condensate pump with a float cut-out. Pans and trays grow biofilm and drip on canopy; blocked drains shut units down (good ones) or overflow into ceilings (the rest). Tray and drain cleaning is an IPM task, not optional housekeeping.
• A free instrument: metered condensate is your measured latent load. Falling condensate at constant irrigation means either runoff went up or removal capacity went down, both worth knowing by breakfast.
• A resource: most of the irrigation water you paid for comes back as near-distilled condensate, and it can be captured and reused^[7].

Redundancy planning

Run the failure before it runs you. Example room, week 6, 23:00: two 80 L/day units are carrying the night at ≈⅔ nameplate. One trips on a failed capacitor. Do the arithmetic: the room’s ~150 m³ of air at 19 °C (66 °F) / 60% RH can only absorb about one more litre of water before saturation. And the crop is adding ≈4.4 L every hour. The surviving unit removes barely half of that. RH is against the ceiling within the hour, condensation starts on the coldest surfaces, and nothing else in the room can help because the AC has no sensible load to run against. This is not a slow drift you catch at the morning walk-through. It runs away in minutes.

N+1 is the fix, and it’s exactly what it sounds like: N units cover the design load; you install one more, so any single failure leaves the room fully served. Apply it in order of what kills the crop fastest:

1. Night dehumidification first. No fallback exists, when a dehu dies at night, nothing else removes water. Three 80s where two carry the load.
2. Cooling second. Day cooling failure has a fallback: shed the load. Interlock the lights so a cooling fault dims them to 50% or kills them. You can afford a lost day of photosynthesis; you cannot afford 40 °C over a full photoperiod. Two smaller ACs also beat one big one here.
3. Controls and alarms last but not least, the redundancy you don’t know has failed doesn’t exist. Alarms must reach a human who can act, and must survive the same power event that caused the fault.

HVAC staging and deadbands

A grow room runs heating, cooling and dehumidification within metres of each other, and two of the three make the third’s job worse: cooling raises RH; dehumidifying adds heat. Un-coordinated, they fight, the AC overcools, RH climbs, the dehu heats, the AC returns, around and around, burning power and cycling compressors. The cure is sequencing, not bigger hardware.

• Deadbands: control to a band, not a knife-edge. Day 26 °C (79 °F) ±0.5 °C, RH 55–60% is a structure (yours will differ, setpoints belong to the crop, see the VPD paper; cannabis photosynthesis runs happily around 25–30 °C (77–86 °F)^[13]). Tight bands feel professional and mostly buy you equipment cycling.
• Sequencing: heat and cool must never run together (lockout between them); dehumidification may run alongside either, but its reheat should come from the machine, hot-gas reheat, not from the heaters fighting the AC^[3].
• Day/night setpoints with ramps: separate targets for lights-on and lights-off, connected by 15–30 minute ramps so the transition is driven, not endured (Section 09).
• Control moisture on dew point where you can: %RH swings with every temperature wobble even when the water content hasn’t changed; dew point tracks the actual grams. At the day/night transition, dew-point control is dramatically calmer.
• Sensor placement is a control decision: an aspirated or well-shielded sensor at canopy height, out of any supply jet or dehu plume. The controller can only be as honest as its sensor (Section 12’s quiet killer).

HVAC efficiency, reheat and heat recovery

Climate is where the money goes. Energy runs 30–60% of indoor operating expense^[1], and life-cycle analysis of US indoor production found environmental control the dominant driver of both energy use and emissions, 2,300–5,200 kg CO₂e per kg of dried flower depending on location^[14]. Every design choice in this section moves that number.

• Hot-gas reheat is the flagship move. Dehumidification means cooling air below its dew point, then warming it back so you don’t overcool the room. Electric reheat pays full price (COP 1) for heat you just paid to remove. Hot-gas reheat recycles the compressor’s own rejected heat to do the rewarming, near-free reheat, standard on mid-tier DX and integrated units^[3].
• Right-size rather than oversize. Oversized single-stage equipment short-cycles: worse dehumidification, worse efficiency, shorter compressor life. Margin belongs in staged capacity (two circuits, inverter drive), not in one heroic unit.
• LEDs shift the ratio, not the rules. At equal PPFD, LEDs draw fewer watts, so the sensible load drops, but the crop transpires much the same, so the latent load doesn’t. LED rooms are latent-dominated rooms: expect the dehumidifier spec, and the winter heating question, to matter more, not less.
• Reuse heat you already own. Dehu heat holds the night temperature (Section 09); condenser heat can pre-warm dry rooms or water. Rejecting heat outside all winter while running COP-1 heaters inside is a bill you chose.
• Maintenance is an efficiency program: dirty filters and fouled coils quietly tax capacity and COP for months before anything actually fails. Filters monthly; coils each cycle.

Troubleshooting

Heat and moisture balance

1. Count watts → sensible load. Folklore counts only the lights; you count everything (Section 06).
2. Count litres → latent load. Water in minus runoff, converted to pints for the spec sheet (Sections 05, 07).
3. Night is the design case: no sensible load, no AC help, transpiration continuing. Size dehumidification for lights-off at your night temperature, derated from nameplate.
4. N+1 the night dehumidification; interlock the lights to the cooling. Then break each unit on purpose once, and watch what happens.
5. Control moisture (dew point) with deadbands and sequencing so the machines cooperate; ramp the day/night transition.
6. Meter irrigation, runoff and condensate, the free, continuous load calculation that also tells you when the crop changes.

The climate plant is one subsystem of the room. Read it alongside the grow-room systems guide, the temperature, humidity and VPD paper for what the setpoints should actually be, and the airflow design paper for the last metre of air movement.