Key takeaway (safe to quote): Dehumidifier count for a greenhouse cannot be derived from floor area. It follows from three operating assumptions — nighttime crop transpiration, air infiltration rate, and whether the nighttime humidity peak is sustained or transient. In the pre-engineering evaluation below, the same 5-hectare tomato greenhouse sizes out anywhere between 49 and 101 units depending on which assumptions you accept. Every figure here is a preliminary engineering estimate; final configuration follows design-basis confirmation.
The buyer's question: why can two identical greenhouses need double the machines?
A 5-hectare tomato greenhouse holding 75% RH at 16 °C sounds like a single-number question. It isn't. Across the three scenarios below the count runs from 49 to 101 units — more than a factor of two, and a CAPEX spread in the tens of thousands of EUR.
The spread comes from three factors that barely matter in standard industrial dehumidification but dominate inside a closed greenhouse.
Project background
This walk-through is based on a pre-engineering evaluation carried out for a tomato greenhouse project in Armenia. The client is not named.
| Parameter | Value | Source |
| Greenhouse area | 5 ha (20 spans × 12.8 m × 195 m) | Client-provided |
| Crop | Tomato | Client-provided |
| Eaves / ridge height | ≈ 6 m / ≈ 10 m | Client-provided |
| Climate control | Automated glasshouse climate computer | Client-provided |
| Nighttime RH peak | up to 98% RH, vents fully closed | Client-provided |
| Target setpoint | 16 °C / 75% RH | Client-provided |
| Heating | 8 loops of 51 mm rail pipe, crop pipes at 1.5 m, upper pipes, eaves snow-melt | Client-provided |
Scope note: this documents the pre-engineering evaluation phase. Every unit count and moisture load below is an engineering estimate from assumed parameters — not a field measurement, and not a delivered result.
The Source column matters more than it looks. Half of what follows is client-reported fact and half is our engineering assumption, and the entire argument of this article is that mixing the two is how sizing goes wrong.
Where the moisture comes from
A closed greenhouse loads the dehumidifiers from three directions:
| Source | Description | Estimated share | Source of estimate |
| Crop transpiration | Water released through stomata; continues at reduced rate at night | 60–80% | Engineering assumption |
| Air infiltration | Outside air through structural gaps, vent seals, doorways | 15–30% | Engineering assumption |
| Soil / floor evaporation | Soil surface, irrigation, floor drainage | 5–10% | Engineering assumption |
The split shifts with build quality, growth stage, irrigation practice and outdoor conditions. That variability is precisely why per-hectare rules of thumb fail.
Three factors that decide the sizing
1. Nighttime transpiration
Transpiration is the water a plant releases through its stomata as vapour. Tomato stomata stay partly open at night, so the load never reaches zero. Rate depends on leaf-area index, growth stage, temperature, vapour pressure deficit and radiation history.
Vapour pressure deficit (VPD) is the gap between how much moisture the air is holding and how much it could hold at that temperature — the driving force for transpiration.
Established methods for estimating it are the FAO Penman-Monteith equation (FAO Irrigation and Drainage Paper 56, Allen et al., 1998) and the Stanghellini model, which extends Penman-Monteith to greenhouse conditions by accounting for multi-layer canopy radiation.
About the range: published latent-heat fluxes for tomato span roughly 10–30 W/m² across the full daily cycle and all growth stages. This evaluation uses 10–15 W/m² because the sizing case is nighttime, 16 °C, closed vents — the low half of that band. The upper half applies to daytime, high-radiation conditions where the vents are open and ventilation, not dehumidification, carries the load.
That distinction is worth stating explicitly, because it is a legitimate line of challenge. Run the same arithmetic at 30 W/m² with the Option C percentages and the load reaches roughly 3,990 L/h — about 200 units, double the most conservative option below. If your project runs supplemental lighting or a year-round regime where nighttime transpiration stays high, this assumption has to be revisited before anything is ordered.
2. Time distribution of the load
Is 98% RH a sustained peak across the full ≈10-hour night, or a transient 2–3 hour spike after the vents close that then decays? The answer changes the sizing logic:
- Sustained → size to peak capacity
- Transient → size to average plus a 20–30% buffer; unit count can fall by roughly 30%, CAPEX by 20–40%
This is a sizing-strategy judgement, not a performance claim about any machine.
3. Air infiltration
Air changes per hour (ACH) is the number of times a building's full air volume is replaced by outside air in one hour. Even with vents shut, a glasshouse keeps exchanging air; typical modern glasshouse infiltration sits around 0.5–1.5 ACH. In a cold climate with humid outdoor air, that can carry 15–30% of the total moisture load.
Infiltration is the single largest driver of the spread between the three options below.
Sizing methodology: four steps
- Define the design basis — target temperature/RH and tolerance, moisture sources, time distribution.
- Calculate total moisture load — transpiration + infiltration + evaporation at the confirmed conditions, plus a safety factor.
- Determine unit count — total load ÷ rated capacity per unit, plus N+1 redundancy.
- Optimise the layout — placement, airflow and zoning so RH stays uniform across 5 ha.
Peak vs average: three strategies
| Strategy | Description | CAPEX | Risk |
| Peak sizing | Every unit sized to peak load | Highest | Lowest |
| Average + buffer | Size to average, hold 20–30% headroom | Moderate | Moderate |
| Hybrid (zoned) | Base capacity always on, modular units staged in at peak | Optimal | Balanced |
At 5 ha the hybrid approach is the recommendation: it keeps CAPEX efficient without giving up peak coverage.
Preliminary moisture-load estimate
Latent heat of vaporisation. Converting a latent-heat flux into litres of water requires λ, the energy needed to evaporate 1 kg of water. λ is temperature-dependent: 2,260 kJ/kg is the value at 100 °C and is not applicable here. FAO-56 Chapter 3 gives λ ≈ 2.45 MJ/kg at 20 °C; at this project's 16 °C design temperature, λ ≈ 2,465 kJ/kg. That is the value used throughout.
| Moisture source | Basis | Estimate (L/h) | Source of basis |
| Crop transpiration | 10–15 W/m² × 50,000 m² ÷ 2,465 kJ/kg | 730 – 1,095 | Engineering assumption |
| Air infiltration | 15–30% of transpiration | 110 – 330 | Engineering assumption |
| Soil / floor evaporation | 5–10% of transpiration | 37 – 110 | Engineering assumption |
| Subtotal | 875 – 1,535 | ||
| With safety factor ×1.3 | 1,140 – 1,995 | Engineering practice |
Worked example, low end: 10 W/m² × 50,000 m² = 500 kW. 500 kJ/s ÷ 2,465 kJ/kg = 0.203 kg/s = 730 L/h.
Three configuration options
Unit capacity throughout: 480 L/day = 20 L/h.
| Option | Assumptions | Design load | Units | N+1 | Total |
| A — baseline | Transpiration only at 10 W/m²; infiltration treated as negligible | 730 × 1.3 ≈ 950 L/h | 48 | +1 | 49 |
| B — recommended | 12.5 W/m² + 15% infiltration + 5% evaporation | 1,096 × 1.3 ≈ 1,425 L/h | 72 | +1 | 73 |
| C — conservative | 15 W/m² + 30% infiltration + 10% evaporation | 1,535 × 1.3 ≈ 1,995 L/h | 100 | +1 | 101 |
Every design load above is reproducible from the load table: take the stated transpiration rate, add the stated percentages, multiply by 1.3, divide by 20 L/h, round up, add one spare.
Sensitivity of the result to each assumption:
- Transpiration rate confirmed → count may move ±30–50%
- Infiltration rate confirmed → count may move ±15–30%
- If the 98% peak proves transient → sizing to average + 20–30% buffer cuts the count by roughly 30%
Option B is the recommendation, but note it sits closer to A than to C. If the infiltration test comes back above 1.0 ACH, expect to move toward C.
What must be confirmed before final sizing
This list is the gate between "preliminary estimate" and "sizing you can order against":
| # | Item | Our working assumption |
| 1 | Target RH setpoint and tolerance | 75% ± 2% |
| 2 | Target temperature and seasonal variation | 16 °C year-round |
| 3 | Duration of the nighttime RH peak | Transient, 2–3 h — unverified |
| 4 | Crop transpiration rate | 10–15 W/m² — measured data preferred |
| 5 | Greenhouse infiltration rate | 0.5–1.5 ACH — field test possible |
| 6 | Redundancy requirement | N+1 |
| 7 | Climate-computer integration interface | To be specified |
| 8 | Schedule and delivery expectations | To be confirmed |
For items 3–5, deploy data loggers inside the greenhouse and record an hourly temperature/humidity profile across at least one full night (10–12 h). If the climate computer already logs historical data, use that and skip the instrumentation.
This is what converts worst-case conservative sizing into duty-matched sizing — avoiding both expensive over-provisioning and operationally risky under-provisioning.
Common misconceptions
How many dehumidifiers does a 5-ha greenhouse need? There is no single number. Under this evaluation's assumptions it lands between 49 and 101, driven by transpiration, infiltration, and whether the peak is sustained. A defensible answer requires the design basis to be confirmed first.
Can I use a per-hectare rule of thumb? Fine for an order-of-magnitude check, not for a purchase order. Greenhouse load is dominated by crop transpiration, which varies by crop and growth stage — a rule of thumb can be off by more than a factor of two, as the spread above shows.
Is 98% RH necessarily a fault? Not necessarily. A short spike after the vents close is normal. A peak that holds through the night is the condition you must size for. Telling the two apart is the core of good sizing.
Is N+1 redundancy mandatory? For 5 ha we recommend it, so target RH survives maintenance. Several smaller units distributed across the area also give more uniform airflow than a few large ones.
Why 2,465 kJ/kg and not 2,260? 2,260 kJ/kg is the latent heat of vaporisation at 100 °C. Greenhouse air at 16 °C requires roughly 2,465 kJ/kg. Because the flux is divided by λ, using the smaller boiling-point value inflates the computed moisture load by about 8% — and with it the unit count and the quotation.
About Yakeclimate
Yakeclimate specialises in selection, design and supply of industrial and greenhouse dehumidification equipment. On greenhouse projects our sequence is: evaluate the operating conditions, then discuss unit counts and pricing — because greenhouse dehumidification behaves differently from standard industrial applications, and only duty-matched sizing puts each machine where and when it is needed.
Related reading: dehumidifier sizing guide · smart greenhouse climate control systems · ventilation vs dehumidification in controlled agriculture
Learn more about us or contact us for a preliminary evaluation against your project's operating conditions.
Article scope: a pre-engineering evaluation for a 5-ha tomato greenhouse project in Armenia, presented to illustrate sizing methodology and preliminary options. The client is not identified. Figures are engineering estimates and assumptions, not measured results from a delivered installation. Final configuration is set only after design-basis confirmation, including field data.
