Rooftop unit (RTU) installation and startup field guide

What is a rooftop unit (RTU)?

A rooftop unit is a packaged HVAC system that holds everything in one cabinet and sits on a curb on the roof. Direct-expansion cooling, a heating section, the supply and return blowers, the filters, and usually an economizer all ship as one assembly. You set it on the curb, connect duct, power, and gas, and it conditions the space below. That is the whole appeal: one unit, one set of connections, no field-built mechanical room.

Packaged means the refrigeration circuit is sealed and charged at the factory. Unlike a split system, you are not brazing line sets and pulling a vacuum on a field circuit. The compressors, the condenser coil, the evaporator coil, and the metering device come matched and charged with a holding charge from the plant, which changes what startup is. Your job is less about building the circuit and more about proving the factory circuit, the curb, and the controls are right before the building runs on it.

The thing that bites crews is treating an RTU like a plug-and-play appliance. It is not. A packaged unit that was rigged onto a crooked curb, never had its shipping bolts pulled, runs a compressor backward, and has a condensate pan that cannot drain will still hum and blow air on day one. It just fails slowly, and the callback lands on whoever signed the startup sheet.

RTU types and tonnage

RTUs are sorted by how they heat and how big they are. The heating section is the first split. A gas/electric unit cools on DX and heats with a gas furnace section. An electric/electric unit cools on DX and heats with electric resistance strip heat. A heat pump reverses the refrigerant circuit to heat, often with electric strip backup for the cold end of the range. And a cooling-only unit has no heat at all, used where heat comes from somewhere else or the climate does not need it.

Size is given in tons of cooling, where one ton is 12,000 Btu/h. Light commercial packaged units run roughly 3 to 25 tons as single-zone constant-volume or with a VAV section; above that you are into larger packaged and custom air handling territory. The tonnage drives the curb size, the rigging weight, the electrical service, and the gas input, so the unit on the order has to match the structure, the power, and the gas that were designed for it.

Match the unit to the job before it leaves the supply house. A gas/electric unit needs gas piping run to the roof. A heat pump needs the defrost and backup heat sized for the climate. The voltage and phase on the nameplate have to match the building service. Swapping a unit type or a voltage after the curb is set and the roof is penetrated is an expensive way to learn to read the schedule.

The roof curb: level, gasketed, and watertight

The roof curb is the steel frame that ties the unit to the structure, carries its weight, routes the supply and return duct through the roof, and keeps water out. Get the curb wrong and nothing downstream saves you. The two things that matter most are that it sits dead level and that it is sealed watertight to both the roof and the unit.

Level is not a nicety. A unit set out of level drains condensate to the wrong corner so the pan holds water, it loads the bearings and the compressor unevenly, and it vibrates harder, which works fasteners loose over time. Set the curb with a level across both axes and shim it true before the unit ever lands. On a sloped roof, that means a pitched or stepped curb built to bring the unit to level, not a unit tilted to match the roof.

Watertight is the other half. The curb gets flashed and counterflashed into the roof system so water is shed away from the seam, and a continuous gasket runs along the top rail so the unit seals to the curb when it is set. That gasket is one continuous run with no gaps at the corners, because a gap at a corner is a leak straight into the duct or the building. Set the unit down once onto the gasket. Picking it back up and resetting it crushes the gasket unevenly and you have lost the seal you came for. Confirm the curb dimensions, the gasket, and the flashing detail against the unit manufacturer's curb instructions and the roofing system requirements, because a curb that fits the steel but voids the roof warranty is its own problem.

Rigging the unit and the duct connection

Rigging is the part where weight and gravity stop being abstract. A packaged unit is heavy, the center of gravity is rarely centered, and it has to land on the curb in one controlled set. Use the manufacturer's lifting points and a spreader bar so the slings pull straight up instead of crushing the cabinet, and confirm the rigging weight against the unit's shipping weight, not a guess. The crane pick, the spotter, and the roof crew work to a plan, because a unit swinging in wind over an occupied building is a serious-injury event, not a scheduling inconvenience.

The structure has to carry it. The dead load of the unit, plus the curb, plus snow and live load on the roof, is a structural question the engineer signs off on, and a unit set on a roof that was not designed for it is a real failure, not a paperwork one. Where the existing structure is light, the design adds steel under the curb to spread the load to the joists.

The duct connects through the curb. Supply and return drop through the curb opening into the building ductwork, and the transition has to be sealed and, where the spec calls for it, lined and isolated so the unit's vibration does not telegraph into the building as noise. Internal curb ducting and a curb-mounted economizer hood take up room inside the curb, so the duct drops have to be coordinated with what the curb already holds. Seal the duct-to-curb joint, because air lost here is air the blower moves and the building never gets, and it shows up later as a balance that will not close. The air balancing guide covers proving that airflow at the registers once the unit is running.

Why does an RTU need a condensate trap?

A draw-through RTU needs a condensate trap because the drain pan sits on the suction side of the blower, under negative pressure, and without a trap the fan pulls air up the drain line instead of letting water flow down it. The evaporator pan is downstream of the coil and upstream of the fan, so the fan is trying to suck air in through every opening, including the drain. Plumb an untrapped drain and the negative pressure holds the water in the pan, the pan overflows into the unit and the duct, and you have a water-damage callback that looks like a leak but is a missing trap.

The trap depth has to beat the negative pressure. The common field method is to make the trap seal deeper than the static the blower pulls at the pan, with a safety margin, so a unit pulling 1 in. w.c. negative at the pan wants a trap roughly an inch deeper than that to hold its seal and still let water move. Size it for the worst case, which is a loaded filter and a dirty coil that push the negative static higher than a clean startup reading, not the day-one number. A trap that holds at startup and breaks its seal when the filter loads is a trap that fails exactly when the unit is working hardest. Confirm the depth against the unit's installation instructions, because the manufacturer ties it to the rated static.

Slope the line away from the unit so gravity carries the water, prime the trap before startup so the seal exists before the fan pulls on it, and run a primary drain plus an overflow path so a plugged primary does not put water in the building. Many jobs add a secondary pan or an overflow switch that shuts the unit down on a high level. Pour water in the pan at startup and watch it leave through the trap. That two-minute test catches the missed trap before the ceiling does.

Gas piping, pressure, and combustion

On a gas/electric unit the gas line is run to the roof, sized for the unit's input, and proved before the furnace section ever fires. The pipe is sized for the Btu/h input over the developed length of the run, and a line sized for the appliance but not the run starves the unit at full fire, so size the gas to the input and the distance, not the connection size at the unit. Refer to a gas pipe sizing procedure for the developed-length method and the table that turns input and length into pipe size.

Two pressures matter and they are different. The inlet, or supply, pressure is what the line delivers to the unit, and natural gas commonly arrives around 7 in. w.c. at the unit with the furnace running, while propane runs higher; the manufacturer states the required inlet range and the maximum it will tolerate. The manifold pressure is what the gas valve delivers to the burners, and for natural gas that is commonly set near 3.5 in. w.c., adjusted to the rating plate. Both get measured with a manometer at startup, inlet at the valve inlet tap and manifold at the outlet tap, with the unit firing.

Combustion is the proof the gas side is right. After the manifold pressure is set, check the flame, the draft, and combustion with an analyzer, looking for a clean burn and carbon monoxide within the acceptable range, because a furnace set rich or starved makes CO and a cracked or sooted heat exchanger is a life-safety problem, not a performance one. The gas code and the manufacturer's instructions govern the piping, the pressures, and the venting; set the manifold to the rating plate and confirm everything against the adopted code edition.

Electrical: disconnect, MCA, MOCP, and the whip

The electrical side starts at the nameplate, not the panel. The unit's data plate lists the minimum circuit ampacity (MCA), which sizes the conductors, and the maximum overcurrent protection (MOCP), which caps the breaker or fuse size. You size the wire to the MCA and you do not exceed the MOCP, because the MOCP protects the unit's internal components and going over it defeats that protection. Read both off the plate and size the circuit to them, not to a rule of thumb.

A disconnect within sight of the unit is the rule, so a tech on the roof can kill power without trusting that someone left the panel alone. The unit ties in through a whip, a short flexible conduit run that lets the unit move on its vibration isolation without stressing the conduit. Confirm the disconnect rating, the conductor sizing to the MCA, and the overcurrent device against the MOCP and the adopted electrical code, and verify the equipment grounding conductor is sized and landed, because the ground is the fault path that has to work when something inside the unit fails.

Voltage and phase have to match the nameplate before you energize. A 208 V unit fed 480 V is destroyed in an instant, and a single-phase unit on a three-phase feed, or the reverse, will not run right. Check the supply voltage with a meter at the disconnect, confirm it matches the plate within tolerance, and confirm the phase. On a three-phase unit the next check is rotation, and that one has its own section because it is the most expensive thing a crew gets wrong on an RTU startup.

Why check rotation on a 3-phase RTU?

You check rotation on a three-phase RTU because a three-phase scroll compressor will run backward if the power legs are crossed, and backward it pumps almost nothing, makes noise, and damages itself if it runs that way. The scroll only compresses in one direction. Feed it the wrong phase sequence and the scrolls turn the wrong way, so the compressor draws low current, the suction does not drop, the head does not rise, and it gets loud. The fix is simple, but only if you catch it in the first minute.

The tell is in the pressures and the sound. On a correct start, suction pressure drops and discharge pressure rises as the compressor pulls down and pumps up. On a reversed start there is little or no pressure split, the compressor is abnormally noisy with a knocking or loud hum, and the current draw runs below the nameplate value instead of where it should be. Watch the gauges and listen the moment it starts. The gaugegatedc tool keeps both pressures in front of you so the no-split condition is obvious as it happens.

The correction is to swap any two of the three line legs at the disconnect, with the power off and verified dead, which reverses the phase sequence and the rotation. A scroll run backward for a short spell will not be ruined, and its internal protector trips after a few minutes, but repeated restarts in reverse will kill it, so do not keep cycling it to listen again. Confirm rotation once, correct it if needed, and confirm the supply fan and any three-phase condenser fans turn the right way too, since they are on the same supply and the airflow direction matters as much as the compressor.

Pre-start: what to check before you energize

The pre-start is the walk-around that keeps the first energization from being the first failure. Before any power, the install gets a once-over: the unit level on the curb, the curb sealed, the duct connected and sealed, the condensate trapped and primed, the disconnect and whip landed, the gas piped and leak-checked. Then the unit-specific items that ship with the machine.

Pull the shipping bolts and the shipping blocks. Compressors and sometimes the blower assembly are bolted down or blocked for transport, and a unit that runs with the shipping hold-downs still tight cannot move on its isolators, so it transmits vibration into the structure and stresses the very components the isolation protects. The manufacturer marks them, often with a tag or paint. Find them and remove what the instructions say to remove. Set clearances too, because a unit jammed against a parapet or another unit starves the condenser of air and short-cycles on high head.

Then the mechanical checks. Filters in, clean, and the right size, because a missing or wrong filter fouls the coil from day one and throws off every airflow reading you are about to take. On a belt-drive blower, confirm the belt tension, the sheave alignment, and the rotation; on a direct-drive or ECM blower, confirm it spins free. Confirm the factory holding charge is still in the unit by reading a standing pressure, because a unit that shipped with a leak has lost its charge before you ever started, and you want to know that before you call low pressures a charging problem.

• Unit level on the curb, curb sealed, duct connected and sealed.
• Condensate trapped, primed, sloped, with a primary and an overflow path.
• Shipping bolts and blocks removed per the unit's instructions.
• Clearances around the condenser and the service side meet the unit's minimums.
• Filters installed, clean, and the correct size.
• Belt tension, sheave alignment, and blower rotation confirmed.
• Factory holding charge confirmed by a standing pressure before startup.
• Disconnect, whip, ground, gas, and leak check all complete.

How do you verify the refrigerant charge on an RTU?

You verify the charge on a packaged unit by the metering device: subcooling on a TXV unit, superheat on a fixed-orifice unit, with the system stabilized and the outdoor and indoor conditions in range. The factory charged the circuit, so this is a verification, not a field charge from empty, and the method follows the metering device because that is what the readings actually tell you.

On a thermostatic expansion valve (TXV) the charge is read by subcooling, commonly a target around 10 to 15°F, set to the manufacturer's number. The TXV holds superheat at the evaporator by metering to demand, so superheat does not track charge in a useful way, while subcooling at the condenser outlet rises as you add refrigerant and falls as you remove it. That makes subcooling the charge indicator on a TXV unit. You still read superheat as a check that the valve is feeding right, but you set the charge to subcooling.

On a fixed-orifice or piston unit there is no self-adjusting valve, so the charge is read by superheat, commonly 10 to 20°F, against a target that depends on the indoor wet-bulb and outdoor dry-bulb conditions at the moment. Use the manufacturer's charging chart or a superheat calculation for the conditions, not a single memorized number. Let the system run and stabilize before you trust any reading, check the suction and head pressures against the conditions, and confirm a healthy temperature split across the evaporator coil, commonly in the high teens to low twenties of degrees on a properly charged and properly airflowed unit. The refrigerant charging procedure covers the subcool and superheat methods in detail; the point here is to verify the factory charge by the right method and not to chase pressures with airflow problems still in the system.

Setting airflow: ESP, CFM, and the blower

Airflow gets set before the charge is final, because every cooling and heating reading depends on it. The measure is external static pressure (ESP), the total resistance the blower works against outside the cabinet, read by taking a static reading on the supply side and on the return side at the manufacturer's test points and combining them. You compare that ESP and the measured airflow to the unit's blower table to confirm the fan is delivering the design CFM at the static the duct actually presents.

CFM and ESP move against each other. Higher external static means lower airflow off a given blower setting, and lower static means more. The blower table is the map: find the speed tap, pulley setting, or VFD command that lands the design CFM at the system's real ESP, and stay in the fan's efficient range rather than at the edge of the table. On a belt-drive unit you set airflow with a sheave or pulley change and confirm with a tach and the table; on an ECM or VFD unit you set the speed and confirm against the table. The cooling design often targets a value in the range of 350 to 400 CFM per ton, with the manufacturer's number governing.

An ESP that reads high out of the gate is a duct or filter problem, not a fan problem, and overspeeding the blower to force airflow against a bad duct just burns motor amps and money. Read the ESP, find which side carries the restriction, and fix the duct before you accept the airflow. Proving design airflow at every register is the air balancing job, and the air balancing guide walks setting total air at the fan and proportioning the branches. The startup job is narrower: confirm the unit moves its rated air at the static it sees, so the charge and the temperature rise you set next are set on real airflow.

What is the temperature rise on gas heat?

Temperature rise is the difference between the return air entering the furnace section and the supply air leaving it, and it has to land inside the range stamped on the unit's rating plate, commonly something like 40 to 70°F. The rating plate gives a low number and a high number, and the measured rise has to fall between them. Below the range means too much airflow or too little fire; above it means too little airflow or too much fire, and a furnace running over its rise range overheats the heat exchanger and trips the high limit.

You measure it with the furnace at full fire and the airflow set. Read the return air temperature and the supply air temperature, taking the supply reading far enough downstream that you are not picking up radiant heat straight off the heat exchanger, then subtract. The two levers are gas input, set by the manifold pressure, and airflow, set by the blower. If the rise is high, you usually need more airflow before you touch the gas; if it is low, confirm the airflow first, then the manifold pressure against the rating plate.

The limit switch is the backstop, not the setting. A furnace that only stays in range because the high limit keeps cutting the burners is a furnace running too hot for its airflow, and the limit cycling is a symptom to fix, not a control to rely on. Set the airflow, set the manifold pressure to the plate, confirm the rise lands in the nameplate range, and verify the flue and combustion are clean. The manufacturer's rise range and manifold pressure govern; the adopted mechanical and gas codes govern the venting and the install.

The economizer on the RTU

Most commercial RTUs ship with or are ordered with an air-side economizer, the linked dampers and changeover control that bring in cool outside air for free cooling when the outdoor conditions allow. On a packaged unit the economizer lives in the mixing section ahead of the coil, with an outside-air hood, a return-air damper, and a relief path, run by the unit controller or a separate economizer control.

At startup the economizer needs three things confirmed: the dampers stroke from minimum to full and back, the changeover high-limit is set to the climate zone and the energy code, and the minimum outside-air position delivers the design ventilation rate. Watch the dampers move at the louver, not just the actuator, because the most common economizer fault is an actuator that strokes while a slipped linkage leaves the blade barely moving. An economizer left unset throws away the free cooling it was paid for and nobody notices, because the space stays comfortable on mechanical cooling.

The full functional test, the high-limit types by climate zone, the minimum outside-air floor from the ventilation standard, the sensor calibration, and demand-control ventilation are their own discipline. The economizer and demand-control ventilation guide covers commissioning the economizer end to end. The RTU startup job is to confirm the economizer is present, the dampers and linkage work, the changeover is set, and the minimum position is measured, then hand the deep functional test to commissioning.

Controls: thermostat, DDC, staging, and the BMS

The unit has to talk to whatever runs the building. The simplest case is a thermostat wired to the unit's terminal board, calling cooling and heating stages and the fan. The more common commercial case is a direct digital control (DDC) interface, where a unit controller takes setpoints and occupancy from a building management system (BMS) over a control network and stages the unit to match.

Staging is where comfort and equipment life are won or lost. A two-stage cooling unit should bring on the second compressor or the second circuit only when the first cannot hold setpoint, and the staging delays and deadbands keep the unit from short-cycling, which is what wears compressors out. Confirm the stages come on in order, that the anti-short-cycle timers are set, and that the heating and cooling do not fight across the deadband. On a unit with economizer, confirm the staging integrates with free cooling so the economizer carries what it can before mechanical cooling stages on.

Occupancy is the other control that saves money quietly. The unit should set back or shut down outside occupied hours and return to setpoint before the building fills, and the occupied and unoccupied schedules, the night setback, and any optimal-start logic get confirmed against what the BMS is sending. A unit running occupied setpoints around the clock because the schedule was never set burns energy every night with nobody in the building to benefit. Confirm the controller is addressed on the network, the points map to the BMS, and the schedule is real.

Duct smoke detector and unit shutdown

Larger RTUs carry a duct smoke detector that shuts the unit down when it senses smoke in the airstream, so the air handling system does not pump smoke through the building during a fire. The detector samples the duct air, and on a smoke signal it stops the fan and closes the unit down through the shutdown relay, often while signaling the building fire alarm.

The threshold is a code question tied to airflow. Both the mechanical code and the air-conditioning systems standard commonly require a duct smoke detector on systems above 2,000 CFM, with the detector placement and the supply-versus-return location set by the adopted code, and a return-side detector typically required upstream of any filters or outside-air connection. Larger systems and systems serving more than one story carry additional requirements. Confirm the threshold, the location, and whether supply, return, or both are required against the adopted mechanical code and the air-conditioning systems standard, because the exact trigger and placement vary by edition and jurisdiction.

At startup the detector is not just installed, it is proved. The detector is wired and installed to the fire alarm standard, tied to the unit shutdown, and functionally tested so a smoke signal actually stops the fan and reports where it should. A duct detector that is mounted but never shuts the unit down is a checkbox, not a safety device. The fire alarm interface, the detector listing, and the wiring follow the fire alarm code and the integration with the building system; coordinate the shutdown and the alarm signal with the fire alarm scope by topic, and prove the shutdown function during startup.

Efficiency: SEER, IEER, and why startup protects it

The rated efficiency on the nameplate is a lab number, and the unit only delivers it if the charge and the airflow are right. Packaged units are rated by AHRI, with SEER2 describing seasonal cooling efficiency and IEER describing part-load efficiency across the operating range, which matters because a commercial unit spends most of its hours at part load, not full load. Those ratings assume the unit is charged and airflowed to spec.

Undercharge, overcharge, and low airflow each erode the rated efficiency in the field, and they stack. A unit low on charge runs poor capacity and high superheat; a unit high on charge runs high head and wastes power; a unit starved for airflow by a dirty filter or a bad duct loses both capacity and efficiency and ices the coil. The startup readings, subcooling or superheat, ESP and CFM, and the temperature split, are exactly the parameters that hold the unit to its rating. Set them right and the unit performs near its rating; skip them and the building pays the difference on every bill for the life of the equipment.

This is the argument for commissioning over a quick turn-on. The efficiency that justified buying a higher-rated unit only shows up if startup proves the charge and the airflow, because the rating is a promise the install has to keep.

Commissioning readings and the startup report

Commissioning an RTU is taking the readings, comparing each to its target, and writing it down so the unit's condition the day it was handed over is on record. A startup that produced no numbers is not a startup. The report is the proof the unit was set, the baseline the next tech measures against, and the document that answers the warranty question when the unit acts up in a year.

Capture the charge readings, subcooling or superheat with the metering device noted, the suction and head pressures with the outdoor and indoor conditions, the evaporator temperature split, the supply and return ESP and the measured CFM, the compressor and fan motor amps against the nameplate, the rotation confirmation, the gas manifold pressure and the measured temperature rise on a gas unit, and the economizer minimum and changeover. Note what passed, what failed, and what you corrected, because a startup sheet with every box checked and no findings on a real unit gets the same hard look a TAB report does.

The manufacturer's startup form is usually the right base to work from, because it lists the parameters that unit needs and ties the warranty to them. Fill it completely, attach the readings, and keep it. The tradeos tool keeps the unit data, the readings, and the photos together so the startup record is one package instead of a clipboard that gets lost between the roof and the office.

What the owner has to maintain

A clean startup hands the owner a unit, and the unit hands the owner a maintenance list. Tell them what it is, because the items that fail an RTU early are the ones nobody does. The startup is the moment to set the expectation, since the parameters you just recorded are the same ones maintenance has to hold.

Filters are first and most ignored. A loaded filter raises the ESP, drops the airflow, ices the coil in cooling, and pushes the temperature rise high in heating, so a filter schedule is not optional housekeeping, it is what keeps the unit at the performance you just set. The belt on a belt-drive unit wears and slips and needs tension checks and replacement. The condenser and evaporator coils foul and need cleaning to hold capacity and head pressure. The condensate trap and pan need clearing before they plug and overflow. And the economizer, the silent failure, needs its dampers, linkage, and sensors checked, because a stuck economizer wastes free cooling without a single complaint.

Frame it as protecting the efficiency the unit was bought for. The owner paid for a rated unit and a commissioned startup; the maintenance is what keeps both from drifting away over the next few years.

Process and data center cooling units by topic

Not every packaged unit cools people. Process cooling and data center applications run packaged and split units to hold a tight temperature and humidity band on equipment, and the install and startup discipline carries over with the stakes raised. A data center cannot ride out a cooling failure the way an office can, so redundancy, continuous run, and tight control replace comfort tolerance.

The differences that matter are sensible load, tight control, and free cooling. IT load is almost all sensible heat with little latent, so these units run high sensible capacity and the humidity control is about holding a band, not wringing out moisture. The control band is tighter than comfort cooling, and the thermal envelope the equipment can accept governs how much free cooling is available, which is why economization is a large lever in this space. The same curb, condensate, rotation, charge, and airflow checks apply, with redundancy and continuous operation layered on top.

The deep treatment of free cooling and the equipment thermal envelope lives in the economizer and demand-control ventilation guide, which covers the data center envelope and how it sets economizer hours. The point here is that an RTU or packaged unit serving a process or data center load is the same machine with tighter targets and less tolerance for a startup that skipped a step.

What to document

The startup record is a table of readings against targets, with a pass or fail on each, plus the unit data and the corrections you made. Record enough that the next tech can reconcile the unit's condition and the warranty claim has the numbers behind it. The table below is the core reading set; the unit nameplate data, the conditions at the time, and the deficiency list ride alongside it.

Take the readings with the system stabilized and the conditions noted, because a subcooling or superheat number without the outdoor and indoor conditions cannot be reproduced, and a temperature rise without the airflow is half a measurement. Note the metering device, the fuel, and whether the unit has an economizer, so the reader knows which targets apply.

Common mistakes

• Plumbing the condensate drain with no trap on a draw-through unit, so the pan holds water and overflows.
• Setting the unit on a curb that is not level or not sealed watertight, causing a drain that will not clear and roof leaks.
• Running a three-phase scroll compressor backward because the rotation was never checked at the disconnect.
• Accepting low pressures as a charging problem when airflow was never set and the coil is starved.
• Setting the charge by superheat on a TXV unit or by subcooling on a fixed-orifice unit, the wrong method for the device.
• Leaving the gas temperature rise out of the nameplate range and relying on the high limit to cut the burners.
• Overspeeding the blower against a high ESP from a bad duct instead of fixing the duct.
• Leaving the shipping bolts in, so the unit cannot move on its isolators and shakes the structure.
• Installing a duct smoke detector but never proving it shuts the unit down.
• Calling the unit done with no startup readings recorded, so there is no baseline and no warranty proof.

Field checklist

Standards and references

The manufacturer's installation and startup instructions govern the RTU. They set the curb dimensions and gasket, the condensate trap depth tied to the rated static, the refrigerant charge targets by metering device, the airflow and blower table, the gas manifold pressure and temperature-rise range, and the MCA and MOCP on the nameplate. When any other document and the manufacturer disagree on a unit setpoint, the manufacturer's instructions and the listing control, and going against them voids the warranty.

Around that sit the codes and rating bodies, each owning its piece. AHRI rates the packaged equipment, with SEER2 and IEER describing the efficiency the startup has to protect. ASHRAE Standard 90.1 and the adopted energy code drive the economizer requirement and the unit efficiency floor. The National Electrical Code (NFPA 70) governs the disconnect, the conductors to the MCA, the overcurrent to the MOCP, and the grounding. The mechanical code and the fuel gas code govern the duct, the combustion air and venting, the gas piping, and the duct smoke detector, with the air-conditioning systems standard (NFPA 90A) behind the smoke detection and shutdown, installed to the fire alarm code (NFPA 72).

The project specification ties it together and can be stricter than any of them on tolerances, commissioning scope, and acceptance. Cite the body that owns the point, set every unit value to the manufacturer's instructions, and confirm the code requirements against the adopted edition and local amendments, because these documents revise on their own cycles and the adopted version controls.

Units, terms, and definitions

RTU work carries its own vocabulary and a few unit systems, so the same value reads differently across a nameplate, a spec, and a metric drawing.

Cooling capacity is in tons, where one ton is 12,000 Btu/h, or in kilowatts in metric work. Airflow is CFM, cubic feet per minute, in the field and liters per second or cubic meters per hour in metric sources. Static pressure is inches of water column, in. w.c. or in. wg, where 1 in. w.c. is about 249 pascals. Gas pressure is also in inches of water column. Temperatures are in degrees Fahrenheit on most US nameplates and Celsius in metric work. The electrical limits on the plate are the MCA and the MOCP, and the unit is a packaged unit, an RTU, or a rooftop.

RTU / packaged unit

A self-contained HVAC unit with cooling, heating, fans, and often an economizer in one cabinet on a roof curb

Roof curb

The steel frame that carries the unit, routes the duct through the roof, and seals it watertight

Condensate trap

The drain seal that lets a draw-through pan drain against the blower's negative pressure

MCA / MOCP

Minimum circuit ampacity sizes the conductors; maximum overcurrent protection caps the breaker or fuse

ESP

External static pressure, the resistance the blower works against outside the cabinet, in inches of water column

Subcooling / superheat

The charge-verification readings: subcooling for a TXV unit, superheat for a fixed-orifice unit

Temperature rise

Supply minus return air temperature on heating, held inside the nameplate range

SEER2 / IEER

AHRI efficiency ratings for seasonal and part-load cooling that proper charge and airflow protect

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is air balancing?
• What is an air-side economizer?
• What are the types of HVAC systems?
• What does commercial kitchen equipment installation involve?
• What is a fan coil unit?