Chiller plant startup and commissioning field guide

The chiller, the plant, and the gate on the first start

The chiller is the heart of the chilled-water plant. It pulls heat out of the building's chilled water and dumps that heat, plus the work its own compressor adds, into the condenser water that goes out to the cooling tower. Everything else in the plant exists to feed it: the pumps that move the water, the tower that rejects the heat, the controls that stage it, and the power that turns the motor. Commissioning the plant means commissioning all of it, but the start that gates the schedule is the chiller's.

That first start is not yours to make. On almost every water-cooled chiller the warranty requires the manufacturer's factory-authorized technician to perform the initial startup, and they will not do it until the contractor has the machine ready: the water flowing through both barrels, the power and the starter or drive proven, the controls live, and the condenser-water side available with the tower running. The factory tech checks the refrigerant charge, the oil, megs the motor, and sets the machine. The contractor supports. Start it yourself to save a day and you can void the warranty on a piece of equipment worth more than the rest of the mechanical room.

Commissioning is the layer on top of the startup. The factory startup proves the chiller will run safely. Commissioning proves it does the job: that it makes its tons at the design chilled-water and condenser-water conditions, that it holds its efficiency in kW per ton, that the safeties trip, that the staging works across multiple machines, and that the plant rides through a failure without losing the load. The two get confused on every job, and the cost of confusing them is a chiller that runs but never gets proven against the number it was bought to hit.

Which chiller is on the job?

Water-cooled chillers split first by how they compress the refrigerant. Scroll machines orbit two interleaved scrolls and run at the small end, commonly up to a few hundred tons, often in multiple-compressor packages so the plant stages by cutting compressors in and out. Screw machines mesh two helical rotors and cover the middle, commonly the few-hundred-ton range, and they are forgiving of load swings and dirty conditions. Centrifugal machines spin an impeller that flings refrigerant outward and live at the large end, commonly above 500 tons, where they are the most efficient choice and the only one of the group that can surge. The break points overlap and the selection controls, but that is the rough map.

Absorption is the outlier. An absorption chiller has no mechanical compressor at all. It uses a heat source, steam, hot water, or a gas flame, to drive a refrigerant-absorbent pair, usually lithium bromide and water, around a thermal cycle. You see it where waste heat or cheap steam is available and electricity is expensive or limited, and it trades efficiency for that flexibility, running a much lower coefficient of performance than a compression machine. Most building and data-center plants run electric compression chillers, and this guide stays on the water-cooled centrifugal and screw machines that anchor those plants.

Water-cooled versus air-cooled is the other fork. A water-cooled chiller rejects its heat to a condenser-water loop and a cooling tower, runs cooler condensing temperatures, and is more efficient, which is why large central plants and data centers use it. An air-cooled chiller rejects straight to ambient air with no tower and no condenser pumps, simpler to install and run but less efficient because it is limited by the dry-bulb. The data center and large central plant are the home turf for water-cooled, and the redundancy and free-cooling questions that come with that scale are their own subject later in this guide.

The refrigeration cycle inside the machine

A water-cooled chiller is a refrigeration machine wrapped around two water-to-refrigerant heat exchangers. In the evaporator, the chilled water gives up its heat to boiling refrigerant on the other side of the tubes. The refrigerant boils at a low temperature and pressure, and the latent heat of that boiling is what pulls the chilled water down to its leaving setpoint, commonly around 44 degrees F on a standard design though the project controls the number. The chilled water leaves cold and goes back out to the building's coils.

The compressor takes the low-pressure vapor off the evaporator and raises it to a high pressure and temperature, and that pressure lift is the work the machine has to do. The hot high-pressure vapor goes to the condenser, where it gives up its heat to the condenser water and condenses back to a liquid. The condenser water, now warmer, leaves for the cooling tower, rejects the heat to the air by evaporation, and comes back to do it again. The tower side and its water treatment are a full subject of their own, covered in the companion cooling-tower commissioning guide, because that loop is where half the chiller's performance is won or lost.

The liquid refrigerant then drops back through an expansion device, an orifice, a float, or an electronic valve depending on the machine, which throttles it down to the low evaporator pressure and meters the flow. That closes the loop and the cycle starts over. The refrigerant itself is part of the commissioning conversation, because a large water-cooled chiller holds a large charge, and a large charge in an occupied building brings the machine-room safety rules into play.

What does a chilled-water plant need before startup?

Before the factory tech ever shows up, the contractor owns a readiness list, and the factory startup is a wasted trip if the list is not done. The piping has to be pressure-tested, flushed, and clean. Construction leaves mill scale, weld slag, cutting oil, and pipe dope in every field-built loop, and that debris chokes strainers, jams control valves, and fouls the evaporator and condenser tubes the chiller depends on. The pressure proof comes first, witnessed before the pipe is insulated, which is the chilled-water hydro-test package covered in the companion guide. Then the flush, run at a velocity high enough to scour the walls and carry the debris to the strainers, until the water runs clear and the iron count drops.

With the pipe clean, the rest of the list is about proving the chiller has what it needs to run. Strainers in and screens clean. Both water loops able to deliver design flow, chilled water through the evaporator and condenser water through the condenser, with the flow proven, not assumed. The pumps wired, rotating the right direction, and started. The power landed and the starter or variable-frequency drive checked out. The controls live and talking to the building management system. The condenser-water side available, which means the tower running and able to make condenser water, because a chiller cannot start into a dead condenser loop.

The order is not negotiable, and the most common way the schedule blows up is starting the readiness work too late. You cannot prove flow before the pumps run, you cannot run the pumps before the loop is flushed, and you cannot flush before the pipe passes its pressure test. Each step gates the next, and the factory startup sits at the far end of the chain. Get the readiness items signed off as a hold point before the factory visit is booked, not the morning the tech is standing in the mechanical room.

What is a factory authorized startup?

A factory authorized startup is the first start of the chiller performed by the manufacturer's own technician or a factory-certified service agent, and on most water-cooled machines it is a condition of the warranty. The contractor cannot make the first start and keep the warranty intact. The manufacturer wants their person to confirm the machine arrived charged and tight, that nobody damaged it during rigging and installation, and that it is set up to their numbers before it ever turns over under power.

The factory tech runs a specific list. They verify or top the refrigerant charge and leak-check the machine, confirm the oil level and that the oil heater has been energized long enough, megger the motor windings to ground to prove the insulation is sound before applying power, check the starter or drive settings, and set the safeties and the control parameters to the factory configuration. Then they bring it up, watch it stabilize, and log the startup readings. The startup sheet they complete is the document that activates the warranty, and many manufacturers require it returned within a set window, commonly on the order of 30 days of the machine's delivery or start.

The contractor's role is support, and it is real work. You provide the proven water flow, the power, the clean loops, and a crew to operate the pumps and the tower while the tech works the machine. You do not start the chiller to test your controls before the tech arrives. Improper startup, a wrong refrigerant charge, or running the machine outside its rated conditions are exactly the things a manufacturer points to when they deny a warranty claim, so the factory startup is not a formality to schedule around. It is the event the whole readiness sequence is built to feed.

Refrigerant, oil, and the machine-room safety rules

A large water-cooled chiller holds a large refrigerant charge, and that charge changes the rules for the room it sits in. ANSI/ASHRAE Standard 15, the safety standard for refrigeration systems, drives the machine-room requirements: a refrigerant detector located where a leak will collect, set to alarm and start mechanical ventilation at a defined concentration, with the alarm annunciating both inside the room and outside every entrance so nobody walks into an oxygen-displaced space. The emergency ventilation is sized off the mass of the largest single charge in the room, and the exhaust needs makeup air so the fan is not starved. Larger refrigerant systems may also need a pressure-relief path piped to the outside. Confirm the exact provisions against the adopted edition of Standard 15 and the local mechanical code, because the thresholds and the calculation move between editions.

Commissioning the refrigerant monitor is a step crews skip and regret. The detector and its ventilation interlock are a life-safety system, and an alarm and a fan that were wired but never functionally tested are an alarm and a fan that will not run on the day a relief lifts. Prove the monitor reads, prove it starts the exhaust and annunciates inside and out, and log the calibration. A chiller startup that does not include a working refrigerant monitor is not a finished startup.

The oil is the other half. On a centrifugal machine, refrigerant migrates into the oil sump during shutdown, because refrigerant has an affinity for the oil when it is cold, and a sump full of refrigerant-diluted oil cannot hold a film on the bearings at start. That is what the oil sump heater is for: it keeps the oil warm so the refrigerant boils back out before the machine runs. The factory procedure typically calls for the oil to reach a minimum temperature, commonly on the order of 130 degrees F, before the compressor is allowed to start, and after a long shutdown the heater may need many hours, often 8 to 12, to drive the refrigerant out. Start a centrifugal compressor on cold, refrigerant-laden oil and you wipe the bearings in minutes. It is one of the few startup mistakes that is both catastrophic and unambiguous.

Why must flow be proven before the compressor runs?

No flow through the evaporator is the fastest way to wreck a chiller. With the compressor pulling heat out of water that is not moving, the water in the tubes drops toward freezing, and ice in the evaporator splits tubes and ends the machine. So both water loops have to prove flow before the compressor is allowed to run, and the proof is an interlock, not an assumption. The chiller controller will not let the compressor start, and trips it if it is running, until it sees flow in the evaporator and, on a water-cooled machine, in the condenser as well.

The proof has two parts that belong in series. The pump starter auxiliary contact says the pump is energized, and a flow-proving device, a paddle flow switch or a differential-pressure switch across the barrel, says water is actually moving. A running pump is not the same as flow. A closed valve, an air-bound loop, or a backward-rotating pump gives you a happy starter contact and no water, which is exactly the case the flow switch is there to catch. The differential-pressure switch is set to the pressure drop that corresponds to minimum flow off the chiller's pressure-drop curve, and below that the contact opens and the compressor stays down.

Commission the interlock by proving it the hard way: kill the flow and confirm the chiller faults and will not start, not just that it runs when flow is present. The number that matters is the manufacturer's minimum evaporator flow, which the variable-flow plants have to respect with a bypass when the building throttles down. Prove the flow switch protects the minimum, and you have protected the most expensive part of the plant from the most common way it dies.

What is chiller approach?

Approach is the gap between the water temperature and the refrigerant saturation temperature in a barrel, and it is the field measure of how well that heat exchanger is working. Evaporator approach is the leaving chilled-water temperature minus the saturated refrigerant temperature in the evaporator. Condenser approach is the saturated condensing temperature minus the leaving condenser-water temperature. A small approach means the tubes are transferring heat well. A growing approach means something is in the way, and on the condenser side that something is almost always fouling or scale on the water-side of the tubes.

Condenser approach is the one to watch over time, because it is the water treatment talking through the chiller. A clean condenser commonly runs an approach of just a few degrees F; when it creeps up at steady load and steady conditions, the tubes are scaling or fouling, the machine is working harder to reject the same heat, and the kW per ton climbs with it. Trend it from day one. The increase is gradual and easy to miss until it has cost real energy, which is why the commissioning baseline matters: you cannot tell the tubes are fouling if you never recorded what clean looked like.

Delta-T is the load side of the same story. The chilled-water delta-T, the return minus the supply, is set by the building load and the flow, commonly designed in the range of 10 to 16 degrees F, and the condenser-water delta-T runs similar on a standard selection. Confirming the chiller makes its capacity means confirming it holds the design delta-T at the design flow and the design entering conditions, because a machine running a low delta-T at high flow is moving water without moving its rated tons. That low delta-T problem is a plant-wide disease, and it usually starts at the coils and valves, not the chiller.

Approach

Gap between barrel water temperature and refrigerant saturation temperature; small is good, growing means fouling

Delta-T

Return minus supply chilled-water temperature, set by load and flow, commonly 10 to 16 degrees F by design

Lift

Difference between condensing and evaporating pressure or temperature; the work the compressor has to do

What is chiller surge?

Surge is a flow reversal in a centrifugal compressor, and it only happens on centrifugal machines. The impeller develops pressure by spinning refrigerant outward, but at light load and high lift it runs out of the flow it needs to hold that pressure, and the high condenser pressure shoves the refrigerant backward through the wheel. Then the wheel rebuilds pressure, pushes forward, runs short again, and reverses. That oscillation, several times a second, is surge, and it hammers the bearings and the impeller. Run a machine in surge and you are counting down to a failure.

You hear it and you see it. Surge sounds like the machine is breathing hard, a rhythmic whoosh or rumble, often with the sound of refrigerant moving the wrong way, and you watch it on the gauges as the discharge pressure and the motor current swing in time with the noise. Lift drives it: lift is the difference between the condensing and evaporating pressures, and surge shows up when lift goes high while load goes low. The classic setup is a cold morning with the tower making very cold condenser water and the building barely calling for cooling. High lift, low flow, surge.

The controls fight it two ways. Hot-gas bypass recirculates a stream of hot discharge gas back to the evaporator to keep a minimum flow through the impeller when the real load is too small to provide it, trading efficiency for stability at the low end. And head-pressure control keeps the lift from going higher than it needs to, by holding the condenser water from going too cold through condenser-water reset at the tower. Commissioning a centrifugal means proving it does not surge across its real operating envelope, especially at the low-load, cold-condenser corner where it is most exposed.

Low-load operation and head-pressure control

A chiller spends almost none of its life at full load. The building load swings with weather and occupancy, and the plant runs at part load the large majority of the hours, which is why part-load efficiency, not full-load, is what decides the energy bill. The metric is the integrated part-load value, IPLV, or the non-standard part-load value, NPLV, both defined by AHRI Standard 550/590 as a single weighted number across four load points. The standard weights the points heavily toward part load, roughly 1 percent at full load, 42 percent at 75 percent load, 45 percent at 50 percent load, and 12 percent at 25 percent load, because that is where the machine actually lives. IPLV uses the standard rating conditions; NPLV is the same idea calculated at the project's own conditions.

Head-pressure control is the lever that makes part load efficient. The lower the lift the compressor has to overcome, the less work it does for the same tons, so the plant gives the chiller the coldest condenser water it can safely use by resetting the tower's leaving-water setpoint down as conditions allow. The catch is the floor: every chiller has a minimum lift or minimum entering-condenser-water temperature it needs to keep oil moving and, on a centrifugal, to stay out of surge. Reset the condenser water below that floor and you trade an efficiency gain for a surge or a lubrication problem. The manufacturer sets the minimum, and the reset strategy has to respect it.

Variable-speed-drive chillers change the math at the low end. A fixed-speed centrifugal unloads with inlet vanes and loses efficiency as it throttles, and it surges more easily at high lift. A VSD machine slows the impeller as the load drops, which is far more efficient at part load and at low lift, and it pushes the surge line back. The VSD pays off most where lift is low for many hours, the same cool-weather, cold-condenser conditions that punish a fixed-speed machine, which is why the VSD chiller has taken over the large water-cooled plant.

Primary-secondary or variable primary flow?

The plant moves chilled water in one of two arrangements, and commissioning them is different work. Primary-secondary splits the flow into two loops joined by a decoupler, a short common pipe. The primary loop runs a constant-flow pump dedicated to each chiller, so every running chiller always sees its design flow. The secondary loop runs variable-speed pumps that vary flow out to the building as the load changes. The decoupler absorbs the difference between what the chillers pump and what the building draws, and the direction and magnitude of flow in that decoupler is exactly what you watch to stage chillers and to confirm the loops are not fighting.

Variable primary flow does away with the second set of pumps. One set of variable-speed pumps varies flow through the chillers and out to the building together, which saves pump energy and first cost. The price is control complexity and a hard limit you cannot cross: the chiller's minimum evaporator flow. When the building throttles down past the minimum the running chillers need, a bypass valve has to open and recirculate enough water to keep each machine above its floor, or you trip the flow interlock and lose the chiller. That bypass commonly holds flow at roughly 40 to 50 percent of a chiller's design flow, set by the manufacturer's minimum.

Variable primary saves energy and primary-secondary is simpler to run and to troubleshoot, and the spec made the call long before you got there. What commissioning owes either one is the staging proof. Add and shed chillers across the load range and confirm the plant brings the next machine on before the running ones run out of capacity, sheds one without dumping the load, holds minimum flow through every transition, and rotates the lead and lag so the hours even out. Staging that was never tested across a real load swing is where a plant quietly fails to hold the building on the first hot afternoon.

The pumps and the flow balance

The chiller only makes its rated tons at its rated flow, so the pumps and the balance are not a separate trade you can wave through. Both loops have to deliver design flow to the barrels: the chilled-water flow through the evaporator and the condenser-water flow through the condenser, each within the chiller's allowable range. Too little flow and the chiller cannot move its heat and the flow interlock may trip it. Too much flow and you risk tube erosion and you waste pump energy for tons you were already making.

Balancing the plant is a hydronic balancing job, and it is where the test-and-balance contractor earns the line item. The condenser-water flow is commonly selected near a few gallons per minute per ton on a standard design, though the chiller and tower selection set the real figure, and the chilled-water flow follows the design delta-T. Prove the flows with a calibrated method, not the pump curve and a prayer, and confirm the delta-T across each barrel lands at design when the machine is loaded. A balance that hits flow but misses delta-T, or hits delta-T at the wrong flow, has not actually proven the chiller can make its capacity.

The pumps themselves get the basics confirmed at startup: rotation, that they are not running backward, that the variable-speed drives ramp and hold a differential-pressure setpoint, and that the strainers ahead of them are clean after the flush. A backward condenser pump and a blocked suction strainer both starve the chiller the same way the flow switch is meant to catch, and both are findings you want before the factory tech is on the clock.

Controls, the sequence of operation, and the safeties

The controls are where the plant becomes a plant instead of a pile of equipment, and each loop gets proven, not assumed. The chiller holds its leaving chilled-water temperature to setpoint, commonly with a reset that raises the setpoint when the building load is light to save compressor work, and the plant stages chillers, pumps, and the tower together. The sequence of operation in the spec is the script: it says what starts what, in what order, with what delays, and what resets against what. Commissioning works the sequence line by line and proves the building management system actually does what the sequence says, including the start and stop interlocks that bring the condenser pump and tower up before the chiller and keep them running after it stops.

The safeties are the part you prove by making them trip. A chiller carries a stack of protective cutouts: low chilled-water temperature and a freeze protection to keep the evaporator from icing, low oil pressure or oil temperature, high condenser pressure, low evaporator pressure, motor overload and the winding or bearing temperature protection, and the flow interlocks on both loops. Each one exists because the failure it catches is expensive, and an untested safety is a safety you are trusting on faith. Functionally test the ones you safely can, and confirm the rest are configured to the factory settings and annunciate where an operator will see them.

Then prove the alarms reach a person. An alarm that lights a board nobody watches is not protection. The alarms that matter most over the plant's life are the ones that say it is about to hurt itself, the low chilled-water temperature, the loss of flow, the oil and bearing faults, and these should annunciate at the operator's station and, on a critical plant, page out. Test them. The functional test is not done when the chiller runs. It is done when you have proven the plant protects itself and tells someone when it cannot.

How is chiller capacity and efficiency proven?

You prove capacity and efficiency by measuring them, not by reading the nameplate. Capacity is the tons the chiller actually moves, computed from the chilled-water flow and the delta-T across the evaporator. Efficiency is the power it spends to do it, the compressor kilowatts divided by the tons, the kW per ton that decides the energy bill. A standard water-cooled centrifugal commonly lands somewhere in the range of 0.5 to 0.6 kW per ton at full-load design conditions, with VSD machines lower and part-load IPLV lower still, but the submittal and AHRI rating for the specific machine are the numbers you test against, not a rule of thumb.

The reference is AHRI Standard 550/590, which sets how a water-chilling package is rated and the tolerances a test is read against. The standard pins the test conditions and the allowable bands: chilled-water and condenser-water flows within about 5 percent, the leaving evaporator and entering condenser water temperatures within about half a degree F of target, voltage within about 10 percent of nameplate, and tolerances on capacity and efficiency that depend on the load point. A field test that drifts outside those bands is not a valid test, the same way a cooling-tower thermal test run far off the design wet-bulb is not valid. Hold the conditions, or correct to them, before you read the result.

Decide early what level of proof the spec actually buys. A witnessed factory performance test, where the manufacturer runs the specific machine on a calibrated test stand to 550/590 before it ships, is the strongest proof and the most expensive. A witnessed field test at the plant proves the installed machine in its real loops but is harder to hold at design conditions. A functional check at whatever conditions the commissioning window offers is the weakest. Those are three different commitments, and the contract decides which one you owe. Whichever it is, trend the readings over the test so a flat, stable result backs the single number you log, and tie the acceptance test into the broader commissioning process the way the companion guides describe.

kW/ton

Compressor power per ton of cooling; the lower the number, the more efficient the machine

IPLV / NPLV

AHRI 550/590 weighted part-load efficiency, at standard (IPLV) or project (NPLV) conditions

Ton

12,000 BTU per hour of cooling, the unit of chiller capacity

Water treatment on both loops

The chiller rides on two water loops and both need treatment, for different reasons. The condenser-water loop is open to the air at the tower, so it scales, corrodes, fouls, and grows Legionella, and it is the loop where the chiller's condenser tubes pick up the deposit that drives the condenser approach up. That loop's treatment, its cycles of concentration, its scale and corrosion inhibitors, its biocide program, and the ASHRAE 188 water management plan, is a full subject covered in the companion cooling-tower commissioning guide, and it is not optional to the chiller. The condenser approach you trend is the chiller reporting on how well that program is working.

The chilled-water loop is a closed loop, and a closed loop is a different animal. It does not evaporate, so it does not concentrate solids or grow biology the way the open tower does, but it still needs a corrosion inhibitor and often a closed-loop biocide, because oxygen ingress and the mix of metals in the loop will corrode it slowly if the chemistry is left alone. A neglected closed loop fills with corrosion products and biofilm that foul the evaporator tubes from the inside and plug the small passages in the building's coils and valves. The treatment dose is small compared to the tower, but a closed loop running with no inhibitor is a loop quietly eating itself.

Commissioning ties the two together through the approach. When the condenser approach climbs, the condenser-water treatment is failing. When the evaporator approach climbs without a flow or charge problem, look at the closed-loop chemistry and the cleanliness of the chilled-water side. The chiller is the instrument that reads both loops, and the day-one baseline is what lets you read the trend.

The chiller plant in a data center

In a data center the chiller plant runs to a harder standard, because the load never goes away and a lost plant is a lost hall. Redundancy is the first difference. The plant is built with spare capacity, commonly N+1 or better, so a chiller or a pump can fail or come down for service and the remaining machines still hold the full load. Commissioning a redundant plant means proving the redundancy, not just the running machines: pull a chiller offline at load and confirm the standby stages in and the hall stays inside its temperature envelope through the swap.

Efficiency is the second difference, because the plant runs every hour of the year and the energy is enormous. That puts a premium on part-load efficiency and on free cooling. A waterside economizer puts a heat exchanger between the condenser-water loop and the chilled-water loop so that when the outdoor wet-bulb is low enough, the cooling tower makes cold enough condenser water to cool the building directly and the chiller compressors can shut off or unload hard. In a cold climate that recovers many hours a year of nearly free cooling, and commissioning it means proving the changeover into and out of economizer mode happens cleanly without dropping the load or short-cycling the machines.

How the rejected heat finally leaves the building, and how the airflow on the hall side is managed, is its own subject in the cooling pillar. The chiller plant is one stage of that chain. What it owes the data center is the same as what it owes any building, proven harder: capacity, efficiency, staging, and the ability to ride through a failure, with no acceptable window where the load goes uncovered.

The integrated test and the failure scenarios

A chiller that starts and makes its tons in isolation is not a commissioned plant. The integrated systems test is where the plant has to prove it behaves as a system when something goes wrong, and it is the test that catches the problems no single-piece functional test ever will. You run the plant at load and then break things on purpose: drop a chiller, drop a pump, fail a tower cell, and on a critical facility, drop the utility and force the plant onto generator power. The question the test answers is whether the plant rides through, recovers, and holds the load, or whether one failure cascades into a lost building.

The failure that hurts on a real facility is the power event, because the cooling plant and the electrical plant have to ride through it together. When the utility drops and the generators pick up, the chillers shut down on the power loss and then have to restart in sequence as power stabilizes, and the building's thermal mass is the only thing holding the load during the gap. On a data center that gap is measured against how fast the hall heats up with the plant down, which is why the cooling integrated test runs alongside the electrical integrated test, and why the two pillars get commissioned as one event. A plant that cannot restart cleanly after a power blip is not commissioned, no matter how good its kW per ton looked on a calm afternoon.

Script the scenarios from the design's stated failure modes, witness them at load, and document what the plant actually did against what the sequence said it would do. The integrated test is the last and hardest gate before turnover, and it is the one that turns a collection of proven boxes into a plant the owner can trust.

What to document

The commissioning record is the baseline the plant gets run and judged against for its whole life, and the day-one numbers do not exist anywhere else once the factory tech drives away. Capture what the machine is, how it was started, what it proved, and the conditions it proved it at, so the next engineer can tell a year from now whether a creeping approach or a climbing kW per ton is a real problem or just where the machine started.

Record the chiller make, model, refrigerant, and charge, and that the factory startup was performed and the warranty sheet returned. Record the refrigerant leak check and that the machine-room monitor and ventilation were proven. Record the evaporator and condenser flow and delta-T at test, the evaporator and condenser approach, the measured capacity in tons and efficiency in kW per ton against the AHRI rating, and the conditions they were measured at. Record that the surge behavior and the safeties were tested, and that the staging and integrated failure scenarios were witnessed. The plant the owner gets is only as good as the record that proves where it started.

Common mistakes

• Starting the chiller before the loops are flushed and the flow is proven, freezing or fouling the machine on its first run.
• Making the first start without the factory authorized technician, voiding the warranty on the most expensive equipment in the plant.
• Energizing the compressor before the oil heater has driven refrigerant out of the sump, wiping the bearings in minutes.
• Wiring the refrigerant monitor and emergency ventilation but never functionally testing them, leaving the machine room's life-safety unproven.
• Trusting a pump starter contact as proof of flow instead of a flow switch, so a closed valve or backward pump starves the barrel undetected.
• Letting condenser-water reset drive the water below the chiller's minimum lift, pushing a centrifugal into surge to chase efficiency.
• Ignoring surge noise and current swings as a quirk instead of the bearing-and-impeller damage it is.
• Accepting a kW/ton result taken outside the AHRI 550/590 condition bands and calling the machine proven.
• Skipping the staging and integrated failure test, so the plant first fails to hold the load on a real hot afternoon or power event.
• Never recording the clean baseline approach, so there is no way to tell later that the tubes have fouled.

Field checklist

Standards and references

The chiller manufacturer governs the startup and the warranty, and that is the first authority to name on any chiller job. The factory startup requirement, the oil and refrigerant procedures, the minimum flows and lifts, and the safety setpoints come from the manufacturer's installation, operation, and maintenance documents, and they override a general rule of thumb every time. The contract decides the level of performance proof, but the manufacturer decides how the machine is started and kept under warranty.

ANSI/ASHRAE Standard 15, Safety Standard for Refrigeration Systems, governs the machine-room safety case: the refrigerant detector, the alarm and emergency ventilation, the makeup air, and the relief path, sized off the refrigerant charge. ASHRAE Standard 90.1 sets the minimum chiller efficiency, in full-load kW per ton and in IPLV, that the selection has to meet. AHRI Standard 550/590 is the performance rating standard for water-chilling packages, defining the rating conditions, the IPLV and NPLV calculation, and the test tolerances a capacity and efficiency test is read against. Confirm the adopted edition of each, because the thresholds and the calculations move between cycles.

ASHRAE Guideline 0 frames the commissioning process itself, the owner's project requirements, the functional and integrated testing, and the documentation, and the project's commissioning specification builds on it. The test-and-balance work on the water loops follows NEBB or AABC procedures for hydronic balancing. Pre-commission cleaning and flushing of the closed loops follows the project specification and references such as BSRIA BG 29 in some markets. Name the standard that actually governs the point, let the project specification and the manufacturer's documents override a general figure, and confirm the edition before you cite a number on a submittal.

Units, terms, and acronyms

Chiller work mixes refrigeration terms, hydronic terms, and the rating-standard vocabulary, and the same quantity shows up in different units across a submittal, a balancing report, and a commissioning script. The terms below travel across the whole startup and commissioning package.

Ton of refrigeration

12,000 BTU per hour of cooling; the unit a chiller's capacity is rated in

kW/ton

Compressor kilowatts per ton of cooling; the efficiency measure, lower is better

Approach

Temperature gap between barrel water and refrigerant saturation; a rising condenser approach signals fouling

Delta-T

Return minus supply chilled-water temperature, commonly 10 to 16 degrees F by design

Lift

Difference between condensing and evaporating pressure or temperature, the compressor's workload

Surge

Flow reversal in a centrifugal compressor at low load and high lift; damaging and centrifugal-only

IPLV / NPLV

AHRI 550/590 weighted part-load efficiency at standard or project conditions

ASHRAE 15

Refrigeration safety standard driving the machine-room detector, ventilation, and relief

Factory startup

The manufacturer-authorized first start required to keep the chiller warranty intact

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What are range and approach on a cooling tower?
• What pressure do you test chilled-water piping to?
• What is chiller plant sequencing?
• What is boiler commissioning?
• What is a VRF system?