You can usually tell within a minute of walking into a plant room whether the people who run the building understand their fire pump. The ones who do treat it as the most important asset on the register. The ones who do not treat it as just another pump and store paint tins beside it. This guide explains what a fire pump is, why it exists, the types you will meet, the parts of the installation, and the logic that decides when the machine runs.
The message up front: a fire pump is a standby machine. It must start reliably after long periods of doing nothing, then run under load for a long time in the worst conditions the building will ever see. Almost everything that makes fire pumps strange compared with ordinary pumps, the listed assemblies, the duplicated power sources, the stripped-down protection in the controller, the heavy testing regime, follows from that one requirement.
What this article is, and is not
This is a general explanation of how fire pump installations work, not design, acceptance, testing or compliance guidance. It publishes no pressures, flows, sizes, capacities, temperatures, durations or frequencies, because every one of those is set by the standard adopted in your jurisdiction and by the specific listed equipment in front of you. Design, alteration, testing and maintenance must follow the standards adopted by your authority having jurisdiction and be carried out by competent people. Read your own adopted edition and the manufacturer's data for your own set.
1. Why a fire pump exists at all
Start here, because everything else follows from it, and most articles on this subject go straight to a list of pump types.
A sprinkler system, a standpipe and hose system or a private hydrant main does not simply need water. It needs water at a pressure and flow that will operate the number of devices the design assumes, at the most hydraulically remote point of the building, for as long as the design assumes. That is a hydraulic demand, calculated by the designer of the water-based system.
The incoming supply has a capability of its own, and frequently it does not meet that demand. A town main may be weak, or adequate at flow but not at pressure, or fine at street level and hopeless high up a tower where static height has eaten the available pressure. A tank at grade feeding a high-rise riser has no useful pressure at all at the top of the building. The fire pump makes up the difference: it takes what the supply can give and lifts it to what the system demands. It is not a water source, it is a booster between a source and a system, which is why suction conditions matter so much later on. A pump can only boost what reaches it.
2. The consequence: a standby machine that must work first time
A chilled water pump runs most of the year, so if it degrades you notice: the building gets warm and somebody complains. Degradation is self-reporting.
A fire pump is the opposite. It may sit still for its entire service life and never be called on in anger. It has no operational feedback loop, nobody complains when it quietly becomes incapable, and on the day it is needed there is no second attempt. On that day it must start on a signal, come up to duty, and run under load for a long time while the building is on fire, the supply may be compromised, the plant room may be hot, and nobody is beside it adjusting anything.
The organising insight
Everything distinctive about fire pumps is a response to standby duty. Listed assemblies exist because you cannot validate reliability by observing a machine that never runs. Duplicated power sources exist because the event that calls the pump is also the event most likely to take out its supply. A controller that will not trip the pump exists because a stopped pump is worse than a damaged pump. A heavy inspection and testing regime exists because standby equipment gives you no other way of knowing it still works. None of it is bureaucratic. It is the same answer to the same problem.
3. Listing and approval: why you cannot just specify a good pump
A fire pump, its driver and especially its controller are not general-purpose equipment chosen on the designer's judgement and a datasheet. They are listed or approved assemblies: evaluated as a product, by a recognised body, against a published test regime, and appearing in that body's listing for fire pump service. What matters on site is the consequences.
- The assembly is the product, not the parts. A listing usually covers a pump, driver and controller arrangement as evaluated. Substituting an apparently equivalent component, or modifying the assembly in the field, can put the installation outside its listing even where the substitute is sound engineering.
- Non-listed equivalents are not equivalents. A pump that would suit a process duty is not automatically acceptable here, because what is certified is suitability for standby fire duty, not general hydraulic competence.
- Which listings are acceptable is jurisdictional. The authority having jurisdiction decides which listing and approval marks it accepts, and in some markets local type approval is expected on top. Ask before you procure.
- Documentation is part of the asset. Listing certificates, manufacturer's data and the accepted pump curve are the evidence that the installed machine is the one approved, and the first thing an inspector asks for.
I am deliberately not naming any certification body's requirements or values. The principle transfers between jurisdictions; the specifics do not.
4. The pump types and what each one suits
Fire pumps are overwhelmingly centrifugal machines, with one special case at the end. The differences are mostly geometric: how the casing splits, whether the shaft is horizontal or vertical, and where the water sits relative to the pump.
- Horizontal split case. The traditional workhorse. The casing splits along the shaft axis, so the top half lifts off and the rotating element can be inspected or replaced without disturbing the pipework. It needs a flooded suction, meaning water arrives under positive pressure. Access is excellent, which matters on an asset you must keep serviceable. Its cost is footprint.
- Vertical split case. The same idea with a vertical shaft and the motor above. The answer where floor area is constrained and height is available, at the price of more awkward access.
- End suction. Suction into the end of the casing, discharge from the side, simpler single-casing construction. Common on smaller duties. Servicing generally means breaking into the suction side.
- Vertical in-line. Suction and discharge in line and a vertical motor above, so the unit drops into the pipe run with a very small footprint. Compact, with the same access trade-off.
- Vertical turbine. The important one. The pump bowls sit submerged, at the bottom of a wet pit, tank or open source, with a long shaft up to a driver on the floor above. Because the impellers are already under water, the pump is permanently primed and does not have to lift water into itself before it can generate pressure. That is why it is effectively the option wherever the water level is below the pump and suction lift is involved: a surface centrifugal pump there depends on priming and on holding that prime, and a standby machine that must start unattended is the worst possible place to depend on either. The trade-offs are real. Submerged bowls and shafting are hard to inspect, the assembly must be pulled to be examined, and the pit becomes part of the asset you have to keep clean and full.
- Multistage and multi-outlet arrangements. Where high pressure is required, as in a tall building, stages are added in series so each impeller contributes part of the total. Multi-outlet variants take off at more than one point in the stage sequence, so different zones can be fed at different pressures from one machine rather than from cascaded sets. That can simplify a high-rise riser strategy, at the cost of a more complex machine.
- Positive displacement pumps. A genuine special case. Where the duty is injecting foam concentrate at a controlled proportion, or generating the conditions a water mist system needs, a positive displacement machine is used because the duty is about delivering a defined volume per revolution rather than developing pressure on a centrifugal curve. These are not a substitute for a centrifugal pump on standard sprinkler or hydrant service.
| Type | Arrangement | Suits | Main trade-off |
|---|---|---|---|
| Horizontal split case | Horizontal shaft, casing splits along the axis | Flooded suction, larger duties, plant rooms with floor space | Largest footprint of the common types |
| Vertical split case | Vertical shaft, driver above, split casing | Flooded suction where floor area is limited | Access to the rotating element is harder |
| End suction | Axial suction, radial discharge, single casing | Smaller duties, simpler installations | Servicing usually disturbs the suction pipework |
| Vertical in-line | Suction and discharge in line, motor above | Tight plant rooms, straightforward pipe runs | Compactness bought with maintenance access |
| Vertical turbine | Bowls submerged in the source, shaft up to the driver | Wet pit, tank or open source below the pump; suction lift | Submerged parts must be pulled to be inspected |
| Multistage / multi-outlet | Impellers in series, sometimes with intermediate take-offs | High pressure duties, tall buildings, zoned risers | More complex machine and more complex failure modes |
| Positive displacement | Fixed volume per revolution | Foam concentrate injection, water mist duties | Not a substitute for centrifugal sprinkler or hydrant service |
5. The drivers: electric, diesel and the legacy steam turbine
The pump is only half the machine. What turns it is the decision that shapes the reliability of the whole installation.
The electric motor is the simplest and cleanest option: no fuel to age, no cooling circuit of its own, no battery, no exhaust, a modest maintenance burden. Its single weakness is total. It is only as available as the supply feeding it, and the event that calls the pump is precisely the event most likely to compromise that supply. This is why the arrangements around the supply matter more than the motor: how the feed is routed, protected and separated from the building's other distribution, and whether there is an alternate source with automatic transfer, decide whether an electric set is reliable or a single point of failure with a listing certificate.
The diesel engine is independent of the mains, and that independence is the whole argument for it. Where the supply cannot be relied upon, or the authority having jurisdiction is not satisfied with the supply arrangement, a diesel set is the answer. The cost is a much larger maintenance and reliability burden: a fuel system whose contents age, absorb water and grow contamination while sitting still; batteries and charging, among the most common causes of a failure to start; a cooling system, often heat exchanger cooled from the pump discharge; an exhaust route, combustion air and ventilation; and a starting system that must work unattended, first time. That is not a reason to avoid a diesel set, it is a reason to budget its maintenance before choosing one.
The steam turbine drive is a legacy and industrial rarity, surviving where a site already has a reliable steam header for process reasons. It is not realistic in ordinary commercial or residential buildings.
Where reliability demands it, the common answer is both: a duplicated set, one electric driven and one diesel driven, arranged so the failure mode of each is covered by the other. The electric set is the day-to-day machine; the diesel set does not care what has happened to the switchboard. That is the standby logic of section two applied directly. You cannot test your way to certainty about an unattended machine, so you build in a second, independent path.
| Consideration | Electric motor driver | Diesel engine driver |
|---|---|---|
| Reliability character | Very high in itself; inherits the reliability of its supply | Self-contained, but depends on disciplined upkeep |
| Key dependency | The electrical supply and its transfer arrangements | Fuel condition, batteries, cooling and starting systems |
| Behaviour in a building fire | At risk if the fire or its response affects distribution | Independent of mains distribution |
| Maintenance burden | Comparatively light | Substantially heavier and more skill dependent |
| Typical failure to start causes | Loss of supply, transfer failure, control circuit faults | Flat or sulphated batteries, degraded or contaminated fuel, air in fuel, cooling faults |
| Plant room implications | Ventilation and access; no combustion services | Combustion air, exhaust route, fuel storage, heat rejection, noise |
| Where it fits | Where the supply arrangement is genuinely dependable | Where the supply cannot be relied on, or as the second of a pair |
6. The jockey pump, and what it is not
A jockey pump, also called a pressure maintenance pump, holds the system at its standing pressure. Every water-filled fire main leaks a little, through gland packing, fittings, and valves that pass slightly. If nothing made that up, pressure would drift down and eventually the main fire pump would start, for no fire, with all the wear and false confidence a spurious start brings. The jockey pump makes up small losses so the main pump is left alone until there is a real demand.
A jockey pump is not a fire pump. It contributes nothing to firefighting flow, it is not sized to supply any part of the system demand, and it plays no part in the hydraulic performance of the installation. It is a housekeeping device. Calling it the small fire pump, or counting it when describing the installation's capability, is a real error, and in the asset registers I work with in maintenance systems it has been written in that way more than once.
The most ignored signal in the plant room
A jockey pump that starts often, or runs more than it used to, is telling you something. Typically the system is losing water: a leak on the main, a passing valve, a weeping test connection, a drain left slightly open. It is free, continuous condition monitoring on the integrity of the whole water-filled system, and it is routinely dismissed as normal noise because the jockey pump is small. If your operators can tell you whether its behaviour has changed, you are ahead of most sites.
7. The components of the installation
A fire pump is an installation, not a machine. Each part exists for a reason, and knowing the reasons turns a plant room walk from sightseeing into inspection.
The suction arrangement comes first, because suction conditions govern whether the pump can perform at all. Everything upstream, the source, the tank or pit, the strainer, the pipework and its valves, determines what reaches the impeller, and a pump starved on suction cannot make up for it on discharge however well it was selected. When an installation fails to reach its accepted performance, the suction side is where an experienced engineer looks first.
The discharge assembly carries the devices that make the installation controllable and testable: check valves against reverse flow, isolation valves whose position is among the most important things anyone can verify because a closed one silently disables the installation, a relief valve against pressures the system was not designed to contain, and the circulation and casing reliefs that stop the pump overheating when it runs against a closed system. The test header and flow measuring arrangement let the pump run against a measurable flow without discharging into the protected system, the only way to prove hydraulic performance rather than merely prove it starts. The pressure sensing line to the controller is the nerve that makes automatic starting possible. The table below lists each item with what it is for.
Finally, the pump room is a component, not a container: a protected space, accessible during an incident, drained so water from testing and relief devices goes somewhere sensible, ventilated, and kept within the conditions the equipment requires. Those conditions are set by the adopted standard and the equipment, so read yours. The principle is that a pump room is part of the fire protection system and must be maintained as such.
| Component | What it is for |
|---|---|
| Water source and suction arrangement | Delivers water to the pump; its condition governs whether the pump can perform at all |
| Suction strainer | Keeps debris out of the pump; a common and easily missed cause of lost performance |
| Fire pump | Raises the supply to the pressure and flow the water-based system demands |
| Driver (electric, diesel or steam turbine) | Turns the pump; the choice sets the installation's dependency profile |
| Coupling and, for vertical turbines, the shaft assembly | Transmits torque from driver to pump |
| Jockey / pressure maintenance pump | Holds standing pressure against small losses; contributes no firefighting flow |
| Check valve | Prevents reverse flow between pumps and supplies feeding a common main |
| Isolation valves | Allow the pump to be taken out of service; a closed one silently disables the installation |
| Relief valve | Protects the system from pressure beyond what it is designed to contain |
| Circulation relief | Passes a small flow when the pump runs against a closed system, preventing overheating |
| Casing relief | Equivalent protection on heat exchanger cooled diesel arrangements |
| Test header and flow measuring device | Allow a measured flow test that proves hydraulic performance, not just starting |
| Gauges, pressure sensing lines and sensors | Instrumentation, and the signal path that lets the controller start the pump |
| Fire pump controller | Starts, monitors and signals the pump; prioritises starting over self-protection |
| Power supply, alternate source and transfer switch | Keep an electric driven pump available when normal supply is lost |
| Diesel fuel system | Stores and supplies fuel; ageing and contamination are its defining risks |
| Batteries and charger | Start the diesel engine; among the most common causes of a failure to start |
| Cooling, exhaust and combustion air systems | Let the diesel engine run continuously and safely inside a building |
| Pump room | A protected, accessible, drained and environmentally controlled space, itself part of the system |
8. The controller and the starting logic
How an automatic start happens. The system is held at a standing pressure and the controller watches it through a sensing line. When a sprinkler operates, or a hose valve or hydrant is opened, water leaves the system and pressure falls. The jockey pump is arranged to respond first, to a small fall, so ordinary leakage never reaches the main pump. If the demand is real, the loss outruns the jockey pump and pressure keeps falling until it reaches the point at which the controller starts the fire pump. The settings are fixed by the adopted standard and the specific installation, and publishing values would be worse than useless because yours will differ. The sequence, though, is universal, and it explains almost every fire pump behaviour you will observe.
Automatic and manual starting. An automatically started pump responds to the pressure signal without human intervention, and may respond to other signals depending on the design. A manual start facility exists as well, and some installations are arranged for manual starting only. Which arrangement is required is a question for the adopted standard and the authority having jurisdiction, not the designer's preference.
The controller is designed to start the pump, not to protect it
An ordinary motor starter is full of protective devices whose purpose is to stop the motor before it damages itself. A fire pump controller deliberately does not work that way. Most of the protection you would expect is absent, disabled, or arranged to alarm rather than trip, because the judgement built into the whole discipline is that a fire pump which stops during a fire is worse than one which destroys itself while still delivering water. The pump is expected to run to destruction rather than stop. Once you understand that, the controller stops looking like a badly specified starter and starts looking like the most single-minded piece of equipment in the building.
Sequential starting. Where more than one pump serves an installation, starts are staggered so the electrical and hydraulic transients of one machine coming up do not disturb the others or the supply. The staging is part of the design and part of what acceptance testing proves.
Alarm and status signalling. Pump running, pump failed to start, loss of a power phase, controller in a position other than automatic, and the various diesel specific troubles are signalled to the fire alarm system and on to a constantly attended or supervising location; that framework is covered in the fire alarm system types and components guide. The practical point is blunt: a controller left in manual, or an alarm output disabled because it was nuisance-signalling, converts an automatic installation into one that depends on somebody noticing.
Transfer between power sources. Where an electric driven pump has an alternate supply, a transfer arrangement moves it over automatically on loss of the normal source. That mechanism has its own failure modes, and proving it is part of proving the installation.
9. How the installation is proved, at concept level
This section is deliberately short, because inspection, testing and maintenance is a large subject with its own guide: see fire pump inspection and preventive maintenance for the regime, what is checked, the routines and the failure patterns. Only the concept belongs here, because it completes the explanation of how the installation works.
There are two different kinds of proof. A no-flow run starts the pump, lets it run, and confirms that the starting circuit, the driver and the controller do what they should: it proves the machine will start. A flow test passes measured water through the test header or an equivalent arrangement and compares what the pump delivers against its accepted performance curve: it proves the machine will perform. Confusing the two is the most consequential misunderstanding in fire pump operation.
A pump that starts is not a pump that performs
A routine start proves the starting circuit and nothing whatever about hydraulic capability. A pump with a worn impeller, eroded casing clearances, a partially closed suction valve or a blocked strainer will start beautifully, run smoothly, sound entirely normal, and still fail to deliver what the system was designed around. That is why the periodic run and the flow test are not alternatives and neither substitutes for the other. Frequencies, durations and acceptance points are all set by the standard adopted in your jurisdiction and by your listed equipment, so read yours. What is universal is that starting and performing are separate questions needing separate evidence.
10. The standards and codes, and whose law they are
Three different documents cover three different things, and they are routinely swapped:
- Installation of the pump: NFPA 20, "Standard for the Installation of Stationary Pumps for Fire Protection", current edition 2025. This governs how a stationary fire pump installation is put in.
- Design and installation of the sprinkler system the pump serves: NFPA 13, "Standard for the Installation of Sprinkler Systems", current edition 2025. The sprinkler design demand is what the pump is selected against, so the two documents meet at the hydraulic calculation. For the systems themselves, see the fire sprinkler system types and components guide.
- Inspection, testing and maintenance: NFPA 25, "Standard for the Inspection, Testing, and Maintenance of Water-Based Fire Protection Systems", current edition 2026. Fire pumps are within its scope, alongside sprinklers, standpipe and hose, foam-water and fixed water spray systems, private fire service mains and water storage tanks. If a conversation is about intervals, routines, records or inspections, this is the document, not the installation standard.
- Alarm and signalling: NFPA 72, "National Fire Alarm and Signaling Code", current edition 2025, is the framework behind how pump status and trouble signals are transmitted and supervised.
Now the part that matters legally. NFPA documents are private standards published by a United States body and are not law anywhere by themselves. They acquire legal force only where an authority having jurisdiction adopts them, and adoption is edition specific: an AHJ commonly adopts a named edition several cycles behind the current one, and amendments and errata are issued against named editions. So the edition binding your project may well not be the current one listed above. Check what has been adopted where the building is. NFPA offers free read-only online access, and current editions can be confirmed at nfpa.org .
In the Gulf, where I work, the framework is regional. The UAE Fire and Life Safety Code of Practice, from the UAE Ministry of Interior, Directorate General of Civil Defence, is federal and enforced emirate by emirate by the local Civil Defence authority; the edition confirmable from an official source is September 2018. It is NFPA-referencing. The Dubai Building Code, 2021 edition does not carry the fire provisions; fire defers to the UAE Fire Code and Dubai Civil Defence. By contrast Saudi Arabia's SBC 801, the Saudi Fire Protection Code from the Saudi Building Code National Committee, editions 2018 and 2024, is ICC and IFC based. There is no GCC-wide fire code, and in each case what governs your installation is the code as adopted and enforced locally.
Taking a fire pump out of service carries two hazards: the energy in the set, and an impaired fire protection system in an occupied building. Those disciplines sit in the permit to work guide and the lockout tagout guide. Where a management system is in place, the control selection logic required by ISO 45001:2018 as amended by Amd 1:2024, clause 8.1.2, and by ANSI/ASSP Z10.0-2019 section 8.4 for users of that framework, applies here as to any work. Both are voluntary standards, law nowhere by themselves; status can be confirmed at iso.org .
11. Where this meets facilities and maintenance management
A fire pump is a compliance asset as much as an engineering object, and that changes how it should sit in your systems.
- Criticality is not a judgement call. On almost any ranking a fire pump lands at the top, because its failure consequence is life safety plus regulatory plus insurance. If your register does not reflect that, the register is wrong. See the asset criticality classification guide and the facilities maintenance management guide.
- Records are the asset's second half. The listing documentation, the accepted performance curve, the as-installed drawings, the controller settings as commissioned and the history of every test and defect are what let anyone, years later, tell whether the machine still does what it was accepted as doing. A pump with no baseline curve on file is a pump whose test results cannot be interpreted. Any reasonable maintenance management system will hold them; keeping them complete and findable is the part usually missing. The methods in the condition monitoring guide can supplement that evidence, particularly on the electrical side and the diesel auxiliaries, but a machine that rarely runs gives thin trend data and they never replace the required regime.
- The pump room gets colonised. Pump rooms are quiet, lockable and out of the way, so they fill with spares, cleaning stock, contractor materials and abandoned plant. Access gets blocked, drainage covered, ventilation obstructed, and the room stops being the protected space it was designed as. Of the fire pump defects that get written up, this is among the most common and the easiest to fix. Walk yours.
- It sits in a wider life safety estate. The pump shares its standby character with alarm and detection, suppression, and equipment such as the systems in the elevator fire and life safety system PM guide. Managing them as one group, with one owner and one evidence trail, beats treating each as an isolated contract.
On maintenance itself I stop short deliberately, because it is the sibling guide's territory. The pointer worth leaving: the failure modes that take fire pumps out of service cluster in the auxiliaries rather than the pump. Batteries, fuel, valve position, controller mode, blocked strainers, and the slow drift of a machine nobody has flow tested against its curve. The regime that catches them is set out in fire pump inspection and preventive maintenance, and if you read one more thing after this, read that. Broader framing sits in the preventive maintenance guide.
The idea to walk away with
A fire pump exists because the water supply a building happens to have is frequently not the supply its fire protection system needs. What explains everything else is that the machine bridging that gap is a standby machine: it must start reliably after long inactivity, then run under load, unattended, on the worst day the building ever has. Read the subject through that lens and it becomes coherent. Listed assemblies, because reliability cannot be inferred from a machine that never runs. Electric and diesel drivers, often both, because the event that calls the pump is the event most likely to take out its supply. A jockey pump, so leakage never wears out the machine you keep in reserve. A controller built to start rather than to protect. A flow test as well as a run, because a pump that starts is not a pump that performs.
Final thoughts
If you manage buildings rather than design fire protection, the most useful thing you can do with this article is take it to your own plant room and use it as a walk. Which pump type is it, and does that tell you where the water comes from? Which driver, and what does that make the installation dependent on? Is there a jockey pump, and does anyone know how often it runs? Are the isolation valves open, and can you see that they are? Is the controller in automatic? Does anyone hold the accepted performance curve the pump was signed off against? Can you reach all of it without moving somebody's stored materials?
Those questions need no values, no code lookups and no specialist equipment, and the answers say a great deal about whether the most consequential machine in your building will do its job. The detailed regime comes from the standard your jurisdiction has adopted and from competent people applying it. This article exists so that when you read that standard, you understand what the machine is and why it is built as it is.
Disclosure
Alongside advisory work I also build a CMMS and CAFM platform, so I have a commercial interest in this category. Nothing above is a recommendation for it, and no vendor named here has paid for inclusion or had any editorial input. Weigh the analysis accordingly.
Getting your life safety assets under control?
Independent advisory on asset registers, criticality, compliance evidence and maintenance programme design for fire protection and other life safety plant. Over twenty years across utilities, oil and gas, manufacturing, government and facility operations. No contractor margins, no reseller arrangements.
Book a conversationRelated reading: Fire pump inspection and preventive maintenance, Fire sprinkler system types and components, Fire alarm system types and components, Elevator fire and life safety system PM, Asset criticality classification, Preventive maintenance: the complete guide, Permit to work, Lockout tagout.
Muhammad Abbas
CMMS / CAFM Manager & Independent Advisor · 22+ years across enterprise CMMS, EAM, CAFM and ERP implementations in utilities, oil and gas, manufacturing, government and facility operations.
Work with me