UPS backup time calculation

[sirhc0109]

Hi everyone! I hope everyone is well.

I am new in the UPS(uninterruptible power supply) industry. I would like to know how to compute the backup time/run time. As well as for battery Ah calculation.
I hope someone can educate me on these calculations.

Thank you in advance for your response.

-Chris

 

[Hyderz]

I dont know much about battery time usage but i would start here

Back-UPS | APC USA

Battery Backup & Surge Protector for Electronics and Computers | APC USA

www.apc.com

 

[ir_cow]

Most UPS products will tell you exactly how long it will last at what load. For example a 900W APC at my work will last just under a minute. With 18min for a light load. So it just depends how big you go and what is being loaded at the moment you loose power.

UPS for computers are not meant for long power outage. Just enough to save your work and turn it off. Or handle quick brownouts.

 

[Arctucas]

Most UPS products will tell you exactly how long it will last at what load. For example a 900W APC at my work will last just under a minute. With 18min for a light load. So it just depends how big you go and what is being loaded at the moment you loose power.

UPS for computers are not meant for long power outage. Just enough to save your work and turn it off. Or handle quick brownouts.

^This^

HWiNFO will tell you the estimated runtime at the current battery charge level and power load.

 

[9087125]

The following is off the top of my head

First off, you need to understand the diff between energy (joules or watt-hours) and capacity (coulombs or amps-hours), with a lead acid battery which is what most UPS use you ALWAYS want the amp-hours, unlike a nikel based or li-ion where Ah and Wh are both much more acceptable.
Joules = Electromotive Force * Coulombs, or
Watt-hours = Potential difference * Amps integrated overtime.

The capacity (and hence energy) that a LA provides is actually a function of the discharge rate, known as the Peukert empirical Law. The max amps you can pull from a battery (and hence power) is dictated by the internal diffusion current which in a LA is very bad due to the terrible diffusion coeffect of the liquid/starved electrolyte, intermediate reactions, the nature of the porous lead with the precipitation build up, and the surface area-surface area is a BIG deal, as it dictates the current density and hence impedance loses and other losses. So as the internal diffusion current drops, the cell drift current and hence cell voltage must drop, and voltage or potential difference is the driving force at the load. Like the kinetic energy of water molecules imparted to a windmill. Not enough voltage and hence electron-volts and the inverter will shut off.
So bigger loads reduce the capacity/runtime, aswell colder temps and age too. All these factors are derived by the battery manufacturer design and quality, simply put, performance can never ever be cheated, amp-hours can, but performance never.

You take the PC power draw and divide by the eff of the PSU (~0.85), then divide again by the UPS inverter eff (~0.85) then you divide by 10.5V (for 12V battery system), NOT 12V!!! rookie mistake.

That gives your c-rate or amps the battery is supplying, and then we plug that into Peukerts law rearranged:

Available Ah = battery's rated Ah x ( battery's rated Ah/(discharge current * battery's hour rate) ^(peukert exponent-1). hour rate is always 20hours for Lead acid.

what the equation tells you is you want your c-rate to be as close to the 20hour rating as possible.
Also most of these batteries have poor internal resistance and the ohmic loss (power loss) is another big factor that will drop the voltage even further. So throw all these in to the mix and the inverter can simple shut off straight away, or within a few minutes at most, despite all those coulombs still being locked up in the battery.

A typical cheap VRLA (SLI) LA is ~PE 1.4, factor in age and sulphation so try 1.6.

So an example: my PC setup with 3 large monitors and misc stuff uses 600watts while doing medium multitasking.
600/0.85/0.85/10.5= 79A!!!!!!!

a typical sized 12V SLI is what? 12Ah, so 79/12=6.5C...

12 * ( (12/(79*20) ) ^0.6 = 0.64Ah = inverter shuts off = zero runtime. Game over son. Of course that was an extreme example but you get the point?

Another way to look at it is the voltage drop, but off the top of my head I don't know the IR of these batteries. But a fully charged SLI has an open-circuit of ~12.85V minus overcharge from a float charge, so you add up the battery's IR and R of circuit which for both these cheap batteries and poor cheap wiring will be large. Use the formula: OCV -( current x (R + IR)). if it gets down towards 10.5V your in trouble. Remember cold and age cripple a LA as the diffusion current and aswell electrolyte ion mobility and hence concentration are affected primarily by their kinetic energy. Aged batteries usually have a sulphation build up that impedes ions to electrodes and increases contact resistance and polarization (more voltage drop) and since the build up is the active mass it actually decreases the capacity. You have to trust that the charging circuitry was made by someone who actually knew about batteries, many electrical engineers do not know anything sadly. Wrong voltages can dry out the cell slowly boiling off the electrolyte.

So I have given enough objective science to show the "runtimes" from the product are not very good to go by. Do pros go by them at all? No. no way, they use actual science, not assertions.
But ignorant and reckless society throw batteries in landfill and dont care. These things are an environmental blight. A good LA for partial deep cycling is suppose to last 10years on average, some do 20years. Stand-by/float service 20-40... Lead Acid, even VRLA are high maintenance and if you dont maintain them properly they simply wont last long.

People think this is all bull, but that is why those battery's have to be replaced every so often, . Environmental blights.

Should I unload on Li-ion and diy UPS??????? Its too easy... Should I pull out the LiFePO4....They never stood a chance

 

[Bill_Bright]

I would like to know how to compute the backup time/run time.

You can't. If you were using the UPS to power a lightbulb, this would be easy because a lightbulb consumes power at a constant rate and therefore places a consistent demand/drain on the UPS battery.

But a computer's power consumption is constantly changing from idle to near capacity as you, the user, ask it to perform different tasks. Even when you, the user are idle, power demands vary greatly because when the user goes idle, the operating system starts doing various housekeeping chores like indexing drives, checking for updates, installing updates, running defrag on hard drives and TRIM and wear leveling tasks on SSDs. When we are idle, our security programs update and start running scans. Other programs may check for and run updates too.

So because the load is constantly varying, there is no way compute battery runtime. The best we can due is a rough guesstimate based on past history with that computer and that user.

There are two relatively easy solutions. The first is to buy a UPS that has an integrated LCD status display panel, as seen on this APC UPS. It provides a wealth of information including the incoming line/grid voltage, charge status, current load, and "estimated" remaining runtime based on the current demand.

The second easy way is to buy an UPS that supports communications with the computer (typically via a USB cable) that is running the UPS maker's monitoring software. Most major UPS makers have such software and most of their upper tier UPS support such intercommunications. The software typically tells you everything the LCD status display tells you, and often times more. You also typically can initiate a test through the software.

The biggest advantage to using the software is you can tell the software to automatically and safely shutdown Windows and the computer BEFORE the batteries runout. But again, it require communications between the UPS and PC via a USB cable. The UPS will still protect the connected devices from surges, spikes, sags, dips and brownouts without the cable. But if a total power outage lasts longer than the batteries, the computer will crash when the batteries run out. Not good.

As well as for battery Ah calculation.

Not sure what you mean by this. The Ah is typically a specification for the battery or cells within the UPS. This will vary depending on the capacity and design of the UPS. Typical UPS use 12V 7Ah or 12V 9Ah SLA (sealed lead-acid) batteries.

If you meant the VA (volt amps) rating, that is usually determined by the UPS maker and is dependent on the value for PF (power factor) it chooses. Sadly, there is no industry standard for this value. Some makers use .8 others may use .9 and another maker may use some other value. But if not stated, going with .65 is commonly used as that conservative value ensures the calculation will provide a safe result.

To convert VA to watts, you go by the formula:

Watts = VoltAmps x Power Factor (W = VA x PF)

So, for example, if you have a 1500VA UPS and want to know its wattage capacity and the PF is .65, it would be

W = 1500 x .65
W = 975

Or to go from VA to watts, you use the formula: VA = W/PF

VA = 975/.65
VA = 1500

HWiNFO will tell you the estimated runtime at the current battery charge level and power load.

Ummm, sorry but not true. It will for laptops, but not PCs.

 

[sirhc0109]

Thank you for all your responses.
@ir_cow I understand your point. However, a few UPS manufacturers still don't include the run time in their user manual/document.
@9087125 I understand some of your points and still trying to absorb the others.
@Bill_Bright What I mean is that if I were to base the runtime on the customer's load requirement, how would I compute it for the runtime given the UPS capacity and the battery bank capacity? For the Ah calculation.. For example, I have the info on the UPS capacity and the load requirements, how can I compute to know which Ah capacity I can use? Because there are higher VA UPS that can adjust to different battery Ah. For example, 12v 7ah/9ah are common for a 10kva but can also handle others like 12ah, 18ah, etc.

What I meant for my post is that I want to know how to calculate the runtime or backup time based on the information I have like UPS capacity, battery bank capacity(v,ah), and load requirements.
The same goes for the Ah calculation. After knowing the UPS capacity, the battery bank voltage, and the load requirements. How can I decide which Ah capacity is required?

Please let me know if my query is still not clear.

Thanks again.

-Chris

 

[Arctucas]

<SNIP>


Ummm, sorry but not true. It will for laptops, but not PCs.

OK.

 

[Hankieroseman]

Hi everyone! I hope everyone is well.

I am new in the UPS(uninterruptible power supply) industry. I would like to know how to compute the backup time/run time. As well as for battery Ah calculation.
I hope someone can educate me on these calculations.

Thank you in advance for your response.

-Chris

My rule of thumb, and I could be wrong, but if I get 1200Watt PSU, I need a 1200watt UPS. I have one on every computer and tv set in the house. Surges have been bad here. I have an APC that shows the approx. load and how long it will run on the display. APC BACK-UPS 1350 model for reference.

 

[chrcoluk]

As I understand it at lower loads you lose efficiency so the run time doesnt go up linearly, condition of battery is also a factor of course.

As said in the first reply, consumer UPS typically is designed as a safe shutdown rather than continue working thing, although in very low power loads they might last a while, and in moderate loads will get you through short lived power cuts, brownouts etc.

I have become very used to using UPS devices now. Almost all of my networking and computing equipment either has its own battery or is connected to a UPS.

 

[Bill_Bright]

What I mean is that if I were to base the runtime on the customer's load requirement, how would I compute it for the runtime given the UPS capacity and the battery bank capacity?

Once again, you can't. You are not powering a lightbulb that presents a constant load. The load created by a computer is constantly varying from a tiny load at near idle, to maximum load with tasked. And that swing can happen several times a minute - without the user even knowing. So there is no way to easily determine the average load and therefore, no way to determine estimated runtimes - especially when you don't yet know the size of the UPS.

You can figure out the minimum size UPS you need, however. The easy way is to use the eXtreme OuterVision PSU Calculator. Plug in all the current components. If upgrades are planned in the future, use them. The calculator will then determine the minimum PSU size needed for that computer. HOWEVER, the nice thing about that eXtreme PSU calculator is it will also recommend a UPS size too.

Add extra for each additional device that will be supported by that UPS. For example, if the monitor, router, modem will be supported, look on their power supplies for the wattage and add that to the UPS.

That said, if more runtime is needed, simply get a bigger UPS than recommended.

You can't go wrong with a 1500VA UPS. For one, the bigger UPS tend to have the nicer features, better AVR, and faster cutover times. But also, obviously, support bigger loads and/or provide longer battery runtimes. My APC 1500VA UPS for examples, supports this computer, two 24" monitors, my wireless router and my modem. If there is a power outage, if I quickly shutdown my computer and power off my monitors, my modem and router will keep my network alive for over 2 hours. This is nice because then I still have Internet access via wifi with my other devices.

Technically, you should add together the maximum demand each connected power supply could, in theory, demand and that is the minimum UPS you should get. But this method is flawed because most users buy computer PSUs bigger (often much bigger) than their computer will ever actually demand.

@Arctucas - my bad. You are correct. My apologies. I was not looking in the right place for one , but also I was not being clear in explaining myself.

But again, since the image clearly shows the model number of the UPS attached, that indicates there must be an interconnecting cable attached to an USP that already exists - which is really the point I was getting at. And that still does not help the OP determine what size UPS he needs - which was his point. And by the way, just keeping that screen up with an eye on it and tasking this computer to do different things, I see my estimated run time varies from 55 minutes while simply typing this sentence to 23 minutes as I start other additional tasks, then back up again as those tasks complete. That is a wide swing, IMO. And those monitors have no clue how long the user or computer will be sitting at those load levels.

 

[Arctucas]

<SNIP>

@Arctucas - my bad. You are correct. My apologies. I was not looking in the right place for one , but also I was not being clear in explaining myself.

But again, since the image clearly shows the model number of the UPS attached, that indicates there must be an interconnecting cable attached to an USP that already exists - which is really the point I was getting at. And that still does not help the OP determine what size UPS he needs - which was his point. And by the way, just keeping that screen up with an eye on it and tasking this computer to do different things, I see my estimated run time varies from 55 minutes while simply typing this sentence to 23 minutes as I start other additional tasks, then back up again as those tasks complete. That is a wide swing, IMO. And those monitors have no clue how long the user or computer will be sitting at those load levels.

Yes,

It has a USB cable connecting it to the PC.

The UPS itself also has a Runtime on the LCD display.

 

[Bill_Bright]

The UPS itself also has a Runtime on the LCD display.

I won't have an UPS without such a display. They can be so informative. For example, you can see your current load. You can check the line voltage and frequency and percent of charge on the batteries.

You can even use it to see how much power individual components use. For example, I can see that right now my system is demanding 158W of power. But if I turn off one of my monitors, I see that drops to 123W.

Minus 7 carry the 1 and I see that my monitor consumes 35W. Good to known. And I didn't even have to mess with plugs and my kill-a-watt meter.

 

[eidairaman1]

You can't. If you were using the UPS to power a lightbulb, this would be easy because a lightbulb consumes power at a constant rate and therefore places a consistent demand/drain on the UPS battery.

But a computer's power consumption is constantly changing from idle to near capacity as you, the user, ask it to perform different tasks. Even when you, the user are idle, power demands vary greatly because when the user goes idle, the operating system starts doing various housekeeping chores like indexing drives, checking for updates, installing updates, running defrag on hard drives and TRIM and wear leveling tasks on SSDs. When we are idle, our security programs update and start running scans. Other programs may check for and run updates too.

So because the load is constantly varying, there is no way compute battery runtime. The best we can due is a rough guesstimate based on past history with that computer and that user.

There are two relatively easy solutions. The first is to buy a UPS that has an integrated LCD status display panel, as seen on this APC UPS. It provides a wealth of information including the incoming line/grid voltage, charge status, current load, and "estimated" remaining runtime based on the current demand.

The second easy way is to buy an UPS that supports communications with the computer (typically via a USB cable) that is running the UPS maker's monitoring software. Most major UPS makers have such software and most of their upper tier UPS support such intercommunications. The software typically tells you everything the LCD status display tells you, and often times more. You also typically can initiate a test through the software.

The biggest advantage to using the software is you can tell the software to automatically and safely shutdown Windows and the computer BEFORE the batteries runout. But again, it require communications between the UPS and PC via a USB cable. The UPS will still protect the connected devices from surges, spikes, sags, dips and brownouts without the cable. But if a total power outage lasts longer than the batteries, the computer will crash when the batteries run out. Not good.


Not sure what you mean by this. The Ah is typically a specification for the battery or cells within the UPS. This will vary depending on the capacity and design of the UPS. Typical UPS use 12V 7Ah or 12V 9Ah SLA (sealed lead-acid) batteries.

If you meant the VA (volt amps) rating, that is usually determined by the UPS maker and is dependent on the value for PF (power factor) it chooses. Sadly, there is no industry standard for this value. Some makers use .8 others may use .9 and another maker may use some other value. But if not stated, going with .65 is commonly used as that conservative value ensures the calculation will provide a safe result.

To convert VA to watts, you go by the formula:

Watts = VoltAmps x Power Factor (W = VA x PF)

So, for example, if you have a 1500VA UPS and want to know its wattage capacity and the PF is .65, it would be

W = 1500 x .65
W = 975

Or to go from VA to watts, you use the formula: VA = W/PF

VA = 975/.65
VA = 1500


Ummm, sorry but not true. It will for laptops, but not PCs.

The function of a ups is like a laptop battery, power settings in windows will dictate, so once battery life is low enough windows will force shut down if person ignors warning alarm to save work and close everything...

 

[Bill_Bright]

The function of a ups is like a laptop battery

That's one function but in my opinion, a minor, almost bonus function.

By far, the greater, most important function of a "good" UPS is the AVR - automatic voltage regulation feature. A surge and spike protector is little more than a fancy, expensive extension cord as they do nothing for low voltage events like dips (opposite of spikes), sags (opposite of surges) or brownouts (long duration sags). And for excessive surges and spikes, they simply cut power (“IF” working properly), crashing your computer - never good.

The AVR of a "good" UPS will easily compensate for all those anomalies - often without even needed to kick over to battery. So IMO, back up battery power during a full power outage is just a minor bonus feature.

But yes, to your point, the UPS will initiate a "graceful" shut down of Windows and the computer when the batteries run low AS LONG AS the intercommunications cable between the UPS and computer is connected. If no cable, when the batteries run down, the computer will crash.

 

[eidairaman1]

That's one function but in my opinion, a minor, almost bonus function.

By far, the greater, most important function of a "good" UPS is the AVR - automatic voltage regulation feature. A surge and spike protector is little more than a fancy, expensive extension cord as they do nothing for low voltage events like dips (opposite of spikes), sags (opposite of surges) or brownouts (long duration sags). And for excessive surges and spikes, they simply cut power (“IF” working properly), crashing your computer - never good.

The AVR of a "good" UPS will easily compensate for all those anomalies - often without even needed to kick over to battery. So IMO, back up battery power during a full power outage is just a minor bonus feature.

But yes, to your point, the UPS will initiate a "graceful" shut down of Windows and the computer when the batteries run low AS LONG AS the intercommunications cable between the UPS and computer is connected. If no cable, when the batteries run down, the computer will crash.

I remember ups using the RS-232 DB9 Seial ports before they were switched to usb, heck they might have used the parallel port at 1 time.

 

[Bill_Bright]

I remember those days too. In fact, I have a 30+ year old APC Smart-UPS 900 supporting my garage door opener that uses a RS-232 port. I think it is on its 6th set of batteries but still works great. I re-assigned it out to the garage about 10 years ago because (1) it is that old, ugly original beige computer case color and (2) its main transformer buzzes all the time. But again, it still works great.

 

[Chrispy_]

Hi everyone! I hope everyone is well.

I am new in the UPS(uninterruptible power supply) industry. I would like to know how to compute the backup time/run time. As well as for battery Ah calculation.
I hope someone can educate me on these calculations.

Thank you in advance for your response.

-Chris

Are you talking about personal UPS for your own system, or industrial UPS for server room(s) or datacenter?

I have plenty of real world experience with a couple of the larger Borri Ingenio systems (120kVA 3-phase 230V units that require a room to themselves and are several meters in two dimensions) and smaller single-phase systems up to 20kVA from Legrand like the Trimod HE. I also have history with plenty of inherited rackmount UPS and would advise against them all unless you are talking very small installations like a single 42U server rack.

For home stuff, I have no clue other than to find one that offers true sine-wave output, since some of the cheaper offerings use square-wave AC that f*cks up modern switching PSUs.

Runtimes depend on load. It's not a linear relationship between capacity and runtime - the more load you put on the cells, the less efficient they are (more energy is lost as heat, which is why large-scale UPS require plenty of ventilation and ideally their own air conditioning.

UPS are typically sized in kVA and rather than guessing you should just use an online calculator like https://powercontrol.co.uk/resources/kva-calculator/

 

[Bill_Bright]

For home stuff, I have no clue other than to find one that offers true sine-wave output, since some of the cheaper offerings use square-wave AC that f*cks up modern switching PSUs.

This is not true - it is pure marketing hogwash by the pure sinewave UPS makers and I really wish folks would stop repeating that myth.

First, just as I would never recommend a budget, entry level power supply, I would never recommend a budget, entry level UPS either for our computers (or home theater systems) either. But regardless, in NO WAY do even cheap UPSs intended for computer support output square-waves! They output approximated (AKA stepped, modified or simulated) sinewaves that look like this.

A square wave looks like this. See the difference?

Second, any 1/2 way decent, modern switching PSU can easily deal with simulated sinewaves. They do it all the time and have for decades! It is silly to suggest or even think they couldn't. There are entire regions of the world that deliver power to their customers through generators connected to inverters that then drop power into homes and offices. Boats, RVs, and motor homes use inverters. And what do inverters output? Simulated sine waves.

EVGA FAQ and Seasonic on stepped approximation output with their PSUs. EVGA says their APFC PSUs work fine with simulated sinewaves.

Q: Do EVGA Power Supplies support UPS backup devices that support Line-Interactive AVR UPS which uses a simulated/artificial sine wave?

A: Yes, all EVGA power supplies support Active PFC and UPS backup devices.

Seasonic also does not have a problem. While Seasonic recommends pure sinewave UPS, they also clearly say,

a high quality simulated sinewave UPS from a reputable manufacturer could also be a possible solution to be used with our power supplies.

Third, it is critical to remember that for the vast majority of users the vast majority of the time, the output of the UPS is the exact same (or slightly regulated if needed by the AVR) as supplied by the grid - whether that is inverter created or not. Only when the UPS is providing backup power with its batteries does the UPS actually convert the DC from the batteries to AC. This means for the typical user, it is a few minutes per month, or even less.

Note Eaton, arguably one of the best UPS manufacturers out there says (my bold underline added),

It’s important to understand that both types of UPS system produce true sine wave output more than 99% of the time.

Modified sine wave UPS systems typically protect PCs

The truth is, today's modern switching power supplies are even more robust than those made in past years. And those made in past years ONLY had approximated sine wave UPS to work with and they did so with aplomb. So today's can too.

So please, let's stop spewing and propagating the false marketing hype. It is just that, false marketing hype intended solely to take sales away from stepped output UPS.

FTR, I am NOT suggesting pure sine wave UPS should be avoided. On the contrary - if you can find one at a competitive price that meets your needs, then go for it. But do NOT dismiss an approximated sinewave UPS just because it does not output a pure sinewave.

 

[eidairaman1]

I remember those days too. In fact, I have a 30+ year old APC Smart-UPS 900 supporting my garage door opener that uses a RS-232 port. I think it is on its 6th set of batteries but still works great. I re-assigned it out to the garage about 10 years ago because (1) it is that old, ugly original beige computer case color and (2) its main transformer buzzes all the time. But again, it still works great.

The subwoofer which is also the power supply for the speakers uses a db9 rs-232 connector layout, i needed an extension cable so I found a serial cable and it works perfectly.

 

[Bill_Bright]

i needed an extension cable so I found a serial cable and it works perfectly.

Makes sense. There would be no need to "crossover" any wires within the cable (making it proprietary) so, as long as the connectors are standard, and all the necessary pins in the connectors are used, a standard serial cable would work.

Come to think of it, I have an old set of Polk Audio 2.1 computer speakers that I believe also use db9 rs-232 connectors. I will have to keep your solution in mind if I ever need an extension cable.

 

[eidairaman1]

Makes sense. There would be no need to "crossover" any wires within the cable (making it proprietary) so, as long as the connectors are standard, and all the necessary pins in the connectors are used, a standard serial cable would work.

Come to think of it, I have an old set of Polk Audio 2.1 computer speakers that I believe also use db9 rs-232 connectors. I will have to keep your solution in mind if I ever need an extension cable.

X-230s, these were pretty good considering there was a 5.1 and 7.1 line, before these i had Labetecs and were awesome

 

[sirhc0109]

Are you talking about personal UPS for your own system, or industrial UPS for server room(s) or datacenter?

I have plenty of real world experience with a couple of the larger Borri Ingenio systems (120kVA 3-phase 230V units that require a room to themselves and are several meters in two dimensions) and smaller single-phase systems up to 20kVA from Legrand like the Trimod HE. I also have history with plenty of inherited rackmount UPS and would advise against them all unless you are talking very small installations like a single 42U server rack.

For home stuff, I have no clue other than to find one that offers true sine-wave output, since some of the cheaper offerings use square-wave AC that f*cks up modern switching PSUs.

Runtimes depend on load. It's not a linear relationship between capacity and runtime - the more load you put on the cells, the less efficient they are (more energy is lost as heat, which is why large-scale UPS require plenty of ventilation and ideally their own air conditioning.

UPS are typically sized in kVA and rather than guessing you should just use an online calculator like https://powercontrol.co.uk/resources/kva-calculator/

I am talking about industrial UPS.
For example:
The customer told me that his total load requires 15.5kW(Max load) and he wants 60 minutes of backup time.
How can I compute how to achieve this output and be able to provide detailed information [UPS capacity, battery capacity, quantity, configuration] to the customer?

Can I use a 20kVA UPS with a PF of 0.9?
What about the battery capacity and quantity I need/can use to achieve the backup time?
What is the calculation/formula to get the details for the battery bank?
Is there any formula to calculate the backup time[when I know the UPS kVA capacity, load(kW) required, and battery bank capacity] and battery capacity[when I only know the desired backup time(ex. 45 mins), UPS kVA capacity, load(kW) required)]?
How can I calculate the Ah required to provide the desired backup time and how many external banks/packs do I need to match the overall requirement?

 

[Waldorf]

might be better to go with a generator, and smaller ups (just to cover "switching")?

 

[Chrispy_]

I am talking about industrial UPS.
For example:
The customer told me that his total load requires 15.5kW(Max load) and he wants 60 minutes of backup time.
How can I compute how to achieve this output and be able to provide detailed information [UPS capacity, battery capacity, quantity, configuration] to the customer?

Can I use a 20kVA UPS with a PF of 0.9?
What about the battery capacity and quantity I need/can use to achieve the backup time?
What is the calculation/formula to get the details for the battery bank?
Is there any formula to calculate the backup time[when I know the UPS kVA capacity, load(kW) required, and battery bank capacity] and battery capacity[when I only know the desired backup time(ex. 45 mins), UPS kVA capacity, load(kW) required)]?
How can I calculate the Ah required to provide the desired backup time and how many external banks/packs do I need to match the overall requirement?

kVA refers to whether the UPS can handle the load or not. So yes, a 20kVA UPS should be appropriate for 15.5kW max load.

Runtime is simply a matter of adding battery shelves or additional battery cabinets. When you buy an industrial UPS the conversation will be around how many batteries and battery cabinets you want. Speak to a reseller local to your country - they'll be able to advise. Here in the UK, the Legrand UPSs seem to be relatively good value and often come back as the most competitive option or at least one of the more competitive options when I'm doing my due diligence and getting multiple quotes from different providers.

To give you a very rough estimate, if you were in the UK a 20kVA UPS able to run at 80% for an hour is going to be a relatively small unit, possibly self-contained as a single floorstanding tower the size of a fridge-freezer, with about 1000kg of batteries. Pricing is likely in the £10000 ballpark, before tax.

If you want more specifics then it really is time to just start getting a couple of quotes from UPS installers in your region. A lot of specifics will come down to the site you're intending to use this in, such as whether there's three-phase power available, whether you already have or need to reinforce the floor tiles, whether you already have circuits with high-current breakers installed that can be used. A non-trivial cost of the industrial UPS cost is the electrical work and physical prep of the area the UPS is going to need.

might be better to go with a generator, and smaller ups (just to cover "switching")?

This is honestly why many huge (million dollar) UPSs are only expected to run for a few minutes. They're normally datacenter scale, or building-scale UPS though.

For the 20kVA ballpark OP is looking at, there are plenty of options. I had a 40kVA unit that could do 6 hours on one site that was subject to annual half-day outages. It's simply a case of having enough battery cabinets to meet the runtime requirements - and they typically all still use lead-acid batteries so they're big and heavy but not actually that expensive.