When a modular UPS makes sense
When a modular UPS makes sense
Last updated: 3 July 2026
Rakesh had already made up his mind. He is the facilities head at a Pune auto components group, and he wanted a 60 kVA monolithic on the wall. His argument was simple. It works, it is cheaper on paper, and the vendor next door has been calling him for months.
Nitya is their CIO. She wanted a modular frame. Her argument was that the load was going to double in two years and triple in four, and she did not want to sit through another power-room shutdown to swap a bigger UPS.
They asked us to referee. Here is how that conversation went, and the answer we landed on.
The Pune server room, before the argument
Their setup was standard for a mid-size manufacturer. One 10 kVA online UPS from an earlier vintage, feeding a small server rack, an ERP box, an on-prem AD, and the SCADA gateway for the shop floor. Present load, sixteen kilowatts at peak. Room, about twelve square metres. Cooling, one four-tonne precision unit that was working harder than anyone liked to admit.
The trigger was a signed IT roadmap. Two new production lines by 2028. A private cloud migration for the design team. Six more virtualisation hosts. Their own IT team had costed the future load at forty to forty-five kilowatts. Nitya said this was firm. Rakesh said IT plans always shrink at budget time.
Both of them were partially right. That is exactly why this decision is hard.
What Rakesh saw when he priced a monolithic
A 60 kVA online double-conversion tower from a good Indian OEM, wall installed, three-phase, comes in around ₹ 4.5 to 5.5 lakh for the frame. Add batteries for 30 minutes of runtime and you are landing at ₹ 7 to 8 lakh all in. His logic was fair. Buy the future capacity today, get it discounted, and stop thinking about it for eight years.
The hole in the plan was efficiency. A 60 kVA tower running a 16 kW load is running at about twenty-six percent load. At that point most online units drop to 88 or 89 percent efficiency. The delta is small on paper. It is not small on a monthly bill. On a 24×7 load, a two percent efficiency gap is roughly ₹ 2,500 to ₹ 3,000 a month, every month, until the load catches up to the box. That is money he never got back.
He also had the same problem every facilities lead has. He was buying a single monolithic path. If the rectifier board failed on year six, the whole box went to bypass and the servers ran on raw mains until an engineer arrived. In Pune, in July, at three in the morning. Bas.
See a real number on a real load
What Nitya saw when she priced a modular frame
She asked for a 90 kVA modular frame populated with three 25 kW power modules on day one. That is a 75 kW installed capacity, sized to a present sixteen-kilowatt load plus two more modules of headroom for the growth she had already signed. Later she could add a fourth and a fifth module without touching the frame.
The sticker was worse. A modular frame from a mainstream brand, plus three modules, plus batteries, was landing around ₹ 12 to 14 lakh in Pune pricing at the time of the discussion. Roughly 60 to 70 percent more than the monolithic quote. Rakesh took one look at that number and shook his head.
What the modular quote bought her was different. Efficiency stayed near 94 to 96 percent even at a light load, because the frame controller balanced the load across three modules and idled the rest. Redundancy came from the module count itself. A failed 25 kW module took itself offline and the other two carried the load without a blink. And if a module ever went sick at three in the morning, a service engineer could slide it out on rails and slide a spare in, without touching the load.
The number that ended the argument
We ran a five-year power ledger for both options against their actual planned load curve. Sixteen kilowatts today, twenty-eight in year two, thirty-six in year three, forty-two in year four. Same site, same tariff, same runtime.
| Metric | 60 kVA monolithic | 90 kVA modular (3 x 25 kW day one) |
|---|---|---|
| Day-zero capex, all in | ~ ₹ 7.5 lakh | ~ ₹ 13 lakh |
| Efficiency at year 1 load (~16 kW) | ~ 88 percent | ~ 94 percent |
| Efficiency at year 4 load (~42 kW) | ~ 93 percent | ~ 96 percent |
| Losses billed over 5 years | ~ ₹ 5.4 lakh | ~ ₹ 1.8 lakh |
| Planned outage to add capacity | One full swap, day-long | Zero. Slide in a module. |
| MTTR after a hardware fault | Hours, engineer visit | Minutes, hot swap |
Rakesh was quiet for a minute. The sticker gap of about ₹ 5.5 lakh looked ugly on day one. Over five years, the modular option was actually a shade cheaper on a total-cost basis, and it left them with a frame that could keep going to 100 kW without a forklift job. He agreed to the modular. Nitya did not gloat, which I respected.
When a modular UPS actually makes sense
Not for everyone. There is a shape of buyer this decision fits, and a shape it does not. Here is where we land it, after a few dozen of these calls.
Modular is the right buy when your present load is above roughly 20 kW, your growth is written into a plan not just hoped for, and a planned power-room shutdown to add capacity would cost you more than a small vendor visit. It is also the right buy when you have compliance pressure to avoid single points of failure in the UPS path, which most BFSI, healthcare and mid-market SaaS setups do.
Monolithic is the right buy when your load is small and steady. If you are protecting a 6 kW load in a satellite office and it is going to stay a 6 kW load, a properly sized monolithic online UPS from a good brand is cheaper on day one and easier to service in a tier-two town. We size a lot of these. Rakesh was not wrong about monolithic. He was wrong about that room.
Modular becomes a mistake when it is sold as a status buy. If a vendor is pushing a 300 kW frame for a 20 kW room “because you might grow one day”, they are selling you empty air with a controller on top. Ask for the load-growth curve on paper before the frame gets picked.
Get a straight modular vs monolithic quote
What we install, and how we scope it
We deploy modular frames from BPE and a couple of other Indian and global brands, sized against a real load survey rather than a wishlist. The scope of work is usually the same. A one-day site walk to meter the actual load, a signed growth plan from the customer, a room heat calculation, and then a written recommendation with both a modular and a monolithic option on the same page. The customer picks. We install.
The BPE modular family runs from about 20 kW up to well above 200 kW in a single frame, with hot-swap power modules, N+1 redundancy at the module level, and lithium battery support if the room can carry it. For rooms that do not need modular, the BPE standalone online range is what we spec instead. Both go through the same warranty and the same AMC. The rest of what we do for a UPS refresh sits on our BPE UPS India page.
The three questions we ask before we quote
These are the questions we ask every buyer before we send a quote. If your vendor is not asking them, that is a sign.
First, what is your metered present load, not your name-plate load. Most Indian server rooms are running at 40 to 55 percent of what the electrical single-line diagram says. If you buy against the SLD, you overpay every month for the life of the box. We cover this in detail in the sibling piece on sizing a UPS for an Indian office.
Second, what does the 36-month load curve look like on paper. Not a hope. A number Nitya could put next to her name and defend at a board meeting. That number is what tells us how many modules to populate on day one, and how many bays to leave empty. Our how to size your UPS without overbuying note has the sizing math we use.
Third, what battery chemistry does the room want. That is a whole conversation on its own. A lithium pack changes the modular economics. It also changes the AMC. Our field notes are in lithium vs VRLA batteries for UPS.
The one-line answer, if you skimmed
Modular UPS makes sense when your load is going to move, when the growth is real not aspirational, and when a planned outage to add capacity would cost you more than the sticker gap. Below 20 kW steady load, buy a good monolithic. Above that, with a signed growth plan, buy modular and pay less over five years.
Start with a quick site survey
We have sized a hundred and forty UPS refreshes in India in the last two years. Sixty of them went modular. Eighty went monolithic. The right answer sits in the load curve, not in the brochure. Reach us on WhatsApp at +91 91375 93228 between 10 and 7 IST, or use the button above.
P.S. Sudeep here. Rakesh called me six months after that installation went live. He said the power bill for the server room had dropped by about ₹ 8,000 a month against the same season the previous year, and the auditor had stopped writing “single point of failure” in his UPS row. His line to me was, “next time I argue with Nitya, I will lose faster.” Chalo, that is progress.
