Dipak Sen Choudhury weighs up the pros and cons of lithium and lead-acid for Indian data centres.
The lead-acid technology has a challenge, and hence a chance. It must last more than the 3–5 years’ life perception it suffers from. Lithium life is, 10 years, I suppose. You talk to anyone in the country who is planning back-up power for a data centre, and this is the wall you hit.
India has a vast and fast-growing UPS battery market. In a simple back-of-the-envelope estimation, the current market size is around 10GWh annually and growing at a CAGR of 7–8%. This is a market entirely served by lead-acid technology, cutting across applications. They range from small 350VA offline UPSs using a 12V 7Ah to a large 500KVA or even bigger line-interactive UPSs for large banks and manufacturing set-ups which use multiple banks of 12V 100–200Ah in parallel.
UPS dominated
The UPS business in the country is dominated by global brands like Delta, Eaton, Vertiv, Socomec, Fuji and Indian manufacturers such as Numeric Power, a brand now owned by the French major Legrand, Switching and Riello Power. Users are mostly in the banking and financial sectors, health, commercial and residential property and manufacturing.
They all need back-up power to all electricity-actuated operations, keeping in mind the generally unreliable grid power availability. The two Indian majors meet almost the entire battery requirement – Exide Industries and Amararaja Batteries. Less than 10% of the overall market goes to other, smaller manufacturers.
Battery life varies between 2–5 years, depending on two operational parameters – how much and how frequent the power outages are. The outages cause the battery to ‘cycle’. Whether the battery banks are in air-conditioned or non-air-conditioned environment matters too.
There are some murmurings about battery life being only around three years. But by and large, users are reconciled to the fact that once battery bank life crosses three years, the battery may have to be replaced any day.
UPS largely lead-acid
The smallest segment, an absolute disposable commodity market, is flooded with cheap batteries imported from China, Vietnam and Thailand. They are of very poor quality, sell with six months’ warranty and last for 6–12 months only.
Here, traders produced a ‘game-changing’ lithium promise of long life, though at 2–3 times the cost. They have been able to corner a reasonable percentage of the market, delivering 1+ year of life.
Safety, with respect to lithium packs, is of least consideration as traders generally deal with the bottom-of-the-rung quality lithium imports, mostly from China, with more inferior BMS bought ready-made from various dubious sources there. At the other extreme, for the larger UPS ratings, say in the 200–500KVA ranges, there is some conversion to lithium. This is in part due to footprint advantage and also due to the purported 10 years’ assured life at a considerably higher capex, justified by the total annual cost of ownership (TCO) calculations.
Data centres’ entry
Against this background, the data centres have made their entry, creating a new niche for power back-up – one that is built around stories of reliability and sustainability. Oddly enough, the two attributes are quite opposed but then these are the two parameters on which energy planning – particularly back-up energy – for data centres is being decided.
Indians are one of the highest consumers of data amongst comparable countries with an estimated 19GB per capita per month. This is mainly because of ever-increasing digital penetration the length and breadth of the country, cutting across diverse economic and social status.
In a country with a population of 1.5 billion, the number of mobile phone connections is close to 1.2 billion. Even amongst the rural poor, digital payment by mobile phone is commonplace.
Paradoxically, despite all the country’s economic and social challenges, digitisation has happened at an astonishing pace and is a part of daily life. The internet is now a part, unknowingly perhaps, and significantly facilitates the daily life of even the most ordinary Indian.
India formally has 880 million internet users at the last estimate in late 2022. An astonishing quantum of data is constantly circulating, touching the lives of every citizen, as the government has given a massive boost to digitisation of every basic need. No one needs to visit anywhere anymore to collect wages, pay a bill or seek a certificate. Almost everything is available online.
Covid itself has been one major booster to online transactions and today long after the abatement of the dreaded disease, the benefits of virtual working have been permanently adopted by different institutions.
Data centres taking control
And now, with AI lurking in the corner, the incremental data creation, transmission and storage will go up by several orders of magnitude and data centres will take complete control of all our daily transactions.
India is the world’s 13th or 14th largest data centre market, with more than 140 such facilities already in operation. An additional 50 or so new co-location data centres, with a cumulative 2GW of demand, are planned to open by 2028. However, in view of rapid digital penetration, there will be additional demand of anywhere between another 2–4GW by 2030 or sooner.
Most of the big names of the IT world – Amazon, Google Microsoft, Oracle, Yotta Infrastructure, Blackstone’s Lumina and Cloudinfra – are all here. Since 2023, these global hyperscalers are fast building up significant data-handling capacity in the country at a very rapid pace.
Presently, Google has two of its 26 global cloud regions in India while Microsoft Azure has three of its 65 global cloud regions in India. Indian corporate houses too have quickly taken up the opportunity and big names like Tata Communication, Larsen & Toubro, as well as the inevitable Adanis and Ambanis, are all moving towards either independently or jointly developing data centres in India.
International partners like Brookfield Infrastructure Partners LP and Digital Realty Trust Inc are getting involved in joint ventures. Their aspirations are big as they find the opportunities to be enormous – so much so that Adani alone plans to put up a 1GW data centre platform within the next decade.
UPS at the eye
Coming to the overall energy planning in a data centre, one must first realise that for conventional residential, commercial, or even industrial application, the UPS is a part of the auxiliary support system. This ensures that the overall system downtime is minimised, if not eliminated. In the case of a data centre, the UPS sits right at the eye of the business and ensures more than 100% zero downtime.
Moving towards the future where autonomous driving or remote enabled surgery may become the order of the day, where AI would predominate much of our current practices, zero downtime is no longer a term that can be discussed or negotiated. It is intrinsic to the definition of back-up. Reliability of energy storage technology thus takes a different dimension here.
Data centres are mostly in Tier I and Tier II cities where real estate costs are a decisive parameter on decisions pertaining to investments. The battery footprint obviously becomes a major technical parameter for choice of preferred technology, and it is here that lithium has a major advantage over lead-acid, although at considerably higher capex.
This is where the claimed long (cycle) life of lithium, generally equated with 8–10 years’ field service life, has turned the ownership cost equations towards lithium.
A well-engineered lead-acid system, under similar air-conditioned operation with periodic short discharges, can easily deliver 7–8 years’ life. But it suffers from a perception deficit that it belongs to a lead-acid family which in UPS applications are known to deliver 3–5 years of life only.
Nevertheless, in this segment lead-acid imports from European manufacturers – with much improved Peukert factors based on thin-plate designs – have made an impact and are quite sought after by data centre engineers.
Cannot compete
Lead-acid for large industrial UPS applications, say for even 2200KVA installations in conventional as well as nuclear power stations have comfortably delivered service life well above 10 years, also under a non-airconditioned environment. But then these are basically 2V battery banks and due to footprint reasons, cannot compete with indoor lithium installations.
A final point which goes against lead-acid in data centre application is the Peukert factor. As data centre backups are all short duration (15–30 minutes) the Peukert index of lead-acid, as compared to lithium, plays a major role in final battery sizing. This is the KVAh or KWh of the proposed battery bank, calculated on the basis of the specified load and planned backup. The much larger VAH numbers of the sized battery, compared to lithium, takes away the cost advantage of lead-acid to a considerable extent.
Started in a big way
Data centres in India have started to happen in a big way. There is a huge opportunity for energy storage solution providers to get a major chunk of business from this segment. As of now, while lead-acid and lithium are both being used, lead-acid will quickly lose the segment unless it is able to substantially improve the Peukert index, footprint and, of course, the life.
A quantum improvement in some or all these parameters can deliver a much more sustainable energy solution to data centres. In terms of the environment, a lead-acid solution is far ahead of any competition.
