Dr. Md Rishad Ahmed
In late 2010 I worked in an office in Banani, Dhaka. My job that winter was to build a spreadsheet for Total Gaz of France. The company wanted to know one thing: how much heavy furnace oil would Bangladesh need to buy over the next ten years?
I took the government’s generation expansion plan. I listed every power plant that was planned or under construction. I worked out how much fuel each one would burn at its rated output. Then I gave the client a number.
I was pleased with that model at the time. It was careful work. It included the storage capacity at the ports, the discharge rates at the jetties, and the way BPC priced liquid fuel for power plants. What I did not think about, at twenty-something, was what the number really meant. I had not calculated a business opportunity. I had calculated a bill. That bill would come due in dollars, every month, for the next twenty years, in a country that had no steady way of earning dollars.
Those were the years of the Speedy Supply of Power and Energy Act. Rental and quick rental plans were approved very fast. The argument for them was reasonable at the time. The lights were off, factories were idle, and an oil-fired engine could arrive on a barge and start running within a year. Nobody in that office thought we were building something permanent. We thought it was a temporary bridge.
Fifteen years later, 62 of the 143 power plants on the national list are short of fuel. Twenty-six of them run on gas, 33 on liquid fuel and two on coal. Load shedding has been averaging around 3,000MW. I know how we got here, because I did some of the arithmetic.
We have the power plants. We cannot buy the fuel.
Most discussion of this crisis starts with capacity. That is the wrong starting point.
Our grid-based installed capacity was 28,919MW in May 2026. Counting captive industrial plants and off-grid renewables, it reaches 32,332MW. Estimated peak demand this summer is 18,000 to 18,500MW. On 8 August the system produced only 13,201MW against demand of 15,264MW. We were not short of machines. We were short of fuel to run them.
The immediate cause is clear. Gas supply through the pipeline fell by more than 17 percent after the accident on 22 July at the floating LNG terminal off Cox’s Bazar, operated by Excelerate Energy. But the weakness that turned one equipment failure into a national shortage took fifteen years to build.
IEEFA has calculated that the power sector’s dependence on imports rose from about 5 per cent in FY2009-10 to around 65 per cent in FY2024-25. This counts imported electricity, coal, LNG and fuel oil together. Over the same period, the cost of generating electricity rose more than fourfold. Between FY2020-21 and FY2024-25 alone, average generation cost went from Tk6.61 to Tk12.10 per unit.
The results show up in the accounts. Oil-fired generation costs about Tk27.5 per unit. Even so, oil supplied 10.7 percent of our electricity in FY24-25. In India, the figure is 0.02 percent. Power sector subsidies rose from Tk79.7 billion in FY18-19 to Tk386.7 billion in FY24-25. We now owe power producers around Tk500 billion, and the furnace oil operators have warned that they may have to stop generating.
So, we run our most expensive plants at the worst hours, using borrowed dollars, and we pay capacity charges on the plants we cannot run at all.
A simple way to picture it- let’s use a water-tap analogy
Think of a neighbourhood that gets its water from a row of storage tanks. Each tank has its own tap. Over fifteen years we built a lot of taps (let’s say 32). On paper we have almost twice as many as we need (18).
But a tap only gives what is inside its tank. Our tanks are not filled by rain water. They are filled by a tanker, and the tanker has to be paid in dollars. Every megawatt we added came with a foreign currency bill attached to it. When the pipeline breaks, or the bank cannot open the letter of credit, the tap is still there. There is simply nothing behind it.
This is why building more capacity does not solve the problem. Adding a 33rd tank to a row of empty tanks only adds rent. The committee formed under the interim government to review unsolicited contracts found that inflated capacity charges have cost Bangladesh as much as $1.5 billion a year.
Now, solar and wind power work differently. They are like rain falling on our own roofs. There is no tanker and no invoice. Their weakness is real, but people often overstate it. The rain falls mostly in the middle of the day, while the neighbourhood queues at the tap between seven and eleven at night (evening peak demand). A battery is the tank that stores rainwater and closes that gap. It does not create water. It moves free water to the hour when we would otherwise burn our most expensive fuel. On 8 August, furnace oil plants ran at about 1,700MW on average, but at more than 3,100MW during the evening peak. That evening block is exactly what a storage tank (i.e. battery) is for.
Rooppur is a well not another tank
Rooppur changes this picture, and it is worth being clear about how. It is not another tank. It is a well.
A tank needs refilling by tanker on a schedule we do not control. A well is dug once, at great expense, and then gives a steady flow without a monthly invoice at the gate. Fuel loading at Unit 1 began on 28 April. The first connection to the grid is expected at the end of this month, at about 300MW, with full output of 1,200MW after eight to ten months. For the first time, a large block of our electricity will not depend on the spot price of LNG.
But three points need to be said alongside this. First, we have already committed to the well. The project now costs about Tk1.39 trillion, financed mostly by Russian credit, and those payments fall due in dollars whether the plant runs or not. We have not escaped our dollar problem. We have changed it from a monthly fuel bill into a long-term loan. Second, the output arrives slowly, so nobody should expect 2,400MW this winter. Third, a well flows at one rate. Rooppur will raise the base level of our supply. It will not follow the evening queue at the tap. A single unit of that size also becomes the largest thing that can trip on our grid at any moment (grid failure), which means we need more standby capacity in reserve, not less.
What India did that we did not
The useful comparison with India is not that they built more solar. It is how they buy electricity.
Last week Solar Energy Corporation of India finished a tender for 1,000MW of firm and dispatchable renewable energy, supplied round the clock. These are renewable projects with batteries built alongside them. The winning price was INR5.25 per unit, about 5.5 US cents, under 25-year contracts. Compare that with our own figures. It is far below our oil-fired cost of Tk27.5 per unit and below our average generation cost of Tk12.1. And the price is fixed for twenty-five years, with no fuel risk at all.
The method matters more than the price. India is not buying a technology. It is buying a delivery promise, and letting developers decide what mix of solar, wind and batteries will meet it. Three supports sit around this. The Union Cabinet approved a second round of viability gap funding in May 2026, worth INR5,400 crore, to support 30GWh of storage. The first round in 2023 supported only about 4GWh. Inter-state transmission charges are waived for storage built alongside renewable plants and commissioned by June 2028. And renewable-heavy states now carry rolling storage obligations.
We should copy the method, but we should also read India’s record honestly. Tendered storage capacity there rose from 6.8GW in 2018 to 90.7GW in 2025. Yet only about 1.8GWh of grid-scale batteries were actually installed by March 2026. IEEFA and JMK Research warn that some 2025 bids were priced too low to be viable. The lesson is not that tenders solve everything. The lesson is that open competition on long fixed-price contracts is a better way to buy electricity than unsolicited deals where the consumer carries the fuel price risk. It still has to be delivered.
Imports help only if the rules are clear
Our links at Baharampur-Bheramara and Tripura-Comilla carry around 1,160MW, and the Adani plant at Godda supplies close to a tenth of our daily electricity through a dedicated line.
What that means became clear in June. Nepal sells us 40MW seasonally at 6.40 US cents per unit, wheeled across Indian territory. Both governments agreed in November 2025 to raise it to 60MW. India’s Central Electricity Authority did not approve the extra 20MW, citing transmission limits. The amount is trivial. What it shows about who decides is not.
Britain buys power differently. Nine subsea HVDC links carry close to 10,000MW from France, the Netherlands, Belgium, Norway, Denmark and Ireland, and capacity on them is sold through auctions and market rules. No single neighbour can switch it off. We have one neighbour and one direction of flow. If we raise the import share of our mix without clear transit rules, we will only trade our dependence on LNG for dependence on one transit country.
One line in the budget needs another look
The FY26-27 budget deserves real credit. Import duty, regulatory duty, supplementary duty and advance tax on the main solar components have been cut to zero until June 2031. Solar generation is tax-free until 2035. Consumers get a 5 per cent rebate on solar electricity bills. Our target is 20 per cent renewable electricity by 2030 and 10,000MW of solar. We have about 1,700MW today. IEEFA’s modelling suggests we need to add roughly 760MW every year until 2030. For the first time, the tax system no longer makes clean energy more expensive than imported fuel.
There is one exception. The duty concessions on mounting structures, lithium cells, battery packs and battery storage systems are due to be withdrawn after June 2028. The reasoning is that this will encourage local manufacturing. I would ask the finance adviser to look at that date again. We will not have a domestic battery cell industry by 2028. But 2028 is roughly when solar output will be large enough that the evening peak cannot be managed without storage. It is also the point at which India has chosen to keep its support going rather than end it. We have given the rain a tax holiday and put an expiry date on the tank that stores it.
Fifteen years ago, I helped work out how much oil this country would have to import. I would rather the next group of Bangladeshi engineers spent their twenties calculating something we can afford to pay for.
Dr. Md Rishad Ahmed
Associate Professor at University of Nottingham.
The views expressed here are writer’s personal opinion
The writer is an Associate Professor at the Power Electronics, Machines and Control Research Institute, University of Nottingham. Between 2010 and 2012 he worked as an Associate at Keystone Business Support Company Ltd, Dhaka, on power sector assignments including work for General Electric International, POWER CELL and Total Gaz. Keystone was founded by Dr M Fouzul Kabir Khan, who served as Adviser for Power, Energy and Mineral Resources until February 2026.
