EV charging project feasibility

Module 01

What you are actually deciding

Almost every bad EV charging investment we see began with a single question: “is this a good business?” That question cannot be answered, because it is three questions wearing one coat.

These have three different answers and they routinely disagree. A site can be operationally profitable and still fail to clear your hurdle. It can clear your hurdle over ten years and still be unfundable because year two is thin. Any document that gives you one verdict has silently chosen which question to answer.

The order that saves money

Work outside-in, cheapest test first. Geometry and electrical feasibility before demand; demand before economics; economics before financing; financing before documents. Each stage can kill the project, and every stage you skip is one you will pay to discover later.

That ordering is the entire logic of the tools this course uses, and it is why screening is free: there is no sense charging for a test whose main value is telling most people to stop.

Module 02

Screening a plot before you fall in love with it

The first question is not commercial. It is whether the physical plot can hold the installation you have in mind and get the power to run it.

What a screen can settle

• Geometry — whether the bays, turning circles, queuing space and equipment footprint physically fit, and what is left over once they do.
• Electrical sizing — what load the configuration implies.
• Transformer selection — what plant that load requires.
• What is already there — existing chargers nearby, from the station registry.

GeoSite does this for a specific plot, free and without an account, and returns a go/no-go with a single-line diagram and layout sketch. The screening is free and it is final if the answer is no — a plot that cannot hold the layout does not become viable further down the funnel.

What a screen cannot settle

It is a geometry and sizing tool, not a site visit. It does not know your soil, your drainage, your neighbour’s objection, whether the DISCOM has capacity at that feeder, or whether the land title is clean. Those need measured inputs. Treat a pass as permission to spend the next increment of effort, not as validation.

Space matters for a reason that is easy to miss: it is where accessibility lives. Sites that are marginal on space are where accessible bays quietly get designed out.

Module 03

The demand assumption, and why it is where projects die

Every projection you will be shown rests on one number: how much energy the site sells. It is almost always the least evidenced number in the document, and it is the one a credit officer will find first.

The convenient default

Vendor quotations routinely build the revenue case on an assumed utilisation of around three hours of charging per day. It produces a satisfying payback. In our experience modelling supplied quotations against lender criteria, that basis is usually not bankable — not because the arithmetic is wrong, but because the single input the case rests on has nothing measured behind it. That is an advisory observation from the files we have read, not a survey.

Building a defensible basis

A demand assumption is defensible when you can say where it came from. In descending order of strength: measured traffic or throughput at the site; observed throughput at a comparable site you can name; a registry-based view of what exists nearby and how saturated it is; a stated policy or fleet commitment; and last, a sector average. Most documents use the last and present it as the first.

Our opportunity maps show where demand appears under-served across states and UTs, and the methodology page states plainly what the index does and does not establish — including where its own backtest failed. Use it to generate candidates and to argue about relative attractiveness, not to assert a throughput figure.

The test to apply to any number you are given

Ask what the verdict becomes if throughput is 30% below plan. If the answer is materially different and nobody has shown you that case, the document is incomplete.

Module 04

The rulebook that caps your margin

This sector has an unusual feature: your margin per unit is capped by regulation. That single fact reshapes the entire business case, because it removes the lever most business plans reach for when returns look thin.

What the national guidelines set

Under the Ministry of Power’s 2024 guidelines for EV charging infrastructure:

• the supply tariff to public charging stations is capped at the Average Cost of Supply until 31 March 2028;
• the service charge — your margin over the cost of electricity — is subject to ceilings: ₹3/kWh (AC, solar hours) and ₹4/kWh (AC, non-solar); ₹11/kWh (DC, solar hours) and ₹13/kWh (DC, non-solar);
• where public land is used, a revenue-share reference of ₹1/kWh applies.

Separately, EV battery charging is de-licensed nationally when the applicable technical, safety and operating requirements are met — you are not applying for an electricity licence, which is the single most common misconception about entering this sector.

What it does to the AC/DC choice

On the non-solar figures, DC permits ₹13 against AC’s ₹4 — more than three times the margin on the same unit of energy. That does not make DC the answer: DC needs far more load, far more capital and a different dwell pattern. It does mean the two configurations are different businesses rather than different speeds, which our article on AC or DC works through.

Note the date. The ACoS cap runs to 31 March 2028. A model whose viability depends on it should say what happens afterwards, rather than projecting it silently to year ten.

Module 05

What a unit of electricity actually costs you

The energy charge is the number everyone quotes and the one you almost never pay. Getting from the published rate to a landed cost is the most common place a model goes quietly wrong.

Four things to assemble

1. Every variable line, not just energy. Maharashtra’s MSEDCL LT VIII category publishes energy at ₹7.73/kVAh and wheeling at ₹1.52/kVAh, with a stated total variable of ₹9.26/kVAh. Quoting ₹7.73 as “the tariff” understates cost by about a fifth.
2. The unit. kWh bills real energy; kVAh bills apparent energy, which charges you for power factor as well as consumption. Telangana’s ₹6/kWh and Maharashtra’s ₹9.26/kVAh cannot be compared directly, and converting at parity always understates the kVAh side.
3. The standing charge. Telangana’s LT-IX carries nil fixed charge while its HT-IX carries ₹100 per kVA per month — payable whether or not you sell anything. Goa charges on both routes. Haryana’s schedule leaves the cell blank, which is an unknown, not a zero.
4. Time of day. Telangana applies flat rupee adders, Madhya Pradesh symmetric percentages, Maharashtra seasonal percentages. Same phrase, three mechanisms.

Fuel and power-purchase adjustments, electricity duty and local cess sit outside the schedule entirely and move independently. They are real money and they are not in the table you read.

The full method is in how to read a DISCOM tariff order, and the connection-voltage decision it feeds — including the arithmetic that settles LT versus HT, and the state where HT is dearer — is in LT or HT.

Module 06

Capital cost, and what quotations leave out

Capex is the number people believe they already know, because a vendor has given them one. The quotation is usually accurate about what it covers. The problem is scope.

The categories that go missing

• Connection and augmentation. Security deposit, service-line and development charges, and any network augmentation are licensee-specific, load-dependent and quoted only against a real application. They are frequently the largest single surprise.
• The safety scope. Madhya Pradesh’s published requirements give a realistic inventory: in-cabinet clean-agent gas fire suppression on the main panels, transformer safety per CEA regulations, a fire suppression system with an automated emergency button linked to the nearest fire department, CCTV, and spatial requirements under the state building rules. Count how many appear in a typical quotation.
• Civil and site works. Pads, drainage, lighting, signage, access, bollards and the surfacing an accessible route needs.
• The interval before energisation. Between delivery and permission to charge you hold depreciating assets and usually service debt on a site earning nothing. That is a real cost even though no one invoices you for it.

A labelled reference case

So the scale is not abstract, here is one real franchise offer supplied to us. It is one deal, not a market average — treat it as a scale check only: a 320 kW DC cluster at ₹38,00,000 ex-GST plus 5% GST, on a 60-month term, with a ₹57,000/month minimum guarantee against a 22,800 kWh/month threshold, ₹2.60/kWh above the threshold, 50% buyback after five years and maintenance included.

What is instructive is not the headline but what happens when it is modelled properly, which is the subject of the next two modules and of our article on evaluating a franchise offer.

Module 07

Running the model: three verdicts, not one

Now the three questions from module one get three answers. The ROI calculator computes them with every assumption visible and adjustable, which is the property that matters — a model you cannot flex is a brochure.

Operational profitability

Revenue less operating cost, before financing. Energy at your landed cost, service charge within the ceiling, plus rent, power, staffing, maintenance, connectivity, payment charges and insurance. If this is negative, nothing downstream can rescue it.

Investment return

NPV, IRR and payback against your hurdle rate. The hurdle is a genuine input, not a convention: the same project is a good investment at 10% and a poor one at 18%. State it before you see the answer, not after.

Debt serviceability

Covered in the next module, and the one most likely to be missing entirely.

Ownership or franchise

The calculator handles both, and they are genuinely different structures rather than the same project with different paperwork. A franchise moves some risk and most upside; whether that trade is good depends on numbers you can now compute rather than on the strength of a pitch.

Module 08

Bankability: the ratio that decides funding

A lender is not asking whether EV charging is a good sector. They are asking one arithmetic question, and most projections fail it.

Debt Service Coverage Ratio

Take the cash the project generates in a year, after operating costs but before loan repayments. Divide by everything owed on the loan that year — interest plus principal. At 1.00 the project pays the bank exactly, with no room for a bad month. Below 1.00 it cannot pay from its own earnings.

Why coverage collapses faster than revenue

Revenue in this business is close to linear in throughput while debt service is fixed. So a 30% shortfall in throughput does not produce a 30% fall in coverage — it produces a much larger one, because the fixed obligation does not move. This is why the demand basis from module three is the input a credit officer tests first.

You can run this on your own project without buying anything: the bankability check returns the coverage verdict against the benchmark above, with the warnings that produced it, with no sign-up and no charge. Our strong advice is to run it on your vendor’s own numbers first, unchanged. If it comes back short, you have learned the most valuable thing available at this stage — before committing capital rather than after.

The reasoning behind the benchmark is set out in will a bank fund your EV charging station?

Module 09

Sensitivity, and the honest base case

A single-scenario model is an opinion. A lender will run sensitivities whether or not you do; the only choice is whether you see the results first.

What to flex, and roughly in this order

1. Throughput — the input with the least evidence and the most leverage. Test a 30% shortfall as standard.
2. Landed energy cost — tariff revision, duty, and the time-of-day mix actually realised rather than assumed.
3. Commissioning date — a two-quarter slip moves revenue right while leaving debt service where it was.
4. Capex — particularly connection, augmentation and the safety scope.
5. Financing terms — rate and tenor, since coverage is a function of both.

Two rules that keep a base case honest

The question sensitivity is really answering

Not “what is the worst case?” but “what would have to be true for this to fail?” If the answer is something plausible — a competitor opening nearby, a tariff revision, a six-month delay — you have found the thing to manage, which is more useful than a number.

Module 10

Turning a verdict into a document a lender can act on

A verdict is not a document. When you go to a bank you need the reasoning, not the conclusion, and in a form a credit committee can read.

What a bank-ready DPR contains that a quotation does not

• A stated, defensible demand basis — where the throughput assumption came from, not merely what it is.
• Year-by-year DSCR, for every year of the loan, not a single blended figure.
• Sensitivity analysis — the cases from the previous module, run and shown.
• The regulatory position stated correctly, including that the service charge is capped so the case cannot be rescued by assuming higher pricing later.
• Separated verdicts — operational, investment and debt, answered independently.
• Honest labelling of what is unverified, so the reader knows which numbers are measured and which are assumed.

Our Bank-Ready DPR produces this from your own figures, with financial annexures downloadable as CSV. It comes in two forms — one generated on the page from your inputs, and a higher tier reviewed and refined by us before submission. Current pricing is on the pricing page.

Module 11

Approvals, and the cost of time

Everything above assumes the site opens. Three separate approval queues decide when, and they are held by three different bodies.

The published statutory connection clocks vary enormously — 3 days in a metro area under the JERC supply code that applies in Goa and Puducherry, against 30 working days in Haryana where supply is feasible from the existing network, and 97 working days there for a 33 kV connection. Every one of them runs from receipt of an application complete in all respects, which means the elapsed time you actually experience is mostly the part before the file is accepted as complete. That part is yours to control.

One trap deserves a place in your model: under that same supply code, where recorded maximum demand exceeds sanctioned load the billing cycle is computed on actual recorded demand, and exceeding it three times in a financial year triggers a revision of the sanction. Fast charging is spiky, so under-declaring sanctioned load to reduce a standing charge has a short life.

The detail is in from application to energisation and the approvals that stop you switching on. State-by-state authorities and timelines are in the EV charging rules reference.

Module 12

Where to take this

If you have worked through the modules in order, you now have the shape of a defensible case: a plot that passes a physical screen, a demand basis you can source, a landed energy cost with its attributes recorded, a capital plan with the usually-missing categories in it, three separate verdicts, and a sensitivity set that shows what would have to be true for it to fail.

The tools, in the order this course used them

• GeoSite — geometry, electrical sizing, transformer selection and a go/no-go for one specific plot. Free, no account.
• Opportunity maps and station registries — where demand appears under-served, and what already exists.
• ROI calculator — the economics, with every assumption visible and adjustable, for ownership and franchise structures.
• Bankability check and Bank-Ready DPR — the coverage verdict free, and the document when you need one.

The reference material

• EV charging rules — state-by-state tariffs, approvals, connection routes and subsidies, each linked to an archived official source.
• Source register — every document we cite, with authority, retrieval date and checksum.
• Articles — the individual decisions worked through in depth.

And the other course

This course covers whether to build and what to build. The installation course covers how it actually gets built — seventeen modules written for the person holding the crimping tool. They answer different questions and neither substitutes for the other.