How to select a chemical project when the spread is thin
This is a decision framework for navigating the chemical business through market cycles — how to read a project when margins are thin, pick the right manufacturing route, time the cycle, choose the ground, sequence the money, and know your exits before you enter. Aniline is used purely as a worked example: the same logic applies to any commodity chemical. Every number below recomputes live as you move the sliders, so you can swap in your own case.
The situation — what “thin” means
Aniline here is only the vehicle for the argument — the subject is how to run a chemical business across the cycle. Every commodity chemical is priced by one number: the spread — product price minus the stoichiometric feedstock behind it. For aniline, that is aniline − 0.87 × benzene. Eight months ago it was $819/t. Today it is ≈ $487. That 41% collapse — benzene squeezed, aniline flat — is what “thin” means, and it is exactly when most committees make their worst decisions: they either freeze completely, or they chase whatever still looks profitable on last year’s screen.
Set the market — drag the squeeze yourself
Aniline fell 10%; the spread fell 41%. Every producer sells the identical molecule at the identical price — some are printing cash, some are dying, and the screen will never tell you which. Nobody sets the price; the marginal producer does. So a thin spread is not a reason to stop deciding. It is the moment the decision sequence below exists for.
The sequence this manual walks, in order: learn to read an NPV the way a process engineer reads a heat balance (01) → use it to pick the manufacturing route (02) → place that route on the cycle and set your triggers (03) → choose the ground (04) → sequence the money so the risk is staged (05) → watch the trajectory turn into secured returns (06) → and write the assumptions and exits down before anyone signs anything (07).
First, learn to read an NPV
This is where engineering arguments die: you prove the plant is better, finance says no, and you don’t know why. So learn the language first — it takes four minutes, because you already know all of it under different names.
NPV is a mass balance on cash. In = out + accumulation. The one difference: a tonne of benzene is the same tonne in 2026 or 2031 — a rupee is not. Cash has a time dimension mass does not, and NPV is the balance done with that dimension included.
The discount rate is a driving force. Heat flows down a temperature gradient; money flows down a return gradient. Your WACC — weighted average cost of capital, the blended return your lenders and shareholders demand — is the ambient temperature of your money. A project earning below it is a heat exchanger running backwards: you are pumping heat uphill and paying for the compressor.
Discounting is first-order decay. PV = FV·e−rt — the same algebra as catalyst deactivation. The discount rate is a deactivation constant for money.
The annuity factor is a residence time. A plant running N years at cash flow C is worth C × AF today, where AF = [1 − (1+r)−N]/r. And put the capital charge on the datasheet: capex × ~17%/yr ÷ annual tonnes is a cost line like steam — on aniline it is bigger than the utility bill.
Finance inputs
At 12%, money’s half-life is 6.1 years — the second decade is worth a quarter of the first, and a 3-year delay destroys 29% of the whole project. And NPV = 0 does not mean “break even” — it means the shareholder does as well in a bond. So when the spread is thin, stop asking “is this a good project?” Ask “at what spread does it become one — and what must I own today to move the moment it does?” That converts a yes/no into a trigger. The rest of this manual is the machinery around that trigger.
Choose the manufacturing route — batch or continuous
The route decision is physics first, money second. Physics: an autoclave’s heat generation grows with volume (D³) while its cooling grows with area (D²) — so batch has a thermal ceiling, dilutes with solvent to live under it, and pays for that solvent on every line of the cost sheet. Money: the full head-cost comparison, live, below.
Autoclave thermal inputs
A neat-charged 10 m³ autoclave generates 2.02 MW against 1.60 MW of cooling — over its limit. The solvent is not chemistry. It is thermal ballast — heated, cooled, distilled and recovered, six thousand times a year. Every batch cost line descends from this.
The head cost sheet, live
Three plants. A — continuous, fully integrated (own nitric acid, own hydrogen, own nitrobenzene). B — the same integration in autoclaves (nobody builds it; it isolates the reactor). C — what India actually runs: batch, buying merchant nitrobenzene.
Feedstock & capital inputs
| Head cost, $/t aniline | A · Continuous integrated | B · Batch integrated | C · Batch, buys NB |
|---|---|---|---|
| Benzene / nitrobenzene feed | 914 | 945 | 1,159 |
| Nitric acid (captive, ex-ammonia) | 120 | 123 | — |
| Hydrogen | 122 | 131 | 158 |
| Nitration section opex | 89 | 105 | — |
| Catalyst & chemicals | 15 | 42 | 42 |
| Power & cooling water | 16 | 61 | 61 |
| Effluent & waste | 4 | 22 | 22 |
| Steam | −21 exported | +68 bought | +68 bought |
| Labour · maintenance · other fixed | 40 | 68 | 60 |
| CASH COST | 1,299 | 1,566 | 1,570 |
| Capital charge @ 17% | 134 | 78 | 68 |
| FULL COST | 1,434 | 1,644 | 1,638 |
| Cash margin at current aniline price | +101 | −166 | −170 |
Columns B and C land on the same number ($1,566 vs $1,570). Integration only pays if the whole train is continuous. A batch nitration section is nearly as bad as paying the merchant’s margin — India’s batch sector buys its nitrobenzene, and it is economically right to.
The route verdict — read it off your own inputs
Continuous · integrated
+$101/t cash margin now SURVIVES THE TROUGHNeeds spread ≥ $492 to earn its capital. Sanction into the trough, 36–48 months to product. Sells tonnes; wins by cost.
Batch · merchant NB
−$170/t cash margin now CASH-NEGATIVECash breakeven at spread $657; earns capital only above $710. Commissions in 12–18 months — built to chase peaks, and only peaks.
When the spread is thin, the route question answers itself: only the low-cost route can be sanctioned at the bottom — at today’s $487, continuous is cash-positive and batch is bleeding. Batch is not therefore useless: it is the correct instrument for young, small, multi-product molecules chased at the top of the cycle on a 12–18 month build. The selection rule is one line: commodity tonnes at the trough → continuous, at scale, integrated. Specialty value at the peak → batch, small, fast. Never the reverse — and never a small continuous plant, which is a batch cost structure bought with a continuous cheque.
Navigate the cycle — the knife edge and the playbook
Now put 01 and 02 together. Plot each route’s NPV against the spread; find where each crosses zero; mark where today sits. The chart is the whole strategy — and below it, the playbook: what a disciplined owner does in each band of the spread, with the current band lit.
The cycle playbook — where are you, and what do you do there
Below everyone’s cash cost. Nobody builds. Buy distressed assets and consents — this is M&A season, not capex season.
Cash-positive, NPV-negative. Buy land, file EIAs, tie up the berth, sign the ammonia MoU. Own everything that takes 18 months and costs little.
Continuous clears its hurdle while the crowd still sees losses. Sanction the continuous complex now — into the tail of the trough.
Continuous prints; batch still underwater. Debottleneck, expand, lock long offtakes at prices that look mediocre and will look brilliant.
Batch FIDs appear everywhere. You sell — capacity, offtake, maybe equity. Never buy here. This band is where the next trough is manufactured.
BANDS RECOMPUTE FROM YOUR INPUTS · HIGHLIGHT = TODAY’S SPREAD
Batch chases the peak; continuous is sanctioned into the trough. A 36–48 month build means the only way to own capacity in the recovery is to sign in the option zone or the FID window — which is why the pre-permitted site in 05 exists. Sanction when it looks obviously attractive (the euphoria band) and you commission into the next trough, every cycle. At today’s NPV −33 ₹ Cr, the correct act is not “build” and not “wait” — it is “own the trigger”: the land, the clearance, the berth agreement, so that the day the spread crosses $492, you are 18 months ahead of everyone who is only now calling their consultant.
Select the location — Weber, then the four conditions
Weber’s material index (1909) = weight of localised inputs ÷ weight of product. Above 1.0 the process loses weight — locate at the feedstock. Aniline: nitric acid ships at ~65%, so 0.71 t of acid is 1.09 t of tanker — MI = 2.03 buying acid, 1.23 making it. Captive acid cuts inbound freight 40% for the same reason it cuts cost $221/t. Both ledgers point the same way; when they agree, the answer is real.
Freight inputs
Paradip by sea costs $17/t more than Dahej in freight — and buys $221/t of captive nitric acid. By road it costs $76/t more and half the advantage is gone. Odisha is a port cluster, or it is nothing. The four conditions no other Indian site will hold simultaneously: benzene at the fence (IOCL Paradip aromatics) · ammonia by rail (Talcher, ~180 km) · a non-cyclical acid customer (the mining belt’s ammonium nitrate) · a deep-water port. Dahej has 1½. Paradip will have four.
Sequence the investment — options first, base-load second, cyclical last
Never buy the whole thesis in one cheque. Stage it so each tranche is protected by the one before it — and price the first tranche as what it is: an option.
Option & sequencing inputs
| Phase | Action | What protects it |
|---|---|---|
| Phase 1 | Buy + fully permit the plot. File the EIA now. | Gate 0 kill switch: IOCL Paradip petrochemical FID. No aromatics → sell the land, exit. |
| Phase 2 | Nitric acid, 400 TPD, on Talcher ammonia. Not aniline. | Mining-belt AN demand — local, contracted, uncorrelated with benzene. Gate on Talcher commissioning. |
| Phase 3 | Sanction the continuous aniline complex — in the option zone / FID window. | The acid’s steam, acid and cash. The berth agreement. The pre-held EC. |
| Phase 4 | Aniline onstream into the recovery. Then migrate ageing batch molecules onto the site. | Commissioning into a rising spread — the entire point of the sequence. |
The land is the only line where being wrong is cheap (₹70 Cr) and being late is fatal (₹386 Cr — 3.9× the land price). The acid goes second because a cluster is seeded by a net producer of utilities and because its customer ignores the petrochemical cycle — it is the base-load that pays the bills while you wait. The aniline goes last because it is the call option: enormous when the spread recovers, worthless to build before it. Options → base-load → cyclical. Every tranche is underwritten by the one before it.
The trajectory — how it materialises into secured returns
Run the sequence forward, year by year. Grey bars are capex going out. Green is the secured layer — acid EBITDA against mining-belt demand, indifferent to benzene. Blue is the cyclical layer — aniline at your recovery spread. The amber line is cumulative cash: where it crosses zero is your payback.
Not a hedge-fund guarantee — a structure. By 2029 the acid is earning ₹95 Cr/yr against demand that does not read the benzene screen; it services the debt and keeps the site warm whether the aniline spread recovers in 2031 or three years later. The aniline layer is deliberately unsecured — it is the option you bought all this to hold — and if the recovery slips, you slip Phase 2 with it and lose carrying cost, not capital. Drag the recovery year to 2033 and watch: the drawdown barely deepens, because the sequence refuses to spend the big money before the trigger fires. That refusal is the security.
Core assumptions — and the exit plan
A plan that cannot state the condition under which it stops is not a plan; it is a pitch. Here is where this one stops, at each gate, with the money you walk away with.
| Gate | Trigger to proceed | If it fails — the exit | You walk away with |
|---|---|---|---|
| 0 · Aromatics | IOCL Paradip petrochemical FID | Sell the plot — notified PCPIR land beside a refinery holds value | −₹70 Cr |
| 1 · Ammonia | Talcher commissioning, not announcement | Defer acid or run merchant NH₃; land keeps appreciating | −₹70 Cr worst case |
| 2 · The berth | Chemical berth + coastal parcel service committed | Shelve aniline permanently; keep the acid — a bond-like business | acid NPV +₹280 Cr |
| 3 · The spread | Spread ≥ continuous breakeven, held one quarter | Hold at Phase 1 indefinitely; acid pays the carry | acid running + option live |
| Post-2031 | — | Operating integrated complex: strategic sale, JV, or listing — the exits multiply once it runs | market |
The assumptions the whole thing rests on
Thermodynamics (unarguable): hydrogenation −463.6 kJ/mol from heats of formation → 4.98 GJ/t; nitration ≈ 1.26 GJ/t; total 6.2 GJ/t of aniline. Attackable: autoclave U = 500 W/m²K and ΔT = 80 K (push them with the sliders — the ceiling moves, the conclusion doesn’t); merchant NB at $840/t, which currently flatters batch — drag it to $950 and watch column C; the ~90 kTPA minimum scale behind captive nitric acid (an economic 60 TPD ammonia-oxidation train would change this conclusion — if someone can cost one, I want to see it); acid EBITDA margin vs AN netback, which you should mark to a real offtake term-sheet before Phase 1. Finance: cash margins only, no terminal value, no tax shield — deliberately austere; adding them improves every number above.
Paradip sits on the Bay of Bengal belt — Fani, Amphan, Yaas. Going east trades Arabian Sea flood exposure for cyclone exposure; the derisking here is regulatory, feedstock and technological — not climatic. Elevated plinths, cyclone-rated structures, 72-hour utility autonomy. Saying this plainly is what earns the right to be believed about everything else.
Planning a chemical project in Odisha?
Whether you are choosing a site, sizing the first plant, or deciding how to move through the cycle — a short conversation can save a lot of capital. I help with buying industrial land in Odisha and with setting the project and market-cycle strategy.
Let’s talk about your project
Tell me a little about your project — I help with buying industrial land in Odisha and with selecting a strategy for chemical projects across the market cycle.
Your details are sent directly to me and treated in confidence.