Concept

Discounted cash flow (DCF)

Overview

Discounted cash flow (DCF) values a project by projecting its net cash flow in every year of its life and discounting each back to a present value, summed into a net present value (NPV). It’s the time-resolved alternative to annualizing capital with a capital recovery factor: where the CRF assumes one up-front outlay and flat operation, a DCF carries the actual year-by-year timing of spending, production, and prices.

Body

What it is. Lay out, year by year, the net cash flow CFₜ — revenue − operating cost − capital spend − tax, plus end-of-life salvage — and discount each to the present at the discount rate i (the cost of capital):

NPV = Σ  CFₜ / (1 + i)ᵗ
       t

Positive NPV means the project beats the discount rate; IRR is the i where NPV = 0; payback is when cumulative discounted cash flow turns positive — all outputs of the same projected table.

What it captures that a CRF can’t. Resolving each year separately represents timing a single annualization smears flat: staged capital, a multi-year ramp, changing prices, mid-life replacements, salvage, and tax/depreciation schedules. It can also reflect the financing structure (split and timing of debt and equity).

Relation to the CRF. The CRF is the steady-state special case: for a single up-front capital and flat cash flow, annualizing with CRF and forming a levelized cost ranks projects identically to computing NPVs. The DCF generalizes the same discounting to arbitrary, non-uniform timelines.

What it produces. Questions of value and return over time — NPV, IRR, payback, the effect of phasing — rather than the steady-state cost-per-unit a levelized cost reports. Complementary: levelized cost asks “what does a unit cost to make,” a DCF asks “what is the investment worth, and when.”

Limits & typical error

• More structure is more assumptions, not more accuracy — a price path, ramp profile, tax schedule, and terminal value are each a guess; if no better grounded than a CRF estimate’s inputs, the elaborate table just delivers false precision.
• The timing assumptions tend to dominate — NPV is highly sensitive to the discount rate and the terminal/salvage value; a speculative ramp or price path can move NPV more than any physical cost.
• IRR has well-known traps — streams that change sign more than once can have multiple IRRs or none, and IRR implicitly assumes reinvestment at the IRR itself, so ranking by IRR can mislead where NPV wouldn’t.
• Real-vs-nominal consistency must hold throughout — as with the CRF, a nominal rate needs nominal cash flows and a real rate real ones; mixing biases every term.
• It adds no physical accuracy — a finance overlay on the same cost and performance estimates; it re-times and values them but can’t make an uncertain capex or yield more correct.

Mini-example

For the steady green ammonia plant, a DCF adds little. With capital spent essentially up front and output, prices, and costs roughly constant, discounting each year and solving for NPV gives the same per-tonne economics as annualizing total capex with the CRF (the ~$355/t capital share) and adding opex — the flat timeline is exactly what the CRF shortcut is built for.

Edge case: a phased build — electrolyzer capacity added in tranches as offtake grows, output ramping toward nameplate — breaks the flat-timeline assumption, and only a DCF resolves the early, capital-heavy, revenue-light years a single CRF charge spreads uniformly. That’s where the year-by-year detail earns its cost.

See also

• Capital recovery factor (CRF)
• Cost of capital
• Total capex (battery-limits → total plant)
• Levelized cost (e.g., $/t NH₃)
• Capex
• False precision