Techno-economic assessment

A complete TEA has four interlocking components. Each can be done well or poorly, and a weakness in any one component propagates through the model into the final result.

Process engineering and mass-and-energy balance. The TEA starts from a description of what the plant does: what comes in, what comes out, what conversion is happening in between, and at what efficiency. For a mature technology with established commercial reference plants, the mass-and-energy balance can be lifted from published data. For an early-stage technology, the balance is constructed from process simulation tools (Aspen Plus, Aspen HYSYS, PRO/II, and others) using laboratory or pilot data extrapolated to commercial scale. The extrapolation is the principal source of risk at this layer. Pilot-scale yields, selectivities, and utility consumptions do not always translate cleanly to commercial scale, and the extrapolation method should be disclosed and defensible.

Capital cost estimation. The total installed cost of the plant is built from equipment costs (the inside-battery-limits scope), plus supporting infrastructure (the outside-battery-limits scope including utilities, storage, civil works), plus indirect costs (engineering, procurement, construction management, owner's costs), plus contingency. The methodology used to estimate these depends on the maturity of the project. Early-stage estimates use factor methods (Lang factors, Hand factors, Guthrie correlations) where the equipment cost is multiplied by a factor reflecting typical industry-wide ratios of equipment to total installed cost. Later-stage estimates use bottom-up methods with vendor quotations for major equipment, location-adjusted construction labour rates, and quantified bulk-material takeoffs. The accuracy of the estimate depends on the methodology and on the maturity of the engineering deliverables that support it.

Operating cost and revenue estimation. The operating cost includes feedstock, utilities, labour, maintenance, catalyst replacement, insurance, and other recurrent costs. The revenue includes the product, byproducts, and any policy support (carbon credits, mandates, grants). Both sides are sensitive to assumptions that the analyst chooses: the long-term price of the feedstock, the long-term price the product will clear at, the operating hours per year (a function of plant availability and market demand), the cost of policy support over the project life. Operating cost and revenue assumptions are the principal lever for moving a TEA's conclusion, and they should receive the most scrutiny in any independent review.

Financial modelling. The capex and the annual operating cash flows are combined into a project financial model, typically a discounted cash flow analysis over a 15 to 25 year horizon, that produces the metrics decision-makers use. Net present value, internal rate of return, levelised cost of product, payback period, debt service coverage ratio, and similar metrics emerge from this layer. The financial assumptions (discount rate, debt-to-equity structure, depreciation method, tax treatment, terminal value) are the layer of the TEA that finance professionals usually focus on, and they can shift the headline metrics substantially with reasonable variations within the defensible range.

These four components are linked. A change in the process engineering layer (an improvement in yield, for example) cascades through the operating cost into the financial metrics. A change in the financial layer (a tighter cost of capital) shifts the headline metrics without any change to the underlying physics. The TEA model is the structured machinery that propagates these changes through to the final numbers, and the value of having a model at all (rather than back-of-envelope arithmetic) is the disciplined propagation of consequences across layers.

The single most useful skill in working with TEAs is the ability to recognise the recurrent failure modes that produce confident but wrong answers. The failure modes are familiar across the industry, and they recur because the structural incentives that produce them have not changed. There are five worth naming explicitly.

False precision. A TEA model with four-decimal-place outputs derived from inputs known to the nearest order of magnitude is presenting a degree of certainty that the underlying data does not support. The appearance of precision tends to anchor decisions on the headline number, even when the analyst knows, and sometimes explicitly notes, that the headline is materially uncertain. The remedy is to insist on ranges rather than point estimates for the principal uncertain inputs, to propagate those ranges through the model rather than presenting only the base-case output, and to be honest in the executive summary about what the model can and cannot answer.

The FOAK-NOAK gap. Many TEAs implicitly or explicitly assume nth-of-a-kind cost performance, meaning the cost level achievable after the industry has built many plants and the learning effects have worked through. First-of-a-kind plants, the actual plants being built today in most emerging energy technology categories, are typically 1.5 to 4 times more expensive than the eventual NOAK plant, with the multiplier depending on the technology maturity, the supply chain readiness, and the integration complexity. A TEA that quotes NOAK economics for a FOAK project, without explicit treatment of the gap, is presenting a future-industry scenario as a current-project assessment. This is one of the most common, and most consequential, errors in early-stage energy project TEAs, and it is responsible for a substantial fraction of the cost-overrun history that the industry now carries.

Optimism bias. The academic literature on megaproject delivery, principally the work of Bent Flyvbjerg and collaborators, documents persistent and substantial cost and schedule overruns across project categories: rail, dams, mines, energy plants, and most other large infrastructure. The overruns are not random; they are systematically biased upward, meaning realised costs exceed estimates much more often than they fall short of them. This optimism bias is not the fault of individual estimators acting in bad faith. It is a structural feature of the project development process, in which projects compete for capital on the basis of their projected economics, and the projects that present the most favourable economics tend to get funded. The selection pressure is towards optimism. Good TEA practice acknowledges this directly, often by applying an explicit optimism-correction factor based on outturn data from comparable projects.

Sensitivity versus scenario confusion. A standard sensitivity analysis in a TEA varies each input one at a time by some symmetric range (typically plus or minus 10 or 20 percent) and observes the effect on the output. This is useful for understanding which inputs the model is most sensitive to. It is not, however, a treatment of risk. Real-world inputs are correlated. A recession that depresses product prices typically also depresses feedstock prices and capex. A policy change that affects carbon credit value typically also affects mandate-driven revenue. A proper scenario analysis groups correlated inputs into coherent scenarios (a high-renewable-buildout scenario, a slow-transition scenario, a high-policy-support scenario, a low-policy-support scenario) and runs the model under each. Many TEAs stop at sensitivity and never get to scenario, which means they understate the realistic range of project outcomes.

Scope drift between TEA and execution. A TEA is prepared with a defined scope: which plant elements are included, which are assumed to be provided by others, which are deferred to later phases. The scope of the actual project that gets built may differ from the scope of the TEA, sometimes substantially. Utilities that the TEA assumed would be provided by an adjacent host site turn out to need to be built. A water supply assumed to be cheap turns out to require a treatment plant. A grid connection assumed to be straightforward turns out to involve substantial upgrade contributions. Each of these scope drifts is an addition to the actual project cost that does not appear in the TEA. The remedy is rigorous scope definition at the TEA stage, explicit identification of what is excluded and the assumption that backs the exclusion, and disciplined change management when the project moves into execution.

These five failure modes are familiar to anyone who has worked in industrial project development for any length of time. They are not exotic risks. They are the normal, predictable, recurrent reasons that projects come in over budget, on optimistic timelines, with disappointing economics. A TEA that has been prepared with active attention to these failure modes is significantly more useful than one that has not. A TEA that is silent on them should be assumed to be subject to them.

A project does not have one TEA. It has a series of TEAs, prepared at different stages of project maturity, with different purposes and different levels of confidence. Understanding where in the lifecycle a TEA sits is necessary for interpreting it correctly.

In the concept stage, the project exists as an idea, perhaps a block diagram on a slide. The TEA is Class 5, prepared by a small team in a few weeks, with the purpose of screening the concept against the most obvious commercial gates. Does the project plausibly have positive economics under reasonable assumptions? If the answer is clearly no, the concept is killed or restructured. If the answer is plausibly yes, the project moves forward.

In the pre-feasibility stage, the project is being developed but not yet committed. The TEA is Class 4, prepared over several months, with substantial input from process engineering, business development, and finance. The purpose is to test whether the project warrants the substantial investment required to proceed to feasibility study. The output supports a stage-gate decision, often called "FEL-1 to FEL-2" or "Gate 1 to Gate 2" depending on the company's terminology.

In the feasibility stage, the project is being developed in earnest. The TEA is Class 3, prepared by a substantial team over six to twelve months, with input from external technology providers, EPC contractors, and financial advisers. The purpose is to support a budget authorisation decision: should the project proceed to FEED engineering and pre-construction work? At this stage, the TEA starts to be supplemented by external review: independent technology assessments, financial adviser reviews, and lender's engineering reviews.

In the FEED stage, the project is approaching final investment decision. The TEA is Class 2, prepared with substantial engineering input and firm vendor quotations for major equipment. The purpose is to support the FID itself. The TEA at this stage is typically subject to extensive external due diligence by lenders, equity investors, and offtakers.

After FID, the TEA's role shifts. It becomes a baseline against which actual execution is measured. Change orders, scope adjustments, and cost overruns are tracked against the FID TEA. The project's economic performance over its life is measured against the FID assumptions, and any material deviation triggers governance review.

This lifecycle pattern is well-established but routinely violated in practice. Common dysfunctions include TEAs labelled at one class but built on engineering of a lower class; TEAs that get more optimistic as the project approaches FID rather than more conservative; TEAs prepared for fundraising purposes presented as if they were independent assessments; and TEAs from the early stages being relied on for execution decisions for which they were never intended. Disciplined project development requires disciplined TEA practice across the lifecycle, with the maturity of the TEA tracking the maturity of the engineering and the maturity of the commercial structuring.

The discipline of techno-economic assessment is being reshaped by three forces that will define its practice over the next decade.

First, the proliferation of first-of-a-kind energy technologies. The energy transition involves the deployment of many technologies that are commercially novel at scale: large-scale electrolysis, integrated Power-to-X projects, advanced biofuels, carbon capture at scale, direct air capture, advanced nuclear. Each of these is currently being built or proposed in projects whose TEAs are being prepared against a thin or absent reference base of operating commercial-scale plants. The risk of FOAK-NOAK confusion is therefore particularly acute. The practice of TEA in this environment requires more explicit treatment of technology maturity, more disciplined use of analogous historical cost data from related technology categories, and more honest disclosure of the uncertainty bands. The TEAs being prepared for emerging energy projects today will be the historical record against which actual outturn is judged over the next decade, and the discipline applied to them now matters for the credibility of the discipline going forward.

Second, the integration of carbon and policy accounting. Modern TEAs for energy and climate projects involve not just commodity prices and capital costs but also carbon credits, mandate-driven revenue, lifecycle emissions accounting, and certification compliance. These revenue and cost lines are policy-dependent, are subject to change, and are fragmented across jurisdictions. The TEA practice required to handle them well is structurally different from the practice that handled the same projects in a fossil-fuel world. Carbon-revenue sensitivity is now one of the largest uncertainties in many project models, and the practice of how to characterise this uncertainty is still maturing.

Third, the increasing role of machine-readable and standardised TEA. The industry has historically delivered TEAs as bespoke spreadsheets, with no standardised data model, limited reusability, and weak comparability across projects. There is increasing pressure, from regulators, lenders, and policy bodies, for TEA outputs to be structured in standardised, machine-readable forms that allow systematic comparison and audit. Initiatives like the IEA's clean energy investment frameworks, the European hydrogen-specific reporting requirements, and various national clean-fuels programmes are pushing in this direction. The TEA practice of the next decade will be more standardised, more transparent, and more directly subject to external benchmarking than the practice of the past decade.

The honest assessment is that TEA as a discipline is improving, but the structural pressures that produce optimism in project assessment have not been resolved. The selection bias that funds projects with the most favourable presented economics, the misalignment between developer and lender perspectives, and the difficulty of forecasting first-of-a-kind costs all remain. A careful reader of TEAs in 2026 should hold the same critical disposition as a careful reader in 2010 or 1995. The frameworks have matured. The structural incentives have not.