There’s a particular kind of pressure that builds as a FEED phase winds down. Drawings are maturing, equipment lists are getting firmer, and somewhere in the background, the project controls team is assembling the cost estimate that will go in front of the client — or more accurately, in front of the client’s investment committee. That estimate will be used to sanction the project. It will shape the EPC contract strategy, influence the financing structure, and set the cost baseline against which every change order and claim will be measured for the next three to five years.
I’ve been through this phase enough times — on projects for SABIC, Saudi Aramco, KNPC, Total, and others — to know that the end-of-FEED cost estimate is one of the most consequential documents produced during the entire front-end phase. And yet it rarely gets the engineering rigour it deserves, because by the time FEED is closing out, everyone is exhausted and attention has already shifted to transition planning and EPC contract packaging. That’s a mistake worth talking about openly.
Understanding the Estimate Classification First
Before anything else, it’s worth being precise about what kind of estimate we’re actually producing. Under AACE International Recommended Practice 18R-97, the estimate produced at end of FEED sits firmly in the Class 3 category — commonly referred to as a budget authorization, appropriation, or project sanction estimate. The required level of project definition is 10% to 40% complete, which at this stage should mean you have approved PFDs, preliminary P&IDs at least issued for internal review, a developed plot plan and equipment layout, a substantially complete equipment list, and the early outcomes of hazardous area classification and relevant safety studies.
Typical accuracy for a Class 3 estimate is -10% to -20% on the low side and +10% to +30% on the high side. That range sounds wide to anyone outside the industry, but for those who’ve lived through FEED-to-EPC transitions, it’s actually a meaningful tightening from where you were at mid-FEED. A Class 4 estimate — where you might be at preliminary FEED stage — carries -15% to -30% on the downside and +20% to +50% on the upside. The convergence happens because by end of FEED, the fundamental engineering decisions are locked: process configuration, equipment selection, plant layout philosophy, and battery limit definitions. What remains undefined is the engineering detail that only emerges through full execution — every spool, every pipe support, every instrument loop. That residual uncertainty is precisely what the accuracy range represents, and what a properly determined contingency needs to account for.
It’s also worth understanding what AACE 18R-97 makes clear: estimate accuracy is not driven solely by design completeness. It is also significantly influenced by the quality of reference cost data, the complexity of the technology involved, the skill and experience of the estimating team, and the assumptions embedded in the estimate. Two Class 3 estimates on two different projects can have very different reliability profiles even at nominally the same percentage of engineering complete. New process technology, a challenging site location, or a thin historical cost database can all push the accuracy range toward the wider end — and that needs to be reflected honestly in the contingency determination.
The Engineering-Estimate Connection Most People Underestimate
Here is something that doesn’t get said often enough in our industry: the estimate is only as good as the engineering deliverables behind it. That’s not a truism — it has concrete, practical implications for how engineers hand off their work to the cost team, and it places a direct professional responsibility on every discipline lead at FEED closeout.
Your plot plan and equipment layout are among the most critical inputs to the bulk material estimate. A well-developed plot plan at end of FEED should reflect confirmed battery limits, main pipe rack routing and elevation, access and maintenance philosophy, major equipment footprints, paving and roads, and the broad structure of underground drainage and piping systems. If that layout is still evolving — if major equipment orientations haven’t been settled, if there’s still a live discussion about rack consolidation, or if underground routing is genuinely unknown — then the cost team’s takeoffs for structural steel, piping, civil foundations, and underground will carry hidden uncertainty. Those bulk material categories typically represent 30 to 45% of total installed cost on a process plant. A 15% error in that portion alone is enough to take you outside your accuracy range before contingency is even applied.
Your 3D model completion status is equally important context for the estimating team. If your model is at 20 to 30% completion at FEED closeout — which is not unusual — major process lines on the main pipe rack might be reasonably well defined, but small-bore branch piping, underground routing, and utility distribution are likely still absent. As per the guidance in AACE RP 36R-08, the estimating team needs to know exactly what’s been modelled and what hasn’t, because that determines where quantity takeoffs can be used directly versus where design development allowances need to be applied. For piping bulk quantities, industry-accepted MTO allowances at FEED stage typically range from 5 to 10% for large-bore stick-build piping, and can be higher for small-bore and traced lines where definition is lower. These allowances must be documented and justified in the estimate basis — not applied silently.
I’ve seen the consequences of this interface being poorly managed. On one major refinery project, the estimating team was working off a plot plan that was two full revisions behind the current engineering status. The difference wasn’t cosmetic — two major vessels had been relocated and a utility substation had shifted. Nobody flagged it because the mismatch happened during a pressure-filled closeout. The estimate was sanctioned, execution began, and the cost growth that emerged in the first nine months of detail engineering was significant enough to require a formal rebaseline. The engineers weren’t wrong and the estimators weren’t incompetent. The interface between design and estimating simply wasn’t managed as a shared engineering responsibility.
What a Proper Basis of Estimate Must Cover
The Basis of Estimate — the BOE — is the document that explains the foundations behind every number in the estimate. AACE RP 34R-05 characterises it as a required component of any cost estimate and, more significantly, identifies it as the one deliverable that defines the scope of the project, and ultimately becomes the basis for change management throughout the project lifecycle. If a scope dispute emerges six months into EPC execution, the BOE is what gets opened. It is not a project controls internal document — it is a shared project document that carries engineering authority.
A well-prepared BOE for an end-of-FEED Class 3 estimate needs to cover several interconnected areas. The design basis section should list every engineering document used to prepare the estimate with specific revision numbers and issue dates — not generic references to “latest revision.” P&IDs change frequently during FEED closeout. Using the wrong revision can invalidate entire sections of the piping and instrumentation bulk estimate without anyone realising it until the damage is done.
The methodology section should explain how quantities were derived for each cost category. For major process equipment, budgetary vendor quotations are expected at this stage — factored estimates based on historical data are not sufficient for tagged equipment. For bulk piping, a tiered approach is defensible: model-extracted quantities for defined large-bore lines with a documented design development allowance, and factored estimates for small-bore piping as a percentage of large-bore material cost, with that percentage and its basis stated explicitly. For civil and structural, preliminary quantities from the developed layout with stated allowances for undefined tertiary steel and foundations.
The inclusions and exclusions section is where many estimates fall short. Battery limits are the obvious starting point, but the detail matters significantly. Are vendor package unit costs inclusive of instrumentation, local panel, and first-fill utilities? Are pre-commissioning and commissioning activities in scope? What about catalyst charges, chemical first fills, operator training, and initial spare parts provisioning? On international projects, import duties, port handling, inland freight, and host country taxes need explicit treatment. Clients like Saudi Aramco and KNPC have very specific expectations about ISBL versus OSBL scope boundaries, and any departure from their standard definitions must be flagged formally in the BOE — not managed informally during estimate review.
Escalation is another area requiring deliberate attention. On a major project with 12 to 18 months between FEED completion and EPC award, followed by three to four years of construction, escalation on labour and bulk materials can be significant. The basis for any escalation indices applied — whether published indices, company historical data, or market intelligence — must be documented, along with the assumed escalation profile over the execution schedule.
Contingency at Class 3 level is not a standard percentage applied to make the total look reasonable for the investment committee. It is a quantified allowance for defined uncertainties, derived either from a deterministic risk register assessment — working through identified risks with probability-weighted cost impacts — or from a probabilistic analysis using Monte Carlo simulation. The method used must be stated in the BOE. On projects where the client has their own contingency methodology requirements, as is the case with many major national oil companies and international operators, the BOE needs to demonstrate explicit alignment with those requirements before the estimate is submitted.
Key Takeaways
The end-of-FEED cost estimate is the financial expression of your engineering work to date. Its credibility — and ultimately the project sanction decision — rests on the maturity of the design deliverables behind it and the integrity of the Basis of Estimate that accompanies it. Per AACE 18R-97, a legitimate Class 3 estimate requires confirmed PFDs, reviewed P&IDs, a developed plot plan, and a substantially complete equipment list. Anything less, and you are effectively producing a Class 4 estimate with a Class 3 label — a situation that creates cost growth during EPC that is difficult to explain and harder to recover from.
The BOE is not an administrative formality. It is the document that separates estimates that hold through execution from those that begin to unravel the moment scope firms up in detail engineering. In Part 2 of this series, I’ll go through exactly what each discipline team should — and shouldn’t — be doing before handing off to project controls at end of FEED.
Leave a Reply