Most of the cost of a process plant is fixed by decisions made on paper, long before any equipment is ordered. A stage-gate process puts those decisions in order. At each gate the owner asks one question and spends only enough to answer it.
This series follows one example: a specialty chemical producer planning a new plant with a first estimate of $100M. This post covers the five stages. Post 2 covers the design basis, and Post 3 covers the FEED package.
The five stages
Companies use different names for the stages, but the sequence is the same. The first three are called front-end loading (FEL). FEED stands for front-end engineering design and is the last of them.
Table 1. What each stage answers and delivers
| Stage | Also called | Question it answers | Main deliverables |
|---|---|---|---|
| 1. Concept | FEL-1, appraise | Is there a business case worth studying? | Block flow diagram, market and site assumptions, Class 5 estimate |
| 2. Pre-FEED | FEL-2, select, feasibility study | Which process route, capacity and site? | Options compared, preliminary process flow diagrams (PFDs), design basis draft, Class 4 estimate |
| 3. FEED | FEL-3, define, basic engineering | What exactly will we build, and what will it cost? | Piping and instrumentation diagrams (P&IDs), equipment list, layout, hazard review, Class 3 estimate, schedule |
| 4. Execution | Detailed engineering, procurement and construction (EPC) | Is it built to the design, on budget? | Construction drawings, purchase orders, installed plant |
| 5. Start-up | Commissioning | Does it make product at the design rate? | Tested systems, trained operators, performance test |
Each gate has three possible outcomes: proceed, go back and rework, or stop. A project that stops at Gate 1 or Gate 2 has done its job. It cost little and prevented a larger loss.
How accurate is the estimate at each stage?
An estimate can only be as accurate as the design behind it. AACE International groups estimates into five classes by how much of the project is defined. Class 5 is the roughest and Class 1 is the most detailed. The ranges below are the widest typical values for process plants.
Table 2. Estimate classes applied to a $100M estimate
| Class | Stage | Typical accuracy | Low | High | Spread |
|---|---|---|---|---|---|
| Class 5 | Concept | −50% to +100% | $50M | $200M | $150M |
| Class 4 | Pre-FEED | −30% to +50% | $70M | $150M | $80M |
| Class 3 | FEED | −20% to +30% | $80M | $130M | $50M |
| Class 2 | Detailed design | −15% to +20% | $85M | $120M | $35M |
| Class 1 | Construction | −10% to +15% | $90M | $115M | $25M |
Two points follow from the table.
- A Class 5 number is a screening figure. A $100M concept estimate means the plant will probably cost between $50M and $200M. It is enough to decide whether to study the project. It is too wide to set a budget.
- Class 3 is the usual basis for approval. After FEED, the range narrows to about $80M to $130M. Most owners and lenders take the final investment decision at this point. For more on how these estimates are built, see Engineering CAPEX and OPEX Estimation for Process Plants.
Why the early stages matter most
Early decisions are cheap to change and affect almost everything. A change of process route at the concept stage costs a few days of study. The same change during construction means scrapped equipment and rework.
By the final investment decision, the owner has spent a few percent of the project cost and has already fixed most of the final cost. Front-end work is therefore the least expensive place to reduce cost and risk.
How much to spend before committing
A common rule of thumb is that front-end work costs 3% to 5% of the installed cost, with FEED as the largest part. The percentages run higher on small or first-of-a-kind projects.
Table 3. Typical front-end spend for the $100M example
| Stage | Share of installed cost | Spend | What the money buys |
|---|---|---|---|
| 1. Concept | 0.25% to 0.5% | $250,000 to $500,000 | A decision to study the project or drop it |
| 2. Pre-FEED | 0.5% to 1% | $500,000 to $1,000,000 | One selected option and a design basis |
| 3. FEED | 2% to 3% | $2,000,000 to $3,000,000 | A defined plant, a Class 3 estimate and an execution plan |
| Total before approval | 2.75% to 4.5% | $2,750,000 to $4,500,000 |
Compare that with the estimate range. Moving from Class 4 to Class 3 narrows the spread from $80M to $50M. The owner pays about $2.5M for FEED and removes $30M of uncertainty before committing $100M.
Four common mistakes
- Skipping pre-FEED. The team designs the first option in detail and never compares it with the alternatives.
- Using a Class 5 number as the budget. The first figure given to the board becomes the target, and every later estimate looks like an overrun.
- Starting FEED with an open design basis. Capacity or product specifications change during FEED and the work is done twice. Post 2 lists the decisions to fix first.
- Ordering equipment before the P&IDs are frozen. Long-lead items bought early often need modification after the hazard review.
What smaller projects can do
A $5M project does not need five formal gates. Stages can be combined, and concept and pre-FEED often become one study. The questions should still be answered in the same order: business case, option, definition, then money.
Takeaway. Spend in stages and match the decision to the accuracy of the estimate. On a $100M plant, about $2.75M to $4.5M of front-end work takes the estimate from a range of $50M to $200M down to $80M to $130M.
The plant, its capacity and its costs in this series are an illustrative example. The ranges and percentages are typical industry values and do not replace an estimate prepared for your project.
