FEED and Process Design Fundamentals

A practical introduction to front-end engineering design (FEED) for owners, project managers and engineers of process plants in chemicals, energy, food, materials and similar industries. You will learn the project stage gates, the design basis, mass and energy balances, PFDs, equipment sizing, P&IDs, control philosophy and Canadian codes and approvals, then process hazard analysis, relief, layers of protection, cost estimate classes, schedule and the FEED package that supports the investment decision. The course is educational and does not replace qualified engineering or the codes and regulations that apply to a project.

Course structure

• Module 1: Projects and FEED (4 lessons). The project stages and gates, what FEED fixes and why it pays, roles and contracts, and the design basis.
• Module 2: Process Definition (4 lessons). Mass and energy balances, process flow diagrams, equipment sizing and datasheets, and utilities, offsites and battery limits.
• Module 3: Engineering Documents (4 lessons). P&IDs, control philosophy and interlocks, lists and layouts, and the codes, standards and approvals that apply in Canada.
• Module 4: Safety and Risk (4 lessons). HAZID and HAZOP, relief and overpressure protection, layers of protection and safety instrumented systems, and constructability, operability and maintainability.
• Module 5: Cost, Schedule and the Decision (4 lessons). Estimate classes and contingency, schedule and long-lead procurement, the FEED package and gate review, and the path to start-up.

Interactive labs

• Which document decides it?: Sort twelve design decisions by the FEED document where they are first fixed: the design basis, the PFD or the P&ID. Then check your answers.
• Ranking hazard scenarios: Place seven hazard scenarios from a HAZOP on a likelihood and consequence grid to see which need urgent risk reduction, which need further review and which are tolerable with existing safeguards.
• Estimate class and plausible cost range: Enter a base cost estimate, its class and the contingency carried to see the typical accuracy range for that class and the plausible low and high costs.

How it works

• Work through 20 short lessons at your own pace. Each has key takeaways, an exercise for your own workplace and a quick knowledge check.
• Try the three interactive labs as you reach them in the course.
• Pass the 20-question final exam (70% or higher, unlimited retakes) to unlock your Chmltech certificate of completion.
• Free, with no sign-up. Your progress is saved in this browser.

Process plant projects move through a series of stages, each ending in a gate: a decision to continue, change course or stop. Each stage spends more money and fixes more decisions. The stage names vary between companies, but the logic is the same.

Why the early stages matter most

The ability to influence cost and performance is highest at the start, when changes are made on paper, and falls quickly once equipment is ordered and steel is up. The cost of a change rises at the same time: moving a line on a P&ID during FEED takes an hour; moving it after the pipe is installed takes cutting, welding, re-testing and schedule delay. Front-end work is cheap insurance against expensive surprises later.

About this course

This course explains how front-end engineering works so owners, project managers and engineers can plan, review and use FEED well. It is educational. It does not replace a qualified engineering team or the codes, standards and regulators that apply to your project. The Insights article From concept to construction covers the stages in more depth.

Front-end engineering design (FEED), sometimes called basic engineering or the define stage, develops the selected process into a design complete enough to estimate cost reliably, plan the schedule, identify the main risks and support the final investment decision (FID).

What FEED fixes and what it leaves for later

Why it pays

• Better estimates: FEED typically moves the cost estimate to an accuracy suitable for funding decisions, which lenders and boards expect.
• Fewer late changes: decisions are made while they are cheap to change.
• Earlier procurement: long-lead equipment can be specified and sometimes ordered.
• Safer design: hazards are found and designed out on paper.
• Competitive bids: a complete FEED package lets contractors bid on the same scope.

Cutting FEED short to save time or money is one of the most common reasons projects overrun later. The Insights article What a FEED package contains lists the documents in detail.

Several parties contribute to FEED, and the contract structure shapes how they work together.

Common contracting models after FEED

• EPC (engineering, procurement and construction): one contractor delivers the plant, often lump sum. The owner gets cost certainty but pays for the contractor's risk, and the FEED must be solid enough to bid on.
• EPCM (engineering, procurement and construction management): the contractor engineers and manages on the owner's behalf; the owner holds construction contracts and more cost risk, with more control.
• Owner-managed: common for smaller projects; the owner coordinates engineering firms and contractors directly.

The owner's job in FEED

FEED is not something to hand over and collect at the end. The owner must provide the design basis inputs, answer questions quickly, make decisions on options, involve operations and maintenance staff in reviews, and approve key documents. Slow owner decisions are a common cause of FEED delays. Name one owner's representative with authority to decide, and agree a weekly decision log with the engineering team.

The design basis is the document that records the assumptions and requirements every engineer designs to. Agreeing it at the start of FEED prevents rework later, because a change in capacity or feed composition ripples through every calculation and drawing.

What a design basis covers

Margins: agree them once

If each discipline adds its own safety margin, the plant ends up oversized and expensive. Agree in the design basis where margins apply and how large they are, and record the reason. Oversized equipment is not always safer: pumps and control valves that run far below their design point can operate poorly.

Keep it controlled

Issue the design basis with a revision number, have the owner sign it off and route any change through a formal change process with an impact check on cost and schedule. The Insights article The design basis: ten decisions goes through each in more detail.

Every process design starts with the mass and energy balance: what goes in, what comes out and what happens in between, stream by stream. It is the foundation for sizing every piece of equipment and every utility.

The principles

• Mass balance: at steady state, mass in = mass out for every unit and for the whole plant. For reacting systems, balance each element or use reaction stoichiometry and conversion.
• Energy balance: heat in + work in = heat out + work out + change in enthalpy. It gives heating and cooling duties, which size exchangers and utilities.

Where the data come from

For established processes, data come from licensors, literature and process simulators. For new or scaled-up processes, they come from pilot plant and laboratory work: conversions, selectivities, yields, residence times, heats of reaction and physical properties. Gaps in data are a major FEED risk. Identify them early and decide whether to fill them with more testing or carry them as a design margin.

The stream table

The balance is presented as a stream table: for each numbered stream, its flow, composition, temperature, pressure and phase. Stream numbers link the table to the process flow diagram.

Check the balance closes: here feed (1,000 kg/h) equals product (950) plus a 50 kg/h purge not shown, and the reactor outlet equals feed plus recycle. A balance that does not close points to an error or a missing stream.

Cases, not just one point

Design for more than the normal case: maximum and minimum throughput, start-up, different feed compositions and summer and winter conditions. Equipment is sized to the governing case for each item. The Insights article From pilot data to a mass and energy balance works through an example.

The process flow diagram (PFD) shows how the process works: the major equipment, the main process streams and how they connect, the key operating conditions and the main control loops. It is the bridge between the balance and the detailed design.

What a PFD shows, and what it leaves out

Reading a PFD

Follow the main stream from feed to product. At each piece of equipment, ask what it does to the stream: heat it, react it, separate it, pump it. Note recycles and purges, which often control how impurities build up. Look at the control loops: what is being controlled and what is manipulated.

Why the PFD matters in FEED

The PFD is where process options are compared and fixed: one reactor or two in series, which separation method, how heat is recovered between streams. Changes are easy at this stage. Once P&IDs and equipment datasheets are developed from the PFD, changes become slower and costlier, so review the PFD carefully with operations and the owner before moving on. The Insights article How to read a PFD and a P&ID has worked drawings.

With balances and the PFD fixed, the process engineer sizes each piece of equipment and records the requirements on a process datasheet. Mechanical engineers and vendors then complete the mechanical design.

What gets sized, and the key inputs

An example: heat exchanger duty and area

To cool 10,000 kg/h of a liquid with heat capacity 2.5 kJ/kg·K from 120 °C to 50 °C, the duty is Q = m × c_p × ΔT = (10,000 ÷ 3,600) × 2.5 × 70 ≈ 486 kW. Area follows from Q = U × A × LMTD, with the overall coefficient U from experience or detailed rating. The Insights article Sizing a heat exchanger completes the calculation.

Design conditions

Each item needs a design pressure and design temperature above the maximum it could see in operation, upsets and start-up, set by an agreed method. They drive wall thickness, flange ratings, relief settings and cost, so set them carefully and consistently.

Long-lead and critical equipment

Flag equipment with long delivery times, such as compressors, large or alloy vessels, specialty reactors and large electrical equipment, so vendor quotes and data come in early.

A process unit cannot run without utilities and offsites. They are often a large part of the total installed cost, and they are easy to underestimate when attention is on the main process.

Utility balances

Sum each utility's demand from every consumer in each operating case, including start-up and peak cases, and compare with what is available. For a project on an existing site, a utility constraint such as limited electrical supply can drive cost and schedule more than the process itself, because utility upgrades can take a long time to arrange.

Battery limits

Define exactly where the project's scope ends: each line, cable and service crossing the boundary, with its conditions at the tie-in point. Unclear battery limits cause gaps, where nobody designs a needed item, and overlaps, where two parties design the same thing. A battery limit table, signed off by both sides, prevents both.

The piping and instrumentation diagram (P&ID) is the central engineering document of a process plant. It shows every piece of equipment, line, valve, instrument and control and safety function, and how they connect. Operators, engineers, maintenance staff and hazard reviews all rely on it for the life of the plant.

What a P&ID shows

• All equipment with tags, including spares and packaged units.
• All process and utility lines with line numbers that encode size, service, sequence and piping class.
• Manual valves, control valves and their fail positions, check valves, drains and vents.
• Instruments and their functions using standard symbols and tag letters, commonly based on ISA-5.1.
• Pressure relief devices with set pressures.
• Interlocks and safety instrumented functions.
• Specification breaks, insulation and heat tracing, and sample points.

Reading instrument tags

The first letter is the measured variable and the following letters the function. FIC-101 is a flow indicating controller; LT-205 a level transmitter; PSV-301 a pressure safety valve; TAH a temperature alarm high. Learn the common letters and most P&IDs become readable.

P&ID development in FEED

P&IDs go through several issues: for review, for HAZOP, after HAZOP and approved for design. Each review adds detail and resolves comments. By the end of FEED they should be stable enough that detailed engineering adds detail rather than changing the design. After that, any change follows management of change.

Operators and maintenance staff should review P&IDs before the HAZOP: they spot missing drains, unreachable valves and isolation that does not allow safe maintenance. The lab after this lesson tests which FEED document records which decision.

How the plant will be operated and controlled is decided in FEED and recorded in a control philosophy (or control narrative) and related documents.

What the control philosophy covers

• The level of automation and what operators do manually.
• The main control loops and strategies: what is measured and what is adjusted to hold each key variable.
• Start-up, normal operation, shutdown and emergency shutdown sequences.
• The control system architecture: basic process control system (BPCS), any separate safety instrumented system (SIS), and links to packaged equipment controls.
• Operator interface, alarm philosophy and data historian.

Alarms

An alarm should require an operator action. Too many alarms, or alarms that come in floods during upsets, lead operators to miss the important ones. An alarm philosophy sets rules for when an alarm is justified, how it is prioritized and how operators should respond. Recognized practices for alarm management, such as ISA-18.2, give a structured approach.

Interlocks and cause-and-effect charts

Interlocks take automatic action when a condition occurs: trip the feed pump on high-high level, close the fuel valve on low air flow. A cause-and-effect chart lists causes (initiating conditions) in rows and effects (actions) in columns, with a mark where a cause triggers an effect. It is a compact, reviewable record of all automatic actions and is used later to program and test the system.

Separate control from safety

Normal control keeps the process at target. Safety functions bring it to a safe state when control fails. Where a safety function is needed to reduce risk to a tolerable level, it is typically implemented in an independent SIS designed to a recognized standard such as IEC 61511, covered later in the course.

Alongside the drawings, FEED produces lists and layouts that turn the design into quantities for estimating, procurement and construction planning.

Layout decisions with lasting effects

The plot plan affects safety, operation, maintenance and cost for decades. Consider:

• Separation distances between hazardous equipment, occupied buildings and site boundaries.
• Prevailing wind for gas releases and the location of control rooms and air intakes.
• Access for cranes, maintenance, tube bundle pulling and emergency response.
• Gravity flow and pump suction requirements, which may fix equipment elevations.
• Space for future expansion.

3D models

Many FEED projects build a 3D model of the main equipment and large piping. Model reviews with operations and maintenance catch access problems that are hard to see on 2D drawings, such as a valve that can only be reached with a ladder.

Process plants in Canada are designed and built under provincial and federal requirements, with recognized codes and standards adopted by reference. FEED must identify what applies early, because some approvals take months.

What FEED should produce

• A codes and standards list naming editions, agreed with the owner.
• A regulatory and permitting plan with lead times, required studies and responsibilities.
• Early meetings with authorities on novel aspects, such as new technology or unusual materials.

Requirements differ between provinces and change over time, and some sites have specific municipal or Indigenous consultation requirements. Confirm the current requirements for your project with the authorities and qualified professionals involved; this table is an orientation, not a compliance checklist.

FEED is the best time to find and design out hazards. Two structured reviews are common.

HAZID

A hazard identification review early in FEED looks at the whole facility: the materials, site, layout, external events and major hazard scenarios such as fires, explosions and toxic releases. It shapes layout, separation distances and the overall safety concept.

HAZOP

A hazard and operability study examines the P&IDs line by line. The plant is split into nodes, sections with a defined design intent. For each node, the team applies guidewords to process parameters to generate deviations:

For each credible deviation the team records causes, consequences, existing safeguards and, where safeguards are not enough, recommendations. Each recommendation gets an owner and is tracked to closure.

Running a good HAZOP

• An experienced, independent facilitator and a scribe.
• Process, operations, instrumentation, mechanical and safety participants.
• P&IDs at the HAZOP issue, with the design basis, cause-and-effect charts and datasheets available.
• Realistic scheduling: tired teams miss things, so keep sessions to a sensible length.

HAZOP also covers operability: start-up, shutdown, maintenance and sampling. Many recommendations improve how the plant will be run, not only its safety. The Insights article A worked HAZOP walks through a node in detail. The lab after this lesson asks you to place hazard scenarios on a risk grid.

Every piece of equipment that could be overpressured needs protection. FEED identifies the overpressure scenarios, sizes the relief devices and designs the system that safely disposes of what they release.

Common overpressure scenarios

• Blocked outlet: a valve closes downstream while a pump or compressor keeps feeding.
• External fire: heat input to a vessel causes liquid to boil and pressure to rise.
• Control valve failure: a valve fails open and lets a high-pressure source into a lower-pressure system.
• Tube rupture: high-pressure fluid leaks into the low-pressure side of a heat exchanger.
• Utility failure: loss of cooling water or power, which can stop condensers or cooling.
• Runaway reaction: an exothermic reaction accelerates beyond cooling capacity.
• Thermal expansion: liquid trapped between closed valves is heated.

Relief design steps

1. List credible scenarios for each piece of equipment.
2. Calculate the required relief rate for each; the largest credible case governs the size.
3. Select and size the device, a pressure safety valve or rupture disc, commonly using methods in API 520 and API 521.
4. Design the inlet and outlet piping so pressure losses do not stop the device from working properly.
5. Design the disposal system: to atmosphere where safe and allowed, or to a closed system, knock-out drum, flare or scrubber.

Runaway reactions need special care

Relief for runaway reactions can involve two-phase flow and very large rates. Get reaction calorimetry data and specialist input early. The Insights articles on runaway reactions and pressure relief basics cover these topics further.

Relief design is a specialist, code-governed task. It must be done and checked by qualified engineers.

No single safeguard is perfect. Process safety relies on several independent layers of protection, each able to stop a hazard scenario from becoming an incident.

Inherently safer design comes first

The most reliable layer is a hazard that is not there. During FEED, ask whether you can eliminate, reduce, substitute or simplify: store less, operate at milder conditions, choose a less hazardous material, or design equipment that cannot be overpressured. These choices are cheap on paper and expensive later.

LOPA and SIL

Layer of protection analysis (LOPA) takes high-consequence scenarios from the HAZOP and checks, semi-quantitatively, whether the independent layers reduce the frequency to a tolerable level. If not, a safety instrumented function may be needed, with a safety integrity level (SIL) that sets how reliable it must be. Safety instrumented systems for the process industry are commonly designed under IEC 61511, which covers the whole lifecycle from specification to testing and maintenance.

A layer only counts if it is independent of the cause and of the other layers. A high-level alarm from the same transmitter that failed and caused the overflow is not independent.

A plant can be safe and correctly sized on paper and still be hard to build, run and maintain. FEED reviews these qualities while changes are still cheap.

Constructability

• Can equipment be transported to site (road limits, bridges, rail) and lifted into place?
• Can parts be modularized or pre-assembled off-site to reduce site work and schedule risk?
• Is there laydown space and crane access?
• Does the sequence allow tie-ins to an operating plant with minimum shutdown time?

Operability

• Can operators start up, shut down and handle upsets with the controls and instruments provided?
• Are sample points, sight glasses and local indicators where operators need them?
• Is the operating window wide enough for real feed variations?

Maintainability

• Can each item be isolated, drained, purged and made safe for maintenance?
• Is there access for removing exchanger bundles, pump motors and agitators?
• Are spares, sparing philosophy and maintenance intervals defined?

The best reviewers are experienced operators and maintenance technicians from a similar plant. Hold structured reviews on the P&IDs and the 3D model, and record actions like HAZOP recommendations.

Change control starts in FEED

Once key documents are approved, changes should go through a design change process: describe the change, assess impact on safety, cost and schedule, get approval and update every affected document. The same discipline continues through detailed engineering, construction and operation as management of change. The Insights article Management of change for small plants covers the operating phase.

A cost estimate is only as good as the design behind it. The AACE International recommended practice 18R-97 defines estimate classes for process industries by how complete the project definition is.

The ranges are wide and asymmetric: estimates are more likely to come in under-scoped than over. The actual range for a project depends on its technical and execution risks.

What a FEED estimate includes

• Equipment costs from vendor quotes for major items and estimating data for the rest.
• Bulk materials: piping, steel, concrete, electrical, instrumentation, from the lists and layouts.
• Construction labour, productivity and site conditions.
• Indirect costs: engineering, construction management, temporary facilities, commissioning, spares, owner's costs.
• Contingency: an amount for costs that will occur within the defined scope but cannot be itemized yet.

Contingency and escalation

Contingency covers uncertainty within scope; it is not for scope changes. Base it on a risk analysis rather than a fixed percentage habit. Separately, allow for escalation: price changes between the estimate date and when money is spent. The lab after this lesson shows the plausible range around a base estimate for each class.

FEED produces a project schedule that the investment decision relies on. The critical path in process projects often runs through engineering and procurement of a few key items, not through construction.

Typical long-lead items

• Compressors and large rotating equipment.
• Large, thick-walled or alloy pressure vessels and reactors.
• Specialty heat exchangers.
• Large transformers, switchgear and utility connections.
• Control and safety systems, when highly customized.
• Items made from specialty materials with limited supply.

Lead times change with market conditions, so get current vendor quotes during FEED rather than relying on past experience.

Early procurement

To protect the schedule, the owner may authorize ordering long-lead equipment before FID. This commits money early and is only sensible when the design of that item is firm and the risk of cancellation is acceptable. Cancellation terms should be agreed with the vendor.

The FEED schedule should show

• Engineering durations and key document issues.
• Vendor data dependencies: detailed engineering needs certified vendor drawings.
• Procurement of each major item, from enquiry to delivery.
• Permits and approvals with their lead times.
• Construction, mechanical completion, commissioning and start-up.
• Planned shutdowns for tie-ins on existing sites.

Realism over optimism

Compare the schedule with similar completed projects. Optimistic durations become a problem later, when costs are committed. Identify the critical path and the few items that, if late, delay everything, and manage them closely.

At the end of FEED, the deliverables are assembled into a FEED package that supports the investment decision and becomes the basis for detailed engineering or for EPC bids.

Typical contents

The FEED gate review

Before FID, a review, ideally with people independent of the project team, checks:

• Is the design complete enough for the estimate class claimed?
• Are HAZOP and other review actions closed or tracked?
• Are the main risks identified with owners and responses?
• Are schedule and cost realistic compared with similar projects?
• Do the economics still work at the high end of the estimate range?

The risk register

List each significant risk with its likelihood, impact, owner and response: avoid, reduce, transfer or accept. Typical FEED-stage risks include scale-up uncertainty, feedstock variability, long-lead delivery, permitting delays, labour availability and interface risks on existing sites.

A decision not to proceed is a legitimate outcome of FEED. Stopping a weak project before construction is far cheaper than finishing it.

After FID, the project moves into execution. Knowing what follows helps you make FEED complete.

A FEED checklist for owners

• Design basis signed off and under change control.
• Pilot and data gaps identified and resolved or covered by margins.
• Balances closed for all key cases; utilities confirmed.
• P&IDs through HAZOP with actions tracked; relief study done.
• Operations and maintenance staff took part in reviews.
• Codes, standards and permitting plan agreed.
• Class 3 estimate with risk-based contingency; realistic schedule with long-lead items.
• Risk register with owners; independent gate review completed.

The Insights articles on commissioning, the pre-startup safety review and start-up and performance testing continue the story. This course is educational; real projects need qualified engineers and must meet the codes and regulations that apply.