The design process and project stages

The ethyl acetate line has been in Theo Laurent's working life for three years. In CHE-345 he sized the resin pre-reactor that starts the esterification. In CHE-350 he worked out the purification train: the reactive distillation column, the decanter, the ester column and the water stripper. In CHE-374 he wrote its control strategy, and in CHE-380 he sat on the hazard review that found an undersized relief valve and an overfill scenario on the product tanks. Every one of those studies answered a technical question. None of them answered the question that Kingfisher Bioenergy's owners care about most: should the company spend tens of millions of Canadian dollars to build the line at all?

This course follows Theo as the line goes through its last study stage before that decision. Malik Osei, who runs process engineering at the Sarnia site, has been named project sponsor. Ingrid Solheim, Theo's mentor since his first work term, leads the design team. Lucia Moreau, a cost estimator from the engineering contractor the company has hired for the front-end work, joins them, and so do the specialists Theo already knows: Nadia Brandt for process safety, Wren Halvorsen for environment and permits, Priya Venkat for industrial hygiene, Dev in the laboratory and Carla Fontaine, the senior panel operator, who will run the line if it is built. The result of the study will be a design report and a recommendation to Kingfisher's board.

Process design is the work of turning a chemical idea into a plant that can be built, operated safely and run at a profit. It draws on everything a chemical engineer has learned: thermodynamics to predict phase behaviour, kinetics and reactor design to size the reaction section, separations to purify the product, heat transfer and fluid mechanics to size equipment, control and safety to make the plant operable, and economics to decide whether it is worth building. What makes design different from the courses that taught those subjects is that the problem is open. There is no single right answer, information is always incomplete, and every decision has to be made before the next piece of information arrives.

Why design is done in stages

Large capital projects are developed in stages, with a decision point, often called a gate, at the end of each. At each gate the owner decides whether to spend the money needed for the next stage, change the scope, or stop. The logic is simple. Early studies are cheap and decisions taken early have the greatest influence on the final cost and performance of the plant. Later studies are expensive, and by the time detailed design and construction begin, most of the cost has been locked in by choices made long before. A staged process lets the owner spend a little money to learn a lot before committing a lot of money.

Many companies call the early stages front-end loading, or FEL. In the first stage, sometimes called appraise or FEL-1, the team checks whether the opportunity is real: is there a market, does a feasible process exist, is the order of magnitude of cost and profit attractive? In the second stage, select or FEL-2, the team compares alternative process routes, sites, capacities and technologies, and picks one. In the third stage, define or FEL-3, the chosen option is developed in enough detail to support a budget-quality cost estimate, a schedule and a funding request. After the sanction gate, the project moves into execution: detailed engineering, procurement, construction and commissioning. Other companies use different names, such as conceptual, pre-feasibility, feasibility and basic engineering, but the idea is the same.

Studies cost money. The code below uses rule-of-thumb shares of installed cost for each front-end stage, applied to the screening estimate of about 42 million Canadian dollars that Malik's team produced last year by scaling the cost of a larger published plant. Even at FEL-3, the study cost is a few percent of the project. That is cheap insurance against building the wrong plant.

# Design basis for the ethyl acetate line and what each project stage costs to study (illustrative)
ester_kmol_h = 26.1                 # product rate from the CHE-350 purification train
M_ester = 88.11                     # kg/kmol
hours = 8400                        # on-stream hours per year (about 96 % of the year)
t_h = ester_kmol_h * M_ester / 1000
print(f"product rate     {t_h:.2f} t/h")
print(f"nameplate        {t_h*hours:,.0f} t/yr at {hours} h/yr")
print(f"at 90 % of hours {t_h*hours*0.90:,.0f} t/yr")
# AACE-style estimate classes: (stage, project definition %, low %, high %) for a point estimate
classes = [("Class 5  concept screening", "0-2", -40, 80), ("Class 4  feasibility (FEL-1)", "1-15", -25, 45),
           ("Class 3  budget / sanction (FEL-3)", "10-40", -15, 25), ("Class 2  control", "30-75", -10, 15)]
point = 42.4                        # million, the FEL-1 screening estimate scaled from a reference plant
print("\nexpected range around a point estimate of", point, "million")
for name, defn, lo, hi in classes:
    print(f"  {name:36s} definition {defn:>6s} %   {point*(1+lo/100):5.1f} to {point*(1+hi/100):5.1f} million")
# Front-end loading: study cost as a share of installed cost (rule-of-thumb ranges, illustrative mid-points)
fel = [("FEL-1 appraise", 0.005), ("FEL-2 select", 0.015), ("FEL-3 define", 0.040)]
total = 0
print("\nfront-end study cost")
for name, f in fel:
    total += f * point
    print(f"  {name:16s} {f*100:4.1f} % of installed cost = {f*point:5.2f} million (cumulative {total:5.2f})")

product rate     2.30 t/h
nameplate        19,317 t/yr at 8400 h/yr
at 90 % of hours 17,386 t/yr

expected range around a point estimate of 42.4 million
  Class 5  concept screening           definition    0-2 %    25.4 to  76.3 million
  Class 4  feasibility (FEL-1)         definition   1-15 %    31.8 to  61.5 million
  Class 3  budget / sanction (FEL-3)   definition  10-40 %    36.0 to  53.0 million
  Class 2  control                     definition  30-75 %    38.2 to  48.8 million

front-end study cost
  FEL-1 appraise    0.5 % of installed cost =  0.21 million (cumulative  0.21)
  FEL-2 select      1.5 % of installed cost =  0.64 million (cumulative  0.85)
  FEL-3 define      4.0 % of installed cost =  1.70 million (cumulative  2.54)

The output shows two things Theo needs to keep in mind. First, the nameplate capacity of about 19,300 t/yr assumes the line runs 8,400 hours a year, which leaves about 360 hours, or two weeks, for shutdowns and upsets. Marketing should be told whether the sales forecast is based on the nameplate or on a lower realistic figure. Second, a screening estimate of 42 million could turn out anywhere from about 25 to 76 million. Nobody would sanction a plant on that basis. The purpose of FEL-3 is to narrow the range to roughly minus 15 to plus 25 percent, which is what most companies require before they release funds.

Estimate classes

The accuracy of a cost estimate depends on how much of the project has been defined. The Association for the Advancement of Cost Engineering, AACE International, publishes a widely used classification that links the maturity of the design to the expected accuracy. A Class 5 estimate is prepared when less than about 2 % of the engineering is done, often by scaling from a similar plant, and may be off by a factor of two. A Class 3 estimate, typical of the sanction gate, is based on a complete process design with equipment sized and priced, and is expected to fall within roughly minus 10 to 20 % and plus 10 to 30 % of the final cost. The exact ranges vary with the technology and with how new the process is; a process with novel steps carries a wider range than a copy of a proven plant.

The estimate class is not just a statement of uncertainty. It is a statement of what has been done. An estimator who calls a number Class 3 should be able to show the process flow diagrams, the heat and material balances, the equipment list with sizes, the plot plan and the vendor quotations behind it. Lucia's first request to Theo is for exactly those documents. Without them, she says, any number she produces is still a Class 4 number with false precision.

The design basis

The first document of FEL-3 is the design basis memorandum. It records the decisions and data that every other document will depend on, so that the dozens of people working on the project use the same assumptions. Theo drafts it from the earlier studies and circulates it for comment. Ingrid insists that every number in it has a source, and that numbers which are assumptions are labelled as such, so that they can be checked later.

A design basis normally covers the product and its specification, the capacity and the operating hours, the feedstocks and their specifications, the utilities available at the site and their conditions and prices, the site conditions such as ambient temperatures, elevation, wind and seismic data, the design codes and standards to be used, environmental limits, the plant's operating philosophy, and the economic basis for evaluating the project. For the ester line, the product is ethyl acetate of at least 99.7 % purity with low water and acidity, meeting the specification of the coatings and adhesives customers Marketing has lined up. The feeds are glacial acetic acid, bought from an outside supplier and delivered by rail, and anhydrous ethanol from Kingfisher Bioenergy's own molecular sieves.

The site conditions matter more than students expect. The design must allow for a Sarnia winter, so acetic acid, which freezes at about 17 °C, needs heated and insulated storage and piping. Cooling water is warmer in August than in January, so condensers are sized for summer. The steam system at Sarnia supplies low-pressure steam from the existing boilers, and the ethanol plant has spare capacity only up to a point. The power supply, the fire water system, the nitrogen supply and the wastewater treatment capacity all set limits that the new line has to fit within.

Codes, standards and the people who approve the plant

A plant is not only designed to work. It is designed to comply. The design basis lists the codes and standards that apply, and these come from law, from the company's own engineering standards and from the requirements of insurers. In Ontario, the pressure vessels and piping must meet CSA B51 and the TSSA pressure equipment rules, which adopt the ASME Boiler and Pressure Vessel Code. Electrical equipment in areas where flammable vapours may be present must follow Section 18 of the Canadian Electrical Code, which uses the Zone 0, 1 and 2 system. Process safety management follows CSA Z767, the Canadian standard for process safety management, applied alongside Ontario's Occupational Health and Safety Act and the Technical Standards and Safety Authority's rules for pressure equipment. The project needs an amended Environmental Compliance Approval from the Ontario Ministry of the Environment, Conservation and Parks covering the new air emissions, registration of the pressure vessels and piping with the Technical Standards and Safety Authority, a building permit from the city, and an update to the site emergency plan. Lesson 11 returns to permits, because they can set the schedule of the whole project.

Engineering documents for the line must be sealed by a P.Eng., who takes professional responsibility for them under the rules of Professional Engineers Ontario. As an engineer-in-training, Theo can prepare calculations and drawings, but they must be checked and sealed by a licensed engineer, in this case Ingrid. That is not a formality. The seal means the engineer has satisfied herself that the work is competent and that the public is protected. Ingrid checks Theo's work with that in mind, and she expects him to show his assumptions so that she can.

How a design team works

Design is iterative. A first pass through the flowsheet produces sizes and costs; the costs show which parts of the plant matter most; the team returns to those parts and refines them. The rule Ingrid repeats is to spend effort where it changes the answer. A pump that costs a few thousand Canadian dollars can be sized in minutes from a rule of thumb. A distillation column that dominates the energy bill deserves careful work. Lessons 2 to 6 build the technical design, from the choice of flowsheet through balances, heat integration and equipment sizing. Lessons 7 to 9 turn the design into capital and operating costs and then into a measure of profitability. Lessons 10 and 11 bring in sustainability, safety and permits, and lesson 12 assembles the design report and the recommendation.

The most common failure in early design is not a wrong calculation. It is an unexamined assumption: a feed purity nobody confirmed, a utility that turns out to be fully used, a product specification that a key customer will not accept, or a cost basis from a different country and year. A good design basis exposes those assumptions so they can be challenged. Carla's first comment on Theo's draft is a good example. The draft assumes the line will be run from the existing control room by the current crew with no extra operators. She asks who will do the field rounds on the new equipment during a turnaround of the ethanol plant, and the answer changes the labour cost in lesson 8.