Materials selection for process plants

The thin stillage evaporator at the Kingfisher Bioenergy plant in Sarnia has been leaking again. Thin stillage is the liquid left after the whole stillage from the distillation columns has been centrifuged: a warm, acidic, slightly salty soup of dissolved solids, organic acids, yeast fragments and chloride from the corn and the process water. The evaporator concentrates it into syrup by boiling off water under vacuum, and its tubes and shell were built from type 304 stainless steel when the plant opened. Twice in the last eighteen months, a tube has developed a pinhole leak that let syrup into the steam side. Each leak meant a shutdown, a plugged tube and a lost day of production worth more than the cost of the tube bundle itself.

Theo Laurent, the co-op student who followed the plant through thermodynamics, physical chemistry and the analytical lab, has been assigned to the plant's new materials and integrity engineer, Rafael Mendes, for this term. Rafael's first question to Theo is the question this course is about: when a piece of equipment fails by corrosion, was it the wrong material, the wrong design, the wrong operation or bad luck? Answering it needs a working knowledge of the materials used to build process plants, the electrochemistry that drives corrosion, the forms corrosion takes, the ways to prevent it and the inspection programs that find it before it causes harm. This course follows Theo and Rafael through all five, with Ingrid Solheim, the process engineer, supplying the process conditions.

Why materials selection matters

Corrosion is the deterioration of a material by reaction with its environment. In a chemical plant it is not a side issue. Studies of the cost of corrosion in industrial economies have repeatedly put it at roughly three to four percent of gross domestic product, and a large share of that cost falls on process industries, pipelines and utilities. The direct costs are replacement equipment, coatings, inhibitors and inspection. The indirect costs are usually larger: lost production, contaminated product, leaks of flammable or toxic material, environmental releases and, in the worst cases, fires, explosions and injuries. Many of the serious process safety incidents of the last fifty years involved equipment that had thinned or cracked over years without anyone noticing.

Materials selection is the first and cheapest line of defence. A material chosen well at the design stage may run for decades with little attention. A material chosen poorly may fail in months, and replacing it in an operating plant costs many times more than choosing correctly in the first place. Selection is never just about corrosion resistance, though. The engineer must balance mechanical properties, fabricability, availability, cost, product purity, code requirements and the consequences of failure, and must do so with incomplete information about how the process will actually run.

The questions behind every selection

Rafael teaches Theo to start every selection with a short list of questions about the service. What is the fluid, and what is in it besides the main components? Trace contaminants often decide corrosion behaviour: a few hundred milligrams per litre of chloride can change a stainless steel from immune to pitting, and a trace of oxygen can turn a harmless condensate into an aggressive one. What are the temperature and pressure, including start-up, shutdown and upset conditions? What is the velocity, and are there solids that can erode? Is the equipment wetted continuously, intermittently or only during cleaning? What cleaning chemicals will it see? A tank that holds a mild fluid for eleven months and a hot caustic cleaning solution for one day each month must survive both.

The second set of questions concerns the consequences. What happens if this item leaks? A leak of cooling water from a heat exchanger is a nuisance; a leak of ethanol vapour near an ignition source is a fire; a leak of anhydrous ammonia is a toxic release. What is the required life? A pilot plant may need three years; a pressure vessel in a base-load plant may need thirty. Can the item be inspected and repaired in service, or does any failure mean a shutdown? Is the item regulated? In Ontario, pressure equipment falls under the Technical Standards and Safety Authority (TSSA), which administers Ontario's boilers and pressure vessels regulation and adopts CSA B51 and the ASME code, and the design must follow the referenced code, including its rules on allowable stresses and corrosion allowance.

Sources of corrosion data

Corrosion behaviour depends on so many variables that no table can predict it perfectly, but several sources give a good start. Corrosion data handbooks and supplier tables list corrosion rates or ratings for common alloys in hundreds of chemicals at various concentrations and temperatures, usually in pure solutions in the laboratory. Isocorrosion diagrams show, for one alloy in one chemical, the lines of constant corrosion rate (often 0.1 mm/y) on a plot of temperature against concentration. They are especially useful for acids such as sulfuric acid, where the behaviour of an alloy changes sharply with concentration. Standards and recommended practices from industry bodies, such as the American Petroleum Institute and AMPP (the Association for Materials Protection and Performance, formed in 2021 by the merger of NACE International and SSPC), summarize decades of field experience for particular services.

The most valuable data, however, come from the plant itself. Rafael's files contain inspection records for every vessel and line, corrosion coupon results from the cooling towers and condensate system, failure analysis reports, and the plant's own corrosion loop data. When the evaporator tubes failed, the tube samples went to a laboratory for metallography, which showed pits starting under deposits on the process side. That one report told Rafael more about the evaporator than any handbook could. Plant experience is especially important for mixed or unusual streams such as thin stillage, which no handbook lists.

Laboratory and field testing fill the gaps. Immersion coupon tests expose weighed specimens to the actual process fluid for weeks or months and measure the weight lost. Electrochemical tests, described in Lesson 6, give faster answers about pitting and passivity. Test racks installed inside operating equipment during a shutdown give the most realistic results of all, because they see the real temperature, flow, deposits and upsets. Whatever the source, the engineer should ask how closely the test matched the real service, especially in temperature, concentration of minor species and the presence of crevices and deposits.

Corrosion rate units and what they mean

Uniform corrosion rates are reported as a thickness loss per year. In Canada and most of the world, the unit is millimetres per year (mm/y); in the United States, mils per year (mpy) is common, where one mil is a thousandth of an inch, so 1 mm/y is about 39.4 mpy. A rough guide to what the numbers mean for carbon steel in a process plant: below 0.025 mm/y (1 mpy) is excellent; 0.025 to 0.1 mm/y is good and easily covered by a normal corrosion allowance; 0.1 to 0.5 mm/y is acceptable only with a generous allowance and regular inspection; and above 0.5 mm/y usually calls for a different material or a change in the environment. For stainless steels and nickel alloys, which are expected to be nearly immune, any measurable uniform rate is a warning that the passive film has broken down.

These guides apply only to uniform corrosion. A material that corrodes uniformly at 0.05 mm/y may be perfectly safe for twenty years, while a material that loses almost no weight but suffers pitting or cracking may fail in weeks. That is why the evaporator's 304 tubes, which showed negligible general corrosion, still leaked. Lessons 7 to 10 cover the forms of corrosion, and the most important lesson of the whole course is that the form often matters more than the rate.

The main families of materials

Process plants use a fairly small set of material families, each with its typical services. Carbon steel is the default for most vessels, tanks, piping and structural work because it is strong, tough, weldable, available and cheap; its weakness is corrosion, which designers handle with corrosion allowances, coatings, inhibitors and control of the environment. Low-alloy steels add chromium and molybdenum for strength at high temperature and resistance to hydrogen attack. Stainless steels resist a wide range of mild to moderately aggressive environments and keep products clean, which is why the fermenters, beer well, distillation columns and evaporators at an ethanol plant are usually stainless. Nickel alloys handle the most aggressive acids and chlorides at high cost. Copper alloys, aluminium alloys and titanium have important niches in heat exchangers and special services. Plastics, elastomers, fibre-reinforced plastics and linings handle many acids and salts that would destroy metals at modest temperatures. Lessons 2 to 5 examine each family.

Life-cycle cost

The cheapest material to buy is rarely the cheapest to own. Life-cycle cost analysis adds the purchase and installation cost to the present value of maintenance, inspection, downtime and replacement over the study period, and subtracts any remaining value at the end. Because money spent in the future is worth less than money spent today, future costs are discounted. With a discount rate \(r\), a cost \(C\) incurred in year \(n\) has a present value of

For the thin stillage service, Rafael and Theo compare four options over thirty years: carbon steel with an epoxy lining, which needs relining every twelve years; 304L, which the plant's experience suggests will last about fifteen years before the pitting becomes intolerable; 316L, expected to last the full thirty years in this duty; and duplex 2205, which would comfortably outlast the study period. The program below does the arithmetic, including a salvage credit for any remaining life at the end. It also runs a simple weighted scoring of the non-cost factors, because a choice that looks best on cost may still lose on lead time, fabrication or cleanability.

# Weighted scoring and life-cycle cost for a new thin stillage tank
# Candidates: carbon steel with epoxy lining, 304L, 316L, duplex 2205
cands = {
    "CS + epoxy lining": {"capex": 180_000, "life": 12, "maint": 9_000},
    "304L":              {"capex": 260_000, "life": 15, "maint": 2_000},
    "316L":              {"capex": 300_000, "life": 30, "maint": 1_500},
    "duplex 2205":       {"capex": 340_000, "life": 40, "maint": 1_000},
}
rate = 0.07          # discount rate per year
horizon = 30         # study period, years

def pv(amount, year):
    return amount / (1 + rate) ** year

for name, c in cands.items():
    total = 0.0
    year = 0
    while year < horizon:              # buy, then replace at end of each life
        total += pv(c["capex"], year)
        year += c["life"]
    total += sum(pv(c["maint"], y) for y in range(1, horizon + 1))
    # salvage credit for remaining life at the end of the study period
    used = horizon % c["life"]
    if used:
        last = horizon - used
        total -= pv(c["capex"] * (c["life"] - used) / c["life"], horizon)
    c["lcc"] = total
    print(f"{name:18s} capex {c['capex']:>8,}  life {c['life']:>2} y  life-cycle cost {total:>10,.0f}")

best = min(cands, key=lambda n: cands[n]["lcc"])
print("lowest life-cycle cost:", best)

# Weighted scoring on non-cost criteria (scores 1-5)
weights = {"chloride resistance": 0.35, "cleanability": 0.15, "fabrication": 0.20, "lead time": 0.10, "cost": 0.20}
scores = {
    "CS + epoxy lining": [2, 3, 5, 5, 4],
    "304L":              [2, 5, 4, 4, 3],
    "316L":              [4, 5, 4, 4, 3],
    "duplex 2205":       [5, 5, 3, 2, 3],
}
for name, sc in scores.items():
    total = sum(w * s for w, s in zip(weights.values(), sc))
    print(f"{name:18s} weighted score {total:.2f}")

CS + epoxy lining  capex  180,000  life 12 y  life-cycle cost    395,267
304L               capex  260,000  life 15 y  life-cycle cost    379,054
316L               capex  300,000  life 30 y  life-cycle cost    318,614
duplex 2205        capex  340,000  life 40 y  life-cycle cost    341,243
lowest life-cycle cost: 316L
CS + epoxy lining  weighted score 3.45
304L               weighted score 3.25
316L               weighted score 3.95
duplex 2205        weighted score 3.90

The lowest purchase price, the lined carbon steel tank, has the highest life-cycle cost once relining and downtime are counted, and 304L, the plant's original choice, is not much better. The 316L option wins on both cost and score here, though duplex is close behind and would be favoured if chloride levels rose. The answer is only as good as its inputs, of course, and the life estimates are the softest numbers in the analysis. Rafael's practice is to rerun the comparison with pessimistic lives for the better alloys and optimistic lives for the cheaper ones; if the ranking survives that test, the choice is robust.

Documenting the selection

A materials selection is not finished until it is written down. Rafael's plant keeps a materials selection report for each new item and a corrosion loop diagram for the whole plant. A corrosion loop groups equipment and piping that share the same process fluid and the same likely damage mechanisms, so that inspection and monitoring can be planned loop by loop. The report records the service conditions assumed, the damage mechanisms considered, the material chosen, the corrosion allowance, any limits on operating conditions (for example, a maximum chloride level or temperature) and the inspection plan. Those limits matter as much as the material. If the evaporator is upgraded to 316L on the assumption that chloride stays below a certain level, the plant must monitor chloride and act if it rises; otherwise the new tubes may fail as the old ones did.

Operating limits written into a selection report are often called integrity operating windows. Each window states a variable (chloride, pH, temperature, oxygen, velocity), its normal range, and what operators must do if it drifts outside the range: adjust the process, notify the integrity engineer, or shut down. Lesson 12 returns to integrity operating windows as part of the plant's corrosion management system.