01Executive summary
Three materials milestones fall within fifty years of each other, and between them they define what engineers have been able to build ever since: bulk steel, synthetic polymers and engineered wood products.
The common thread is not chemistry. It is the replacement of found properties with specified properties. Wrought iron was whatever the puddler produced that day. Timber was whatever the tree grew. The engineering of this period consisted of taking each of those variable natural inputs and building a process around it that delivered a number you could put in a calculation — and, critically, a number you could hold a supplier to.
A material becomes an engineering material at the moment its properties become reproducible and warranted, not at the moment it is discovered.
02Bulk steel: the Bessemer converter, 1855
Before the mid-1850s steel existed but was a luxury. The cementation and crucible routes produced good material in kilogram quantities at prices that made structural use absurd. Everything large was built from cast iron (strong in compression, brittle and unreliable in tension) or wrought iron (tough and fibrous but laborious to make and inconsistent).
Henry Bessemer’s insight was to stop adding heat and start using the charge itself as the fuel. Blow air through molten pig iron and the dissolved carbon, silicon and manganese oxidise. Those oxidation reactions are strongly exothermic — enough to keep the bath molten without external firing while the carbon burns out as carbon monoxide. A blow that converted several tonnes took roughly twenty minutes.
Why the process needed a second invention
The converter as first built worked only on low-phosphorus ores, because the acid (siliceous) lining could not remove phosphorus and phosphorus makes steel cold-short — brittle at ambient temperature. The basic lining developed by Thomas and Gilchrist about twenty years later, using dolomite and a lime flux, fixed this and opened the process to the ore bodies of continental Europe. It is a useful reminder that the headline invention is often only viable after an unglamorous metallurgical fix that nobody remembers.
- Ore to ironBlast furnace reduces iron oxide with coke; output is pig iron at roughly 4–5% carbon, saturated and brittle.
- DecarburiseOxygen (air, then later pure oxygen) burns out carbon and tramp elements; exothermic reactions sustain the melt.
- Alloy and deoxidiseControlled additions set the final chemistry — carbon, manganese, chromium, molybdenum, vanadium.
- Cast, roll, heat-treatSection geometry and thermal history set the delivered mechanical properties.
What the carbon content actually buys
Carbon in steel is an interstitial solute that impedes dislocation motion and, on quenching, permits the formation of martensite. More carbon therefore means higher hardness and strength and lower ductility and weldability. The relationship is monotonic and brutally simple, which is why it is the first number on any steel certificate.
| Class | Carbon | Behaviour | Typical use |
|---|---|---|---|
| Low / mild | < 0.25% | Ductile, readily welded, modest strength | Structural sections, plate, pipe |
| Medium | 0.25–0.60% | Responds to quench and temper; weldability needs care | Shafts, gears, rails |
| High | 0.60–1.0% | Hard, wear-resistant, notch-sensitive | Springs, cutting edges, wire |
| Tool / alloy | varies | Carbide formers dominate; hardness retained hot | Dies, tooling, hot-work parts |
Case hardening: two properties in one part
Where a component must resist wear at the surface but survive impact through the section, carburising or nitriding raises carbon or nitrogen in the outer layer only. The result is a hard case over a tough core, with the additional benefit that the transformation places the surface in residual compression — which directly improves fatigue life, because fatigue cracks initiate at surfaces in tension. This is one of the neatest examples in engineering of using a process to defeat a material trade-off rather than accepting it.
Structural steel supply is governed by AS/NZS 3678 (plate), AS/NZS 3679.1
(hot-rolled bars and sections) and AS/NZS 1163 (hollow sections), with design to
AS 4100 and welding to AS/NZS 1554. Two practical cautions: grade
designations refer to yield strength in megapascals, so a 300-grade section and a 350-grade
section of identical geometry are not interchangeable in a strength check; and imported
sections require certification traceable to a recognised test regime before they are treated
as compliant. Non-conforming structural steel has been a recurring issue in Australian
construction and is worth an explicit hold point in procurement.
03Synthetic polymers: Parkesine, 1856
Alexander Parkes patented a nitrocellulose compound, plasticised and mouldable when warm, that he exhibited publicly in the early 1860s. Commercially it failed — it was expensive, it shrank and, being nitrocellulose, it was alarmingly flammable. Technically it established the proposition that mattered: a material could be synthesised to a specification rather than harvested.
The modern significance is structural rather than historical. Polymer properties are governed by a small number of molecular parameters, and an engineer who understands those four levers can predict behaviour across an enormous product range.
Chain length
Higher molecular weight raises melt viscosity, toughness and creep resistance, and lowers processability. Ultra-high molecular weight polyethylene and a supermarket bag differ mainly in this one parameter.
Branching and crystallinity
Linear chains pack into crystallites; branched chains cannot. Crystallinity raises stiffness, density, and chemical and barrier resistance while reducing clarity and impact toughness.
Cross-linking
Covalent links between chains convert a thermoplastic into a thermoset or an elastomer. The material can no longer be re-melted — which is exactly what a gasket, a tyre or an epoxy joint requires, and exactly what makes recycling hard.
Reinforcement and fillers
Glass, carbon or mineral fillers move stiffness and thermal expansion toward the filler and away from the matrix, at the cost of anisotropy and, usually, of toughness.
The design failure mode engineers most often miss
Polymers are viscoelastic. A polymer part under sustained load does not simply deflect and stop; it creeps, and the deflection continues for the life of the load. Two consequences follow that catch out designers used to metals:
- Design to a creep modulus, not the datasheet modulus. The short-term tensile modulus quoted on a supplier sheet is measured in seconds. A part loaded for ten years may exhibit an effective modulus a small fraction of that value.
- Bolted plastic joints relax. Preload decays through stress relaxation, so a plastic flange that was tight at handover will not be tight in three years unless the joint is designed with a metal compression limiter or a spring element.
The properties that make polymers superb engineering materials — chemical inertness, low density, resistance to degradation — are precisely the properties that make them a persistent environmental problem. A materials selection exercise that scores only cost, mass and mechanical performance is incomplete. End-of-life pathway, additive content and the realistic recyclability of the specific grade belong in the same table as tensile strength.
04Engineered wood: commercial plywood, 1905
Sawn timber is a beautiful structural material with an inconvenient property: it is strongly anisotropic and its defects are randomly distributed. Strength along the grain may be an order of magnitude greater than across it, and a single knot in the wrong place can halve the capacity of a member. Engineered wood products exist to average that variability away.
| Product | Construction | Problem solved |
|---|---|---|
| Plywood | Cross-laminated rotary-peeled veneers, odd number of plies | Removes in-plane anisotropy; gives dimensional stability and panel shear capacity |
| Laminated veneer lumber | Parallel-laminated veneers | Disperses defects statistically; delivers a tight, predictable bending strength distribution |
| Glued laminated timber | Finger-jointed, face-glued laminations | Allows long spans and curved members from short, small-section feedstock |
| Oriented strand board | Aligned flakes in bonded layers | Converts small and low-grade material into a structural sheet |
| Timber I-joist | LVL flanges, structural panel web | Places material where bending stress is highest; suppresses shrinkage and creaking |
| Prefabricated truss | Jig-assembled, nail-plated | Repeatable geometry, factory tolerance, rapid site erection |
Why the I-joist is the clearest lesson
In a beam in bending, longitudinal stress varies approximately linearly from the neutral axis, so material near the centroid contributes almost nothing to moment capacity but all of its mass. Shear, conversely, peaks at the neutral axis. The rational answer is stiff flanges far from the axis and a thin web to carry shear — the I-section. A solid timber beam cannot do this because it must be cut from a single piece; an engineered one can, because the flanges and web are separately specified materials bonded together. The same logic produced the steel universal beam and the sandwich panel.
Timber structures are designed to AS 1720.1, with structural plywood to
AS/NZS 2269, glulam to AS/NZS 1328 and LVL to AS/NZS 4357.
Three items that consistently cause trouble on Australian projects: the durability and
preservative treatment class must suit the hazard class of the actual service location, not the
building type; moisture content at installation drives most subsequent dimensional complaints;
and bushfire-prone area construction to AS 3959 can override an otherwise sound
timber selection entirely. Specify the treatment class and the moisture condition explicitly —
“treated pine” is not a specification.
05Selection: how the three actually compete
Indicative comparison only, on a normalised scale. Real selection needs project-specific data, but the shape of the trade-off is stable.
Specific stiffness (stiffness per unit mass)
Predictability of delivered properties
06Takeaways
Process defines the property
Chemistry sets the possible range; thermal and mechanical history sets what you actually receive. Always specify both.
Variability is the real enemy
Every one of these three milestones is a method for narrowing a distribution, not for raising a mean.
Anisotropy is a tool
Cross-lamination, fibre alignment and case hardening all deliberately engineer direction-dependent properties rather than avoiding them.
Certificates are engineering documents
A mill certificate, a grade stamp and a batch traceability record are load-bearing parts of the design. Treat them as such.