Three main types of coke are produced from coking coals:
  • Metallurgical coke is produced in coke ovens and is mainly used in the reduction of iron ore to pig iron in blast furnaces. It is also consumed in blast and electric furnaces for ferro- alloy production, reduction of metal oxides to metals and chlorides, reduction of phosphates and sulphates, and in the reduction of carbonates to carbides.
  • Foundry coke is produced in beehive or non-recovery coke ovens and is used at foundries to melt iron and various copper, lead, tin and zinc alloys in cupolas. The basic coke requirements are the same as for metallurgical coke but the size specification varies, depending on the size of the cupola. Foundry coke is almost always of larger size than metallurgical coke.
  • Domestic coke, or more often semi-coke, is used as a fuel. A low ash, easily ignited coke of high specific energy with a very low sulphur content is required.
The coking processes may be carried out in either slot-type ovens, beehive ovens, newer slot-type ovens with preheating of the charge, travelling grate ovens, rotary kilns or in formed-coke plants. Only the first two processes have widespread commercial application. Slot ovens are generally located adjacent to or close to the steelworks where valuable by-product gases can be collected and used. Beehive (non- recovery) ovens that produce larger and less reactive cokes for foundry use can be located at any convenient site. Formed-coke processes have been developed but have not been commercially implemented because the qualities of that coke have been proven to be inferior to slot-oven and beehive oven cokes.
Aspects of current coking practice that influence coke properties are heating rate and duration, charge bulk density, final temperature, and degree of preheating the charge.  
The value of by-products from carbonisation fluctuates, depending on the oil price, and is secondary to the value of the coke. Nevertheless, the production of by-products is important to the overall economics of a steelworks. The composition and level of production of the by-products is dependent on the types of coals used in the coking blend.
Metallurgical coke
The steel industry is the largest consumer of coke, using it to reduce iron ore to pig iron in blast furnaces where the coke has three roles:
  • source of heat through combustion with the hot air blast at the tuyeres
  • source of reducing gas after reaction with the hot air blast to form carbon monoxide, or by reaction with carbon dioxide produced during the high temperature stage of iron ore reduction
  • permeable granular material with sufficient strength to support raw materials (known as the burden) in the blast furnace through which gas can percolate, particularly in the lower regions of the furnace. Poor permeability affects furnace stability, output and fuel efficiency.
During the movement of the burden down the furnace, the coke within it is subjected to mechanical degradation and chemical attack. Coke size decreases by:
  • reaction with carbon dioxide, which occurs in the temperature range 900–1100°C
  • reaction with alkali metal vapours at temperatures of up to 1450°C, which cause a reduction in the abrasion resistance of the coke
  • thermal effects at temperatures of up to 1500°C, which may further weaken the strength of the coke and reduce its size.
Increasing trends by the steel industry to use pulverised coal injection (PCI) to reduce coke requirements has placed more stringent requirements on the burden quality, in particular on the coke quality. The industry uses layer charging of different coke size ranges. It is also recognised that larger coke particles with a relatively uniform size distribution are required to reduce blast resistance. It is also important that the coke has high and consistent impact strength and abrasion resistance at the furnace operating temperature, as the presence of fines increases the burden resistance, increases slag viscosity and increases carbon loss through the off-gases. Low and consistent coke moisture content is desirable as water vapour catalyses the oxidation of carbon monoxide to carbon dioxide and reduces the extent of direct reduction of the iron ore. Also, high moisture diminishes the carbon feed rate for a constant gravimetric or volumetric coke input. Lower coke reactivity is also required at high PCI rates.
Reifenstein (2005) reviewed several coke quality prediction models. The introduction of this report backgrounds the importance of coal rank and coal macerals on coke quality. The author then explored most coking indices, based on coal properties, used in coking coal trade.
General relationships exits between coke quality parameters and blast furnace performance, but these relationships are weak due to the complexities caused by
  • how the different macerals of the blend coals fused together which influences coke strength and fissure formation,
  • the role mineral matter plays during theplastic stage of coking and the impact of mineral matter on cake reactivity and
  • changes in the coke structure as it moves down the shaft of the blast furnace.
Coking properties are mainly affected by rank, maceral types and inorganic matter of the coal. Rank is best determined by the mean maximum reflectance of vitrinite (Rv,max) in the coal. Volatile matter on a dry mineral matter free basis is also an indicator of coal rank. Measurement of the plastic properties during carbonisation (e.g. Gieseler fluidity) is commonly used in evaluating component coals to be used in a coking blend, though there is some debate on how fluidity influences the important coke properties of coke strength after reaction (CSR) and the size of the stabilised coke.  
The table below lists the main quality parameters commonly used in the evaluation of coking coals.

Coke property
J.I.S. drum indices
AS 1038.13
Micum indices
Reactivity to carbon dioxide (CRI)
Strength after reaction (CSR)
Size range
AS 1038.18
Mean size
Size <25mm
AS 1038.2
AS 1038.4
Total sulphur
AS 1038.6.3.2
In blending coals for coke manufacture, the properties of the blend can be estimated as follows:
  • Rv,max is determined from the percentage-weighted average of vitrinite determined on the volume percentage.
  • assuming the log of the Gieseler maximum fluidity is additive, though this is not strictly correct for all blends
  • the proximate and ultimate analyses are additive.