Caherconnell Iron Smelt - Sunday August 27

Team : Daniel Jeffery (lead)
    Colilin O Duhghail
     Darrell Markewitz

Furnace : Irish red clay with approximately 1/3 course sand
    (Initial build by inexperienced workers, resulting in considerable distortion from planned design of 25 cm ID.)
    Interior diameter = 35 x 30 cm, oval long axis from tuyere
    Top Interior diameter = 29 cm, circular
    Shaft height = 69 cm
    Stack height = 55 cm
    Base depth = 11 cm
    Packing = 10 cm of charcoal fines as bowl shape (before pre-heat)

    Tuyere : supplied copper / ID = 2 cm (furnace end)
    Insert = 4.5 cm
    Angle = 20° down

Air : Supplied boxed vacuum blower / volume unknown
    It was found later that this unit was set to only allow about 3/4 available input air by earlier users.

Ore : Irish Antrim - red hematite
    Ore was roughly roasted, then broken into ‘pea to rice + dust’ size

Total Ore : 25.5 kg
    Ore was added in ever increasing amounts, starting with 1 kg charges, to 1.5 / 2 kg
    Ore charges broken into smaller additions, each applied with partial charcoal amounts

Total Charcoal : 26 kg, mixed Irish types (primarily ash?)
    Added as roughly 1 kg charges (measured by marked volume)

Main Sequence : 120 minutes, using 14 kg charcoal
     average charcoal consumption = 7.6 min/kg,
     increasing from 6 at start to 9 at end
     average ore addition 4.8 min/kg

Return : Bloom = 1437 gm + fragmented gromps totalling 1025 gm
    Yield = 6 % (bloom only)
    No obvious bloom at initial extraction, cold slag mass was broken to free contained metal

image 1- Working team (R-L) Daniel / Darrell / Colilin
Image by Kelly Probyn-Smith


It is worth stressing that experimentation at earlier Furnace Festivals had established both a furnace design and working process that suits the unusually rich Irish bog iron ore (at 93% Fe2O3 / 65% Fe). Primarily due to the experience of Jens Olesen from Denmark, these furnaces are very short, at 30 cm stack height, use the bellows plate and blast hole air input system, and run 25 cm ID (at tuyere level). These would be the bulk of the furnaces built in 2023, with this ore the primary one employed. Additionally, ore amounts were kept small, 10 kg addition being the most common. The overall results of these smelts has not been reported publicly at date of writing, but casual observation suggested 1.5 - 2 kg appeared most typical (sometimes to 3 kg), with the masses being crumbly in texture. (1)

This smelt was an attempt to test one of the unproven ore types made available at the Festival in 2023, a red and visibly sand containing stone material described as Antrim hematite. Earlier assessment of the ore (roasted sample) indicated a low available iron content, with Fe2O3  at 59% , SiO2 at 16%, and Al2O3 at 7% +. (2) This represents an available iron content of only 41%, which based on past experience is considered marginal at best. The high combined slag creating components suggested that considerable slag would be generated. If an iron bloom could be created from this ore at all, certainly a low finished yield would be expected.
Both Daniel and Darrell have experience with lower iron content rock based ores, and specifically with slag tapping furnaces using inserted tuyeres.

The furnace used was one built on Friday, then fired on Saturday, by two complete newcomers to iron smelting, who would not accept instruction, and so operated without any significant guidance. (3) As a result, the furnace as built had a significantly larger interior cross section than the 25 cm ID standard (at 490 cm2), the 30 x 35 cm oval creating 824 cm2 for this furnace as constructed. As it has been well demonstrated that ideal air volume input is related to the area at tuyere level (4) and the available air blowers produced a fixed amount only, it was fully expected this build distortion from the plan would have a negative impact on bloom formation. The initial build had been over the top of the remains of a furnace used in 2022. This had a bowl shaped base formed of clay, instead of the flat lower chamber set on the earth typically used by members of this team in the past.

The furnace had suffered considerable damage after it’s last use on Saturday, with the bellows plate broken way, leaving a large hole in the bottom. Several loops of fencing wire had already been added to brace up extensive cracking from earlier use. After some discussion, it was decided to re-set the air input to use one of the copper plate tuyeres that were on hand, using a layout well understood by the more experienced members the team.
The first step was clearing out slag remains in the interior that had accumulated from the smelt attempts the day before. The lower bowl shaped depression was filled with charcoal fines to create a flat surface, adding this packing has been found to make eventual breaking away of a slag bowl, leading to bottom extraction of the bloom, easier. As well, errors in working with the damp clay resulted in the lower part of the furnace slumping considerably during construction. The effect was that there was a limited base depth (below air input) of only 11 cm available, a measurement considered marginal for correct slag bowl and bloom formation, especially if any volume of slag was produced from the silica content of a rock based ore. In expectation of needing to tap accumulated slag, a small port was cut into the furnace base, plugged with damp clay.


image 2 – furnace as re-designed, during pre-heat
tapping port just to the right of the scale, sealed with wet clay

The furnace was given a fairly standard pre-heat period using split wood, for about one hour. After being filled with charcoal, another six, 1kg fuel charges were added to allow the interior to fully ignite and then establish a consistent burn rate, establishing about 6 – 7 minutes per kg consumed. The first five charges of ore combined with 1kg charcoal were added as 1 kg amounts, then three as 1.5, the following eight as 2 kg amounts, then one last charcoal addition and allowed to burn down. Into the later additions of 2 kg ore charges, the burn rate dropped to average 9 min/kg, likely a reflection of the larger furnace cross section against available air volume.
Later in the main sequence, the furnace would expel contained slag through two separate self tapping events of typical black and fluid iron rich slag. There was no observed impact on air flow from accumulating slag, so it was decided not to intervene beyond this natural process.
Late in the burn down phase it was observed that the portion of the charcoal on the side away from the tuyere was not fully ignited, as might have been expected by the large cross section against the air volume available.


image 3 – opening up the bottom of the furnace after removing tuyere
image by Kelly Probyn-Smith

The entire extraction process was drawn out, resulting in the interior cooling below what would be the ideal, causing slag to congeal. An attempt was made to compact the hoped for bloom mass in place inside the furnace, using a roughly 7-8 cm diameter length of pine fence post. Although the process of a top extraction continued, no clearly defined bloom could be located. In the end the tuyere was removed, the front wall broken open, and a large slaggy mass was extracted from the bottom. When hammered on a compaction stump this proved to be little more than slag and gromp fragments.
Although the team was greatly disappointed at this result, the next morning it was decided to pull free the mostly cool lower slag bowl and break it apart. The result was the recovery of one small but significant metallic bloom mass at 1.4 kg, plus the collection of a quantity of gromp fragments. The bloom clearly still contained voids and considerable slag, this partially due to the lack of any initial compaction.


image 4 - final iron bloom recovered
likely the upper surface as placed inside furnace, wall side and tuyere to top

Conclusions :

The lobed shape of the bloom as seen in image 3 does suggest it formed somewhat too high in the interior slag bowl, with the air blast starting to cut away the iron mass as it was deposited. The crescent shape also suggests that there was limited penetration of air into the interior, likely an effect of the expanded interior measurements against what was too low a volume (perhaps even pressure) of the available blower.


image 5 : bloom sliced down the long axis
image by Daniel Jeffery

Before any compaction work was undertaken, the volume of the bloom was measured via water displacement. This gave a result of 380 cc for 1404 gm, so a density of 3.7 gm/cc. This is significantly lower than typical results (5), but it does need to be remembered that this bloom had not been subjected to the normal initial compacting on extraction.
 As the first step of the compaction to bar, the bloom was sliced using a cutting disk on an angle grinder.  Important to any assessment of overall results needs to be not only the effective yield of ore to bloom size, but also some examination of the qualities of the bloom produced. (Remembering that the desired final product of the smelting effort is not a bloom, but a solidly compacted working bar.) This bloom was retained by Daniel, with the expectation of measuring volume and so calculating density, spark testing for carbon content, and recording the returns on the eventual compression into a solid bar. (6)

At best, this specific test smelt can only be considered a very rough proof of concept. That the ore used is able to be reduced into an iron bloom is demonstrated, but the low available elemental iron content does not suggest this is a truly effective ore type, especially compared to other, far richer, available Irish ores. The uncontrolled size of the furnace, complicated by a limited air supply, suggests at the very least some experimentation remains to determine an ideal furnace design to achieve the best possible results. It is suggested that the silica component of the hematite ore provided will certainly generate significant slag, likely to require a slag tapping design. It would be expected that the less porous consistency of the hematite will also require a taller stack height, a minimum of 40 cm, but ideally 50 cm, as proven with other attempts smelting with similarly poor materials. Even still, past experience suggests that overall yields are more likely to remain in the range of less than 10 % for ore amounts in the 25 - 30 kg range.

Although iron can be produced using the Antrim hematite, the low yields expected do not balance the cost in effort and other expended materials.

Appendix A – Field drawing of furnace
Appendix B – Sequence data

 1) Individual teams were left to their own decisions about record keeping, with the overall result that few teams made any records (nothing was available at time of writing).
 2) Data via analysis reported by Paul Rondelez in a group e-mail sent 7/11/23
 3) The two individuals working this furnace on Saturday failed to make any recordings of process or results. Although they appear to have produced several tiny bloom masses, it is significant that they were working exclusively with the massively pure Irish bog iron ore.
 4) See Sauder & Williams :
 5) For comparison, see ‘How Dense are You’ :
 6) Measurements of density and recording of the losses at bloom to bar are considered of importance, as the quality of the other blooms created at Caherconnell has not been clearly assessed. Simple observation of surface textures, combined by the results of compaction attempts by highly skilled blacksmiths at the event, suggests that bloom quality against raw yield numbers needs to be correctly assessed.

        Two bloom fragments, an estimated 750 gms total, were recovered from the working area on Sunday morning, almost certainly discarded by those working this furnace on Saturday. These are highly fragmented in texture. A report on the qualities and eventual compaction of the larger of the two (at 570 gms) is under preparation.

© Darrell Markewitz - 2023