Iron production weekend Experiment - May 2003

Date: May & June 2003

Location: Vinderheima

Premise: If they made iron -- then so can we using the same techniques

Conclusion: We can't do it this way either but this is much closer.

The Team:

Staff	Dave Kevin Gus
Recorder	Neil Peterson
Leader	Darrell Markewitz

Reports of all of our iron smelting efforts along with more articles and information are available on the "Iron Smelting in the Viking Age" CD from the Wareham Forge.  Copies of the CD can be purchased here.

This year marked our second attempt to produce iron from ore. We made some changes from the previous year. Several of our people had been down to visit Skip & Lee. Based on some of their results we altered the furnace design and attempted to narrow the number of variables facing us.

This time around we eliminated the issue of randomly variable airflow by using a vacuum cleaner motor. To allow a controlled but variable air flow a metal plate over the intake was used.

The smelter itself was constructed differently this time using as a template a roman smelter that came into in Scandinavia use towards the end of the Viking Period. We began by digging a small level area and a slightly deeper area for the run off. The first layer of the furnace was a collection of firebrick. On top of this was laid a pad of jewelry investment (think high temperature plaster of paris). This pad was shaped and sloped slightly to aid the flow of slag. When it was dry we used a rolled up sheet of metal to provide a form. 3.7 to 5 cm of refractory concrete we laid in around the form. Outside the concrete we laid on a layer of cobb. This was a blend of clay, straw, and sand. 4 to 6 cm of cobb were used. A space was left in the front to fit two firebricks as a removable door. The smelter itself was built a couple of the weeks before the experiment and allowed to air dry for that time (covered during any rain).

Photos of the Smelter build weekend

Chopping straw	Internal form	Mixing for size & consistancy	The cement and cobb overlays
The tower grows	Finished smelter drying	Smelter from top	Front of the smelter
Overview of the smelter	Air hose and viewing hole

The day of the Smelt

At this point we split to work two sets of tasks at the same time. Preparing the ore and preparing the smelter and charcoal.

This is a good place to express our thanks to Royal Oak Charcoal who donated the charcoal used in this year's experiment. We have used Royal Oak Charcoal in both our experiments as they are a hardwood lump charcoal (pretty much what the Norse would have produced) and their quality is consistently high. From all of us -- Thank You!

Charcoal	Overview	Adding charcoal

Preparing the Smelter and Charcoal

We began with a small fire to gradually heat up the smelter then added charcoal (at 1349h) and turned the air flow on to bring the temperature up. We adjusted the airflow (to 3/8s flow at 1401h, then down again to 1/4 at 1412h, down to 1/8th at 1418h and back up to 1/4 at 1420h). Charcoal was added repeatedly as this was one hungry burn -- half buckets were added at 1404h, 1411h, and 1419h. A temperature check with asurplus thermocouple was tried at 1414h but we couldn't get a reading.

The charcoal itself was screened to a consistent size (about 5cm diameter) larger chunks were broken up. We screened and prepared 55 Kg of charcoal and came quite close to burning it all. A small addition of cobb was made at 1359h around the doors as we were noticing some flames coming through.

Preparing the Ore

While the smelter was heating we weighed out 18.8 Kg of rock ore and set up a stack fire to roast the ore. It was run through a good long fire then pulled out and broken up to a 2cm diameter size. We weighted out 6 charges of 1.36 Kg of the roasted ore. This left us roasted ore left over for future experiments.

When the ore was pulled from roasting fire we had three options. We could have dumped it all in water which would have broken the ore up and cooled everything down but then we would have had to wait to dry it before the next step. We could also have added gasoline to the fire and burned the remains hard to remove all the remaining charcoal leaving only ore -- this also would have required more time. Instead we pulled out what we could as chunks & used all the charcoal (and no doubt small bits of ore) in the smelter. This produces two errors in the ore weights (in the total amount of roasted ore and in the ore added to the smelt) we do not feel, however, that either weight is in any way significant.

Starting pile of ore	Layering wood and ore	Starting the burn	Cooling ore
Crushing ore	Weighing ore

Then it was time for the actual smelt.

Actual Time	Elapsed Time	Notes
1424h		The first charge was added along with a 1.5 Kg of charcoal.
1435h	T+0:11	The charcoal leftovers from the ore roasting -- along with no doubt some ore was added.
1436h	T+0:12	The second charge with 1.5 Kg of charcoal was added
1439h	T+0:15	0.75 Kg of charcoal was added.
1448h	T+0:24	3 Kg of chorcoal was added.
1452h	T+0:28	The third charge was added after another 3 Kg of charcoal.
1457h	T+0:33	We altered the angle of air pipe by raising the back end 1 cm. Some more cobb was added to adjust the seal of the pipe.
1501h	T+0:37	3 Kg of charcoal added.
1506h	T+0:42	We noticed a small amount of slag appearing by the fire bricks.
1508h	T+0:44	The fourth charge was added along with 3 Kg of charcoal.
1512h	T+0:48	The tuyure was cleared of slag by rodding it out.
1517h	T+0:53	The tuyure was cleared of slag by rodding it out.
1520h	T+0:56	Another 1.5 Kg of charcoal.
1523h	T+0:59	The fifth charge and another 3 Kg of charcoal was added.
1528h	T+1:04	The tuyure was rodded out.
1535h	T+1:11	3 Kg of charcoal was added.
1538h	T+1:14	The tuyure was rodded out.
1539h	T+1:15	The tuyure was rodded out.
1540h	T+1:16	1.5 Kg of charcoal was added.
1545h	T+1:21	The tuyure was rodded out.
1550h	T+1:26	The air was run up to 1/2
1552h	T+1:28	Air down to zero and the doors were opened to let the slag out. We were overjoyed to see liquid pour out (ok perhaps pour is an overstatement as the first slag tapping was rather viscous). In hindsight it was our guess that this was the still liquid parts of the bowl itself flowing, rather than the liquid slag that forms above of the bowl. Some was set aside as samples the rest was (as planned) added back into the top of the smelter. The doors were closed, air returned to 1/8th, and another long burn occurred.
1556h	T+1:32	The tuyure was cleared again.
1601h	T+1:37	The wedge was removed from the tuyure to raise the air over the slag better and the air was bumped up to 5/16ths.
1606h	T+1:42	1.5 Kg of charcoal was added.
1610h	T+1:46	Air was adjusted to 1/2 flow.
1614h	T+1:50	Air was adjusted to full.
1617h	T+1:53	1.5 Kg of charcoal was added.
1620h	T+1:56	Air was adjusted to 1/2.
1623h	T+1:59	1.5 Kg of charcoal was added.
1624h	T+2:00	The air was set to zero, the doors were opened and out came the slag (much runnier this time). This time we think we actually tapped into the slag we wanted. As of this point things were going exactly as expected. The slag was sampled and the remainder added back in to the top.
1632h	T+2:08	The air was turned back up to 1/2.
1636h	T+2:12	3 Kg of charcoal was added.
1637h	T+2:13	The doors were closed.
1641h	T+2:17	The tuyure was rodded out.
1644h	T+2:20	The tuyure was rodded out.
		Now we come to the painful part. Something went wrong. We aren't sure why but the slag pool built up enough to partially block the airway. This deflected the air upwards allowing a cool spot to form immediately below the airway. This caused more blockage of the airway. In addition the deflected airflow produced a hot spot above the airway. In this picture you can see the beginning of the burn through. We reacted to attempt to block the burn through but it was in vain. All that came of this was a heat sink that reduced the smelter temperature to a point that would not finish the smelt.
1650h	T+2:26	More charcoal was added.
1701h	T+2:37	Another 3 Kg of charcoal was added.
1705h	T+2:41	The air was raised to 1/2.
1708h	T+2:44	2.25 Kg of charcoal was added.
1713h	T+2:49	The next tap was done with a good slag flow.
1716h	T+2:52	Closed the tap hole and added 1.5 Kg of charcoal. There had been a little slag from the first tap sitting without coming out. In this tap it came out. This let us have a comparison of the first and third taps side by side. The third tap was noticably less dense.
1725h	T+3:01	The turrie was cleaned again.
1727h	T+3:03	1.5 Kg of charcoal added. At this point we became concerned that the volume of slag was so large that it might not be possible for the reducing charge to work. We decided to tap off some of the slag to make room.
1737h	T+3:13	We added more charcoal and began the tap. Unfortunately this tap really didn't produce any slag.
1741h	T+3:17	We added the sixth (reducing) charge. The tuyure was cleared again then charcoal added to top up the furnace. Three more turrie clearings were done in the next 9 minutes.
1756h	T+3:32	Another 1.5 Kg of charcoal was added.
1759h	T+3:35	The air was cranked up to full as air did not seem to be moving through the turrie due to the damage to the wall.
1800h	T+3:36	We tried another slag tapping to lower the level and try and get air through the turrie. This time we did not put the top brick back in.
1806h	T+3:42	Another 1.5Kg of charcoal was added.
1814h	T+3:50	The bottom brick was removed and a tap done. This did reduce the level clearing the turrie and letting the furnace begin to roar again.
1816h	T+3:52	Another 1.5Kg of charcoal was added.
1826h	T+4:02	We fired up a propane forge to take the bloom and keep it hot -- as you will see though we never did find the bloom.
1830h	T+4:06	We went to do the final pull of the slag. None came pouring out and we had to reach far up into the smelter attempting to pull any out. It was, however, too cold. We opened a larger door in the smelter to pull out the slag and bloom. (Next time we need to have two doors -- a small one for the slag and a large one for the bloom). Although we got a large cascade of very hot charcoal that wasn't quite what we were after. The smelter was knocked over. It broke apart and some materials were removed. Samples of these have been sent out for testing. All appear rather glasslike but their weights are shockingly different. Clearly some are "pure" slag while others have a high metal content.

A layer of ore	Adding more charcoal	Air inlet pipe	First sign of slag
First slag tap	Thicker slag	Second slag tap	Repair the burnthrough
Repair overheating	Pulling Slag	Cascade of charcoal	Cooling down
Smelter tipped over	Base pealed away	Smelter cooling	Pieces cooling

Here are some "after" photos

End of the airpipe	Inside the fallen smelter	One of the burnthroughs	The open interior
The pipe protruded 5 cm	Wall with cement and cobb	Wall from another burn through	Concrete and cob piece
A view down the pipe	Piece showing different heats

Conclusions

The furnace shape and size seemed to work fine. We were burning about 1.5 Kg (about a gallon) of charcoal every 12 - 15 minutes. The total charge of ore was about 18 lbs - added in 3 lb lots. On the face of it, the furnace seemed to be working fine. We got liquid slag at the point we expected. This material was pretty viscus, and still had considerable iron content. Over the second  tap of slag, the material become more fluid, more glass like and far less dense, At that point we were pretty excited - as we knew the iron had to have gone somewhere.

The problem: First, the liquid slag started to pool up high enough to clog the air intake. As the air was hitting the liquid slag it was solidifying in a curved shape extending from the bottom. The hole for air moved towards the top of our pipe - in fact deflecting the air up. This was despite poking with a rod - eventually the air pipe was actually inside the slag pool.

We're not sure why the 'bowl' formed so high - there was about 25 cm between the floor of the furnace and the air intlet. It may have been that the angle was not directed down enough, or perhaps the air flow should have been higher. We came to the air volume based on Lee and Skip's notes and visible consumption rates. We were adding charcoal every time the level dropped down about 6 - 8 inches in the stack. We tried 'full blast'  for a minute or two, and you could see the fuel level dropping. My guess is at that air flow we would have been adding 3 lbs charcoal every 3 - 5 minutes. Our intake pipe was made of straight 1" ID steel (schedule 40) - if the end had been slightly crimped down to closer to 3/4" we might have had better delivery  pressure.

What seems to have happened is that the iron rich slag tapped out eariler was becoming semi molten and then re-freezing when it hit the spot where the furnace wall burned out. The slag poured out near the end of the process was likely liquified furnace wall.   There was a large mass of 'gromp'  attached to the furnace wall at the burn out point. We are hoping we may have something like spongy cast iron mixed in with the gromp. We did test over the hot surfaces of the grapefruit sized gromps with a magnet, and the pull is incosistant (which gives some hope).

We did learn a lot in terms of physical process and set up. The overall design of the furnace is sound - but needs some tuning in details (wall thickness / air intake angle / air pressure)

Reports of all of our iron smelting efforts along with more articles and information are available on the "Iron Smelting in the Viking Age" CD from the Wareham Forge.  Copies of the CD can be purchased here.