Thursday, July 06, 2017

a 'Celtic Iron Age' Bloomery Furnace

If you have not been following recent posts, you might want to look back to see the development of this prototype...

As a fast introduction, this is my suggested furnace build for the demonstration upcoming at the Scottish Crannog Centre event, August 5 & 6, 2017

I have decided to keep to a known and proven layout :
- 40 cm stack above tuyere
- 20 - 25 ° down angle on tuyere
- at least 10 cm base (floor to tuyere)
- set the tuyere proud of inner wall

What will mark this system as 'Celtic' is :
- use of semi-drum bellows
- slag pit design

I see the most likely 'design flaw' is the available air volume from the 'semi-drum' bellows.
Because of limited workers, and the very real transport restrictions, I am going with a single tuyere / bellows combination.
To try to balance that, I've reduced the interior size of the furnace :
- at 20 cm the ideal air would be 375 - 470 LpM
- at 22.5 cm the ideal air would be 500 - 620 LpM
My theoretical output on the semi-drum bellows I built is about 500 LpM. So this suggests with the smaller furnace diameter (and good workers!) this system should function as hoped.
Honestly, I feel that creation of a lacy, lower yield bloom is more likely. Not really a problem of itself, as this is more like to represent actual Celtic Iron Age results.
There is a balance between interior furnace volume (heat created) and exterior surface (loss through radiation). I have successfully run furnaces at 20 cm ID before. There is a tipping point some place between 15 cm and 20 cm where there is too much loss off the exterior against limited heat volume in the interior.

I'd like to try the slag pit filled with bundled grasses. The ideal (from an experimental archaeology stand point) would be to be able to cut a sheaf of oats or barley. I have run this same system (slag pit with thin clay cap) a couple of times using small sticks - with good results.

The leather Y tube is something first made up for the Vinland Series, originally intended to replace the modern steel pipe fittings that normally are used. (The purpose of the Y is to allow a straight line down into the tuyere, allowing for visual inspection and clearing of blockages as needed.) A second advantage proved to be providing a flexible coupling between moving bellows and static (and fragile) furnace. Working in a more historic context, a simple wooden plug seals one branch of the T. Sound then becomes the indicator of when the slag builds too high or freezes in a drip, either case partially restricting air flow.

Monday, July 03, 2017

Iron Smelting in the Celtic Age (three PLUS)

Continuing from the post on June 23...

Thijs van de Manakker returned my inquiry on some technical details of the smelt series seen on the videos documenting his process.
(Remember Thijs' primary language is not English! Any modifications / comments in italics)

1) The height of our furnace is 80 cm.
Interior diam bottom 38 cm. (a)
Interior diam top    27cm.

2) The slagpit is under the whole furnace.

(the third image shows the 'basket' covered with a roughly 1 cm layer of clay)

 3) One bellow is 36 Liter is 1/2 furnace content. (b) (**)
A furnace with two tuyères with Stiphout bog ore runs best at 17 strokes (of 2 bellows) per minute  17 x 72 Liter = 1224 L/min.
With that tempo we can charge every ten minutes: 1,5 KG (6 L) charcoal and 1,5 KG ore (2L), so  both charcoal and ore 9 KG /hour. (c)

4) The furnace on the film was not documented on paper, so I don't know the exact figures.
The average yield of our furnace is 15% raw iron . (d)
The highest amount of ore we ever could load was  103  KG .
Bloomforging gives 30% lost of iron, so bar yield of the furnace is 10%. (e)

5) Our blooms have  0,06 % Carbon. (f)
To get steel  we double an iron bar several times, to get rid of slag as much as possible, than the bar of 5 mm thickness  is packed with charcoal powder inside a  loam (mud)  cover and fired in a woodfire for 25 hours.

Thijs van de Manakker.
the notes / comments

(a) At 38 cm the interior cross section will be 1330 cm2
Our standard furnaces run between 25 - 30 cm ID, so 490 - 710 cm2
(see how even a small increase in diameter really will impact air requirements)

(b) My original WAG was really off on the math and actual reported numbers! I had estimated the furnace with a bigger diameter and a lower produced volume on the bellows.
Using the 'Sauder & Williams Method', the 'ideal' air volume for this furnace would be 1600 to 2000 LpM. 

(c) There is still no standard used throughout the experimental community on exactly how to measure and compare burning rates.
Kilograms of charcoal per hour / minutes per standard charge are good indicators.
Between the two, I suggest minutes per kg (charge) is the single best. During the smelt itself, this measure also gives the smelt master the best indication of just what is happening within the furnace as the smelt progresses. The consumption of standard charges will certainly vary over the course of any smelt event.
This makes this experiment with an average burn rate of 6 - 7 minutes per kg.
Our standard is to aim for 4 - 6 minutes per kg. (The impact of higher air volumes).

(d) Given an effective furnace design, and suitable air volumes being delivered, the next big impact on smelting yields is the quality of the ore itself.
So its hard to say if the yield reported is basically due to the iron content of the ore (not given here).
Our normal expectation is for 20 - 30% return from ore to bloom. Larger ore volume smelts typically produce higher yield results. I do note that our own efforts using human powered bellows typically yield down into the 15 % range as well.
(A hint to hopeful experimenters : For good results, your ideal ore should have +50% Fe / +65% Fe2O3. Sticking to a single ore type / source really helps your initial learning curve!)

(e) Bloom to Bar represents a loss that some people completely forget about in this whole process. Remember the objective of ancient iron masters was the * bar * - this the finished product then sold to the blacksmith (a separate craftsman).
Others working with historic methods have reported the same kind of results (roughly ore to bar = 10% return) The extensive pioneering work by Peter Crew should be mentioned.
I (shamefully) admit that I have not converted so many of my blooms up into working bars. Although I have been attempting to record some of the data of this progress systematically, this project lags (!). Generally I am getting much the same loss rate of about 30% from bloom to bar. Lee Sauder has told me his results are in the same loss range.
Our overall ore to bar results are a bit better, as the high volume air method normally used does increase the yield at the ore to bloom stage.

(f) My own data on iron content is far more varied - and honestly I would suggest less reliable. Although the potential sample size is good (see table o' blooms 2012), I have used a large number of quite different ores, with the expected variations in results. Our objective is always a 'nice soft iron', sometimes higher carbon (into the range of 0.5%) has been the result.

(g) Slap on the forehead time!
I have a copy of this paper (which I suggest all read) via Early Iron Production, Lars Nørbach (editor), ISBN 87-87567-40-7. I had read this back in my initial research work into bloomery iron smelting, well over a decade ago. (It is where our use of the 'thumper' came from!)
The volume however has a mistake in naming authors in the primary index, swapping the contributors of two of the papers. So I ended up missing this reference when I started researching this specific project.

(**) A later addition
I was thinking about what Thijs had said here.
My standard, based on Lee's work and example, has been to set air volume via the formula V = (ID area x 1.2 to 1.5) as LpM

I have never considered applied air against internal furnace volume.

The reduction process is a 'gas over surface' reaction. Although increasing air will create higher temperatures, the space inside the furnace actually involved in the reduction process is actually more important. (That is why the stack height is important. Too little and there is not enough space to utilize the reactive gases. Building too high does not greatly increase the furnace effectiveness.)

That being said - there may be some revealed wisdom in what Thijs has said.
Bears checking back with my own past furnace builds and smelts? I will be curious to see if bellows volume to furnace volume has any relationship...

Saturday, July 01, 2017

DARC returns to VINLAND - (past view 1)

DARC will be returning to Vinland!

Ragnar Ragnarson will once again be gathering his band of friends and heading of into the West. (You think by now we would have learned not to trust his navigation skills!)

Members of DARC will be expanding the regular Encampment program at L'Anse aux Meadows NHSC from July 15 through 23, 2017. The highlight will be July 16 - with a full re-creation of the first iron smelt in North America (originally undertaken by Leif Eirikson's crew some time about 1000 AD).

To give you a hint at what you might see in this special presentation for Parks Canada and Canada 150 - here are some images from past voyages:


All images by Paul Halasz - © 2010

Friday, June 23, 2017

Iron Smelting in the Celtic Age (three)

In casting around for prototypes for the upcoming demonstration project at the Scottish Crannog Centre, I keep coming back to the work of Thijs van de Manakker.

Thijs works at / with the Eindhoven Museum in the Netherlands. This is a living history museum, centred from pre-history to the Medieval period.
One of the activities there, which Thijs has lead over the years, is experimental iron smelting based on the Celtic Iron Age. You may note that the video record below is from 1999 - two years before I even started inv

(Below ported over from YouTube - you may have to click on the title to get the intended content. The full set of videos are on Thijs' web site.)

Looking over the process illustrated above - these are the things I notice (*):

1) Build : The mix is a fairly rough blend of chopped straw, what appears to be locally dug clay, and sand. The consistency is softer than I normally use, with what appears to be a sequence of thick rings applied, then left to sun dry to firm up before the next layer is added.
- This method results in very thick wall, with clear breaks between the layers. The outside is clearly blended, but the interior has been left very rough. I did wonder at this when I initially watched this first video (but the logic becomes clear later).

2) Layout : The completed furnace appears to stand roughly 70 cm tall. It is clearly flask shaped, looking about the same outside diameter at the base as the height.
- Given the thick walls, this suggests an interior diameter at tuyere level of about 50 cm, perhaps 30 cm at the top opening. (All WAG).
- The furnace is a slag pit type (seen briefly at the start of the construction phase, later in the smelt when slag is drained). It looks like this is a smaller pit, lined with sticks, is placed to the front of the furnace below the tap arch. (Rather than a full pit under the entire furnace?).
- There are two tuyeres, set at base level, opposite each other and so also 90° to the tap arch. These are both basically set dead flat. (I would be concerned about slag levels.)
- The actual tuyeres appear to be lengths of modern steel pipe. These are quite long, the purpose appearing to keep the bellows operators well back from the furnace itself. (Likely done for both safety and to keep the working area around the furnace clear. This becomes especially helpful during the extraction phase.)
- There is no specific way to tell if these were set proud in the interior - or how far they may have burned back during the smelt.

3) Air : There are two good sized leather drum bellows supplying air. Obviously there is a flap input valve on the top. It is not clear if there is an exhaust valve.
- From the video you can see the pump rate is about one stroke per second each.
Also that the two bellows are being blown identically, with one operator setting the pace, the second (less experienced?) following.
- A (very WAG) guess on the size of each is about 30 cm wide, with about 30 - 35 cm height of air being expelled each stroke. This (extremely WAG) suggests about 20 litres per stroke / 600 litres per minute / 1200 LpM combined. Taken against that estimated 50 cm ID, I would consider this marginal air volume, and likely to effect the bloom yield and density.
(corrected against bad original bad math!)

4) Slag Management : The tuyeres are set up fixed and in a straight line. (Not the Y tube with removable plug / view port that we use.) This means all the slag control is in the hands, and ears (!) of the smelt master.
- As the tuyeres are set flat to the ground, this is certain to result in the slag bowls forming high, requiring careful monitoring to ensure the tuyeres are not 'drowned'.
- Given the lower volumes of air blast, I would expect either two separate slag bowls, or jointed to a lobed shape (like a kidney bean), shallower in the centre.
- In the later stages of the smelt, you can see several slag taps. The bowl is punctured at the front edge - I would suspect on a diagonal back to the individual slag bowls. The excess slag runs to drip into the front pit.

5) Extraction : The logic behind the construction style of the furnace, with the corregated interior, becomes clear here. The lines between the individual rings of clay are easily broken free, allowing the furnace to be dismantled in easy steps as the charcoal burns down.
- The use of the wooden chisel tool is a nice touch. A bit lost when the clearly modern metal tongs are used to pick up the clay pieces. (But certainly much easier than using a more historic wood shovel or wooden folded tongs.)
- Once the charcoal is burned down close to tuyere level, you can clearly see the 'bright spot' outlining the two separate bloom masses. As expected, these are located just to the front of each of the two tuyeres.
- On the extraction and first quick cleaning / compaction of each, it is clear that the iron produced is mainly a number of smaller pieces most significantly quite lacy in consistency. (Note how much slaggy 'mother' breaks away on hammering, and how quickly the hot iron core collapses into a very small mass.) The last piece extracted appears to be as large as all the others combined. I'm not surprised it was formed on the tuyere side of the 'lead' bellows operator.

6) Compaction : The use of the hand held stones initially is certainly a 'primitive' touch. I'm not sure exactly how accurate that might be historically? (Use of hand held stones was recorded in African traditional method, but wood hafted stone hammers are certainly part of the archaeological record by the Celtic Iron Age.)
- The use of a second, quite different, 'slot' forge is clear for the compaction phase. - Quickly the hammer stones are abandoned in favour of metal head sledges.
The piece being worked is appears to be the largest iron mass extracted. Over the sequence you can see it compacting nicely to a dense iron billet. It certainly looks to be working up like good soft iron, judging from the effect of the hammering!

The end result (2 kg?) is worked to resemble the 'double pyramid' trade bars known from the Celtic Iron Age.

(*) Although I know Thijs loosely through past e-mail conversations, I freely admit at the point of writing this I have not specifically approached him on more accurate details. Specifically for exact measurements and things like consumption rates and yields. I will be doing so!

WAG = Wild Ass Guess

February 15 - May 15, 2012 : Supported by a Crafts Projects - Creation and Development Grant

COPYRIGHT NOTICE - All posted text and images @ Darrell Markewitz.
No duplication, in whole or in part, is permitted without the author's expressed written permission.
For a detailed copyright statement : go HERE