Friday, June 26, 2015

Icelandic 6 - excavated

Monday, with the remains of Saturday's smelt expected to be cool, I 'excavated' the remains.
Top view (tuyere to left)
3/4 view (tuyere to left)
Neil Peterson's correction from yesterday (see comments) - that the earth was seen to collapse significantly more as the furnace was cooling, had continued overnight. The earth surfaces would fall away at the lightest touch. This would distort the inner surface from the shaft size present during the smelt itself.
Some idea of how much material collapsed is seen in the 3/4 view above. At the rear of the shaft (opposite the tuyere) as much as 5 - 6 inches of material had collapsed in. The 'average' collapse was bout 3 inches. This material was baked visibly to a reddish brown colour (all organics missing).

Loose debris removed to expose slag mass (tuyere to left)
 Eventually I was able to scoop out all the loose debris (ash, charcoal, loose slag pieces, collapsed earth).  For the archaeologists, this was done by hand 'polishing' the bottom surface of the shaft. I did not touch the walls themselves other than to remove a couple of the top stones (to keep them from falling).
As expected from the burn pattern during the smelt, there were larger charcoal pieces remaining in the back 1/3 of the furnace. (Some of these were larger pieces from the first filling with ungraded fuel - and some still burning slightly.)

To free the slag mass, the front area above the tuyere was cut away. First the cap stone was lifted, then the sod pieces over the tuyere down to the tap arch top stone were cleared.
Then the sods / earth from the sides were cut back to form an open slot, down to the top lintel stone of the tap arch. At this point the slag mass was removed as one piece. 

Front view (towards the tuyere)

Rear view - from the back of the furnace

Side view - Right from the front of the furnace

Top View (tuyere seen to left)
Field drawing with measurements (imperial / not to scale)

What remains is a very large mass - the expected bloom still locked inside the slag.
• The bowl itself has formed a bit lower than expected. The 'soft base' of charcoal fines was set at 4 inches / 10 cm below the lower edge of the lintel stone, but you can see the slag burned down a further 5 cm. This puts the bottom of the slag bowl roughly 10 inches / 25 cm below the bottom edge of the tuyere.
• The usual hot spot in an oval pattern around the tuyere has resulted in a wall of slag, fusing earth behind it. It is a circular pattern, roughly 7 inches / 18 cm above and about 8 1/2 inches / 22 cm to each side of the tuyere. This wall varies in thickness, thinner at the top and side edges, to about 1 1/2 inches / 4 cm at its centre.
• The slag bowl penetrated to about 9 inches / 23 cm from the front of the furnace. This left a gap about  5 inches / 13 cm from  the back wall. Some fragments of partially sintered, but reduced, iron were gathered from this rear area. It does appear however that the burning pattern inside the shaft did in fact pull ore into the heat zone (no ore fragments were recovered).
• There is a central pillar of slag rising above the normal flat surface of the slag bowl. This pillar is about 4 inches / 10 cm wide at the base and about 6 inches / 15 cm high, coming to a rough point.
It is thought that this slag was dragged up off the liquid slag bowl during attempts to hook the edge of the bloom during the failed top extraction attempt.
 • The ceramic tuyere has a final length of 6 1/2 inches / 16.5 cm, from a starting length of 8 inches / 20 cm. This represents only about 3.5 cm of erosion, more typical has been 4 - 5 cm in past smelts.

Almost all of the slag has been recovered. There are surprisingly few 'gromps' (partially sintered and reduced iron encased in slag that did not fall to incorporate into the bloom).
It is likely that most of the slag present is the result of silica mixed with iron from the ore. In this case there was no clay involved, which usually does erode to add to the slag produced.  One remaining task is to weigh all the slag produced.

At this point it has been decided to ensure the size and weight of the slag mass is recorded. To determine the presence of the likely bloom, it will be necessary to break up the slag to free the metal.  (This most likely to be attempted over the weekend - not an easy process with the slag mass cold!)
The hope is that the comparison of ore against slag and bloom produced will be of some use in understanding the production cycle estimated at Hals.

Wednesday, June 24, 2015

Icelandic 6 - the Smelt

Here are some actual photographs from the smelt itself:

Down the interior of the shaft - with the metal form removed.

Furnace - front view. Stone framed tap arch below.

Working area - The build was framed 2 x 2 m by rail ties.

Start of main sequence. Ceramic tuyere installed with air supply. Tap arch blocked with sods.

Into main sequence, ore analog added.

End of smlet : View down the opened tap arch, showing the slag bowl in place.

Two views down the inside of the furnace at the end of the smelt:
Showing the variations in temperature.
Highlighting the hot spot of the bloom.
As was remarked in the comments on the last posting (by Neil Peterson) - As the charcoal level dropped during burn down, the earth walls were seen to collapse on to the top of the descending charcoal. The result was very heavy erosion of the earth walls.

One overall problem was a lack of full consideration of the diameter of the furnace - and what that implied in terms of both charcoal requirements and air volumes.
Our normal furnaces vary from about 25 to 30 cm internal diameter. This build started using a 33 cm diameter metal form (1). After the short (30 minute) pre-heat, the walls had shrunk back somewhat, resulting in an ID at the start of the smelt of  36 cm.

A) Effect on air delivery volumes:

ID = 30 cm : area at tuyere = 700 cm2
air required (at 1.2 to 1.5 l/m/cm2) = 840 to 1050 l/m
ID = 36 cm : area at tuyere =  1020 cm2
air required (at 1.2 to 1.5 l/m/cm2) = 1225 to 1530 l/m

With the start of the main sequence (charcoal addition)  the available electric blower was set wide open. (That volume is thought to be 1400 l/m, but this has not been tested 'in line'.)
Another variable is delivery pressure, which in this case could not be modified. (Pressure effects how far into the charcoal mass of the furnace the air will penetrate.) The concern was that even if sufficient air volume was available, the air might not penetrate right across the furnace diameter.

B) Charcoal requirements :
Although the height of various furnaces varies, internal volume is a cube function.
Our typical furnaces require 4 - 5 standard buckets (about 1.8 kg) of charcoal to fill.
For this build the volume of charcoal needed was almost double - a total of 10 standard buckets.
This would certainly effect the 'hang time' of any individual ore particle. (The time the material was inside the reduction chemistry of the furnace.)
The consumption of a standard bucket of charcoal was inside the normal range considered effective on past bloomery iron smelts. The furnace ran a bit hot at first, at a rate of 6 - 8 minutes per bucket. This shifted (as expected) with the addition of the first ore charges. The average was about 10 minutes per bucket, increasing slightly with the later additions of larger ore charges. Ore charges started at 1 kg per bucket, and increased smoothly to 2.5 kg per 1.8 kg charcoal at the final stages.

C) Tuyere :
One effect that was not anticipated was the shifting of the tuyere angle over the course of the smelt. At the start of the smelt, the tuyere was set to the standard 22 degree down angle. However as the smelt progressed, this angle flattened out considerably. By the 2 1/2 hour point into the main sequence, the angle was reduced to 17 degrees down. Measured at the end of the smelt, the tuyere angle was at only 12 degrees down.
This appears to have been caused by the sods collapsing as they dried / organic materials burned away. The tuyere had been placed so that the interior end was resting on the stone slab lintel of the tap arch, but the outside end was only supported by grass sods.
This shifting could have been prevented by placing a small stone support on the outside end of the ceramic tube used.

Next : 'Excavating the furnace & extracting the slag / bloom mass

(1) The reason for using the larger form was an initial intention to line the sod shaft with a thin clay layer. This was suggested by clay fragments found at Hals, ranging up to as much as 2 cm thick. With that clay liner, this furnace would have in fact had an ID closer to 30 cm. Time (and weather) limitations resulted in this clay liner not being added to this furnace.
A successful test of a thin clay liner in an earth cut cylinder was made with 'Icelandic 5', May 2012

Monday, June 22, 2015

Icelandic / Hals Iron Smelt - full build

Readers may have been following the recent progress of the bloomery iron smelt based on the work of Kevin Smith at Hals in Iceland. Hals was a long duration, 'industrial' level iron smelting site, dating from the Viking Age.

As described in recent posts, the experiment on June 20 was to build and operate a full version of the type of iron furnaces at Hals. These are basically a cone constructed of grass sods, the shaft dug into the centre of that cone. The cone would have been framed in a box made of timbers. The space above the cone would have been levelled off with earth, creating a flat working surface around the top of the furnace shaft.

In a departure from the usual reporting method, here are a pair of illustrations of the smelting furnace during the later part of the main sequence. Illustrations by Margaret Gissing:

• The interior diameter of the final build was in the range of 33 - 35 cm.
This represents a considerable (!) increase in volume over our standard layout (at closer to 25 cm).
• Stack height (at the start) was 50 cm.
• Tuyere was the standard ceramic tube (at 2 cm ID), set at 22 degrees down, about 6 cm proud.
• Air was supplied by the normal industrial electric blower, in this case wide open to maximum.

• A total of roughly 31 kg of slightly enriched DD1 ore analog was added.
• Total charcoal consumed was roughly 76 kg.
• Total elapsed time (from short pre-heat to extraction) was about 7 hours.
• Ore charging followed a fairly standard sequence, increasing smoothly from 1 kg up to 2.5 kg per charge.
• The consumption rate for a standard charcoal measure (graded maple at 1.82 kg each) was an average of 10 minutes.

• Initial extraction attempt was made via the top of the furnace. It was clear that there was a nugget of bloom present, but it did not prove possible to get a hook on to the bloom itself to pull it free of the slag bowl.
• At this point an attempt was made to extract via the tap arch. It became clear the slag bowl had solidly attached itself to the stones framing the tap arch.
• Tapping with a steel rod from the top suggests the bloom in place is a good size - perhaps as large as 15 x 20 cm. (Admittedly, this is a vague impression at best - and may represent wishful expectations!)
• A decision was made to leave the slag bowl with bloom in place. The overall intent of this experiment is to measure the functioning furnace so as to be able to gain insight how the excavated remains at Hals reflect the working furnaces there. This will be done best by more carefully examining / 'excavating' the complete furnace.

One thing that became obvious as the empty furnace cooled. The bare, dry earth walls started collapsing inwards as soon as the supporting charcoal burned away.

Expect a more detailed examination / photographs of both the smelt and the final furnace, shortly.

Wednesday, June 17, 2015

Return (really) to Icelandic Iron Furnace

(no - seriously this time!)

June 20 - Saturday
Organization start : 9 AM
Pre-heat start : 10 AM
Main Sequence start : 11 am
Estimated Extraction : 4 - 5 PM

Furnace is a full dress of the Icelandic / Halls grass sod construction
Ore will be enriched DD1 analog, roughly 25 kg
Air will be electric blower
Start of the build - Fall 2014
See more details from past postings

Icelandic Grass Sod Furnace  

'Working towards an Icelandic Viking Age Smelt'

This will be the sixth experiment in the larger Icelandic pattern furnace sequence.
(Interrupted by Turf to Tools - and ongoing personal disasters)

Open Event - please feel free to come.
There will be grunt work to be undertaken!

Followed up with BYO (pot luck style) BBQ

Wednesday, June 10, 2015

Norse Lock - Construction

From the earlier post on this project, I had mentioned the problem with using internet sources as references.
The illustration sent me does not actually show an actual artifact, but is merely a drawing showing the working mechanism.
How were the artifacts actually constructed?

Remember the source artifact that I have detailed information on is one from Coppergate at York :
Click on image to see it as life sized

Clearing away the decorative elements, I made a direct tracing of the 'working' construction:
Click on image to see it as life sized
There is reference in the written description of this lock of bronze braising being used in the construction. This most certainly applies to the method that the various small forged iron wire decorative elements were attached to the outside of the case.
The combination of wall thickness and smaller circular diameters of both the main case body and the upper shackle tube would mean that these would not basically require any solid joint along their seams. These could of course also be bronze braised - but from the description if this was originally done is not clear. (A colour photo might show this - and examination of the actual artifact certainly would indicate it!)
It appears from the cross section that there is a short flat plate that links the shackle tube and the main case body. I would guess that this would be placed into the seam of the tube and case (although this is not indicated in the cross section drawing).

More importantly, the method of joining the end plates to the case body is seen as a cage made from a series of small square rods, basically working as long rivets along the side of the case.
Another possibility would be to braise the two end plates on to the case and tube. It would certainly be a lot easier to fix all those separate decorative elements in place using the bar/rivet method. (A Norse smith would have to heat the entire case in a fire to melt the bronze to braise attach those elements. Heating for one element might easily loosen some pieces already attached.)

Thursday, June 04, 2015

Iron Smelting - Art or Science?

" I am not recommending this process to anyone is dirty and a lot of work..and may only work with some ores. "

(Quote from a recent post on the Bladesmith Forum)

I wanted to pipe in here.
I've maybe worked with more ore types than most others reading here, primarily because there is no naturally occurring iron ore in my local region - due to geography.
Certainly, * any * natural ore will vary considerably in potential iron content, oxide type, silica combination, dynamic impurities, structural form.
One of the source 'ores' I had access to for a while was processed hematite  as fine particle blasting grit. I have worked with this stuff at least 5 + times. I know both Antoine Marcel and Jesus Hernandez have used this material with success as well.

On a guess the original comment refers to using either hematite or 'iron sand' as the ore?

My own experience  is that the small particle size of the hematite grit tends to create a metallic iron particle size that also is quite fine. Because of this, temperature control is critical. The iron will very rapidly absorb * too  much * carbon. Unless you are very careful, you end up with cast iron.
As I understand it, this absorbtion is a surface area effect. Those small starting particles just have a lot of surface area compared to volume.

Your images of the results of this smelt certainly look like my own results sometimes - with this fine material at least.
The iron produced (if everything is going well) tends to be a crumbly texture, looking much like dampened dark brown sugar. This also tends to be a higher carbon content iron as well.  The image you provided of the (potentially) forgeable iron bloom certainly looks the same to my eye as my own results ( )
And yes - I certainly have gotten identical cast iron formed below the slag bowl.

One key in your own comment: "...may only work with some ores."
For any given ore - there will be  * one * ideal design of furnace.
The problem here may be in attempting to utilize a single static furnace layout - then attempting * that * to dictate the process (and expectation of results).

My consideration of ancient furnaces has caused me to appreciate the basic wisdom of a clay construction for the furnace. This material functions well, especially when modifications of mix using various proportions of sand / addition of organic materials is undertaken. This construction is relatively fast, simple and cheap. The furnaces can be fairly durable under normal operations, especially if a little care is taken with the construction - and especially the extraction method used.
More importantly, it is relatively simple to modify the base design of the furnace (stack height) - or simple enough to just build a whole new furnace body (Two or three bags of powdered clay and a couple of hours!)

When you examine the archaeology, one thing is commonly seen at larger scale, long term use iron smelting sites. There is often two or three 'unique' furnaces, commonly that show often only single firings. Then there are a whole pile of escentially identical furnances, usually with more durable construction and showing multiple uses.
Iron smelting was located most commonly at the site of the ore body.
Furnaces were then adapted from a theoretical template to suite the exact combination of ore (and likely charcoal) available at that location.

It also needs to be constantly remembered that the historic objective of a direct process bloomery furnace was dead soft * easy to forge * iron. Meaning no or next to no carbon content.

As many of the 'old hands' will tell you - bloomery iron making is more an art than a science!

This is a piece I started working up before ICMS in early May - that got caught in draft form.
I have just posted it without further editing

Wednesday, June 03, 2015

Viking Age Padlock

I had been contacted a while back by a fellow re-enactor of the Viking Age who undertakes presentations at a major Medieval / Renaissance Festival in the USA. He has a security problem (sounds like absolutely huge crowds!). He has ordered a pair of Norse era padlocks, with the associated strap hinge and hasp assemblies to fit on to sea chests.

This is the reference image he had sourced (off the internet):
So - here is the first problem. The internet as source.

The image is actually taken from (and not credited to) Anglo-Scandinavian Ironwork from Coppergate by Patrick Ottaway.

Like usual for a replica project like this one, I checked my own reference library. A second source I checked was Ironwork in Medieval Britian by Ian Goodall. Both are primary sources - as they are first generation archaeological reports. They include the original detailed scale drawings of the objects listed (but neither has photographs).
There are a couple of possible configurations of the basic 'compress a spring' mechanism in barrel padlocks. (There is an alternate construction method - the box padlock.)
These are split between cases with a hole in the end plate that use a fork, and those with a key rectangular hole in the end. An alternative is a case with a T shaped slot along the lower edge, also using a key with a shaped hole in it.

Goodall illustrates a number of barrel padlocks. However almost all are post 1100.  Almost all are also of the T slot type. There is one that fits the mechanism type and date requirements : # 16 - Winchester, Hampshire, c 1100. Unfortunately the scale of the illustration (roughly 1:3) is too small to show much detail.

Ottaway reports that at Coppergate there was only one complete barrel padlock found, plus two other case fragments.

Direct scan from Anglo-Scandinavian Ironwork from Coppergate
My normal practice is to take a scan of the artifact image, then run it through Photoshop to increase the image to life sized.  This allows me to take direct measurements right off the illustration, and most certainly helps with getting proportions and shapes more accurately.

First consideration for me - the customer has NOT asked for a direct reproduction / replica of this specific object. The artifact is highly ornamented - and I certainly did not quote a price for this level of decoration.
Second 'problem' is that the customer's source is * not actually an artifact *. That reference drawing is only a * diagram * - meant to illustrate the function of the lock mechanism.

What we see from the actual artifact (#3610):
case diameter = 4.5 cm / 1 3/4 inch
case length = 7 cm / 2 3/4 inch
metal thickness = 3 mm / about 1/8 inch
upper cylinder ID = 8 mm / 5/16 inch (suggests upper bar diameter at 6 mm / 1/4 inch)

Next Post : Forging materials to dimension

Regular Readers - may notice there has been a big gap in postings. Early May I attended the International Medieval Studies Conference for a week. On arriving back to Wareham, we found our satellite uplink hardware no longer receiving a signal. Through various failures, it was May 29 until internet access was restored here.

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