How is a Hall-Herault aluminum smelting cell started up
and shut down?

All the descriptions and diagrams I've seen show it in
steady state operation, with no hint how the electrolyte
is melted initially, nor how it's tapped off before cooling
so the structure of the cell doesn't turn into a big rock
on cooldown.

Thanks for reading, and any hints!

bob prohaska

bob prohaska <bp@www.zefox.net> wrote:
> How is a Hall-Herault aluminum smelting cell started up
> and shut down?
>
> All the descriptions and diagrams I've seen show it in
> steady state operation, with no hint how the electrolyte
> is melted initially, nor how it's tapped off before cooling
> so the structure of the cell doesn't turn into a big rock
> on cooldown.

Link 1 below has an abstract and references for a chapter called "Cell
Preheat/Start-up and Early Operationon", in a book on light metals
production. Springer wants $29.95 for a pdf of the chapter, or $319
for the book but the chapter's references are in publications
engineering libraries might have on hand. Gas preheating using heater
assemblies on wheels seems to be common. About getting the aluminum
out, per wikipedia (link 2) "The liquid aluminium is removed from the
cell via a siphon every 1 to 3 days in order to avoid having to use
extremely high temperature valves and pumps. Alumina is added to the
cells as the aluminum is removed." Of course, then aluminium is being
siphoned out, they usually stop short of getting it all because the
molten cryolite that floats on the molten aluminium needs to be left
in the cell. But if the cell is being shut down for cathode changing,
they probably need to siphon everything out.

Aluminum smelters usually have lots of Hall-Héroult cells hooked in
series, with about 5 volts DC (at 100-300 kA) across each cell, and
each cell operates full time except when the cathode is being changed
out. Per wikipedia (link 2) "Cathodes are typically replaced every
2-6 years. This requires the whole cell to be shut down." (Anodes are
much smaller, wear much faster, and are replaced far more frequently,
without stopping the cell.) For cathode replacement, I imagine that
cells are shut down in rotation, rather than shutting down the whole
line, but don't know for sure. Just attach a foot-thick bussbar
across the cell's connections, while avoiding ground, and start
siphoning out?

1. <https://link.springer.com/chapter/10.1007/978-3-319-48156-2_107>
2. <https://en.wikipedia.org/wiki/Hall???Héroult_process>

James Waldby <j-waldby@no.no> wrote:
>
> Link 1 below has an abstract and references for a chapter called "Cell
> Preheat/Start-up and Early Operationon", in a book on light metals
> production. Springer wants $29.95 for a pdf of the chapter, or $319
> for the book but the chapter's references are in publications
> engineering libraries might have on hand.

A bit spendy for a matter of curiosity. I'm surprised there's not more
online.

> Gas preheating using heater
> assemblies on wheels seems to be common.

That's very much a surprise. Given that aluminum smelters run
on electricity I expected an electric preheat, maybe using
SiC resistance elements in the cell walls. Gas seems alien.

> About getting the aluminum
> out, per wikipedia (link 2) "The liquid aluminium is removed from the
> cell via a siphon every 1 to 3 days in order to avoid having to use
> extremely high temperature valves and pumps. Alumina is added to the
> cells as the aluminum is removed." Of course, then aluminium is being
> siphoned out, they usually stop short of getting it all because the
> molten cryolite that floats on the molten aluminium needs to be left
> in the cell. But if the cell is being shut down for cathode changing,
> they probably need to siphon everything out.
>

Perhaps that's the answer to my shutdown question. Just tap off all
the aluminum, and then keep going. I guess it depends on how the
siphon is arranged.

> 1. <https://link.springer.com/chapter/10.1007/978-3-319-48156-2_107>

The link worked, but the pdf preview wouldn't download, reporting one
page requested and three as minimum.

Thanks for posting!

bob prohaska

On Sat, 2 Oct 2021 01:50:34 -0000 (UTC), bob prohaska
<bp@www.zefox.net> wrote:

>How is a Hall-Herault aluminum smelting cell started up
>and shut down?
>
>All the descriptions and diagrams I've seen show it in
>steady state operation, with no hint how the electrolyte
>is melted initially, nor how it's tapped off before cooling
>so the structure of the cell doesn't turn into a big rock
>on cooldown.
>
>Thanks for reading, and any hints!
>
>bob prohaska

"Presently there are two main preheating methods being used, i.e.
gas preheating/baking and resistor bed heating/baking."

The resistor bed method is based on using a layer of coke or
graphite particles between the anodes and cathode block surface
to provide ohmic voltage drop and act as a heating element
.. Some plants use shunts to deflect a part of the electric
current directly to the next pol without passing through the
resistor bed. The shunts will enable a more gentle start of the
preheating period and by gradually increasing the load passing
through the resistor bed the preheating time is extended
A typical gas or fuel bake equipment consists of either two large
propane, LNG or oil burners or multiple small gas
nozzle/bumers . Preferably, steel sheets are used to protect the
cathode surface from direct flame exposure

--
Best regards,
Spehro Pefhany

bob prohaska <bp@www.zefox.net> wrote:
> James Waldby <j-waldby@no.no> wrote:
>>
>> Link 1 below has an abstract and references for a chapter called "Cell
>> Preheat/Start-up and Early Operationon", in a book on light metals
>> production. Springer wants $29.95 for a pdf of the chapter, or $319
>> for the book but the chapter's references are in publications
>> engineering libraries might have on hand.
>
> A bit spendy for a matter of curiosity. I'm surprised there's not more
> online.

>> Gas preheating using heater assemblies on wheels seems to be
>> common.

> That's very much a surprise. Given that aluminum smelters run
> on electricity I expected an electric preheat, maybe using
> SiC resistance elements in the cell walls. Gas seems alien.

Big gas ovens are used to pre-bake some anodes, vs either gas or
electric for preheating the cell and cathode; don't know which is more
common. Link 3 has some Egyptalum Co. electric preheat startup
details, such as time/temperature for the 70-hour cell startup
process; voltages, currents, power used (eg 18 MWh at Egyptalum, vs 30
MWh for twice-as-big cells at Hydro Rheinwerk, link 4); resistors, eg
nichrome, stainless steel, graphite.

Both gas and electric preheaters are computer controlled because a
slow and even temperature rampup is needed. Per 3, 2-12°C/h is
"best", vs 10-19°C/h too fast, eg causing thermal shock and later some
"pot instability" leading to shorter pot life. Note, the heatup phase
is to bake (cook out the tar or pitch resin binder) the new pressed
graphite cathode and bring it up to operating temperature, after which
some tons of molten cryolite, alumina, etc are added to the cell and
its electrolysis process starts up.

3. <https://jpme.journals.ekb.eg/article_96413_1fe96828620111dc4c49dc8e80a67568.pdf>
4. <https://www.researchgate.net/publication/268293355_Improvements_in_the_electrical_preheating_of_hall_-_Heroult_pots> abstract

>> About getting the aluminum out, per wikipedia (link 2) "The liquid
>> aluminium is removed from the cell via a siphon every 1 to 3 days
>> in order to avoid having to use extremely high temperature valves
>> and pumps. Alumina is added to the cells as the aluminum is
>> removed." Of course, when aluminum is being siphoned out, they
>> usually stop short of getting it all because the molten cryolite
>> that floats on the molten aluminium needs to be left in the cell.
>> But if the cell is being shut down for cathode changing, they
>> probably need to siphon everything out.

> Perhaps that's the answer to my shutdown question. Just tap off all
> the aluminum, and then keep going. I guess it depends on how the
> siphon is arranged.

>> 1. <https://link.springer.com/chapter/10.1007/978-3-319-48156-2_107>
>
> The link worked, but the pdf preview wouldn't download, reporting one
> page requested and three as minimum.

Spehro Pefhany <speffSNIP@interlogdotyou.knowwhat> wrote:
> On Sat, 2 Oct 2021 01:50:34 -0000 (UTC), bob prohaska
> <bp@www.zefox.net> wrote:
>
>>How is a Hall-Herault aluminum smelting cell started up
>>and shut down?
>>
>
> "Presently there are two main preheating methods being used, i.e.
> gas preheating/baking and resistor bed heating/baking."
>
> The resistor bed method is based on using a layer of coke or
> graphite particles between the anodes and cathode block surface
> to provide ohmic voltage drop and act as a heating element
> . Some plants use shunts to deflect a part of the electric
> current directly to the next pol without passing through the
> resistor bed. The shunts will enable a more gentle start of the
> preheating period and by gradually increasing the load passing
> through the resistor bed the preheating time is extended
>
> A typical gas or fuel bake equipment consists of either two large
> propane, LNG or oil burners or multiple small gas
> nozzle/bumers . Preferably, steel sheets are used to protect the
> cathode surface from direct flame exposure
>

The mental picture is coming together. Preheating with fuel-air
flames, granulated coke resistors and pre-melting the cryolite
before pouring it into the preheated electrolytic cell were all
things I didn't think of.

Thanks for posting!

bob prohaska

"bob prohaska" wrote in message news:sjbdga$gm7$1@dont-email.me...

The mental picture is coming together. Preheating with fuel-air
flames, granulated coke resistors and pre-melting the cryolite
before pouring it into the preheated electrolytic cell were all
things I didn't think of.

Thanks for posting!

bob prohaska

----------------

How will you power this project?

I've been looking into loads that utilize alternate energy efficiently as
it's generated and so far the only practical one for me is DC refrigeration,
with an Alpicool freezer. Heating water is out because the elements in an
electric water heater can fail by shorting to the water and thus the tank,
which I've seen happen without tripping the breaker, and could possibly put
120VAC on my otherwise Low Voltage (<50V) solar panel wiring.

Carbon fiber mat has low electrical resistance and excellent high
temperature properties and might make a good distributed heating element.
The sample I got from a vacuum oven maker's scrap heap measures 3 - 5 Ohms
through its 1" thickness.

Jim Wilkins <muratlanne@gmail.com> wrote:
> "bob prohaska" wrote in message news:sjbdga$gm7$1@dont-email.me...
>
> The mental picture is coming together. Preheating with fuel-air
> flames, granulated coke resistors and pre-melting the cryolite
> before pouring it into the preheated electrolytic cell were all
> things I didn't think of.
>
>
> ----------------
>
> How will you power this project?

No project involved. I'm not contemplating making an aluminum
smelter, just curious about how they're operated.

>
> I've been looking into loads that utilize alternate energy efficiently as
> it's generated and so far the only practical one for me is DC refrigeration,
> with an Alpicool freezer.

> Heating water is out because the elements in an
> electric water heater can fail by shorting to the water and thus the tank,
> which I've seen happen without tripping the breaker, and could possibly put
> 120VAC on my otherwise Low Voltage (<50V) solar panel wiring.

Not quite following you. An isolation transformer on the 120VAC side will
protect the DC side from electrical mischief.

If I gather right, you're looking for useful applications for intermittently
available electric power. There aren't very many. Pumped water stores well
and keeps well, geography permitting. Purifying water via reverse osmosis
works well and keeps well, if you need purified water. Charging batteries,
of course, but they're expensive per unit of energy stored.

I don't think any high-temp process will be practical on a small scale. Too
hard to insulate, and too much mechanical stress from thermal cycling. Even
reverse osmosis has some limits; the membranes don't like pressure changes.
They're more tolerant than most things, but far from perfect.
> Carbon fiber mat has low electrical resistance and excellent high
> temperature properties and might make a good distributed heating element.
> The sample I got from a vacuum oven maker's scrap heap measures 3 - 5 Ohms
> through its 1" thickness.
>

The carbon fibers I've handled (25 years ago) were all pretty fragile
outside of a resin matrix. It didn't take much flexing to break them.

How much energy are you trying to utilize and over what timescale?

bob prohaska

"bob prohaska" wrote in message news:sjdb94$h1q$1@dont-email.me...

Jim Wilkins <muratlanne@gmail.com> wrote:
> "bob prohaska" wrote in message news:sjbdga$gm7$1@dont-email.me...
>
> The mental picture is coming together. Preheating with fuel-air
> flames, granulated coke resistors and pre-melting the cryolite
> before pouring it into the preheated electrolytic cell were all
> things I didn't think of.
>
>
> ----------------
>
> How will you power this project?

No project involved. I'm not contemplating making an aluminum
smelter, just curious about how they're operated.

>
> I've been looking into loads that utilize alternate energy efficiently as
> it's generated and so far the only practical one for me is DC
> refrigeration,
> with an Alpicool freezer.

> Heating water is out because the elements in an
> electric water heater can fail by shorting to the water and thus the tank,
> which I've seen happen without tripping the breaker, and could possibly
> put
> 120VAC on my otherwise Low Voltage (<50V) solar panel wiring.

Not quite following you. An isolation transformer on the 120VAC side will
protect the DC side from electrical mischief.

If I gather right, you're looking for useful applications for intermittently
available electric power. There aren't very many. Pumped water stores well
and keeps well, geography permitting. Purifying water via reverse osmosis
works well and keeps well, if you need purified water. Charging batteries,
of course, but they're expensive per unit of energy stored.

I don't think any high-temp process will be practical on a small scale. Too
hard to insulate, and too much mechanical stress from thermal cycling. Even
reverse osmosis has some limits; the membranes don't like pressure changes.
They're more tolerant than most things, but far from perfect.

> Carbon fiber mat has low electrical resistance and excellent high
> temperature properties and might make a good distributed heating element.
> The sample I got from a vacuum oven maker's scrap heap measures 3 - 5 Ohms
> through its 1" thickness.
>

The carbon fibers I've handled (25 years ago) were all pretty fragile
outside of a resin matrix. It didn't take much flexing to break them.

How much energy are you trying to utilize and over what timescale?

bob prohaska
------------------------------

Due to surrounding trees and the orientation of my roof I've found
semi-permanent homes for only about 150W of panels. If needed I can quickly
set up another 400W in the yard or driveway and manually re-aim them every
few hours. If the sky is clear that appears to be enough to support my low
winter electricity demand, since my wood stove provides heat, cooking and
hot water, even for laundry and showers.

Using the solar power as DC instead of AC would eliminate the ~50W constant
loss in the inverter, reducing battery cost by decreasing the nighttime
depth of discharge and the number of batteries required. My compact
refrigerator consumes about 100W half the time so the inverter doubles the
overnight drain. Panels have fallen below $1/Watt and it's the cost of
batteries that makes my solar more expensive than grid power.

If the inverter could power up when it senses load demand its loss would be
halved, but my (free) APC 1400 UPS doesn't support turn-on from idle in its
command set and would require hacking the circuit board to accomplish it.
Tripp Lite inverters can sense load if you have to buy. Apparently all
true-sine inverters have significant no-load power demand and the cheaper,
more efficient square wave ones aren't suited to refrigerator compressors,
other AC motors or capacitive-input (light weight) power converters. A scope
shows that "modified sine" means a pulse-width-modified square wave.

I'd like to create a low cost home solar system that someone else could
assemble from purchased or salvaged components without needing electrical
engineering and technician skills. So far mine is cheap and effective, but
not simple or easy and I don't operate some parts of it unattended or
overnight.

I've been testing how second-hand AGMs hold up. My experience so far agrees
with Internet advice that their storage life averages about half what a
well-maintained flooded marine battery can give. A marine battery I bought
circa 2008 still runs my DC freezer for 24 hours.
https://www.mastervolt.com/determining-the-lifespan-of-a-battery/
"If kept in a charged state when unused, the common lifespan of a 12-volt
Gel or AGM battery is up to six years."
AGMs appear to fail by irreversible sulfation (or contact corrosion?) that
increases internal resistance although the voltage seems normal, whereas
I've had reasonably good results from desulfating an old wet battery with a
current-limited higher charging voltage, a good side job for solar power. If
you can't design and build a DC-powered metered adjustable power supply this
looks good so far:
https://www.aliexpress.com/item/4001230106243.html
I haven't gotten more than 2.7A continuously from it but that's enough. 14
AWG silicone wire was slightly too big for the terminals.

The only advantage I see for AGMs is that they can be fully recharged
quickly without releasing hydrogen. I charge my wet batteries at the float
voltage where they don't gas, or fully recharge during winter daylight. Most
of my industrial experience is with Lithiums and I'd use them if they
weren't so expensive.

jsw

"Jim Wilkins" wrote in message news:sjet80$gcm$1@dont-email.me...

"bob prohaska" wrote in message news:sjdb94$h1q$1@dont-email.me...

> Heating water is out because the elements in an
> electric water heater can fail by shorting to the water and thus the tank,
> which I've seen happen without tripping the breaker, and could possibly
> put
> 120VAC on my otherwise Low Voltage (<50V) solar panel wiring.

Not quite following you. An isolation transformer on the 120VAC side will
protect the DC side from electrical mischief.

---------------------------------

That's a 5 KVA transformer whose cost I'd never come close to recovering,
plus the tank is grounded in uncertain ways through the plumbing. I found
that out with a clamp-on ammeter when the Neutral connection at the
weatherhead corroded.

I don't leave second-hand equipment of unknown history or condition
permanently powered up, for instance when not in active use the air
compressor's breaker is Off. I can test for HiPot leakage but not burnt
insulation etc.

The solar controllers I have now switch the negative side and can't regulate
current if both the source and load are grounded. My solar power system
comes under the exemptions for Low Voltage and Separately Derived, isolated
from the grid, and I intend to keep it so.

Jim Wilkins <muratlanne@gmail.com> wrote:
>
> Due to surrounding trees and the orientation of my roof I've found
> semi-permanent homes for only about 150W of panels. If needed I can quickly
> set up another 400W in the yard or driveway and manually re-aim them every
> few hours. If the sky is clear that appears to be enough to support my low
> winter electricity demand, since my wood stove provides heat, cooking and
> hot water, even for laundry and showers.
>

On that size scale battery charging is probably the best possible
use for excess solar power.

> Using the solar power as DC instead of AC would eliminate the ~50W constant
> loss in the inverter, reducing battery cost by decreasing the nighttime
> depth of discharge and the number of batteries required. My compact
> refrigerator consumes about 100W half the time so the inverter doubles the
> overnight drain. Panels have fallen below $1/Watt and it's the cost of
> batteries that makes my solar more expensive than grid power.
>

Solar power is far from cheap. People get seduced by the "sun is free"
and forget the cost of infrastructure, intermittency and upkeep. That's
why we're so fond of fossil fuels 8-)
> If the inverter could power up when it senses load demand its loss would be
> halved, but my (free) APC 1400 UPS doesn't support turn-on from idle in its
> command set and would require hacking the circuit board to accomplish it.
> Tripp Lite inverters can sense load if you have to buy. Apparently all
> true-sine inverters have significant no-load power demand and the cheaper,
> more efficient square wave ones aren't suited to refrigerator compressors,
> other AC motors or capacitive-input (light weight) power converters. A scope
> shows that "modified sine" means a pulse-width-modified square wave.
>

Sounds like the "free" UPS is the wrong tool for the job. A DC thermostat
circuit controlling an individual inverter for the fridge alone is closer
to an appropriate setup. Hacking still required, but on the fridge, not
the power supply. And, the inverter can be tuned to the load.

> I'd like to create a low cost home solar system that someone else could
> assemble from purchased or salvaged components without needing electrical
> engineering and technician skills. So far mine is cheap and effective, but
> not simple or easy and I don't operate some parts of it unattended or
> overnight.
>

The problem you're solving is a close cousin to that faced by long-distance
sailboaters. You don't need to worry so much about weight, but in most other
respects the solutions will be similar, and similarly expensive. Either in
know-how or money paid for someone else's know-how.

Wish I had some better ideas!

bob prohaska

"bob prohaska" wrote in message news:sjfv26$til$1@dont-email.me...

Jim Wilkins <muratlanne@gmail.com> wrote:
>
Sounds like the "free" UPS is the wrong tool for the job. A DC thermostat
circuit controlling an individual inverter for the fridge alone is closer
to an appropriate setup. Hacking still required, but on the fridge, not
the power supply. And, the inverter can be tuned to the load.

---------------------

Unlike the boater I can risk inadequate or unreliable solutions because I
still have the house's standard appliances and electric heat. My electronic
know-how is enough to design computerized devices and DC power supplies but
not switchers, though some very nice digital regulators are now available at
a reasonable price, such as the DPS5020, which enable building powerful and
efficient lab supplies with flea-market variacs and transformers or solar
input. I used to order less capable Agilent lab supplies that cost $1000.
The high-end automated industrial test equipment I used to build consisted
mostly of digitally controlled 4-quadrant (+/-V, +/-I) power supplies. The
Keithley Sourcemeter is a good example though these were designed in-house.

I knew from the start that I didn't have a good location for enough roof
panels to support daily cycling, and designed a small system mainly to
experiment with and measure performance. Though it isn't what I would have
bought new (Tripplite?) the free 900W APC UPS is quite adequate for
experimenting and large enough to start and run my compact refrigerator. I
have a 2200W HF inverter generator which I think is about the minimum
practical size to support a kitchen, as long as the fridge is unplugged when
running the microwave or coffee pot. The APC tends to reject non-inverter
generator AC and revert to battery even at its lowest sensitivity setting.

As it presently exists, 900W of AC with 2 KWH of battery energy, the system
is able to run my refrigeration and laptop off-grid during daytime and
overnight cold front thunderstorms. Other loads have their outlet strips
shut off. I consider the laptop essential for Internet weather radar which
is far more useful than radio reports. It appears able to support a long
term winter grid outage if the sun is out, with generator support at meal
times, but it needs attention and manual panel tracking. It was relatively
inexpensive and enough to keep me functioning fairly normally when
everything is shut down and I have nothing better to do. Part of the
experiment is determining an acceptable level of alternate energy
inconvenience.

The APC can be started remotely with a FET in parallel with the On button
and a logic input from the beeper driver since a cold-start-on-battery
requires pressing the button until the beeper sounds. These might need
optical isolation to avoid ground loops.

I could add a thermostat or thermocouple to the fridge without drilling by
slipping the wires through a caulked slit in the door gasket.

A relay at the fridge's AC plug would permit either sensing resistance or
connecting to power. I put a $15 current-sensing relay on the fridge cord to
signal visually and electrically whether its compressor is on or off. The
relay LED shows when to wait a few minutes before moving the plug from grid
to inverter, to reduce the starting surge.

I could set up an Arduino to sense fridge demand, start the APC and switch
the fridge plug from sensing to power, then put the APC back to sleep by
serial command when the fridge motor stops. Unfortunately there isn't a
wake-up command that works when it's on battery. At least this older APC
model doesn't have a shutdown timer and can operate indefinitely at ~75%
power.

You are correct that it isn't the right tool, for someone who wants a simple
and painless substitute for grid power. The serial connector does NOT take a
standard cable, it has battery voltage coming out (or going in, at test?) on
it. I wrote a program to talk to it.

For me testing, measuring and fine-tuning is what I do.

jsw

"Jim Wilkins" wrote in message news:sjhmjq$art$1@dont-email.me...
....
You are correct that it isn't the right tool, for someone who wants a simple
and painless substitute for grid power. The serial connector does NOT take a
standard cable, it has battery voltage coming out (or going in, at test?) on
it. I wrote a program to talk to it.

-------------------

I forgot to mention that as received it needed its battery constant and
float voltage reprogrammed, with a user-wired serial cable.

jsw