Re: Energy requirements of additive manufacturing?
12 views
Skip to first unread message
Christian Siefkes
Aug 27, 2009, 9:06:20 AM8/27/09
to openmanufacturing
ben lipkowitz wrote:
> On Tue, 18 Aug 2009, Christian Siefkes wrote:
> [...] I'm doubtful regarding the
>> energy requirements of these techniques: will they *ever* be able to
>> compete with traditional techniques in terms of energy usage, or will their
>> energy usage stay much higher, even after the technique has matured?
>
> I dont know about ultimate ultimates but for real life numbers all we have
> to do is take a look at the available data:
[...]
> so for reprap that yields 6kWh/kg = 2e7 J/kg
>
> compare to various other processes in this paper: (see page 12)
> http://web.mit.edu/2.810/www/lecture/Gutowski - Thermo Analysis.pdf
>
> injection molding about 2e6 J/kg
> casting about 2e6 J/kg
> rough machining about 1e7 J/kg
> finish machining about 5e8 J/kg
>
> so reprap uses 10 times the energy of injection molding or about on par
> with rough machining.
>
> as the chart shows it's not enough to compare simply by kg of material
> produced but also the precision with which it is produced and the rate.
> On Tue, 18 Aug 2009, Christian Siefkes wrote:
> [...] I'm doubtful regarding the
>> energy requirements of these techniques: will they *ever* be able to
>> compete with traditional techniques in terms of energy usage, or will their
>> energy usage stay much higher, even after the technique has matured?
>
> I dont know about ultimate ultimates but for real life numbers all we have
> to do is take a look at the available data:
[...]
> so for reprap that yields 6kWh/kg = 2e7 J/kg
>
> compare to various other processes in this paper: (see page 12)
> http://web.mit.edu/2.810/www/lecture/Gutowski - Thermo Analysis.pdf
>
> injection molding about 2e6 J/kg
> casting about 2e6 J/kg
> rough machining about 1e7 J/kg
> finish machining about 5e8 J/kg
>
> so reprap uses 10 times the energy of injection molding or about on par
> with rough machining.
>
> as the chart shows it's not enough to compare simply by kg of material
> produced but also the precision with which it is produced and the rate.
Thanks for this analysis! That's interesting to know. So while additive
techniques certainly still have other problems (slowness, poor surface
precision, cost), energy usage seems to be less of an issue...
Thanks and best regards
Christian
--
|-------- Dr. Christian Siefkes --------- chri...@siefkes.net ---------
| Homepage: http://www.siefkes.net/ | Blog: http://www.keimform.de/
| Better Bayesian Analysis: | Peer Production Everywhere:
| http://bart-project.com/ | http://peerconomy.org/wiki/
|------------------------------------------ OpenPGP Key ID: 0x346452D8 --
Computers are not intelligent. They only think they are.
Erik
Aug 29, 2009, 6:26:20 PM8/29/09
to Open Manufacturing
Thanks. It's nice to have some data for comparison. Only, it's a tiny
fraction of the bigger picture. I took some time to do a more
sophisticated analysis, but it's nowhere near accurate or complete. I
do want to have a more accurate and unbiased comparison, so comments
are highly appreciated!
To have a fair analysis of the actual impact of different production
paradigms, you have to look at the entire lifecycle of products,
including not only production, but everything resulting from the
production, consumption and post-consumption.
Setting up, maintaining and decommissioning an entire production line
should be included if it is product specific. Per product this might
not be much, but considering that there are instances of production
lines for the vast amounts of products that exist, it is an enormous
environmental burden. Instead, flexible production machines can be
used for many different product categories. Granted, you would need a
machine per neighbourhood for an alternative to be feasible to the
general public. Yet this is entirely within the realm of possibility.
Distributing goods is usually done through wholesale, retail and
involve energy costs in terms of transportation, powering lighting in
warehouses, require packaging (for transit) and repackaging (for
retail), lighting and heating/ air conditioning in retail stores, etc.
It becomes much worse, but admittedly bit awkward if you include
personnel of each of these organisations need to go to work, which
uses energy. Warehousing of thousands of digital products could take
as much as a small (virtualised) server with a little free hard disk
space. Going to the store for a product, will cost energy (if you
combine purchases, this cost is shared among them). Some people even
have a tiny warehouse at home. A room that's used for storage can't be
used for something else. We need on average a slightly bigger house
because we store everything.
And then, a fraction of products is overproduced. For these there is
an environmental burden that should be added to the products that were
consumed. With shortening product cycles for almost any product type,
more products are introduced every time and demand is increasingly
difficult to forecast, often resulting in over or underproduction.
Products that are not really wanted but are bought at huge price cuts
(perhaps even at a loss in many cases), are often tossed away the
earliest or will start collecting dust in your attic. At least they
are not really valued which makes the environmental impact less than
worth it.
For products that are 'made to order' by a local machine, there is an
almost certain demand for it. So, products that are produced on
demand for someone, are products that someone actually wants.
Disposal of products should be taken into account. While we cannot
claim yet that RepRap products are more eco-friendly in disposal
(unless you print in biodegradable PLA or CAPA (from disposable milk-
bottles), there certainly is big potential for making cradle to cradle
products. You can close the cycle of your material flows by recycling
products that you don't use any more. Normal disposal results in
transportation and processing of domestic waste, which has its
environmental costs. In the RepRap scenario there would be less
disposal because 'waste equals food'.
From the point of view of the consumer, many trips to stores can be
saved if the free commons of designs of physical products is extended.
The premium of almost 'instant gratification' can also be an
incentive. As a RepRap owner I can think of many reasons, but I'll
save them for another discussion. From the point of view of the
environment, most of the previously mentioned costs can be saved.
Distribution of a single type of homogeneous stock material is not
that expensive. RepRap and other additive production processes use
material efficiently, it only adds material where needed. In contrast
to subtractive machining which removes material until the desired
shape is achieved. Stock material needs to be bigger than the desired
object and planning is needed to use material efficiently. Excess
material is often waste and needs to find its way through recycling or
to the landfill.
To highlight that the specific heat of plastics (J/kg/K) and
production efficiency are negligible in many cases, here's an extreme
example, printing a button on your RepRap or Makerbot at home (as done
and described here: http://www.thingiverse.com/thing:950 ) is more
efficient than going to a store. I've created a 'Spare part' tag for
some digital designs of products that can be locally produced:
http://www.thingiverse.com/tag:spare_part
This extreme example seems unfair, but in fact it is not. This would
be an example product category for which centralised production has no
raison d'être. Admittedly, for buttons on more centrally manufactured
clothing it will still be feasible, but not for replacements that
people need only occasionally. There are also categories for which the
converse is true. Of course an efficient system is a hybrid of many
different modes of distribution. Some products will be distributed
digitally and produced locally. Others will be produced centrally,
distributed and consumed. For product categories where centralised
production and injection moulding is more efficient (on the supply
side), and where there is little need homogeneity for this case (on
the demand side), mass production is a better candidate.
Production of RepRap machines, by contrast is very efficient. It
consists of two types of parts: parts that can efficiently be made
centrally and products that should be made in a dense, distributed
network. Nuts and bolts, semiconductors, etc. should be made centrally
for now. So we get them from the local hardware store. These are so
generic that we need stocks of these things for many purposes anyway.
It doesn't make economical or ecological sense to make those at home.
The custom parts are subject to improvements because the design
evolves, so making moulds for central production would be impractical
and would inhibit innovation because you might choose to make less
improvements. If they are produced flexibly, this could be in a
unified network of fabricator operators, so that consumption is close
to production. RepRap has to potential to use this hybrid model in
foreseeable time. Even when limited to thermodynamics of production in
a RepRap compared to injection moulding, RepRap might eventually win.
Consider that heating the thermoplastics costs at least as much as the
specific heat times the amount of material used. This is the
theoretical minimum. Recovery of heat (e.g. heating the house, which
RepRap currently does) could even make this energy 100% efficiently
used during two seasons. Injection moulding often involves use of
cooling channels through which a coolant will flow after injection.
This is to speed up the process and quickly make the mould available
for the next product. It's an art in itself to cool as quickly as
possible, but I would think that cooling costs energy. The only
exception is if a heat exchanger were used and heat is useful
somewhere else in the factory, but I would be positively surprised if
this were the case in most manufacturing plants. Still, in a RepRap,
cooling is often passive (just by convection) or at most involves a
ca. 1 Watt fan. Motion of the RepRap could be minimal if the moving
mass is minimised. In fact, I'm implementing this in my RepRap at the
moment. This would result in an even more desktop friendly machine,
with much smaller motors. You could then use any stepper from a
decommissioned product or very small and cheap motors. Oh yeah, I
forgot to mention that many people use things that would otherwise be
'junk' as useful parts in a RepRap (PSU, printer stepper motors, most
cables). By the way my estimate is that most RepRaps use less than
50W, which is not all that bad at this moment. Electricity costs are
no more than a few cents even for larger objects.
Again, your feedback is greatly appreciated! I intend to post this on
some blogs after your feedback.
Erik de Bruijn
http://blog.erikdebruijn.nl/
(B.t.w. Christian, I'll be looking into your book on peer production
after I finished this post!)
On Aug 27, 3:06 pm, Christian Siefkes <christ...@siefkes.net> wrote:
> ben lipkowitz wrote:
> > On Tue, 18 Aug 2009, Christian Siefkes wrote:
> > [...] I'm doubtful regarding the
> >> energy requirements of these techniques: will they *ever* be able to
> >> compete with traditional techniques in terms of energy usage, or will their
> >> energy usage stay much higher, even after the technique has matured?
>
> > I dont know about ultimate ultimates but for real life numbers all we have
> > to do is take a look at the available data:
> [...]
> > so for reprap that yields 6kWh/kg = 2e7 J/kg
>
> > compare to various other processes in this paper: (see page 12)
> >http://web.mit.edu/2.810/www/lecture/Gutowski- Thermo Analysis.pdf
> < 1KViewDownload
fraction of the bigger picture. I took some time to do a more
sophisticated analysis, but it's nowhere near accurate or complete. I
do want to have a more accurate and unbiased comparison, so comments
are highly appreciated!
To have a fair analysis of the actual impact of different production
paradigms, you have to look at the entire lifecycle of products,
including not only production, but everything resulting from the
production, consumption and post-consumption.
Setting up, maintaining and decommissioning an entire production line
should be included if it is product specific. Per product this might
not be much, but considering that there are instances of production
lines for the vast amounts of products that exist, it is an enormous
environmental burden. Instead, flexible production machines can be
used for many different product categories. Granted, you would need a
machine per neighbourhood for an alternative to be feasible to the
general public. Yet this is entirely within the realm of possibility.
Distributing goods is usually done through wholesale, retail and
involve energy costs in terms of transportation, powering lighting in
warehouses, require packaging (for transit) and repackaging (for
retail), lighting and heating/ air conditioning in retail stores, etc.
It becomes much worse, but admittedly bit awkward if you include
personnel of each of these organisations need to go to work, which
uses energy. Warehousing of thousands of digital products could take
as much as a small (virtualised) server with a little free hard disk
space. Going to the store for a product, will cost energy (if you
combine purchases, this cost is shared among them). Some people even
have a tiny warehouse at home. A room that's used for storage can't be
used for something else. We need on average a slightly bigger house
because we store everything.
And then, a fraction of products is overproduced. For these there is
an environmental burden that should be added to the products that were
consumed. With shortening product cycles for almost any product type,
more products are introduced every time and demand is increasingly
difficult to forecast, often resulting in over or underproduction.
Products that are not really wanted but are bought at huge price cuts
(perhaps even at a loss in many cases), are often tossed away the
earliest or will start collecting dust in your attic. At least they
are not really valued which makes the environmental impact less than
worth it.
For products that are 'made to order' by a local machine, there is an
almost certain demand for it. So, products that are produced on
demand for someone, are products that someone actually wants.
Disposal of products should be taken into account. While we cannot
claim yet that RepRap products are more eco-friendly in disposal
(unless you print in biodegradable PLA or CAPA (from disposable milk-
bottles), there certainly is big potential for making cradle to cradle
products. You can close the cycle of your material flows by recycling
products that you don't use any more. Normal disposal results in
transportation and processing of domestic waste, which has its
environmental costs. In the RepRap scenario there would be less
disposal because 'waste equals food'.
From the point of view of the consumer, many trips to stores can be
saved if the free commons of designs of physical products is extended.
The premium of almost 'instant gratification' can also be an
incentive. As a RepRap owner I can think of many reasons, but I'll
save them for another discussion. From the point of view of the
environment, most of the previously mentioned costs can be saved.
Distribution of a single type of homogeneous stock material is not
that expensive. RepRap and other additive production processes use
material efficiently, it only adds material where needed. In contrast
to subtractive machining which removes material until the desired
shape is achieved. Stock material needs to be bigger than the desired
object and planning is needed to use material efficiently. Excess
material is often waste and needs to find its way through recycling or
to the landfill.
To highlight that the specific heat of plastics (J/kg/K) and
production efficiency are negligible in many cases, here's an extreme
example, printing a button on your RepRap or Makerbot at home (as done
and described here: http://www.thingiverse.com/thing:950 ) is more
efficient than going to a store. I've created a 'Spare part' tag for
some digital designs of products that can be locally produced:
http://www.thingiverse.com/tag:spare_part
This extreme example seems unfair, but in fact it is not. This would
be an example product category for which centralised production has no
raison d'être. Admittedly, for buttons on more centrally manufactured
clothing it will still be feasible, but not for replacements that
people need only occasionally. There are also categories for which the
converse is true. Of course an efficient system is a hybrid of many
different modes of distribution. Some products will be distributed
digitally and produced locally. Others will be produced centrally,
distributed and consumed. For product categories where centralised
production and injection moulding is more efficient (on the supply
side), and where there is little need homogeneity for this case (on
the demand side), mass production is a better candidate.
Production of RepRap machines, by contrast is very efficient. It
consists of two types of parts: parts that can efficiently be made
centrally and products that should be made in a dense, distributed
network. Nuts and bolts, semiconductors, etc. should be made centrally
for now. So we get them from the local hardware store. These are so
generic that we need stocks of these things for many purposes anyway.
It doesn't make economical or ecological sense to make those at home.
The custom parts are subject to improvements because the design
evolves, so making moulds for central production would be impractical
and would inhibit innovation because you might choose to make less
improvements. If they are produced flexibly, this could be in a
unified network of fabricator operators, so that consumption is close
to production. RepRap has to potential to use this hybrid model in
foreseeable time. Even when limited to thermodynamics of production in
a RepRap compared to injection moulding, RepRap might eventually win.
Consider that heating the thermoplastics costs at least as much as the
specific heat times the amount of material used. This is the
theoretical minimum. Recovery of heat (e.g. heating the house, which
RepRap currently does) could even make this energy 100% efficiently
used during two seasons. Injection moulding often involves use of
cooling channels through which a coolant will flow after injection.
This is to speed up the process and quickly make the mould available
for the next product. It's an art in itself to cool as quickly as
possible, but I would think that cooling costs energy. The only
exception is if a heat exchanger were used and heat is useful
somewhere else in the factory, but I would be positively surprised if
this were the case in most manufacturing plants. Still, in a RepRap,
cooling is often passive (just by convection) or at most involves a
ca. 1 Watt fan. Motion of the RepRap could be minimal if the moving
mass is minimised. In fact, I'm implementing this in my RepRap at the
moment. This would result in an even more desktop friendly machine,
with much smaller motors. You could then use any stepper from a
decommissioned product or very small and cheap motors. Oh yeah, I
forgot to mention that many people use things that would otherwise be
'junk' as useful parts in a RepRap (PSU, printer stepper motors, most
cables). By the way my estimate is that most RepRaps use less than
50W, which is not all that bad at this moment. Electricity costs are
no more than a few cents even for larger objects.
Again, your feedback is greatly appreciated! I intend to post this on
some blogs after your feedback.
Erik de Bruijn
http://blog.erikdebruijn.nl/
(B.t.w. Christian, I'll be looking into your book on peer production
after I finished this post!)
On Aug 27, 3:06 pm, Christian Siefkes <christ...@siefkes.net> wrote:
> ben lipkowitz wrote:
> > On Tue, 18 Aug 2009, Christian Siefkes wrote:
> > [...] I'm doubtful regarding the
> >> energy requirements of these techniques: will they *ever* be able to
> >> compete with traditional techniques in terms of energy usage, or will their
> >> energy usage stay much higher, even after the technique has matured?
>
> > I dont know about ultimate ultimates but for real life numbers all we have
> > to do is take a look at the available data:
> [...]
> > so for reprap that yields 6kWh/kg = 2e7 J/kg
>
> > compare to various other processes in this paper: (see page 12)
>
> > injection molding about 2e6 J/kg
> > casting about 2e6 J/kg
> > rough machining about 1e7 J/kg
> > finish machining about 5e8 J/kg
>
> > so reprap uses 10 times the energy of injection molding or about on par
> > with rough machining.
>
> > as the chart shows it's not enough to compare simply by kg of material
> > produced but also the precision with which it is produced and the rate.
>
> Thanks for this analysis! That's interesting to know. So while additive
> techniques certainly still have other problems (slowness, poor surface
> precision, cost), energy usage seems to be less of an issue...
>
> Thanks and best regards
> Christian
>
> --
> |-------- Dr. Christian Siefkes --------- christ...@siefkes.net ---------
> > injection molding about 2e6 J/kg
> > casting about 2e6 J/kg
> > rough machining about 1e7 J/kg
> > finish machining about 5e8 J/kg
>
> > so reprap uses 10 times the energy of injection molding or about on par
> > with rough machining.
>
> > as the chart shows it's not enough to compare simply by kg of material
> > produced but also the precision with which it is produced and the rate.
>
> Thanks for this analysis! That's interesting to know. So while additive
> techniques certainly still have other problems (slowness, poor surface
> precision, cost), energy usage seems to be less of an issue...
>
> Thanks and best regards
> Christian
>
> --
> | Homepage:http://www.siefkes.net/ | Blog:http://www.keimform.de/
> | Better Bayesian Analysis: | Peer Production Everywhere:
> | http://bart-project.com/ | http://peerconomy.org/wiki/
> |------------------------------------------ OpenPGP Key ID: 0x346452D8 --
> Computers are not intelligent. They only think they are.
>
> signature.asc
> | Better Bayesian Analysis: | Peer Production Everywhere:
> | http://bart-project.com/ | http://peerconomy.org/wiki/
> |------------------------------------------ OpenPGP Key ID: 0x346452D8 --
> Computers are not intelligent. They only think they are.
>
> < 1KViewDownload
Reply all
Reply to author
Forward
0 new messages