Sunday, September 10, 2006

50 Gallons of Gasoline in Each PC

Well not exactly, but there the energy equivalent of 50 gallons used in the production of a PC and 17" CRT monitor, according to a report by Eric Williams of the United Nations University. He calculates that it takes 6,400MJ to produce them. 1 gallon of gasoline contains 131MJ of energy, so this works out to 48.9 gallons. This is approximately the amount of gasoline that an average American uses to fuel their car each month.

Another surprising statistic from the report is that in the standard 3 year lifespan of a PC and monitor, 81% of the energy is used to create them and only 19% is used to run them. To contrast that, a refrigerator uses 11% of its life cycle energy to be produced and 89% to run it. A car typically uses around 20% to produce.

According to the report a typical Pentium III system with 17-in. CRT monitor consumes on average 128 W when fully on 3 h use per day full-on (no standby). This works out to 384Wh a day or 140kWh a year or 420kWh over a standard 3 year lifetime of a PC. 420kWh is equivalent to 1500MJ, so it takes 6400MJ to produce a computer, 1500MJ to run it for a total of 7900MJ or 2633MJ/year of use (or 20 gallons of gasoline equivalent).

The implication of this is that to reduce the amount of energy a PC uses it is much more important to focus on manufacturing then on end usage. The Energy Star programs rates PCs and monitors based on how much energy they use, but says nothing about how much energy it takes to produce them. EPEAT rates computers based on how environmentally aware they are, but does not take energy of production into account. Here is a chart of where the energy goes.

direct fossil (MJ)electricity use (kWh)total energy (MJ)% of production% of total
process analysis
printed circuit boards26.77.7540.8%0.7%
CRT manufacture/assembly21012.52554.0%3.2%
bulk materials - control unit0077012.0%9.7%
bulk materials - CRT0080012.5%10.1%
silicon wafers038.11372.1%1.7%
computer assembly35.351.22193.4%2.8%
IO analysis
electronic chemicals38118.54477.0%5.7%
semiconductor manufacturing equipment39229.44987.8%6.3%
passive components10910.31462.3%1.8%
disk drives and other parts365234487.0%5.7%
packaging, documentation1204.81372.1%1.7%
other processes97361119218.6%15.1%
total production32484306400100%
use phase: home user (3 yrs)420150019.0%

Semiconductors and bulk materials for the control unit and CRT makeup 31% of the total lifetime energy and look like a place to focus for reduction. I don't know how exactly to improve this, but I would guess using more recyclable metals and plastics would help.

Another option is to lengthen the time of ownership. If a PC is used for 5 years instead of 3, the amount of energy needed per year (taking total lifetime energy and dividing it by total years) goes down from 2633MJ/yr to 1783MJ/yr or a decrease of 32%.

While I think this analysis is very important, I think that some of the numbers are suspect and I hope somebody undertakes another study of the life cycle energy use of PCs soon.

The first issue I have is that this was based on a 2000 PC and 17" CRT. I would like to see the numbers for a 2006 PC and a 17" or 19" LCD. I don't know exactly how it would turn out. PCs and monitors are much cheaper now, which I would think means they are made more efficiently and might therefore use less energy per PC to make. But, monitors have gotten larger and PCs faster which might offset this gain. Also, PCs run faster and use more electricity now, so the 81% to 19% ratio of production to usage might tilt more towards usage. My personal computer (based on Kill-a-Watt readings) uses 85W when nothing is running and 135 when the CPU is at 100% and the DVD is running, so lets call it 100W average. My 19" monitor uses on average 70W, so my 170W is a bit over the 128 W they state.

I would also be interested in the numbers for laptops as more people are using laptops now. My laptop uses 43W, so I would think production energy is much more important for laptops in the life cycle analysis.

I think the methodology he uses might overstate the energy usage of production.
To sum up, pessimistic assumptions on the accuracy of process-sum and IO parts of the analysis yield a possible range of 5000-16 000 MJ (base result: 6400 MJ) for the total energy required to manufacture a desktop system.

The hybrid result of 6400 MJ required to produce a desktop system is considerably higher than the process sum result of the 1998 EU-sponsored study of 3630 MJ.
Other models have had results that are just about 1/2 of what he states, and there is a big margin of error for the number he does report. The economic input-output (IO) methodology also comes up with higher values by including more of the indirect energy costs then the process sum method. In doing an apple to apple comparison I think this might end up estimating on the high side.

When he uses electricity he does not take into account the upstream usage of coal or other fossil fuels that are needed to produce the electricity which typically are only around 30% efficient. Making that change would tilt the energy balance back towards usage.

Another issue is that this assumes the monitor is replaced with the computer. In my experience the monitor usually sticks around for a longer amount of time then the PC does.

Also, now PCs are connected to the internet and there is energy being used by the network and the server machines. The energy usage of my PC should probably take into account this additional energy that is used as part of my computing experience.

The amount of time that you use the PC greatly shifts the balance. Lets look at 3 cases: the typical 3 hours of home use, the 8 hrs a working day for business use and 24 hours for server computers. And lets look at the impact of using it for 5 years instead of the typical 3.

3 years of use5 years of use
hours of use a dayelectricity usage (kwh)% of total energyelectricity usage (kwh)% of total energy
3 hrs4201970028
8 hrs768301,28042
24 hrs3,360655,60076

As the graph shows, the amount of energy used in usage vs. production goes up greatly for business and server machines. A server machine that is used for 5 years has 76% of its life cycle energy in usage and only 24% in production. That goes right along with this article on Google:
He estimates the system power consumption of a single dual-core processor system - which he described as a "successful attempt to reduce processors' runaway energy consumption" - at around 265 watts, which requires another 135 watts of power to cool the system down within a data center. "Over four years, the power costs of running a PC can add up to half of the hardware cost," he writes, and adds: "Saving power is still the name of the game, even to the extent that we shut off the lights in them when no-one is there."
To sum up, unlike cars or refrigerators for home usage of PCs most of the lifetime energy is used in production of the PC rather than in the electricity to run it. That amount of energy for a 2000 PC and 17" CRT monitor was estimated at 6400MJ which is the energy equivalent of almost 50 gallons of gasoline. To reduce energy usage on PCs it makes sense to focus as much on making the production of them efficient as it does on making them run efficiently. It is also important to try and keep a machine running for as long as possible without upgrading. The usage of the machine as a home, business or server machine also makes a huge difference in the life cycle energy analysis. The methodology and data of this report are suspect so I would like to see a new version of this analysis run with 2006 machines, both desktop and laptop.


Anonymous said...

Great post. Realistically, do you think that the gasonline equivalent is an accurate way to measure c02 emissions from a computer? This method compared to using a coefficient to convert kwh to c02 yields much different answers.

Fat Knowledge said...

No, I would think that the coefficient to convert kwh is closer. I was just looking at total energy, not co2 emissions here. As electricity is usually less co2 intensive (unless coal is being used) I would think there is less co2 than what would be emitted by 50 gallons of gasoline.

And I should also note that this info is out of date and I think the number are much lower now, as computers have become cheaper, and I am fairly certain, and less energy intensive to produce.

Post a Comment

Note: Only a member of this blog may post a comment.