Monitoring GB grid carbon-intensity in real time

[DamonHD]

Here's a mockup of the 'traffic light' grid status page I'm creating:

http://www.earth.org.uk/_trafficLightsOutput.html

Soon that page will update every 5 minutes or so, and will be driven by live data from the GB (ie all of the UK except Northern Ireland) grid managers to compute its result.

The idea is that, to help save the planet (ie generate less CO2 for whatever it is you want to do) you don't start your dishwasher or washing machine or vacuum cleaner or whatever if that shows RED.

I'm going to add more techie details at the bottom of the page for example, but is the basic idea clear? Might you use it or something like it?

There will be an accompanying flag file/URL that can be tested with your client getting a 404 if GREEN, else a 200, so you can automate servers, etc, to use less power when the grid is not in good shape.

Rgds

Damon

[wooferhound]

I think you should include some helpful information, like what time of day is the grid normally overloaded, and when is the grid normally freely available. Maybe a graph of the last week or the last few days. Possibly even predictions of problems Times.

 

[DamonHD]

Hi,

All good suggestions!  Thanks.

Generating the graphs on the fly is going to be the hardest bit of that for me, and might scare Jon Doe, though remains an attractive idea.

Predictions are also hard, especially of the future, but yes, I'm going to have a go soon.

It should be simple for me to at least add the time of the last maximum and minimum seen, though I can tell you in advance that 4pm to 9pm is bad and midnight to 4am is good most days in the UK!  B^>

I may do another mockup page with more info over the weekend.

Rgds

Damon

[DamonHD]

Hi,

I've updated the mock-up with times of maximum and minimum, though I need to do better than that to cover the points you make.

Rgds

Damon

[DamonHD]

Hi,

Next attempt... Mockup page at URL above now includes a chart near the bottom of red/yellow/green status by hour, which is a reasonable prediction without actually being a prediction!

The data on that page is not quite real yet.

Rgds

Damon

[wooferhound]

Cool

Maybe an explanation of some of the terms, Like: what does "420gCO2/kWh" really mean. a few other terms on there that might need some definition.

You need not be really accurate with your predictions. I'm just suggesting that you may show people what the Trends are.

[ghurd]

Damon,

Thought you may like this,

"In terms of savings alone, it's possible to save over $2.8 billion in the US alone through turning off your PC while in the UK it's nearer £300 million and €918 million in Germany."

http://www.1e.com/energycampaign/index.aspx

G-

[DamonHD]

Indeed yes.

Tried to persuade a previous client to turn off their screens etc but the entire company went down the tube.  Will now work on my current very large IT-driven client given that the worldwide head of IT sits about 5 yards from me...  B^>

Rgds

Damon

[DamonHD]

Thanks again for all your input.

Here's a 'beta' version of the final page, not quite official yet for a number of reasons.

http://www.earth.org.uk/_gridCarbonIntensityGB.html

Any further comments would be welcome.

Rgds

Damon

[DamonHD]

FWIW, the page has officially gone live...  B^>

Rgds

Damon

[DamonHD]

Anyone (esp in the UK) want to try this URL on their mobile phone:

www.earth.org.uk/_gridCarbonIntensityGB.xhtml

I will provide a shorter alias URL one way or another in future if it works.

Rgds

Damon

[DamonHD]

As well as one of my UK-based servers (in co-lo) that turns off all non-essential activity unless my grid intensity status is 'green', ie below average CO2 emissions per kWh used, I have now dug out my old X10 kit and installed HEYU (www.heyu.com) on my Linux server in order to turn on my dishwasher when the status is 'green' overnight, thus deferring one of our largest movable loads automatically.

Tonight is the first attempt, so I may burn the house down! %-P

(The HEYU stuff worked right out of the box with no pain at all, BTW.)

I also set one of the digital outputs on my k8055 USB I/O board at the same time, so I could use that to drive an opto-isolated SSR directly. I can't instantly find a suitable device at RS (the dishwasher takes the whole 13A) but it is the basis of a fall-back plan.

Wiring up the X10 means that I could reactivate the PIR sensors to create an automatic heating set-back mechanism in winter when everyone is out, for example.

Rgds

Damon

PS. I may have double-posted this, but I think the previous copy went AWOL...

[DamonHD]

The house didn't burn down, nor on the following nights when we've done the same, and last night we had the dishwasher and washing machine on overnight thus deferring maybe 1/3rd of our daily load off-peak and to low carbon intensity, so it can be done!

The only problem with the washing machine is ending up with more-creased-than-usual clothes even with rinse-hold, so the trick there seems to be timing the cycle to finish not long before you're up and about to take the clothes out, but still starting and doing the energy-intensive water heating before intensity starts to rise again in the morning.

(Oh, and today I met Dale Vince, CEO of Ecotricity (http://www.ecotricity.co.uk/) which was interesting as you might imagine!)

Rgds

Damon

[DamonHD]

OK, for those of you (a) in the UK and (b) on Twitter, you might like to look at the new channel/ID I have just set up @EarthOrgUK which will send an update each time grid status changes between RED / YELLOW / GREEN states.

You can even have the message forwarded directly to your mobile phone if you're on Vodaphone.

I'd be pleased to get some feedback!

Rgds

Damon

PS. The normal Web page is http://www.earth.org.uk/_gridCarbonIntensityGB.html

[Hairypizza]

This could become a victim of it's own success.  If everyone switches on appliances on receipt of a tweet the surge would have to be made up from somewhere.  Maybe a while before that's a problem though.  I do like the idea though, I'll be following it.

Also, not just Vodaphone - anyone with an iPhone, Blackberry, HTC (eg G1) can download the necessary apps.

[DamonHD]

The current grid demand/intensity pattern is still true after 30% of UK domestic demand is already shifted off-peak with special tariffs such as "Economy 7".  And if we do get 33GW of windpower (which is more than 50% of peak demand) then there will start to be a lot of cheap energy to be soaked up at short notice on occasion during the day as well as overnight, backed by dynamic-pricing tariffs to business and retail IMHO.  People will then be watching their current price from their supplier rather than anything I produce except possibly some dedicated greenies!

I won't move the market, regrettably, but I will maybe very slightly nudge it in the right direction, as I hope I did in a previous life with UK Internet services...  B^>

Rgds

Damon

PS. Hope to have an announcement very soon about one of the UK green 'leccy suppliers doing a version of my calculator, and it may have slightly different criteria to mine, already helping to spread any eco-rush to flip on the dishwasher!

[Madscientist267]

I have to agree with hairypizza in the sense that this seems to be self-negating. I want to say up front that I'm not exactly bagging the concept, but more scrutinizing it's usefulness...

Power conservation should be done on a more subconsious level by everyone all of the time. Given the fact that this won't happen since the majority of energy consumers aren't really even aware of what their appliances actually use, they don't know which ones are best to turn off during peak times.

A real world example of this hit me like a brick when I was setting up a 47" LCD TV last year... It draws almost 400 watts - 50 more than the 27" Tube version it replaced. I would have never guessed, and I've been dinking with electronics for 25+ years and know the key differences in the technologies!

Now imagine that most consumers only see the "energy star" and leave it pretty much at that in their minds, assuming that it is helping to "save the planet"... Follow me on this seemingly ridiculous example...

If the entire world were made up of people of the kind found on this forum, this would be an EXCELLENT idea, since intelligence could prevail.

Since this isn't the case, it seems to me all it will do is peak people's interest, since "green" is such a buzz word these days, and have the effect of "turn on the computer - I want to see if I can zap my coffee in the microwave".

The result would be, in the time it would take to boot the machine, get on the site, and find out the current state, the computer will have drawn a few times what the microwave would have to warm the coffee, all because of ignorance. They know that 1000W means the microwave uses a lot of power, but most people don't know that between the computer and monitor, they could be approaching 500W. If your intent is to zap your coffee for 30 seconds, you'd have to be able to boot up your machine, get your info, and shut it back down in under a minute to see any benefits. Not likely.

Add that to the idea that firing up the computer is "wasted energy" (doesn't help heat the coffee), and you're out that much more!

Just my two cents...

Steve

[DamonHD]

I certainly don't want people to be cranking up their 1kW games PC to see when to make  a toasted sandwich: thus the alternative delivery mechanisms of phone-viewable Web pages and SMS 'tweets', flags to allow on-line servers to automatically regulate their consumption, and graphs, etc, to help people predict when the next good time is without turning on any computers at all!

Also, more subtly, this is not about consumption per se, but reducing emissions from an appliance you were going to run anyway.  You consume the same kWh but cause the emission of less CO2.

I agree it's a tricky message to get across and I'm still working on it...

Rgds

Damon

[Madscientist267]

Dont get me wrong, its an intriguing idea. It will be interesting to see how it works out in the end.

I suppose, actually, this could turn into an automated system, given enough resources and interest... Machines that sleep during peak times, monitoring the grid usage and using trending information to schedule operation.

I'm just not so sure of the usefulness when the human element is involved.

Steve

[DamonHD]

Hi,

I am already doing exactly this with one of my servers (and if you have any UK servers I can tell you how to hitch them to the same signal)...  Indeed all my servers worldwide try to minimise peak-time power consumption, though the rest are purely done by time, postponing anything non-essential until 11pm--6am local time.

As for the human element: awareness.  Really the main message is "Don't run your dishwasher right after dinner because that's probably the worst possible time; stick it on a timer and run it overnight."  All the rest is technobabble for now...  B^>

Rgds

Damon

[Madscientist267]

Right, thats more along the lines of what I meant by automation... LOL not just computers.

A dishwasher or washing machine that waits until the lowest trended usage time (based on say, the past week) and then runs it's load. Even refrigeration/AC maybe - During peak times it allows for more tolerance to temps, not running unless it absolutely is getting to a critical level.

I was heading more in the direction of just an embedded module that connects to the web. A bit futuristic at this point, but not out of reach. A TV that keeps tabs; if usage gets above a threshold, it warns the 'audience' that maybe TiVo would be a better choice at the moment...

Steve

[DamonHD]

Hi,

Kewl: my stuff finally made it live at the site of the UK green energy supplier Ecotricity:

http://www.ecotricity.co.uk/about/live-grid-carbon-intensity

This is a somewhat trimmed page, but we'll work on fleshing it out (including a credit to me!) later...

Rgds

Damon

[Madscientist267]

Nice... Its coming together... Reading the light, I see a potential addition... ""... How about a history graph? Takes the guesswork out.

Steve

[DamonHD]

Hi,

My version of the page does have such a graph, and I agree that some more human form of such a graph should on any such page.  I've already sent a note to that effect to the CEO!  B^>

The difficulty is finding a form that is instantly understandable for the average Joe and has useful predictive power.  I think my bar chart isn't awful, and I have an idea for the Ecotricity page, but I'd really like to hear any ideas that others here have...

Rgds

Damon

[Madscientist267]

Just something showing the trending of the last 24 hours, stock exchange or google trends style. It doesn't need to provide much more than an indication of waxing/waning energy usage, so that users can gauge for themselves what the best time will be based on history. Maybe even more than 24 hours, say 72 (or even a week?) so that they can see what day even is best; you know, for things that can be put off for extended periods.

Something like a 3 color horizontal background, with the graph fluctuating, using actual value (not a 'binary' state taken from just the 'traffic light').

Steve

[DamonHD]

But do ordinary non-geeky people in the street, not with the benefit of degree-level education, read graphs?

I just don't know, but I suspect that graphs are just wiggly lines to many people even though you and I are fortunate enough to be able to take them in quickly, which is why I came up with the basic traffic-light view of carbon intensity.

As I say, I have a couple of charts on my version of the page (and indeed a 7-day chart would be a better tools than a 24-hour one if it doesn't simply overwhelm, since weekends are fairly different to weekdays in intensity terms):

http://www.earth.org.uk/_gridCarbonIntensityGB.html

But is it really simple to understand?  I doubt it.  I have to think about it quite carefully sometimes and I produce the damn things...

Rgds

Damon

[DamonHD]

And it got a decent write-up on TreeHugger:

http://www.treehugger.com/files/2009/07/uk-grid-carbon-emissions.php

Rgds

Damon

[Madscientist267]

The bar graph version is a bit intense, I must admit. Even knowing what I'm looking at, it still has the immediate tendency to overwhelm with TMI. However, a line graph with nothing other than vertical day lines and the 3 'states' of the grid (by background color) wont be intimidating.

For those that are more curious of the details, a "click here for more detail" link is appropriate, and that is where the detailed bar graph with the extra data would make more sense.

Something like this... Forgive the crudeness - I trended "renewable energy" for the graph and then tweaked it rather quickly in gimp to give the basic concept. The finished product tho really doesn't need any more detail than this:

Know what I mean? Super simple, and anyone can tell if it is on its way up or down...

Right "now", it's Saturday, and looks to be an ideal time to fire up the dishwasher and washing machine!

(And I'm curious after toying with this, why lack of search volume for renewables right before a new year? Hmmm...)

Steve

[DamonHD]

Understand that the green/red boundaries are not fixed, and represent (currently) the upper and lower quartiles.  That makes it more robust in the face of quite large dynamic changes in the absolute numbers, such as between weekdays and weekends, and even errors in the data (which happen), but nixes your nice layout.

But keep coming with the ideas.

Rgds

Damon

[Madscientist267]

Correct me where I'm wrong here, but if I understand the whole point of the project, dependent on how one looks at this, there may be too much 'outside the box' thinking, as applies to complexity and accuracy of the instrumentation for Joe Q. Average.

At some point, you had to decide how the thresholds were defined, and what makes a given set of production parameters or degree of load produce more and less CO2. But the average user will not, as much as the site is to the contrary, be concerned with dead-on accuracy. For a trending graph, you're really only concerned with whether it is headed up or down. The colors just give them an idea of how close they are to the thresholds. A baseline.

While I'm not schooled on the ins and outs of industrial power generation, distribution, and monitoring, I'd be willing to bet that the total raw energy consumption roughly follows the CO2 emissions for any level of power used. RE is such a small contributor, and represents the most dynamic component, yes? If you were closer to 50% (or even 25) RE vs coal/nuclear, then the graph would have a significant error to be dealt with on a continuous basis.

However, if your boundaries are dynamic, then since your trend graph isn't plotting any numbers, just have the entire graph follow the current quartile thresholds and be done with it. Trending is about ease of perception, not nitty gritty details. If you're worried about someone calling you out on it, throw a disclaimer out there with it, stating that there is a tolerance that you are allowing for spontaneous ratio changes in the sources of the total energy. I would imagine that 5-10% for something like this is probably perfectly acceptable. I know I personally wouldn't question it, and I'm imagining myself somewhere between you and Joe geek-wise.

Then, when they beg with the 'better details' link, you can hammer them with all the data they could ever want, and after Joe clicks the 'back' button on his browser, he still understands that it's not the best time for putting the dog in the dishwasher. Maybe after dinner.

Steve

[DamonHD]

Hi,

You are entirely right that:

1. Intensity closely follows consumption at the moment in the UK (though it's messier than you might imagine due do things like pumped storage and a significant interconnector to France's mainly-nuke grid and our "Economy 7" off-peak scheme that eats about 30% of domestic usage, and we're building RE quickly enough that I saw some very strange effects around midnight only a few days ago).
   
2. Absolute accuracy in numbers is not critical (but be aware that attempting to set thresholds by hand is already a fool's errand, so we have to have some way of doing it automagically, thus the interquartile nonsense), though it is interesting to see that gCO2/kWh is almost always higher than the UK government's preferred figure of 430gCO2/kWh.
   
3. A 'more details' link and a disclaimer are a good idea.
   
   

My 'traffic light' view was my main attempt to be 'out of the box'.

The heading up/down in the short term is unfortunately not a very good guide to anything.  At the 5-minute level of the data granularity it's essentially random for example.  But the predictive power of 24h or 7d ago is quite good.

But yes, we have to find the right simplifications, not get too hung up on the details, and disclaim and describe methodologies for those who are interested.

Rgds

Damon

[Madscientist267]

The traffic light is an excellent, already 'established' concept, that works excellent for at a glance for the moment. By "out of the box" I was referring to how far you want to go with details and accuracy for just the trends.

As far as updates, I would think just an average of what the grid has done hourly would suffice to give a granularity of 24 units per day, adequate for a trend, and not so much that the graph is obnoxious in size. Using a single column per hour would yield a graph that is 168 px wide, a good balance for any browser/bandwidth combo.

Basically, sounds like you're dead on with the whole idea. I'm fascinated - its interesting to watch this grow.

Steve

[DamonHD]

BTW, Prof MacKay doesn't agree but as it happens I think he's wrong B^> and exchanged a couple of emails with him.  He was kind enough at least not to knock my two efforts directly!

http://withouthotair.blogspot.com/2009/07/wouldnt-it-be-better-no-i-dont-think-so.html

It's a strange situation where it's the third-order effects that may important (like avoiding infrastructure build and use of storage) that may be more important than the apparent headline intensity numbers.

Anyway, robust feedback like MacKay's is good, and it forces a little bit more hard thinking on all sides.

Rgds

Damon