Graham White: My Notes

Way back at the end of July I started looking into how to graph current cost data. I was looking at this from the point of view of putting the information together on a web page somewhere and potentially sending out useful information via e-mail to me and Beth on our power usage.

Recently I've been looking into this again and over the past couple of weeks we've been receiving a daily HTML formatted e-mail with an attached graph image that might look something like this:


This is a really handy reminder each day to try and be as green as possible but also to show us how we did on the previous day. The graph shows two lines, the red one showing the actual power usage in watts and a straight blue line showing our average usage for that day.

Also attached to the e-mail are the charges and a few summary statistics for the day. So on the example above you can see we averaged 240 watts throughout the day with our minimum usage recorded as 61 watts and maximum usage as 2359. The total usage doesn't mean very much without time associated with it, but I've listed it there all the same. The total can be used to work out the kWh reading for the day, total divided by 1440 minutes in the day (and points on the graph) divided by 60 minutes in an hour.

The costs are rather crudely calculated. We're charged for electricity on two levels, our first 225kWh per quarter are at a higher rate than any usage thereafter. For us, this averages out at around 10 pence per kWh cost so that rough average is used to calculate an estimate for the total cost in GBP.

This only a really simple way in which the current cost data can be used, Dale is way ahead of me on other applications for the data. I intend to extend the e-mail we're getting at the moment to integrate into a little web system we can use at home to log in and look through historic records of our usage and also produce other summaries such as weekly or quarterly usage. Also, the ability to add notes to each day will needed so when we look back over historic data we might have recorded what particular spikes in our usage are. All this is an effort to educate ourselves more in our power usage so we can change our behavior to be more efficient with it.

After hooking up my Current Cost Meter to a database recently I've been logging my power usage so the next step is to look at what I can do with the data. As I mentioned when I introduced my meter lots of other people have been looking into this as well.

I like Dale's idea for creating a daily energy bill so I've decided to aim towards that but include a graph of the days usage and possibly a link to a web page where I can look at more details of that days use.

With the web page idea in mind I decided to look into what I could do graph wise in PHP since that's my web language of choice. After looking through a few options I plumped for the obvious library, that of PHP Image Graph as it's part of PEAR so should be fairly decent, complete, well used and flexible. Here's where I'm at so far:
That's just a small snipped of 1 days power usage when I was cooking dinner. You can see I turned on my steamer (approx 800 watts) at about 5:30pm then went on to turn on the oven which heated up then cut in and out while it was cooking before I turned it and the steamer off just before 6pm.

As you can see, I'm pretty much there with the graphing side of things. The only issue left is to sort out the X axis. While it might appear to be a bit crazy, the current cost meter only samples approximately every 6 seconds. Because of the different number of samples in each hour and with a few rounding errors, the X axis appears not to be linear. Hopefully, I'll solve this by ensuring a consistent number of samples in each time slot.

Overall, PHP Image Graph seems very powerful. You can graph pretty much anything with it in all sorts of formats and have relatively complete control over the way the graph looks. I'm not at all artistic so mine looks very plane! There is, however, a reasonable barrier to entry. Documentation for the API is relatively complete if you know where to start, but documentation in general seems really rather thin on the ground and very sketchy. But, thankfully someone produced a great set of demos for a PHP conference that really show off a lot of the features. The source is available too so it's easier to work out what to do with your own graphing and how to get started. I'll probably blog my source code at some point, it's not very long, but definitely requires some tidying before it's ready for a public outing.

What I want to do next is take the graph and produce some sort of daily "bill" e-mail. That will have a link back to a private area on my web site that would allow closer inspection of certain areas of the graph. This is just the next step in education for how to save electricity by learning how much different combinations of devices use together. For example, is it better to steam fish in a steamer, grill, oven cook, etc. Hopefully, I'll be able to apply those sorts of questions all around the house to work out how to be as green as possible.

A couple of guys at work have managed to lay their hands on a funky new device called a Current Cost meter. It's pretty simple to understand, it measures your household electricity usage and displays it on the screen (left). To do this it comes in two parts, you hook a plastic loop around the live feed going into your fuse box (mine is in the garage) which is connected to some electrical wizardry I don't understand to measure the current without interfering with the wire, and a wireless transmitter device to send the data to the unit you just plug in somewhere in your house.

I jumped at the chance to get one of these things and was one of the first batch to get them at work and a little community of current cost users has built up pretty quickly. That said, pretty much everyone else has beaten me to blogging about it with some very cool results. So here's the list (that I know of) right now, in no particular order we have...

• Nick O'Leary
  
• Roo Reynolds
  
• James Taylor
  
• Dale Lane
  
• Rich Cumbers
  
• Andy Piper
  
• Ian Hughes

As an aside, Dale's just joined my department at work (horay), and Roo is leaving IBM (boo and much sobbing).

There's lots of other people looking into these things too, but those are just the blogs I know about. This type of device isn't exactly new, but what makes this one exciting is the data port and connectivity to a computer. With such a community there's been some cool work done such as graph power usage and send yourself a daily electricity bill. The device itself can't really save you money and definitely doesn't save energy (you have to plug it in) but what it does do a great job of is educating you about exactly what all your household devices like to eat for breakfast.

I'm just starting out at home having had the screen in the kitchen for a couple of months already with looking at what I can do with it while connected to a computer. I'm now logging all the data to a database so I guess one of the first things I'll be doing is graphing my own data. From that I hope to learn even more about what my house does with its power. For now though, the only graph I have (above) is that produced by sending my data via a Nanobroker to Andy Standford-Clark's server. This should hopefully be another fruitful little project for my new (and now you know why it's low power) home server.