Every night at RIT I would walk by closed computers labs, but to my amazement the computer and monitors remained powered on. As an institution that promoted its sustainable practices, not turning the computers off when the labs were closed seemed hypocritical. Every year students would complain about this issue, but nothing was ever done. My roommate, Ian Mikutel, and I decided that we would do something about RIT not having a computer power policy.
Ian and I met with the Chief Technology Officer of RIT, and she agreed that the idea of either turning computers off or having them sleep when they were not utilized was a great idea. We pursued the idea further and started meeting with ITS, the organization that managed a majority of the RIT computer labs and FMS, which was responsible for providing electricity to the buildings. Ian and I decided to implement an actually computer lab trial to demonstrate that a computer power policy could be successfully implemented and to measure the power savings. After many meetings we decided that putting the computers into S3 sleep mode after 20 minutes was the best way to maximize student productivity while minimizing power consumption. Having the computers sleep instead of turning them off ensured that computers would still receive security updates and patches in the middle of the night.
FMS installed a power meter to monitor the power consumption of the 16 Dell desktop computers in the lab. This allowed us to quantize the power savings of having the computers automatically sleep after 20 minutes. After modifying some files and fixing a couple configuration issues the computers automatically went to sleep and automatically awoke in the middle of the night to install security updates. We monitored the power consumption for around 10 weeks, and discovered that having the computers sleep reduced the labs power consumption by 66%! Everyone working on the project was amazed at the results, and when we extrapolated this to the thousands of computers at RIT, the environmental and monetary impacts are massive. Ian and I presented to the RIT’s Committee on Sustainable Practices which was very impressed with our presentation and decided to begin implementing computer sleep mode across campus. Ian and I have both recently graduated so we will not see the end results of our work, but we are both extremely proud to have made such a difference in reducing the power consumption across RIT.
In the past month my friend David Brenner and I have made a great deal of progress with the wind turbine remote Arduino power meter. The idea behind the project is to log the amount of power generated from the wind turbine and have the data automatically be uploaded online where it can graphed. There are similar products on the market, but they are designed for much larger turbine and solar arrays, and thus are expensive. We wanted to utilize the Arduino micro-controller as it versatile and inexpensive. With all the parts the costs is around $80, which is a fraction of the cost of other solutions.
The power meter works by reading the current flowing from the turbine. We are using a Hall Effect current sensor as it is very efficient and can handle up 30 Amps. Based on the amount of current, the sensor outputs a voltage which can then be read by the Arduino’s analog to digital input. An issue with the Arduino’s ADC is that it is limited to only 1024 values ranging from 0-5 volts. This means that the Arduino is not very accurate and can only sense a 4.8mV change. The issue we had was that the current sensor’s output voltage changes by a very small amount and the change couldn’t be read by the Arduino. To correct this, we fed the current sensor into an amplifier stage to increase the output voltage so the Arduino could read smaller current changes. The current sensor outputs 1.5v with no current, so to ensure that we weren’t amplifying the voltage when there was no current we needed to implement a 1.5v offset.
The Arduino microcontroller reads the data from the ADC input and then sends the data through the Internet to a server. We are utilizing the wireless repeater portion of the wind turbine to transport the data through the Internet. The server then reads the data and stores it in a mysql data base so graphs can be created. The Arduino sends a new power reading every one second, so the graphs can be extremely accurate if we wish. The remote power meter is currently in field trials to ensure the basic functionally works correctly.
This weekend I displayed my wind powered wireless network repeater at the RIT Innovation Festival. The turnout was great but the weather consisted of light showers and the wind was very calm. I explained the project to many people and showed how it was operational by surfing the net on an OLPC by RIT’s quarter mile. Here are the posters I displayed at my booth explaining the project.
There are a lot of great sites on the net that explain how to build your own wind turbine. My favorite is by Michael Davis and you can view his site here. Another great place is gotwind.org and the site also has a discussion form with many other DIY wind turbine builders.
How the Project Works:
My wireless network repeater consists of four parts: wind turbine, batteries, charge controller, and wireless network gear. The wind turbine consists of an electric DC motor which is connected to three blades. When the wind blows, it causes the blades to spin. This produces a DC voltage which depends on how fast the generator is spinning. My turbine produces 12v at around 300RPM. Above this speed causes current to flow into the batteries which charge them. The issue on days with light wind is that the blades won’t spin fast enough to reach 12v, so the batteries won’t charge. I am investigating a voltage boosting circuits which would trickle charge the batteries on days with light wind. To prevent the batteries from overcharging on windy days I use ghurd’s charge controller. When the batteries voltage reaches a certain level, the controller dumps the power to a set of load resistors.
For the wireless network gear, I use two Linksys WRT54G routers with the DDWRT firmware. The first router acts as the backhaul link and is setup in client bridge mode. It has a 18dBi 2.4Ghz parabolic antenna so if can communicate over far distances. The second router is connected to the first one via an Ethernet cable, and is set up in access point mode. DHCP is disabled on both units, so any devices connecting to the repeater will pull an IP address from the DHCP server from the host network. In my case I connect to the wind turbine network which had an SSID of “Wind Turbine,” and I receive an IP address from RIT’s network. My laptop looked/worked like any other computer on the network even though I was located in a field far away from RIT’s network infrastructure.
Last month I installed my wind turbine wireless repeater project at RIT. I planned on adding information regarding the project to this site, but my friend Dave Brenner beat me to it. Dave wrote an article on his site regarding the installing of the turbine and Make Magazine linked to it on their blog.
