Student Presentations: June 8, 2011
An Alternative Approach to Demand Response
Srikanth Iyer, Brian Lumpkins, and Matthew Murray
Currently, the infrastructure for power delivery does not allow for dynamic allocation of the load based on demand. Most of the research on demand response has focused on designing sensor systems that would be placed in consumer devices in order to estimate the power required by each device. This system requires the design and installation of complicated sensors in nearly every consumer device. We believe that such systems are too complex to be implemented rapidly in large numbers. We are proposing a system that uses embedded client controls to allow for the continuous monitoring of power usage as well as the ability for consumers to send requests in order to use high load devices in scenarios where turn on time is not a high priority. In this way, the power generator can determine when to distribute power to these devices in a way that ensures the load remains constant, essentially eliminating spikes in the demand response that result when many consumers turn on devices simultaneously. This system would replace the old one because less power would be needed at the source. The source will be notified via a request that additional power will needed to be generated. Then that additional power will be distributed through the grid, with a response being sent from the source to the device indicating that it now has the power it needs to operate.
This approach is unique in that we are not attempting to predict demand. Instead, demand is explicitly stated and then satisfied as soon as possible. The key to this strategy is that it will only be applied to power-hungry devices for which response time is not important. For instance, it is not terribly important whether your washer machine starts its load immediately after you press start or after some brief signaling delay, whereas it is important that a light bulb turn on immediately after you flip the switch. Thus, the washer machine would be a candidate for our system, whereas light bulbs would not. It is important to note that some minimum amount of power be continuously supplied to each building in order to power devices where response time is important.
In addition to design and implementation of the control system, the research must also analyze the cost benefits of this new system versus the current approach. It is important that the costs of designing, building, installing and maintaining the control hardware and sensors are compensated, in a reasonable amount of time, by the reduction of load at peak demand times. The delay in the response of the system to consumer requests must be kept small in order to ensure that consumers are satisfied with the performance of the power distribution system. However, this should not be too much of an issue since the only devices that would use this system are relatively insensitive to response time.
We believe that this system has several important advantages over systems currently being researched. First, it requires minimal set-up. Other systems require that devices be connected to the network and that sensors be designed and installed. Our system only requires the development of a simple protocol to allow each device to request a certain amount of power and to allow the source to signal when the power is available. Second, this system maximizes performance improvement per cost. We are only interested in power-hungry devices where there is potential for large improvements in the performance of the grid. Also, since our system does not require any sensors, it is relatively cheap. We believe that this minimal set-up overhead and good performance improvement per dollar of cost could enable rapid and effective deployment of systems designed using the ideas outlined above.
Presentation Charts
Presentation Charts in PDF