Solar Array Control at Imperial College, London
It is proposed that the student lab demonstration of photovoltaic energy generation is used as the basis for a remote experiment. The ultimate aim of the experiment will be to investigate the effect of network latencies and errors on our ability to control equipment remotely. The initial aim will be to allow remote users to access and use the Imperial College equipment for their own student lab demonstrations.
Imperial College has a solar panel lab demonstration that is used to teach students about photovoltaic energy. In its current form, the lab is supervisor led. The equipment is set up before the students arrive and the students then make estimates of the efficiency and energy output of the solar array under different conditions. It is only minimally interactive but is a good introduction to solar power and interfacing to the grid and does involved remote monitoring of actual hardware.
The main learning points from the lab are: how energy yield is affected by time of day and weather conditions; how the orientation of the panel helps to improve the yield; how the power electronics can control the output of the panel and how shading of the panel affects the VI characteristic.
The panels are connected to the grid via an inverter that can be used to sweep through the panel's VI characteristic (the results are displayed on the Student View application). This process is fully automated and the only action that can be carried out on the inverter is to start or stop this process.
The Solar panel system consists of 4 frames with 2 actuators each (see above). All four panels are controlled by a National Instruments Compact RIO that runs a real-time OS that operates headless (i.e. no GUI). There are two user interfaces, demonstrator view and student view that communicate with the cRIO using Network Shared Variables. The two interfaces are standalone Windows applications that only need the free Labview Runtime Engine installed. Network Shared Variables use a proprietary protocol that is constrained to operate within a closed network; National Instruments has chosen not to document the packet contents and message passing.
It is proposed to bridge the communications between the hardware and the GUI using a secure protocol that will allow us to run the experiment from another country. Delft has expressed an interest in running the demonstration as an actual student lab for Delft students as nothing similar exists there at this time.
For the initial implementation, secure communication will be achieved using a VPN. Future work will involve interfacing the CompactRIO to Agentscape and using the secure communications of AgentScape to transport the relevant data.
The proposal for this remote experiment was only raised in August, so progress has so far been limited. A mechanism for reading and setting LabView shared variables using Java has been implemented. The next step is to make this into a separate module that can host the shared variables at multiple locations so that many Student View applications can run in many locations at once. This will minimise the amount of work needed to make the user interface compatible with the Remote Experiment.
It is hoped that a simple implementation using a VPN to transport the data between the Shared Variable Hosts will be achieved before the year end (see below). This will allow us to start testing communications latency. Next year will involve putting the secure communication through AgentScape. The work of characterising the communications link between variable hosts will continue; this work will then be able to include the effect of AgentScape on the communication link.