MODELING, ANALYSIS AND SIMULATION OF A DC GRID SINGLE ENDED PRIMARY INDUCTANCE CONVERTER FOR DC LOAD
MODELING, ANALYSIS AND SIMULATION OF A DC GRID SINGLE ENDED PRIMARY INDUCTANCE CONVERTER FOR DC LOAD
MODELING, ANALYSIS AND SIMULATION OF A DC GRID SINGLE ENDED PRIMARY INDUCTANCE CONVERTER FOR DC LOAD
Abstract:
This research studied a circuit topology of a solar based Single Ended Primary Inductance Converter (SEPIC) system and a battery furnishing a permanent magnet dc motor with current via a dc grid. The complete model equations of the entire system were derived. Harmonic balance technique…
This summer, during the Dutch hackercamp SHA2017, we implemented an actually decentralized DC microgrid as part of altpwr.net. The grid functioned on a nominal voltage of 42 VDC, because < 50V is safe and, because.
One of three 42V connection boxes; this is where users could connect cables to their tents and appliances.
In recent years, in IT as well as in other industries it is becoming increasingly apparent that a swift transition to renewable energy sources is the only viable option. Google, Apple and Microsoft have all committed themselves towards (at least) 100% renewables.
Now, a group within the international hacker community, traditionally at the forefront of experimentation, creativity and critique with regards to technology, has initiated an experiment that simultaneously makes the safe and ad-hoc distribution of (sustainable) energy possible while being significantly easier to implement and maintain.
AC/DC, hierarchy and decentralization
A traditional AC grid would follow a top-down architecture, where every aspect of power distribution needs to be calculated and designed beforehand. There would typically be one or several units of production (i.e. generators) which would have to be synchronized and exchange of energy between end users is only possible by power through hubs higher in the hierarchy.
Hierarchical network 'AC grid'
In a DC grid, however, power can be exchanged by any two connected nodes, in any direction. Furthermore, no synchronization is required; the flow of electrical power behaves quite a lot like water in a system with pumps, dams and channels. Moreover, because production, storage and consumption can be integrated anywhere within the grid, the storage and production can be moved as close as possible to consumption, significantly reducing the amount of energy that needs to be exchanged between remote parts of the grid.
Distributed network 'DC grid'
A Proof of Concept
Now those of you who have been following the Decentralized Society Research Project for a while, might remember an earlier post, nailing the microgrid about DC microgrids. In this post (the result of research done for Metabolic), I made the case for a truly decentralized DC grid and explore a possible architecture for such a microgrid.
The group who built the microgrid at the SHA hackercamp adhered to some of the same design principles and, effectively, performed somewhat of a 'smoke test' for a simplified version grid -- where energy flowed 'freely' rather than being actively routed across the grid, controlled by voltage and Ohm's law alone.
The grid successfully harvested energy from the sun, using panels delivering 72V, which was downconverted to 46V and then directly fed to the grid. Besides this there was a windmill feeding into the grid, there were some lead-acid batteries directly on the grid and there was a 'long distance' link (90m) to a remote area where electric carts where being charged.
As consumers, some laptops where being charged, there were facilities for phone charging, someone operated a modified oven and there was a fountain pump 'dummy load' available for testing.
Universal 42V DC laptop charger and modified 42V oven
High voltage interlinks
Besides the above, we were loaned two EZA2500 bi-directional 48 VDC/320 VDC converters from TDK-Lambda which allowed us to do experiments with a high-voltage interlink between two nodes in the grid. The higher voltage makes sense, as the cable losses with 42V can easily amount to tens of percents as does the voltage drop; a lot of energy gets lost trying to push more electrons through the cable rather than getting them to move with higher potential energy.
Schematic rendering of DC microgrid at SHA with HV interlink
These converters from and to a much higher voltage, would in theory allow for a self-regulating distributed grid spanning a significant part of the hacker camp and it's power hungry citizens. Likewise, similar technology could be used to create ad-hoc power networks for action camps, refugee camps or to create temporary power grids for hazard relief.
Within households, or in other user-facing situations, only the safe 42V would be exposed for 'hot-plugging' of devices. This, in theory, allows for essentially unqualified people to manage electricity within the house. Similarly, the high voltage interlinks between houses, or parts of a camp, could simply be connected -- although safety protocols should be adhered to more strictly in this case.
Experiences with high voltage interconnects
During the event, we attempted to wire up the two identical bi-directional converters such that we could transfer electrical energy from one part of the camp across a 90m distance to a (relatively) remote location. As previously mentioned, there was an actual use case as this particular interlink was costing us a hardly acceptable loss.
First, we succesfully powered one of the converters from the 42V grid. However, the converters required an (initial) setup over an RS-485 bus before they do anything. Although the protocol is fully documented, for this experiment we chose to run the accompanying configuration tool in a Windows virtual machine in order to save precious time during the event. After all, things like figuring out the (non-standardized) wiring for RS-485 through the UTP connector took some time.
After we managed to start up the first converter, it slowly ramped up the voltage from ~20V to 350V -- as reliably shown by the configuration tool. This, in turn, kicked up the other unit, connected to it over the 350V bus.
Which, in turn, started to ramp up the voltage on the 48V side until it produced a 'battery undervoltage alarm'. Apparently, these converters assume the LV bus to be connected to a 48V battery while the 350V is assumed to be a 'grid voltage'. In this way, the units could be used as some sort of UPS, although for our particular use case it took a lot of work before we figured out how to override this particular alarm. (It seemed that the integer value we entered into the configuration tool was not converted to the particular settings bitmap in a particularly straight-forward way.)
The two units interconnected through a long cable over the 350V bus.
Having successfully coerced the units to convert 42V to 350V and back down to 42V, we managed to reliably connect small loads, such as this USB car charger (after first converting it down to 12V).
Unresolved issues and research opportunities
Sadly enough, as we later tried to connect a higher load to the downconverted 42V bus (a 230V AC inverter driving a fountain pump) we had repeated failures of the fuses on the 42V side of the first converter. For some reason, the 42V to 350V conversion generated not only more load than we had configured it to, it also seemed to outstretch the limit of the fuse, tipping it in a matter of seconds after the pump started working.
Moreover, we had hoped to be able to fully configure the bi-directional operation of these units (scarce time and difficulties with setup proved prohibitive). In this way, the unit would autonomously determine it's direction of operation depending on the LV and HV grid voltages. When the LV would be high enough (above 42V) and the HV would be low enough (below 350V) it would start feeding the HV from the LV side. When the grid voltage would be equal to or above 350V and the LV is below 42V it should feed the LV from the HV grid.
In this case, it could be expected that some feedback loops and/or resonances occur, which would require careful tuning of parameters such as drooping. This should yield interesting results, particularly whether the unit's built-in modes of operation are supportive of such as scheme or whether active control from an external microcontroller is necessary.
As an added benefit, such an external microcontroller might allow for remote monitoring and operation of the units over a wireless link such that live experiments would be possible and realtime statistics could be gather. Such a system was already set up for the low-voltage side of the microgrid, although it wasn’t yet visualized in realtime. (It’s available in it’s raw form here.)
Towards a decentralized future
Overall, it can be said that the first application of the 42V altpwr grid was a great success. The intention of some of the initiators was to raise interest and test the hardware and concepts, in order to create a long-lived project that would extend over multiple hacker events and, through it's Open Source nature, to other events as well. Hence, we can expect the same setup to appear, in improved and refined form, at future hacker events.
As for the high voltage interlinks, we hope that future events will give us chances to continue the experiments we started and, ideally, that more converters will be available. This would allow us to extend the DC microgrid and provide us with insights in actual dynamics of the grid as well as stability of its operation.
