A dispatchable datacentre energy system is provided. The system comprises a power conditioning system for providing conditioned power to a datacentre; wherein the power conditioning system includes a primary battery system for providing a primary energy reserve to the datacentre and being available to supply power to a grid operably connected to the datacentre in response to a dispatch request from a grid operator. A secondary battery system provides a secondary energy reserve to the datacentre and being available to supply power to the grid in response to the dispatch request. A power generation system provides a third energy reserve to the datacentre and being available to supply power to the grid in response to the dispatch request. A controller is provided for predicting grid conditions and being configured for selectively controlling at least one of the primary battery system; the secondary battery system and the power generation system in response to the predicted grid conditions; and wherein the controller is responsive to the dispatch request to adjust power consumption of the datacentre from the grid or power supply from at least one of the primary battery system, the secondary battery system and the power generation to the grid.

A plurality of DC power supply apparatuses which are CVCC power supplies have output sides connected in parallel to a load. In each of the plurality of DC power supply apparatuses, an upper limit current maintained in a CC mode is designed to be equal to or lower than a rated current. Output voltages from the plurality of DC power supply apparatuses are actually different from equivalently set reference voltages Vr, due to variation within an allowable voltage range of the load 120. The plurality of DC power supply apparatuses operate to supply currents, as being accompanied by transition from a CV mode to the CC mode, sequentially in the descending order of magnitude of actual output voltages thereof, with increase in output current to the load.

In systems and methods, a power distribution system provides power for multiple chassis installed in a rack. Two or more power supply units (PSUs) are installed in the chassis and may draw power redundantly from separate power grids supplying power to the rack. A first PSU of the chassis is coupled to one power grid and a second PSU of the same chassis is coupled to another power grid. Upon a failure in the second power grid, power drawn from the first power grid by the first PSU is limited according to a first current limit specified in a first emergency rack protection policy of the rack. Upon a failure in the first power grid, power drawn from the second power grid by the second PSU is limited according to a second current limit specified in a second emergency rack protection policy of the rack.

Aspects of the disclosed technology include techniques and mechanisms for controlling energy usage within data centers. An energy control system (ECS) may use measurement sensors that are integrated within different energy sources and different sources of energy consumption to gather energy usage information. The ECS may analyze the energy usage information using control algorithms. The control algorithms may analyze a current operational state of the data center and identify control modes containing control actions for enhancing and managing the performance of the data center. The ECS may elect an identified control mode and may transmit, via the measurement sensors, instructions for the energy sources and the sources of energy consumption to execute the control actions outlined within the elected control mode.

Power draw stabilization is provided. A target power consumption of a source load is determined. The source load is generated by electronics supplied power by a primary power source through a power rail. The power rail is coupled to a capacitor bank by a bi-directional converter configured to smooth fluctuations in power drawn from the primary power source by performing mode switch operations. The mode switch operations include, in response to the source load exceeding a target power consumption, controllably switching to a second directional mode that directs current released from the capacitor bank to the power rail. The mode switch operations further include, in response to the source load dropping below the target power consumption, controllably switching the operational mode of the bi-directional converter to a first directional mode to direct current from the power rail into the capacitor bank.

Systems and methods are described for powering a load, such as a data center, with renewable energy from a renewable energy source. When the renewable energy is greater than a demand of the load, excess renewable energy is used to power a hydrogen production device or charge a battery depending on whether or not the charge level of the battery satisfies an upper threshold charge level, respectively. When the renewable energy is less than the demand of the load and the charge level of the battery satisfies a lower threshold charge level, the load is powered with energy from the battery. When the renewable energy is less than the demand of the load and the charge level of the battery does not satisfy the lower threshold charge level, the load is powered and the battery is charged with energy generated by the hydrogen-based energy generator.

In some aspects, techniques may include monitoring a primary load of a datacenter and a reserve load of the datacenter. The primary load and reserve load can be monitored by a computing device. The primary load of the datacenter can be configured to be powered by one or more primary generator blocks having a primary capacity, and the reserve load of the datacenter can be configured to be powered by one or more reserve generator blocks having a reserve capacity. Also, the techniques may include detecting that the primary load of the datacenter exceeds the primary capacity. In addition, the techniques may include connecting the reserve generator blocks to at least one of the primary generator blocks and the primary load using a computing device switch.

The present disclosure is directed to a system that employs fuel cell-based power generation for various loads, such as data centers for artificial intelligence (AI) model training. The system utilizes various modules, such as different types of energy storage devices, to supplement power output by the fuel cells, as well as store any excess power generated by the fuel cell systems. As a result, swings in the power output by the fuel cells are minimized and the life of the fuel cells may be extended.