US10608433B1 — Methods and systems for adjusting power consumption based on a fixed-duration power option agreement (Part 2 of 2)
Document text
Research, not advice. Part of the Bitcoin research archive (October 2026). Claims labelled unverified, contested or fringe are reported, not endorsed; statuses of bills and rules are as of the date checked. Government, court and patent records are public domain; the research notes are CC BY 4.0.
gen
phase of three -phase AC voltage to the third group of six eration station control system 414 , the remote master control
racks 610. In other embodiments , the quantity of racks and 40 system 300 , or the grid operator 702 to go offline or reduce
computing systems can vary . power. As such , the datacenter control system 504 may
FIG . 7 shows a control distribution system 700 of the enable 714 the power input system 502 to provide power to
flexible datacenter 500 according to one or more example the power distribution system 506 to power the computing
embodiments . The system 700 includes a grid operator 702 , systems 512 or a subset thereof.
a generation station control system 216 , a remote master 45 The datacenter control system 504 may optionally direct
control system 300 , and a flexible datacenter 500. As such , one or more computing systems 512 to perform predeter
the system 700 represents one example configuration for mined computational operations (e.g., distributed computing
controlling operations of the flexible datacenter 500 , but processes). For example , if the one or more computing
other configurations may includemore or fewer components systems 512 are configured to perform blockchain hashing
in other arrangements . 50 operations, the datacenter control system 504 may direct
The datacenter control system 504 may independently or them to perform blockchain hashing operations for a specific
cooperatively with one or more of the generation station blockchain application , such as, for example , Bitcoin , Lite
control system 414 , the remote master control system 300 , coin , or Ethereum . Alternatively , one or more computing
and / or the grid operator 702 modulate power at the flexible systems 512 may be configured to perform high -throughput
datacenter 500. During operations, the power delivery to the 55 computing operations and /or high performance computing
flexible datacenter 500 may be dynamically adjusted based operations.
on conditions or operational directives. The conditions may The remote master control system 300 may specify to the
correspond to economic conditions (e.g., cost for power, datacenter control system 504 what sufficient behind -the
aspects of computational operations to be performed ), meter power availability constitutes, or the datacenter con
power -related conditions (e.g. , availability of the power, the 60 trol system 504 may be programmed with a predetermined
sources offering power ), demand response, and /or weather preference or criteria on which to make the determination
related conditions, among others. independently . For example, in certain circumstances, suf
The generation station control system 414 may be one or ficient behind - the -meter power availability may be less than
more computing systems configured to control various that required to fully power the entire flexible datacenter
aspects of a generation station (not independently illustrated , 65 500. In such circumstances , the datacenter control system
e.g., 216 or 400 ). As such , the generation station control 504 may provide power to only a subset of computing
system 414 may communicate with the remote master systems, or operate the plurality of computing systems in a
US 10,608,433 B1
37 38
lower power mode , that is within the sufficient, but less than systems by reducing their operating frequency or forcing
full , range of power that is available . In addition , the them into a lower power mode through a network directive .
computing systems 512 may adjust operational frequency, Similarly, the flexible datacenter 500 may ramp up power
such as performing more or less processes during a given consumption based on various conditions. For instance , the
duration . The computing systems512 may also adjust inter- 5 datacenter control system 504 may determine, or the gen
nal clocks via over-clocking or under -clocking when per eration control system 414 , the remote master control sys
tem 300, or the grid operator 702 may communicate, that a
forming operations.
While the flexible datacenter 500 is online and opera generation change in local conditions may result in greater power
tional , a datacenter ramp-down condition may be met when , availability , or economic feasibility . In such
there is insufficient or anticipated to be insufficient, behind- 10 tosituations
increase, thepower
datacenter control system
consumption by the 504 may datacenter
flexible take steps
the -meter power availability or there is an operational direc 500 .
tive from the generation station control system 414 , the Alternatively , the generation station control system 414 ,
remote master control system 300, or the grid operator 702 . the remote master
The datacenter control system 504 may monitor and deter 15 702, may issue an control system 300 , or the grid operator
operational directive to increase power
mine when there is insufficient, or anticipated to be insuf consumption for any reason , the cause of which may be
ficient, behind -the -meter power availability . As noted above, unknown . In response , the datacenter control system 504
sufficiency may be specified by the remote master control may dynamically increase power delivery to one or more
system 300 or the datacenter control system 504 may be computing systems 512 (or operations at the computing
programmed with a predetermined preference or criteria on 20 systems 512) to meet the dictate . For instance, one or more
which to make the determination independently. computing systems 512 may transition into a higher power
An operational directive may be based on current dis mode, which may involve increasing power consumption
patch - ability , forward looking forecasts for when behind and/or operation frequency .
the-meter power is, or is expected to be, available, economic One of ordinary skill in the art will recognize that data
considerations, reliability considerations, operational con- 25 center control system 504 may be configured to have a
siderations, or the discretion of the generation station control number ofdifferent configurations, such as a number or type
system 414 , the remote master control system 300 , or the or kind of the computing systems512 that may be powered ,
grid operator 702. For example, the generation station and in what operating mode, that correspond to a number of
control system 414 , the remote master control system 300 , different ranges of sufficient and available behind -the -meter
or the grid operator 702 may issue an operational directive 30 power. As such , the datacenter control system 504 may
to flexible datacenter 500 to go offline and power down . modulate power delivery over a variety of ranges of suffi
When the datacenter ramp-down condition is met, the data cient and available unutilized behind -the -meter power avail
center control system 504 may disable power delivery the ability .
plurality of computing systems (e.g., 512 ). The datacenter FIG . 8 shows a control distribution system 800 of a fleet
control system 504 may disable 714 the power input system 35 of flexible datacenters according to one or more example
502 from providing power (e.g., three -phase nominal AC embodiments. The control distribution system 800 of the
voltage ) to the power distribution system 506 to power down flexible datacenter 500 shown and described with respect to
the computing systems 512 while the datacenter control FIG . 7 may be extended to a fleet of flexible datacenters as
system 504 remains powered and is capable of returning illustrated in FIG . 8. For example , a first generation station
service to operating mode at the flexible datacenter 500 40 (not independently illustrated ), such as a wind farm , may
when behind -the -meter power becomes available again . include a first plurality of flexible datacenters 802 , which
While the flexible datacenter 500 is online and opera may be collocated or distributed across the generation
tional, changed conditions or an operational directive may station . A second generation station (not independently
cause the datacenter control system 504 to modulate power illustrated ), such as another wind farm or a solar farm ,may
consumption by the flexible datacenter 500. The datacenter 45 include a second plurality of flexible datacenters 804 , which
control system 504 may determine, or the generation station may be collocated or distributed across the generation
control system 414 , the remote master control system 300 , station . One ofordinary skill in the art will recognize that the
or the grid operator 702 may communicate, that a change in number of flexible datacenters deployed at a given station
local conditions may result in less power generation , avail and the number of stations within the fleet may vary based
ability , or economic feasibility, than would be necessary to 50 on an application or design in accordance with one or more
fully power the flexible datacenter 500. In such situations , example embodiments .
the datacenter control system 504 may take steps to reduce The remote master control system 300 may provide
or stop power consumption by the flexible datacenter 500 directive to datacenter control systems of the fleet of flexible
(other than that required to maintain operation of datacenter datacenters in a similar manner to that shown and described
control system 504 ). 55 with respect to FIG . 7, with the added flexibility to make
Alternatively , the generation station control system 414 , high level decisions with respect to fleet that may be
the remote master control system 300, or the grid operator counterintuitive to a given station . The remote master con
702, may issue an operational directive to reduce power trol system 300 may make decisions regarding the issuance
consumption for any reason , the cause of which may be of operational directives to a given generation station based
unknown . In response, the datacenter control system 504 60 on , for example , the status of each generation station where
may dynamically reduce or withdraw power delivery to one flexible datacenters are deployed , the workload distributed
or more computing systems 512 to meet the dictate . The across fleet, and the expected computational demand
datacenter control system 504 may controllably provide required for one or both of the expected workload and
three -phase nominal AC voltage to a smaller subset of predicted power availability . In addition , the remote master
computing systems ( e.g., 512 ) to reduce power consump- 65 control system 300 may shift workloads from the first
tion. The datacenter control system 504 may dynamically plurality of flexible datacenters 802 to the second plurality
reduce the power consumption of one or more computing of flexible datacenters 804 for any reason , including , for
US 10,608,433 B1
39 40
example , a loss ofBTM power availability at one generation able), type of computational operations available, estimated
station and the availability of BTM power at another gen cost to perform the computational operations at the flexible
eration station . As such , the remote master control system datacenter 500 , cost for power, cost for power relative to cost
300 may communicate with the generation station control for grid power, and instructions from other components
systems 806A, 806B to obtain information that can be used 5 within the system , among others. The datacenter control
to organize and distribute computational operations to the system 504 may analyze one or more of the factors when
fleets of flexible datacenters 802 , 804 . determining whether to obtain a new set of computational
FIG . 9 shows a queue distribution arrangement for a operations for the computing systems 512 to perform . In
traditional datacenter 902 and a flexible datacenter 500, such a configuration , the datacenter control system 504
according to one or more example embodiments. The 10 manages the activity of the flexible datacenter 500 , includ
arrangement of FIG . 9 includes a flexible datacenter 500, a ing determining
traditional datacenter 902, a queue system 312 , a set of operations when when to acquire new sets of computational
capacity among the computing systems512
communication links 916 , 918, 920A , 920B , and the remote permit.
master control system 300. The arrangement of FIG . 9
represents an example configuration scheme that can be used 15 In other examples, a component ( e.g., the remote master
to distribute computing operations using a queue system 312 control system 300 ) within the system may assign or dis
between the traditional datacenter 902 and one or more tribute one or more sets of computational operations orga
flexible datacenters . In other examples , the arrangement of nized by the queue system 312 to the flexible datacenter 500 .
FIG . 9 may include more or fewer components in other For example , the remote master control system 300 may
potential configurations. For instance , the arrangement of 20 manage the queue system 312 , including the distribution of
FIG . 9 may not include the queue system 312 ormay include computational operations organized by the queue system
routes that bypass the queue system 312 . 312 to the flexible datacenter 500 and the traditional data
The arrangement of FIG . 9 may enable computational center 902. The remote master control system 300 may
operations requested to be performed by entities (e.g., com utilize to information described with respect to the Figures
panies). As such , the arrangement of FIG . 9 may use the 25 above to determine when to assign computational operations
queue system 312 to organize incoming computational to the flexible datacenter 500 .
operations requests to enable efficient distribution to the The traditional datacenter 902 may include a power input
flexible datacenter 500 and the critical traditional datacenter system 930 , a power distribution system 932, a datacenter
902. Particularly , the arrangement of FIG . 9 may use the control system 936 , and a setof computing systems 934. The
queue system 312 to organize sets of computational opera- 30 power input system 930 may be configured to receive power
tions thereby increasing the speed of distribution and per from a power grid and distribute the power to the computing
formance of the different computational operations among systems 934 via the power distribution system 932. The
datacenters. As a result , the use of the queue system 312 may datacenter control system 936 may monitor activity of the
reduce time to complete operations and reduce costs. computing systems 934 and obtain computational operations
In some examples , one or more components , such as the 35 to perform from the queue system 312. The datacenter
datacenter control system 504 , the remote master control control system 936 may analyze various factors prior to
system 300, the queue system 312 , or the control system requesting or accessing a set of computational operations or
936 , may be configured to identify situations that may arise an indication of the computational operations for the com
where using the flexible datacenter 500 can reduce costs or puting systems 934 to perform . A component ( e.g., the
increase productivity of the system , as compared to using the 40 remote master control system 300 ) within the arrangement
traditional datacenter 902 for computational operations. For of FIG . 9 may assign or distribute one or more sets of
example , a component within the arrangement of FIG . 9 computational operations organized by the queue system
may identify when using behind -the -meter power to power 312 to the traditional datacenter 902 .
the computing systems 512 within the flexible datacenter The communication link 916 represents one or more links
500 is at a lower cost compared to using the computing 45 that may serve to connect the flexible datacenter 500 , the
systems 934 within the traditional datacenter 902 that are traditional datacenter 902, and other components within the
powered by grid power. Additionally , a component in the system (e.g., the remote master control system 300, the
arrangement of FIG . 9 may be configured to determine queue system 312 — connections not shown). In particular,
situations when offloading computational operations from the communication link 916 may enable direct or indirect
the traditional datacenter 902 indirectly ( i.e., via the queue 50 communication between the flexible datacenter 500 and the
system 312 ) or directly (i.e. , bypassing the queue system traditional datacenter 902. The type of communication link
312 ) to the flexible datacenter 500 can increase the perfor 916 may depend on the locations of the flexible datacenter
mance allotted to the computational operations requested by 500 and the traditional datacenter 902. Within embodiments,
an entity (e.g. , reduce the time required to complete time different types of communication links can be used , includ
sensitive computational operations). 55 ing but not limited to WAN connectivity , cloud -based con
In some examples, the datacenter control system 504 may nectivity , and wired and wireless communication links .
monitor activity of the computing systems 512 within the The queue system 312 represents an abstract data type
flexible datacenter 500 and use the respective activity levels capable of organizing computational operation requests
to determine when to obtain computational operations from received from entities. As each request for computational
the queue system 312. For instance, the datacenter control 60 operations are received , the queue system 312 may organize
system 504 may analyze various factors prior to requesting the request in some manner for subsequent distribution to a
or accessing a set of computational operations or an indi datacenter. Different types of queues can make up the queue
cation of the computational operations for the computing system 312 within embodiments . The queue system 312 may
systems 512 to perform . The various factors may include be a centralized queue that organizes all requests for com
power availability at the flexible datacenter 500 ( e.g. , either 65 putational operations . As a centralized queue, all incoming
stored or from a BTM source ), availability of the computing requests for computational operations may be organized by
systems 512 (e.g., percentage of computing systems avail the centralized queue.
US 10,608,433 B1
41 42
In other examples, the queue system 312 may be distrib offloaded computational operations until a flexible datacen
uted consisting ofmultiple queue sub -systems. In the dis ter (e.g., the flexible datacenter 500 ) is available to perform
tributed configuration , the queue system 312 may use mul them . The flexible datacenter 500 consumes behind - the
tiple queue sub-systems to organize different sets of meter power without transmission or distribution costs,
computational operations. Each queue sub - system may be 5 which lowers the costs associated with performing compu
used to organize computational operations based on various tational operations originally assigned to the traditional
factors, such as according to deadlines for completing each datacenter 902. The remote master control system 300 may
set of computational operations, locations of enterprises further communicate with the flexible datacenter 500 via
submitting the computational operations, economic value communication link 922 and the traditional datacenter 902
associated with the completion of computational operations, 10 via the communication link 924 .
and quantity of computing resources required for performing FIG . 10A shows method 1000 of dynamic power con
each set of computational operations. For instance , a first sumption at a flexible datacenter using behind -the -meter
queue sub -system may organize sets of non - intensive com power according to one or more example embodiments .
putational operations and a second queue sub -system may Other example methods may be used to manipulate the
organize sets of intensive computational operations. In some 15 power delivery to one or more flexible datacenters .
examples, the queue system 312 may include queue sub In step 1010 , the datacenter control system , the remote
systems located at each datacenter. This way , each datacen master control system , or another computing system may
ter (e.g., via a datacenter control system ) may organize monitor behind - the -meter power availability . In some
computational operations obtained at the datacenter until embodiments, monitoring may include receiving informa
computing systems are able to start executing the compu- 20 tion or an operational directive from the generation station
tational operations. In some examples, the queue system 312 control system or the grid operator corresponding to behind
may move computational operations between different com the -meter power availability .
puting systems or different datacenters in real-time. In step 1020, the datacenter control system or the remote
Within the arrangement of FIG . 9 , the queue system 312 master control system 300 may determine when a datacenter
is shown connected to the remote master control system 300 25 ramp -up condition is met . In some embodiments, the data
via the communication link 918. In addition, the queue center ramp -up condition may be met when there is suffi
system 312 is also shown connected to the flexible datacen cient behind -the -meter power availability and there is no
ter via the communication 920A and to the traditional operational directive from the generation station to go offline
datacenter 902 via the communication link 920B . The com or reduce power.
munication links 918 , 920A , 920B may be similar to the 30 In step 1030 , the datacenter control system may enable
communication link 916 and can be various types of com behind -the -meter power delivery to one or more computing
munication links within examples . systems. In some instances, the remote mater control system
The queue system 312 may include a computing system may directly enable BTM power delivery to computing
configured to organize and maintain queues within the queue systems within the flexible system without instructing the
system 312. In another example , one or more other compo- 35 datacenter control system .
nents of the system may maintain and support queues within In step 1040, once ramped - up , the datacenter control
the queue system 312. For instance, the remote master system or the remote master control system may direct one
control system 300 may maintain and support the queue or more computing systems to perform predetermined com
system 312. In other examples, multiple components may putational operations. In some embodiments , the predeter
maintain and support the queue system 312 in a distributed 40 mined computational operations may include the execution
manner , such as a blockchain configuration . of one or more distributed computing processes, parallel
In some embodiments , the remote master control system processes, and /or hashing functions, among other types of
300 may serve as an intermediary that facilitates all com processes .
munication between flexible datacenter 500 and the tradi While operational, the datacenter control system , the
tional datacenter 902.Particularly , the traditional datacenter 45 remote master control system , or another computing system
902 or the flexible datacenter 500 might need to transmit may receive an operational directive to modulate power
communications to the remote master control system 300 in consumption. In some embodiments, the operational direc
order to communicate with the other datacenter . As also tive may be a directive to reduce power consumption . In
shown, the remote master control system 300 may connect such embodiments , the datacenter control system or the
to the queue system 312 via the communication link 918. 50 remote master control system may dynamically reduce
Computational operations may be distributed between the power delivery to one or more computing systems or
queue system 312 and the remotemaster control system 300 dynamically reduce power consumption of one or more
via the communication link 918. The computational opera computing systems. In other embodiments, the operational
tions may be transferred in real-time and mid -performance directive may be a directive to provide a power factor
from one datacenter to another (e.g., from the traditional 55 correction factor. In such embodiments, the datacenter con
datacenter 902 to the flexible datacenter 500 ). In addition , trol system or the remote master control system may
the remote master control system 300 may manage the dynamically adjust power delivery to one or more comput
queue system 312 , including providing resources to support ing systems to achieve a desired power factor correction
queues within the queue system 312 . factor. In still other embodiments , the operational directive
As a result, the remote master control system 300 may 60 may be a directive to go offline or power down. In such
offload some or all of the computational operations assigned embodiments, the datacenter control system may disable
to the traditional datacenter 902 to the flexible datacenter power delivery to one or more computing systems.
500. This way , the flexible datacenter 500 can reduce overall FIG . 10B showsmethod 1050 of dynamic power delivery
computational costs by using the behind -the -meter power to to a flexible datacenter using behind -the-meter power
provide computational resources to assist traditional data- 65 according to one or more embodiments . In step 1060 , the
center 902. The remote master control system 300 may use datacenter control system or the remote master control
the queue system 312 to temporarily store and organize the system may monitor behind - the-meter power availability. In
US 10,608,433 B1
43 44
certain embodiments , monitoring may include receiving at a reduced price and/or monetary payment if the option is
information or an operational directive from the generation exercised by the power entity ).
station control system or the grid operator corresponding to The power option agreementmay be used by the power
behind - the -meter power availability . entity 1140 to reserve the right to reduce the amount of grid
In step 1070 , the datacenter control system or the remote 5 power delivered to the load during a set time frame ( e.g., the
master control system may determine when a datacenter next 24 hours). For instance , the power entity 1140 may
ramp -down condition is met. In certain embodiments , the exercise a predefined power option to reduce the amount of
datacenter ramp -down condition may be met when there is power grid power delivered to the load during a timewhen the grid
insufficient behind -the-meter power availability or antici powermay be better redirected to other loads coupled to the
pated to be insufficient behind - the -meter power availability 10 power grid . As such , the power entity 1140 may exercise
option agreements to balance loads coupled to the
or there is an operational directive from the generation power grid . In some embodiments, a power option agree
station to go offline or reduce power. ment may also specify other parameters, such as costs
In step 1080 , the datacenter control system may disable associated with different levels of power consumption and/
behind -the-meter power delivery to one or more computing 15 or maximum power thresholds for the load to operate
systems. In step 1090, once ramped - down , the datacenter according to .
control system remains powered and in communication with To illustrate an example , a power option agreementmay
the remotemaster control system so that it may dynamically specify that a load ( e.g., the datacenters 1102-1106 ) is
power the flexible datacenter when conditions change . required to use at least 10 MW or more at all times during
One of ordinary skill in the art will recognize that a 20 the next 12 hours. Thus, the minimum power threshold
datacenter control system may dynamically modulate power according to the power option agreement is 10 MW and this
delivery to one or more computing systems of a flexible minimum power threshold extends across the time interval
datacenter based on behind -the-meter power availability or of the next 12 hours. In order to comply with the agreement,
an operational directive. The flexible datacenter may tran the load must subsequently operate using 10 MW or more
sition between a fully powered down state (while the data- 25 power at all times during the next 12 hours. This way, the
center control system remains powered ), a fully powered up load can accommodate a situation where the power entity
state , and various intermediate states in between . In addi 1140 exercises the option . Particularly , exercising the option
tion , flexible datacenter may have a blackout state, where all may trigger the load to reduce the amount of power it
power consumption , including that of the datacenter control consumes by an amount up to 10 MW at any point during the
system is halted . However, once the flexible datacenter 30 12 hour interval. By establishing this power option agree
enters the blackout state , it will have to be manually ment, the power entity 1140 can manipulate the amount of
rebooted to restore power to datacenter control system . power consumed at the load during the next 12 hours by up
Generation station conditions or operational directives may to 10 MW if power needs to be redirected another load or
cause flexible datacenter to ramp- up , reduce power con a reduction in power consumption is needed for other
sumption , change power factor, or ramp -down. 35 reasons .
FIG . 11 illustrates a block diagram of a system for In the example arrangement of the system 1100 shown in
implementing control strategies based on a power option FIG . 11, one or more of the datacenters ( e.g., the flexible
agreement, according to one or more embodiments. The datacenters 1102 , 1104 , and the traditional datacenter 1106 )
system 1100 represents an example arrangement that may operate as the load that is subject to a power option
includes a control system (e.g., the remote master control 40 agreement. As the load that is subject to the power option
system 262), a load ( e.g., one or more of the datacenters agreement, the datacenters 1102-1106 may execute control
1102, 1104 , and 1106 ), and a power entity 1140 , which may instructions in accordance with power target consumption
establish and operate in accordance with a power option targets that meet or exceed the minimum power thresholds
agreement. Additional arrangements are possible within based on the power option agreement.
examples . 45 As shown in FIG . 11 , each datacenter 1102-1106 may
In general, a power option agreement is an agreement include a set of computing systems configured to perform
between a power entity 1140 associated with the delivery of computational operations using power from one or more
power to a load (e.g., a grid operator, power generation power sources (e.g., BTM power, grid power, and/or grid
station , or local control station ) and the load (e.g., the power subject to a power option agreement). In particular,
datacenters 1102-1106 ). As part of the power option agree- 50 the flexible datacenter 1102 includes computing systems
ment, the load (e.g., load operator, contracting agent for the 1108 arranged into a first set 1114A , a second set 1114B , and
load , semi-automated control system associated with the a third set 1114C , the flexible datacenter 1104 includes
load , and /or automated control system associated with the computing systems 1110 arranged into a first set 1116A , a
load ) provides the power entity 1140 with the right, but not second set 1116B , and a third set 1118B , and the traditional
obligation , to reduce the amount of power delivered (e.g. , 55 datacenter 1106 includes computing systems 1112 arranged
grid power) to the load up to an agreed amount of power into a first set 1118A , a second set 1118B , and a third set
during an agreed upon time interval. In order to provide the 1118C . Each set of computing systemsmay include various
power entity 1140 with this option , the load needs to be types of computing systems that can operate in one or more
using at least the amount of power subject to the option (e.g., modes .
a minimum power threshold ). For instance , the load may 60 The different sets of computing systems as well as the
agree to use at least 1 MW of grid power at all times during multiple datacenters are included in FIG . 11 for illustration
a specified 24-hour time interval to provide the power entity purposes. In particular , the variety of computing systems
1140 with the option ofbeing able to reduce the amount of represent different configurations that a load may take while
power delivered to the load by any amount up to 1 MW at operating in accordance with a power option agreement, and
any point during the specified 24 -hour time interval . The 65 each configuration (as detailed herein ) may include ramping
load may grant the power entity 1140 with this option in up or down power consumption and transferring and per
exchange for a monetary consideration (e.g. , receive power forming computational operations between sets of comput
US 10,608,433 B1
45 46
ing systems and /or datacenters. In other examples, the load the datacenters 1102-1106 . In particular, the power option
that is subject to a power option agreement may take on data may specify the minimum power threshold or thresh
other configurations (e.g., a single datacenter 1102-1106 , olds associated with one or more time intervals for the load
and /or a single set of computing systems). to operate at in accordance with based on the power option
The remote master control system 262 may serve as a 5 agreement. The power option data may also specify other
control system that can determine performance strategies constraints that the load should operate in accordance with .
and provide control instructions to the load ( e.g., one or In some examples, the power option data may also include
more of the datacenters 1102-1106 ). In particular, the remote an indication of a monetary penalty that would be imposed
master control system 262 can monitor conditions in concert upon the load for failure to operate as agreed upon for the
with the minimum power thresholds and time intervals (e.g., 10 power option agreement. In addition , the power option data
power option data ) set forth in , and/or derived from , one or may also include an indication of a monetary benefit pro
more power option agreements to determine performance vided to the load operating at power consumption levels that
strategies that can enable the load to meet the expectations are in accordance with a power option agreement. For
of the power option agreement(s ) while also efficiently using instance , monetary benefits could include reduced prices for
power to accomplish computational operations. In some 15 power, credits for power, and/or monetary payments . In
instances, the remote master control system 262 may also be addition , the power option data may include further con
subject to the power option agreement and may adjust its straints upon power use, such as one or more maximum
own power consumption based on the power option agree power thresholds and corresponding time intervals for the
ment ( e.g., ramp up or down power consumption based on maximum power thresholds.
the defined minimum power thresholds during time inter- 20 In some embodiments , the power entity 1140 may corre
vals ). spond to a qualified scheduling entity (QSE). A QSE may
To establish a power option agreement, the remote master submit bids and offers on behalf of resource entities (RES)
control system 262 (or another computing system ) may or load serving entities (LSEs), such as retail electric pro
communicate with the power entity 1140. For instance , the viders (REPs). QSEsmay submit offers to sell and/or bids to
remote master control system 262 may provide a request 25 buy power (energy ) in the Day -Ahead Market (e.g. , the next
(e.g., a signal and /or a bid ) to the power entity 1140 and 24 hours ) and the Real- Time Market. As such , the remote
receive the terms of one or more power option agreements , master control system 262 or another computing system may
or power option data related to power option agreements communicate with one or more QSEs to engage and control
(e.g. , data such as minimum power thresholds and time one or more loads in accordance with one or more power
intervals , but not all terms contained within a potential 30 option agreements.
power option agreement) in response. In some examples, the In some examples, a power option agreement may take
remote master control system 262 may evaluate one or more the form of a fixed duration power option agreement 1144.
conditions prior to establishing a power option agreement to The fixed duration power option agreement 1144 may
ensure that the conditions could enable the load ( e.g., the specify a set ofminimum power thresholds and a set of time
datacenters 1102-1106 ) to operate in accordance with the 35 intervals in advance for an upcoming fixed duration of time
power option agreement. For instance , the remote master covered by the agreement. Each minimum power threshold
control system 262 may check the quantity and deadlines in the set of minimum power thresholds may be associated
associated with computational operations assigned to spe with a time interval in the set of time intervals . Examples of
cific datacenters prior to establishing specific datacenters as such association are provided in FIG . 12. The fixed duration
a load subject to a power option agreement. In some cases, 40 power option agreement may be established in advanced of
multiple power option agreements may be established . For the time period covered by the setof time intervals to enable
example , each datacenter 1102-1106 may be subject to a the remote master control system 262 to prepare perfor
different power option agreement, which may result in the mance strategies for the load (e.g., the datacenter(s)) asso
remote master control system 262 managing the power ciated with the power option agreement. Thus, the remote
consumption at each of the datacenters 1102-1106 differ- 45 master control system 262 may evaluate the fixed duration
ently . power option and other monitored conditions to determine
Within the system 1100 shown in FIG . 11 , the power performance strategies for a set of computing systems (e.g.,
entity 1140 may represent any type of power entity associ one or more datacenters ) during the different intervals that
ated with the delivery of power to the load that is subject to satisfy the minimum power thresholds .
a power option agreement. For instance, the power entity 50 In other examples, a power option agreementmay take the
1140 may be a local station control system , a grid operator, form of a dynamic power option agreement 1146. For a
or a power generation source. As such , the power entity 1140 dynamic power option agreement 1146 , minimum power
may establish power option agreements with the loads via thresholds may be provided to the remote master control
communication with the loads and /or the remote master system 262 in real- time (or near real-time ). For instance , a
controlsystem 262. For example , the power entity 1140 may 55 dynamic power option agreement may specify that the
obtain and accept a bid from a load trying to engage in a power entity 1140 may provide adjustments to minimum
power option agreement with the power entity 1140. The power thresholds and corresponding time intervals in real
power entity 1140 is shown with a power option module time to the remote master control system 262. For example ,
1142, which may be used to establish power option agree a dynamic power option agreement may provide power
ments (e.g. , fixed -duration 1144 and /or dynamic 1146 ). 60 option data that specifies a minimum power threshold for
Once a power option agreement is established, the remote immediate adjustments (e.g., for the next hour ).
master control system 262 may obtain power option data In an embodiment, a dynamic power option agreement
from the power entity 1140 ( or another source ) that specifies 1146 may involve repeat communication between the
the power and time expectations of the power entity 1140 . remotemaster control system 262 and the power entity 1140.
Asshown in FIG . 11, the power entity 1140 includes a power 65 Particularly , the power entity 1140 may provide signals to
option module 1142, which may be used to provide power the remote master control system 262 that request power
option data to the remote master control system 262 and /or consumption adjustments to be initiated at one or more
US 10,608,433 B1
47 48
datacenters by the remote master control system 262 over power, and /or a battery system ) that a datacenter has avail
short time intervals, such as across minutes or seconds . For able . Power prices 1122 may involve an analysis of the
example , the power entity 1140 may communicate to the different costs associated with powering a set of computing
remote master control system 262 to ramp power consump systems. For instance , the remote master control system 262
tion down to a particular level within the next 5 minutes. As 5 may determine cost ofpower from the grid without a power
a result , the remote master control system 262 may provide option agreement relative to the cost power from the grid
instructions to one or more datacenters to ramp down power under the power option agreement. In addition , the remote
consumption using a linear ramp over the next 5 minutes to master control system 262 may also compare the costof grid
meet the particular level specified by the power entity 1140 . power relative to the cost of BTM power when available at
The remote master control system 262 may monitor the 10 a datacenter. The power prices 1122 may also involve
linear ramp down of power consumption and increase or comparing the cost of using power at differentdatacenters to
decrease the rate that the datacenter( s ) ramp down power use determine which datacenter may perform computational
based on projections and updates received from the power operations at a lower cost.
entity 1140. As a result, although the ramp down of power Monitoring computing system parameters 1124 may
consumption may initially be performed in a linear manner 15 involve determining parameters related to the computing
to meet a power target threshold , the remote master control systems at one or more datacenters. For instance , the remote
system 262 may adjust the rate of power consumption master control system 262 may monitor various parameters
decrease based on updates from the power entity 1140. For of the computing systems at a datacenter, such as the
example, 25 percent of the overall power consumption ramp abilities and availability of various computing systems, the
down may occur during a first period (e.g., 4 minutes 30 20 status of the queue used to store computational operations
seconds) of the 5 minutes and the remaining 75 percent of awaiting performance by the computing systems. The
the overall power consumption ramp down may occur remote master control system 262 may determine types and
during the remaining period of the 5 minutes ( e.g., the final operation modes of the computing systems, including which
30 seconds). The example percentages are included for computing systems could operate in different modes (e.g., a
illustration purposes and can vary within examples based on 25 higher power or a lower power mode) and /or at different
various parameters , such as additional communication (e.g., hash rates and /or frequencies. The remote master control
adjustments) provided by the power entity 1140 . system 262 may also estimate when computing systems may
In further examples, a power option agreement may complete current computational operations and/ or how
operate similarly to both a fixed -duration 1144 and a many computational operations are assigned to computing
dynamic power option agreement 1146. Particularly , power 30 systems.
option data specifying minimum power thresholds and cor Monitoring cryptocurrency prices 1126 may involve
responding time intervals may be provided in advance for monitoring the current price of one or more cryptocurren
the entire fixed -duration of time ( e.g., the next 24 hours). cies , the hash rate and /or estimated power consumption
Additional power option data may then be subsequently associated with mining each cryptocurrency, and other fac
provided enabling the remote master control system 262 to 35 tors associated with the cryptocurrencies. The remote master
make one or more adjustments to accommodate any changes control system 262 may use data related to monitoring
specified within the additional power option data . For cryptocurrency prices 1126 to determine whether using
instance , additional power option data may indicate that a computing systemsto mine a cryptocurrency generates more
power entity exercised its option to deliver less power to the revenue than the cost of power required for performance of
load . As a result , the remote master control system may 40 the mining operations.
instruct the load to adjust power consumption based on the The remote master control system 262 may monitor
power entity reducing the power threshold minimum via parameters related to computational operations (e.g., com
exercising the option . putational operation parameters 1128 ). For example, the
As indicated above , the remote master control system 262 remote master control system 262 may monitor parameters
may monitor conditions in addition to the constraints set 45 related to the computational operations requiring perfor
forth in power option data received from the power entity mance and currently being performed , such quantity of
1140. Particularly , the remote master control system 262 operations, estimated time to complete, cost to perform each
may monitor and analyze a set of conditions ( including the computational operation , deadlines and priorities associated
power option data ) to determine strategies for assigning, with each computational operation . In addition , the remote
transferring , and otherwise managing computational opera- 50 master control system 262 may analyze computational
tions using the one or more datacenters 1102-1106 . The operations to determine if a particular type of computing
determined strategies may enable efficient operation by the system may perform the computational operation better than
datacenters while also ensuring that the datacenters operate other types of computing systems.
at target power consumption levels thatmeet or exceed the Monitoring weather conditions 1129 may include moni
minimum power thresholds set forth within one or more 55 toring for any potential power generation disruption due to
power option agreements . emergencies or other events , and changes in temperatures or
Example monitored conditions include , but are not lim weather conditions at power generators or datacenters that
ited to , power availability 1120, power prices 1122 , com could affect power generation . As such , the operations and
puting systems parameters 1124 , cryptocurrency prices environment analysis module (or another component) of the
1126 , computational operation parameters 1128, and 60 remote master control system 262 may be configured to
weather conditions 1129. Power availability 1120 may monitor one or more conditions described above.
include determining power consumption ranges at a set of The performance strategy determined by the remote mas
computing systems and/or at one or more datacenters. In ter control system 262 based on the monitored conditions
addition , power availability 1120 may also involve deter and / or power option data can include control instructions for
mining the source or sources of power available at a data- 65 the load (e.g., the datacenters and/or one or more sets of
center. For instance, the remote master control system 262 computing systems). For instance , a performance strategy
may identify the types of power sources (e.g. , BTM , grid can specify operating parameters, such as operating frequen
US 10,608,433 B1
49 50
cies, power consumption targets , operating modes , power using enough power to meet the minimum power threshold
on /off and / or standby states, and other operation aspects for
or thresholds set forth in the power option agreement.
computing systems at a datacenter . In some examples , the remote master control system 262
The performance strategy can also involve aspects related may monitor the grid frequency signal received from the
to the assignment, transfer, and performance of computa- 5 power entity 1140. When the frequency of the grid deviates
tional operations at the computing systems. For instance, the athethreshold amount (e.g., 0.036 Hz above or below 60 Hz),
remote master control system 262 may adjust perfor
performance strategy may specify computational operations
to be performed at the computing systems, an order for mance strategies at the load . In some cases, the remote
master control system 262 may adjust the power consump
completing computational operations based on priorities
associated with the computational operations, and an iden- 10 tion
systemsat )theoperating
load, atthethenumber
load , andof/ orminers (or computing
the frequency or hash
tification of which computing systems should perform rate
which computational operations . In some instances, priori control system may readjust performance strategiesmaster
, among other possible changes . The remote
ties may depend on revenue associated with completing each load in response to receiving additional power optionatdata the
computational operation and deadlines for each computa- 15 from the power entity 1140 (e.g., an indication that the
tional operation .
The monitored conditions may enable efficient distribu
frequency of the grid is back to 60 Hz ). In addition , the
remote master control system 262 may communicate
tion and performance of computational operations among changes in operations at the load to the power entity 1140.
computing systems at one or more datacenters ( e.g., data This way, the power entity 1140 may obtain confirmation
centers 1102-1106 ) in ways that can reduce costs and/or time 20 that the load is adjusting in accordance with a power option
to perform computational operations, take advantage of agreement.
availability and abilities of computing systems at the data In some embodiments, a power generation source ( e.g. ,
centers 1102-1106 , and /or take advantage in changes in the the generation station 400 shown in FIG . 4 ) may enter into
cost for power at the datacenters 1102-1106 . In addition , the a power option agreement with a grid operator, which may
monitored conditions may also involve consideration of the 25 provide the grid operator with the option to reduce the
power option data to ensure that the computing systems amount of power that the power source generator can deliver
consume enough power to meet minimum power thresholds to the grid during a defined time interval. For instance , a
set forth in one or more power option agreements. wind generation farm may enter into the power option
The variousmonitored conditions described above as well agreement with the grid operator. In addition , the remote
as other potential conditions may change dynamically and 30 master control system 262 may also enter into a power
with great frequency . Thus, to enable efficient distribution option agreement with the power generation source (e.g., the
and performance of the computational operations at the wind farm ) to provide a load that can receive excess power
datacenters, the re te master control system 262 may be from the power generation source when the grid operator
configured to monitor changes in the various conditions to exercises the option and lowers the amount of power that the
assist with the efficient management and operations of the 35 power generation source can deliver to the grid . Thus, rather
computing systems at each datacenter. For instance, the than reducing the amount of power produced , the power
remote master control system 262 may engage in wired or generation source could exercise an option in the agreement
wireless communication 1130 with datacenter control sys with remote master control system 262 and redirect excess
tems ( e.g., datacenter control system 504) at each datacenter power to one or more loads (e.g., a set of computing
as well as other sources ( e.g., the power entity 1140) to 40 systems) that could ramp up power consumption in
monitor for changes in the conditions . response . In such situations, the remote master control
The remote master control system 262 may analyze the system 262 maybe able to use the excess power from the
different conditions in real-time to modulate operating attri power generation source (e.g., BTM power) to perform
butes of computing systems at one ormore of the datacen operations at one or more loads at a low cost ( or no cost at
ters . By using the monitored conditions, the remote master 45 all). In addition , the power generation source may benefit
control system 262 may increase revenue , decrease costs, from the power option agreement by directing excess power
and /or increase performance of computational operations to the load instead of temporarily halting power production .
via various modifications, such as transferring computa In some examples, a power option agreementmay depend
tional operations between datacenters or sets of computing on parameters associated balancing grid capacity and
systems within a datacenter and adjusting performance at 50 demand. For instance , power option agreements may incen
one or more sets of computing systems (e.g., switching to a tivize power consumption ramping during periods of peak
low power mode ). grid power use.
In some examples, the traditional datacenter 1106 may be FIG . 12 shows a graph representing power option data
the load subject to a power option agreement. As such , the based on a power option agreement, according to one or
remote master control system 262 may factor the power 55 more embodiments. The graph 1200 shows power option
option agreement when determining whether to perform data arranged according to power 1204 over time 1202. As
computational operations using the computing systems 1112 shown in FIG . 12 , time 1202 increases along the X - axis and
at the traditional datacenter 1106 and/ or transfer computa minimum power thresholds 1204 increase along the Y -axis
tional operations to the computing systems 1108 , 1110 at the of the graph 1200. In the example embodiment shown in
flexible datacenters 1102 , 1104. For instance, the monitored 60 FIG . 12 , the time 1202 increases up to a full day ( e.g., 24
conditions may indicate that the price of grid power is hours ) in 4 hour increments and the power is shown in MW
substantially higher than BTM power. As a result, the remote increasing in intervals of 5 MW . The 24 duration and
master control system 262may transfer a subset of compu example minimum power thresholds can differ in other
tational operations from the traditional datacenter 1106 to embodiments. Particularly , these values may depend on the
the flexible datacenters 1102, 1104. The traditional datacen- 65 terms set forth within the power option agreement.
ter 1106 may still have some computational operations to The graph line 1206 represents sets of minimum power
perform to ensure that the traditional datacenter 1106 is thresholds 1206A , 1206B , 1206C that are specified by
US 10,608,433 B1
51 52
power option data based on the power option agreement. As The last minimum power threshold 1206C is associated
shown, the graph line 1206 extends the entire 24 hour with the time interval that starts at hour 16 and extends until
duration , which indicates that the set of time intervals hour 24. Similar to the initial minimum power threshold
associated with minimum power thresholds add up to 24 1206A associated with the beginning of the graph line 1206 ,
hours . In other examples, the power option agreement may 5 the last minimum power threshold 1206 is also set at 5 MW .
not include a minimum power threshold during a portion of As such , at any point during this interval (hour 16 to hour
the duration . 24 ) the loads may consume 5 MW or more to operate in
The graph line 1206 of the graph 1200 is further used to accordance with the power option agreement. As discussed
illustrate power consumption levels that one or more loads above, by operating at 5 MW or more, the load enables the
(e.g., a setof computing systems) operating according to the 10 power consumed from the power grid to be reduced any
power option agreement may utilize during the 24 hour amount from zero up to 5 MW during this time interval.
duration. Particularly , the power quantities above the graph When determining the power consumption strategy for a
line 1206 represents power levels that the load (s) may load , a computing system (e.g., the remote master control
consume from the power grid during the 24 hour duration system 262) may consider various conditions in addition to
that would satisfy the requirements (i.e. , the minimum 15 the power option data received based on one or more power
power thresholds 1206A - 1206C ) set forth by the power option agreements. Particularly , the computing system may
option agreement. In particular, the power quantities above consider and weigh different conditions in addition to the
the graph line 1206 include any power quantity that meets or power option data to determine power consumption targets
exceeds the minimum power threshold at that time. By and / or other control instructions for a load . The conditions
extension , the power quantities positioned below the graph 20 may include, but are not limited to , the price of grid power,
line 1206 represents the amount of power that the load could the price of alternative power sources (e.g., BTM power,
be directed to reduce power consumption by per the power stored energy ), the revenue associated with mining for one
option agreement. or more cryptocurrencies, parameters related to the compu
To further illustrate , an initial minimum power threshold tational operations requiring performance (e.g. , priorities,
1206A is shown associated with the time interval starting at 25 deadlines, status of the queue organizing the operations,
hour 0 and extending to hour 8. In particular, the minimum and /or revenue associated with completing each computa
power threshold 1206A is set at 5 MW during this time tional operation ), parameters related to the set of computing
interval. Thus,based on the power option data shown in FIG . systems (e.g., types and availabilities of computing sys
12 , the loads must be able to operate at a target power tems), and other conditions (e.g. , penalties if a minimum
consumption level that is equal to or greater than the 5 MW 30 power threshold is not met and /or monetary benefits from
minimum power threshold 1206A at all times during the operating under a power option agreement). By weighing
time interval extending from hour 0 to hour 8, in order to be various conditions, the computing system may efficiently
able to satisfy the power option if it is exercised for that time manage the set of computing systems, including enabling
interval. Similarly, the power entity could reduce the power performance of computational operations cost effectively
consumed by loads by any amount up to 5 MW at any point 35 and/or ensuring at that computing systems operate at target
during the time interval from hour 0 to hour 8 in accordance power consumption levels that one or more satisfy power
with the power option agreement. For instance , the power option agreements .
entity could exercise its option at any point during this time In some examples, the computing system may decrease
interval to reduce the power consumed by the loads by 3 the amount of power that a set of computing systems
MW as a way to load balance the power grid . In response to 40 consumes from one source and while also increasing the
the power entity exercising its option , the load may then amount of power that the set consumes from another source .
operate using 3 MW less power and /or another strategy For instance , the computing system may determine that the
determined by a control system factoring additional condi price of power grid power is above a threshold price that
tions (e.g., the price of grid power , the revenue that could be makes computationaloperations relatively expensive to per
generated from mining a cryptocurrency , and /or parameters 45 form using grid power . As a result , the computing system
associated with computational operations awaiting perfor may provide control instructions for the computing systems
mance ) to consume power grid power that matches a minimum
As further shown in the graph 1200 illustrated in FIG . 12 , power threshold specified by power option data. This may
the next minimum power threshold 1206B is associated with enable the computing systems to satisfy the power option
the following time interval, which starts at hour 8 and 50 agreement while also avoiding using pricey grid power
extends until hour 16. During this time interval (hour 8 to beyond the minimum amount required per the power option
hour 16 ), the load (s ) may consume 10 MW or more power data . In addition , the computing system may instruct some
since the minimum power threshold 1206B is now set at 10 computing systems to switch to a low power mode or
MW as shown on the Y -axis of the graph 1200. In light of temporarily stop until the price of power from the grid
the power option data , a control system may determine and 55 decreases. The computing system may instruct one or more
provide a performance strategy to the load (e.g., a set of computing systems to operate using power from another
computing systems) that includes a power consumption source (e.g., BTM power and /or stored energy from a battery
target that meets or exceeds the minimum power threshold system ) and / or transfer one or more computational opera
1206B ( i.e., 10 MW ). The performance strategy may depend tions to another set of computing systems ( e.g., a different
on the power option data as well as other possible condi- 60 datacenter ).
tions, such as the price of grid power, the availability of When the power option agreement is a fixed duration
computing systems, and /or the type of computing opera power option agreement, the computing system may receive
tions, etc. In addition , the power entity could exercise its an indication of all the minimum power thresholds 1206A
option to reduce the amount of power consumed by the load 1206C and an indication of the associated time interval
by 10 MW or less as represented by the power levels under 65 altogether and in advance of the duration associated with the
the minimum threshold 1206B that extend during the time power option agreement. By providing all of the minimum
interval of hour 8 to hour 16 . power thresholds 1206A - 1206C and the time intervals in
US 10,608,433 B1
53 54
advance , the computing system may determine a perfor duration power option agreement. In other instances, the
mance strategy for the load that can extend across the entire computing system may receive power option data dynami
duration . Particularly , the computing system may factor the cally and adjust operations in real-time (or near real-time).
minimum power thresholds and associated time intervals as For instance, the computing system may receive a series of
well as other monitored conditions to determine the perfor- 5 power option data that each specifies minimum power
mance strategy for the total duration. This can enable the threshold changes during the duration set forth in the
computing system to accept and assign computational opera dynamic power option agreement. To illustrate an example ,
tions to computing systems in advance while also using a the computing system may receive power option data during
performance strategy thatmeets the expectations of a power hour 1 that specifies the minimum power threshold for hour
option agreement. 10 2 , power option data during hour 2 that specifies the mini
In some examples, the performance strategy determined mum power threshold for hour 3 , and so on across the
by the computing system may include control instructions duration of the dynamic power option agreement.
for the set ofcomputing systems to execute if a power option In some examples , the minimum power threshold for a
is exercised . For instance, the performance strategy may time interval may be zero during the duration of a power
specify different power consumption targets for the comput- 15 option agreement. As such , the load may use any amount of
ing systems that depend on whether a power option is power from the power grid in accordance with the power
exercised during each time interval. option agreement, including no power at all during this time
In some instances, the computing system may modify the interval. When the price for power is high during this time
performance strategy when one or more conditions change frame, the load may ramp down power usage to zero MW to
enough to warrant a modification . For instance, the com- 20 avoid paying the high price for power while still being in
puting system may receive an indication of a change in a compliance with the power option agreement.
minimum power threshold ( e.g., a decrease in the minimum FIG . 13 illustrates a method for implementing control
power threshold ) and determine one or more modifications strategies based on a fixed -duration power option agree
based on the new minimum power threshold and/or other ment, according to one or more embodiments . The method
conditions ( e.g., a change in the price of power). 25 1300 serves as an example and may include other steps
In other examples, the power option agreement may be a within other embodiments . A control system (e.g., the
dynamic power option agreement. Particularly, the loadmay remote master control system 262) may be configured to
be subject to a changing minimum power threshold that can perform one or more steps of the method 1300. As such , the
vary during a predefined duration associated with the power control system may take various forms of a computing
option agreement. For example , a dynamic power option 30 system , such as a mobile computing device, a wearable
agreementmay specify that the load is subject to a minimum computing device, a network of computing systems, etc.
power threshold that may vary from 0 MW up to 5 MW At step 1302 , the method 1300 involves monitoring a set
during the next 24 hours and the particular minimum thresh of conditions. For instance, a computing system (e.g., a
old for each hourmay depend on power option data received control system ) may monitor various conditions that could
from the power entity during the prior hour. The dynamic 35 impact the performance of operations at one or more loads,
power option agreement may further specify the expected including the power consumption targets at the loads. The
response time from the load . For instance, the power option set of monitored conditions may include a variety of infor
agreement may indicate that an indication of a new mini mation obtained from one or more external sources, such as
mum power threshold will be provided an hour prior to the one or more datacenters, databases, power generation sta
start of the minimum power threshold . The computing 40 tions, or types of sources.
system , for example, may receive an indication at hour 7 Some example conditions include, but are not limited to ,
about the increase in the minimum power threshold 1206B the price of grid power, the price and availability of alter
starting at hour 8. The indication may (or may not) specify native power options (e.g. BTM power, and /or stored
the total time interval associated with a new minimum energy ), parameters of the load (e.g., ramping abilities , type
power threshold . For instance , the indication received by the 45 of computing systems, operation modes, etc. ), parameters of
computing system may specify that the 10 MW minimum tasks to be performed using the power at the load ( e.g.,
power threshold 1206B extends from hour 8 until hour 16 . types , deadlines , priorities, and /or revenue associated with
In other instances, the power option data may indicate that computational operations), availability of other computing
the computing system should abide by the new minimum systems and their associated costs , and /or revenue associ
power threshold until receiving further power option data 50 ated with mining a cryptocurrency. The computing system
indicating a change to another new minimum power thresh may monitor one or more of these conditions as well as
old . others.
In some examples, the power option data may arrive at the At step 1304 , the method 1300 involves receiving power
computing system in an unknown order from the power option data based , at least in part, on a power option
entity with expectations of swift power consumption adjust- 55 agreement.As discussed above , the computing system (e.g.,
ments by the load . As a result, the power option agreement a remote master control system ) may engage in a power
may require fast ramping of the load to meet changes. option agreement with a power entity . As a result , the
Ramping may involve ramping up or down power consump computing system may control a load (e.g. , a set of com
tion as well as ramping operating techniques (e.g. , adjusting puting systems) in accordance with power thresholds and
frequency or operation mode ). 60 time intervals received from the power entity based on the
In some embodiments , the type of power option power power option agreement.
agreementmay depend on the delivery and content of power In some examples , the power option data may specify a
option data provided to the load (or a control system set ofminimum power thresholds and a set of time intervals.
controlling the load ). For instance, a computing system may Each minimum power threshold in the set of minimum
receive minimum power thresholds set across an entire 65 power thresholds may be associated with a time interval in
duration associated with a power option agreement in the set of time intervals. To illustrate an example , the power
advance when the power option agreement is a fixed option data may specify a first minimum power threshold
US 10,608,433 B1
55 56
associated with a first time interval and a second minimum set of power thresholds. Responsive to receiving the subse
power threshold associated with a second time interval , with quent power option data, the performance strategy for the set
the second time interval subsequent to the first time interval. of computing systems may be modified based on a combi
The set of time intervals may add up to the duration nation of at least a portion of the subsequent power option
represented by the power option agreement. For instance, 5 data and one or more conditions of the monitored conditions .
the total duration of the set of time intervals may correspond The modified performance strategy may include one or more
to a twenty -four hour period (e.g., the next day ). In other reduced
examples, the power option agreement may span across a systems.power consumption targets for the set of computing
The amount of the reduction in a power consump
different duration ( e.g. , 12 hours). In additional embodi
ments , the power option data may specify other information , 10 tion target may depend linearly with the amount that the
corresponding minimum power threshold was reduced by .
such as monetary incentives associated with parameters of For instance, when a minimum power threshold for a time
the power option agreement and/or one or more maximum interval is reduced from 10 MW to 5 MW , the power
power thresholds. consumption target for that time interval may be reduced
At step 1306 , the method 1300 involves determining a
performance strategy for the set of computing systems based 15 setfromof 10computing
MW to 5 systems
MW. Instructions
to performmay be provided opera
computational to the
on a combination of at least a portion of the power option
data and at least one condition in the set of conditions. The tions based on the modified performance strategy .
performance strategy may be determined responsive to FIG . 14 illustrates a method for implementing control
receiving the power option data . In addition , the perfor strategies based on a dynamic power option agreement,
mance strategy may include a power consumption target for 20 according to one or more embodiments . The method 1400
the set of computing systems for each time interval in the set serves as an example and may include other steps within
of time intervals. In some examples , each power consump other embodiments . Similar to the method 1400 , a control
tion target is equal to or greater than the minimum power system (e.g., the remote master control system 262) may be
threshold associated with each time interval. configured to perform one or more steps of the method 1400 .
As an example, the performance strategy may specify a 25 As such , the control system may take various forms of a
first power consumption target for the set of computing computing system , such as a mobile computing device , a
systems for a first time interval such that the first power wearable computing device , a network of computing sys
consumption target is equal to or greater than a first mini tems, etc.
mum power threshold associated with the first time interval At block 1402, the method 1400 involves monitoring a set
and a second power consumption target for the set for a 30 of conditions. Similar to block 1302 of themethod 1300, a
second time interval in a similar manner (i.e., the second computing system may monitor various conditions to deter
power consumption target is equal to or greater than a mine instructions for controlling a setof computing systems.
second minimum power threshold ). At block the method 1400 involves receiving first
In some examples, the performance strategy may include power option data based , at least in part, on a power option
an sequence for the set of computing systems to follow when 35 agreement while monitoring the set of conditions. The first
performing computational operations. The sequence, for power option data may specify a first minimum power
example, may be based on priorities associated with the threshold associated with a first time interval. For example ,
computational operations. In addition , the performance the first power option data may specify a minimum power
strategy may include one or more power consumption threshold of 10 MW for the next hour, which may start in an
targets that are greater than the minimum power thresholds 40 hour or less.
when the price of power from the power grid is below a The power option agreement may correspond to a
threshold price during the time intervals associated with the dynamic power option agreement in some examples. When
minimum power thresholds . managing a load with respect to a dynamic power option
The performance strategy may also involve transferring , agreement, a computing system may receive power option
delaying, or adjusting one or more computational operations 45 data specifying changes in minimum power thresholds that
performed at the set of computing systems. In addition , the a load ( e.g., the set of computing systems) may be desig
performance strategy may involve adjusting operations at nated to use in the near term (e.g., the next hour ). For
the computing systems. For instance , one or more comput example, the computing system may receive power option
ing systems may switch modes (e.g., operate at a higher data during each hour of the duration specified by a power
frequency or switch to a low power mode). 50 option agreement that indicates a minimum power threshold
In addition , the performance strategy may also specify for the next hour.
power consumption targets for the set of computing systems At block 1406 , the method 1400 involves providing first
to use if the power option is exercised during an interval. control instructions for a set of computing systems based on
This way, the computing systems may continue to perform a combination of at least a portion of the first power option
computational operations (or suspend performance) based 55 data and at least one condition . The first control instructions
on the power option being exercised . may be provided responsive to receiving the first power
At step 1308 , the method 1300 involves providing option data .
instructions to the set of computing systems to perform one The first control instructions may include a first power
or more computational operations based on the performance consumption target for the set of computing systems for the
strategy. For example, the set of computing systems may 60 first time interval. Particularly , the first power consumption
operate according to the performance strategy to ensure that target may be equal to or greater than the first minimum
the minimum power thresholds are met during the defined power threshold associated with the first time interval. For
time intervals based on the power option agreement. example , the first power consumption target may be greater
Some examples may further involve receiving subsequent than the first minimum power threshold when a cost of
power option data based , at least in part, on the power option 65 power from the power grid is below a threshold price during
agreement. The subsequent power option data may specify the first time interval. In other instances, the first power
to decrease one or more minimum power thresholds of the consumption target may be equal to the firstminimum power
US 10,608,433 B1
57 58
threshold when the cost of power from the power grid is Advantages of one or more embodiments of the present
greater than the threshold price . invention may include one or more of the following :
In some examples, control instructions may specify a One or more embodiments of the present invention pro
sequence for the computing systems to follow when per vides a green solution to two prominent problems: the
forming computational operations. The sequence may be 5 exponential increase in power required for growing block
based on priorities associated with each computational chain operations and the unutilized and typically wasted
operation . energy generated from renewable energy sources .
The first control instructions may be determined based on One ormore embodiments of the present invention allows
a combination of the first power option data, the price of 10 for the rapid deployment of mobile datacenters to local
power from the power grid , and parameters associated with stations. The mobile datacenters may be deployed on site ,
computational operations to be performed at the set of near the source of power generation , and receive low costor
computing systems. unutilized power behind - the -meter when it is available .
In some examples, the first control instructions may One or more embodiments of the present invention pro
involve ramping up or down power consumption at the set 15 vide the use of a queue system to organize computational
of computing systems. The power consumption may be operations and enable efficient distribution of the computa
ramped up or down based on the first minimum power tional operations across multiple datacenters.
threshold and one ormore other conditions (e.g., the price of One ormore embodiments of the present invention enable
power ). datacenters to access and obtain computational operations
At block 1408 , the method 1400 involves receiving sec- 20 organized by a queue system .
ond power option data based , at least in part , on the power One ormore embodiments of the present invention allows
option agreement while monitoring the set of conditions. for the power delivery to the datacenter to be modulated
The computing system may receive the second power option based on conditions or an operational directive received
data subsequent to receiving the first power option data . The from the local station or the grid operator.
second power option data may specify a second minimum 25 One or more embodiments of the present invention may
power threshold associated with a second time interval. For dynamically adjust power consumption by ramping-up ,
example , the second minimum power threshold may be 7 ramping -down, or adjusting the power consumption of one
MW over the duration of the upcoming hour. In other or more computing systems within the flexible datacenter.
examples, the second minimum power threshold may differ One or more embodiments of the present invention may
as shown in FIG . 12 . 30 be powered by behind - the -meter power that is free from
In some instances, the computing system may receive the transmission and distribution costs. As such , the flexible
second power option data during the first time interval such datacenter may perform computational operations, such as
that the second time interval overlaps the first time interval. distributed computing processes, with little to no energy
For instance , the computing system may receive the second cost.
power option data to enable real-time adjustments to be 35 One or more embodiments of the present invention pro
made to the power consumed at the set of computing vides a number of benefits to the hosting local station . The
systems. local station may use the flexible datacenter to adjust a load ,
At block 1410 , the method 1400 involves providing provide a power factor correction , to offload power, or
second control instructions for the set of computing systems operate in a manner that invokes a production tax credit
based on a combination of at least a portion of the second 40 and / or generates incremental revenue.
power option data and at least one condition. The second One or more embodiments of the present invention allows
control instructionsmay be provided responsive to receiving for continued shunting of behind-the-meter power into a
the second power option data. The second control instruc storage solution when a flexible datacenter cannot fully
tions may specify a second power consumption target for the utilize excess generated behind-the -meter power.
set of computing systems for the second time interval. The 45 One or more embodiments of the present invention allows
second power consumption target may be equal to or greater for continued use of stored behind-the -meter power when a
than the second minimum power threshold associated with flexible datacenter can be operational but there is not an
the second time interval. excess of generated behind -the -meter power.
In some examples, the computing system may provide a One or more embodiments of the present invention allows
request to a QSE to determine the power option agreement. 50 for management and distribution of computational opera
As such , the computing system may receive power option tions at computing systems across a fleet of datacenters such
data ( e.g., the first and second power option data ) in that the performance of the computational operations take
response to providing the request to the QSE . advantages of increased efficiency and decreased costs.
The computing system may monitor the price of power It will also be recognized by the skilled worker that, in
from the power grid, and the global mining hash rate and a 55 addition to improved efficiencies in controlling power deliv
price for a cryptocurrency ( e.g., Bitcoin ), among other ery from intermittent generation sources, such as wind farms
conditions. The computing system may determine control and solar panel arrays, to regulated power grids, the inven
instructions (e.g., the first and/or second control instruc tion provides more economically efficient control and sta
tions) based on a combination of power option data, the bility of such power grids in the implementation of the
price of power from the power grid , and the global mining 60 technical features as set forth herein .
hash rate and the price for the cryptocurrency. For instance , While the present invention has been described with
the computing system may cause one or more computing respect to the above -noted embodiments , those skilled in the
systems ( e.g., a subset of computing systems) to perform art, having the benefit of this disclosure, will recognize that
mining operations for the cryptocurrency when the price of other embodiments may be devised that are within the scope
power from the power grid is equal to or less than a revenue 65 of the invention as disclosed herein . Accordingly , the scope
obtained by performing the mining operations for the cryp of the invention should be limited only by the appended
tocurrency . claims.
US 10,608,433 B1
59 60
What is claimed is : portion of the subsequent power option data and at least
1. A system comprising : one condition in the set of conditions,
a set of computing systems, wherein the set of computing wherein themodified performance strategy comprises one
systems is configured to perform computational opera or more reduced power consumption targets for the set
tions using power from a power grid ; 5 of computing systems.
a control system configured to : 8. The system of claim 7, wherein the control system is
monitor a set of conditions; further configured to :
receive power option data based , at least in part, on a provide instructions to the set of computing systems to
power option agreement, wherein the power option perform the one or more computational operations
data specify : (i) a set ofminimum power thresholds, 10 based on the modified performance strategy .
and ( ii ) a set of time intervals, wherein each mini
mum power threshold in the set of minimum power remoteThemaster 9. system of claim 1, wherein the control system is a
control system positioned remotely from the
thresholds is associated with a time interval in the set
of time intervals ; set of computing systems.
responsive to receiving the power option data , deter- 15 a mobile computingof device
10. The system claim 1 , wherein the control system is
.
mine a performance strategy for the set of computing
systemsbased on a combination of at least a portion 11. The system of claim 1, wherein the control system is
of the power option data and at least one condition in configured to receive the power option data while monitor
the set of conditions, wherein the performance strat ing the set of conditions.
egy comprises a power consumption target for the set 20 12. The system of claim 1 , wherein the control system is
of computing systems for each time interval in the further configured to :
set of time intervals, wherein each power consump provide a request to a qualified scheduling entity (QSE ) to
tion target is equal to or greater than the minimum determine the power option agreement; and
power threshold associated with each time interval; 25 receive power option data in response to providing the
and request to the QSE .
provide instructions to the set of computing systems to 13. The system of claim 1 ,wherein the power option data
perform one or more computationaloperations based specify: (i) a first minimum power threshold associated with
on the performance strategy . a first time interval in the set of time intervals, and (ii) a
2. The system of claim 1, wherein the control system is second minimum power threshold associated with a second
configured to monitor the set of conditions comprising : 30 time interval in the set of time intervals,
a price of power from the power grid ; and wherein the second time interval is subsequent to the first
a plurality of parameters associated with one or more time interval.
computational operations to be performed at the set of 14. The system of claim 13, wherein the control system is
computing systems.
3. The system of claim 2 , wherein the control system is 35 configured to :
configured to : determine the performance strategy for the set of com
determine the performance strategy for the set of com puting systems such that the performance strategy
puting systems based on a combination of at least the comprises:
portion option data, the price of power from the power a first power consumption target for the set of computing
grid , and the plurality of parameters associated with the 40 systems for the first time interval, wherein the first
one or more computational operations. power consumption target is equal to or greater than the
4. The system of claim 3 , wherein the performance firstminimum power threshold ; and
strategy further comprises: a second power consumption target for the set of com
an order for the set of computing systems to follow when puting systems for the second time interval, wherein
performing the one or more computational operations, 45 the second power consumption target is equal to or
wherein the order is based on respective priorities greater than the second minimum power threshold .
associated with the one or more computational opera 15. The system of claim 1, wherein a total duration of the
tions. set of time intervals corresponds to a twenty -four hour
5. The system of claim 4 , wherein the performance period .
strategy further comprises : 50
16. The system of claim 1, wherein the set of conditions
at least one power consumption target that is greater than monitored by the control system further comprise :
a minimum power threshold when the price of power a price of power from the power grid ; and
from the power grid is below a threshold price during a global mining hash rate and a price for a cryptocurrency ;
the time interval associated with the minimum power and
threshold . 55
6. The system of claim 1, wherein the control system is wherein the control system is configured to :
further configured to : determine the performance strategy for the set of com
receive subsequent power option data based , at least in puting systems based on a combination ofat the portion
part, on the power option agreement, of the power option data , the price of power from the
wherein the subsequent power option data specify to 60 power grid, the global mining hash rate and the price
decrease one ormore minimum power thresholds of the for the cryptocurrency ,
set of minimum power thresholds. wherein the performance strategy specifies for at least a
7. The system of claim 6 , wherein the control system is subset of the set of computing systems to perform
further configured to : mining operations for the cryptocurrency when the
responsive to receiving the subsequent power option data, 65 price of power from the power grid is equal to or less
modify the performance strategy for the set of com than a revenue obtained by performing the mining
puting systems based on a combination of at least the operations for the cryptocurrency.
US 10,608,433 B1
61 62
17. A method comprising: responsive to receiving the subsequent power option data ,
monitoring, by a computing system , a set of conditions; modifying the performance strategy for the set of
receiving , at the computing system , power option data computing systems based on a combination of at least
based , at least in part, on a power option agreement, the portion of the subsequent power option data and at
wherein the power option data specify: (i) a set of 5 least one condition in the set of conditions, wherein the
minimum power thresholds, and (ii) a set of time modified performance strategy comprises one or more
intervals, wherein each minimum power threshold in reduced power consumption targets for the set of com
the set ofminimum power thresholds is associated with puting systems, and
a time interval in the set of time intervals;
responsive to receiving the power option data, determin- 10 providing instructions to the set of computing systems to
perform the one or more computational operations
ing a performance strategy for a set of computing based on the modified performance strategy.
systems based on a combination of at least a portion of
the power option data and at least one condition in the stored Atherein
20. non -transitory computer readable medium having
instructions executable by one or more pro
set of conditions, wherein the performance strategy
comprises a power consumption target for the set of 15 cessors to cause a computing system to perform functions
computing systems for each time interval in the set of comprising :
time intervals , wherein each power consumption target monitoring a set of conditions ;
is equal to or greater than the minimum power thresh receiving power option data based , at least in part, on a
old associated with each time interval; and power option agreement, wherein the power option
providing instructions to the set of computing systems to 20 data specify : (i) a set of minimum power thresholds ,
perform one or more computational operations based and (ii ) a set of time intervals , wherein each minimum
on the performance strategy . power threshold in the set of minimum power thresh
18. The method of claim 17, wherein determining the olds is associated with a time interval in the set of time
performance strategy for the set of computing systems intervals;
comprises: 25 responsive to receiving the power option data , determin
identifying information about the set of computing sys ing a performance strategy for a set of computing
tems; and systemsbased on a combination of at least a portion of
determining the performance strategy to further comprise the power option data and at least one condition in the
instructions for at least a subset of the setof computing set of conditions, wherein the performance strategy
systems to operate at an increased frequency based on 30 comprises a power consumption target for the set of
a combination of at least the portion of the power computing systems for each time interval in the set of
option data and the information about the set of com time intervals , wherein each power consumption target
puting systems. is equal to or greater than the minimum power thresh
19. The method of claim 17 , further comprising: old associated with each time interval; and
receiving subsequent power option data based , at least in 35 providing instructions to the set of computing systems to
part, on the power option agreement, wherein the perform one or more computational operations based
subsequent power option data specify to decrease one on the performance strategy .
or more minimum power thresholds of the set of
minimum power thresholds ;