JP2021189845A - Data center system and operation method thereof, as well as apparatus and method for supporting data center system hub design - Google Patents

Abstract

To provide a data center system configured to improve economical efficiency in utilizing electric power generated by renewable energy power sources disposed in hubs for drive power of the data center system composed of a plurality of data centers, and an operation method of the data center system, as well as an apparatus and method for supporting data center system hub design.SOLUTION: In a data center system, data centers DC in a plurality of hubs and a data center system energy management device EM are connected by communication. Servers SV of the data centers each include a renewable energy power source G to interchange electric power between the hubs. The data center system energy management device includes task allocation adjustment means which adjusts hubs to execute information processing so as to suppress execution of the information processing in a data center which is likely to be short of renewable energy in the hub, and to increase execution of the information processing in a data center which is likely to have a surplus of renewable energy in the hub, to adjust task allocation in the data centers in the hubs.SELECTED DRAWING: Figure 1

Description

The present invention relates to a data center system that manages the drive power of the entire data center system composed of a plurality of data centers, an operation method thereof, and a data center system base design support device and method.

With the recent development of communication technology and cloud computing, data centers that process various types of data are increasingly installed in multiple locations to build a data center system that is configured in a distributed format for centralized management via communication. ing.

On the other hand, with the increase in information traffic, the power consumption in the data center is increasing, and the reduction of the environmental load of the driving power has become an issue. Even in a distributed data center system, improving the total energy efficiency of the data center is one of the countermeasures, and replacing the information equipment and air conditioning equipment installed in the data center with energy-efficient ones. And so on.

Regarding these points, in Non-Patent Document 1, the operating location of information processing is moved among a plurality of servers installed in a plurality of data centers, and the total number of servers required for information processing is reduced. It is stated that the server that is turned on increases the operating rate, and the server that is not used is turned off to improve energy efficiency.

"Energy Saving of Cloud Computing", AIST TODAY, VOL. 10-2, 2010.2

In addition to improving the energy efficiency of equipment in order to reduce the environmental load of drive power in data center systems, drive power is referred to as renewable energy power sources such as solar power generation and wind power generation (hereinafter referred to as renewable energy power sources). There is a way to cover it with the power generated in). This method uses the power generated by installing a renewable energy power source on-site at the data center base or the power procured from the renewable energy power source installed off-site at another location to use various devices in the data center. It is the one that drives.

However, the on-site re-energy power source has a insufficient power generation amount depending on the weather conditions such as the amount of solar radiation and the air volume, and the off-site re-energy power source has an access cost of the power system, which deteriorates the economic efficiency.

From the above, it is an object of the present invention to improve economic efficiency in utilizing the power generated by the renewable energy power source installed at each base for the drive power of the data center system composed of a plurality of data centers. The purpose is to provide a data center system and its operation method, as well as a data center system base design support device and method.

From the above, the present invention states that "the data centers at a plurality of sites and the data center system energy management device are connected by communication, and the data centers at a plurality of sites including a server for data calculation are each provided with a renewable energy power supply. It is a data center system configured to exchange power between bases, and the data center energy management device suppresses the execution of information processing in the data center where the shortage of renewable energy power in the base is predicted, and the base Equipped with a task allocation adjustment means to adjust the execution base of information processing so as to increase the execution of information processing in the data center where the surplus of renewable energy power is predicted, and adjust the task allocation in the data centers of multiple bases. It is a "data center system characterized by doing".

Further, the present invention states that "a data center at a plurality of bases and a data center system energy management device are connected by communication, and a data center at a plurality of bases equipped with a server for data calculation is provided with a renewable energy power supply and power is supplied between the bases. It is an operation method of a data center system configured to be flexible, and the data center energy management device suppresses the execution of information processing in the data center where the shortage of renewable energy power in the base is predicted, and in the base. The operation method of the data center system, which is characterized by adjusting the execution base of the information processing so as to increase the execution of the information processing in the data center where the surplus of the renewable energy power is predicted. "

Further, the present invention states that "a data center at a plurality of bases and a data center system energy management device are connected by communication, and a data center at a plurality of bases equipped with a server for data calculation is provided with a renewable energy power supply and power is supplied between the bases. A data center system base design support device to support the base design of a data center system configured to be flexible, and the data center is arranged so as to reduce the total power amount or total power cost of the data center system. Set the location of each base and the power generation capacity of the renewable energy power generation equipment installed at each base, and based on the difference in energy efficiency of information processing depending on the region and the difference in the annual power generation rate of the renewable energy power generation equipment, the data center system It is a data center system base design support device characterized by calculating the total amount of renewable energy power that drives the entire system. "

Further, the present invention states that "a data center at a plurality of bases and a data center system energy management device are connected by communication, and a data center at a plurality of bases equipped with a server for data calculation is provided with a renewable energy power supply and power is supplied between the bases. It is a data center system base design support method to support the base design of a data center system configured to be flexible, and the data center is arranged so as to reduce the total power amount or the total power cost of the data center system. Set the location of each base and the power generation capacity of the renewable energy power generation equipment installed at each base, and based on the difference in energy efficiency of information processing depending on the region and the difference in the annual power generation rate of the renewable energy power generation equipment, the data center system It is a data center system base design support method characterized by calculating the total amount of renewable energy power that drives the whole. "

According to the present invention, among the generated power of the renewable energy power source used for the drive power of the data center, the amount procured from the off-site renewable energy power source in another place is reduced, and the on-site of the data center base is used. It is possible to increase the amount of power generated by renewable energy sources.

Issues, configurations and effects other than those described above will be clarified by the following description of the embodiments.

It is a figure which shows the whole structure example of the data center system to which this invention is applied. The functional block diagram which shows the structural example of the data center energy management apparatus which concerns on Example 1 of this invention. The functional block diagram which shows the structural example of the base management apparatus which concerns on Example 1 of this invention. The figure which shows the flow of the whole processing by the data center energy management apparatus and the base management apparatus which concerns on Example 1 of this invention. The figure which shows the composition example of the base supply-demand record data D11a. The figure which shows the example of the structure of the base supply and demand forecast data D11b. The figure which shows the configuration example of the task allocation constraint data D12. The figure which shows the extended configuration example of the task allocation constraint data D12. The figure which shows the structural example of the electric power procurement risk data D13. The functional block diagram which shows the configuration example of the data center system base design support apparatus SP which concerns on Example 2. The figure which shows the flow of the whole processing of the data center system base design support apparatus SP which concerns on Example 2. FIG. The figure which shows the flow of the optimization method using a genetic algorithm.

Next, a mode for carrying out the present invention will be described in detail with reference to the drawings as appropriate.

In the first embodiment, a data center system and a method will be described. FIG. 1 is a diagram showing an overall configuration example of a data center system to which the present invention is applied.

The data center system shown in FIG. 1 combines the energies of a plurality of data center DCs installed at a plurality of points (in the example of the figure, an example composed of three DCa, DCb, and DCc) and a plurality of data centers. It is composed of a data center system energy management device EM to be managed, an information network Nw and a power network Pw connecting between a plurality of data center DCs and the data center system energy management device EM. As a result, information sharing and power interchange are achieved in the data center system.

In the data center DC, a server SV (SVa, SVb, SVc) for performing data processing, a renewable energy power supply G (Ga, Gb, Gc), and a base management device C (Ca, Cb, Cc) are installed. Further, a power supply bus B (Ba, Bb, Bc) is arranged in the data center DC, and the power supply bus B is connected to a renewable energy power supply G, a base management device C, and a power network Pw. Further, various devices in the data center are connected to the information network Nw. In the following, the expression "base" shall be used as an expression meaning the place where the data center is located or the data center itself. Renewable energy refers to renewable energy, but here it is abbreviated as renewable energy.

Hereinafter, each device constituting the data center DC of FIG. 1 and its functions will be described in detail. First, the data center DC is a facility of a geographical scale such as in one building or on the premises, and is a facility configured by installing a server SV, a base management device C, and the like. The data center system energy management device EM may be installed in any of these facilities, or may be installed in another facility.

The server SV (SVa, SVb, SVc) is an information processing resource responsible for storing and processing information, and includes, for example, a storage device, a calculation processing device, an external input / output interface, and the like as devices necessary for the information processing resource.

Since the power for driving the server SV (SVa, SVb, SVc) is supplied from the power supply bus B in the data center DC, in the example shown in the figure, the power is supplied via the renewable energy power supply G in the self-data center DC or the power network Pw. It is possible to supply power from the renewable energy power supply G or the external power supply GO in the data center DC of another base. As a result, power is basically supplied within the own base, but if necessary, supply support from other bases and supply support from an external power source GO are possible. Further, in this example, the renewable energy power source G is exemplified as the distributed power source of each base, but this may be configured to generate power by installing a distributed power source such as wind power generation or cogeneration equipment. FIG. 1 shows an example including a renewable energy power source G as such a configuration example.

The base management device C (Ca, Cb, Cc) is connected to various devices in the data center DC in addition to the data center system energy management device EM by the information network Nw. As a result, the base management device C monitors and records the status of various devices (server SV, renewable energy power supply G) in its own base (data center DC), and also controls and manages them. For example, the power consumption of the server SV is monitored and recorded, the power generation amount of the renewable energy power source G is monitored and recorded, and the start and stop of these facilities are controlled.

The data center system energy management device EM operates in cooperation with the base management device C of each data center DC, and manages the supply and consumption of power that drives the entire data center system.

The equipment constituting the data center system of FIG. 1 (a plurality of data center DCs and a data center system energy management device EM) is connected via an information network Nw and is configured to be able to operate in cooperation with each other. Specifically, when a request for information processing (hereinafter referred to as "task") to the data center system is received by the server SV (for example, SVa) of one of the bases, the request of the one of the bases is based on the situation of each base. It is assigned to a server SV (for example, SVb), and the requested task is executed on the assigned server SVb.

FIG. 2 is a functional block diagram showing a configuration example of the data center energy management device EM according to the first embodiment. The data center energy management device EM composed of computer devices includes a power supply / demand management unit 12, a task allocation adjustment unit 13, a power procurement risk management unit 14, a base information collection unit 15, and a data management unit DB1 composed of a storage device. It is configured to include the overall control unit 11 and the input / output unit 16, and each functional block operates in cooperation with each other. The data management unit DB1 is further configured to include base supply / demand status data D11, task allocation constraint data D12, and power procurement risk data D13.

FIG. 3 is a functional block diagram showing a configuration example of the base management device C according to the first embodiment. The site management device C composed of computer devices includes a site status monitoring unit 26, a site equipment control unit 22, a task execution determination unit 23, a data arrangement adjustment unit 24, a data management unit DB2 composed of a storage device, and the whole. It is configured to include a control unit 21 and an input / output unit 25, and each functional block operates in cooperation with each other. The data management unit DB2 is further configured to include past actual data D21 and task allocation constraint data D22.

FIG. 4 is a diagram showing the overall processing flow by the data center energy management device EM and the site management device DC according to the first embodiment of the present invention. Hereinafter, the processing of each functional block shown in FIGS. 2 and 3 will be described according to the flow of FIG.

The overall control unit 11 of the data center system energy management device EM starts the entire process by receiving the process start input from the input / output unit 16 of the data center system energy management device EM. Then, the task allocation constraint data D12 managed by the data management unit DB1 is initialized.

FIG. 6a is a diagram showing a configuration example of the task allocation constraint data D12. As shown in FIG. 6a, the task allocation constraint data D12 is configured by associating a task allocation allowance rate D122 with a base identifier D121 for distinguishing bases (data centers DC (DCa, DCb, DCc)). .. In the initialization process, the value TAR (DC) of the task allocation allowance D122 is set to 1 for all the bases. The above is the content to be executed in the first processing step S100 in the flow of FIG.

Subsequently, the overall control unit 11 monitors the current time in the processing step S101, and when it reaches every hour or every 30 minutes, commands each functional block to proceed with the following processing.

First, the overall control unit 11 collects base information. This process is process step S102. Here, the base information collecting unit 15 is instructed to collect the past actual data D21 regarding the supply and demand situation of each base. The past actual data D21 to be collected is more specifically data of the amount of re-energy power generation, the amount of power consumption, the amount of re-energy surplus power, the weather, the temperature, and the wind speed at each date and time in the past of each base. In response to the command, the base information collecting unit 15 acquires the past actual data D21 managed by the base management device C of each base, and the acquired result is the base supply / demand status data D11 managed by the data management unit DB1. It is stored in the base supply / demand record data D11a constituting the above.

FIG. 5a is a diagram showing an example of the configuration of the base supply / demand actual data D11a. The base supply / demand actual data D11a is composed of the base identifier D11a1, the date and time D11a2, and the actual data D11a3, and as the actual data D11a3 of each date and time of each base, the re-energy power generation amount D11a31, the power consumption amount D11a32, and the re-energy surplus power amount D11a33. , Weather D11a34, temperature D11a35, and wind speed D11a36 are associated with each other.

Next, the overall control unit 11 predicts the power supply and demand situation from the processing step S103 to the processing step S105. Therefore, the power supply and demand management unit 12 is sequentially instructed to predict the amount of re-energy power generation, the amount of power consumption, and the amount of re-energy surplus power for 30 minutes from 30 minutes after each base.

The electric power supply and demand management unit 12 stores the result of executing each prediction process in response to these commands in the base supply and demand forecast data D11b constituting the base supply and demand status data D11. FIG. 5b is a diagram showing an example of the configuration of the base supply and demand forecast data D11b. The base supply / demand forecast data D11b is composed of the base identifier D11b1 and the forecast data D11b2, and further, as the forecast data D11b2 of each base, each predicted value of the re-energy power generation amount D11b21, the power consumption amount D11b22, and the re-energy surplus power amount D12b23 is used. It is configured in association with each other.

Of these, the prediction of the amount of renewable energy power generation D11b21 is executed as follows in the processing step S103. Forecast information (weather, temperature, wind speed, etc.) about the weather after t = 30 minutes of each base (S) is obtained, and a plurality of similar past date and time candidates are searched for in the base supply / demand record data D11a of FIG. 5a. The re-energy power generation amount D11a31 associated with the past date and time is extracted, the average value thereof is calculated, and this is used as the predicted value of the re-energy power generation amount Gre_f (S). The weather forecast information after 30 minutes is obtained from an external weather service.

The prediction of the power consumption D11b22 is executed in the processing step S104 as follows. Forecast information (weather, temperature, wind speed, etc.) about the weather after t = 30 minutes of each base (S) is obtained, and a plurality of similar past date and time candidates are searched for in the base supply / demand actual data D11a in FIG. 5a. The power consumption D11a32 associated with the past date and time is extracted, the average value thereof is calculated, and the predicted value Gdc_f (S) of the power consumption is used as the base. Here, the candidate for which the average value is calculated may be limited to the data of the past date and time in which the month and the day of the week at the time after 30 minutes match. Then, the allocation tolerance D122 of each base D121 is referred to from the allocation constraint data D12 of FIG. 6a, and the previously obtained power consumption forecast value base is multiplied by this allocation allowance D122 to obtain the predicted value of the power consumption. ..

The prediction of the amount of surplus renewable energy is executed in the processing step S105 as follows. Using the predicted value of re-energy power generation and the predicted value of power consumption obtained above, subtract the predicted value of power consumption from the predicted value of re-energy power generation, and use it as the predicted value of re-energy surplus power Sre_f (S) = Let it be Gre_f (S) -Gdc_f (S).

Next, the overall control unit 11 executes the process for task allocation adjustment in the process step S106. Here, with reference to the base supply / demand forecast data D11b in FIG. 5b, the bases having the maximum and minimum renewable energy surplus electric energy forecast values Sre_f (S) are extracted, and the identifiers of those bases are set to S_sremax and S_sremin, respectively, and these bases are used. Let Sre_f (S_sremax) and Sre_f (S_sremin) be the predicted values of the amount of surplus power generated by the renewable energy.

Then, if Sre_f (S_sremax) is negative in the processing step S107, or if Sre_f (S_sremin) is positive in the processing step S108, the overall control unit 11 considers the processing at that time to be complete and moves to the processing step S110. .. In the processing step S110, if there is a processing stop input from the input / output unit 16, the entire processing is stopped, and if there is no processing stop input, the process returns to the step of monitoring the current time.

Further, when Sre_f (S_sremax) is positive and Sre_f (S_sremin) is negative in a series of processes of the process step S107 and the process step S108, the overall control unit 11 is sent to the task allocation adjustment unit 13 for 30 minutes. Order the adjustment of the task allocation for 30 minutes later. In response to the command in the process step S109, the task allocation adjustment unit 13 updates the task allocation allowance rate D122 of those bases stored in the task allocation constraint data D12 of FIG. 6a as follows. The task allocation allowance of the base S_sremin with the minimum surplus power is updated with the value obtained by subtracting Δ from that value and the larger value of 0, and the task allocation allowance of the base S_sremax with the largest surplus power is that. Update with the value obtained by adding Δ to the value and the smaller value of 1. Here, Δ is a predetermined real number constant of 1 or less. After the task allocation adjustment process, the overall processing unit 11 returns to the process step S101 for predicting the power consumption and repeats the process.

With this configuration, the task allocation allowance is set smaller for the bases where the renewable energy power is insufficient, and the task allocation allowance is set larger for the bases where the renewable energy power is surplus.

The above is the processing operation in the data center system energy management device EM up to the task allocation. In response to this result, the data center system energy management device EM gives a task request to the data center DC of the base where the task allocation allowance is set to be large. The operation and processing in the data center DC after receiving the task request will be described below.

Upon receiving the task request, the server SV of the data center DC at each site instructs the site management device C at the site to determine whether or not to allow the execution of the task. The site management device C obtains the task allocation constraint data D12 of FIG. 6a from the data center system energy management device EM, sequentially copies it to the data management unit DB2, and manages it locally as the task allocation constraint data D22. When the site management device C receives a command for determining task execution, it refers to the locally managed task allocation constraint data D22, refers to the task allocation allowance rate D122 of its own site set therein, and refers to the task. A random number is used to determine whether or not to allow the execution of the task so that the task request is allowed with the probability of the allocation allowance rate D122.

When the site management device C determines that the task execution is permitted, the site management device C instructs the server SV of the site to execute the task. If it is determined not to allow it, another base is randomly selected, and the server SV of the own base is instructed to forward the task request to the server SV of the selected different base. The server SV of the own base executes the process specified by the command from the base management device C.

With this configuration, the lower the task allocation allowance rate D122, the more the task execution amount can be expected to decrease. The task allocation allowance rate D122 is set smaller for the bases where the re-energy power shortage is predicted and larger for the bases where the re-energy power surplus is predicted by the above-mentioned method for creating the task allocation constraint data D12.

Therefore, it can be expected that the task execution amount will decrease for the bases where the re-energy power shortage is predicted, and the task execution amount will increase for the bases where the re-energy power surplus is predicted. , The effect of shifting task allocation can be expected to the bases where surplus renewable electricity is predicted. In other words, it is possible to increase the ratio of the power that drives each site to be covered by the on-site re-energy power source of that site, and suppress the generation of power system access costs required when procuring off-site re-energy power. The effect of

In task allocation, if most of the data required to process the task is stored in a site other than the site where the task is executed, it takes time to access the required data from the server that executes the task. It is necessary and may increase the time for task processing. In order to avoid this as much as possible, it may be extended to the following configuration considering related data access restrictions.

First, the task request to the data center system DC is configured with the information classification of the data required for the task processing. The information classification shall be given by the user who requested the task.

Therefore, it is preferable to use the configuration of FIG. 6b, which is an extension of the task allocation constraint data of FIG. 6a. FIG. 6b is a diagram showing an example of the expanded configuration. Here, the stored information classification identifier D123 is added in association with each base. The stored information classification identifier D123 is data indicating the information classification of the data stored in each site.

In the example of FIG. 6b, it is shown that the data of the information classification IC1, IC2, and IC3 are stored in the base DCa, and the data of the information classification IC1, IC4, IC5, and IC6 are stored in the base DCb. The server SV of each base is configured to notify the base information collection unit 15 of the data center system energy management device EM when the data is stored in the base. The site information collection unit 15 of the data center system energy management device EM stores the stored information classification identifier data D123 corresponding to the site of the task allocation constraint data D12 in accordance with the notification received from the server of each site. It is configured to include the information classification identifier D123 of the data and update it.

Then, the processing of the task execution determination by the base management device C is changed as follows. Upon receiving a command for determining task execution, the site management device C determines whether or not to allow the execution of the task. However, when the task allocation constraint data D22 managed locally is referred to, the stored information classification identifier D123 of the own base set therein is referred to, and the information classification assigned to the task is included. Determines to allow the execution of this task. If this is not the case, the task allocation allowance rate D122 of the own base is referred to, and a random number is used to determine so that the task request is allowed with the probability of the task allocation allowance rate. When the site management device C determines that the task execution is permitted, the site management device C instructs the server SV of the site to execute the task. If it is determined not to allow it, one of the other bases is randomly selected, and the server SV of the own base is instructed to forward the task request to the server SV of the other base.

By configuring in this way, in the case of a task for which the information classification of the required information is explicitly specified, the effect of increasing the accuracy of allocating the task to the base where the required information is likely to be stored can be improved. Obtainable.

Furthermore, in task allocation, in order to increase the probability that the base where the data required for task processing is stored and the surplus of renewable energy power is predicted to be selected as the base to be assigned, more tasks will be performed. The information of the required information classification may be expanded to the following configuration with the addition of a data arrangement adjustment function that copies and stores the information of the required information classification in advance at the base where the surplus of the renewable energy power is generated in advance.

In this case, if the site management device C determines in the task execution determination process that the task execution at the site is permitted by a random number based on the standard of the task allocation allowance rate D122 of the site, the site management device C manages the task at its own site. With reference to the task allocation constraint data, the bases that include the information classification assigned to the task in the stored information classification identifier D123 are extracted, and the information corresponding to the information classification is extracted from the server SV of the base. It is configured to execute the process of acquiring, copying and storing in the server of the relevant site.

With this configuration, there are more opportunities to copy the information used for task processing to a site with a higher task allocation allowance, that is, a base with a larger amount of renewable energy surplus power, and with more task allocation. The more information in the information classification referenced, the more opportunities to copy the information, and therefore the opportunity to copy the information in the information classification needed for more tasks to a site with more surplus renewable energy. Can be obtained.

Next, an example of expansion that considers uncertainty in the prediction of the amount of surplus renewable energy at each site will be described.

There is uncertainty in the forecast of renewable power generation at each site. Fluctuations in power generation due to uncertainties in weather forecasts, power generation curtailment based on the status of the power system to which renewable energy power sources are interconnected, etc. There are also uncertainties in the power consumption forecast, such as fluctuations in air conditioning power due to weather conditions. Therefore, there is uncertainty in predicting the amount of surplus electricity for renewable energy. If the amount of re-energy power generation is lower than expected at one base and the re-energy power is insufficient, in order to cover the base drive power with the re-energy power, it is necessary to procure the re-energy power from another base, etc. Cost is incurred. On the contrary, if the amount of re-energy power generation increases more than expected at one base and the re-energy power becomes surplus, the cost of power system access still occurs in order to utilize the surplus power at another base.

The uncertainty of the renewable energy surplus power forecast is more likely to be upside or downside, and its magnitude varies depending on the time and place. Therefore, if more tasks are assigned to bases that are expected to be upside and less tasks are assigned to bases that are expected to be downside, the cost of power system access will increase. The possibility of runout can be suppressed as a whole.

In order to be able to make adjustments in response to such power procurement risk, the following configuration may be used.

The data management unit DB1 of the data center system energy management device EM includes the power procurement risk data D13. FIG. 7 is a diagram showing an example of the configuration of the power procurement risk data D13. The power procurement impact data D13 is configured by associating each data of the expected value of the surplus power of renewable energy, the expected value of the downside of the surplus power of renewable energy, and the expected value of the unit price of procurement cost for 30 minutes from 30 minutes later. do. The expected value of surplus re-energy power is the expected value of the rate at which the amount of surplus re-energy power exceeds the predicted value. The expected value of the downside of the re-energy surplus power is the expected value [kWh] of the ratio of the amount of the re-energy surplus power to the downside from the predicted value. The expected unit price of the procurement cost is the expected unit price [¥ / kWh] per unit power amount of the procurement cost such as the power system access cost generated when procuring the renewable energy power of another base or the like.

The electric power procurement risk management unit 14 sets the electric power procurement risk data D13. Prediction information (weather, temperature, wind speed, etc.) about the weather 30 minutes after each base is obtained, and multiple candidates for the past date and time similar to it are searched for in the base supply and demand actual data and associated with the past date and time. Extract the amount of renewable energy power generation, calculate the average value for the deviation on the upside side and the average value for the deviation on the downside side from the average value, and calculate the expected value for the upside and the amount of renewable energy power generation, respectively. Expected downside. The expected value of the upside of the re-energy surplus power is set by the value obtained by dividing the expected value of the upside of the amount of re-energy power generation by the predicted value of the amount of re-energy surplus power, and the expected value of the downside of the re-energy surplus power is the re-energy power generation. Set by the value obtained by dividing the expected value of downside of the amount by the predicted value of the amount of surplus electricity generated. The expected procurement cost unit price is set in advance via the input / output unit.

Then, before starting the process of the task allocation adjustment, the task allocation allowance rate of each base is updated as follows. In other words, if the value obtained by subtracting the expected value of the re-energy surplus power downside from the expected value of the re-energy surplus power upside of each base is positive, then Δ'is added to the task allocation allowance and it is negative. In that case, subtract Δ'to set.

With this configuration, more tasks are assigned to the bases where it is expected that there is a high possibility that the amount of surplus re-energy power will be upside, and it is expected that there is a high possibility that the amount of surplus re-energy power will be downside. The effect of allocating less tasks to the bases can be obtained, and the effect of suppressing the possibility of an increase in the cost of power system access can be obtained as a whole.

In the second embodiment, a data center system base design support device and a method will be described. In the second embodiment, in the first embodiment, the geographical arrangement of each base of the data center and the capacity of the renewable energy power generation facility G attached to each base of the DS are adjusted, and the electric power driven by the entire data center system is regenerated. It supports the design to reduce the amount of renewable energy procured to cover 100% with energy and the procurement cost. The electric power to be procured includes the electric power generated by the renewable energy power generation facilities attached to each base, and the electric power generated by the renewable energy purchased from other power generation companies and the electric power trading market.

The base of the data center is located in the area where the task processing efficiency (task processing amount that can be driven by the unit power amount) is large, and the renewable energy power generation rate (the amount of power when 8760 hours per year is generated at the rated power generation capacity) is compared with the power amount in the area. A large-capacity renewable energy power generation facility will be installed at a base with a large power generation rate (ratio of power generation when generating power according to the solar radiation pattern of 8760 hours per year), and a large-capacity power generation will be re-installed at a base with high transmission accessibility to other bases. If an energy power generation facility is installed, it can be expected to reduce the total amount of renewable energy procured and the cost of procuring it.

However, the area with high task processing efficiency and the area with high renewable energy power generation rate do not always match. For example, suppose that a data center base is located in an area with high task processing efficiency. Since the task processing efficiency is high, it can be expected that the required driving power will be reduced. However, the base may have a low renewable energy generation rate. In that case, considering that the power generated by the renewable energy power generation equipment installed at the base is to be used, it is necessary to install a large-capacity renewable energy power generation equipment, and the equipment cost increases accordingly. On the other hand, in order to avoid the cost increase due to the increase in installed capacity, use the generated power of the re-energized power generation equipment of other bases with high re-energized power generation rate, and re-energized power of other power generation companies and the power trading market. It is conceivable to purchase and utilize electricity, but in that case, there is a cost of procuring renewable energy such as transmission loss, consignment fee, and renewable energy purchase fee.

Therefore, it is necessary to quantitatively analyze the above-mentioned total cost in appropriately designing the geographical arrangement of each base of the data center and the capacity of the renewable energy power generation facility attached to each base. In the second embodiment, as this quantitative analysis, when the placement candidate area of each data center base is set, the re-energy power procured in order to cover 100% of the drive power of the entire data center system with the re-energy in the placement pattern. It is related to the data center system base design support device that seeks the capacity of the renewable energy power generation equipment attached to each base so as to reduce the amount of power and its procurement cost.

FIG. 8 is a functional block diagram showing a configuration example of the data center system base design support device SP according to the second embodiment. The data center system base design support device SP includes a design condition setting unit 32, a model management unit 33, a data management unit DB3 composed of a storage device, an optimization calculation unit 34, an overall control unit 31, and an input / output unit 35. It is configured and each functional block operates in cooperation. The data management unit DB3 is further configured to include area characteristic data D31 and overall characteristic data D32.

The area characteristic data D31 is data for managing the conditions of the placement candidate area of each base of the data center, and is managed including data of task processing efficiency, renewable energy power generation rate, and power network connection performance. Of these, the task processing efficiency and the renewable energy power generation rate are as described above: (task processing amount that can be driven by a unit power amount) and (annual 8760 for the area concerned with respect to the power amount when 8760 hours per year are generated at the rated power generation capacity). It is the ratio of the amount of electric power when power is generated according to the solar radiation amount pattern of time). In addition, the power network connection performance data is further data on the "transmission loss rate", which is the ratio of the amount of power lost when transmitting a unit amount of power, and the "consignment unit price", which is the charge paid to the power network operator when transmitting a unit amount of power. Consists of including.

The overall characteristic data D32 is data that manages conditions common to all areas, and manages data of the assumed total task amount, the re-energy power purchase unit price, the re-energy power generation cost unit price, and the re-energy equipment cost unit price. The estimated total task amount is data representing the total amount of tasks processed by the entire data center system. The re-energy power purchase unit price is the unit price per unit power amount when procuring re-energy power from other businesses or the power trading market. The unit price of renewable energy power generation is the cost per unit power generation amount related to the operation cost of the renewable energy power generation facility attached to the data center base. The unit cost of renewable energy equipment is the cost per unit power generation capacity related to the initial cost required to install the renewable energy power generation equipment attached to the data center base.

FIG. 9 shows the overall processing flow of the data center system base design support device SP according to the second embodiment. Hereinafter, the processing of each functional block shown in FIG. 8 will be described according to the flow of FIG.

The overall control unit 31 of the data center system base design support device SP receives the processing start input from the input / output unit 35 and starts the processing of the data center system base design support.

First, the overall control unit 31 receives the setting input of the area characteristic data D31 of the Delia candidate for arranging the data center base from the user via the input / output unit 35. As described above, the area characteristic data D31 is data including each data of task processing efficiency and renewable energy power generation rate. This processing content is carried out in the processing step S200 of FIG.

Next, the setting input of the overall characteristic data D32 is received from the user via the input / output unit 35. Overall characteristic data D32 is data composed of data including estimated total task amount, re-energy power purchase unit price, re-energy power generation cost unit price, re-energy equipment cost unit price, and power network connection performance (transmission loss rate, consignment unit price). Is. This processing content is carried out in the processing step S201 of FIG.

The model management unit 33 determines the amount of drive power of each base, such as self-consumption of the power generated by the renewable energy power generation facility installed at that base, the consignment of the generated power of the renewable energy power generation facility installed at another base, and other operations. Manage the mathematical model of the optimization problem that determines the allocation of electricity purchased from power generators and the electricity trading market in combination. Therefore, it is defined in advance as follows.

First, the variable names related to the base i are defined as follows. The task processing efficiency is covered by TPRi, the re-energy power generation rate is covered by RGRi, the amount of power generated by the re-energy power source of the own base among the drive power is covered by SCPi, and the power generated by the re-energy power source of the base j is covered by the consignment of the drive power. The electric energy is TPji, the electric energy of the driving power is MPi, the electric energy that is covered by the purchase from other companies and the electric power trading market is OPi, the electric energy is OPi, the task amount is TVi, the re-energy power generation amount is RGi, and the re-energy installed capacity is. RCi, renewable energy equipment cost is expressed by RCCi, and power generation cost is expressed by PCi.

In addition, the variable names for the entire data center system are defined as follows. The estimated total task amount is TT, the re-energy power purchase unit price is RPUP, the re-energy power generation cost unit price is RGUP, the re-energy equipment cost unit price is RCUP, the transmission loss rate is TLR, the consignment unit price is TPUP, and the total power procurement amount is TPA. The power cost is expressed in TPC.

Using these variables, the optimization model is defined as follows. The optimization variables are SCPi (the amount of power generated by the renewable energy power source of the own base among the driving power) and TPji (the power generated by the renewable energy power source of the base j among the driving power) for all the bases i. Amount), MPi (amount of driving power that can be covered by purchasing power from other businesses or the power trading market).

Here, the drive power OPi, task amount TVi, re-energy power generation amount RGi, re-energy equipment capacity RCi, re-energy equipment cost RCCi, and power procurement cost PCi are calculated using other variables for all i. It is a dependent variable to be used. Specifically, it can be calculated from the equation (1) to the equation (6). However, Σj means that the sum is taken for the base j.
[Number 1]
OPi = SCPi + Σj (TPji × (1-TLR)) + MPi (1)
[Number 2]
TVi = OPi x TPRi (2)
[Number 3]
RGi = SCPi + TPji (3)
[Number 4]
RCi = RGi ÷ RGRi ÷ 8760 (4)
[Number 5]
RCCi = RCi × RCUP (5)
[Number 6]
PCi = RGi x RGUP + Σj (TPji x (1-TLR) x TPUP) (6)
The total power procurement amount TPA and the total power cost TPC are also dependent variables whose values are determined by calculation using other variables. Specifically, it can be calculated by the equations (7) and (8). However, Σi means that the sum is taken for the base i. An inequality constraint of TT ≦ Σi (TVi) is defined as a constraint condition to be satisfied. There are two objective functions, total power procurement amount TPA and total power cost TPC, both of which are aimed at minimizing.
[Number 7]
TPA = Σi (RGi) + Σi (MPi) (7)
[Number 8]
TPC = Σi (RCCi) + Σi (PCi) (8)
Then, the overall control unit 31 commands the optimization calculation unit 34 to execute the optimization calculation using the optimization problem mathematical model managed by the model management unit 33. This processing content is carried out in the processing step S202 of FIG. The present invention does not limit the method of optimization calculation, but here, as an example, a processing example of using a general genetic algorithm is shown in FIG.

In the optimization method using the genetic algorithm of FIG. 10, the optimization variable is coded as a gene in the first processing step S300. In the processing step S301, a plurality of genes are generated by random numbers to form a gene pool which is a set thereof.

The evaluation value of each gene is calculated in the processing step S302. The evaluation value is calculated by adding a predetermined penalty value to the value of the objective function if the constraint condition is not satisfied. In the processing step S303, the gene having the smallest evaluation value is selected as the optimal gene.

In the processing step S304, it is confirmed that the number of genes generated after the start of optimization has reached a certain number, or that the evaluation value of the optimum gene satisfies a certain predetermined excellent condition. If it is satisfied (yes), it moves to the processing step S305 as the optimum or the end of the calculation, and outputs the optimization calculation result. This output is a combination of the amount of re-energy self-consumption, the amount of re-energy consignment utilization power at other bases, and the amount of power purchased from other power generation companies or the power trading market, which corresponds to the optimum gene and covers the drive power of each base. These are the breakdown values of, the renewable energy installed capacity of each base, the total amount of electricity procured, and the total electricity cost.

If the judgment of the processing step S304 is not satisfied (No), the process proceeds to the processing step S306 to generate a new gene by genetic processing of crossover or mutation based on the evaluation value, and genetic selection based on the evaluation value. By combining the erasure of genes that will not be left in the next generation by processing, genes that will be left as next-generation genes are generated.

In the processing step S307, a new gene pool is composed of the generated plurality of genes, the process returns to the processing of the processing step S306, and the process is repeated until the optimization calculation is completed by the determination of the processing step S304.

In the above, the re-energy power generation cost unit price and the re-energy equipment cost unit price are common values regardless of the base, but different values may be used depending on the base. Similarly, for the transmission loss rate and the consignment unit price, common values are used regardless of the consignment route (start point and end point), but different values may be used depending on the consignment route.

With this configuration, when all the drive power required for each base is covered by renewable energy, the power generated by the renewable energy power generation equipment at each base and the power generated by the renewable energy power generation equipment at other bases can be utilized by consignment. And how to combine the purchase of re-energy power from other power generation companies and the power trading market, and appropriately adjust the installed capacity of the re-energy power generation equipment installed at each base to reduce the total cost of re-energy power. You can get the effect of

DC (DCa, DCb, DCc): Data center DB1, DB2, DB3: Data management unit EM: Data center system for centrally managing energy Energy management device Nw: Information network Pw: Power network SV (SVa, SVb, SVc): Server SP: Data center system base design support device G (Ga, Gb, Gc): Re-energy power supply C (Ca, Cb, Cc): Base management device B (Ba, Bb, Bc): Power supply bus lines 11, 21, 31 : Overall control unit 12: Power supply and demand management unit 13: Task allocation adjustment unit 14: Power procurement risk management unit 15: Base information collection unit 16, 25, 35: Input / output unit 22: In-site equipment control unit 23: Task execution judgment Unit 24: Data layout adjustment unit 26: In-site status monitoring unit 32: Design condition setting unit 33: Model management unit 34: Optimization calculation unit

Claims (10)

The data centers of multiple bases and the data center system energy management device are connected by communication, and the data centers of multiple bases equipped with servers for data calculation are each equipped with a renewable energy power source so that power can be interchanged between the bases. It is a configured data center system
The data center energy management device suppresses the execution of information processing in the data center where the shortage of the renewable energy power in the base is predicted, and the information processing in the data center where the surplus of the renewable energy power in the base is predicted. A data center system that is equipped with task allocation adjustment means that adjusts information processing execution bases so as to increase the execution of data, and is characterized by adjusting task allocations in data centers at multiple bases. The data center system according to claim 1.
A data center system characterized by being provided with task allocation constraint means that increase the possibility of selecting a data center in which information related to the information processing exists when determining a data center for executing information processing. The data center system according to claim 2.
A data center system characterized by being provided with a data arrangement adjusting means for storing frequently-used information required for information processing in a data center where a surplus of renewable energy is frequently generated. The data center system according to any one of claims 1 to 3.
The task allocation is provided with a power procurement risk management means for managing the power procurement risk, which is the possibility that the actual surplus amount may be above or below the predicted value of the surplus of renewable energy power of each data center. A data center system in which the adjusting means adjusts the execution location of information processing based on the power procurement risk. The data center system according to any one of claims 1 to 4.
A data center system characterized in that a data center that receives a task request based on a task allocation determines whether or not to allow the execution of the task, and if the execution is refused, returns a selection of another data center. The data centers of multiple bases and the data center system energy management device are connected by communication, and the data centers of multiple bases equipped with servers for data calculation are each equipped with a renewable energy power source so that power can be interchanged between the bases. It is an operation method of the configured data center system.
The data center energy management device suppresses the execution of information processing in the data center where the shortage of the renewable energy power in the base is predicted, and the information processing in the data center where the surplus of the renewable energy power in the base is predicted. An operation method of a data center system characterized by adjusting the execution base of information processing so as to increase the execution of information processing. The data centers of multiple bases and the data center system energy management device are connected by communication, and the data centers of multiple bases equipped with a server for data calculation are each equipped with a renewable energy power source so that power can be interchanged between the bases. It is a data center system base design support device for supporting the base design of the configured data center system.
In order to reduce the total electric energy or total power cost of the data center system, the location of each base where the data center is located and the power generation capacity of the renewable energy power generation equipment installed at each base are set, and the energy of information processing by the region is set. A data center system base design support device characterized by calculating the total amount of renewable energy that drives the entire data center system based on the difference in efficiency and the difference in the annual power generation rate of the renewable energy power generation facility. The data center system base design support device according to claim 7.
A data center system base design support device characterized in that the total amount of renewable energy that drives the entire data center system is calculated using the transmission loss when power is interchanged between the data centers of each base. The data center system base design support device according to claim 8.
A data center system base design support device characterized by calculating the total power cost of renewable energy power that drives the entire data center system, including the charge for power transmission services when power is interchanged between data centers at each base. The data centers of multiple bases and the data center system energy management device are connected by communication, and the data centers of multiple bases equipped with a server for data calculation are each equipped with a renewable energy power source so that power can be interchanged between the bases. It is a data center system base design support method to support the base design of the configured data center system.
In order to reduce the total electric energy or total power cost of the data center system, the location of each base where the data center is located and the power generation capacity of the renewable energy power generation equipment installed at each base are set, and the energy for information processing by the region is set. A data center system base design support method characterized by calculating the total electric energy of the renewable energy power that drives the entire data center system based on the difference in efficiency and the difference in the annual power generation rate of the renewable energy power generation facility.

Images