Method for evaluating power system operation reliability considering data center load flexibility

Abstract

The application discloses a data center energy efficiency optimization method considering multi-period device coupling, and a power system operation reliability evaluation method considering data center load flexibility, and comprises the following steps: 1) establishing a data center load flexibility model; 2) establishing a power system operation reliability index system considering data center load flexibility; 3) proposing a power system operation reliability evaluation model and method considering data center load flexibility, and obtaining power system operation reliability indexes. The application establishes a power system operation reliability evaluation model considering data center load flexibility, comprehensively analyzes the influence of the data center on the power system operation reliability on the basis of the data center load flexibility model, and establishes a power system operation reliability evaluation model considering data center load flexibility with the total cost of the power system operation as an objective function.

Description

Technical Field

[0001] This invention relates to the field of power system operation reliability assessment, specifically a power system operation reliability assessment method that takes into account the load flexibility of data centers. Background Technology

[0002] With the widespread application of technologies such as the Internet of Things, cloud computing, and artificial intelligence, my country's data centers have developed rapidly. As of 2021, my country had the second largest hyperscale data center capacity after the United States, accounting for 15% of the world's total. The high energy consumption of data centers has also attracted increasing attention. In 2021, the total electricity consumption of data centers nationwide was 225.6 billion kilowatt-hours, accounting for approximately 2.99% of the total social electricity consumption, which is 2.18 times the power generation of the Three Gorges Dam that year. It is projected that by 2025, the total electricity consumption of data centers nationwide will reach 395.2 billion kilowatt-hours, accounting for approximately 4.10% of the total social electricity consumption.

[0003] Data centers are not only energy-intensive, centrally distributed, and easy to control, but they also possess strong energy flexibility. Furthermore, relying on cloud interconnection and data center network (DCN) technology, they can achieve interconnection between data centers and the remote transfer of computing loads, altering the load distribution in the power grid and realizing the integration of "data-energy" dual networks. This adds more flexibility to the power system operation, alleviates transmission line congestion during peak hours, and improves the overall reliability of the system. With the increasing proportion of renewable energy and the growing randomness and volatility of the power grid, data centers can effectively play a role in demand-side management, maintaining the safe and stable operation of the power system. - .

[0004] With the large-scale integration of data centers into the power grid, their high energy consumption, high elasticity, and interconnected energy characteristics will bring unprecedented challenges and opportunities to the safe and reliable operation of both themselves and the power grid. On the one hand, data centers typically have large fluctuations in computing load, thus requiring redundant hardware configurations. For example, e-commerce platforms configure their data centers according to the computing power demands of "Double 11" to cope with peak computing loads, resulting in high equipment redundancy, low utilization, high energy consumption, and strong fluctuations in computing load, posing challenges to the reliable operation of the power grid. On the other hand, low utilization also means a large space for energy consumption adjustment. Large telecommunications operators and internet companies often deploy data centers in different regions, and can achieve the remote transfer of their computing load through DCN (Distributed Data Network), which can change the load distribution in the power network (under the condition that the operating constraints allowed by the hardware equipment are followed, the remote transfer of computing load can be achieved instantly without incurring additional costs), providing a huge opportunity to maintain the safe and stable operation of the power grid.

[0005] While existing power system reliability assessment methods are mature for conventional power systems, data centers exhibit significant energy flexibility, a steadily increasing proportion of electricity consumption, and highly interconnected energy characteristics. In scenarios involving large-scale data center grid integration, these high energy consumption, high flexibility, and interconnectivity can lead to inaccurate results from existing power system reliability assessment methods. Current indicator systems cannot reflect the energy efficiency status of data centers and their impact on grid reliability. Furthermore, there is currently no literature researching power system reliability assessment methods that consider the load flexibility of data centers. Therefore, to ensure the safe and reliable operation of power systems, it is necessary to establish power system reliability assessment methods that take into account the load flexibility of data centers. Summary of the Invention

[0006] The purpose of this invention is to provide a method for assessing the operational reliability of power systems that takes into account the load flexibility of data centers, comprising the following steps:

[0007] 1) Establish an energy consumption model for data center equipment;

[0008] The data center equipment energy consumption model includes a single data center computing power balance equation considering the characteristics of computing power load transfer across different locations, a total computing power balance equation for multiple data centers, upper and lower limits of computing power constraints for a single data center, and upper and lower limits of computing power demand transfer constraints between data centers.

[0009] The computing power balance equation for a single data center considering the characteristics of computing load migration:

[0010]

[0011] In the formula: u kl (t) represents the computing power demand transferred from data center k to data center l during time period t; D IT,k (t) represents the computing power requirement of data center k during time period t.

[0012] Overall computing power balance equation for multiple data centers:

[0013]

[0014] Where: N DC This represents the total number of data centers.

[0015] Upper and lower limits of computing power constraints for a single data center:

[0016] u total,k,min ≤u total,k (t)≤u total,k,max (3)

[0017] In the formula: u total,k,max and u total,k,minThese represent the upper and lower limits of the total computing power of all IT devices in data center k. The upper and lower limits of computing power demand transfer constraints between data centers are also included.

[0018] -u kl,max ≤u kl (t)≤u kl,max (4)

[0019] In the formula: u kl,max This represents the upper limit of the computing power demand transfer between data centers k and l.

[0020] 2) Establish a power system operation reliability assessment index system that takes into account the load flexibility of data centers;

[0021] The power system operation reliability assessment index system that takes into account the load flexibility of data centers includes system-level operation reliability indexes, data center energy efficiency, and data center contribution indexes to power system operation reliability.

[0022] a. System-level operational reliability indicators

[0023] The indicators for expected demand not supplied (EDNS, MW), probability of load curtailments (PLC), expected wind power abandoned (EWPA, MW), and expected solar power abandoned (ESPA, MW) are as follows:

[0024]

[0025]

[0026]

[0027]

[0028] In the formula: P(x) is the probability of system state x; G E G W G S C represents the set of states where load reduction, wind curtailment, and solar curtailment occur; E (x), ΔP W (x) and ΔP S (x) represents the load reduction, wind curtailment, and solar curtailment for system state x, respectively.

[0029] b. Data Center Energy Efficiency Indicators

[0030] Average Power Usage Effectiveness (APUE) for time period t:

[0031]

[0032] In the formula: P DC,k (t) and P IT,k (t) represents the total energy consumption of data center k and the total energy consumption of IT equipment during time period t, respectively.

[0033] c. Data center contribution indicators to power system operational reliability

[0034] Contribution coefficients of data center to EDNS / PLC / EWPA / ESPA, CDED / CDPL / CDEW / CDES: (This refers to the expected power shortage, load shedding probability, expected wind curtailment, and expected solar curtailment.)

[0035] CDED=(EDNS0-EDNS1) / EDNS0 (10)

[0036] CDPL=(PLC0-PLC1) / PLC0 (11)

[0037] CDEW=(EWPA0-EWPA1) / EWPA0 (12)

[0038] CDES=(ESPA0-ESPA1) / ESPA0 (13)

[0039] In the formula: the subscripts 0 and 1 of each system-level indicator represent not considering and considering data center load flexibility, respectively.

[0040] 3) Establish a power system operation reliability assessment model that takes into account the load flexibility of data centers;

[0041] The power system operation reliability assessment model that considers data center load flexibility includes an objective function as follows:

[0042]

[0043] Where: N b P represents the total number of nodes in the power grid. D,j (t), ΔP W,j (t), ΔP S,j (t) and C E,j(t) represents the active load, wind curtailment, solar curtailment, and load reduction of node j in time period t, respectively; a, b, c, and d are the load electricity cost coefficient, wind curtailment cost coefficient, solar curtailment cost coefficient, and load reduction cost coefficient, respectively; Δt is the interval time period.

[0044] The constraints include data center load flexibility constraints, data center operation constraints, and power grid operation constraints.

[0045] a. Data center load flexibility constraints:

[0046] The computing power balance equation for a single data center considering the characteristics of computing load migration:

[0047]

[0048] In the formula: u kl (t) represents the computing power demand transferred from data center k to data center l during time period t; D IT,k (t) represents the computing power requirement of data center k during time period t.

[0049] Overall computing power balance equation for multiple data centers:

[0050]

[0051] Where: N DC This represents the total number of data centers.

[0052] Upper and lower limits of computing power constraints for a single data center:

[0053] u total,k,min ≤u total,k (t)≤u total,k,max (17)

[0054] In the formula: u total,k,max and u total,k,min These represent the upper and lower limits of the total computing power of all IT devices in data center k. The upper and lower limits of computing power demand transfer constraints between data centers are also included.

[0055] -u kl,max ≤u kl (t)≤u kl,max (18)

[0056] In the formula: u kl,max This represents the upper limit of the computing power demand transfer between data centers k and l.

[0057] b. Data center operational constraints

[0058] The power ramp-up value of IT equipment should be less than its upper limit:

[0059]

[0060] In the formula: R represents the power consumption of IT device i during time period t when i is in the startup state; IT This represents the upper limit of the climbing slope.

[0061] The shortest time required for IT equipment to go from a shutdown state to a power-on state:

[0062] T IT-off,i (t)≥T IT-off,imin (20)

[0063] In the formula: T IT-off,i (t) represents the downtime of IT device i; T IT-off,imin This is the minimum downtime, typically 1 to 2 minutes.

[0064] Energy conservation constraint:

[0065] T IT,i (t) = 97%P IT,i (t)R in +T in,i (t) (21)

[0066] T in,i (t)=T out (t)+DP IT,i (t) (22)

[0067] In the formula: R in The equivalent thermal resistance (K / W) for convective heat transfer at the IT equipment inlet; D is the heat transfer coefficient; T out (t) represents the supply air temperature of the air conditioning system; 97% P IT,i (t) represents the heat generated by IT device i; T in,i (t) represents the inlet airflow temperature of IT device i.

[0068] The chip temperature of IT device i should be lower than its upper limit:

[0069] T IT,i (t)≤T IT,max (twenty three)

[0070] In the formula: T IT,max It is the upper limit of chip temperature.

[0071] IT equipment inlet airflow temperature constraints:

[0072] T in,min ≤T in,i (t)≤T in,max (twenty four)

[0073] In the formula: T in,max and T in,minThese are the upper and lower limits of the inlet airflow temperature for IT equipment, and are an internationally recognized important reference for data center design.

[0074] The supply air temperature of the air conditioning system in data center i should meet its upper and lower limits:

[0075] T out,min ≤T out,i (t)≤T out,max (25)

[0076] In the formula: T out,max and T out,min These are the upper and lower limits of the air supply temperature of the air conditioning system, and their values ​​depend on the operating parameters of the air conditioner.

[0077] c. Power grid operation constraints

[0078] Active and reactive power balance constraints in power systems:

[0079]

[0080]

[0081] In the formula: P G,i (t), Q G,i (t) represents the active and reactive power outputs of the generator unit at node i in the power system during time period t; P D,i (t), Q D,i (t) represents the active and reactive loads of node i in the power system during time period t; V i (t) is the voltage amplitude at node i during time period t; θ ij (t) is the voltage phase angle difference between nodes i and j in the power system during time period t; G ij and B ij These are the real and imaginary parts of the element in the i-th row and j-th column of the nodal admittance matrix, respectively.

[0082] Shear load C E,j (t), wind curtailment volume ΔP W,m (t) and waste light amount ΔP S,n (t) should satisfy their respective upper and lower bound constraints:

[0083] 0≤C E,j (t)≤P D,j (t), j = 1, 2, ..., N b (28)

[0084] 0≤ΔP W,m (t)≤P W,m (t), m=1,2,...,N w (29)

[0085] 0≤ΔP S,n(t)≤P S,n (t), n=1,2,...,N s (30)

[0086] In the formula: P W,m (t) and P S,n (t) represents the predicted active power generated by wind farm m and photovoltaic power station n, respectively; N w and N s These are the number of wind farms and photovoltaic power plants, respectively.

[0087] Power system node voltage amplitude V j (t) and unit output P G,i (t) should satisfy their respective upper and lower bound constraints:

[0088] V j,min ≤V j (t)≤V j,max j = 1, 2, ..., N b (31)

[0089] P G,i,min ≤P G,i (t)≤P G,i,max i = 1, 2, ..., N g (32)

[0090] In the formula: the subscripts max and min represent its upper and lower limits, respectively.

[0091] 4) Propose a power system operation reliability assessment method that considers data center load flexibility.

[0092] The power system operation reliability assessment method that takes into account data center load flexibility includes the following specific steps:

[0093] a. State Sampling: Using the Monte Carlo method to sample the state, load, and data center computing load of system components (generator sets, transmission lines, transformers, etc.). i (t), ambient temperature coefficient k T (t) Active power P of the wind farm W,m (t) and the active power P of the photovoltaic power station S,n A system state is determined by sampling random variables such as (t).

[0094] b. Topology Analysis and Power Flow Calculation: Connectivity identification is performed on the sampled system state, analyzing how many subsystems are formed by the system's nodes connected by branches. Power flow calculations are performed on the power system to determine if there are problems such as node voltage exceeding limits or line exceeding limits.

[0095] c. Minimum load shedding calculation: For systems with power flow issues, perform minimum load shedding calculations considering data center load flexibility, and calculate the load shedding amount C. E,j (t), wind curtailment volume ΔP W,m (t) and waste light amount ΔP S,n (t).

[0096] d. Calculation of operational reliability indicators: Based on the minimum load shedding solution, calculate the power system operational reliability indicators EDNS, PLC, EWPA, ESPA, APUE, CDED, CDPL, CDEW, and CDES, taking into account the load flexibility of data centers.

[0097] e. Convergence judgment: The maximum variance coefficient β of the operational reliability index is used as the convergence criterion. If the convergence accuracy is met, the calculation ends; otherwise, the next sampling is performed and the process returns to step 1).

[0098] It is worth noting that the technical solution of this invention first establishes the energy consumption characteristics and heat exchange coupling relationship of data center IT and air conditioning equipment, and establishes individual energy consumption and heat exchange models for IT and air conditioning equipment. Then, it introduces inter-time period coupling relationships to establish a data center energy efficiency optimization model based on multi-time period equipment coupling, achieving multi-time period energy efficiency optimization of the data center. Based on look-ahead decoupling and equipment aggregation algorithms, a multi-time period data center energy efficiency optimization solution method is proposed, providing theoretical support for data center energy efficiency management system-level modeling and operational optimization.

[0099] The technical effects of this invention are undeniable, and its beneficial effects are as follows:

[0100] This invention proposes a data center load flexibility model, which is based on the interconnection and interoperability between data centers and the characteristics of remote transfer of computing load.

[0101] This invention proposes a power system operation reliability assessment index system that includes indicators such as the energy efficiency and operational reliability contribution of data centers. Addressing the problem that existing power system operation reliability assessment index systems cannot fully reflect the energy efficiency status of data centers and their impact on power system operation reliability, this invention proposes a power system operation reliability assessment index system that considers the load flexibility of data centers from both the power system and data center perspectives. This system is more comprehensive and more suitable for scenarios involving large-scale data center grid connections.

[0102] This invention proposes a power system operation reliability assessment model and algorithm that considers the load flexibility of data centers. Addressing the problem that existing power system operation reliability assessment models and methods often fail to consider the load flexibility of data centers in scenarios involving large-scale data center grid connection, leading to inaccurate results, this invention proposes a power system operation reliability assessment model and method that considers the load flexibility of data centers, based on the proposed data center load flexibility model and operation reliability indicators, and outputs grid operation reliability indicators.

[0103] This invention has strong potential for industrial applications. Existing power system operational reliability assessment models and methods often fail to consider the load flexibility of data centers, leading to inaccurate results. This method, based on the interconnectivity between data centers and the characteristics of off-site computing load transfer, comprehensively considers the data center load flexibility model and proposes a power system operational reliability assessment model and method that takes into account the load flexibility of data centers, which has significant practical implications. Attached Figure Description

[0104] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0105] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0106] Example 1:

[0107] See Figure 1 A method for assessing the operational reliability of power systems that considers data center load flexibility includes the following steps:

[0108] 1) State Sampling: Using the Monte Carlo method to sample the state, load, and data center computing load of system components (generator sets, transmission lines, transformers, etc.). i (t), ambient temperature coefficient k T (t) Active power P of the wind farm W,m (t) and the active power P of the photovoltaic power station S,n A system state is determined by sampling random variables such as (t).

[0109] 2) Topology Analysis and Power Flow Calculation: Connectivity is identified from the sampled system state, analyzing how many subsystems are formed by the system's nodes connected by branches. Power flow calculations are performed on the power system to determine if there are issues such as node voltage exceeding limits or line exceeding limits.

[0110]

[0111]

[0112] In the formula: P G,i (t), Q G,i (t) represents the active and reactive power outputs of the generator unit at node i in the power system during time period t; P D,i (t), Q D,i (t) represents the active and reactive loads of node i in the power system during time period t; V i (t) is the voltage amplitude at node i during time period t; θ ij (t) is the voltage phase angle difference between nodes i and j in the power system during time period t; G ij and B ij These are the real and imaginary parts of the element in the i-th row and j-th column of the nodal admittance matrix, respectively.

[0113] 3) Minimum load shedding calculation: For systems with power flow issues, perform minimum load shedding calculations considering data center load flexibility. This is because the IT device i's state variable I... i Given the existence of (t), this problem is a mixed integer optimization problem. It is solved using the branch and bound method and an improved PSG (partial surrogate cuts) algorithm to calculate the load shedding amount C. E,j (t), wind curtailment volume ΔP W,m (t) and waste light amount ΔP S,n (t).

[0114] The objective function is to minimize the total operating cost C of the power system, which mainly includes the cost of electricity consumption by the load, the cost of wind curtailment, the cost of solar curtailment, and the cost of load shedding.

[0115]

[0116] Where: N b P represents the total number of nodes in the power grid. D,j (t), ΔP W,j (t), ΔP S,j (t) and C E,j (t) represents the active load, wind curtailment, solar curtailment, and load reduction of node j in time period t, respectively; a, b, c, and d are the load electricity cost coefficient, wind curtailment cost coefficient, solar curtailment cost coefficient, and load reduction cost coefficient, respectively; Δt is the interval time period.

[0117] Constraints

[0118] a. Data center load flexibility constraints

[0119] Energy consumption of a single IT device

[0120] The energy consumption of the i-th IT device during time period t can be expressed as:

[0121]

[0122] In the formula: I i (t) represents the state variable of the i-th IT device during time period t. i (t) takes values ​​of 2, 1, and 0, representing the three states of on, standby, and off, respectively; P sleep P represents the standby power of IT equipment. idle,i P represents the static power of the i-th IT device. max,i u represents the full-load power of the i-th IT device; i (t) represents the computing power utilization rate of the i-th IT device during time period t; b1 and b2 are leakage current constants.

[26] ;T IT,i (t) represents the chip temperature of the i-th IT device during time period t.

[0123] Energy consumption of air conditioning system in a single data center

[0124] The energy consumption of the air conditioning system during time period t can be expressed as:

[0125]

[0126] In the formula: Q IT This refers to the heat dissipation of IT equipment (most IT equipment manufacturers can provide information on the power consumption and heat dissipation of their IT equipment; heat dissipation accounts for approximately 97% of the power consumption).

[24] );k COP It is the coefficient of performance (COP) of the air conditioning system; k T The ambient temperature coefficient (kW / m) 2 );S dc This refers to the area of ​​the computer room.

[0127] In summary, the total energy consumption P of a single data center DC Total computing power u total for:

[0128]

[0129] Where: N IT This represents the total number of IT devices in the data center.

[0130] Computing power balance equation for a single data center considering the characteristics of computing power load migration

[0131] After the computing power demand shifts during time period t, the computing power of a single data center should equal the computing power demand.

[0132]

[0133] In the formula: ukl (t) represents the computing power demand transferred from data center k to data center l during time period t; D IT,k (t) represents the computing power requirement of data center k during time period t.

[0134] Overall computing power balance equation of multiple data centers

[0135] The total computing power of all data centers during time period t should be equal to the total computing power demand.

[0136]

[0137] Where: N DC This represents the total number of data centers.

[0138] Upper and lower limits of computing power for a single data center

[0139] The total computing power of IT equipment in a single data center during time period t should meet its upper and lower limits:

[0140] u total,k,min ≤u total,k (t)≤u total,k,max (9)

[0141] In the formula: u total,k,max and u total,k,min These represent the upper and lower limits of the total computing power of all IT devices in data center k. The upper and lower limits of computing power demand transfer constraints between data centers are also included.

[0142] -u kl,max ≤u kl (t)≤u kl,max (10)

[0143] b. Data center operational constraints

[0144] The power ramp-up value of IT equipment should be less than its upper limit:

[0145]

[0146] In the formula: R represents the power consumption of IT device i during time period t when i is in the startup state; IT This represents the upper limit of the climbing slope.

[0147] The shortest time required for IT equipment to go from a shutdown state to a power-on state:

[0148] T IT-off,i (t)≥T IT-off,imin (12)

[0149] In the formula: T IT-off,i (t) represents the downtime of IT device i; T IT-off,imin This is the minimum downtime, typically 1 to 2 minutes.

[0150] Data center operation must comply with the law of conservation of energy. When IT equipment reaches thermal equilibrium, the sum of the heat emitted by the equipment and the heat of the incoming cold air equals the heat of the outgoing hot air.

[0151] T IT,i (t) = 97%P IT,i (t)R in +T in,i (t) (13)

[0152] T in,i (t)=T out (t)+DP IT,i (t) (14)

[0153] In the formula: R in The equivalent thermal resistance (K / W) for convective heat transfer at the IT equipment inlet; D is the heat transfer coefficient; T out (t) represents the supply air temperature of the air conditioning system; 97% P IT,i (t) represents the heat generated by IT device i; T in,i (t) represents the inlet airflow temperature of IT device i.

[0154] The chip temperature of IT device i should be lower than its upper limit:

[0155] T IT,i (t)≤T IT,max (15)

[0156] In the formula: T IT,max It is the upper limit of chip temperature.

[0157] ASHRAE (Institute of Refrigeration and Air-Conditioning Engineers) guidelines stipulate that the inlet airflow temperature of IT equipment should meet its upper and lower limits:

[0158] T in,min ≤T in,i (t)≤T in,max (16)

[0159] In the formula: T in,max and T in,min These are the upper and lower limits of the inlet airflow temperature for IT equipment, and are an internationally recognized important reference for data center design.

[0160] The supply air temperature of the air conditioning system in data center i should meet its upper and lower limits:

[0161] T out,min ≤T out,i (t)≤T out,max (17)

[0162] In the formula: T out,max and Tout,min These are the upper and lower limits of the air supply temperature of the air conditioning system, and their values ​​depend on the operating parameters of the air conditioner.

[0163] ③ Power grid operation constraints

[0164] Active and reactive power balance constraints in power systems:

[0165]

[0166]

[0167] In the formula: P G,i (t), Q G,i (t) represents the active and reactive power outputs of the generator unit at node i in the power system during time period t; P D,i (t), Q D,i (t) represents the active and reactive loads of node i in the power system during time period t; V i (t) is the voltage amplitude at node i during time period t; θ ij (t) is the voltage phase angle difference between nodes i and j in the power system during time period t; G ij and B ij These are the real and imaginary parts of the element in the i-th row and j-th column of the nodal admittance matrix, respectively.

[0168] Shear load C E,j (t), wind curtailment volume ΔP W,m (t) and waste light amount ΔP S,n (t) should satisfy their respective upper and lower bound constraints:

[0169] 0≤C E,j (t)≤P D,j (t), j = 1, 2, ..., N b (20)

[0170] 0≤ΔP W,m (t)≤P W,m (t), m=1,2,...,N w (twenty one)

[0171] 0≤ΔP S,n (t)≤P S,n (t), n=1,2,...,N s (twenty two)

[0172] In the formula: P W,m (t) and P S,n (t) represents the predicted active power generated by wind farm m and photovoltaic power station n, respectively; N w and N s These are the number of wind farms and photovoltaic power plants, respectively.

[0173] Power system node voltage amplitude V j (t) and unit output P G,i (t) should satisfy their respective upper and lower bound constraints:

[0174] V j,min ≤V j (t)≤V j,max, j = 1, 2, ..., N b (twenty three)

[0175] P G,i,min ≤P G,i (t)≤P G,i,max i = 1, 2, ..., N g (twenty four)

[0176] In the formula: the subscripts max and min represent its upper and lower limits, respectively.

[0177] 4) Calculation of operational reliability indicators: Based on the minimum load shedding solution, calculate the power system operational reliability indicators EDNS, PLC, EWPA, ESPA, APUE, CDED, CDPL, CDEW and CDES that take into account the load flexibility of data centers.

[0178]

[0179]

[0180]

[0181]

[0182] In the formula: P(x) is the probability of system state x; G E G W G S C represents the set of states where load reduction, wind curtailment, and solar curtailment occur; E (x), ΔP W (x) and ΔP S (x) represents the load reduction, wind curtailment, and solar curtailment for system state x, respectively.

[0183]

[0184] In the formula: P DC,k (t) and P IT,k (t) represents the total energy consumption of data center k and the total energy consumption of IT equipment during time period t, respectively.

[0185] CDED=(EDNS0-EDNS1) / EDNS0 (30)

[0186] CDPL=(PLC0-PLC1) / PLC0 (31)

[0187] CDEW=(EWPA0-EWPA1) / EWPA0 (32)

[0188] CDES=(ESPA0-ESPA1) / ESPA0 (33)

[0189] In the formula: the subscripts 0 and 1 of each system-level indicator represent not considering and considering data center load flexibility, respectively.

[0190] 5) Convergence judgment: The maximum variance coefficient β of the operational reliability index is used as the convergence criterion. If the convergence accuracy is met, the calculation ends; otherwise, the next sampling is performed and the process returns to step 1).

Claims

1. A method for assessing the operational reliability of a power system considering the load flexibility of data centers, characterized in that, Includes the following steps: 1) Establish a data center load flexibility model; 2) Establish a power system operation reliability assessment index system that takes into account the load flexibility of data centers; 3) A power system operation reliability assessment model considering data center load flexibility was established, and a corresponding assessment method was proposed; The power system operation reliability assessment model established in step 3) that considers the flexibility of data center load has the following objective function: (14) Where: N b P represents the total number of nodes in the power grid. D,j (t), ∆P W,j (t), ∆P S,j (t) and C E,j (t) represents the active load, wind curtailment, solar curtailment, and load shedding at node j during time period t; a, b, c, and d are the load electricity cost coefficient, wind curtailment cost coefficient, solar curtailment cost coefficient, and load reduction cost coefficient, respectively; Δt is the interval time period; The power system operation reliability assessment model established in step 3) considering the data center load flexibility includes constraints such as data center load flexibility constraints, data center operation constraints, and power grid operation constraints. The data center load flexibility constraint refers to: The computing power balance equation for a single data center considering the characteristics of computing load migration: (15) In the formula: u kl (t) represents the computing power demand transferred from data center k to data center l during time period t; D IT,k (t) represents the computing power requirement of data center k during time period t; Overall computing power balance equation for multiple data centers: (16) Where: N DC Total number of data centers; Upper and lower limits of computing power constraints for a single data center: (17) In the formula: u total,k,max and u total,k,min These represent the upper and lower limits of the total computing power of all IT devices in data center k, respectively. Upper and lower limits of computing power demand transfer between data centers (18) In the formula: u kl,max This represents the upper limit for the transfer of computing power demand between data centers k and l; The data center operational constraints refer to: The power ramp-up value of IT equipment should be less than its upper limit: (19) In the formula: R represents the power consumption of IT device i during time period t when i is in the startup state; IT This is the upper limit of the climbing slope; The shortest time required for IT equipment to go from a shutdown state to a power-on state: (20) In the formula: For IT equipment downtime; The minimum downtime is 1 to 2 minutes; Energy conservation constraint: (21) (22) In the formula: R in The equivalent thermal resistance for convective heat transfer at the IT equipment inlet; D is the heat transfer coefficient; T out (t) represents the supply air temperature of the air conditioning system; 97%P IT,i (t) represents the heat generated by IT device i; T in,i (t) represents the inlet airflow temperature of IT device i; The chip temperature of IT device i should be lower than its upper limit: (23) In the formula: T IT,max This is the upper limit of the chip's temperature; IT equipment inlet airflow temperature constraints: (24) In the formula: T in,max and T in,min These are the upper and lower limits of the inlet airflow temperature for IT equipment; The supply air temperature of the air conditioning system in data center i should meet its upper and lower limits: (25) In the formula: T out,max and T out,min These are the upper and lower limits of the air supply temperature of the air conditioning system, and their values ​​depend on the operating parameters of the air conditioner. The power grid operation constraints refer to: Active and reactive power balance constraints in power systems: (26) (27) In the formula: P G,i (t), Q G,i (t) represents the active and reactive power outputs of the generator unit at node i in the power system during time period t; P D,i (t), Q D,i (t) represents the active and reactive loads of node i in the power system during time period t; V i (t) is the voltage amplitude at node i during time period t; θ ij (t) is the voltage phase angle difference between nodes i and j in the power system during time period t; G ij and B ij Let be the real and imaginary parts of the element in the i-th row and j-th column of the nodal admittance matrix, respectively. Shear load C E,j (t), wind curtailment volume ΔP W,m (t) and waste light amount ΔP S,n (t) should satisfy their respective upper and lower bound constraints: (28) (29) (30) In the formula: P W,m (t) and P S,n (t) represents the predicted active power generated by wind farm m and photovoltaic power station n, respectively; N w and N s These are the numbers of wind farms and photovoltaic power plants, respectively. Power system node voltage amplitude With the active power output of the generator unit Their respective upper and lower bound constraints should be satisfied: (31) (32) In the formula: the subscripts max and min represent its upper and lower limits, respectively.

2. The power system operation reliability assessment method considering data center load flexibility according to claim 1, characterized in that, The data center load flexibility model established in step 1) includes data center interconnection and computing load transfer characteristics.

3. The power system operation reliability assessment method considering data center load flexibility according to claim 1, characterized in that, The data center load flexibility model established in step 1) also includes a single data center computing power balance equation that considers the characteristics of computing power load transfer across different locations, a total computing power balance equation for multiple data centers, upper and lower limits of computing power constraints for a single data center, and upper and lower limits of computing power demand transfer between data centers.

4. The power system operation reliability assessment method considering data center load flexibility according to claim 3, characterized in that, The computing power balance equation for a single data center considering the characteristics of computing load migration: (1) In the formula: u kl (t) represents the computing power demand transferred from data center k to data center l during time period t; D IT,k (t) represents the computing power requirement of data center k during time period t; Overall computing power balance equation for multiple data centers: (2) Where: N DC Total number of data centers; Upper and lower limits of computing power constraints for a single data center: (3) In the formula: u total,k,max and u total,k,min These represent the upper and lower limits of the total computing power of all IT devices in data center k, respectively. Upper and lower limits of computing power demand transfer between data centers (4) In the formula: u kl,max This represents the upper limit of the computing power demand transfer between data centers k and l.

5. The power system operation reliability assessment method considering data center load flexibility according to claim 1, characterized in that, The power system operation reliability assessment index system established in step 2) that takes into account the load flexibility of data centers includes system-level operation reliability indexes, data center energy efficiency, and data center contribution indexes to power system operation reliability.

6. The power system operation reliability assessment method considering data center load flexibility according to claim 5, characterized in that, The system-level operational reliability index refers to: The following indicators are considered: Expected Demand Not Supplied (EDDNS, MW), Probability of Load Curtailments (PLC), Expected Wind Power Abandoned (EWPA, MW), and Expected Solar Power Abandoned (ESA, MW): (5) (6) (7) (8) In the formula: P(x) is the probability of system state x; G E G W G S C represents the set of states where load reduction, wind curtailment, and solar curtailment occur; E (x), ∆P W (x) and ∆P S (x) represents the load reduction, wind curtailment, and solar curtailment for system state x, respectively; The energy efficiency of the data center refers to: Average Power Usage Effectiveness (APUE) for time period t: (9) In the formula: P DC,k (t) and P IT,k (t) represents the total energy consumption of data center k and the total energy consumption of IT equipment during time period t, respectively; The contribution indicator of the data center to the reliability of power system operation refers to: Contribution coefficients of data center to EDNS / PLC / EWPA / ESA, CDED / CDPL / CDEW / CDES: (This is a calculation of the data center's contribution to system power shortages, load shedding probability, wind curtailment, and solar power curtailment.) (10) (11) (12) (13) In the formula: the subscripts 0 and 1 of each system-level indicator represent not considering and considering data center load flexibility, respectively.

7. The power system operation reliability assessment method considering data center load flexibility according to claim 6, characterized in that, The specific steps are as follows: 1) State sampling: Using the Monte Carlo method to sample the state of system components, load, and data center computing power load. i (t), ambient temperature coefficient k T (t) Predicted active power P generated by wind farm m W,m (t) and the predicted active power P generated by the photovoltaic power station n. S,n (t) is sampled to determine a system state; 2) Topology analysis and power flow calculation: Connectivity identification is performed on the sampled system state, the number of subsystems connected by branches of the system nodes is analyzed, power flow calculation is performed on the power system, and it is determined whether there are node voltage over-limit or line over-limit problems in the system. 3) Minimum load shedding calculation: For systems with power flow issues, perform minimum load shedding calculations considering data center load flexibility, and calculate the load shedding amount C. E,j (t), wind curtailment volume ΔP W,m (t) and waste light amount ΔP S,n (t); 4) Calculation of operational reliability indicators: Based on the minimum load shedding solution, calculate the power system operational reliability indicators EDNS, PLC, EWPA, ESPA, APUE, CDED, CDPL, CDEW and CDES, taking into account the load flexibility of data centers. 5) Convergence judgment: The maximum variance coefficient β of the operational reliability index is used as the convergence criterion. If the convergence accuracy is met, the calculation ends; otherwise, the next sampling is performed and the process returns to step 1).

Images