ECE 422/522 Power System Operations & Planning/ Power Systems Analysis II 3 Load Modeling

Transcription

1 ECE 422/522 Power System Operations & Planning/ Power Systems Analysis II 3 Load Modeling Spring 2014 Instructor: Kai Sun 1

2 References 1. Load Performance for Dynamic Performance Analysis, IEEE Committee Report, IEEE Trans. on Power Systems, Vol. 8, No. 2, pp , May Bibliography on Load Models for Power Flow and Dynamic Performance Simulation, Vol. 10, No. 1, pp , Feb Standard Load Models for Power Flow and Dynamic Performance Simulation, IEEE Trans. on Power Systems, Vol. 10, No. 2, pp , Aug P. Pourbeik and B. Agrawal, A hybrid model for representing air-conditioner compressor motor behavior in power system studies, in Proc. IEEE PES General Meeting, Jul EPRI Report: Load Modeling for Power Flow and Transient Stability Computer Studies, Vol. 2: Load-Modeling Reference Manual, Product ID: EL-5003-CCMV2, EPRI Report: Measurement-Based Load Modeling, Product ID: ,

3 Background The representation of loads has not received as much attention as other components in power systems, but it may have significant impacts on stability analysis results. Without motor load dynamics modeled With motor load dynamics modeled (Source: B. Sapkota, et al, Dynamic VAr Planning in a Large power System Using Trajectory Sensitivities, IEEE Trans. Power Systems, v25, n1, Feb, 2010) 3

4 (Source : Measurement-Based Load Modeling, EPRI Report, Product ID: , 2006) 4

5 Background (cont d) Accurate load modeling is difficult because: At high-voltage levels, loads must be aggregated for stability studies A power system has a large number of diverse load components Ownerships and locations of load devices are not directly accessible to the electricity utility There is no precise information on load composition, which changes with time of day and week, seasons and weather. Load characteristics are uncertain, particularly for large voltage or frequency variations 5

6 Basic Concepts 6

7 Bus Load A portion of the actual system, which is not explicitly represented in the system model, but treated as a single power-consuming device connected to a bus in the system model Therefore in this context, load includes connected load devices together with some or all of the following: Substation step-down transformers Sub-transmission feeders Primary distribution feeders Distributions transformers Secondary distribution feeders Shunt capacitors Voltage regulators Customer wiring, transformers and capacitors If we want to represent the load accurately, the elements listed above must be accounted for. To a large extent, this depends on what is and is not represented in the system model. For bulk power system studies, much of the sub-transmission as well as the distribution system are omitted. 7

8 Load Characteristics: A set of parameters or functions, e.g. power factor, P(V,f) and Q(V,f), describing behaviors of a specified load. This could be applied to either a specific load device or an aggregation of multiple load devices (e.g. a load component, a load class and the total bus load) Load Component: The aggregate equivalent of all load devices of similar types, e.g. heater, air conditioner, lighting, etc. Load Composition: Fractional composition of the load by Load Components Could be fraction applied to a Bus Load or to a specific Load Class As an example, composition for an internet data center could be 40~50% computer servers, 40~50% HVAC loads and 10~15% lighting loads Load Class: A category of load, e.g. residential, commercial or industrial. For load modeling purposes, it is useful to group loads into several classes, which each have similar load composition and characteristics Load class mix: Fractional composition of the Bus Load by Load Class. 8

9 Load Bus Load Class Mix, e.g. 80% residential + 20% commercial Load Class, e.g. Residential Commercial Industrial Load Composition e.g. 20% lighting + 40% heating + 40% AC Load Component, e.g. Heater load Air Conditioner load Lighting load Load Characteristics e.g. Power factor P(V,f), Q(V,f) Motor Load Device, e.g. Heater Air Conditioner Light 9

10 Load Model Load Model represents the relationship between P&Q of the load, bus voltage magnitude and frequency Static Load Models: P & Q at t expressed as P(V(t), f(t)) and Q(V(t), f(t)), i.e. algebraic functions of bus voltage magnitude V and frequency f. Used to represent static loads like resistive and lighting loads, and as an approximation to dynamic loads. Dynamic Load Models: P & Q at t expressed as functions of V and f at past instants in time and at t. Usually represented by difference or differential equations. 10

11 Static Load Models Polynomial load model (ZIP Model): P 0, Q 0 and V 0 are rated values or the values at the initial operating condition V = V V 0 P = P 0 p 1 V 2 + p 2 V + p 3 Q = Q 0 (q 1 V 2 + q 2 V + q 3 ) It is composed of constant impedance (Z), constant current (I) and constant power (P/Q) components. Exponential load model P = P 0 V a Q = Q 0 V b With a (or b) = 0, 1 or 2, the model represents constant power, constant current and constant impedance, respectively. Around V 0, P V a Q V b Usually, a=0.5~1.8 and b=1.5~6 11

12 Frequency Dependency With a frequency deviation f = f f 0 P = P 0 V a 1 + K pp f Q = Q 0 V b (1 + K qf f) Typically, K pf = P/ f= 0~3.0, K qf = Q/ f= -2.0~0 P = P 0 (p 1 V 2 + p 2 V + p 3 ) (1 + K pp f) Q = Q 0 (q 1 V 2 + q 2 V + q 3 ) (1 + K qq f) Frequency of the bus voltage is not a state variable in the system model for stability analysis. It can be computed by taking the numerical derivative of the voltage phase angle. Comprehensive static model: P = P 0 (P ZZZ + P EEE + P EE2 ) 12

13 Static Load Models a b K pf K qf a b K pf K qf P = P 0 V a 1 + K pp f Q = Q 0 V b (1 + K qq f) 13

14 a b P = P 0 V a 1 + K pp f Q = Q 0 V b (1 + K qq f) 14 14

15 Common Practice in Stability Studies In the absence of information on load characteristics: When V is around V 0 a=1: P as constant current b=2: Q as constant impedance When V is below a threshold (e.g. 0.6~0.7V 0 ), convert the load to constant impedance to avoid computational problems See the snapshots on TSAT 15

16 Induction Motors Motor loads consume 60-70% of the total energy supplied by a power system Many industrial motors have starters with electromagnetically held contactors For 2.3-4kV motors, contactors drop at 65-75% of rated voltage For 460V or lower motors, contactors drop at 55-65% of rated voltage Motors could drop 70% of system load during a fault or during voltage excursions following the fault For sending-end generators, this will impact stability since lost loads will increase acceleration. For receiving-end generators, this will improve stability See Kundur s section 7.2 for modeling of induction motors 16

17 Air-Conditioner Compressor Motors Up to 50% of system load in summer Referred to as a prone to stall motor with low inertia. Similar to a standard induction motor under normal operating conditions It stalls if the voltage is low (e.g. 50~65% of the nominal voltage for 3 cycles) due to a fault Once stalled, it behaves like a constant impedance load (with a low power factor) drawing 2-3 times of the rated current. The thermal overload protection disconnects the motor if it remains stalled for considerable amount of time (typically, 15s if stalled at 50% voltage) It recovers if voltage goes back to 70% of the nominal voltage. That delays the voltage recovery. In areas with a high percentage of air conditioner loads, delayed voltage recovery, short-term voltage stability and fast voltage collapse are main stability concerns. 17

18 Southern California Edison has experienced delayed voltage recovery problems due to its high percentage of air conditioner loads. The high load currents and VAR demands caused up to 30s delays for the voltage to recover following the fault clearing operation 18

19 Discharge Lighting Loads Mercury vapor, sodium vapor, etc. About 20% of the load in commercial areas Extinguishes at voltages of 70~80% of rated voltage This causes load to drop following fault clearing. That will improve or impact stability depending on the location When the voltage recovers, they restart after 1-2 sec delay 19

20 Thermostat-Controlled Loads Space heaters, water heaters, molding and packaging machines, soldering machines, etc. Short term: constant resistance. Long term: constant power with a suitable time constant 20

21 Approaches for Load Modeling Load aggregation can be performed in two ways Bottom-up (theoretical aggregation): Analytically, by lumping similar loads based on the load type and then using pre-determined values for each parameter of the load Referred to as component-based approach Top-down (identification): Selecting a load model structure and then performing parameter estimation using an appropriate identification technique based on field measurements Referred to as measurement-based approach 21

22 Component-based Approach for Load Modeling The load supplied at a bulk power delivery point is categorized into load classes and further into load components Study the characteristics of individual appliances in detail and aggregate individual loads to produce a composite load mode. 22

23 Measurement-based Approach for Load Modeling Load characteristics are measured at representative substations and feeders at selected times of the day and season. Steady-state P/ V, P/ f, Q/ V and Q/ f Dynamic load-voltage characteristics These are used to extrapolate the parameters of loads throughout the system 23

24 Data Collection 24

25 Measurement-based Load Modeling by EPRI [6] 25

26 Model Parameters for Identification Slip s= ω s ω r ω s 26

27 Field Test 27

28 Results 28