ECE 5984: Power Distribution System Analysis

Lecture 2: Load Allocation across a Feeder

Reference: Textbook, Chapter 2
Instructor: V. Kekatos

1
Motivation for allocating load

Facts
•  In transmission systems, load forecasting is pretty accurate (law of large numbers)
•  In distribution systems, load exhibits higher variability

Problem
•  Study variations of individual and aggregate loads, in particular their max values

Why?
•  To size equipment (transformers) and study voltage drops

2
Individual customer load

•  Demand: load (kW, kVA, kVAR, A) averaged over a time period (e.g., 15 min)

example: the 15-min demand ending at 6:30 pm is 4.75 kW

•  Individual loads vary significantly; interested in the maximum

•  Characterize variability for transformer sizing; load allocation; voltage drops
3
Average and maximum demand

•  Collect demands {dn,t }Tt=1 across time for customer n

XT
1
•  Average demand d˜n := dn,t example : d˜1 = 2.46 kW
T t=1
<latexit sha1_base64="x5wsFQ0W8L/+Y6tVMcASJ9ZBy9I=">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</latexit>

•  Maximum demand d¯n := max{dn,t } example : d¯1 = 6.19 kW at 13:15

t
<latexit sha1_base64="eEVtCYxHrDv6YiEGW0eP2HizzUA=">AAACI3icZZDNThsxEMe9fBQaaAn02MuKqBIHFK3b8imhIvXCESRCkEgUzTqTxIrtXdmziGi1vA1Pg3qp4MCh74I3WSQoI1nz84zHM/OPUyUdRdFTMDe/sPhhafljbWX10+e1+vrGhUsyK7AlEpXYyxgcKmmwRZIUXqYWQccK2/H4d5lvX6N1MjHnNEmxq2Fo5EAKIB/q1X91CG8oxxvQqcLi8LYTg837RY8f7Tb5we0sPW4XFQG9EP9xyHeKXr0RNaOphe+BV9BglZ326n86/URkGg0JBc5d8Silbg6WpPAD1DqZwxTEGIZ45dGARtfNp4sW4Tcf6YeDxPpjKJxGX1fkoJ0GGm177yY6Lj2N9PYQTXn1H5RZ97YJDfa7uTRpRmjErMcgUyElYSlY2JcWBamJBxBW+jFDMQILgrysNb8//3/b93DxvcmjJj/72TjmlRLL7CvbZFuMsz12zE7YKWsxwe7YPXtgj8FdcB/8DR5mT+eCquYLe2PBv2dESKXB</latexit>

<latexit sha1_base64="rpgqjKcwJotPZ2A11WoCQRci+Ao=">AAACA3icZVBNS8QwEE39dv2qevRgcBE8LEsrgiIIghePCq4u2FLSNLsbTNKSTMWl9Li/Zm+iR3+CP8B/Y7rbg+sOhHl5M5PJe3EmuAHP+3EWFpeWV1bX1hsbm1vbO+7u3qNJc01Zh6Yi1d2YGCa4Yh3gIFg304zIWLCn+OWmqj+9Mm14qh5gmLFQkr7iPU4JWCpyD4OY6CIpI3V5FUjyFhVQ4qBIokK1oAzKyG16bW8SeB74NWiiOu4i9ztIUppLpoAKYsyz72UQFkQDp4KVjSA3LCP0hfTZs4WKSGbCYiKkxMeWSXAv1fYowBP270RBpJEEBi2bzVDGVYaBbPWZqq72gapqZpdA7yIsuMpyYIpOd/RygSHFlSE44ZpREEMLCNXcfhPTAdGEgrWtYfX7/9XOg8fTtu+1/fuz5rVfO7GGDtAROkE+OkfX6BbdoQ6iaITG6AN9OiNn7Lw7H9PWBaee2Ucz4Xz9AqoamBM=</latexit>

•  How to calculate total energy consumed?

4
Load factor and demand factor

d˜n
•  Load factor LFn := ¯  1 [why?]
dn

d˜1 2.46 kW
example : LF1 = ¯ = = 0.40
d1 6.19 kW

LF=1 is ideal for better utilization of facilities

•  Demand factor
maximum customer demand 6.19 kW
Demand factor := = = 0.18  1
sum of device ratings 35 kW
percentage of electrical devices that are on when maximum demand occurs

Characterized load for one customer. How about aggregated load in a distribution
transformer serving 5-50 customers?

5
Load surveys

•  We are metering energy monthly (billing); but do not know user maximum
•  Interested in maxima to determine transformer ratings

d¯n ' ↵ · d˜n +

<latexit sha1_base64="EyAmX+IV+J30jNxMeVqe3VL+poA=">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</latexit>

•  Install meters over some customers and use linear regression to find a linear
transformation between energy consumption and maximum demand
6
 Customer #1
Load at distribution transformer

Customer #2

Customer #3

Diversified demand: sum of customer demands

Customer #4
N
X
d= dn
n=1

7
Load duration curve

•  Sort diversified demand to evaluate transformer stress

•  Each bar corresponds to 15min/24h=1.04%

•  Example: 22% of the time, the transformer serves more than 12 kW

8
Maximum demand for load aggregations

•  Maximum diversified demand (max of sum)

(N )
X
d¯ := max{dt } = max dn,t example : d¯ = 16.16 kW at 17:30
t t
n=1

•  Maximum non-coincident diversified demand (sum of max)

N
X N
X
dˆ := d¯n = max{dn,t } example : dˆ = 6.18 + 6.82 + 4.93 + 7.05 = 24.98 kW
t
n=1 n=1

Per-customer maxima do not necessarily occur at the same time …

dˆ dˆ
•  Diversity factor DF := ¯ 1 [why?] example : DF = ¯ = 1.55
d d

•  Load diversity = dˆ <latexit sha1_base64="4fjyj3e75Lwecw8AfoN7SUJFldE=">AAAB9nicZVDLSsNAFJ34rPUV69JNsAguaklE0I1ScOOygn1AE8pkMmmGTiZh5kZaQn+lO6lLf8QP8G+ctFlYe2E4Z86dO/fe46ecKbDtH2Nre2d3b79yUD08Oj45Nc9qXZVkktAOSXgi+z5WlDNBO8CA034qKY59Tnv++LnI996pVCwRbzBNqRfjkWAhIxi0NDRrj26EIQ9mN66PpcYne2jW7aa9DGuTOCWpozLaQ/PbDRKSxVQA4VipgWOn4OVYAiOczqpupmiKyRiP6EBTgWOqvHw5+8y60kpghYnUR4C1VP9W5DhWMYaooVFNY79AiOLGiIriqj8osmq9CYQPXs5EmgEVZNUjzLgFiVV4YAVMUgJ8qgkmkukxLRJhiQlop6p6f+f/tpuke9t07KbzeldvOaUTFXSBLtE1ctA9aqEX1EYdRNAEzdECfRoTY258GIvV0y2jrDlHa2F8/QIUWJIU</latexit>

d¯ > 0 example : 24.98 16.16 = 8.82 kW

Why do we care about the max non-coincident diversified demand?

9
Diversity factors
•  Maximum diversified demand is hard to measure

•  We usually know max demand per customer

(metered or via linear regression from monthly energy)

•  How to translate from dˆ ! d?

¯

•  Calculate diversity factors experimentally

§  record d¯ at specific network locations
§  record {d¯n } for all customers downstream these locations

DFN

N 10
Recap
1.  For each customer, either directly meter max demand d¯n , or use linear
regression from customer’s total demand d˜n
<latexit sha1_base64="k0uaCk9VoaLCVEXWuU4+98+QzvE=">AAAB63icZVBNSwMxEM3Wr1q/qh69LBbBQym7Iuix4MVjBfsh7VKy2WkbmmSXZFYoS39Fb1KP/hx/gP/GbLsHax+EeXmTycybMBHcoOf9OKWd3b39g/Jh5ej45PSsen7RMXGqGbRZLGLdC6kBwRW0kaOAXqKBylBAN5w+5fnuO2jDY/WKswQCSceKjzijaKW3QUh1Fs2HaliteQ1vBXeb+AWpkQKtYfV7EMUslaCQCWpM3/cSDDKqkTMB88ogNZBQNqVj6FuqqAQTZKuB5+6NVSJ3FGt7FLor9W9FRqWRFCd1G81MhnnEiayPQeVX+0GeNZtNcPQYZFwlKYJi6x6jVLgYu7lxN+IaGIqZJZRpbsd02YRqytCup2L9+//dbpPOXcP3Gv7Lfa3pF5sokytyTW6JTx5IkzyTFmkTRiRZkCX5dKSzcD6c5fppySlqLskGnK9fp0yOiQ==</latexit>

<latexit sha1_base64="b8ovXiyISZXm+r08DQ+dGBuV8Fs=">AAAB7XicZVBNSwMxEM36WetX1aOXxSJ4KGVXBD0WvHisYD+gXUo2O9uGZpMlmRXK0p/Rm9Sjv8Yf4L8x2+7B2gdhXt5kMvMmTAU36Hk/zs7u3v7BYeWoenxyenZeu7jsGpVpBh2mhNL9kBoQXEIHOQropxpoEgrohdPnIt97B224km84SyFI6FjymDOKVhoMkYsI8mg+kqNa3Wt6K7jbxC9JnZRoj2rfw0ixLAGJTFBjBr6XYpBTjZwJmFeHmYGUsikdw8BSSRMwQb4aee7eWiVyY6Xtkeiu1L8VOU1MQnHSsNHMkrCIOEkaY5DF1X5QZM1mE4yfgpzLNEOQbN0jzoSLyi2suxHXwFDMLKFMczumyyZUU4Z2QVXr3//vdpt075u+1/RfH+otv9xEhVyTG3JHfPJIWuSFtEmHMKLIgizJp6OchfPhLNdPd5yy5opswPn6BUQYj3o=</latexit>

2.  Sum per-customer maxima to find

N
X
d ˆN = d¯n
<latexit sha1_base64="B/OBTkIwCMWNwOENqsjcMLfQCoQ=">AAACA3icZVDLSsNAFJ3UV62vqksXBovgopREBN0UCm5cSQX7gCaWyXTSDJ2ZhJkboYQu+zXdSV36CX6Af+Ok7cLaC8M999zX3BMknGlwnB+rsLW9s7tX3C8dHB4dn5RPz9o6ThWhLRLzWHUDrClnkraAAafdRFEsAk47wegxz3feqdIslq8wTqgv8FCykBEMhuqXL70IQzaYvD3XPZ2Kfibrrgm8ACvD9k1Fxak5C7M3gbsCFbSyZr/87Q1ikgoqgXCsdc91EvAzrIARTiclL9U0wWSEh7RnoMSCaj9bHDKxrw0zsMNYmSfBXrB/OzIstMAQVY3XYxHkHiJRHVKZh2ZAntXrSyB88DMmkxSoJMsdYcptiO1cEHvAFCXAxwZgopj5pk0irDABI1vJ3O/+v3YTtG9rrlNzX+4qDXelRBFdoCt0g1x0jxroCTVRCxE0RTM0R5/W1JpZH9Z8WVqwVj3naM2sr1+TT5gE</latexit>
n=1

3.  Use experimentally estimated DFN to find max diversified demand

¯N dˆN
d =
DFN
<latexit sha1_base64="zQ9ZxLKpM6CL56rRDawxo7FwU88=">AAACC3icZVDLSsNAFJ3UV62vqks3wSK6KCURQTeCoIirUsE+wNQymd60g5NJmLkRS8gn9Gu6k7p06Qf4N07aLnxcGO65586dO+f4seAaHefLKiwtr6yuFddLG5tb2zvl3b2WjhLFoMkiEamOTzUILqGJHAV0YgU09AW0/efrvN9+AaV5JB9wFEM3pAPJA84oGqpXPvZ8qtJ+9lS/9AJFWeoNKc7qLPUQXjG9uc169axXrjg1Zxb2f+AuQIUsotErf3r9iCUhSGSCav3oOjF2U6qQMwFZyUs0xJQ90wE8GihpCLqbzgRl9pFh+nYQKXMk2jP250RKQx1SHFZN1qPQzzMOw+oAZF6aB/Ku/r0Eg4tuymWcIEg23xEkwsbIzo2x+1wBQzEygDLFzTdtNqTGEjT2lYx+96/a/6B1WnOdmnt/VrlyF04UyQE5JCfEJefkityRBmkSRsZkQqbk3RpbE+vNms6vFqzFzD75FdbHN3cWm9s=</latexit>

11
Load allocation using diversity factors
Example 2.1: Analyze a single-phase lateral.
Given: a) monthly energy usage per customer; assume 0.9 PF lagging
b) load survey shows that d¯n = 0.2 + 0.008 · Monthly energy [kWh]
c) diversity factors (see graph on previous slide)
Wanted: max diversified demand for each transformer and line segment

103.22 kVA 80.77 kVA 54.4 kVA

33.7 kVA 39.4 kVA 54.4 kVA

Transformer ratings:
T1: 25 kVA
T2: 37.5 kVA
T3: 50 kVA
KCL load does not apply with max aggregated (real or apparent) powers!
12
Voltage drop using diversity factors
Example: Find voltage at each transformer assuming VN1=2.4 kV

5, 000 ft 500 ft 750 ft

103.22 kVA 80.77 kVA 54.4 kVA
1 mile = 5, 280 ft
33.7 kVA 39.4 kVA 54.4 kVA

232.15 V [3.27%] 233.1 V [2.92%] 232.7 V [3.04%]

zline = 0.3 + j0.6 ⌦/mile

T1 : 25 kVA, 2400 240 V, Z = 1.8\40 %

T2 : 37.5 kVA, 2400 240 V, Z = 1.9\45 %
T3 : 50 kVA, 2400 240 V, Z = 2.0\50 %

13
Transformer load management
•  The method of diversity factors requires knowing
d1) DFN’s for different N
d2) customer maxima; and
d3) customer assignment to transformers

•  Transformer load management is a simpler but less accurate method

–  relies on the fact that transformers are metered in greater detail

•  Training stage
-  Collect historical data from transformers
-  Fit a linear regression model fitting monthly energy served to peak demand

•  Operational stage
-  Knowing d3) and customer bills, find total energy to be served
-  Using the trained model, predict max demand

14
Allocation factors

Distribute maximum feeder demand on transformers based on their rating

kVATk
d¯Tk = d¯feeder ⇥ P allocation factor
m kVATm

Example: Re-allocate load assuming maximum demand at feeder is 92.9 kW

22.94 kVA 34.41 kVA 45.88 kVA

Much easier to implement; more conservative; but relies on reliable ratings

15
Voltage drop using allocation factors
Example: Find voltage at each transformer using allocation factors

92.9 + j45 kVA 72.3 + j35 kVA 41.3 + j20 kVA

21 + j10 kVA 31 + j15 kVA 41.3 + j20 kVA

234.0\ 0.6 V 233.7\ 0.8 V 233.5\ 0.9 V

•  Due to linearity of allocation factors, KCL now holds for line segments

•  Results obtained are similar; this method may be less conservative

•  Note small voltage angles

Does this method yield the actual quantities assuming demands are correct?
16
Feeder load
•  Smoother demand curve
•  Define statistics similar to transformers (maximum diversified demand)

17
Summary
•  Allocate load to size transformer and study voltage drops

linear
X DFN
•  Method 1: diversity factors d˜n ! d¯n ! dˆTk = d¯n ! d¯Tk
regression
<latexit sha1_base64="FkLtPCQU0gUYFJ0fQENKHqZPZzA=">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</latexit>
n2Tk

X linear
•  Method 2: transformer load management d˜n ! d˜n ! d¯Tk
regression
<latexit sha1_base64="0pJTaMv4pebYGXcWUCJo7nfRfio=">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</latexit>
n2Tk

kVATk
•  Method 3: allocation factors d¯Tk = P d¯feeder
m kVATm
<latexit sha1_base64="qRIDU41CROdZFgxlPoLCdCVIhs0=">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</latexit>

•  Requirements for every method

•  M3 features linearity in flows; none of the methods considers losses

18