Sunteți pe pagina 1din 39

Evaluating Conservation Voltage Reduction with WindMil

Greg Shirek
Milsoft

1.0 What is Conservation Voltage Reduction (CVR)?


CVR is the concept of lowering the utilization voltage to end-use consumers such that their
demands, and energy consumption, decreases, with the most noteworthy benefit being the demand
reduction during peak periods in which the cost of power is at its maximum due to the high cost to
run peaking generation units. Additionally, if kVAR demands are reduced this provides loss
reduction incentives as well, so typically CVR projects make use of both capacitors and voltage
regulators. This is sometimes known as VOLT/VAR optimization in the power industry. This paper
will discuss the technical side of why CVR works and also illustrate it with an example on a
distribution system by using WindMil. Economic benefits will also be calculated.

2.0 Load Mix Definitions and Equations


One must understand the nature of loads to be able to understand why conservation voltage
reduction benefits the system, or conversely worsens the situation. As explained earlier, the goal of
CVR is to reduce the amount of energy and demand that the system has to deliver to the end-use
customer, thus reducing demand and energy costs that the utility needs to pay to the power
supplier. Typically demand costs during peak periods are of a significant amount since generation
peaking units cost more to operate than base load units.

Loads are voltage dependent. The power consumed by a load is simply P=VI, so as the supply
voltage changes, the current may changes and the power and energy consumed may change, all
depending upon which load type is being addressed. The three generally accepted load types are:

1. Constant Power (PQ)


2. Constant Impedance (Z)
3. Constant Current (I)

The equations used for analyzing each of these along with numerical examples are as follows.

2.1 Constant PQ (or kVA)


S is fixed or constant for this load type.

*
S S
IL = = / = IL / (1)
V V

= line-to-neutral voltage angle

= power factor angle

pg. 1 Copyright 2011 Milsoft Utility Solutions, Inc.


S , which is the complex power is held constant while V changes during iterations

From equation (1), it can be seen that as V changes, the load current IL must change inversely to
hold S constant.

*
S
IL = S = IL*V
V

Consider an example of 1000W load rated 120V, if the voltage were lowered 10% to 108 volts, the
current would have to increase by 0.9(-1) or 1.11 p.u. as below.

1000
IL= =8.33 amps (rated current)
120

1000
IL = =9.26 amps
108

S
IL = S = IL * V 1000W = 9.26 A * 108 V
V

So this shows that constant PQ type load rules are that the load current must change inversely
to the change in delivered voltage to hold the power (PQ) constant.

2.2 Constant Impedance (Z)


2 2
V V
Z= = / = Z / (2)
S* S

For a wye connected load, the complex power S is specified at a rated line-to-neutral voltage V .
This data is then used to calculate the impedance of the load Z .

Then Z is used in the following equation to find load current IL at rated voltage.

V V
IL = = / = IL / (3)
Z Z

From equation (2) it can be seen that a Z type load varies quadratically with voltage. As an
example, if the rated voltage for a 1000 Watt load was 120 volts, the calculated Z would be:

2 2
V 120
Z= = = 14.4
S 1000

pg. 2 Copyright 2011 Milsoft Utility Solutions, Inc.


This Z value is then a constant and used in subsequent calculations. The load current IL would
then be:

V 120
IL = = = 8.33
Z 14.4

If the voltage delivered to this load were reduced by 10% to 108 volts from the 120 volt rated, the
new load current would be:

V 108
IL = = = 7.5
Z 14.4

And the new effective calculated power would be P=VI=108* 7.5 = 810 Watts

So this shows that constant impedance type load rules are that the load current changes
linearly with the change in delivered voltage and the demand varies as a squared function of
the voltage change.

2.3 Constant Current (I)


Constant Current loads are calculated according to:

*
S S
IL = = / = IL / (4)
V V

IL = IL / (5)

= line-to-neutral voltage angle

= power factor angle

IL is held constant while the angle of the voltage changes resulting in a changed angle on the
current while the power factor of the load is held and remains a constant.

From equation (4), it can be seen that the power S varies linearly with the change in voltage since
ILa remains constant. That is, for a change in voltage of X percent, the current will change X
percent.

As an example, 1000W load at 120 volt rated would give load current of:

1000
IL = =8.33 amps
120

pg. 3 Copyright 2011 Milsoft Utility Solutions, Inc.


If the voltage were lowered 10% to 108 volts, then power of the load will have to change to keep
the current constant.

S =V *IL =108 * 8.33=900 W

So this shows that constant current type load rules are that the power delivered to the load
varies linearly with the change in voltage delivered to the load.

3.0 WindMil Calculation Examples


It is important to understand how the load mix definitions affect WindMils allocation and voltage
drop results. A simple model is used to show how the change in voltage affects the allocated load
and the WindMil category called adjustments. Each load mix type is set up for a simple model with
a load control point value set to 100 kW at 100 percent power factor. The source element output
voltage is varied to show the voltage dependencies of the loads.

3.1 Constant kVA (PQ)


The consumer is set with a constant kVA (PQ) load mix type and the results are shown below. The
source output voltage is set at 120 volts. Notice that the amount of allocated load is a straight-
forward 100.0 kW.

pg. 4 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 1 Constant kVA at 1 p.u. Voltage

Also of interest are the 13.9 Amps of load current. Basically 13.89 amps multiplied by the 120 volts
equals 100 kW.

Now lower the voltage by 10 percent to 108 volts and rerun load allocation.

pg. 5 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 2 Constant kVA at 0.90 p.u. Voltage

The amount of allocated load to the consumer shown in the calculated load tab remains at 100 kW
and is not dependent upon the voltage delivered, thus proving constant PQ load mix is voltage
independent. The current had to increase by 11 percent (108/120)-1 * 13.890 amps to get the
15.433 amps to equal the constant PQ load total of 100 kW.

pg. 6 Copyright 2011 Milsoft Utility Solutions, Inc.


3.2 Constant Z
The consumer is now set with a constant Z load mix type and the substation output voltage is set at
120 volts and then lowered by 5 percent to 114 volts to see the results.

Since no losses are introduced into this circuit, allocating 100 kW of LCP load should show 100 kW
allocated to the consumer as shown below. The consumer demand is calculated at the base voltage
of 120 volts, which is also delivered to the consumer.

Figure 3 Constant Z at 1 p.u. Voltage

pg. 7 Copyright 2011 Milsoft Utility Solutions, Inc.


Now reduce the source voltage by 10 percent to 108 volts and rerun allocation.

Figure 4 Constant Z at 0.90 p.u. Voltage

Notice that the amount of allocated load is now different since the voltage delivered to the
consumer is lower than 120 volts which was the voltage used to calculate Z. This Z value is then
used subsequently in the power flow equations as shown in equation 2 during the allocation
process. The calculated load tab shows that 123.4 kW has been allocated. So why can the allocated

pg. 8 Copyright 2011 Milsoft Utility Solutions, Inc.


load be greater than the load control point kW of 100 kW? WindMil uses an adjustments category
that is populated with a value equal to the difference in kW due to this lowered voltage.

To explain numerically, whatever increase above the LCP kW added to the consumers calculated
load will have to be subtracted from the total allocated load and will be shown in the adjustments
category of the load allocation report. Note the adjustments value of -123.4 kW in the allocation
report. This was calculated as follows:

1. The Z of the load was calculated at 120V base (7200 V L-G) as described in equation 2.
2 2
V 7.20
Z= = / = 0.5184 / ohms
S* 100

Current calculated for this impedance.

V 7.2
IL = = = 13.89amps
Z 0.5184

2. The voltage delivered to the load is 108 V (6.48 kV L-G), so the current decreases

V 6.48
IL = = = 12.5 amps
Z 0.5184

3. The power delivered to the load is then

P=VI=6.48 * 12.5 = 81 kW

What occurs in WindMil is a somewhat inverse relationship. That is, WindMil assigns a
calculated load that is larger than what the truly effective load is at the consumer, effectively
meaning it would be recorded with a power meter. WindMil then subtracts this additional kW
value assigned to the consumer to the adjustments category in the allocation report.

Since WindMil allocates load based on a set LCP value (which is converted to amps at the
defined source voltage, 108 volts), in this case:

LCP kW
IL = = 15.43amps
( BusVolt / 120 )* 7.2V (7)

WindMil then assigns a calculated load at the consumer based on this amp value and the voltage
ratio factor:

pg. 9 Copyright 2011 Milsoft Utility Solutions, Inc.


BaseVoltage 7200
Calculated kW = * IL = * 15.43 amps = 123.4 kW
ConsumerVoltage 6480 (8)

4. The adjustment value is then:


= 1 = 1.234 1 100 = 23.4 (9)

This adjustment value is the reduced kW demand needed by the consumer when the voltage is
lowered.

3.3 Constant I
The allocation process for constant current is comparable to the above constant Z method, however
the allocated kW value is a linear voltage adjustment rather than the exponential adjustment for the
Z load mix. The WindMil results for the 10 percent voltage decrease are seen below.

pg. 10 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 5 Constant I at 0.9 p.u. Voltage

4.0 What Load Mixture Should I Use?


Now that the equations have been shown that relate to each type, the next question is what types of
loads are in each one of these categories and which ones should I use in my model? Loads can be
lumped into one of these three types; however other loads may be a combination of these three.

The following chart lists some common loads and their load mix type.

pg. 11 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 1 - Common Load Types

Constant Power Constant Impedance Constant Current


Motors (at rated load) Incandescent Lighting Welding Units
Power supplies Resistive water heaters Electroplating Processes
Fluorescent Lighting Electric stoves
Washing Machines Clothes Dryers

A good way to illustrate the voltage change versus demand change can be seen in the following
graph.

Figure 6 - Voltage vs. Demand for Load Mix Types

Demand vs. Voltage


1.20
1.15
1.10
kW (P.U.)

1.05
1.00 Constant PQ
0.95 Constant I
0.90
Constant Z
0.85
0.80
0.9 0.92 0.94 0.96 0.98 1.0 1.02 1.04 1.06 1.08 1.1

Voltage (P.U.)

4.1 Proven and Reasonable Percentages


It seems as though every year due to new technologies for the common household, that load mixes
are slightly changing. However, for the most part, the largest demand loads for a residential class of
end-use consumers remains the same, regardless if winter or summer peaking. Such loads are
electric heat, heat pumps, air-conditioning, forced-air heating, washers/dryers, ovens, etc. With
these types of loads, new products may have higher efficiencies and other enhancements, but
overall the load mix type they fit into does not change. Therefore reasonable documented load mix
percentages developed over the past decade or so still seem to apply.

Many references exist with these load mix percentages based on load class, winter vs. summer
peak, etc. Below is a list of some rules of thumb:

pg. 12 Copyright 2011 Milsoft Utility Solutions, Inc.


Residential summer peaking - 70/30 power/impedance
Residential summer peaking with minimal air-conditioning 30/70 power/impedance
Residential winter peaking - 30/70 power/impedance
Industrial - 80/20 power/impedance
Commercial 50/50 power/impedance (or can use 100 percent constant current)

Web searches or other industry publications should be used as guidance for setting up your load
mix percentages in WindMil.

4.2 Pacific Northwest National Laboratory Testing 1


Lab testing by Pacific Northwest National Laboratory (PNNL) showed that some common load
types for typical residential are a blended ZIP (Z=Impedance, I=Current, P=Power). A chart for
some of the loads tested with results is shown below.

This chart shows that new technologies such as plasma and LCD TVs as well as CFL light bulbs will
need to be incorporated into the total diversity of loads served in a typical household therefore
affecting the blended load mix of the entire household load. This should be noted as new system
models are created with new loads allocated in WindMil.

Table 2 Zip Values for End-Use Loads


Z% I% PQ%
FAN 73 25 2
DELL LCD Monitor -41 47 95
Sony Plasma TV -32 48 84
Clarity LCD TV -4 4 100
CFL 13 W 40 0 60
CFL 42 W 49 -38 89
CFL 20 W -1 100 1
Incandescent 75 W 58 42 0
TV CRT 0 83 18
LED TV High Quality -43 87 56
LED TV Med Quality -45 45 100
LED TV Low Quality 100 0 0
Ceiling Fan and 20 W CFL 56 51 -7
Source: Pacific Northwest National Laboratory

Overall, a blended ZIP model needs to be developed in the WindMil model and it is up to the user to
perform the research necessary to establish the load mix equipment definitions. Many references
with reasonable ZIP values per load type that have been used throughout the past thirty years are
still for the most part valid, however as the above results from PNNL indicate, new technologies
may justify some slight adjustments. The weighting factor needs to consider what percentage of the
loads contribute to the system peak kW values. For example, in the winter is electric heat a large
percentage of the load and in the summer are air conditioning units the largest load on a residential

pg. 13 Copyright 2011 Milsoft Utility Solutions, Inc.


circuit. Furthermore, have residences shifted to a larger percentage of CFL light bulbs versus the
old incandescent types. All these factors need to be addressed when establishing the ZIP models.

4.3 Motor Load Misconceptions


One may hear that motor loads should be modeled as constant kVA load mix. This is only true if
more supporting data is included with this statement. Testing has demonstrated that motor loads
may actually require less demand as their utilization voltage is decreased, somewhat representative
of constant current or impedance load types. Many times, you will hear misconceptions that
motors are to be modeled as a constant kVA (PQ type) load mix. Well, this is only partly true and is
dependent upon the motor type and where it is operating along its torque curve. The
understanding is that a motors kVA demand was constant over a typical range of utilization
voltages, for example at 0.9 to 1.1 p.u. voltage the demand would not change. Thus, decreasing the
terminal voltage supplied increased motor current. Testing has found that the kVA demand from
the motors were dependent upon the percent loading or torque of the motor.(1) See below graph:

pg. 14 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 7 Motor Power, Efficiency and Current Graphs

Still to this day, however, the assumption of constant kVA representing motor loads may be
somewhat correct as long as the motor is running at rated torque or above as shown in the graphs
on the red or green lines. However, residential motor loads for HVAC, air conditioning or heating,
will most likely be running at less than rated load, especially if the manufacturer correctly and

pg. 15 Copyright 2011 Milsoft Utility Solutions, Inc.


conservatively designed the system. Yes, there will be instances where residential motor loads will
run at or above rated output, but not normally the case. One case is where older industrial
processes motor loads may be running at or above nameplate rating if load has been added to
motors over the years. So overall defining a load mix type for motors depends on many factors such
as what type of process it is running and at what percent of nameplate rating it will be running at
over its useful life. Realistically a motors load mix definition will be constantly changing based
upon its percent loading, but having multiple mixes is not worthwhile to manage in the WindMil
equipment load mix definition, so making a sound engineering decision on a single definition is
satisfactory.

5.0 Capacitors Effects on CVR


Capacitors reduce the kVAR demand on the system, and since the total kVA demand consists of a
real and reactive component, the capacitors provide the needed kVAR to the loads rather than the
reactive power coming from the source and flowing through the distribution lines. Load amps are a
function of the total kVA, and reducing the kVAR reduces the kVA, thus reducing the amount of
current and line losses. Losses are function of current squared, so that is another important benefit
of using capacitors to deliver kVAR to loads rather than from the generation source.

In addition to reducing the amount of line losses, capacitors help to increase line voltage delivered
to the end-use consumer. Obviously this can cause an adverse effect during periods of full CVR
implementation since capacitors raise line voltage. The voltage drop through the line is a function
of the voltage drop through the line resistance as well as the drop through the line reactance, V =
IRcos+ IXsin. So reducing the line current will reduce the voltage drop, or conversely stated will
increase the voltage utilized at the loads.

Care should be taken when deciding how many capacitors should be turned on or off during times
of running CVR on the system.

6.0 WindMil Setup and Procedures


The procedure for converting applied kWh or kW billing load to the WindMil circuit model is
accomplished with load allocation. This engineering function allocates a PQ value at a user defined
physical point or location on the power system, most commonly a consumer element. Most
typically the user enters a kW (P) and kVAR (Q) value at the delivery point such as a substation and
then this load is allocated throughout all the connected elements downline. Load amp values can
also be entered as the LCP values.

Load allocation assigns a calculated load to each consumer based upon the consumers billing
information supplied as well as the consumers load mix. Therefore to see the affects of voltage
reduction on the system loading, WindMils subsequent voltage drop analyses after load allocation
will need to be set up to accommodate the voltage reduction via equipment that is capable of
transforming the voltage, most likely a regulator or a source output voltage, or could even be a
transformers input and output values to represent off-nominal taps to increase or decrease
delivered voltage.

pg. 16 Copyright 2011 Milsoft Utility Solutions, Inc.


As previously discussed, as the utilization voltage changes at various loads, the energy and demand
drawn from those loads may change based on their predefined load mix. WindMils load allocation
determines what the correct calculated load, kW and kVAR, should be for all consumers, or load
bearing elements, during the load allocation analysis. Subsequently, after those loads become
established they will be fixed for all other load flow or voltage drop analyses. In other words, once
the calculated loads are established, their values do not change.

6.1 Procedure for CVR Setup and Analysis


Running load allocation that matches metered system quantities establishes the base system and
then subsequently voltage drop analyses are run with modified voltages, as well as percentage of
peak loadings if so desired, to find demand, energy, and loss savings when implementing CVR.

The general procedure for CVR assessment in WindMil is as follows.

1. Run load allocation for existing metered peak demand at source. This will be called the
BASE SYSTEM load.
a. Make sure loads are modeled with a proper load mix
b. Make sure source, transformers rated in/out, and/or regulator voltages used in the
model have output voltage set correctly to coincide with value at time of peak, or the
kW demand or current assigned to the LCP
c. Run load allocation.
i. Note the kW and kVAR values in the ADJUSTMENT category. These are due
to the constant Z or constant I loads. Note if 100 percent constant PQ (kVA)
mix is used, there will be not adjustment values.
ii. Note the load and losses in the load allocation report
iii. Review the load allocation summary report to verify allocated kW and kVAR
results appear reasonably accurate. Pay attention to system loss values as
well.
2. Analyze benefits using Voltage Drop Analysis
a. Change the various elements with voltage settings such as source bus voltage or
nominal voltages, regulator output voltages, or transformer input/output voltages if
representing off-nominal taps or LTCs.
b. Run a Voltage Drop Analysis
c. Note the new values in the ADJUSTMENT category. If voltage reduction was
represented, the ADJUSTMENT value should have decreased (a larger negative
value). This represents the reduction in the total delivered kW and kVAR the loads
would demand from the system for the modified voltage settings.
d. Note the losses category in the Voltage Drop report. Due to the lowering of demand
delivered to the consumers, the line losses may decrease as well.
e. Lastly, note the total kW in the summary report. This is the key value to focus on as
this will be the total demand reduction due to your CVR implementation which will
be a summation of reduced end-use consumer demands, and the reduction in line
losses.

pg. 17 Copyright 2011 Milsoft Utility Solutions, Inc.


3. Use the results in economic analyses or any other calculations needed to show the benefit of
CVR. Some costs and system dynamics that may be needed or considered:
a. Peak kW demand charges from power supplier
b. Wholesale kWh rates (on and off peak)
c. Load Factors
d. Loss Factors
4. Develop action plan for system implementation of Volt/VAR optimization.

7.0 Example System Study


As with any system planning study, various criterion needs to be defined. For this following
example, just voltage criterion at the primary system will be defined to make it easier. The levels
will need to meet ANSI Range A voltage levels, which are 118-126 volts.

Consider a distribution system with the following characteristics:

5. 10,000 kW peak load


6. 5,000 kW minimum load
7. 67 miles of UG
8. 8 miles of OH
9. 147 Large Power Customers (Load Mix - 70/30 kVA/Z)
10. 389 Residential Customers (Load Mix 30/70 kVA/Z)
11. Substation Regulators set at 123 V, no LDC for Base System allocation
12. 2700 kVAR Capacitors

7.1 Load Allocation with Existing System (no LDC)


The first step is to run a load allocation to develop the Base System at peak load, or 10,000 kW.
The substation regulators are set at 123 volts with no line drop compensation.

Table 3 Existing System with Peak Load Allocation

Reviewing the above table, the Adjustments category indicates 91 kW and 35 kVAR. What is
interesting is that the adjustment values are positive. This indicates that the average voltages at the
customers are above 120 volts in which the base Z values were calculated; the constant impedance
portion of the load mix for the two customer classes is not being significantly affected because the
utilization voltages at the customers is more less above the 120 volt base.

pg. 18 Copyright 2011 Milsoft Utility Solutions, Inc.


Notice that the substation regulators are set at 123 volts output and there is also 2700 kVAR of
capacitors installed throughout the system, which will most likely cause voltages to be much higher
than the 118 volt minimum. The figure below shows the system which does not violate voltage
criteria. In fact, the lowest primary voltage is 120.8 volts (120 V Base). The voltage profile begins
at the substation regulators and ends at a far extremity at the northern part of the system, which
will be used as the focal point.

Figure 8- Existing System Peak Load Voltage Min/Max


Min Line = 120.6 volts
Max at Sub Regulator = 123.1 volts

pg. 19 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 9 Base System Voltages at Annual Peak

7.2 Evaluate Various Load Levels


For the economic analysis to be presented later, the kW adjustments will be found at varying load
levels, in ten percent increments down to 5,000 kW. The smallest peak load of 5,800 kW was
experienced in April, so the analysis reviewed kW demand savings down to 5,000 kW.

pg. 20 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 4 Monthly Peak Demands

Month Peak Demand kW


January 9,900
February 8,800
March 6,500
April 5,800
May 6,800
June 5,900
July 9,500
August 10,000
September 8,500
October 7,000
November 7,200
December 9,400

The following table shows the adjustments as well as the total kW losses that were found by
running voltage drops after the initial base allocation at each ten percent increment load level.

Table 5 Existing System Dynamics


Load level Adjustments Losses Focal Point Voltage
(kW) (kW) (kW) (120V base)
5,000 93 34 122.3
6,000 101 47 122.0
7,000 104 63 121.7
8,000 104 82 121.4
9,000 100 104 121.1
10,000 92 128 120.8

pg. 21 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 10 Load Level System Dynamics

140 122.5
120
122
100
80 121.5 End of Line
kW Voltage (120V
60 121 Base)
40
120.5
20
0 120
5,000 6,000 7,000 8,000 9,000 10,000

Load Level (kW)

Adjustments Losses End of Line Voltage

Notice the adjustments value is relatively consistent for each load level, hovering right around 100
kW. It would seem the adjustments value would become less (a larger negative value) as the load is
decreased. However, this is not the case because the substation regulators are still establishing a
123 volt output, and with reduced load on the system, the voltage delivered to the customers will
increase thus negatively affecting any demand reduction required by customers due to the constant
impedance load mix. The voltage is way too high at these light load levels; it meets criteria, but by
way too much. At the 5,000 kW level, the lowest voltage at the focal point is excessively high and
there is less than 1.0 volt drop from substation to end of line as can be seen in the voltage profile.
This signifies that there should be much lower voltage set at the substation due to only a one volt
profile range.

Also make note that the kW losses range from 34 kW at minimum load to 128 kW at maximum load.
The losses are calculated exponentially as a function of the current, I2R, so the values appear to be
reasonable.

pg. 22 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 11 - Voltage Min/Max at 5000 kW Load Level

Min Line = 122.3 volts


Max at Sub Regulator = 123.1 volts

pg. 23 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 12 System Voltages at 5,000 kW

7.3 Explore Voltage Reduction Options


Some voltage reduction measures must now be tested to determine the decrease in needed kW
purchases at the delivery point.. To see the benefits of the voltage reduction, there are more than
just a few options for consideration, with a non-inclusive list such as the following:

1. Lower Voltage Regulator Settings and/or apply LDC


2. Apply Switched of Fixed capacitor banks
3. Use off-nominal taps at the substation and/or at the distribution transformers
4. Alternative feeds or switching options
5. New Substations

pg. 24 Copyright 2011 Milsoft Utility Solutions, Inc.


There are pros and cons with each option as well as a wide range of costs to implement each. For
our example system, we will simply analyze the first option. When conducting a CVR project to
determine best measures, many options should be explored, and benefit/cost metrics are valuable
as well.

7.4 LDC at Substation Regulators


One reasonable option is to implement line drop compensation (LDC) at the substation voltage
regulators. In order to do this, we need to also ensure that at the lightest loaded period, the LDC
settings will be high enough to boost voltage to meet the 118 volts minimum level.

Rather than determining the substation voltage regulator LDCs settings manually, WindMils SET
REGULATION analysis is used. This feature has a setting to calculate the regulator settings to either
hold a high or low set voltage, as well as a most desired voltage. For our purposes, the setting to
hold lowest voltage to greater than minimum, or 118 volts, will be used. Also the most desirable
voltage is 118 volts.

Figure 13 - WindMil Set Regulation

pg. 25 Copyright 2011 Milsoft Utility Solutions, Inc.


The application results in WindMil recommending the following settings.

Table 6 Line Drop Compensation Settings


Voltage = 119V
LDC R = 5.8 V
LDC X = 3.8 V

At this point, a voltage drop analysis with these LDC settings at each load level increment results in
the following.

pg. 26 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 7 Dynamics with LDC Enabled
Load level Adjustments Losses Regulated Voltage End of Line Voltage
(kW) (kW) (kW) (120V base) (120V base)
5,000 -35 33 120.0 119.2
6,000 -53 47 120.0 118.9
7,000 -45 63 120.8 119.4
8,000 -51 82 120.8 119.1
9,000 -74 103 121.5 118.8
10,000 -57 127 121.5 119.3

The kW Adjustments category has now been greatly affected which is expected with a decrease in
voltage. The initial load allocation with the existing system at the peak load level of 10,000 kW
resulted in an adjustment value of 92 kW as was shown in Table 1. With the substation regulator
LDC settings, the value has now decreased to -56 kW. This is a net improvement of 148 kW, or
about 1.4 percent. Overall this indicates power purchases during the annual peak at the substation
level have decreased 1.4 percent.

It is valuable to calculate a demand reduction percentage based on the voltage reduction


percentage; this will be discussed later in the economic analysis. The table below illustrates the
dynamics of the existing system with no LDC and the proposed system with LDC.

A key metric shown is the CVR factor which is a measure of how effective voltage reduction is from
an energy and demand standpoint. It is simply the ratio between the percent demand (or energy)
reduction and the percent voltage reduction.

%
= (10)
%

pg. 27 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 8 Voltage, Adjustments, and CVR Factor With and Without LDC
Load Level Substation Voltage Adjustments CVR Factor
(kW) (120V base) (kW) (kW/V)
no LDC 123 93
5,000 w/ LDC 120.0 -35
% Change 2.44 2.56 1.05
no LDC 123 101
6,000 w/ LDC 120.0 -53
% Change 2.44 2.57 1.05
no LDC 123 104
7,000 w/ LDC 120.8 -45
% Change 1.79 2.13 1.19
no LDC 123 104
8,000 w/ LDC 120.8 -51
% Change 1.79 1.94 1.08
no LDC 123 100
9,000 w/ LDC 121.5 -74
% Change 1.22 1.93 1.59
no LDC 123 92
10,000 w/ LDC 121.5 -57
% Change 1.22 1.49 1.22

Notice throughout the min to max load range, the CVR Factor does not change that significantly, and
averages about 1.2.

7.5 Losses
With constant impedance loads, the line losses decrease proportional to the square of the voltage
decrease. With constant kVA or PQ type loads, the line losses will increase proportionally to the
square of the voltage decrease since the current must increase to maintain the constant kVA. Since
the sample system contained a mixture of constant Z and constant kVA load types, the system losses
were very unaffected by the small voltage change percentages due to the LDC.

The reason there is really no difference in losses between the base system with no LDC and the
system with LDC is that the total demands required by the consumers have only been reduced by a
small percentage, 1 to 2 percent, so the line current does not change much either; since line losses
are for the most part I2R, there is not much change. This indicates there is very little, if any at all,
line loss savings with such a small voltage change with the example system CVR plan.

pg. 28 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 9 Losses Pre and Post LDC Settings
Load level Losses w/o LDC Losses w/ LDC
(kW) (kW) (kW)
5,000 34 33
6,000 47 47
7,000 63 63
8,000 82 82
9,000 104 103
10,000 128 127

During peak conditions, voltage criteria are still met with these LDC settings, with the lowest
voltage being 119.3 volts in the far north part of the system, while the LDC settings caused the
voltage regulator to boost to 121.5 volts during peak load conditions.

7.6 Voltage Levels with LDC


Primary distribution system voltage levels were calculated with the LDC settings at min and max
loading to ensure voltage criteria was met. The following graphs show the results.

For the annual peak of 10,000 kW the voltage at the regulators and end of line at the far north part
of the system are 121.5 and 119.3 volts.

pg. 29 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 10 Peak Load with LDC Settings

Min Line = 119.3 volts


Max at Sub Regulator = 121.5 volts

pg. 30 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 14 Peak Load with LDC

Lastly, the minimum voltage at the end of the north feeder still meets the 118 volt minimum criteria
during the minimum load period at 5,000 kW with the LDC set. Also notice the voltage profile
shows only a 1.7 volt range.

pg. 31 Copyright 2011 Milsoft Utility Solutions, Inc.


Min Line = 118.3 volts
Max at Sub Regulator = 120.0 volts

pg. 32 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 15 - Minimum Load with LDC

8.0 Economic Analysis


As shown, reducing system voltage with a simple implementation of using LDC at the substation
regulators proves beneficial for reducing the demand requirements for end-use customers. The
next step is to apply economic analysis to the results to determine the bottom-line monetary
benefits.

pg. 33 Copyright 2011 Milsoft Utility Solutions, Inc.


8.1 Economic Factors
Previously, system loading scenarios with and without CVR were evaluated with WindMil. This was
needed to determine the demand savings at each incremental load level.

For typical planning study economic analysis, developing a loss savings approach is needed, and
usually the kW loss savings are calculated at peak for such items as load balancing a feeder, line
upgrades, new substations, etc. With that approach those peak kW demand savings are converted
to an annual loss savings by using a loss factor as shown below.

Floss_factor=0.15 Fload_factor+ 0.85Fload_factor2 (11)

This works reasonably well for figuring savings for typical system upgrades or other loss saving
measures since the equation is based upon the fact that as the load decreases in the study area, the
line current decreases as well, but by a squared factor.

However, for the CVR economics, the above equation does not pertain because as shown previously,
the demand savings at the lower load levels are still large and does not decrease as a squared
function like typical I2R line losses do. We are dealing with a savings in required kW and kWh
purchases at the delivery point rather than line losses.

The following diagrams show the kW adjustments with and without LDC. The net is the potential
savings in power purchases at those load levels if CVR is used via the LDC.

Table 11 Net kW Savings with LDC


Load level Adjustments (No LDC) Adjustments (LDC) Net Savings
(kW) (kW) (kW) (kW)
5,000 93 -35 128
6,000 101 -53 154
7,000 104 -45 149
8,000 104 -51 155
9,000 100 -74 174
10,000 92 -57 149
Average 151.5

pg. 34 Copyright 2011 Milsoft Utility Solutions, Inc.


Figure 16 Adjustments and Savings with LDC

200

150

100

kW 50

-50

-100
5,000 6,000 7,000 8,000 9,000 10,000
Load Level (kW)

Adjustments (No LDC) Adjustments (LDC) Net Savings

Of interest, is that the net savings stay within a rather small band, from 128 kW to 174 kW. Again,
the reason for this is that the LDC boosts varying amounts based on the current or load through the
regulator. Furthermore, it is not a linear function due to the mixture of constant kVA and constant Z
type loads.

Fortunately, this makes the economic analysis much simpler. An engineering judgment can be
made to use a linear relationship between each interval to find the kW savings according to the
peak demands per month.

The monthly peak demands are as follows along with the kW savings based on the linear
approximations per interval.

pg. 35 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 12 Demand Savings with Linear Approximations
Month Peak Demand kW kW savings
January 9900 151.5
February 8800 170.2
March 6500 146.0
April 5800 148.8
May 6800 150.0
June 5900 151.4
July 9500 161.5
August 10000 149.0
September 8500 164.5
October 7000 149.0
November 7200 150.2
December 9400 164.0

8.2 Demand Savings


The costs used in the economics are as follows:

Table 13 Utility Rates


Peak Demand Costs Per Month
Spring (March, April, May) $5/kW
Summer (June, July, August) $10/kW
Fall (September, October) $6/kW
Winter (Nov., Dec., Jan., Feb.) $8/kW
Wholesale Energy Cost $0.06/kWh
Retail Energy Cost $0.10/kWh

Since the utility incurs a monthly peak demand charge, it is beneficial to implement the LDC
settings on the substation regulators during what could be the anticipated peak period each month.
As shown earlier, the kW demand reduction can be calculated for given monthly peak from
historical data according to the difference in kW with and without LDC.

Very simply, the peak demand cost savings are calculated by:

= (12)

Using this equation, the system demand reduction benefit per month is listed below and totals
$13,811 in savings.

pg. 36 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 14 Net Demand Reduction Benefit
Month Peak Demand kW kW savings Demand Savings
January 9,900 151.5 $ 1,212
February 8,800 170.2 $ 1,362
March 6,500 146.0 $ 730
April 5,800 148.8 $ 744
May 6,800 150.0 $ 750
June 5,900 151.4 $ 1,514
July 9,500 161.5 $ 1,615
August 10,000 149.0 $ 1,490
September 8,500 164.5 $ 987
October 7,000 149.0 $ 894
November 7,200 150.2 $ 1,202
December 9,400 164.0 $ 1,312
Total $ 13,811

8.3 Lost kWh Sales


As the voltage is reduced, the demand is reduced for constant Z and constant current loads as has
been shown. Counter intuitively this can have an adverse affect on kWh sales for the utility. As
with this example system, there is approximately a 150 kW reduction at the substation during each
interval from 5,000 MW to 10,000 MW when using LDC at the regulators. This obviously helps with
demand charge savings, but has a negative effect on kWh sales.

Since the wholesale power charges to the utility are less than the retail energy charges for the end-
use consumers, reducing the voltage will reduce the number of kWh sales during CVR
implementation. For this example, the kWh revenue for the utility per kWh is simply the difference
between the wholesale and retail rates, $0.10 - $0.06, or $0.04.

Therefore, the longer CVR is employed on the system the more potential kWh sales are lost. To
calculate those values, simply use the following formula:

= (13)

Consider the utility uses CVR two times per month for six hours each time, or twelve total hours per
month. For January, that would amount to:

= . $. $. = $

Using this same procedure per all twelve months results in the following:

pg. 37 Copyright 2011 Milsoft Utility Solutions, Inc.


Table 15 Annual Lost Revenue Sales with CVR
Month Peak Demand kW Hours of CVR kW savings Lost Energy Lost kWh sales
January 9,900 12 151.5 1,818 $ 73
February 8,800 12 170.2 2,042 $ 82
March 6,500 12 146.0 1,752 $ 70
April 5,800 12 148.8 1,786 $ 71
May 6,800 12 150.0 1,800 $ 72
June 5,900 12 151.4 1,817 $ 73
July 9,500 12 161.5 1,938 $ 78
August 10,000 12 149.0 1,788 $ 72
September 8,500 12 164.5 1,974 $ 79
October 7,000 12 149.0 1,788 $ 72
November 7,200 12 150.2 1,802 $ 72
December 9,400 12 164.0 1,968 $ 79
Total $ 891

The lost sales metric is a very minor part of the overall economics, but is a simple calculation and
may want to be included depending upon the amount of detail needed.

8.4 Total Annual Cost Benefit


The overall annual cost savings of $12,920 due to CVR is reflected in the following table.

Table 16 Annual Savings with CVR


Month Peak Demand kW kW savings Demand Savings Lost kWh sales
January 9900 151.5 $ 1,212 $ 73
February 8800 170.2 $ 1,362 $ 82
March 6500 146.0 $ 730 $ 70
April 5800 148.8 $ 744 $ 71
May 6800 150.0 $ 750 $ 72
June 5900 151.4 $ 1,514 $ 73
July 9500 161.5 $ 1,615 $ 78
August 10000 149.0 $ 1,490 $ 72
September 8500 164.5 $ 987 $ 79
October 7000 149.0 $ 894 $ 72
November 7200 150.2 $ 1,202 $ 72
December 9400 164.0 $ 1,312 $ 79
Total $ 13,811 $ 891
Net Savings $ 12,920

9.0 Summary
As demonstrated, one must be aware of how various customer classes and the load mix type
assigned to them affect the load allocation and voltage drop analyses in WindMil. For conservation
voltage reduction studies, this is crucial. Simply using WindMils default load mix type which is set

pg. 38 Copyright 2011 Milsoft Utility Solutions, Inc.


as 1 p.u. kVA will show adverse affects when studying CVR projects. Many users are unaware of
how load mix affects the engineering results and should figure out how to set them appropriately.
One must make good engineering judgment when establishing load mix values. One good option is
to establish the ZIP percentages to each rate class and assign an equipment definition for each.

Keep in mind the effect on kW demand requirements as voltage changes for the three load mix
options.

1. Constant kVA: Power delivered remains constant


2. Constant Current: Power delivered decreases proportionally to the voltage decrease
3. Constant Impedance: Power delivered decreases proportional to the square of the decrease
in voltage.

The example system studied simply made us of more aware of the capabilities of step voltage
regulators. Too often, utilities may set the output voltage at a higher level thus ensuring the ANSI
voltage ranges are maintained at each customer on the substation, but, as demonstrated, this has
adverse benefits for potential kW demand savings during lightly loaded periods.

Regulating line voltage via regulators with LDC in conjunction with capacitors, either fixed or
switched, proves to be one of the most, if not the most, promising actions for Volt/VAR projects
from a cost benefit standpoint. The payback is very short with the LDC implementation since the
only cost is the labor and time for the engineering or operations personnel to develop the correct
value and program them into the controls. There are basically no material costs in this option.

Also as shown, the line loss savings by reducing the voltage and thus load requirements has a very
insignificant benefit compared to the delivered demand reduction.

10.0 References
Schneider, Fuller, Tuffner, Singh, Evaluation of Conservation Voltage Reduction on a National
Level, Prepared by Pacific Northwest National Laboratory for the U.S. Department of Energy, July
2010.

Tom Wilson, Measurement and Verification of Distribution Voltage Optimization Results for the
IEEE Power & Energy Society, presented at the 2010 PES General Meeting, July 2010.

American Public Power Association, Distribution System Performance Improvement Guide, March
1997.

Rob Ardis, Peak Shaving via Voltage Reduction, Presented at the CRN Summit, June 2010.

pg. 39 Copyright 2011 Milsoft Utility Solutions, Inc.

S-ar putea să vă placă și