SELECTION CRITERIA AND APPLICATION OF A 16,000 HP, 4-POLE INDUCTION MOTOR FOR A COMPRESSOR DRIVE

Copyright Material IEEE Paper No. PCIC-98-18

Kenneth Clarkson Member, IEEE (Formerly Messer - AGS) Air Products & Chemicals Calvert City, KY USA

Tim Trumbo Member, IEEE ABB Industrial Systems Inc. North Brunswick. NJ

Lukas Kueng Member, IEEE ABB lndustrie AG CH-5242 Birr

USA

Abstract - Historically, many users have established a somewhat arbitrary line in the range of 5,000 HP to 10,000 HP as the point to switch from an induction motor to a synchronous motor design. This paper covers the selection process of a 16,000 HP, 4-pole, Induction Motor driving a centrifugal compressor. Starting considerations, oscillating torque, power factor, mechanical & electrical design issues, first cost, and life- cycle costs are taken into account.

Discussion points will include the perspective of the User, Air Separation OEM and Motor Manufacturer. We see a growing .trend toward ever-larger horsepower induction motor applications. The intent of this paper is to provide criteria to be examined during motor specification.

1.0 Introduction

In December 1996, Messer-AGS began engineering and procurement of equipment for a 900 metric ton-per-day Air Separation Unit (ASU) to be located near Kuala Lumpur, Malaysia. Principle products of this facility are industrial gases via pipeline to the customer, liquid products for the area merchant market and a cylinder filling operation for these products for local use. The primary user of these products is a large steel manufacturing facility adjacent to the ASU. The ASU is owned and operated by a joint venture, with Messer having partial ownership. The project schedule was aggressive with partial start-up initially scheduled for March 1998 (16 months).

8 8 275W. 50 Hr. 3 RUSE 1

I fi----- 33W. 51 Hz. 3 RUSE I I I

- 1

-6 16000 HP

Switzerland

2.0 Electrical System Constraints

The electrical system is illustrated in Figure 1. The ASU customer is supplying power from a 33kV substation, by contract. A 3 kilometer 33kV feeder was installed by the ASU. The initial conditions given to the motor vendors are listed below:

14,500 BHP (nominal rating 16,000HP) Quadratic curve to 30% FLT, 100% speed WK2 = 58,000 (lb-f?) Altitude (~1000 m) Ambient conditions: 40C 32C cooling water 11 kV, 50Hz 1500 rpm 1500MVA min. three-phase SCC 5% drop at 33kV bus 80% Starting Terminal Voltage 20MVA, Captive Transformer 33kV/ll kV Power Net Present Value (NPV) $lOOO/kW

3.0 Design Requirements

In order to supply firm performance data and pricing, the minimum data required by motor manufacturers is not always available. This is due to user resources, insufficient schedule or incomplete understanding of machine design issues. An increasingly popular trend in the Air Separation Industry is the use of standardized plant designs, where an identical plant is used in several locations around the country, or the world. Within limits of altitude, ambient conditions, and process requirements, the design of the major mechanical components, i.e., the cold box and compressors, are not affected by location. Conversely, the design and cost of the electrical system is greatly affected by the physical location, which alters the electrical supply to the plant. Table 3.1 is a list of the minimum requirements for basic design to begin. This table applies to induction and synchronous machines.

Figure 1: System One-Line Diagram 0-7803-4897-4/98/$10.00 8 1998 IEEE

- 165 - 98CH36234

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

e Horsepower e Voltage

Frequency e Speed m Ambient Conditions (Temperature, Altitude, Safe Area) e Applicable Standards (NEMA, IEC, API) e Load Speed-Torque Characteristic during Starting e Load Inertia e Minimum Network Short Circuit Capacity e Maximum Allowable Network Voltage Dip

MVAsYSEM Motor Rating Motor Inrush M V ~ O T O R (During Start) Voltage Dip

Table 1: Minimum Data Requirements

Example 1 Example 2

200 MVA 100 MVA 4 MVA (4000 kVA) 500% 500% 4 5.0 = 20 MVA 20 I100 = 10%

4 MVA (4000 kVA)

4 5.0 = 20 MVA 20 I200 = 20%

Regardless of specialty, everyone involved understands the ,reason for design differences between a plant supplied at 275kV and one at 13.8kV. They also understand a 10,000 HP induction machine has a very different design than a 1,000 HP induction machine. However, many don't understand why the design and cost changes for the same machine rating between two projects. Specifically, a motor's design is affected by two independent sets of variables, one set associated with the load.and a second set associated with the electrical system.

Two variables related to the load are ones most familiar to non-electrical engineers, the horsepower and speed. There never seems to be a problem getting these pieces of data, although it should be Rointed out that in some cases the machine's speed is uncertain, particularly when a compressor with an integral gear box is involved. Other load related variables are inertia and speed-torque characteristics during starting, which may not be known when the motor design is requested. This could be because the compressor design isn't completed, or it may be because the selection of the compressor vendor is undecided (both of which are common during fast track projects). The impacts these variables have on motor design are discussed in detail below. Based upon applicable experience, we will assume the impact of these variables can result in a change in the cost of the machine by 10% to 35%.

3.1 Short Circuit Capacity

Non-electrical engineers frequently misunderstand the impact of electrical system parameters on the machine's design. Most project people understand the first cost of

a 4000 HP machine will be more when built for 13.8kV than for 4kV, or that a 10,000 HP machine costs less at 13.8kV than 4kV. But, they may not understand it's the thicker insulation, longer end turns and smaller available slot space which result in a larger frame size in the first example, or the huge inrush results in cooling problems during acceleration of the second example. They also understand the costs will increase when a transformer is added to the scope of supply. This leads to the more subtle variations in the electrical network at different sites, such as network short circuit capacity (SCC). For the record, short circuit capacity is the amount of current that will flow from the network in the case of a fault. It also has a direct impact on the voltage dip during motor starting. Typically, it is expressed in thousands of amperes (kA), or millions of volt-amperes (MVA). This parameter varies from site to site because of the configuration of the electrical system, or network, feeding the site. The major variables are the number and size of the generators feeding the network and the system impedance between these generators and the site. The components affecting impedance are the length and size of the cables carrying the power and the transformer ratings (MVA and impedance).

What is the importance of SCC? Many engineers, electrical and non-electrical, understand that this parameter affects the design and cost of the switchgear, particularly in the higher MVA ratings. What frequently is not appreciated is how SCC affects the voltage flicker (or dip) during starting, and how this impacts the motor design. For example, a 4000 kW (for simple calculations, assume 4000 kW = 4000 kVA = 4 MVA) machine with 500% inrush, connected to a network with a SCC of 200MVA at the motor terminals (see Table 3.2). When started, this machine will draw 20 MVA from the network (4.0 x 5.0 = 20). This will cause the network voltage to dip 10% (20 I 200 = 0.10). However, if this same motor is connected to a network with a SCC of 100 MVA, the voltage dip will be 20% (see Table 3.2). A voltage dip of this magnitude can cause problems with sensitive electronic devices, cause electrical contactors to drop out, or have a negative impact upon other plants connected to the same electrical system. Another potential impact of more immediate and critical nature is that the motor may stall during acceleration. In some locations, system voltage flicker requirements are very tight, and when combined with a relatively low SCC, this creates a very difficult design condition.

Table 2: Simplified Voltage Dip Calculations (assumes P.F.=O during starting)

- 166 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

3.3 The Impact of SCC on Torque

Motor Inrush @ 100% V Inrush @ Feeder Voltage Dip @ Motor Voltage Dip Q Feeder

The SCC has a direct impact on a motors torque. The torque produced by a machine (induction or synchronous) is proportional to the square of the voltage. Therefore, the torque produced at 80% voltage is 64% of the torque at 100% voltage. In the example of a 4000 kW machine, lets assume the design produces 40h of rated torque at locked rotor, has a pull-up torque of 40% and a breakdown torque of 200%. At 80% voltage, these values become 26%, 26% and 128Oh, respectively. Figure 2. shows these characteristics graphically for three motor speed-torque curves, at loo%, 90% and 80% voltage, and two load speed-torque curves, with maximum torques during starting of 90% and 60% at full speed. At 90% voltage the machine will be able to start both loads, but at 80% voltage only load # 2.

20 MVA 20 MVA 18 MVA 16 MVA 18 MVA 16 MVA 20 MVA 20 MVA 20 MVA 20 MVA 16.2 MVA 12.8 MVA 16.2 MVA 12.8 MVA

10% 20% 7.5% 6.0% 14.0% 11.4% 7.5% 14.1% 10% 20h 16.8% 25.0% 22.6% 29.1Oh 24.3% 29.6Oh

I I 180 00%

im 00%

3 14000%

8 i m m E moo56

P 12000%

e 80Wk

4000% 2000%

000%

% % R $ $ 6 $ P e $ ~ r d sped

L Figure 2: Motor Starting Torques at 80%, 90% 8 100% plotted against Load Torques Increasing Quadratically to 90h 8 60% of Motor Rated Torque.

One method used to lessen the impact of motor starting on the network is to apply a reduced voltage starter, typically an auto-transformer. However, the auto-

transformer will lower the motor terminal voltage even further, so while the network voltage dip is less, the motor still may not develop sufficient torque. In the 4000 kW motor example, using an auto-transformer with a 90% voltage tap will reduce the motors starting current to 450% (500 x .9 = 450). The inrush seen by the network will be reduced by the .90 transformer ratio, to 405%, or 16.2 MVA. This will result in a network voltage dip of 16.2% on a 100 MVA network. This in turn reduces the motor terminal voltage to 75.4% (0.838 x .9 = .754). The motor terminal voltage will result in a further reduction in inrush, so several more iterations are necessary to calculate the true inrush, resulting voltage flicker, and the resultant available load torque. The problem is the motor didnt develop sufficient torque when the voltage dip was 20%, and puttiMg an auto- transformer in the circuit only decreased the motor torque. To compensate for the lower terminal voltage and lower torque, the motor designer may wish to increase the starting torque, which in turn will increase the inrush. This continues the problem in which the higher inrush leads to a higher voltage drop or lower auto-transformer tap, and subsequently a lower motor torque.

One possible solution to this problem may be to remove the auto-transformer from the circuit before the motor reaches full speed. Switching the auto-transformer out at 80% or 90% speed, when the inrush current is slightly lower may enable the system engineer to meet the voltage dip requirements and allow the motor to develop sufficient torque. Another solution is to apply a dedicated, or captive transformer. Adjusting the transformer rating (MVA) and impedance should allow the engineer to meet the voltage dip requirement, and provide sufficient current to the motor. One of the biggest advantages of a dedicated transformer is the voltage dip at the motor terminals can be allowed to dip much lower than would be possible if the motor shared a bus with other loads.

Table 3: A Comparison of Alternative Starting Methods and the Impact of Voltage Dip at the Feeder and Motor Terminals

- 167 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

NEMA MG-I [I] requires a motor to be capable of accelerating a defined load at 85% voltage. The NEMA MG-1 curve is a somewhat arbitrary starting point for design, and may or may not be representative of the final load speed-torque curve. The load is defined as one with a speed-torque characteristic, which varies quadratically up to 90h of the motor's rated torque at 100% speed and has inertia according to formula. A Table of the results of the formula is also given for various speeds and power ratings. The alternative international standard, IEC 34-1, similarly defines the load speed- torque characteristics and inertia at various power ratings and speeds.

4.0 Technical Comparison of 4-pole synchronous and induction machines

determine if synchronous or induction was the appropriate compressor driver. In this particular case, the evaluated choice indicated that an induction solution was required. The following sections describe design issues, which merit consideration. This machine will be referred to as the initial machine in the following paragraphs.

4.1 Values compared

There are hundreds of machine parameters to compare between synchronous and induction machine designs. The main values are the active volume, i.e. the active length, and the diameter, which affects the material cost of the machine, and efficiency which affects lifecycle costs. A third value is Power Factor which affects lifecycle and investment costs (stronger network).

The 16,000 HP machine specified in Section 2.0 of this paper was technically and commercially evaluated to

Table 4: Comparison of synchronous and induction machine parameters for the same output ratings and side conditions (without exciter).

4.2 'What if' affects on machine design

We will consider the effect of general design changes on the electrical machines. Starting from the given design, the table shows the impact of varying three commonly ignored parameters:

Short circuit capacity, Load speed curve,

0 Inertia

The short circuit capacity of the power supply is considered at three additional points 1.3, 0.7 and 0.3 of the initial value while the voltage dip is kept constant. Next, the load torque curve is changed from quadratic against 30 % of nominal torque at nominal speed (initial machine) to quadratic against 60 %. Finally, the initial value of inertia is doubled. The initial machine is considered to be an optimal design for this machine type with the initial side conditions.

The synchronous machine is less affected by application changes than the induction machine. Since the torque is proportional to the machine voltage, a change of the short circuit capacity has an affect on the machine. An increase in load torque is not that critical, although above a certain limit (60% load torque) synchronization may be

difficult. If the inertia is doubled, the initial synchronous machine becomes overloaded. Rotor temperature, not the synchronizing torque, limits the application of this machine: an auto-transformer is a possible solution.

To the contrary, the induction machine is more affected by application changes, but it has capabilities for a wider range of high inertia loads. Since the torque varies proportionally with the square of the voltage, a difference in short circuit capacity changes the machine design. A larger short circuit capacity improves efficiency, and a lower short circuit capacity lowers efficiency. Normally, instead of using copper for the rotor bars, an alloy with lower electrical conductivity is used. Increasing the load torque affects the induction machine in a similar way. The doubling of the inertia increases the acceleration time. Since a squirrel cage machine has no insulation in the rotor (unlike the synchronous machine), the thermal limit is not limited by insulation. Rather, the thermal limits are restricted by mechanical stresses in the rotor due to different thermal elongation coefficients from copper and steel. The permissible locked rotor time may become lower than the starting time, so monitoring of the machine acceleration is required to prevent damage.

- 168 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

Effects on an Induction Motor Effects on a Synchronous Motor

Initial machine

Permissible Stall Time < Starting

Initial machine, better efficiency (+0.8%) Auto-transformer

Lower efficiency (-0.2%) Permissible Stall Time c Starting Time, lower efficiency (-0.6%) Shorter machine (4%), better efficiency (+0.1%) Longer machine (12%), Lower efficiency (-0.4%) Permissible Stall Time e Starting Time, longer machine (16%) Lower efficiency (-0.2%) Permissible Stall Time < Starting Time, lower efficiency (-0.3%) Lower efficiency (-0.4%) Permissible Stall Time < Starting Time, lower efficiency (-0.6%) Auto-transformer, lower efficiency (-0.6%)

Time Initial machine Auto-transformer

Initial machine

Auto-transformer Initial machine Auto-transformer

Initial machine Auto-transformer

Initial machine Auto-transformer

Initial machine

- 1

1.3

I

30 % 1

0.7

Table 5 Comparison of design changes on motor parameters

30 % 1 2

4.4 Machine Characteristics

0.3

The synchronous machine has a larger starting torque than an induction machine. The effect is a shorter startup time but also a larger torsional stress on the load equipment - apart from the synchronous torque pulsation during startup! For startup characteristics, see following graphs.

5.0 Motor Selection

The following technical and commercial considerations must be evaluated during the initial project phase of a large compressor driver design:

30 O h 1

Capital cost Efficiency at load point Energy cost Maintenance impact Simplicity of design Common spares Machine starting method Machine load requirements Construction/lnstallation Method of starting

- 169 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

1

1.5 - B 1 al 3

0 p 0.5

0 U

1.5 I 0 0.5 1

4 0.5

U

0 5 I O

time [SI

0 5

time [SI 10

Induction Machin startup

2.5 - 2

1.5 3 1 b 0.5

0

al

U

U

Startup 2.5 1 I I I I

- 2 1.5

3 1 I F 1 8 0.5

0 0 5 10 15 20

time [SI

1.5 1

0 5 10 15 20

time [s]

Figure 3: A Comparison of Synchronous and Induction Motor starting characteristics (Load Torque is dashed line)

Due to commercial sensitivity from the user and Most users utilize a net present value of dollars per kW equipment supplier, the evaluation presented here is not ($/kW). This factor ($/kW) is frequently determined from the actual one from this project. In a competitive financial models for the given project utilizing energy business environment, this evaluation should be done to cost, plant depreciation life, internal rate of return on provide the greatest engineering value to a given project. capital, inflation, interest rate for capital, joint venture

financial arrangements, and numerous other items.

- 170 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

In the case of the 16,000 HP machine, the key considerations which drove the decision were a relatively low $/kW, capital cost, and simplicity of design. As can easily be seen, the evaluation process could yield a different result if performed on two projects, which are in different countries, or even in different locations within the same country.

A captive (dedicated) transformer was utilized, due to the available system voltage and the substation design optimization which was possible by utilizing 38 kV metal- clad switchgear. If this motor (as well as the 6,500 HP unit) were installed utilizing an 11 kV-autotransformer start, the electrical system would have included a 38kV breaker, a large substation transformer, and an I l k V switchgear line-up. The voltage drop at the I l k V bus, and effectively the entire plant (ASU) electrical system, would be based upon starting the 16,000 HP motor. In the actual design, this is eliminated, due to the dedicated transformer. This method of starting is often considered to be one of the most expensive, and not generally

Synchronous Motor Efficiency 98.0% Price $560,000 Power Penalty - Spares

Bearings (2), exciter $ 73,000 components, spare exciter rotor

considered to be a reasonable solution for efficient, cost effective motor starting application. In this application, a complete evaluation of various options indicated that captive transformer start was clearly the most economical and technically desirable choice.

6.0 Motor design

Induction Motor 97.2%

$350,000 $ 95,000

$ 11,000

The design of a large 4-pole induction machine is close to the one of a smaller, more commonly used 2-pole machine. The rotor has retaining rings, a spider shaft and needs an analysis of mechanical stresses and critical speeds. However, this machine can be designed to run either above or below the first critical speed. There are no fans mounted on the shaft, due to the fact the rotor itself acts as a fan. The machine is equipped with forced lubricated sleeve bearings mounted on the end- shields. The TEWAC design has a noise level of 85 dB(A). The stator winding is vacuum impregnated.

Torsional Analysis Reduced Voltage Starting Capacitors (including switchgear)

Total Evaluated Costs

- I Not Req'd Not Req'd Not Req'd Not Req'd

$633,000 456,000

& stator, spare field coils I I Additional Switchgear devices - -

7.0 Field Report

To be provided from startup of plant (spring 1998). Pictures will be included.

8.0 Conclusions

When specifying large compressor-drive electrical machines, it is critical that complete electrical system information is obtained as early in the project as possible. The system SCC, as well as the voltage flicker requirements, must be determined prior to formal specification of the machine to ensure that it will be able to accelerate under compressor load requirements. It is quite evident, given the wide range of electrical system parameters that SCC will vary from one geographical site to another. It is a common design philosophy in industry today that production units are copied from one facility to another for economic and schedule reasons. This philosophy must accommodate changes in SCC, and

include detailed engineering study. Even given the requirement that the motor supplier must meet NEMA MG-1 (85Oh Voltage), large compressor drivers must be specified with the actual load curve to ensure appropriate design.

When comparing and evaluating machine designs, a machine designed for operation on one electrical system may not be a suitable choice for installation and operation on another system. The reason is each electrical system has the possibility of having completely different system parameters. Although a higher SCC may allow the use of a less expensive and more efficient machine, it is possible that this same machine design may be inadequate in an identical process unit, which happens to be geographically, and electrically, located in a lower SCC area.

- 171 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.

Figure 4 Motor sitting on Factory Floor

Evaluation of synchronous and induction machines must be done commercially as well as technically. In many large machine applications, the commercial evaluation requires more than the supplier's best price. It involves project economic evaluated power savings ($/kW), motor efficiency, method of starting, installation cost, as well as life cycle operations and maintenance considerations.

Captive (dedicated) transformer motor starting is often considered to be one of the most expensive and least desirable solutions to motor starting problems. It is typically not even considered as an efficient, cost- effective motor starting solution. In this application, a complete evaluation of various options (including captive transformer) indicated that captive start was clearly the most economical and technically desirable choice.

REFERENCES

1. "Motors and Generators" ,ANSI/NEMA Standards Publication MG 1-1993, Rev. 4, June, 1997

2. "Rotating electrical machines Part 1: Rating and performance", IEC Publication 34-1, 1969,

3 Moon H. Yuen: "Short Circuit ABC-Learn It in an Hour, Use It Anywhere, Memorize No Formula" , IEEE Transactions on Industry Applications, March/April 1974

- 172 -

Authorized licensed use limited to: Manuel Briceno Pina. Downloaded on January 21, 2010 at 16:50 from IEEE Xplore. Restrictions apply.