Variable Speed Drive on a Die Casting Machine

Design & Engineering Services Variable Speed Drive on a Die Casting Machine 0 Report Prepared by: Design & Engineering Services Customer Service Business Unit Southern California Edison December 17, 2010

0 Acknowledgements Southern California Edison's Design & Engineering Services (DES) group is responsible for this project. It was developed as part of Southern California Edison s Emerging Technology program under internal project number 0. DES' Charles Kim conducted this technology evaluation with overall guidance and management from Paul Delaney. For more information on this project, contact Charles.Kim@sce.com. Disclaimer This report was prepared by Southern California Edison (SCE) and funded by California utility customers under the auspices of the California Public Utilities Commission. Reproduction or distribution of the whole or any part of the contents of this document without the express written permission of SCE is prohibited. This work was performed with reasonable care and in accordance with professional standards. However, neither SCE nor any entity performing the work pursuant to SCE's authority make any warranty or representation, expressed or implied, with regard to this report, the merchantability or fitness for a particular purpose of the results of the work, or any analyses, or conclusions contained in this report. The results reflected in the work are generally representative of operating conditions; however, the results in any other situation may vary depending upon particular operating conditions. Southern California Edison

0 ABBREVIATIONS AND ACRONYMS DCM DES hp Hz IEEE khz kw kwh lb PLC psig PX RMS SCE V VSD Die casting machine Design and Engineering Services Horsepower Hertz Institute of Electrical and Electronics Engineers Kilohertz Kilowatt Kilowatt hours Pound Programmable Logic Control Pounds per square inch (gauge) Power Explorer meter Root-mean-square Southern California Edison Volt Variable speed drive Southern California Edison Page i

0 FIGURES Figure 1. Simplified DCM's Hydraulic System... 3 Figure 2. A 40 HP Variable Speed Drive... 5 Figure 3. Snapshot of Baseline Production Cycle (Power vs. Time)... 6 Figure 4. Wiring Diagram and Relay Settings... 7 Figure 5. Typical Casting Cycle (Without VSD vs. With VSD)... 10 Figure 6. Histogram of Casting Cycle (Unit: Second)... 11 Figure 7. Regression Analysis of The Casting Production (Energy vs. Casting Cycle)... 12 Figure 8. Time Series Trend Analysis of Casting Production Cycle (with VSD only)... 13 Figure 9. 5-Minute Demand Comparison... 13 Figure 10. Monte Carlo Simulation Results of Daily kwh Savings per DCM... 15 Figure 11. Monte Carlo Simulation Results of Estimating Daily Casting Production... 15 Figure 12. Operating Hours Indicated By DCM kw Recordings from January 7, 2011 Through January 23, 2011.... 16 Figure 13. Operating Hours Indicated By DCM kw Recordings from January 27, 2011 Through February 10, 2011.... 17 TABLES Table 1. Descriptive Statistics for A Typical Casting Cycle Shown in Figure 5 (Unit: kw)... 10 Southern California Edison Page ii

0 CONTENTS EXECUTIVE SUMMARY 1 INTRODUCTION 3 Background... 3 Emerging Technology... 4 Assessment Objectives... 4 TECHNOLOGY/PRODUCT EVALUATION 5 Baseline Technology... 5 Retrofitting a VSD on the DCM... 5 VSD Parameter Settings and Input Signals from DCM... 6 TEST METHODOLOGY 8 Field Testing of Technology... 8 Instrumentation Plan... 9 Error Analysis... 9 Measurement Uncertainty... 9 RESULTS 11 Energy Savings... 11 Potential Demand Reduction... 13 Estimating Annual Savings... 14 RECOMMENDATIONS 18 Southern California Edison Page iii

0 EXECUTIVE SUMMARY A die-casting machine (DCM) is typically driven by a complex hydraulic system to produce dimensional metal parts accurately. The design of a DCM is mainly driven by high-volume production and shot capacity 1 ; the low production cost of each casting allows the DCM owner to make a profit. Until the early 90's, high-volume orders enabled DCM owners to invest in advanced die casting machines. Unfortunately, the global economy has changed the nature of the die casting business in the United States, particularly in California, and now other countries produce high-volume orders; therefore, the survival of die casting companies in California heavily relies on frequent customized designs and small-volume orders. Die-casting companies are looking for methods to reduce the production cost per casting by making various improvements such as increasing energy efficiency. The goal of this project is to improve the energy efficiency of existing aging DCMs by retrofitting a variable speed drive (VSD) on the DCM's hydraulic pump motor. This project evaluates the VSD measures' ability to provide energy savings when the speed of the hydraulic pump motor varies, or is controlled based on the DCM's hydraulic pressure and flow rate requirements. Southern California Edison (SCE) conducted a field evaluation where the energy efficiency of an existing aging DCM was measured. A baseline was established by examining the typical operation of an existing DCM's performance after being retrofitted with a VSD while still maintaining the typical operation. The selected DCM is a modified Harvill 2 Model 406 that is over 20 years old. Its locking pressure capacity is 400 tons (US) and the maximum shot capacity is 14.3 pounds (lbs) of (aluminum). The hydraulic pump is driven by a 6-pole, 40 horsepower (hp), 480 Volt (V), and 3-phase motor. In order to establish a fair comparison, we replaced the existing motor with a National Electrical Manufacturers Association (NEMA) Premium motor with the same specifications. Thus, the baseline is the conventional operation mode of an existing DCM when retrofitted with a new hydraulic pump motor. Because the DCM does not produce the same casting for more than 1 or 2 days, low-volume die casting orders created a measurement challenge during this evaluation period. Therefore, one-day measurements were designed in order to capture a fair comparison between a DCM with the VSD installed and one without. The energy consumption of the DCM without the VSD installed while producing the same number of castings as the DCM with the VSD was monitored. The one-day measurement results reveal an approximate 10% reduction in energy consumption and the potential to reduce demand by 15%, which is a significant change. Despite significant energy savings, retrofitting a VSD on an existing aging DCM is a challenge. The VSD can get overloaded and stop working if the accumulator pressure valve is malfunctioning, as experienced in this field evaluation. Additionally, oil leaks are a common maintenance issue that can change the pressure rise-time and present challenges in maintaining pressure levels. The modified DCM, used as the baseline system, created 1 The "shot capacity" is the maximum weight of molten metal that the DCM can inject into the mold. 2 This DCM manufacturer closed its business about 15 years ago. Southern California Edison Page 1

0 another challenge since its existing design can be significantly different from the original design specifications (e.g., different size motor, pump, or additional accumulator, a new programmable logic control (PLC), etc.). Because retrofitting a VSD on an existing aging DCM requires a solid understanding of its specific hydraulic system's performance and ability to control the DCM through the PLC, it is highly recommended that a DCM expert be consulted prior to installing a VSD on a DCM. In addition, the study recommends that a bypass switch be installed between the motor and the VSD so the operator can bypass the VSD in the event of a malfunction. Southern California Edison Page 2

INTRODUCTION BACKGROUND Die-casting is a fabrication technique that involves a high-pressure injection of molten metal into a cast. The high-pressure injection is driven by a complex hydraulic system that allows a die-casting machine (DCM) to accurately mass produce dimensional metal parts. The DCM's programmable logic control (PLC) allows the hydraulic pump to inject oil into an accumulator until it reaches a desired pressure level (1,500 pounds per square inch (gauge) (psig) to 1,800 psig). Once it reaches the desired pressure level, the accumulator valve closes and the pump is unloaded. The PLC then opens the fluid control valve that causes highly pressured oil to 1) close the die, 2) inject liquefied metal into the die, and 3) open the die. This three-step process takes 9 to 11 seconds. The PCL then 4) controls the DCM and ejects the casting; 5) the DCM operator opens the DCM door and removes the casting, 6) cools the die with the water spray, and 7) closes the DCM door. This process takes 20 to 30 seconds. Therefore, the total cycle takes between 29 and 41 seconds. The production cycle time changes depending on how efficiently the operator removes the cast, cools the die, and opens and closes the die-casting door. The complexity of the casting can also affect the production cycle by adding 1 to 2 seconds to the total cycle time. Figure 1 shows a typical DCM's hydraulic system. FIGURE 1. SIMPLIFIED DCM'S HYDRAULIC SYSTEM The design of a DCM is driven mainly by high-volume production and shot capacity. The low production cost of each casting allows the DCM owner to make a profit. Until Southern California Edison Page 3

the early 90's, high-volume orders enabled DCM owners to invest in advanced die casting machines. Unfortunately, the global economy has changed the die casting business in the United States, particularly in California, and now high-volume orders are produced in other countries; therefore, the survival of die casting companies in California heavily relies on frequent customized designs and small-volume orders. Die-casting companies are looking for ways to reduce the production cost of each casting by making various improvements such as increasing energy efficiency. EMERGING TECHNOLOGY A variable speed drive (VSD) is a device that controls the speed or torque of an induction motor by modulating the input frequency to the motor. The VSD requires an input signal related to the load fluctuations. For a DCM, input signals can be obtained from the existing PLC by installing additional logic relays, if necessary. For example, the PLC can send an input signal by opening or closing a relay to the VSD so the VSD can control the motor to run at high, medium, or low speed, depending on the die casting process modes. The different speeds can be determined by monitoring the pressure rise-time (typically 4 to 5 seconds) and understanding the steps in the die casting process. The die casting process is relatively simple and the steps are outlined below: 1. Die closes: The hydraulic system closes the "cover die" and the "ejector die". 2. Shoots forward: A metal piston forces molten metal into the paired die. The injected metal forms to the shape of the die and solidifies rapidly in the cold die.) 3. Die opens: The hydraulic system separates the "cover die" from the "ejector die". 4. Ejects forward: The hydraulic system ejects the newly formed casting from the "cover die". 5. Shoots back: The metal piston moves back to the original position. 6. Ejects backward: The hydraulic system ejects the newly formed casting from the "ejector die". 7. Idles for cooling: The die is cooled by a water spray. ASSESSMENT OBJECTIVES The goal of this project is to improve the energy efficiency of existing aging DCMs by retrofitting a VSD on the DCM's hydraulic pump motor. This project evaluates the VSD measures' ability to provide energy savings when the speed of the hydraulic pump motor is varied or controlled based on the DCM's hydraulic pressure and flow rate requirements. The retrofitted VSD on the DCM is evaluated based on the following objectives: Evaluate the energy savings, if any, after installing a VSD on the existing aging DCM; Evaluate the potential of a peak demand savings during the field test, if any, after installing a VSD; and Estimate the annual energy savings. Southern California Edison Page 4

TECHNOLOGY/PRODUCT EVALUATION BASELINE TECHNOLOGY Southern California Edison (SCE) conducted a field evaluation where the energy efficiency of an existing aging DCM was measured. A baseline was established by examining the typical operation of an existing DCM's performance after being retrofitted with a VSD. A typical existing DCM does not use a VSD for controlling the pump speed, as is the case among 17 die casting companies within SCE's service territory. The selected DCM for this study is a modified Harvill 3 Model 406 that is over 20 years old. Its locking pressure capacity is 400 tons (US), and the maximum shot capacity is 14.3 pounds (lbs) (aluminum). The hydraulic pump is driven by a 6-pole, 40 horsepower (hp), 480 Volt (V), and 3-phase motor. In order to establish a fair comparison, the existing motor was replaced with a National Electrical Manufacturers Association (NEMA) Premium motor with the same specifications. Thus, the baseline is the conventional operation mode of the existing DCM driven by a new hydraulic pump motor. RETROFITTING A VSD ON THE DCM Figure 2 displays a 40 hp VSD installed to control the 40 hp motor. The performance of the VSD was compared to the baseline in order to evaluate the amount of energy saved when a DCM is retrofitted with a VSD. FIGURE 2. A 40 HP VARIABLE SPEED DRIVE 3 This DCM manufacturer closed its business about 15 years ago. Southern California Edison Page 5

Power (unit: kw) Variable Speed Drive on a Die Casting Machine VSD PARAMETER SETTINGS AND INPUT SIGNALS FROM DCM The VSD parameters were programmed in accordance to the desired DCM performances. The key parameters include: Motor speed Motor acceleration time Motor deceleration time To determine the above parameters, several snapshot power measurements were taken for the baseline condition. Figure 3 shows a snapshot of the baseline production cycle. One Cycle Mode 1 Mode 2 Mode 3 40 35 30 25 20 Oil Leaking 15 10 5 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 Time (unit: sec) FIGURE 3. SNAPSHOT OF BASELINE PRODUCTION CYCLE (POWER VS. TIME) The "Mode 1" section of the graph represents the first three steps of the DCM production (1) Die closes, 2) Shoots forward, and 3) Die opens), which require the most energy. During this mode, the pump reaches the peak within 5 seconds (accelerating time) and slows down within 2 seconds (deceleration time). The "Mode 2" section of the graph represents steps 4, 5, and 6 of the DCM production process (4) Ejects forward, 5) Shoots back, and 6) Ejects backward and rebuilds the pressure in the accumulator). During this mode, the pump reaches the peak within 1 to 2 seconds and slows down at the same speed. The "Mode 3" section of this graph represents step 7 of the DCM production process (7) Idle time for die cooling). During this mode, the existing motor still runs at full speed; however, it is unloaded or lightly loaded. Please note that during this mode there is oil leaking from the DCM's accumulator that creates a noticeable pressure drop (about a 150-psig drop within 1 second); therefore, the motor re-runs in order to build up the pressure. Southern California Edison Page 6

Although this is a maintenance problem, the operator sees this as the typical operating condition 4 for an existing aging DCM. The energy level for each mode can be determined by totaling the kilowatt (kw) values within the specific mode. After the energy levels for each mode are determined, the speed parameter on the VSD can be set. The following parameters were selected based on the analysis results shown in Figure 3: Mode 1 speed: 40 hertz (Hz) Mode 2 speed: 38 Hz Mode 3 speed: 32 Hz Acceleration Time: 2.5 seconds Deceleration Time: 1.5 seconds After the speed parameters are determined, input signals can be provided to the VSD so the VSD can send the appropriate signals to the motor according to the modes defined in Figure 3. This can be accomplished by reprogramming the existing PLC with three additional relays. A typical VSD allows the user to run at the preset value when it receives a relay signal to its terminal block. Figure 4 shows the wiring and logic diagrams. FIGURE 4. WIRING DIAGRAM AND RELAY SETTINGS The PLC can send signals to the VSD by closing or opening Relay 1 through 3. For example, when the PLC closes Relay 1, the PLC informs the VSD that the DCM is in Mode 1 ("Die closes, Shoots forward, and Die opens"). Once the VSD sees that Relay 1 is "closed", the VSD runs the motor at the "Preset Speed #1". Then Relay 2 "closes" and the VSD runs the motor at the "Preset Speed #2". 4 According to the maintenance supervisor, the accumulator valve is not fixed or replaced unless there is an excessive amount of oil leaking. Southern California Edison Page 7

TEST METHODOLOGY FIELD TESTING OF TECHNOLOGY To evaluate the energy savings when retrofitting the VSD onto an existing aging DCM, a power-quality-grade meter was installed to collect kw data. An event data logger was also installed to record the number of castings produced during the test period. Low-volume die casting orders created a measurement challenge because the DCM does not produce the same casting for more than 1 or 2 days. Therefore, two different amounts of measurement time-periods were established: One-day measurements were performed on December 13, 2010 (data is captured every 1 second for 1 to 2 hours.) Four-week measurements (capture data every 30 seconds.) Designed to capture a fair comparison, the one-day measurements monitor energy consumption without the VSD on the DCM while producing the same number of castings. The 4-week measurements are designed to estimate the average energy savings and to determine operating costs. The 4-week monitoring period included 2 weeks with the VSD on the new hydraulic pump motor and 2 weeks without the VSD installed on the DCM. A fair comparison for the 4-week measurement is difficult because of the: Various casting types produced: Some are simple-shaped while others are more complicated; some castings are heavy (over 4 lbs) while others are light (less than 1 lb). Operator performance: Some operators produce more castings than others within a specific shift do. According to the production supervisor, a highly experienced operator can produce approximately 800 castings per shift while most operators produce approximately 600 castings per shift. Wide range of die setup time: When setting up a pair of die (one is called the "cover die" and the other is called the "ejector die"), a simple-shaped die takes 1 to 2 hours while a complicated and specially-designed die can take up to 5 hours. During the die setting, a limited number of castings are produced. Before a die is considered "set", a quality control supervisor must approve the sample casting. Therefore, operators may have to set the die repeatedly, and sometimes only 10-20 castings can be produced per hour. This is compared to 60-80 castings per hour during the production mode. Maintenance issues: Operating an existing aging DCM is also a challenge for the operator. For example, after a few hours of operation, the pair of die can become misaligned. The operator may detect this visually by inspecting the produced casting and either shut down the DCM or allow the machine to run while realigning the die. Realigning the die can take anywhere from 15 minutes to 1 hour. If there is a major problem, the machine can be out of commission for 1 to 2 days. Southern California Edison Page 8

INSTRUMENTATION PLAN At the first site visit, the size of motor was recorded and the location of the circuit breaker for the DCM was identified. As a result, the following instruments were selected for this project: Power measurement: Dranetz BMI PowerXplorer PX5. This device is designed for monitoring power quality. The accuracy of a meter is 0.1% of the true root mean square (RMS) reading. It complies with IEEE 1159, IEC 61000-4-30 Class A and EN50160, and captures IEEE 1459 parameters for non-sinusoidal and advanced power systems. Power measurement sensor: Dranflex 3000XL current probes. Its accuracy level for a 300 Ampere (A) Range is ±1% of reading ± 0.1 A. Its frequency range is 10 Hz to 10 kilohertz (khz) (-1 Decibel). It complies with EMC EN 61326-2-2:2006 and EN 61010-1:2001; EN 61010-031:2002; and EN 61010-2-032:2002 safety standards. Event Measurement: Hobo U11. It can record up to 43,000 state changes or events. It is used for recording each casting production cycle. All of the instruments used in this project are up to calibration date. ERROR ANALYSIS Error analysis for a small number of independent variables can be done as follows; the kw reading, for example, is dependent upon two independent variables, voltage (V) and the current (I), then the error in kw measurement is: kw kw V V 2 I I 2 Therefore, the most inaccurate device will drive the error in kw. In this case, the probe is the most inaccurate device. Its accuracy level is 1%, which implies that the kw error will be less than 1.1%. MEASUREMENT UNCERTAINTY Measurement uncertainty is a key issue for this project. The VSD is programmed to save energy and reduce the demand by controlling the acceleration and deceleration time variables in addition to varying the voltage frequency to the motor. Therefore, the DCM, without the VSD, quickly accelerates the motor pump and reaches the predetermined pressure level while the VSD controls the motor pump and reaches the same pressure level relatively slowly, as illustrated in Figure 5. The descriptive statistics of the typical casting cycles shown in Figure 5 are tabulated in Table 1. When comparing the "mean" and "median" values, it can be easily seen that the casting cycle for "without the VSD" is skewed to the right and the casting cycle data for "with the VSD" is skewed to the left while the mean values for both are relatively close. This means that if the measurement is collected by "averaging samples over 30 seconds", for example, the measurement instrument may not see the significant energy savings. Therefore, the meter must have a capability to capture a snapshot, Southern California Edison Page 9

kw Variable Speed Drive on a Die Casting Machine not averaging samples over the given interval. Furthermore, the tested DCM was not fully automated but rather controlled by an operator. Therefore, the production time for making a casting can vary greatly from one operator to others. The DCM makes various parts as well. Therefore, the energy needed for making a casting can vary as well. Because of these uncertain factors, the four-week measurement (two weeks without the VSD and two weeks with VSD) is designed to verify operating hours. 35 30 Variable without VSD VSD 25 20 15 10 5 0 4 8 12 16 20 24 28 One Casting Cycle (unit: second) 32 36 FIGURE 5. TYPICAL CASTING CYCLE (WITHOUT VSD VS. WITH VSD) TABLE 1. DESCRIPTIVE STATISTICS FOR A TYPICAL CASTING CYCLE SHOWN IN FIGURE 5 (UNIT: KW) VARIABLE SAMPLE SIZE MEAN Standard DEVIATION MINIMUM First QUARTILE MEDIAN Third QUARTILE MAXIMUM Without VSD 38 14.74 12.63 3.67 3.69 8.78 29.87 34.01 VSD 38 13.76 9.16 1.69 2.05 18.38 21.5 23.69 Southern California Edison Page 10

Frequency Variable Speed Drive on a Die Casting Machine RESULTS ENERGY SAVINGS Figure 6 shows the production cycles (in seconds) obtained from the 1-day measurement. The sample size was 522 which means 522 castings were made. The mean production cycle was 36.7 seconds. However, there were outliers resulting from the quality-assurance inspector's activities. During the inspector's visual checkup, the casting cycle ranged from 65 to 280 seconds. The total number of outliers was 29 out of 522 samples. Excluding the outlier, the range of the casting cycle was 26 to 58 seconds. (The total number of samples was 493 of which 343 castings were produced with the VSD, and 150 castings were produced without the VSD.). 160 140 Mean 36.70 StDev 6.813 N 522 120 100 80 60 40 20 0 20 30 40 50 60 70 Casting Cycle 80 90 FIGURE 6. HISTOGRAM OF CASTING CYCLE (UNIT: SECOND) The energy-savings calculation for the baseline data can be accomplished by performing the following steps: 1. For each casting, add the kw values taken every second during the casting cycle. (The sum of the kw values equals the amount of energy used.) 2. Run a regression analysis to determine the amount of energy used per casting cycle for all of the samples. 3. Repeat steps 1 and 2 for the VSD retrofitting condition. Figure 7 shows the results of the above steps. Southern California Edison Page 11

Energy Variable Speed Drive on a Die Casting Machine 700 650 Variable Baseline VSD 600 550 500 450 400 30 35 40 Casting Cycle 45 50 FIGURE 7. REGRESSION ANALYSIS OF THE CASTING PRODUCTION (ENERGY VS. CASTING CYCLE) The regression results are: Baseline (energy) = 274.6 + 7.918 T (unit: kw seconds/casting) where T is the casting cycle in seconds. VSD (energy) = 351.7 + 4.279 T When the average casting cycle of 36.7 seconds, obtained from Figure 6, is applied, the result is: Baseline (Energy) = 274.6 + 7.918 (36.7) = 565.2 (unit: kw seconds/casting) VSD (Energy) = 351.7 + 4.279 (36.7) = 508.7 (unit: kw seconds/casting) Therefore, the average energy savings is the difference of 56.5 (unit: kw seconds/casting) or 0.0157 kilowatt hour (kwh) per casting. This implies that the VSD saves 0.0157 kwh per casting on average. This is equivalent to a 10% energy savings per casting. Figure 7 shows that the casting cycle is reduced for the baseline condition; the average casting cycle in the morning is 37 seconds (using the VSD), and the average casting cycle in the afternoon is approximately 33 seconds (baseline). The average casting cycle difference is mainly due to the operator's performance rather than the performance of the DCM. Because a new production cycle requires new die settings and alignments, the operator spends more time on the visual inspection in the morning than in the afternoon as shown in the Time Series Trend Analysis in Figure 8. Southern California Edison Page 12

5-Minute Average Demand (unit: kw) Casting Cycle (unit: Second) Variable Speed Drive on a Die Casting Machine 50 45 40 Yt = 38.155-0.006611*t Variable Actual Fits Accuracy Measures MAPE 4.73019 MAD 1.75820 MSD 5.95027 35 30 1 34 68 102 136 170 204 238 272 No. of Casting Production in Time Order 306 340 FIGURE 8. TIME SERIES TREND ANALYSIS OF CASTING PRODUCTION CYCLE (WITH VSD ONLY) POTENTIAL DEMAND REDUCTION A 5-minute demand from the 1-day measurement is shown in Figure 9. For a fair comparison, any data resulting from interruptions (e.g., lunchtime, break-time, and the inspector's activities) was excluded. 20 Variable Demand (without VSD) Demand (VSD) 15 10 5 0 2 4 6 8 10 12 14 16 18 No of Samples in Time 20 22 24 FIGURE 9. 5-MINUTE DEMAND COMPARISON The average 5-minute demand for the baseline (e.g., without the VSD) is 16.145 kw while for the VSD retrofitting is 13.425 kw; the difference is 2.72 kw. This implies that there is a potential for a 16.8% reduction in the 5-min demand, which is a significant change when the VSD is installed. Southern California Edison Page 13

ESTIMATING ANNUAL SAVINGS According to the operation supervisor, daily casting production ranges from 600 to 900 castings. The collected data indicates that the casting production ranges from 675 to 877 castings per day. The average number of work hours per day is 16 hours (from 5:00 a.m. to 10:00 p.m.), excluding breakfast, lunch, and dinner breaks. The plant operates 49 weeks per year; therefore, the annual saving range is: 0.0157 kwh per casting (600 castings per day x 5 days per week x 49 weeks per year = 2,308 kwh per year) When there are 900 castings produced per day, the annual energy savings is 3,462 kwh per year. Therefore, the range of annual energy savings is between 2,308 kwh and 3,462 kwh per DCM. A Monte Carlo simulation (a repeated sampling) can also be used to estimate the savings. This diverts from using a single point of estimating the energy savings per casting (e.g., applying 0.0157 kwh per casting factor). The Monte Carlo procedure is as follows: 1. Find a probability distribution function for one of the casting cycles illustrated in Figure 6. 5 2. Use a random number generator for the probability distribution function found in step 1; this is a casting cycle from the probability distribution. 3. Calculate the energy savings for a given casting cycle generated from step 2. See Figure 7. 4. Repeat steps 2 and 3 until the total production time per day is between 7 and 11 hours 6. 5. Repeat step 2 through step 4, 5,000 times. The results of the Monte Carlo Simulation reveal an average daily savings of 13.78 kwh. See Figure 10. When multiplying the daily savings by 5 days per week and 49 weeks per year, the annual savings can be estimated as 3,376 kwh per DCM. This savings is driven by the daily production variation defined in Figure 11 where the daily mean production is 804 castings. This is a conservative estimated value of energy savings and accounts for only 8.5 hours, on average, of continuous production time. This simulation assumes that the die setting takes place daily and will encounter some maintenance issues during the day. Therefore, the actual savings may be greater if the production level increases. 5 The "gamma" function is found to be closest to the casting data collected where the scale is 14.2946, and the shape is 2.6694. The gamma function is defined as: f ( x) 0 r x x r r 1 1 e e x x dx for r >0, λ > 0 and x > 0 The parameters λ and r are often called the "scale" and "shape". 6 The 7 to 11 hours per day accounts for the continuous production time per day with an average production time of 8.5 hours per day. This excludes die setting time, all breaks (excluding breakfast and lunchtime), various inspections, and maintenance times. Southern California Edison Page 14

Frequency Frequency Variable Speed Drive on a Die Casting Machine Currently, there are 17 die casting companies within SCE's service area. If an average of five DCMs per company operates, the estimated annual savings potential can be calculated for retrofitting the VSDs as follows: Market Potential for Retrofitting the VSDs on DCMs = 3,376 kwh per DCM x 5 DCMs per company x 17 companies = 286,960 kwh savings annually. Therefore, the estimated VSD market potential, in the context of energy savings, for retrofitting a VSD is 287,000 kwh annually. 300 250 Mean 13.78 StDev 1.630 N 5000 200 150 100 50 0 8.0 9.6 11.2 12.8 14.4 16.0 Daily kwh Saving per DCM 17.6 19.2 FIGURE 10. MONTE CARLO SIMULATION RESULTS OF DAILY KWH SAVINGS PER DCM 350 300 Mean 803.6 StDev 95.17 N 5000 250 200 150 100 50 0 450 540 630 720 810 900 Daily Casting Production 990 1080 FIGURE 11. MONTE CARLO SIMULATION RESULTS OF ESTIMATING DAILY CASTING PRODUCTION Figure 12 illustrates the test site ran the DCM without VSD about 10 to 11 hours per day, 5 days per week from January 7 through January 20, 2011. Southern California Edison Page 15

1/5/2011 1/7/2011 1/9/2011 1/11/2011 1/13/2011 1/15/2011 1/17/2011 1/19/2011 1/21/2011 1/23/2011 Without VSD (unit: kw) Variable Speed Drive on a Die Casting Machine 40 35 30 25 20 15 10 5 0 Date FIGURE 12. OPERATING HOURS INDICATED BY DCM KW RECORDINGS FROM JANUARY 7, 2011 THROUGH JANUARY 23, 2011. Figure 13 also shows that the site runs its operation closer to 16 hours a day, excluding breakfast, lunch, and dinner breaks, 5 days a week. The shortest working day is the Friday when the operator turns off the machine around noon. Therefore, the average working hour is about 11 hours a day. Figure 13 also shows that there is a significant peak demand reduction as compared to Figure 12. Therefore, 8.5 hours per day is reasonable for estimating annual savings. Southern California Edison Page 16

1/27/2011 1/29/2011 1/31/2011 2/2/2011 2/4/2011 2/6/2011 2/8/2011 2/10/2011 2/12/2011 With VSD (unit: kw) Variable Speed Drive on a Die Casting Machine 40 35 30 25 20 15 10 5 0 Date FIGURE 13. OPERATING HOURS INDICATED BY DCM KW RECORDINGS FROM JANUARY 27, 2011 THROUGH FEBRUARY 10, 2011. The VSD saves approximately 10% of the energy needed to produce castings from the existing DCMs. The VSD can also reshape the input power to the motor to potentially reduce the peak demand, which is approximately 16.8%, based on the 5-minute average peak demand collected data. Overall, the VSD can be an energy efficient measure for existing DCMs since the energy savings can be much greater for the DCMs that do not have an accumulator valve (e.g., the pump is holding the pressure, not the valve) and two to three times more energy savings than a DCM with an accumulator valve. The estimated VSD market potential, in the context of energy savings for this VSD measure, is 287,000 kwh annually within SCE's service territory. Southern California Edison Page 17

RECOMMENDATIONS The VSD saves approximately 10% of the energy needed to produce castings from the existing DCMs. The VSD can also reshape the input power to the motor to potentially reduce the peak demand, which is approximately 16.8%, based on the 5-minute average peak demand collected data. Overall, the VSD can be an energy efficient measure for existing DCMs since the energy savings can be much greater for the DCMs that do not have an accumulator valve (e.g., the pump is holding the pressure, not the valve) and two to three times more energy savings than a DCM with an accumulator valve. The estimated VSD market potential, in the context of energy savings for this VSD measure, is 287,000 kwh annually within SCE's service territory. Despite the significant energy savings it provides, retrofitting a VSD on an existing aging DCM is a challenge. The VSD can be overloaded and stop working if the accumulator pressure valve malfunctions, as experienced in this field evaluation. Additionally, oil leaks are a common maintenance issue that can change the pressure rise-time and cause issues in maintaining pressure levels since its existing design can be significantly different from the original design specifications (e.g., different size motor, pump, or additional accumulator, a new programmable logic control (PLC), etc.). Therefore, a complete field evaluation of the DCM is highly recommended. Because retrofitting a VSD on an existing aging DCM requires a solid understanding of its specific hydraulic system's performance and ability to control the DCM through the PLC, it is highly recommended that a DCM expert be consulted prior to installing a VSD on a DCM. It is also recommended that a bypass switch be installed between the motor and the VSD so the operator can bypass the VSD in the event of a malfunction. Southern California Edison Page 18