Economic and Mechanical Benefits of Utilizing Automatic Balancing and Online Monitoring on Integral Engines and Compressors Gas Machinery Conference 2016 Written by: Kenneth L Anglois DCP Midstream Kent Petersen Windrock, Inc. Technical Assistance: Ray Kimmel DCP Midstream Dylan Abel Windrock, Inc. August 31, 2016
TABLE OF CONTENTS INTRODUCTION... 4 SYSTEM INSTALLATION... 4 CONTROL & MONITORING CAPABILITIES (POWER CYLINDERS)... 7 CASE STUDY #1... 11 CASE STUDY #2... 12 CASE STUDY #3... 14 MONITORING CAPABILITIES (COMPRESSOR CYLINDERS)... 16 CASE STUDY #4... 17 INTEGRATION WITH STATION DCS... 20 ECONOMIC BENEFITS REALIZED... 21 CONCLUSIONS... 22 APPENDIX... 23
EXECUTIVE SUMMARY This paper demonstrates the value of automation monitoring and control upgrades to slow-speed, integral engine/compressors. Manually balancing an integral engine is a time consuming process that is often performed at inadequate intervals to compensate for changes in compressor load, process parameters, ambient conditions or variations in BTU fuel content. The manual task also raises safety issues. Adding automatic balancing and online monitoring to legacy slow-speed machines has been shown to extend the life of equipment, reduce maintenance costs, reduce emissions, improve fuel economy, increase spark plug life and minimize safety issues. DCP Midstream s site Robert s Ranch in Odessa, Texas has historically used manual balance methods for its propane and residue compression. An overview of the system and its installation is given. The monitoring capabilities are also discussed in regards to both the power cylinders and the compressor cylinders. Four separate case studies are provided using data from the DCP site to demonstrate and quantify the advantages of using automation to optimize machinery. Conclusions are given to show DCP s direct economic savings, among other benefits realized. Note: Platinum and AutoBalance are registered trademarks of Windrock, Inc. The Windrock AutoBalance system is protected under US Patent #8522750.
INTRODUCTION Between December, 2015 and January, 2016, DCP Midstream installed Windrock engine AutoBalance and Platinum online monitoring systems on three integral compressors. This paper discusses the economic, performance and maintenance benefits realized with these systems. The systems were installed on two Cooper-Bessemer GMVH-10 and one GMVH-8 integral compressors. These units were built and installed in 1967 at the Roberts Ranch Gas Processing Facility located 9 miles southeast of Odessa, TX. The compressors are dual service and are used to simultaneously compress propane refrigerant for a cryogenic liquefaction process and compression of the resultant low-btu gas into high-pressure gas transportation systems. Please see the appendix for engine specifications and the staging of the compressors. Below is a brief history of the Roberts Ranch Plant and the units: Plant was built in the mid-1950 s by Cities Service Oil Company GMVH units were installed at Roberts Ranch in 1967 Cities Services acquired by Occidental Petroleum in 1982 Duke Energy Field Services (later DCP Midstream) acquired plant in 1998 from Dynegy Plant was shut down in 2005 Restarted in 2010 after complete overhauls including a lean-burn modification with jet cells and upgrade to turbocharger controls SYSTEM INSTALLATION DCP Midstream decided to install the system due to high maintenance costs and significant downtime. These were the results of chronic out-of-balance conditions resulting in repeated head and power cylinder failures on multiple units on multiple cylinders. Manual balancing was done on average once per month due to low availability of qualified personnel with the proper equipment. In addition, DCP Midstream wanted to lower emissions on these units from 3.0 g/bhp-hr to 2.0 g/bhp-hr and reduce the frequency of detonations. This was not achievable with the current balancing methodology. Prior to the installation, all three units would regularly detonate, especially when it was a hot day. The system installation was performed by a team of DCP Midstream station personnel, Windrock personnel and an electrical subcontractor. Power cylinder pressure sensors were installed on each of the power cylinders using dual-port indicator valves. Figure 1 shows an example of one of the online pressure sensors installed on an indicator valve. Note that the dualport indicator valve continues to allow a portable analyzer pressure sensor to be connected for additional diagnostics. In addition, a reference sensor can be connected to the indicator connection to support online calibration of the online sensor.
Figure 1 - Power cylinder pressure sensor The manual fuel balancing valves were replaced with stepper-motor controlled balance valves (Figure 2). Note that no modifications were necessary to the fuel system or the mechanical jet cells. Figure 2 - Original manual balance valve (left) and AutoBalance valve (right)
On the compressor cylinders, pressure sensors were installed on the head-end and crank-end of each throw using dual-ported indicator valves again (Figure 3). Figure 3 - Compressor pressure sensor Accelerometers were installed on the crossheads for each compressor throw as well as on each end of the engine frame (Figure 4) Figure 4 - Accelerometer on crosshead Finally, magnetic pickups were installed at the flywheel to provide speed and phase information for the AutoBalance and monitoring system (Figure 5). Figure 5 - Magnetic pickup installed at flywheel
Cabling for all of the sensors installed on the engines were routed through conduit (Figure 6) and run to the control panels installed on the compressor deck (Figure 7). Figure 6 - Conduit fittings for engine pressure and AutoBalance valve Figure 7 - Platinum cabinet control panel installed on compressor deck CONTROL & MONITORING CAPABILITIES (POWER CYLINDERS) The engine/compressor monitoring system monitors the power cylinders pressure curves continuously. From the pressure curves, peak firing pressures (PFPs) are identified and communicated to the AutoBalance module. The AutoBalance module uses the PFPs to make adjustments to the fuel valves and control the fuel flow to each power cylinder to continuously balance the engine. Some of the key power cylinder parameters that are trended for maintenance and protection purposes are listed in Table 1.
Power Cylinder Parameters Item Description Units 1 AutoBalance valve positions % 2 Fuel flow (from station DCS via MODBUS) MMCFD 3 Fuel heating value BTU/ft 3 4 Mean Peak Firing Pressure (PFP) PSI 5 Compression pressure PSI 6 Mean Peak Firing Angle (PFA) Degrees 7 Standard deviation of PFP PSI 8 Standard deviation of PFA Degrees 9 Percent of poor combustions % 10 Percent of pre-combustion events % 11 Percent over-pressure events % Table 1 - Key power cylinder parameters Figure 8 shows the peak firing pressure trends for a startup for one of the units at Roberts Ranch. Figure 8 - Peak pressure trends for a startup of Unit 12 (GMVH-10) The AutoBalance process was enabled about 30 minutes into the plot in Figure 8. Figures 9, 10 and 11 show the power cylinder pressure curves at the numbered locations in Figure 8.
Figure 9 Power cylinder pressure curves for Time 1 in Figure 8 Figure 10 Power cylinder pressure curves for Time 2 in Figure 8
Figure 11 Power cylinder pressure curves for Time 3 in Figure 8 Figures 9, 10 and 11 show the power cylinder pressure curves as the AutoBalance system adjusts the fuel flow to each cylinder in order to balance the PFPs across the engine. The pressure curves shown above are the average of 30 cycles. The number of cycles to average is user configurable. With each of these cycle averages, the online system calculates a number of parameters and statistics that are trended and are valuable for determining the overall quality of the combustion process in each power cylinder. The station operators and mechanics monitor these parameters either through alarms or by reviewing trends over time. On numerous occasions since the system was installed, operators or mechanics have identified power cylinder issues and resolved them quickly and efficiently because of the additional information provided by the monitoring system. Following are a few examples of these events.
CASE STUDY #1 The trend of compression pressures for all ten of the power cylinders on Unit 12 are shown (Figure 12). Prior to May 3, the compression pressure for Cylinder 1-R was running lower than the other cylinders (approximately 350 PSI compared to approximately 425 PSI for most of the other cylinders). During the shutdown on May 3 and 4, the head gasket on cylinder 1-R was replaced. The compression pressure trend for 1-R after the restart shows that its compression pressure was restored to a value similar to the other cylinders. The same trend data along with the pressure curves before and after the repair is shown in Figure 13. Power Cylinder 1-R Replaced head gasket Figure 12 - Compression pressure trends for Unit 12 Figure 13 - Compression pressure trends with before/after pressure curves
CASE STUDY #2 The peak firing pressure trend for a 24-hour period on May 19 and 20 is shown in Figure 14. At approximately 6:56 am on May 19, the operator received a low PFP alarm for cylinder 4-R (below 500 PSI). When he went out to troubleshoot, he found the secondary lead to the spark plug on the jet cell had come loose. After he reconnected the lead, the PFP returned to the normal value. The pressure curves before and after re-connecting the plug wire show improvement (Figure 15). The sharp rise in Brake Specific Fuel Consumption (BSFC) during the event is also easily seen (Figure 16). The high BSFC would have continued longer without the benefit of the prompt alarm from the monitoring system. Figure 14 - Peak firing pressure trend for a 24 hour period on Unit -11 Figure 15 - Power cylinder pressure curves before and after the event
Figure 16 Brake specific fuel consumption (BSFC) during the event Finally, Figure 17 shows the peak firing pressure trend (zoomed in) around the event. This shows the action of the AutoBalance system recovering the fuel flow after the issue was corrected. While the plug wire was disconnected, the AutoBalance system gradually opened the fuel valve for cylinder 4-R in an attempt to bring its PFP up to the engine average. After the plug wire was reconnected, cylinder 4-R s PFP came up above the engine average and the AutoBalance system gradually reduced the fuel to the cylinder to bring the PFP into the average range. AutoBalance adjusts fuel valve Re-connect plug wire Figure 17 - AutoBalance function during the event
CASE STUDY #3 Figure 18 shows a 3-day trend of percentage of poor combustion events for Unit 11. Poor combustion events are defined as those pressure curves in which there is not a peak pressure higher than the compression pressure. Cylinder 3-L shows intermittently higher percentage of events than the other cylinders. Finally, on May 24, it begins to show consistently higher poor combustion events. Additional symptoms are shown in the Standard Deviation of the PFA in Figure 19. Cylinder 3-L has a consistently higher standard deviation. Figure 18 - Percent poor combustion trend for Unit 11 Figure 19 - PFA standard deviation trend for Unit-11
After the unit was shut down, the jet cell was examined and the port was found to be significantly worn. Figure 20 shows the orifice area of the old and replacement jet cells. The worn port on the jet cell results in poor combustion in the cylinder since the flame front from the jet cell does not propagate through the air/fuel mixture in the cylinder as designed. As shown by the percentage of poor combustion events (Figure 18), a significant percentage of the cycles for this cylinder were poor. This would result in additional fuel consumption as well as increased un-burned fuel in the exhaust for this cylinder. Figure 20 - Old and new Jet Cell orifice surface After the unit was re-started, pressure curves for cylinder 3-L were compared. Figure 21 compares the before and after pressure curves for cylinder 3-L. These plots include the Stat boxes that represent the maximum deviations and the standard deviations of both the PFP and the PFA for the 30 cycle average. Figure 21 - Power pressure curves for cylinder 3-L
Finally, Figure 22 shows how the fuel system was compensating for the poor jet cell performance with the valve position for cylinder 3-L. Figure 22 - Valve position trend for cylinder 3-L MONITORING CAPABILITIES (COMPRESSOR CYLINDERS) Each of the compressor throws are equipped with head-end and crank-end pressure sensors as well as a crosshead accelerometer. These instruments along with the online system provide a complete set of compressor performance monitoring parameters. Some key compressor cylinder parameters that are trended for maintenance and protection purposes are listed in Table 2. Key Compressor Cylinder Parameters Item Description Units 1 Compressor indicated horsepower HP 2 Individual cylinder end HP HP 3 Cylinder end suction/discharge volumetric efficiency % 4 Cylinder end capacities MMSCFD 5 Cylinder end flow balance No units 6 Cylinder end clearances % 7 Rod reversal Degrees 8 Rod loads klbf 9 Leak index % Table 2 - Key compressor cylinder parameters In addition to the trendable parameters, crank-angle curves for cylinder pressures and crosshead vibrations are acquired and stored for analysis.
CASE STUDY #4 Figures 23 and 24 show the eight day trends for Flow Balance and Leak Index for the compressor throws on Unit 13. A suction valve leakage on cylinder 1-CE began showing up at about 12:00 on May 20. Figure 23 - Flow balance trend for Unit 13 Figure 24 - Leak index trend for unit 13
Figures 25, 26 and 27 show the cylinder pressure Pressure-Volume plots along with calculated parameters for points in time before, during and after the suction valve leakage event. The valve was replaced during the shutdown that began on May 24 at about 7:20 am. Figure 25 - P-V curves before the suction valve leakage began Figure 26 - P-V curves during the suction valve leakage
Figure 27 - P-V curves after replacing the leaking suction valve Figure 28 shows the economic calculations from the Windrock MD analysis software. Since the gas that is being compressed with this Integral unit is propane that is used as refrigerant for another process in the plant; it does not have a direct economic value. However, the cost of the engine fuel does have a value. The replacement of the leaking suction valve reduced the Unit fuel cost by about 2.1% for an annual cost savings of approximately $4,000. Figure 28
INTEGRATION WITH STATION DCS In order to provide current information to the unit operators and to allow shutdowns on appropriate parameters, the engine/compressor monitoring system is connected to the Station DCS via MODBUS/TCP communications. Table 3 lists the parameters that are available on the operator s screens from the system: System parameters available to Operators 1 Unit average peak firing pressure 2 Engine balance 3 Combustion statistics 4 Unit fuel usage & BTU value 5 Compressor rod loads 6 Compressor rod reversal 7 Total compressor indicated horsepower 8 Total engine indicated power horsepower 9 Compressor flow balances & leak indices 10 Crosshead vibration 11 Frame vibration Table 3 - System parameters available to operators Figure 29 shows the display screen for the operators that displays live engine balance, PFP standard deviations and alarm status for all three units. The automatic shutdowns that are included in the DCS are High PFP pressures (to avoid detonation) and crosshead/frame vibration. The shutdowns on high PFP and crosshead/frame vibration are accomplished using the alarm relays on the monitoring system that are hard-wire connected to a digital input on the unit PLC. Figure 29
ECONOMIC BENEFITS REALIZED DCP Midstream estimates that they are saving about $250,000 per year, per unit for engine protection and lower maintenance costs. Longer overhaul intervals are projected to save an additional $150,000 per engine, per year. This is due to lower labor hours, lower plant excess emissions, improved efficiency, higher detonation margins, longer spark plug life and other factors (see appendix for full breakdown). With an estimated $276,000 capital cost per unit, the payback period for the systems is less than one year (around 8.3 months). Figure 30 details the capital expenditure of the online system. Figure 30 Capital Expenditure Summary of installing the online monitoring system If the current economic trend continues, the Net Present Value (NPV) of the online monitoring system will be more than $1.3M. The Profitability Index (PI) is the ratio of payoff to initial investment of a proposed project. It is a useful tool for ranking projects because it allows you to quantify the amount of value (cash) created per unit of investment. A ratio of 1.0 is logically the lowest acceptable measure on the index, as any value lower than 1.0 would indicate that the project s cash inflow is less than the initial investment. With a PI of 6.24 and a payback period of 8.3 months, this is a very attractive capital investment.
CONCLUSIONS In conclusion, DCP Midstream has become a significant supporter of utilizing automatic balancing and online monitoring on integral engines and compressors. The system installed on each of three integral compressors was a Platinum online monitoring system with engine AutoBalance. The low cost and ease of the system installation was designed to help the DCP Midstream site Robert s Ranch to reduce high maintenance costs and significant downtime. It would also alleviate the problems that occur with head and power cylinder failures due to chronic out-of-balance conditions. There were emissions benefits as well. DCP tried to lower emission limits from 3.0 g/bhp-hr to 2.0 g/bhp-hr but it was not sustainable prior to the AutoBalance installation. Presently, DCP is able to achieve between 1.2 and 1.8 g/bhp-hr emissions on both NOx and CO without having to tweak the controls or loading or wait until the coolest part of the day. A detonation is now a rare event, regardless of time of day or ambient temperatures. The first case study shows how the online system identified compression pressure problems in a cylinder. The head gasket was replaced and the pressures went back to normal immediately. In the second case study, the online system gives an alarm that the PFP was too low, which allows for an operator to troubleshoot and correct the problem. The third case study shows data that led to worn jet cells causing poor combustion events to be replaced. The combustion process was improved after replacing the jet cells. In case study four, the online system monitors Flow Balance and Leak Index for the compressor throws and shows a suction valve leakage. After replacing the valve, a fuel cost reduction was seen. These case studies and other information provide solid return on investment benefits realized from utilizing this technology. DCP Midstream recommends online monitoring to any company with this type of machinery, regardless of the industry.
APPENDIX Table 1 - Unit Specifications Engine Specifications Unit Model Rated HP (@330) Bore Stroke 11 & 12 C-B GMVH-10 2000 14 14 13 C-B GMVH-8 1600 14 14 Each of the units have 4 compressor throws and Tables 2 and 3 list the staging. Table 2 - Cylinder Staging for the GMVH-10 units Units 11 & 12 (GMVH-10) Cylinder Bore ( ) Service Suction Pressure Discharge Pressure 1 23.5 2 nd Stage Propane 50 PSI 180 PSI 2 12.0 1 st Stage Residue 200 PSI 600 PSI 3 9.25 1 st Stage Residue 200 PSI 600 PSI 4 29.5 1 st Stage Propane 5 PSI 60 PSI Table 3 - Cylinder Staging for the GMVH-8 unit Unit 13 (GMVH-8) Cylinder Bore ( ) Service Suction Pressure Discharge Pressure 1 9.25 1 st Stage Residue 200 PSI 600 PSI 2 9.25 1 st Stage Residue 200 PSI 600 PSI 3 9.25 1 st Stage Residue 200 PSI 600 PSI 4 11.55 1 st Stage Residue 200 PSI 600 PSI Table 4 - Field instruments installed Sensor Power cylinder pressure Stepper motor controlled fuel valves Field Instruments GMVH-10 GMVH-8 Description Quantity Quantity 10 8 AC engine pressure sensor installed on indicator valve at each power cylinder 10 8 Replaced manual balance valves at each cylinder with stepper motor actuated fuel valves controlled by the AutoBalance system Compressor pressure 8 8 Pressure sensor installed on indicator valve on each cylinder head-end and crank-end Accelerometer 6 6 Accelerometer installed on each compressor throw crosshead and each engine frame end TDC mag pickup 1 1 Installed at flywheel. One pulse per revolution Degree mag pickup 1 1 Installed at flywheel. 360 pulses per revolution
Table 5 Factors leading to economic benefits and savings Capital Cost Lower Plant "Excess" Emissions Lower Tons/BHP-Hr Exhaust Emissions (hitting 50% of Permit > lowers cost per year paid to state) Engine Protection (lowers maintenance costs) Improved efficiency (lower fuel costs/better emissions; 10%-15%) Higher Detonation Margin > Higher C2 Content in Fuel Longer Spark Plug Life; at least 2x (140 vs 70 days) Longer pre-chamber life Lower mechanic, EA and operator labor (hrs) $275,000 per engine $100,000 per year, per engine $2,000 per year, per engine $115,000 per year, per engine $20,000 per year, per engine $7,000 per year, per engine $2,500 per year, per engine $600 per year, per engine Protection and Maintenance Savings $247,100 Longer OH Intervals $150,000 per year, per engine Total Savings Estimate (per year) $397,100 8.31 Payback Period (Months) Figure 1 - Engine power cylinder pressures and mag pickups
Figure 2 - Compressor pressures and vibration Figure 3 - AutoBalance and fuel valves The GMVH-8 unit has a similar layout with 8 engine pressure sensors and fuel valves.