How Comprehensive Diesel Engine Analysis Improves Performance and Reliability
Edward P. Kelleher, Windrock, Inc.
This paper presents a proven technology and approach to reducing unexpected engine failures and optimizing performance for all types of diesel engines. Utilizing comprehensive engine analysis and monitoring techniques, a detailed view of the condition of internal parts such as valves and valve train components, injectors, fuel pumps, liners, rings, and bearings can be made in a non-intrusive manner to identify any degradation or change before a component fails. A detailed explanation of the technology as well as case studies from a variety of different engine models will be presented in this paper.
Printed with permission from CE Industry Journal. See the full issue here.Download the White Paper
This paper presents a proven technology that has been utilized in many industries for decades to reduce engine failures and optimize performance on high, medium, and slow speed diesel engines. Engine owners and operators have historically performed engine OEM maintenance activities, oil analysis, and typically obtain indicator card measurements or peak firing pressures on a periodic basis. Even with these procedures and processes in use, diesel engines still have unexpected failures on a regular basis between these scheduled or planned maintenance activities.
Although the cost of fuel today is more than 50% of what it was just a few years ago, fuel is still the biggest cost of operating an engine in the power generation arena. Additionally, when an engine fails un-expectantly, there are lost production costs, parts and labor costs and in some cases, regulatory fines.
Comprehensive Engine Analysis employs the use of traditional engine performance measurements (similar to traditional indicator cards), it also utilizes crank-angle based, degree-by- degree, electronically collected vibration and ultrasonic data. Full utilization of the digitized cylinder pressure data combined with phased vibration and ultrasonic signatures allows for a complete non-intrusive assessment of the internal cylinder components and identification of degraded components.
Most power plants today will use some form of combustion analyzer, whether it is the older indicator card type, a MEP device, a simple electronic peak pressure device or crank angle based combustion analyzers to assess engine balance. The author would ask whether this information is utilized in full and more importantly is it acted on.
The engine indicator and Indicator Card go as far back as the late 1700’s as shown in Figures 1 and 2.
Fig. 1-Engine Indicator
Fig. 2- Indicator Card
The use of electronic or digital engine analyzers began in the late 1980s and started to incorporate analysis software. Peak Pressure gauges are still widely used, however, they provide very limited information and, in some ways, is a step backward from the indicator measurement technique.
With the help of advanced software, more detailed measures are available to an engineer responsible for the safe and economical operation of an engine. Statistical analysis, such as standard deviation or the measure of how consistent a cylinder fires, is a key measure of fuel delivery to the cylinder. For example, a standard deviation that is too high would indicate inconsistent fuel delivery. Another measure of importance is the first derivative curve from the standard Pressure-Time (PT) curve or the maximum pressure rise rate curve and the resultant peak value. This parameter will indicate when fuel is delivered (early or late) and how fast that fuel is burned during the uncontrolled combustion phase.
From diesel engine combustion first principles and the four phases of combustion, Phase 1 (ignition delay phase) and Phase 2 (Rapid or Uncontrolled Combustion Phase) are inherently the most important. In fact, only looking at the peak pressure can be misleading and reliance on exhaust temperatures can be completely inaccurate.
Phased Vibration and Ultrasonic Data
The old way and how this engineer learned “vibration monitoring” was the screwdriver or rubber hose to the forehead to attempt to identify the source of a noise. This, as we know, is very subjective and does not provide clarity to the actual issue.
Using an accelerometer and/or ultrasonic microphone that is band filtered and referenced to crankshaft position a comprehensive and detailed picture can be obtained and analyzed.
As discussed later, this type of vibration analysis is not your typical FFT or Spectrum data that can bring fear and confusion to the non-vibration analyst. Phased Vibration is very clear and concise as to when a mechanical component has degraded or failed.
The following cylinder “signature” (Fig. 3) presents the in-cylinder mechanical data very clearly. With Top Dead Center (TDC) represented in the middle of the plot, the sharp mechanical impacts to the left are the exhaust valve (s) and intake valve (s) closing. Just prior to TDC, the mechanical impact is the fuel injector opening. Finally, to the right of TDC is the exhaust gas noise from the exhaust valve opening.
Fig. 3- Cylinder Signature Example
The above signature, although not familiar to you, is very easy to interpret. Either the valves open and close “when” they are supposed to and with an amplitude like the other cylinders or they don’t. When the don’t, that indicates a difference or change. If the injector does not pop when expected, normal combustion cannot be expected. Further, when unexpected mechanical impacts occur where none are expected, that also indicates an issue. As the data is crank-referenced, the likelihood of identifying the failed component is greatly increased.
Reliability and vibration engineers are likely to be familiar with the P-F curve as it is a well-documented approach that illustrates the behavior of equipment as it approaches failure. This same “curve” and concept for rotating equipment can be applied to Diesel engines. In the example, a failure starts manifesting, the equipment deteriorates to the point at which it can be detected, point (P). If the failure is not detected and mitigated, it continues until a “hard” failure occurs at point (F). This is the window of opportunity during which engine analysis and resultant inspections can likely detect the imminent failure and address it under a planned maintenance activity. The following (Fig. 4) represents a P-F Curve for an unplanned injector “adjustment”.
Fig 4-PF curve for an injector failure
An example of not performing engine analysis post overhaul.
Fig 5-Damaged Head
This damage (Fig. 5) was concluded to be the result of not properly tightening the upper side of the rocker arm. This failure occurred 30 days after a major overhaul and resulted in a $600K failure, as the turbocharger was damaged.
Comprehensive diesel engine analysis and monitoring allow for the early detection of faults utilizing both vibration and ultrasonic detection, albeit in a slightly different manner from traditional vibration techniques. When components in an engine degrade, their mechanical behavior changes which can easily be detected with these technologies when implemented correctly.
Another graph that illustrates the economic benefit of comprehensive engine analysis as part of a condition monitoring program is the Bathtub Curve (Fig. 6). With Preventative Maintenance (PM) strategies in place, these will only “catch” or mitigate the age-related failures which represent 11-22 % of failures (Allen-2006). Whereas Condition Monitoring (CM) or Advanced Engine Analysis can be applied to identify 21% -77% of the random or unexpected failures (Moriarty and Chauvin, 2016) indicated in the Useful Life section of the curve. Further, with a CM approach, post-maintenance testing is performed to identify improper maintenance and reduce the failures from maintenance induced maintenance.
Fig. 6-Bathtub Curve
A Maintenance Induced Maintenance Failure
Following a routine maintenance activity and the subsequent surveillance runs which included the performance of engine analysis on the Emergency Diesel Generator (EDG). It was noted that Cylinder 12 had a peak firing pressure approximately 500PSI lower than the engine average, however the exhaust temperature for cylinder 12 was above average.
Fig. 7-Good Condition
Fig. 8-Misfiring Cylinder
Fig-9 Engine Report with dead cylinder
Further analysis of the phased vibration and ultrasonic data identified late injection as the cause of the low pressure.
Fig. 10 Poor Cylinder
Once an inspection of the fuel pump on Cylinder 12 was conducted, it was easily identified that the “pant leg washer” (Fig. 11 and Fig. 12) and retaining bolt were not properly installed which allowed the pump timing adjustment retainer to fall off. This led to the pump timing change, resulting in the fuel pump timing becoming retarded. Thus, the retarded timing caused the cylinder combustion to start very late, not reaching even compression pressure. However, it created a normal to slightly elevated exhaust temperature 860 Deg. F vs the engine average of 846 deg. F.
Fig. 11 Pump Housing
Fig. 12- Loose Items
A Detailed Review of Cylinder Pressure Data
Figure 13 represents a typical Pressure-Time (PT) Curve and the applicable parameters that can be obtained and utilized for analysis.
Fig. 13-Pressure Time Curve
Figure 14 represents a normal Pressure-Volume or PV Curve. The PV is used to calculate the Indicated Horsepower (IHP) and Indicated Mean Effective Pressure (IMEP).
Fig. 14-Pressure Volume Curve
Figure 15 represents the First Derivative Curve and PT Curve. The First Derivative curve is where the maximum pressure rise rate is determined. This curve is sometimes called the Rise-Rate Curve.
Figure 15- First Derivative Curve
As there is a significant amount of pressure data and calculations to analyze, a summary report is required to capture the important information in a table format. This allows for a very quick assessment of the overall engine performance as well as individual cylinder performance. For example, is the engine “balanced”, is each cylinder firing consistently from the standard deviation calculation or are any of the cylinders exceeding the maximum allowed peak pressure value?
For a Diesel engine, the SD value typically should be in the range of 1% of the average peak firing pressure. When this value is too high, it would indicate a fuel delivery problem.
Engine OEMS have or should have very defined acceptance criteria for engine balance, maximum allowed peak pressures as well as standard deviations.
A typical engine report from the collected pressure data will show items such as IMEP (Indicated Mean Effective Pressure) or Mean Indicated Pressure (MIP), IHP (Indicated Horsepower), Combustion Start Angle, Maximum Pressure Rise Rate (MPRR), Peak Firing Pressure Statistical data from at least 30 cycles, the Delta or difference between the cylinder average and the engine average. Peak Firing Pressure Angle (PFPA), as well as other pressure points along the P-T curve have important value. Typically, your engine OEM manual will represent this acceptance criteria as ±5 bar from the certified value at any one cylinder.
The author frequently reviews combustion-related data where measured plant data is nowhere close to these acceptance criteria, as much as 28% in some instances. Fig. 16 is an example engine report with good peak pressure balance.
Fig. 16-Engine Combustion Report
When collecting and analyzing vibration data from a reciprocating machine, such as a Diesel Engine, there are two basic types of vibration – Free Vibration and Forcing Function Vibration.
Free Vibration is created after a structure vibrates in response to an input excitation and vibrates freely at its natural frequency. This would be similar to a bell being struck by a hammer (Fig. 17). Examples on an engine would be the impact measured in “g” or acceleration of a valve hitting the seat when it is closed.
Another example is when a valve opens and high pressure gas exits through a small opening causing a vibration (Fig. 18).
Fig. 17-Mechanical Impact Example
Fig. 18-Gas Noise Example
Different engine components and their related faults are determined by utilizing different frequency ranges and/or filtering in the system hardware.
When this type of data is “phased” to crankshaft position, the component making the noise can be determined and assessed as to whether the noise is normal or abnormal (Fig. 19). This analysis technique applies to 4-stroke, 2-stroke or high, medium or slow speed engines. In each case, the data acquisition and interpretation process remains the same.
Fig. 19 Multiple Cylinder Vibration Patterns
Once proper data collection is complete, the data will allow the engineer to identify normal behavior and abnormal behavior. Examples include, but are not limited to, the following:
- The condition of the intake and exhaust valves and/or ports
- The timing for all valve and port-related events
- The condition of the fuel injector
- The condition of the rings and cylinder surfaces
- If piston slap is indicated
- The identity of all external leaks
- If bearings, pins and bushings indicate impacting
In Figure 20, an example is shown of a complete cylinder signature with different locations and frequency ranges:
Fig. 20-Abnormal Impact on Cylinder Signature
The arrows (moving from left to right) point out the Exhaust valve closure (1), an abnormal “mechanical knock or impact” (2), Intake valve closure (3), Fuel Injection occurring (4), and Exhaust valve opening and the blowdown event (5).
The different patterns or vibration signatures are High Frequency (Top), Ultrasonic (2nd) as well as Raw vibration (3rd) and Ultrasonic, specifically on the fuel injector or fuel line (Bottom).
The ultrasonic frequency (35-45 KHz), is a very useful frequency for establishing when a fuel injector “opens”, without having the expense or safety issue of installing a valve and pressure transducer into a high-pressure fuel line.
During evaluation, multiple cycles of data are collected for each cylinder and are compared for consistency and the signatures are compared across the engine or banks (Fig. 21).
Fig. 21- Multiple Cylinder High Frequency Vibration
The previous example in Figure 16 is of the High Frequency Test point collected on the cylinder head stud. It is evident that the exhaust valves are not all closing at the same “time” and that some of the cylinders have a mechanical impact that is unexpected.
Reviewing historical data can be very useful in identifying a change to a component and the severity. As components wear, this change can be seen over time as amplitudes get bigger or smaller for a known event. Going back to the P-F curve concept, having trendable data will help determine the rate of change for certain faults and whether immediate action is required.
The other type of vibration commonly utilized is a Forcing Function Vibration. A forcing function vibration is created when a structure vibrates as a result of a periodic external excitation force. The structure vibrates at the frequency of the forcing function. An example of this type of vibration is frame motion due to misalignment or in imbalance in a rotating component. Typically, FFT spectral analysis is used for this type of analysis (Fig 22).
FFT is not very useful in identifying internal issues on reciprocating machines. The following (Fig. 22) is a typical FFT plot with the y-axis in g and the x-axis in Frequency or Hz collected from a cylinder head. There is not much “diagnostic” information in such a plot as only an overall vibration level can be measured. If the overall level changes, this type of data will not identify which component has changed.
Fig. 22-Typical FFT or Spectrum Pattern
Case Study #1-Broken Component
In this case study from a high speed, 2-stroke engine that is an Emergency Diesel Generator (EDG) at a nuclear power plant, the data is taken during a regularly scheduled surveillance test. There was not any concern or awareness of any problem with the engine. Basic engine parameters were available to the control room staff (temperatures, speed, load, etc.). Cylinder Pressures and Phased Vibration data were collected as part of normal plant procedures during this test.
However, when reviewing the combustion or engine report (Fig. 23) for this test date, it is clear that Cylinder 1 Peak Firing Pressure is quite low. However, the Exhaust temperature is on the high side. The only parameter visible to Operations is exhaust temperatures and, although indicated as “HI”, these values are within the allowable range per the operating specification. It should also be noted that, although the STDDEV (Standard Deviation) is the lowest, this would be attributable to the low firing pressure. Also note the PFP or Peak Angle is earliest as indicated here at 5 Degrees after TDC. In reality, the cylinder is firing late and low as shown in Figure 24.
Fig. 23-Engine Combustion Report
Examining the PT Signature for Cylinder 1 (Fig. 24), it is obvious that there is an abnormal “peak”. The peak pressure is very low and late. This is likely caused by a fuel injection or delivery issue. Interestingly, the low peak firing pressure would normally produce a low exhaust temperature, however, because it is late (relative to crank-angle), it causes the temperature reading to increase, masking the problem entirely.
Fig. 24- Cylinder 1 PT Curve
In Figure 25, all of the median PT Curves are stacked on top of each other which makes it very easy to see an abnormal cylinder.
Fig. 25- All Cylinder PT Curves Stacked
When the Phased Vibration trace (Fig.26) is included, it is now very clear that the fuel injection occurs around TDC which is ~12 degrees later than it should for this engine type. As can be seen, it is opening as indicated by the vibration impact at essentially TDC position. The first area to focus on is the fuel delivery system (Injector, Rocker Arm, etc.)
Fig. 26-Phased Vibration and Pressure Time Curve
With the above guidance, the maintenance personnel performed a targeted physical engine examination and identified the cause of the bad cylinder performance as shown in Figure 27.
Fig. 27-Broken Component from rocker assembly
After repair, the pressure value returned to normal as did the Pressure trace and Vibration trace, as indicated in Figure 28 below.
Fig. 28- “As left” Engine Combustion Report
Fig.29-“As Left” Phased Vibration and Pressure Curves
Summary: Without the pressure data, this fault would have gone unnoticed. This case study also highlights why exhaust temperature monitoring can be misleading and actually mask a real problem which can and does lead to improper engine adjustments. The vibration data allowed for quick, pinpoint troubleshooting. As this is plant critical safety related equipment, identification and repair occurred in a short period of time and did not necessitate having to power down the reactor.
Case Study #2-Missing Parts in a fuel pump
Following a periodic engine analysis activity, it was immediately noted by the analyst that Cylinder #2 had a peak firing pressure approximately 134 PSI lower than the engine average, the peak firing pressure was early ~3 degrees and exhaust temperatures were approximately 180 degrees F lower than engine average (Fig. 30).
Fig. 30-Engine Report
The fuel pump ultrasonic data (Fig. 31) identified a mechanical anomaly in the fuel pump.
Fig. 31-All Cylinders Ultrasonic Data
After a root cause analysis and complete disassembly and inspection of the fuel pump (Fig. 32), it was revealed that two internal components were not installed during the last pump rebuild.
This detailed inspection revealed that the Fuel Delivery Valve Stop and Fuel Delivery Fuel Valve Spring were not installed.
Due to the design of this Fuel Delivery Valve, there was enough fuel being delivered for the cylinder to fire but at a reduced pressure. It was also noted that there was a through wall crack in the Delivery Valve Assembly. It is believed that this crack was caused by high pressure being built up in the pump due to the restricted flow area to the fuel injector.
Fig. 31-Fuel Pump-Picture courtesy of MPR Associates
Case Study #3-Valves Hitting Piston
Operations and maintenance personnel identified a mechanical “ticking” sound coming from Cylinder L1. The plant’s peak pressure monitoring also identified cylinder L1 has having the highest peak pressure at 11% higher than the engine average (Fig.32). The increase in pressure was noted after a maintenance activity where the head was removed and reinstalled.
Comprehensive engine analysis was performed which determined that the mechanical ticking was occurring just prior (1 deg.) to Top Dead Center (TDC) on the scavenging stroke (Fig. 33). With the piston at or very near TDC on the scavenging stroke is when both the exhaust valves and intake valves are open. As a result of the analysis performed, a maintenance activity was performed which indicated the valves where touching the piston at TDC. It was also determined that a 1mm shim or spacer was installed, and not the required 2mm spacer.
The diesel generator at this station is one of many at this location that provide power to a Caribbean Island. The engine is a 16 cylinder Sulzer 16ZV40/48 rated at 11,125 BHP. Each cylinder has a 400mm (15.748”) bore and a 480mm (18.989”) stroke and operates at 514 rpm. This engine also employs a rotating piston crown design.
As shown in the engine report below (Fig 32.), it should be noted that even though the peak pressure and compression pressure were the highest, the peak pressure angle was earliest and the exhaust temperature was below the engine average.
This is caused by the higher compression pressure (17%) due to the smaller clearance volume which in turn results in earlier combustion and a higher peak pressure.
Fig. 32-Engine Report
When the cylinder 1L head was removed on Engine E8, it could be seen that the piston crown had a polished ring with no carbon on it indicating that the valves where hitting the piston each scavenge stroke and as the piston crown rotated it kept a ring free of carbon. The valves also exhibited damage. Due to the damage to the valves from being impacted, all were replaced and same head re-installed with the correct 2 mm. After this maintenance, the engine noise was eliminated and peak pressure monitoring indicated the peak pressure was back to normal.
Fig.33 Vibration Signature showing the impact
Utilizing a comprehensive engine analyzer allows for the quick and reliable identification of a fault before the fault creates a catastrophic failure. As these valves were constantly being impacted by the piston, the risk of a valve failure (dropped valve) was very high. In addition, the higher and earlier peak pressure places more strain on the bearings, more stress on the cooling and lube oil systems as well.
It is also very clear from this event that exhaust temperatures are not a reliable indicator of cylinder health and performance
Case Study #4-Lose Fuel Injector Adjustment Lock- Nut
This engine is a two-stroke EMD 645-16 cylinder engine in nuclear emergency diesel generator service. In this example the engine was being tested after a maintenance activity per plant procedures.
As indicated in the report below (Fig. 34), cylinder 11 has the lowest peak firing pressure, the earliest peak pressure angle and the highest exhaust temperature. If only Exhaust temperatures were monitored there would be no compelling evidence, there was an issue as the spread of 130 degrees is within normal acceptance criteria.
Fig. 34-Engine Report
In the two following plots (Figs 35 and 36) the first being the High Frequency Vibration signature and the second one the Ultrasonic Frequency signature, it can be seen clearly that the injector does not “Pop” at the expected time and is extremely late.
Fig. 35-High Frequency Vibration
Fig. 36-Ultrasonic Frequency Vibration
The Exhaust Temperaure bar graph for the right bank is normal as shown in Figure 37.
Fig. 37 Exhaust Temperatures
Upon inspection, it was found that the Locknut the secures that Adjust screw had not been torqued correctly and was found to be loose (Fig. 38 and 39). This allowed the adjusting screw to thread-in which changed the timing of injection. As a result, fuel was extremely retarded resulting in a low and late combustion. The early peak is from compression as the “peak firing pressure” was so late is was below compression pressure. This also led to the slightly elevated exhaust temperature.
Fig. 38 -Rocker Arm Assembly
Fig. 39-Rocker Arm Assembly
And then there is the rest of the story. In the process of correcting this maintenance induced maintenance issue, instead of just correcting the timing, the injector was replaced. During this replacement, the fuel lines were not corrected properly.
This, in turn, resulted in a fuel leak, which was not detected immediately and ultimately let to a complete crankcase lube oil dilution with fuel oil. Therefore a simple maintenance error resulted in significant corrective maintenance actions.
This paper has presented that comprehensive engine analysis and monitoring can benefit the operators of large diesel engines by providing technology that can help avoid engine failures and optimize engine performance with targeted corrective actions. This technology is well proven and is utilized in many other applications (Oil & Gas, Military, and Power Generation, as well as critical applications such as Nuclear Emergency Diesel Generators).
By monitoring engines in a comprehensive manner, small mechanical issues can be identified early on before a small issue causes a more significant engine failure.
The benefits of operating an engine at its optimum condition saves fuel and reduces emissions as well as providing the maximum ”up-time” and availability.
Whether the goal or outcome is to reduce fuel consumption by ensuring optimized operating conditions or to have the parts and personnel available to perform maintenance before a failure, even a “minor” one occurs, the financial benefits can be measured in the hundreds of thousands of dollars.
The reality is that conducting this type of monitoring and analysis on a regular basis is not hard or complicated, nor is it expensive considering the savings present. Whether companies conduct this “in-house” or collect data and send it to others for analysis, there are proven benefits to comprehensive engine analysis.
Kelleher, Economic and Mechanical Benefits of Utilizing Advanced Diesel Engine Analysis and Monitoring Techniques, Paper SMC-089-2016, SNAME 2016
C L Haller and E P Kelleher. ‘Practical Integrated Maintenance and Diagnostics for Medium and Slow Speed Diesel Engines.’
I Mar E Conference on Computers and Ships From Ship Design and Build, Through Automation and Management and on to Support, 1999
Heywood-1988 Internal Combustion Engine Fundamentals MPR-LTR-2062-0002-01, September 3, 2014
About the author
Edward P. Kelleher is the Global Business Development Manager-Diesel for Windrock Inc. (subsidiary of the Dover Corporation). Mr. Kelleher is a licensed Third Assistant Engineer (Steam and Motor, Unlimited Horsepower) and has Bachelor of Science in Marine Engineering from the Massachusetts Maritime Academy. Mr. Kelleher has 30 years of experience in field engineering in naval, nuclear, rail and power generation industries, with 20 years of condition monitoring experience related to reciprocating engines and compressors in various industries. Mr. Kelleher has published papers on condition based maintenance and has presented to a variety of audiences worldwide.
Windrock specializes in the development, manufacture and distribution of online monitoring systems, portable analyzers, software, sensors, and technical services for reciprocating engines, compressors and rotating machinery. Founded in 1996, Windrock, Inc. is headquartered in Knoxville,
- Windrock products are used worldwide by operators, engineers, and maintenance personnel to analyze, monitor, trend, alarm, and automatically diagnose the mechanical condition and performance of reciprocating engines, compressors and rotating machinery