Improving engine performance

Recently, a Cooper GMVA integral reciprocating unit was converted from the OEM original configuration to “enhanced mixing” (EM) to comply with Louisiana Clean Air requirements. This conversion also incorporated a 22% horsepower upgrade.

A key part of this effort was a quality assurance process called the “advanced product quality planning” (or APQP) process, which evaluates the effects of a “design mode effect failure” analysis. This process plans for a number of potentially severe events, including:

• Crankcase explosions
• Seized piston
• Port carboning (intake)
• Detonation
• Pre-ignition, liner scoring, power piston failure
• Abnormally high PP angles (> 22° ATDC, or After Top Dead Center), which equates to a 15° backflow into the fresh air side
• A 55% increase in Standard Deviation (STDEV) (34 to 56 psi)
• A 200-degree increase in average exhaust temperatures
• High pre-turbine temperatures
• Premature ring and liner wear
• High air manifold temperatures
• Unable to balance units
• Unit very unstable at low loads – cannot run without auxiliary air.

Dealing with the changes

In Table 1, the white cells are direct analyzer inputs or outputs, and the gray cells are calculated values. A quick look at the Table shows the many parameter changes that take place when we add additional air.

After the initial installation of the Clean Air compliance equipment, a failure report study identified a number of issues that would cause the problems the station was experiencing. Because the unit upgrade was not previously modeled, we had no target parameters to compare to; so, no one identified the real issues. This lack of target parameters /modeling turned out to be very expensive in time, dollars, and wasted manpower.

We chased a number of perceived issues such as a severely limited air-path/flow within the unit, caused by internal passageways, air chest/cavities, and power cylinder internals that were almost completely plugged with carbon and oil.

Everyone thought that this blockage had caused a disruption of the air flow from cylinder to cylinder. The air chest acts as a pulsation dampener, and when it is clogged, as in this case, causes high pulsations to occur from cylinder to cylinder, which dramatically changes the cylinder to cylinder air charge/balance. This required the turbocharger to over-boost, causing late LOPP (location of peak pressure) high compression pressure, premature port carboning, and detonation. Once the restrictions were removed, it was necessary to lower the turbo head (AMP) by 17.5% which equates to a higher flow which enhances scavenge efficiency and gives more cushion to the detonation limit line. This over-boost condition should/would have alerted us to an issue if we would have had some baseline modeling/engineering data up-front to compare with.

It should be noted here that the entire natural gas industry as a whole has been willing to accept a non-performing unit (i.e., out of OEM specifications) in the name of Clean Air modifications. We should always strive to maintain OEM operating specifications; and if we are not, we should be identifying this information up front before the equipment is installed, and make our decisions from an informed platform. In short, we need to perform the up front modeling. In this case, once we cleaned the passageways and air chest, the engine performance parameters looked quite different, on the surface (Table 2).

Even though the data in Tables 1 and 2 appears to be acceptable, a warning must be given, “operator beware of average data.” Averages can and are very deceiving, and although this unit demonstrated favorable results using the averages, the fallacy in this approach became clear as we continued our analysis.

Engine compression ratio

This is a general statement that this author finds to be generally accurate: “The higher the compression ratio, the higher the charge density, and at equal flame speeds, the higher the heat release rate.” Over the years, this author has found that there are four compression ratio (CR) numbers that can be calculated. However, only the ECRboosted & effective figures are relevant to this discussion. But in order to get everyone on the same page, the following may useful:

1) CR_geo – Compression ratio static – a ratio based solely on the in3 of the cylinder plus clearance volume. (A geometric ratio that does not change unless a physical change is made to the cylinder.)

CR_geo = Max Volume in³ Min Volume in³

(NEED TO FORMAT EQUATION)

2) ECR_geo – The engine effective compression ratio (trapped/geometric) a measure of the CR from exhaust valve or port closing, and is a geometric measure. This value remains constant and will be less than CR_geo.

ECR_geo = Max Vol.@ valve closing in³

Min Volume in³

(NEED TO FORMAT EQUATION)

3) ECR_B – The engine compression ratio effective (trapped/boosted). This is a measure of CR from the port or valve closing of either a 2 or 4-cycle boosted/turbocharged engine. This value is not geometric, but remains constant throughout the boosted range and will normally be less than CR_geo.

^1/gamma

ECR_B = Power Cyl. CP (psia)

AMP_PSIA

(NEED TO FORMAT EQUATION)

4) ECR_DYN – The engine compression ratio effective (trapped/dynamic). This is a measure of the actual (real) compression being applied to the air charge from outside atmospheric pressure to in-cylinder compression pressure. This value takes into ac-count the compression being applied by the “boost” and the cylinder, including all associated temperatures.

^1/gamma

ECR_DYN = Power Cyl. CP (psia)

Atmos_PSIA — Air F. _DP-PSIA

(NEED TO FORMAT EQUATION)

The contention here is that the ECR_DYN is a contributor to engine health (knock limit), and it is an identifying “leading indicator” of engine health which should be monitored, further researched, and documented. It should not be ignored. Figure 1 and Table 3 allow users to estimate ECR_DYN, although inter/after-cooling has an effect on the final number. It is a close estimation, as can be seen in the comparisons at the bottom of Tables 1 and 2.

Experience shows that it has been difficult to get a good “feel” for the limitations or boundaries on some slow-speed units. We can, however, use these numbers to effectively monitor and track the relative changes in the units from both the modification side and any performance changes the unit might undergo over time. By building this database, it should be possible to predict where any abnormal engine behavior will occur, and either avoid this outcome; or recognize it, so it can be used for troubleshooting and failure avoidance.

STDEV and its effects

As can be seen in Figures 2a and 2b, the STDEV on this unit has suffered considerably. It went from 34 to 60, a roughly 76% increase. Because it is some times difficult to explain pure numbers, a visual representation often helps. So, in Figure 3, we can see what STDEV actually looks like, and its effects on engine integrity. It depicts results from a test conducted by SwRI on the DOE infrastructure project. It can be seen that when the unit was both balanced and not-balanced using the peak-firing (PFP) method that “ustrains” – crankshaft deflection damage large enough to cause permanent distortion to the crankshaft – were occurring every 20 minutes or less.

When we balanced using the coefficient of variance (COV) methodology (which minimizes STDEV), it could be seen that there was a steady rise in time between events, to over 100 minutes. This dropped off again when the unit was returned to “automation mode,” and PFP balance resumed. So what causes the STDEV increase in the enhanced fuel mixing scenario? In effect, one could say that the medium/high-pressure systems work too well. These systems completely mix the A/F to a point that there is no stratification and thus no rich pockets for the spark to initiate from. The only way at present to overcome this is with pre-combustion chambers (PCCs).

First, one has to accept the fact that these “old thumpers” were originally designed to have a stratified combustion charge. This involved the creation of a rich pocket of gas at the dome of the head surrounding the spark plugs, which became progressively leaner as you go towards the piston crown. This rich pocket allowed the spark kernel to initiate and propagate to the cylinder outer walls. This was particularly important in slow-speed, large bore units. Secondly, the leaner the charge, the greater the misfires; the greater the STDEV.

In and of itself, STDEV is one of the more damaging aspects of the whole combustion process, as demonstrated in Figure 3. In the past, the engine analyzers reported the STDEV as an average, while the newer units reported the high, low, and mean. This really helps in the interpretation of the data. In addition to the cyclic damage, the STDEV also affects the LOPP. The LOPP has a big influence on the A/F ratio and cylinder scavenge efficiency. STDEV also is directly related to PFP, as well as the forces exerted on the power piston and bearings.

As Table 4 illustrates, the STDEV cyclic characteristics can and do cause high pressure variations and excessive bearing loads. Viewing this table, one can see that:

• The unit prior to conversation produced a much better combustion consistency – 66 of every 100 cycles fired within the + 5% band, with a 135 psi spread.
• The Clean Air converted unit only strikes within the + 5% band 43 out of every 100 cycles, with a 291 psi spread.
• The “bearing loading” spread for the original was 23,070 psi for each 100 cycles; the converted unit went up to 44,778 psi/100 cycles.
• If we would have installed PCCs as our first choice, we would have improved the STDEV to 20 or lower (based on similar testing results), which is 80 out of 100 cycles within the + 5% band. This would have allowed us to reduce the bearing load deviation to approximately 13,519 psi/100 cycles, with all other performance parameters remaining roughly the same.

Looking at Figure 4, one can see the differences in combustion stability that the unit experiences. It is not hard to justify or come to the conclusion that a STDEV of 20 psi is much better than 60 psi. Additionally, this endeavor revealed that the oscillating forces being applied by the excessive STDEV will take a toll on engine integrity. This is especially true for heads, bearings, piston crowns, piston pins and rings, liners, and crankshafts, and can also be seen in out-of-specification LOPP.

As can be seen here, late LOPP can be very detrimental to the overall unit, both in performance and unit integrity. An LOPP of 22 equates to 15 degrees of back-flow, which contributes to the extreme carbon buildup, backpressure, high exhaust temperatures and component failures.

These highly automated “enhanced gas delivery” systems have the potential to make our legacy equipment not only much more Clean Air friendly, but also improve the overall operation and flexibility, if engineered and incorporated correctly into our system by intelligent design. The LOPP is ideally at 12 to 15 degrees ATDC. Interestingly, the aftermarket industry (not the OEMs) have moved this number to 17 to 24 to facilitate the “enhanced mixing” conversions.

One additional thought is to be aware of the system complexities and the ability of station personnel to engage in operations, troubleshooting and repair activities. Many pipeline companies have been forced to deal with reduced manpower. When we install equipment that requires more attention and dollars to maintain, the realities need to be explored upfront.

Conclusions

Have your engine performance operating parameters modeled upfront before the conversion, then you will have something to compare against after the modifications are completed. Additionally, if there are modifications identified in the modeling effort, those parts, modifications, and engineering work can be completed before the unit conversion begins.

The up-front parameters would include (but not limited to): a. Turbocharger ratio b. Air manifold pressure (AMP) c. Air manifold temperature (AMT) d. Peak firing pressure e. Compression pressure f. LOPP g. Ignition timing h. ECRB – Effective compression ratio-boosted i. ECRDYN – Effective compression ratio-dynamic j. STDEV/STDEV spread.

Through this experience, it became evident that we need a process. Advanced Product Quality Planning (APQP) is a quality framework used for developing new products in the automotive industry. It can be applied to any industry, and is similar in many respects to the concept of design for six sigma (DFSS). It does this by focusing on upfront quality planning, and by evaluating the output to determine if goals have been met. The process consists of four phases and five major activities, along with ongoing feedback assessment and corrective action.

Unless there is some overwhelming or compelling reason not to, this author advises that we utilize the PCC methodology as our first line of modification. In addition to NOx reduction to the ~1.5gr or lower area, the PCCs also give us good combustion stability, i.e. STDEV, which translates to a smoother running unit as well as unit integrity. In general, the technology is more mature and causes few startup issues and maintains the OEM operational specifications. Yes, there are still check valve issues, but these can be managed. This author is not aware of any engine type that does not benefit from using PCC technology.

Operators should also acquire accurate pre-conversion data and compare it to OEM specifications. Get the unit into OEM specification before any modifications are completed. If the unit will run within OEM specifications, then the question of unit integrity (carboning, correct parts, port heights, cams etc.) is pretty well answered. If it will not, then maybe a complete topside rebuild is in order.

Do not field modify any OEM part (cylinder, liner, heads, etc.) without a comprehensive modeling effort. You need to have a good handle on how any modification would change the basic engine performance parameters. Also, high STDEV/COV is bad, it equals abnormal wear and tear which equals safety, reliability and maintenance costs.

Acknowledgment

Based on a paper presented at the Gas Machinery Conference, held in Dallas, Texas, October 1-3, 2007.