Editor's note: In the May issue, CT readers had an opportunity to get a snapshot of results obtained during our first engine dyno session, using the GM CT 525 project engine. Because of the abundant test and test-related information gathered, we chose to provide two levels of technical analysis; one developed as an overview (in May) and a second taken to a higher level of discussion involving reasons on which the results are based (in this article). While this latter analysis is intended to be more of a "learning experience" than observational, we believe it will provide a more in-depth understanding of the engine's performance. This is an important consideration because the next dyno session will include in-cylinder analysis of the combustion process using alternative fuels. This is another bold step toward urging circle track racers and engine builders to step into the "green racing" box with us.
What might cause a difference in brake horsepower when switching from carburetion to electronic fuel injection?
Of several reasons, two stand out in particular; intake manifold design and fuel atomization efficiency throughout the rpm range. This observation includes the fact that some degree of combination optimization will tend to minimize the differences but, from a practical and theoretical perspective, the fundamental differences should still exist. Here are some thoughts to consider.
In prior Circle Track editorial material, we've proposed that flow passage cross-section area (intake or exhaust) has a pronounced effect on flow rate. And even though both intake and exhaust systems function in the presence of unsteady, compressible flow, it's still possible and desirable (for purposes of mathematical calculations involving intake and exhaust system modeling) to deal with what we'll term "mean flow velocity." Given this notion, we can also relate mean flow velocity to engine speed.
Consider that there will be a mean flow velocity associated with any given rpm, with a value in the 240-250 feet/second range at peak torque. In fact, we've previously reviewed this concept that we'll now apply to the two manifolds in this project, essentially in terms of passage section area and length. Recall the notion that section area determines mean flow velocity (all else being equal) passage length tends to "rock" the torque (volumetric efficiency) curve about the point of peak torque rpm.
Simply put, shorten runner length and the torque curve "rocks" in a counter-clockwise direction. Lengthen it and the opposite occurs, notably adding torque below the peak and removing it from above (again all else being equal). And while it's reasonable to assume that some of this shifting in torque could be addressed by companion parts changes (notably camshaft), the purpose of our discussion is to compare fundamental differences in manifold design, single 4V carburetor to EFI, as applied to the project at hand.
You may want to refresh your memory on the comparative torque curves shown in May (Figure 1). Since the same set of headers was used for both induction system tests and applying the little equation previously shared in CT "Enginology" materials, we ran the numbers on where the header contribution to the net torque peak should have been, calculating a value of 5,216 rpm. Eyeballing the peaks shown in the graph, you can see the indicated peaks are very close to 5,400 rpm. This suggests that both intake manifold section areas are about the same, but larger than that of the header primary pipes. For purposes of this discussion, that's immaterial.
Figure 1 - This is a dummy caption...st of the housing into place, install the ring gear b
Figure 2 - Time-weighted, steady-state test points (data supplied from a Late Model stock
Figure 3 - Cycle-weighted, steady-state dynamometer emission results. Individual steady-st
Figure 4 - LS3 CT 525 engine, full-load power pull results and comparison of E85 to 100-oc
What is significant and absent any specific section area data comparing the two manifolds but focusing on the material differences in runner lengths, you can visualize how this influences torque output by again viewing the data plots. Specifically, if you will focus on the overall curve shape above and below peak torque rpm, you can see how the EFI manifold's curve was rotated slightly clockwise, compared to that of the carbureted combination. The result was improved torque below its torque peak and a slight loss above this point, while the opposite held the carburetor package. However, you will also observe that the quantitative improvement in torque from the EFI manifold below its torque peak clearly offset losses above the peak, given where an engine of these torque characteristics would like be geared to run.
OK, that's the first reason. The second deals with the fuel quality issue mentioned in May, particularly inherent differences in the conditioning of fuel delivered by fuel injectors versus carburetors. Of course, there are multiple factors involved in how carburetors and FI affect air/fuel charge quality.
It's the very nature of their design and functional differences. We'll just summarize a core difference by mentioning again that if you view how carburetors produce atomization efficiency over a broad range of engine speed versus that of fuel injectors, carburetors are comparatively lacking. This deficiency is particularly evident within the combustion space, cylinder to cylinder, during real-time combustion pressure and temperature histories.
As pointed out elsewhere in this story, the CT team is scheduling a round of engine dyno testing focused on engine cycle analysis (ECA) techniques that's intended to characterize the burn properties of carburetion versus EFI. These tests will be conducted in an academic, scientific environment void of any bias or presumptions about what the data will show. We encourage CT readers to continue following this project for those combustion-related results.
Are there any material differences in torque characteristics when comparing carburetion to EFI systems?
We included this question to emphasize the importance of considering functional differences between the two manifolds used (as previously discussed) while sharing some comments about optimizing companion parts. For example, an honest appraisal of the two induction systems should include exploring what configurations of other parts would enhance either system, particularly the camshaft and headers.
In other words, it's likely that using a cam and header set intended to operate in the engine speed range for which either the carburetor or EFI manifold was designed to operate should help both. As pointed out at the outset of this project, a principle CT goal was to explore viable possibilities for a greener approach to circle track racing. We intend to create a level of awareness whereby racers who are in the sport or business of racing can pick up where we leave off.
Prior to conducting engine tests, we realized that the critical step in any engineering evaluation is to ensure "apples-to-apples" comparisons of the data at hand. Circle Track's green racing team ensured a baseline comparison that could withstand scrutiny through honest and realistic testing procedures by using the resources available at every stage of the project. There was an internal agreement that, regardless of the results, data would be openly shared and published for the reader to review, discuss, and comment on.
In this environment, the first step in the dynamometer portion of the testing required the team to determine appropriate engine speed/load points for the tests. One method of doing this is to select a weighted series of steady-state points under which the engine operates during a given drive cycle, focus on optimization, and development of those points, and use them as an approximation of what the engine would experience in the real world.
Figure 5 - Individual, steady-state points, lambda values. Each number on the X-axis corre
The team was able to collect engine rpm, throttle position, and time utilization data from a late-model stock car making laps at New Smyrna Speedway and apply that information to the CT 525 LS3 test engine. These data were then analyzed to determine the frequency at which the engine operated at specific speeds and throttle positions, and time-weighted histograms at each one of the points were determined. These throttle positions were then used on the dynamometer with the test engine, and the brake torque employed for each run was recorded to be used throughout the testing.
In this case, there were seven test points chosen, as shown in Figure 2. Given these conditions, one can see that for approximately 80 percent of the time, the engine experiences wide open throttle (WOT) conditions, with only a fraction of the time spent under part-load.
Why, the reader might ask, would a race engine-testing group focus on steady-state points rather than peak power? Typically, all the advertising we see for performance parts centers around one number-peak horsepower. The importance of the torque and power curves were discussed by Jim McFarland in multiple issues of Circle Track and the benefits of both fuel injection and E85 over race fuel were openly presented. Now, it is time to use the steady-state testing points to begin explaining their basis of use.
Was there a chance that combustion efficiency might be affected, where in the engine speed range could this occur, and could we link it to the differences between carburetion and EFI?
During the combustion process, it's inevitable that a portion of the fuel does not react to completion and release the full amount of energy present. This is the combined result of unfavorable local mixing consistency within the combustion chamber, cylinder wall wetting effects, and dynamic cycle-to-cycle variations.
At the onset of our engine testing, with regard to combustion efficiency, one question for which we wanted to gain some insight was whether we could make some generalizations between the race fuel and E85, as well as EFI versus carburetion.
Fortunately, by utilizing the Portable Emissions Measuring System (PEMS) unit for the series of seven steady-state operating dynamometer testing points, we could begin to answer some of these questions. For example, by measuring the unburned hydrocarbon (HC) and carbon monoxide (CO) emissions, we were able to gain some insight into the behavior of the various fuels within the engine. Taking the steady-state points presented in Figure 2, and weighting the emissions together, we generated the approximated on-track cycle-weighted emissions with the mass-specific units for each of the criteria emissions, measured in grams of emissions/kg fuel burned.
In Figure 3, we see that both the approximate cycle-weighted HC and CO emissions for the fuel-injected, 100-octane race fuel are lower than for the carbureted version. This indicates that the combustion efficiency for fuel-injection is greater than that for the carbureted version.
As mentioned earlier, mixing efficiency plays a significant role in this. The data show improved mixture preparation for the EFI system compared with carburetion. Since these results hold over a broad, steady-state operating range, it can be concluded that the fuel-injection metering system delivers better proportioned and/or mixed fuel (per cylinder) than the carburetor.
However, comparing the E85 EFI results, one sees higher CO and HC emissions relative to the gasoline-fueled counterpart. Since the energy density of ethanol fuels is lower than the petroleum counterparts, larger volumes must be injected per cycle (in this case, approximately 20 percent more fuel). This larger volume of fuel would require more mixing time, and/or better atomization by optimizing injector size. Something we intend to address during the on-track testing of the project. As the project progresses, the team will be looking further into these results to share with our readers.
Figure 6 - Full-load power pull, X-axis equals time (in seconds), Y-axis equals HC emissio
How much power would be sacrificed when catalytic converters are installed? Was the use of such converters going to make any significant difference in emissions levels from a bona fide racing engine?
Important questions that the Project G.R.E.E.N team had concerning the use of catalytic converters involved the potential power loss and the type of conversion efficiency that would be achieved. In modern vehicles, great effort has gone into designing catalytic converters with high emission-conversion efficiency, while simultaneously minimizing engine power loss and fuel consumption. Would the higher power levels and flow rates associated with larger displacement race engines impact the loss? How effectively would HC, CO, and oxides of nitrogen (NOx) emissions be reduced? Would the catalyst endure strenuous, full-power dynamometer testing?
Comparing the full-power sweep results, we determined a good bit of information on the amount of power loss associated with catalytic convertor usage. For this test matrix, the engine was run utilizing E85 and the fuel injection manifold, along with two different catalyst substrates. One catalyst contained a 100-cell-per-inch (CPI) substrate and the second a 300-CPI substrate. The higher number of cells per substrate equates to an increased catalyst surface area, and, theoretically, would result in increased backpressure. This increase in backpressure should result in a loss of power, the results of which are shown in Figure 4.
In Figure 4, we see that a peak of 1 hp and 1 lb-ft of torque was lost using the 100-CPI catalyst, with almost 4 hp and 10 lb-ft of torque for the 300-CPI catalyst, compared with no catalyst use. Substantially more pressure drop is created with the smaller-celled substrate (300 CPI).
However, peak values do not always tell the complete story, so the entire power curve should be considered. For engine speeds less than 3,250 rpm, there is a slight increase in torque for the 100-CPI catalyst (relative to no catalyst) and almost no loss associated with the 300-CPI catalyst. Reversion and exhaust tuning effects are most likely the culprits, resulting in slight elevations of in-cylinder trapped mass and generating slightly higher torque values. However, as the engine flow (rpm) increases and we observe the area under the power curve that is important on the racetrack (3,500 to redline rpm), a consistent deficit is noted.
In the study of compressible fluid flow, for relatively low velocities (Mach numbers less than 1), as the cross-sectional areas decrease, the resulting pressure of the fluid drops and the flow rate increases. This pressure drop requires energy, and in the case of a reciprocating engine, some of the energy used to push the pistons drives this pressure drop (thus, less power at the crank).
As the exhaust flow enters the catalyst, the variation in flow diameter coupled with the small cells of the substrate causes this pressure drop across the catalyst. In theory, the more cells per substrate, the more restrictive the flow and the higher the loss. This is ultimately what was observed when comparing the 100-CPI, 300-CPI, and no catalyst engine configurations. However, it is telling to note how little power is lost using the higher flow, 100-CPI catalyst; a single lb-ft of torque; and approximate single horsepower across a portion of the powerband.
Final questions concerning the catalysts centered on their effectiveness in converting the criteria emissions of unburned HC, CO, and NOx. Since the engine was tuned for peak power at relatively rich A/F ratios, would there be sufficient oxygen (O2) in the exhaust stream to oxidize the HC and CO, while reducing the NOx? Would the catalysts be effective for such high loads and flow rates, and would the substrate remain durable during repeated full-load testing?
The results were very telling. To frame the discussion, we will review the relative A/F lambda values for which the engine was tested, as shown in Figure 5.
Figure 7 - Full-load power pull, X-axis equals time (in seconds), Y-axis equals CO emissio
It may be seen in Figure 5 that lambda varies from approximately 0.82-0.87 for a majority of the configurations, with greater inconsistency occurring for the carbureted induction system. These relative A/F ratios were established to maximize power in the calibrations for the fuel injection induction systems, and were simply observed from the carburetor tests (after tuning the jets for near maximum torque). From this, many conclusions can be made.
Reviewing the cycle-weighted emissions in Figure 6, we see that there is actually a very modest reduction in CO/HC relative to the engine not utilizing a catalyst (for the 100-octane fuel), and an actual increase in HC emissions for the E85 FI engine when using the catalyst.
However, there is measurable NOx reduction for all cases utilizing the catalysts. In order to shed light on these results, the global chemical reactions for CO/HC oxidation and NOx formation/reduction are provided in the purple box.
Since the reactions required for the catalyst to properly function require a near stoichiometric A/F charge mix (approximately 14.7:1 for gasoline), it's clearly seen from these reactions that there is insufficient O2 in the exhaust stream for the converters to have a high conversion efficiency of unburned HC and CO when the engine is running a rich A/F equivalence ratio. Also from these reactions, we can see that in order to reduce NOx, diatomic O2 is not required. Therefore, the catalyst is effective for NOx reduction. Further evidence of this can be seen in Figures 6, 7, and 8. By comparing the full-load sweep measurements of HC, CO, and NOx emissions over the full-load pull, at best we see only modest oxidation of CO and HC, even though the 300-CPI catalyst had three times the cell count and reactive surface area of the 100-CPI catalyst.
In theory, significant oxidation would have been achieved if the exhaust stream had the proper composition of air and fuel. Because the engine was calibrated for peak torque across its entire operating range (or the carburetor tuned for peak power), additional O2 in the exhaust stream would have been required to improve catalytic convertor efficiency. This topic will be covered further in future testing as the team heads to the track.
How would the engine respond to a switch from conventional racing gasoline to E85?
In the May issue of CT we covered the base engine testing that was conducted at Mast Motorsports, comparing the 100-octane conventional race fuel to E85 in some detail. From those results, there was (as anticipated) a significant gain in torque from the 93- to 100-octane fuel, with another slight gain in torque from 4,000 rpm to approximately 6,250 rpm for the E85-fueled configuration. These tests were all conducted with the electronic fuel-injection induction system. However, it's here noted that the calibrations were not fully optimized for the E85 fuel due to time constraints.
Referring back to Figure 5, one can see that the fueling rate for E85 was slightly lean relative to the race fuel counterpart, resulting in non-optimal maximum brake torque calibrations. This would have resulted in slightly higher torque values than those presented. That said, the high heat of vaporization of E85 relative to 100-octane race fuel suggests increased in-cylinder A/F charge densities. Also, burn curve characteristics of the E85 fuel, coupled with reduced heat transfer rates (due to the cooler intake charge) would also increase the overall power potential.
For the first round of testing, in-cylinder pressure measurements and the corresponding heat release rate data were not available. The CT team is in the stages of planning a second round of engine testing that will include individual cylinder pres-sure measurements, which will provide us with detailed combustion information for an improved understanding of the actual burn process and the thermodynamics involved. The results will be published in the future, as this project continues to unfold.
In fact, would less fuel be required to produce the same or more power when swapping EFI for carburetion?
Other crucial questions posed by the Circle Track team concerned the overall efficiency of the fuel-metering systems. Would the carburetor require more fuel than the EFI system for a given power? How would fuel-injection affect the power curve and overall efficiency? Although the team was unable to answer all the questions first raised in the project due to time and testing limitations, some surprising answers came to light, as well as other questions we are eager to answer. Analysis of the issue regarding the fueling rate between the carburetor and the fuel injection system highlights the complexity of answering such engineering questions and provides guidance for future testing plans.
In Figure 9, a comparison of the A/F lambda values for the carbureted and fuel-injection metering system is shown for each of the seven test points (shown in Figures 2 and 5).
The first thing to notice is that the overall A/F control for carburetors is inconsistent. Lambda values for the seven test points vary from 0.63~0.85. Since the carburetor fuel flow rate depends largely upon the shape of the venturi(s) and airflow velocity, it may be seen that consistent A/F control is not achieved.
At lower intake air velocities, the fluid flow is dominated by viscosity effects resulting in inconsistent metering. For the fuel injection system, it becomes simply a matter of properly calibrating the fueling table, and under certain systems, utilizing closed loop O2 sensor feedback for consistency.
In analyzing the total fuel flow for each load point, it's counterintuitive that the fuel-injection system consumes more fuel. This is apparent in Figure 10, in which each steady-state test point fuel flow is compared. It may be seen that for all load points, the fuel-injection fuel flow rate is higher. Again, comparing Figure 9 with Figure 10, light is shed upon the reason. The fuel-injection system was calibrated for MBT timing and equivalence ratio.
In other words, spark timing was advanced to the knock limit for each load point, while fuel was simultaneously added until torque began to drop off. The discreet set point mapping of fuel-injection allows for each point to be optimized, rather than be a tradeoff for certain load points.
In the May issue, it was shown that there was a significant torque gain across the powerband for fuel-injection relative to the carbureted system, except at peak power due to tuning issues associated with the production fuel-injection intake manifold relative to the carburetor tuned manifold. This torque advantage was attributed to the discreet point tuning that is available with individual-port fuel-injection systems, which is not available with carburetors.
As the project moves to more advanced stages of engine testing, the team will examine more complex in-cylinder behavior to further differentiate the benefits of fuel metering technologies and will openly publish the results. Stay tuned.
Figure 8 - Full-load power pull, X-axis equals time (in seconds), Y-axis equals NOx emissi
Figure 9 - Individual steady-state test point lambda values, carburetor and fuel-injection
Figure 10 - Individual steady-state test point fuel flow rates, carburetor and fuel-inject