Each Part Of A Gt Power Model Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

To study the effect of calibration target parameters on the performance of the simulation and to investigate the most suitable target parameter/s for simulation of a DI jet based diesel engine combustion model.

To simulate the HRR and NOx emission for fuels other than normal diesel using the GT Power model and then study the effect of engine variables on NOx emission.

We have chosen the software GT Power by Gamma Technologies USA to simulate diesel combustion, GT-Power is used by many engine/vehicle manufactures and developers to simulate and analyze the working principles of engines, one-dimensional gas dynamics is the backbone of GT power, and this is used to calculate the fluid flows and heat transfer. Each part of a GT-Power model is separated into many smaller parts. The small parts have a very small volume and in these small volumes the scalar properties (like pressure, temperature, density and internal energy) are assumed to be constant. The small volumes also have vector properties (like mass flux and velocity) that are transferred across the boundaries; GT-Power determines the properties by simultaneously solving the equations for the conservation of mass, momentum and energy and by balancing exergy.

Modelling combustion

There are broadly two ways in which this can be done either with a non predictive or a predictive model we have chosen the DI jet model, this is a predictive model and therefore is more accurate. It basically predicts the burn rate and NOx emissions although it can also calculate the soot emission this function is generally not used because of inaccuracy.

In order to use the DI jet model we require data for its calibration, this data has to be collect so that it can fine tune the model for the case at hand. The most important set of data which is required before di jet modelling is an accurate injection profile.

Description of the DI jet model

The total injected mass of fuel is divided into five radial parts, and then these parts are further subdivided into liquid fuel, unburned vapour fuel, entrained air and burned gasses. The mass injected is specified in terms of mg/ stroke as this fuel move inwards it gets evaporated and entrained and finally burnt. As the plume spreads the penetration distance is decreased.

The unburned mixture will start to burn according to the temperature, pressure and the air-fuel ratio, as the mixture burns the temperature and composition keep on changing, the NOx is calculated in each of the individual burning zone based on the air fuel ratio and temperature

Building the Model

The construction of the DI jet combustion model is fairly simple it consists of an injector, a cylinder and an engine, In the injector we define the injected mass, the injection timing, the fuel properties, nozzle properties like the number of holes and the diameter and the discharge coefficient and the most important the injection profile. In the cylinder object the initial conditions inside the cylinder, the wall temperature, heat transfer and the combustion object are defined. And after that the engine is defined by specifying the type the speed and the cylinder geometry.

Part 1: To investigate the effect of target calibration parameters on the performance of the simulation and to find the most suitable target parameter/s for simulation of a DI jet based diesel engine combustion model.

After defining the model constants and variables like this we need to calibrate the model in order to do this we require some data the most important is the instantaneous pressure profile inside the cylinder, then we also require the injection profile and other data like the injection mass and speed etc.

In order to calibrate the model we first need to set the cylinder pressure analysis mode to measured+predicted now we can enter the values of cylinder pressure and injection profile in the case setup, along with this we need to assign values to the calibration multipliers in order that they show up in the case setup. The values of the breakup time multiplier and the combustion rate multiplier need not be altered during combustion then we enter the initial values of the multipliers and run the simulation. First of all we will have to calibrate the injection delay multiplier.

In order to calibrate the injection delay we first need to select a few points and study the HRR of these points if there is a gap between the take off points of the predicted burn rate and the measured burn rate then we need to adjust the Overall combustion delay multiplier (CIGN1), in order to delay the ignition of the predicted curve we need to increase the Overall combustion delay multiplier (CIGN1), at this point in time only the start of combustion point is important and so we have to concentrate only on bringing the take off points of the curve as close as possible.

Since the Dilution effect multiplier (CIGN8) comes into play only when there is EGR involved we will use the default values for this multiplier which is 0.75.

Now that the ignition delay has been calibrated we move on to the rest of the curve for this we have three more multipliers to enable us to match the burn rates these are CAAIR entrainment multiplier after combustion, CBAIR the entrainment multiplier before combustion and CWALL the entrainment multiplier after impingement fortunately there is an easier way to calibrate these entrainment multipliers we created a DoE model and we varied the values of each of the multipliers over a considerable range, sometimes at a latter stage these ranges had to be expanded after running the DOE experiments we used the DOE post processor to find the optimum values of the multipliers.

By making use of the DOE post processor one may set targets for the model and the multiplier value which lead to the closest burn rate curve is the best target value for calibrating the DI jet model. But before one uses this tool we should filter out some of the cases as some cases may not be suitable for post processing.

The target data available to us was the Maximum cylinder pressure, the maximum rate of change of pressure the crank angle for max burn rate, the crank angles for 2% , 5%, 10% and 50% heat release.

We have made use of these target points in various combinations in order to study about which combinations give satisfactory results and which do not give satisfactory results and consequently find out the best combination for future use.

After the final results have been obtained for the first case ie 20mg we continue to repeat the same for the 30mg case.

The results of this have been shown in the next section.

Part 2: To simulate the HRR and NOx emission for fuels other than normal diesel and to use the GT Power model and study the effect of engine variables on NOx emission.

The cetane number represents the suitability of a fuel for its use in a compression engine, the cetane number is actually a measure of ignition delay and since we can alter the ignition timing by changing the ignition delay multiplier in GT power we will use this in order to match the heat release rate curves obtained from the combustion of these alternate fuels and so even though the exact characteristics of these fuels are not readily available in GT power we shall be able to simulate the DI jet model and predict the NOx emissions and also compare the predicted heat release rate curves and pressure curves.

There are two multipliers available in GT power to calibrate the model for ignition delay; they are the Overall combustion delay multiplier (CIGN1) and the Dilution effect multiplier (CIGN8).

In order to calibrate the injection delay we first need to select a few points and study the HRR of these points if there is a gap between the take off points of the predicted burn rate and the measured burn rate then we need to adjust the Overall combustion delay multiplier (CIGN1), in order to delay the ignition of the predicted curve we need to increase the Overall combustion delay multiplier (CIGN1), at this point in time only the start of combustion point is important and so we have to concentrate only on bringing the take off points of the curve as close as possible.

Since the Dilution effect multiplier (CIGN8) comes into play only when there is EGR involved we will use the default values for this multiplier which is 0.75.

After the combustion delay multipliers have been calibrated we would normally move on to calibrate the other multipliers by conducting a full factorial DoE simulation and then using the DoE post processor to find the best value of the multipliers but since here we already have studied extensively and found out the values for the entrainment multipliers from the base diesel case we shall make use of these multipliers, here it should be noted that we are making the assumption that the only major effect of using a different fuel is that of a different cetane number, which can be compensated by properly calibrating the injection delay multipliers.

Now that the calibration constants have been determined we can enter these values in the combustion model and run the simulation then the resulting graphs mainly pertaining to the pressure, the burn rate and the NOx emission shall be studied to see if the simulation was acceptable the simulation will be deemed to be acceptable if the predicted pressure curves and the predicted Heat release curves are in close correspondence with the curves obtained from the experimental data and the NOx emission predictions are within an acceptable range.

RESULTS

Part I: The main aim of the entire project is to simulate the GT power model of a diesel engine and then find the engine parameters essential for properly simulating the model.

There are two sets of experiments to be simulated for ordinary diesel one is for a 20 mg/stroke case and the other is the 30mg/stroke case.

NOx emissions, pressure curves and heat release curves are the main parameters by which the quality of the simulation will be judged.

The main engine parameters under investigation are maximum cycle pressure, NOx, maximum burn rate, CA at maximum burn rate, crank angle at 2% heat release, crank angle at 5% heat release, crank angle at 10% heat release and crank angle at 50% heat release.

After the engine model is made using the DI jet model it was essential to calibrate the ignition delay and entrainment multipliers.

The ignition delay multipliers are based on the start of ignition and so they were set in order to match the start of ignition.

The entrainment multipliers required us to use the DOE approach here the value of the multipliers was varied over a range and all the simulation results are recorded. Now a DoE post processor is used in order to attain the most suitable multiplier values. Then these values are substituted in the original GT power model in order to confirm the results.

In order to test all the cases the experiments were divided into cases; firstly the experiments are divided into the 20mg/stroke case and the 30mg/stroke case.

Then the main parameters used in the Doe post processing was examined, Pmax, HR2, HR5, HR10, HR50. Initially NOx was also used but since we would like to predict NOx it was later not used.

Also the parameters were first targets individually and then in pairs, the values for groups of three were not good and so they are left out.

Tests:

20mg/stroke, Pressure: 1000 bar

Target: Pmax

The simulated values of maximum pressure are about 30% greater than the target values. NOx values are too high.

Target: Pmax + HR02

The NOx values are still very high almost 50% more than actual values.

Target: Pmax + HR05

High NOx

Target: Pmax + HR10

High NOx

Target: Pmax + HR50

High NOx

Target: HR02

The target values are reached quite accurately. The values of NOx are in the 10% to 20% error range which is ok. The graphs for the cylinder pressure and heat release rate are also quite fine although the profile of the burn rate is a little different but the peaks are close.

The multiplier values for this case are 0.45, 0.995 and 1.014 for caair, cbair and cwall respectively.

This gives the best multiplier values some sample burn rate and pressure simulation curves are shown below

The Nox values calculated by the simulation are 643, 594, 567 and 554 compared to the reference values of 676.3, 589.4, 535.31 and 528.64, therefore the values are quite accurately simulated.

For 20mg and 1000bar this is the best result.

Figure : Cylinder pressure for 20mg/stroke and 1000bar injection pressure when target is HR02

Figure: Burn Rate for 20mg/stroke and 1000bar injection pressure when target is HR02

Target: HR02 + HR05

The pressure curves are ok but the NOx value are very low.

Target: HR02 +HR10

The NOx values are very low.

Target: HR02 + HR50

The pressure curves heat release rates and the NOx emissions are all fine. The NOx predictions get better at latter injection.

The Nox values calculated by the simulation are 543, 509, 525 and 519 compared to the reference values of 676.3, 589.4, 535.31 and 528.64, therefore the values are quite accurately simulated. Of course these values are not as accurate as those which we got from targeting HR 02 but these are also quite good.

Figure : Cylinder pressure for 20mg/stroke and 1000bar injection pressure when target is HR02+HR50

Figure : Burn rate for 20mg/stroke and 1000bar injection pressure when target is HR02+HR50

Target: HR05

The values of NOx are significantly lower than the target values (more than 50% error), but they follow the trend.

All the other combinations also give similar results with very low NOx.

Target: HR10

The NOx values for the DoE post processor are very low almost half of the target values the pressure curves are not good either although the heat release rates are acceptable.

Target: HR 10 + HR 50

The NOx values are good but the pressure and burn rate curves are slightly worse than the better cases.

Target: HR50

NOx values are 20-30% lower than the target values; the predicted pressure graph is a bit lower than the actual graph although the profile is similar.

Table: Representing the improvement or worsening of simulation performance on adding more data. - represents that there is no improvement with respect to the base case, + represents an improvement, 0 is the base case for comparisions and a + superscript represents a satisfactory simulation.

20 mg 1000 bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

-

Ca of max burn rate

0+

HR2

0++

HR5

0

HR10

0

HR50

-

0

20mg/stroke, 1500bar

Target: Pmax

The pressure values simulated are very high compared to the target values. The NOx values are also quite high they are about 50% greater than the target. All the combinations with Pmax also have very high NOx.

Target: NOx

The NOx values coincide but none of the other parameters are even close to the required values.

Target: dP/da

NOx is ok about 20% error, but the heat release curves are not satisfactory.

Target: dP/da +CA of maximum burn rate

The case has gone worse higher NOx and worse graphs.

Target: dP/da + HR2

Even more worse.

Target: dP/da + HR5

Similar to the last case.

Target: dP/da + HR10

Slightly worse NOx results as compared to the base case.

Target: dP/da + HR50

Much worse

Target: CA at maximum burn rate

This is a decent simulation but the burn rate curves are not good. NOx and pressure are ok.

All the combinations are worse than the first case.

Target: HR 02

When simulated with independent optimization the NOx values have a 30-40% error and they do not follow any trend. With sweep the error is only about 10-15% and the trends are also followed. The graphs are also ok but the cylinder maximum pressure is as usual higher than the target. The multipliers are 0.431,0.976,1.005.

All other combinations except for the one with HR50 are not satisfactory.

Figure : Pressure curves for 20mg/stroke and 1500bar injection pressure when target is HR02+HR50

Figure : Burn rate for 20mg/stroke and 1500bar injection pressure when target is HR02+HR50

Target HR05

The trends are followed here also as in the last case but the errors in NOx are higher almost in the 20-30% error range.

The combination with HR02 is better.

Target HR10

The NOx values in the post processor were quite good but the values from GT post were 50% higher than target.

The combination with HR02 is satisfactory.

Target: HR 50

The NOx values in the Doe post follow the trend and are only slightly lower than the target values, the values from the GT post were almost 25% higher and they also followed the trend.

The combination with HR02 is satisfactory.

Table: Representing the improvement or worsening of simulation performance on adding more data. - represents that there is no improvement with respect to the base case, + represents an improvement, 0 is the base case for comparisions and a + superscript represents a satisfactory simulation.

20mg 1500bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

-

Ca of max burn rate

0+

HR2

0++

HR5

0+

HR10

0

HR50

-

0

20mg/stroke, 2000bar

Target: Pmax

The maximum pressure is as usual much higher than the target values. The NOx values are about 50% higher but the values follow the trend. All the combinations involving Pmax have very high NOx.

Target: HR02

The NOx values are about 40-50% higher, therefore no GT post simulation was carried out.

Target HR02+HR10

Good results the simulation is satisfactory. The results are slightly worse for HR02 and HR 50 but still it is satisfactory.

Target HR05

The NOx values have a 33% error on the lower side. But the pressure curves and the heat release rates are very good.

Target HR10

The results are very similar to the last case the error in the NOx values is about 30% and the pressure and heat release rates are very good.

Target HR50

The best results have been obtained at HR 50 the pressure and the heat release rates are very good and the NOx emission values follow the trend and have less than 20% error.

Figure : Pressure curve for 20mg/stroke and 2000bar injection pressure when target is HR50

Figure : Burn rate for 20mg/stroke and 2000bar injection pressure when target is HR50

Table: Representing the improvement or worsening of simulation performance on adding more data. - represents that there is no improvement with respect to the base case, + represents an improvement, 0 is the base case for comparisions and a + superscript represents a satisfactory simulation.

20mg 2000bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

-

Ca of max burn rate

0

HR2

0

HR5

0+

HR10

0

HR50

-

0++

20mg/stroke, 2250bar

Target: Pmax

The Pmax values are off by more than 10%

The NOx values are almost 50% higher than the target, and the rest of the Pmax combinations also have very high NOx.

Target: dP/da

All these combinations have unsatisfactory burn rate predictions.

Target NOx:

The NOx values are very well simulated but the other parameter such are the pressure curve and the heat release curves are not very close.

Target Pmax and NOx:

The values of NOx now have a 40% error and there is no improvement in the values of Pmax.

Target: CA of maximum burn rate

This gives a satisfactory result, the results get slightly worse when HR2 and HR5 are in combination but all three are acceptable.

Target HR02

This is a good simulation close burn rate and good NOx predictions.

The NOx values are close for SOI-4 and SOI-2 but the values get worse as the SOI increases.

Target HR05

The simulated NOx values are ok and the pressure curves and the heat release rates are also quite good. Satisfactory

Target HR50:

The Pmax values are still high, but the values for NOx are very good they have max error of about 20%. This is the best simulation.

Target Hr 50 +HR02

The values are worse than the values for individual cases. The error is greater than 30%.

Table: Representing the improvement or worsening of simulation performance on adding more data. - represents that there is no improvement with respect to the base case, + represents an improvement, 0 is the base case for comparisions and a + superscript represents a satisfactory simulation.

20mg 2250 bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0+

-

Ca of max burn rate

0+

HR2

0+

HR5

0

HR10

0

HR50

-

0++

30mg/stroke, 1000 and 1500 bar

Target: Pmax

The values of NOx are not good there is about 80% error in the simulated values and the graphs for cylinder pressure and heat release rates is also not acceptable.

Target Pmax + HR02

The NOx values are still too high to be acceptable; the values now are almost the same as in the last case. The NOx values are still very high and the graphs for heat release rates are not good.

Target Pmax + HR05

The NOx values are still very high there is no significant improvement in the prediction.

Target Pmax + HR10

Still no improvement in the predicted NOx values they are more than twice the experimental values.

Target Pmax + HR 50

The results are similar to the last case The values of NOx are not good there is about 70% error in the simulated values and the graphs for cylinder pressure and heat release rates is also not acceptable. The values are in the same range as in the prior cases.

Target HR 02

The graphs for burn rate and pressure are ok for 1000bar pressure but the values get worse for 1500 bar. The simulated NOx values are also good for the 1000 bar case and they too get worse for 1500bar the worst value is recorded has a 30% error.

Target HR 02 +HR 05

The NOx emission values are getting worse. The error is slightly higher now.

Target HR 02 +HR10

The NOx emission values are now significantly worse than before.

Target HR 02 +HR50

Te NOx prediction is even worse now.

Target HR05

The pressure curves and burn rate graphs are ok for the first set but the values get worse for the second set (1500bar)

The NOx emission values for the 1000 bar case is generally good the values are more than 30% higher for very early injection but then for the rest the prediction is fairly accurate. With 1500bar injection pressure the NOx prediction for all cases is very good.

Target HR10+ Hr05

The NOx values predicted are slightly worse that those predicted by HR05 alone but the difference is not a lot.

Target HR05 +HR50

The NOx values are worse than before.

Target HR10

The pressure curves are ok for the 1000 bar injection pressure. But the values of NOx have a greater than 50% error.

Target HR10 +HR50

No improvement in NOx values the error is still about 50%

Target HR50

The graphs for pressure at 1000 bar injection are ok, the graphs for 1500bar are also ok but the second peak in the curve is slightly lower.

The burn rate graphs are ok only for the first four cases after that the curves are not good.

The NOx predictions are also good (10-20% error)

Table: Representing the improvement or worsening of simulation performance on adding more data. - represents that there is no improvement with respect to the base case, + represents an improvement, 0 is the base case for comparisions and a + superscript represents a satisfactory simulation.

30mg 1000bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

-

Ca of max burn rate

0+

HR2

0-

HR5

0++

HR10

0+

HR50

-

0+

30mg/stroke 1500bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

-

Ca of max burn rate

0

HR2

0

HR5

0-

HR10

0

HR50

-

0-

30mg/stroke, 2000 and 2250 bar

Target: Pmax

The pressure curves for are acceptable in all the cases.

The NOx values are about twice of the target values. The burn rate curves are not good.

Target: Pmax +HR02

There is a slight improvement in the NOx results but the change is not very significant.

Target: Pmax +HR05

No improvement in NOx prediction. The values now have a significantly higher error.

Target: Pmax +HR10

Much better values the error is now about 20%

Target: Pmax +HR50

No improvement

Target it HR02

The NOx values have an error of greater than 30%.

Target HR02 + HR05

The values are worse than before.

Target HR10+ HR02

The NOx values are in the same range.

The values of NOx are too high to be acceptable.

Target HR02 + HR50

Target: HR05

The NOx values are too low about 50% lower than target.

Target HR 10

The pressure graphs are ok but the NOx values are very low.

Target HR50

The pressure curves and the burn rates are acceptable and the NOx values are also acceptable as the errors are only 25%.

Table: Representing the improvement or worsening of simulation performance on adding more data. - represents that there is no improvement with respect to the base case, + represents an improvement, 0 is the base case for comparisions and a + superscript represents a satisfactory simulation.

30mg 2000bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

+

Ca of max burn rate

0

HR2

0-

HR5

0+

HR10

0

HR50

-

0

30mg 2250 bar

Pmax

Dp/da

CA of max burn rate

HR2

HR5

HR10

HR50

Pmax

0

Dp/da

0

-

Ca of max burn rate

0-

HR2

0-

HR5

0

HR10

0

HR50

-

0

Part II: Simulation of DI jet combustion model for fuels other than normal Diesel, namely Gas to liquids (GTL) diesel fuel and US worst case diesel and to use the GT Power model and study the effect of engine variables on NOx emission.

The cetane number represents the suitability of a fuel for its use in a compression engine, the cetane number is actually a measure of ignition delay and since we can alter the ignition timing by changing the ignition delay multiplier in GT power we will use this in order to match the heat release rate curves obtained from the combustion of these alternate fuels and so even though the exact characteristics of these fuels are not readily available in GT power we shall be able to simulate the DI jet model and predict the NOx emissions and also compare the predicted heat release rate curves and pressure curves.

There are two multipliers available in GT power to calibrate the model for ignition delay; they are the Overall combustion delay multiplier (CIGN1) and the Dilution effect multiplier (CIGN8).

In order to calibrate the injection delay we first need to select a few points and study the HRR of these points if there is a gap between the take off points of the predicted burn rate and the measured burn rate then we need to adjust the Overall combustion delay multiplier (CIGN1), in order to delay the ignition of the predicted curve we need to increase the Overall combustion delay multiplier (CIGN1), at this point in time only the start of combustion point is important and so we have to concentrate only on bringing the take off points of the curve as close as possible.

Since the Dilution effect multiplier (CIGN8) comes into play only when there is EGR involved we will use the default values for this multiplier which is 0.75.

After the combustion delay multipliers have been calibrated we would normally move on to calibrate the other multipliers by conducting a full factorial DoE simulation and then using the DoE post processor to find the best value of the multipliers but since here we already have studied extensively and found out the values for the entrainment multipliers from the base diesel case we shall make use of these multipliers, here it should be noted that we are making the assumption that the only major effect of using a different fuel is that of a different cetane number, which can be compensated by properly calibrating the injection delay multipliers.

Now that the calibration constants have been determined we can enter these values in the combustion model and run the simulation then the resulting graphs mainly pertaining to the pressure, the burn rate and the NOx emission shall be studied to see if the simulation was acceptable the simulation will be deemed to be acceptable if the predicted pressure curves and the predicted Heat release curves are in close correspondence with the curves obtained from the experimental data and the NOx emission predictions are within an acceptable range.

Results

Since there are almost 40 cases showing the all the graphs for all the cases is not possible here therefore only one case from the 20mg and one from the 30mg set are shown over here the rest of the graphs are enclose in the appendix.

Case 1:

Fuel: USWCD

Injected mass: 20mg/stroke

Injection pressure: 1000 bar

SOI: -40 CA

The pressure curves are shown below

Figure: Pressure vs crank angle for USWCD 20mg/stroke, 1000 bar, SOI -40 CA

Figure: Burn rate for USWCD 20mg/stroke, 1000 bar, SOI -40 CA

Since both the pressure curves and the burn rates are quite close we have reason to believe that this is a good simulation.

The NOx predictions will give us the complete picture

Figure: NOx vs crank angle for USWCD 20mg/stroke, 1000 bar, SOI -40 CA

The NOx emission predicted here is 806 ppm which is quite close to the experimentally measured value of 892.7 ppm this gives us an error of 9.7% which is quite acceptable.

The rest of the NOx predictions are shown in the following table:

Table: NOx predicted and actual values for USWCD at 20mg/stroke

No.

Nox [ppm]

rail pressure [bar]

SOI [&deg;CA ATDC)

NOX Predictions

Error %

1

892,7

1000

-4

806

9,72154402

2

846,1

1000

-2

748

11,5466426

3

829,1

1000

0

705

14,9915954

4

735,9

1000

1

773

4,96932256

8

1251,4

1500

-4

1098,71

12,2001359

9

1188,6

1500

-2

1002,08

15,6947333

10

1104,5

1500

0

928

15,9420463

11

995,7

1500

1

959

3,7095531

15

1517,0

2000

-4

1268,98

16,3474163

16

1404,7

2000

-2

1178,27

16,121494

17

1270,9

2000

0

1075,16

15,401781

18

1090,9

2000

1

1061,22

2,72165275

25

1644,4

2250

-4

1387,02

15,6494001

26

1491,3

2250

-2

1269

14,9055898

27

1328,2

2250

0

1145,02

13,7898631

28

1178,9

2250

1

1054,47

10,5529344

The overall error here is 12.1%.

Case 2:

Fuel: GTL

Injected mass: 20mg/stroke

Injection pressure: 2000 bar

SOI: 10 CA

The pressure curves are shown below

Figure: Pressure vs crank angle for GTL 20mg/stroke, 2000 bar, SOI 10 CA

Figure: Burn rate for GTL 20mg/stroke, 2000 bar, SOI 10 CA

Here we can see that the predicted pressure curve is good but the burn rates are not that great, now let us see the NOx prediction

Figure: NOx vs crank angle for GTL 20mg/stroke, 2000 bar, SOI 10 CA

The NOx prediction here is again quite good 794 compared to 814ppm which means an error of only 2.7%

Table: Predicted and experimental NOx for GTL 20mg/stroke.

No.

NOx [ppm]

rail pressure [bar]

SOI [&deg;CA ATDC)

Nox predictions

Error %

1

695,7

1000

-4

525

24,5346129

2

581,1

1000

-2

479

17,6339149

3

504,4

1000

0

459

9,06081454

4

462,9

1000

1

441,27

4,67516648

8

973,5

1500

-4

779,62

19,9117563

9

820,6

1500

-2

721

12,1090755

10

702,5

1500

0

676

3,7979148

11

655,7

1500

1

650

0,87336418

15

1202,5

2000

-4

956,93

20,42397

16

1012,8

2000

-2

889

12,1938732

17

881,0

2000

0

824

6,50460603

18

814,6

2000

1

792

2,73845456

25

1303,1

2250

-4

1110,58

14,7751333

26

1113,6

2250

-2

1023,4

8,09857842

27

956,6

2250

0

949

0,74330869

28

888,6

2250

1

913

2,79274873

Case 3:

Fuel: GTL

Injected mass: 30mg/stroke

Injection pressure: 2000 bar

SOI: 20 CA

The pressure curve is shown below this is a good curve

Figure: Pressure vs crank angle for GTL 30mg/stroke, 2000 bar, SOI 20 CA

Since both the pressure curves and the burn rates are quite accurate this is a good simulation.

Figure: Burn rate vs crank angle for GTL 30mg/stroke, 2000 bar, SOI 20 CA

Figure: NOx vs crank angle for GTL 30mg/stroke, 2000 bar, SOI 20 CA

Here we see that there is only a 12% error in the prediction.

Table: NOx for GTL 30mg/stroke

No.

NOx [ppm]

rail pressure [bar]

SOI [&deg;CA ATDC)

Predicted NOx

Error %

1

954,3

1000

-6

1132,36

18,6529075

2

815,9

1000

-4

1000,84

22,6734881

3

692,5

1000

-2

870

25,5557923

4

590,3

1000

0

758

28,4199401

5

503,8

1000

2

656

30,227746

6

435,7

1000

4

584

34,0000946

13

1393,7

1500

-6

1236,87

11,2514586

14

1172,3

1500

-4

1083,23

7,59779934

15

985,9

1500

-2

950

3,65387962

16

828,8

1500

0

837

1,00299119

17

702,8

1500

2

747

6,26593841

18

608,9

1500

4

693

13,7341563

25

1483,6

2000

-4

1301,97

12,2397567

26

1242,9

2000

-2

1179,8

5,07791062

27

1032,7

2000

0

1067,8

3,39416457

28

880,3

2000

2

987

12,0792506

29

751,7

2000

4

879,58

17,0196248

35

1615,6

2250

-4

1511,65

6,43162742

36

1367,6

2250

-2

1389,8

1,62174096

37

1145,9

2250

0

1268,83

10,7270873

38

975,4

2250

2

1172,42

20,1974963

39

847,3

2250

4

1086,8

28,2647704

Case 4:

Fuel: USWCD

Injected mass: 30mg/stroke

Injection pressure: 1000 bar

SOI: -60 CA

The graphs for pressure and the burn rate are quite good as can be seen from the following graphs

Figure: Pressure vs crank angle for USWCD 30mg/stroke, 1000 bar, SOI -60 C

Figure: Burn rate for USWCD 30mg/stroke, 1000 bar, SOI -60 CA

Figure: NOx emission vs crank angle for USWC diesel 30mg/stroke, 1000 bar and -60CA

The error in the NOx prediction is also not very high its around 14 %.

Figure: NOx for USWCD 30mg/stroke.

No.

NOx [ppm]

rail pressure [bar]

SOI [&deg;CA ATDC)

Predicted NOx

Error %

1

996,6

1000

-6

1143,14

14,7029176

2

824,2

1000

-4

1009,69

22,5079788

3

724,4

1000

-2

876

20,9789908

4

649,1

1000

0

758

16,7352593

5

616,0

1000

2

663

7,61844526

6

652,5

1000

4

586

10,2361051

13

1413,3

1500

-6

1239,31

12,3119596

14

1226,2

1500

-4

1085,63

11,4622903

15

1068,7

1500

-2

957

10,4728146

16

947,8

1500

0

848

10,5143992

17

891,8

1500

2

757

15,1452036

18

929,0

1500

4

699

24,7987441

25

1556,5

2000

-4

1301,03

16,4147866

26

1342,8

2000

-2

1175,33

12,4696658

27

1192,3

2000

0

1066,62

10,5428162

28

1119,3

2000

2

976

12,8072402

29

1130,6

2000

4

900

20,4103231

35

1719,1

2250

-4

1439,59

16,2597755

36

1491,8

2250

-2

1388,33

6,93695507

37

1341,1

2250

0

1269,97

5,30537947

38

1248,3

2250

2

1170,44

6,2392337

39

1245,2

2250

4

1085,78

12,7999868

Now that we have an acceptable simulation we can study the effect of parameters such as pressure and start of injection on the NOx emission of a diesel engine operating on GTL and USWCD.

Effect of Injection pressure:

The simulation was able to correctly predict the trends in the NOx emission, as the injection pressure is increased there is an increase in the NOx emission from the cylinder, this increase in NOx is attributed to the higher pressure attained in the combustion chamber.

Figure: Effect of injection pressure on NOx emissions for different SOI and GTL fuel at 20mg/stroke

Figure: Effect of injection pressure on NOx emissions for different SOI and USWCD fuel at 20mg/stroke

Figure: Effect of injection pressure on NOx emissions for different SOI and GTL fuel at 30mg/stroke

Figure: Effect of injection pressure on NOx emissions for different SOI and USWCD fuel at 30mg/stroke

Effect of injection timing on NOx emission

From the predicted graphs we see that the simulation is able to correctly predict the trend in the variation of injection timing, the result, as expected, is that the NOx emissions are decreased with ignition advance.

This is because when we retard the injection there is more fuel accumulated inside the cylinder this leads to much higher temperature and pressure and thus more NOx.

Figure: The effect of injection timing on NOx emissions for GTL fuel and 20mg/stroke

Figure: The effect of injection timing on NOx emissions for USWCD fuel and 20mg/stroke

Figure: The effect of injection timing on NOx emissions for GTL fuel and 30mg/stroke

Figure: The effect of injection timing on NOx emissions for USWCD fuel and 30mg/stroke

Effect of Cetane number on NOx emission

As the cetane number increases the ignition delay decreases and so there is a decrease in the NOx emission. This is attributed to the fact that there is less time available for the fuel to accumulate and therefore the peak temperature and pressure in the combustion chamber is lower.

This trend can be observed in the graphs below

Figure: effect of cetane number on NOx emission for different injection pressures keeping SOI constant

Figure: Effect of cetane number on NOx emission for different injection pressures keeping SOI constant

Conclusions

The results especially the pressure curves and the heat release rate graphs are better at lower injection pressure.

The results are in general better for SOI near 0 CA.

On increasing the targets to 2 parameters the results do not necessarily improve.

The heat release rates are better for the 20mg/stroke cases.

HR02 and HR50 are the best parameters to predict NOx, HR02 are better for lower pressures and HR 50 is better for higher pressures.

By manually tweaking the entrainment multipliers the NOx prediction can be made more accurate.

On using independent optimization in the DoE post the NOx emissions do not follow the trend and so the results are worse than Sweep optimization.

It is possible to examine the emission performance of different fuels which have different cetane numbers by calibrating the injection delay multipliers so as to nullify the ignition difference.

In order to predict the results in a case where all the experimental target data is not available it may become necessary to assume the values of the multipliers.

On retarding the injection timing there is an increase in the NOx emissions and the effect is slightly more profound at lower cetane numbers.

On increasing the injection pressure there is an increase in the NOx emissions.

As the cetane number increases there is a decrease in NOx emission.